summaryrefslogtreecommitdiffstats
path: root/src/shutdown/shutdown.c
blob: 0d07865542e0f04759b7be6be1fb31212a0d5491 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
/* SPDX-License-Identifier: LGPL-2.1-or-later */
/***
  Copyright © 2010 ProFUSION embedded systems
***/

#include <errno.h>
#include <getopt.h>
#include <linux/reboot.h>
#include <stdbool.h>
#include <stdlib.h>
#include <sys/mman.h>
#include <sys/mount.h>
#include <sys/reboot.h>
#include <sys/stat.h>
#include <unistd.h>

#include "alloc-util.h"
#include "async.h"
#include "binfmt-util.h"
#include "cgroup-setup.h"
#include "cgroup-util.h"
#include "def.h"
#include "exec-util.h"
#include "fd-util.h"
#include "fileio.h"
#include "killall.h"
#include "log.h"
#include "parse-util.h"
#include "process-util.h"
#include "reboot-util.h"
#include "rlimit-util.h"
#include "signal-util.h"
#include "string-util.h"
#include "switch-root.h"
#include "sysctl-util.h"
#include "terminal-util.h"
#include "umount.h"
#include "util.h"
#include "virt.h"
#include "watchdog.h"

#define SYNC_PROGRESS_ATTEMPTS 3
#define SYNC_TIMEOUT_USEC (10*USEC_PER_SEC)

static char* arg_verb;
static uint8_t arg_exit_code;
static usec_t arg_timeout = DEFAULT_TIMEOUT_USEC;

static int parse_argv(int argc, char *argv[]) {
        enum {
                ARG_LOG_LEVEL = 0x100,
                ARG_LOG_TARGET,
                ARG_LOG_COLOR,
                ARG_LOG_LOCATION,
                ARG_LOG_TIME,
                ARG_EXIT_CODE,
                ARG_TIMEOUT,
        };

        static const struct option options[] = {
                { "log-level",     required_argument, NULL, ARG_LOG_LEVEL    },
                { "log-target",    required_argument, NULL, ARG_LOG_TARGET   },
                { "log-color",     optional_argument, NULL, ARG_LOG_COLOR    },
                { "log-location",  optional_argument, NULL, ARG_LOG_LOCATION },
                { "log-time",      optional_argument, NULL, ARG_LOG_TIME     },
                { "exit-code",     required_argument, NULL, ARG_EXIT_CODE    },
                { "timeout",       required_argument, NULL, ARG_TIMEOUT      },
                {}
        };

        int c, r;

        assert(argc >= 1);
        assert(argv);

        /* "-" prevents getopt from permuting argv[] and moving the verb away
         * from argv[1]. Our interface to initrd promises it'll be there. */
        while ((c = getopt_long(argc, argv, "-", options, NULL)) >= 0)
                switch (c) {

                case ARG_LOG_LEVEL:
                        r = log_set_max_level_from_string(optarg);
                        if (r < 0)
                                log_error_errno(r, "Failed to parse log level %s, ignoring: %m", optarg);

                        break;

                case ARG_LOG_TARGET:
                        r = log_set_target_from_string(optarg);
                        if (r < 0)
                                log_error_errno(r, "Failed to parse log target %s, ignoring: %m", optarg);

                        break;

                case ARG_LOG_COLOR:

                        if (optarg) {
                                r = log_show_color_from_string(optarg);
                                if (r < 0)
                                        log_error_errno(r, "Failed to parse log color setting %s, ignoring: %m", optarg);
                        } else
                                log_show_color(true);

                        break;

                case ARG_LOG_LOCATION:
                        if (optarg) {
                                r = log_show_location_from_string(optarg);
                                if (r < 0)
                                        log_error_errno(r, "Failed to parse log location setting %s, ignoring: %m", optarg);
                        } else
                                log_show_location(true);

                        break;

                case ARG_LOG_TIME:

                        if (optarg) {
                                r = log_show_time_from_string(optarg);
                                if (r < 0)
                                        log_error_errno(r, "Failed to parse log time setting %s, ignoring: %m", optarg);
                        } else
                                log_show_time(true);

                        break;

                case ARG_EXIT_CODE:
                        r = safe_atou8(optarg, &arg_exit_code);
                        if (r < 0)
                                log_error_errno(r, "Failed to parse exit code %s, ignoring: %m", optarg);

                        break;

                case ARG_TIMEOUT:
                        r = parse_sec(optarg, &arg_timeout);
                        if (r < 0)
                                log_error_errno(r, "Failed to parse shutdown timeout %s, ignoring: %m", optarg);

                        break;

                case '\001':
                        if (!arg_verb)
                                arg_verb = optarg;
                        else
                                log_error("Excess arguments, ignoring");
                        break;

                case '?':
                        return -EINVAL;

                default:
                        assert_not_reached("Unhandled option code.");
                }

        if (!arg_verb)
                return log_error_errno(SYNTHETIC_ERRNO(EINVAL),
                                       "Verb argument missing.");

        return 0;
}

static int switch_root_initramfs(void) {
        if (mount("/run/initramfs", "/run/initramfs", NULL, MS_BIND, NULL) < 0)
                return log_error_errno(errno, "Failed to mount bind /run/initramfs on /run/initramfs: %m");

        if (mount(NULL, "/run/initramfs", NULL, MS_PRIVATE, NULL) < 0)
                return log_error_errno(errno, "Failed to make /run/initramfs private mount: %m");

        /* switch_root with MS_BIND, because there might still be processes lurking around, which have open file descriptors.
         * /run/initramfs/shutdown will take care of these.
         * Also do not detach the old root, because /run/initramfs/shutdown needs to access it.
         */
        return switch_root("/run/initramfs", "/oldroot", false, MS_BIND);
}

/* Read the following fields from /proc/meminfo:
 *
 *  NFS_Unstable
 *  Writeback
 *  Dirty
 *
 * Return true if the sum of these fields is greater than the previous
 * value input. For all other issues, report the failure and indicate that
 * the sync is not making progress.
 */
static int sync_making_progress(unsigned long long *prev_dirty) {
        _cleanup_fclose_ FILE *f = NULL;
        unsigned long long val = 0;
        int ret;

        f = fopen("/proc/meminfo", "re");
        if (!f)
                return log_warning_errno(errno, "Failed to open /proc/meminfo: %m");

        for (;;) {
                _cleanup_free_ char *line = NULL;
                unsigned long long ull = 0;
                int q;

                q = read_line(f, LONG_LINE_MAX, &line);
                if (q < 0)
                        return log_warning_errno(q, "Failed to parse /proc/meminfo: %m");
                if (q == 0)
                        break;

                if (!first_word(line, "NFS_Unstable:") && !first_word(line, "Writeback:") && !first_word(line, "Dirty:"))
                        continue;

                errno = 0;
                if (sscanf(line, "%*s %llu %*s", &ull) != 1) {
                        if (errno != 0)
                                log_warning_errno(errno, "Failed to parse /proc/meminfo: %m");
                        else
                                log_warning("Failed to parse /proc/meminfo");

                        return false;
                }

                val += ull;
        }

        ret = *prev_dirty > val;
        *prev_dirty = val;
        return ret;
}

static void sync_with_progress(void) {
        unsigned long long dirty = ULLONG_MAX;
        unsigned checks;
        pid_t pid;
        int r;

        BLOCK_SIGNALS(SIGCHLD);

        /* Due to the possibility of the sync operation hanging, we fork a child process and monitor
         * the progress. If the timeout lapses, the assumption is that the particular sync stalled. */

        r = asynchronous_sync(&pid);
        if (r < 0) {
                log_error_errno(r, "Failed to fork sync(): %m");
                return;
        }

        log_info("Syncing filesystems and block devices.");

        /* Start monitoring the sync operation. If more than
         * SYNC_PROGRESS_ATTEMPTS lapse without progress being made,
         * we assume that the sync is stalled */
        for (checks = 0; checks < SYNC_PROGRESS_ATTEMPTS; checks++) {
                r = wait_for_terminate_with_timeout(pid, SYNC_TIMEOUT_USEC);
                if (r == 0)
                        /* Sync finished without error.
                         * (The sync itself does not return an error code) */
                        return;
                else if (r == -ETIMEDOUT) {
                        /* Reset the check counter if the "Dirty" value is
                         * decreasing */
                        if (sync_making_progress(&dirty) > 0)
                                checks = 0;
                } else {
                        log_error_errno(r, "Failed to sync filesystems and block devices: %m");
                        return;
                }
        }

        /* Only reached in the event of a timeout. We should issue a kill
         * to the stray process. */
        log_error("Syncing filesystems and block devices - timed out, issuing SIGKILL to PID "PID_FMT".", pid);
        (void) kill(pid, SIGKILL);
}

static int read_current_sysctl_printk_log_level(void) {
        _cleanup_free_ char *sysctl_printk_vals = NULL, *sysctl_printk_curr = NULL;
        int current_lvl;
        const char *p;
        int r;

        r = sysctl_read("kernel/printk", &sysctl_printk_vals);
        if (r < 0)
                return log_debug_errno(r, "Cannot read sysctl kernel.printk: %m");

        p = sysctl_printk_vals;
        r = extract_first_word(&p, &sysctl_printk_curr, NULL, 0);
        if (r < 0)
                return log_debug_errno(r, "Failed to split out kernel printk priority: %m");
        if (r == 0)
                return log_debug_errno(SYNTHETIC_ERRNO(EINVAL), "Short read while reading kernel.printk sysctl");

        r = safe_atoi(sysctl_printk_curr, &current_lvl);
        if (r < 0)
                return log_debug_errno(r, "Failed to parse kernel.printk sysctl: %s", sysctl_printk_vals);

        return current_lvl;
}

static void bump_sysctl_printk_log_level(int min_level) {
        int current_lvl, r;

        /* Set the logging level to be able to see messages with log level smaller or equal to min_level */

        current_lvl = read_current_sysctl_printk_log_level();
        if (current_lvl < 0 || current_lvl >= min_level + 1)
                return;

        r = sysctl_writef("kernel/printk", "%i", min_level + 1);
        if (r < 0)
                log_debug_errno(r, "Failed to bump kernel.printk to %i: %m", min_level + 1);
}

int main(int argc, char *argv[]) {
        bool need_umount, need_swapoff, need_loop_detach, need_dm_detach, need_md_detach, in_container, use_watchdog = false, can_initrd;
        _cleanup_free_ char *cgroup = NULL;
        char *arguments[3], *watchdog_device;
        int cmd, r, umount_log_level = LOG_INFO;
        static const char* const dirs[] = {SYSTEM_SHUTDOWN_PATH, NULL};

        /* The log target defaults to console, but the original systemd process will pass its log target in through a
         * command line argument, which will override this default. Also, ensure we'll never log to the journal or
         * syslog, as these logging daemons are either already dead or will die very soon. */

        log_set_target(LOG_TARGET_CONSOLE);
        log_set_prohibit_ipc(true);
        log_parse_environment();

        r = parse_argv(argc, argv);
        if (r < 0)
                goto error;

        log_open();

        umask(0022);

        if (getpid_cached() != 1) {
                log_error("Not executed by init (PID 1).");
                r = -EPERM;
                goto error;
        }

        if (streq(arg_verb, "reboot"))
                cmd = RB_AUTOBOOT;
        else if (streq(arg_verb, "poweroff"))
                cmd = RB_POWER_OFF;
        else if (streq(arg_verb, "halt"))
                cmd = RB_HALT_SYSTEM;
        else if (streq(arg_verb, "kexec"))
                cmd = LINUX_REBOOT_CMD_KEXEC;
        else if (streq(arg_verb, "exit"))
                cmd = 0; /* ignored, just checking that arg_verb is valid */
        else {
                log_error("Unknown action '%s'.", arg_verb);
                r = -EINVAL;
                goto error;
        }

        (void) cg_get_root_path(&cgroup);
        in_container = detect_container() > 0;

        /* If the logging messages are going to KMSG, and if we are not running from a container, then try to
         * update the sysctl kernel.printk current value in order to see "info" messages; This current log
         * level is not updated if already big enough.
         */
        if (!in_container &&
            IN_SET(log_get_target(),
                   LOG_TARGET_AUTO,
                   LOG_TARGET_JOURNAL_OR_KMSG,
                   LOG_TARGET_SYSLOG_OR_KMSG,
                   LOG_TARGET_KMSG))
                bump_sysctl_printk_log_level(LOG_WARNING);

        use_watchdog = getenv("WATCHDOG_USEC");
        watchdog_device = getenv("WATCHDOG_DEVICE");
        if (watchdog_device) {
                r = watchdog_set_device(watchdog_device);
                if (r < 0)
                        log_warning_errno(r, "Failed to set watchdog device to %s, ignoring: %m",
                                          watchdog_device);
        }

        /* Lock us into memory */
        (void) mlockall(MCL_CURRENT|MCL_FUTURE);

        /* Synchronize everything that is not written to disk yet at this point already. This is a good idea so that
         * slow IO is processed here already and the final process killing spree is not impacted by processes
         * desperately trying to sync IO to disk within their timeout. Do not remove this sync, data corruption will
         * result. */
        if (!in_container)
                sync_with_progress();

        disable_coredumps();
        disable_binfmt();

        log_info("Sending SIGTERM to remaining processes...");
        broadcast_signal(SIGTERM, true, true, arg_timeout);

        log_info("Sending SIGKILL to remaining processes...");
        broadcast_signal(SIGKILL, true, false, arg_timeout);

        need_umount = !in_container;
        need_swapoff = !in_container;
        need_loop_detach = !in_container;
        need_dm_detach = !in_container;
        need_md_detach = !in_container;
        can_initrd = !in_container && !in_initrd() && access("/run/initramfs/shutdown", X_OK) == 0;

        /* Unmount all mountpoints, swaps, and loopback devices */
        for (;;) {
                bool changed = false;

                if (use_watchdog)
                        (void) watchdog_ping();

                /* Let's trim the cgroup tree on each iteration so
                   that we leave an empty cgroup tree around, so that
                   container managers get a nice notify event when we
                   are down */
                if (cgroup)
                        (void) cg_trim(SYSTEMD_CGROUP_CONTROLLER, cgroup, false);

                if (need_umount) {
                        log_info("Unmounting file systems.");
                        r = umount_all(&changed, umount_log_level);
                        if (r == 0) {
                                need_umount = false;
                                log_info("All filesystems unmounted.");
                        } else if (r > 0)
                                log_info("Not all file systems unmounted, %d left.", r);
                        else
                                log_error_errno(r, "Failed to unmount file systems: %m");
                }

                if (need_swapoff) {
                        log_info("Deactivating swaps.");
                        r = swapoff_all(&changed);
                        if (r == 0) {
                                need_swapoff = false;
                                log_info("All swaps deactivated.");
                        } else if (r > 0)
                                log_info("Not all swaps deactivated, %d left.", r);
                        else
                                log_error_errno(r, "Failed to deactivate swaps: %m");
                }

                if (need_loop_detach) {
                        log_info("Detaching loop devices.");
                        r = loopback_detach_all(&changed, umount_log_level);
                        if (r == 0) {
                                need_loop_detach = false;
                                log_info("All loop devices detached.");
                        } else if (r > 0)
                                log_info("Not all loop devices detached, %d left.", r);
                        else
                                log_error_errno(r, "Failed to detach loop devices: %m");
                }

                if (need_md_detach) {
                        log_info("Stopping MD devices.");
                        r = md_detach_all(&changed, umount_log_level);
                        if (r == 0) {
                                need_md_detach = false;
                                log_info("All MD devices stopped.");
                        } else if (r > 0)
                                log_info("Not all MD devices stopped, %d left.", r);
                        else
                                log_error_errno(r, "Failed to stop MD devices: %m");
                }

                if (need_dm_detach) {
                        log_info("Detaching DM devices.");
                        r = dm_detach_all(&changed, umount_log_level);
                        if (r == 0) {
                                need_dm_detach = false;
                                log_info("All DM devices detached.");
                        } else if (r > 0)
                                log_info("Not all DM devices detached, %d left.", r);
                        else
                                log_error_errno(r, "Failed to detach DM devices: %m");
                }

                if (!need_umount && !need_swapoff && !need_loop_detach && !need_dm_detach
                            && !need_md_detach) {
                        log_info("All filesystems, swaps, loop devices, MD devices and DM devices detached.");
                        /* Yay, done */
                        break;
                }

                if (!changed && umount_log_level == LOG_INFO && !can_initrd) {
                        /* There are things we cannot get rid of. Loop one more time
                         * with LOG_ERR to inform the user. Note that we don't need
                         * to do this if there is a initrd to switch to, because that
                         * one is likely to get rid of the remounting mounts. If not,
                         * it will log about them. */
                        umount_log_level = LOG_ERR;
                        continue;
                }

                /* If in this iteration we didn't manage to
                 * unmount/deactivate anything, we simply give up */
                if (!changed) {
                        log_info("Cannot finalize remaining%s%s%s%s%s continuing.",
                                 need_umount ? " file systems," : "",
                                 need_swapoff ? " swap devices," : "",
                                 need_loop_detach ? " loop devices," : "",
                                 need_dm_detach ? " DM devices," : "",
                                 need_md_detach ? " MD devices," : "");
                        break;
                }

                log_debug("Couldn't finalize remaining %s%s%s%s%s trying again.",
                          need_umount ? " file systems," : "",
                          need_swapoff ? " swap devices," : "",
                          need_loop_detach ? " loop devices," : "",
                          need_dm_detach ? " DM devices," : "",
                          need_md_detach ? " MD devices," : "");
        }

        /* We're done with the watchdog. */
        watchdog_free_device();

        arguments[0] = NULL;
        arguments[1] = arg_verb;
        arguments[2] = NULL;
        (void) execute_directories(dirs, DEFAULT_TIMEOUT_USEC, NULL, NULL, arguments, NULL, EXEC_DIR_PARALLEL | EXEC_DIR_IGNORE_ERRORS);

        (void) rlimit_nofile_safe();

        if (can_initrd) {
                r = switch_root_initramfs();
                if (r >= 0) {
                        argv[0] = (char*) "/shutdown";

                        (void) setsid();
                        (void) make_console_stdio();

                        log_info("Successfully changed into root pivot.\n"
                                 "Returning to initrd...");

                        execv("/shutdown", argv);
                        log_error_errno(errno, "Failed to execute shutdown binary: %m");
                } else
                        log_error_errno(r, "Failed to switch root to \"/run/initramfs\": %m");
        }

        if (need_umount || need_swapoff || need_loop_detach || need_dm_detach || need_md_detach)
                log_error("Failed to finalize%s%s%s%s%s ignoring.",
                          need_umount ? " file systems," : "",
                          need_swapoff ? " swap devices," : "",
                          need_loop_detach ? " loop devices," : "",
                          need_dm_detach ? " DM devices," : "",
                          need_md_detach ? " MD devices," : "");

        /* The kernel will automatically flush ATA disks and suchlike on reboot(), but the file systems need to be
         * sync'ed explicitly in advance. So let's do this here, but not needlessly slow down containers. Note that we
         * sync'ed things already once above, but we did some more work since then which might have caused IO, hence
         * let's do it once more. Do not remove this sync, data corruption will result. */
        if (!in_container)
                sync_with_progress();

        if (streq(arg_verb, "exit")) {
                if (in_container)
                        return arg_exit_code;

                cmd = RB_POWER_OFF; /* We cannot exit() on the host, fallback on another method. */
        }

        switch (cmd) {

        case LINUX_REBOOT_CMD_KEXEC:

                if (!in_container) {
                        /* We cheat and exec kexec to avoid doing all its work */
                        log_info("Rebooting with kexec.");

                        r = safe_fork("(sd-kexec)", FORK_RESET_SIGNALS|FORK_CLOSE_ALL_FDS|FORK_LOG|FORK_WAIT, NULL);
                        if (r == 0) {
                                const char * const args[] = {
                                        KEXEC, "-e", NULL
                                };

                                /* Child */

                                execv(args[0], (char * const *) args);

                                /* execv failed (kexec binary missing?), so try simply reboot(RB_KEXEC) */
                                (void) reboot(cmd);
                                _exit(EXIT_FAILURE);
                        }

                        /* If we are still running, then the kexec can't have worked, let's fall through */
                }

                cmd = RB_AUTOBOOT;
                _fallthrough_;

        case RB_AUTOBOOT:
                (void) reboot_with_parameter(REBOOT_LOG);
                log_info("Rebooting.");
                break;

        case RB_POWER_OFF:
                log_info("Powering off.");
                break;

        case RB_HALT_SYSTEM:
                log_info("Halting system.");
                break;

        default:
                assert_not_reached("Unknown magic");
        }

        (void) reboot(cmd);
        if (errno == EPERM && in_container) {
                /* If we are in a container, and we lacked
                 * CAP_SYS_BOOT just exit, this will kill our
                 * container for good. */
                log_info("Exiting container.");
                return EXIT_SUCCESS;
        }

        r = log_error_errno(errno, "Failed to invoke reboot(): %m");

  error:
        log_emergency_errno(r, "Critical error while doing system shutdown: %m");
        freeze();
}