2 * Copyright (c) 1986, 1988, 1991, 1993
3 * The Regents of the University of California. All rights reserved.
4 * (c) UNIX System Laboratories, Inc.
5 * All or some portions of this file are derived from material licensed
6 * to the University of California by American Telephone and Telegraph
7 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
8 * the permission of UNIX System Laboratories, Inc.
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 4. Neither the name of the University nor the names of its contributors
19 * may be used to endorse or promote products derived from this software
20 * without specific prior written permission.
22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * @(#)kern_shutdown.c 8.3 (Berkeley) 1/21/94
37 #include <sys/cdefs.h>
38 __FBSDID("$FreeBSD$");
42 #include "opt_panic.h"
43 #include "opt_sched.h"
44 #include "opt_watchdog.h"
46 #include <sys/param.h>
47 #include <sys/systm.h>
52 #include <sys/eventhandler.h>
55 #include <sys/kernel.h>
56 #include <sys/kerneldump.h>
57 #include <sys/kthread.h>
59 #include <sys/malloc.h>
60 #include <sys/mount.h>
63 #include <sys/reboot.h>
64 #include <sys/resourcevar.h>
65 #include <sys/rwlock.h>
66 #include <sys/sched.h>
68 #include <sys/sysctl.h>
69 #include <sys/sysproto.h>
70 #include <sys/vnode.h>
71 #include <sys/watchdog.h>
75 #include <machine/cpu.h>
76 #include <machine/dump.h>
77 #include <machine/pcb.h>
78 #include <machine/smp.h>
80 #include <security/mac/mac_framework.h>
83 #include <vm/vm_object.h>
84 #include <vm/vm_page.h>
85 #include <vm/vm_pager.h>
86 #include <vm/swap_pager.h>
88 #include <sys/signalvar.h>
90 #ifndef PANIC_REBOOT_WAIT_TIME
91 #define PANIC_REBOOT_WAIT_TIME 15 /* default to 15 seconds */
93 static int panic_reboot_wait_time = PANIC_REBOOT_WAIT_TIME;
94 SYSCTL_INT(_kern, OID_AUTO, panic_reboot_wait_time, CTLFLAG_RWTUN,
95 &panic_reboot_wait_time, 0,
96 "Seconds to wait before rebooting after a panic");
99 * Note that stdarg.h and the ANSI style va_start macro is used for both
100 * ANSI and traditional C compilers.
102 #include <machine/stdarg.h>
105 #ifdef KDB_UNATTENDED
106 int debugger_on_panic = 0;
108 int debugger_on_panic = 1;
110 SYSCTL_INT(_debug, OID_AUTO, debugger_on_panic,
111 CTLFLAG_RWTUN | CTLFLAG_SECURE,
112 &debugger_on_panic, 0, "Run debugger on kernel panic");
115 static int trace_on_panic = 1;
117 static int trace_on_panic = 0;
119 SYSCTL_INT(_debug, OID_AUTO, trace_on_panic,
120 CTLFLAG_RWTUN | CTLFLAG_SECURE,
121 &trace_on_panic, 0, "Print stack trace on kernel panic");
124 static int sync_on_panic = 0;
125 SYSCTL_INT(_kern, OID_AUTO, sync_on_panic, CTLFLAG_RWTUN,
126 &sync_on_panic, 0, "Do a sync before rebooting from a panic");
128 static SYSCTL_NODE(_kern, OID_AUTO, shutdown, CTLFLAG_RW, 0,
129 "Shutdown environment");
132 static int show_busybufs;
134 static int show_busybufs = 1;
136 SYSCTL_INT(_kern_shutdown, OID_AUTO, show_busybufs, CTLFLAG_RW,
137 &show_busybufs, 0, "");
140 * Variable panicstr contains argument to first call to panic; used as flag
141 * to indicate that the kernel has already called panic.
143 const char *panicstr;
145 int dumping; /* system is dumping */
146 int rebooting; /* system is rebooting */
147 static struct dumperinfo dumper; /* our selected dumper */
149 /* Context information for dump-debuggers. */
150 static struct pcb dumppcb; /* Registers. */
151 lwpid_t dumptid; /* Thread ID. */
153 static void poweroff_wait(void *, int);
154 static void shutdown_halt(void *junk, int howto);
155 static void shutdown_panic(void *junk, int howto);
156 static void shutdown_reset(void *junk, int howto);
158 /* register various local shutdown events */
160 shutdown_conf(void *unused)
163 EVENTHANDLER_REGISTER(shutdown_final, poweroff_wait, NULL,
165 EVENTHANDLER_REGISTER(shutdown_final, shutdown_halt, NULL,
166 SHUTDOWN_PRI_LAST + 100);
167 EVENTHANDLER_REGISTER(shutdown_final, shutdown_panic, NULL,
168 SHUTDOWN_PRI_LAST + 100);
169 EVENTHANDLER_REGISTER(shutdown_final, shutdown_reset, NULL,
170 SHUTDOWN_PRI_LAST + 200);
173 SYSINIT(shutdown_conf, SI_SUB_INTRINSIC, SI_ORDER_ANY, shutdown_conf, NULL);
176 * The system call that results in a reboot.
180 sys_reboot(struct thread *td, struct reboot_args *uap)
186 error = mac_system_check_reboot(td->td_ucred, uap->opt);
189 error = priv_check(td, PRIV_REBOOT);
192 kern_reboot(uap->opt);
199 * Called by events that want to shut down.. e.g <CTL><ALT><DEL> on a PC
202 shutdown_nice(int howto)
205 if (initproc != NULL) {
206 /* Send a signal to init(8) and have it shutdown the world. */
208 if (howto & RB_POWEROFF)
209 kern_psignal(initproc, SIGUSR2);
210 else if (howto & RB_HALT)
211 kern_psignal(initproc, SIGUSR1);
213 kern_psignal(initproc, SIGINT);
214 PROC_UNLOCK(initproc);
216 /* No init(8) running, so simply reboot. */
217 kern_reboot(howto | RB_NOSYNC);
230 if (ts.tv_sec >= 86400) {
231 printf("%ldd", (long)ts.tv_sec / 86400);
235 if (f || ts.tv_sec >= 3600) {
236 printf("%ldh", (long)ts.tv_sec / 3600);
240 if (f || ts.tv_sec >= 60) {
241 printf("%ldm", (long)ts.tv_sec / 60);
245 printf("%lds\n", (long)ts.tv_sec);
249 doadump(boolean_t textdump)
257 if (dumper.dumper == NULL)
261 dumptid = curthread->td_tid;
266 if (textdump && textdump_pending) {
268 textdump_dumpsys(&dumper);
272 error = dumpsys(&dumper);
279 * Shutdown the system cleanly to prepare for reboot, halt, or power off.
282 kern_reboot(int howto)
288 * Bind us to CPU 0 so that all shutdown code runs there. Some
289 * systems don't shutdown properly (i.e., ACPI power off) if we
290 * run on another processor.
292 if (!SCHEDULER_STOPPED()) {
293 thread_lock(curthread);
294 sched_bind(curthread, 0);
295 thread_unlock(curthread);
296 KASSERT(PCPU_GET(cpuid) == 0, ("boot: not running on cpu 0"));
299 /* We're in the process of rebooting. */
302 /* We are out of the debugger now. */
306 * Do any callouts that should be done BEFORE syncing the filesystems.
308 EVENTHANDLER_INVOKE(shutdown_pre_sync, howto);
311 * Now sync filesystems
313 if (!cold && (howto & RB_NOSYNC) == 0 && once == 0) {
315 bufshutdown(show_busybufs);
323 * Ok, now do things that assume all filesystem activity has
326 EVENTHANDLER_INVOKE(shutdown_post_sync, howto);
328 if ((howto & (RB_HALT|RB_DUMP)) == RB_DUMP && !cold && !dumping)
331 /* Now that we're going to really halt the system... */
332 EVENTHANDLER_INVOKE(shutdown_final, howto);
334 for(;;) ; /* safety against shutdown_reset not working */
339 * If the shutdown was a clean halt, behave accordingly.
342 shutdown_halt(void *junk, int howto)
345 if (howto & RB_HALT) {
347 printf("The operating system has halted.\n");
348 printf("Please press any key to reboot.\n\n");
350 case -1: /* No console, just die */
361 * Check to see if the system paniced, pause and then reboot
362 * according to the specified delay.
365 shutdown_panic(void *junk, int howto)
369 if (howto & RB_DUMP) {
370 if (panic_reboot_wait_time != 0) {
371 if (panic_reboot_wait_time != -1) {
372 printf("Automatic reboot in %d seconds - "
373 "press a key on the console to abort\n",
374 panic_reboot_wait_time);
375 for (loop = panic_reboot_wait_time * 10;
377 DELAY(1000 * 100); /* 1/10th second */
378 /* Did user type a key? */
379 if (cncheckc() != -1)
385 } else { /* zero time specified - reboot NOW */
388 printf("--> Press a key on the console to reboot,\n");
389 printf("--> or switch off the system now.\n");
395 * Everything done, now reset
398 shutdown_reset(void *junk, int howto)
401 printf("Rebooting...\n");
402 DELAY(1000000); /* wait 1 sec for printf's to complete and be read */
405 * Acquiring smp_ipi_mtx here has a double effect:
406 * - it disables interrupts avoiding CPU0 preemption
407 * by fast handlers (thus deadlocking against other CPUs)
408 * - it avoids deadlocks against smp_rendezvous() or, more
409 * generally, threads busy-waiting, with this spinlock held,
410 * and waiting for responses by threads on other CPUs
411 * (ie. smp_tlb_shootdown()).
413 * For the !SMP case it just needs to handle the former problem.
416 mtx_lock_spin(&smp_ipi_mtx);
421 /* cpu_boot(howto); */ /* doesn't do anything at the moment */
423 /* NOTREACHED */ /* assuming reset worked */
426 #if defined(WITNESS) || defined(INVARIANTS)
427 static int kassert_warn_only = 0;
429 static int kassert_do_kdb = 0;
432 static int kassert_do_ktr = 0;
434 static int kassert_do_log = 1;
435 static int kassert_log_pps_limit = 4;
436 static int kassert_log_mute_at = 0;
437 static int kassert_log_panic_at = 0;
438 static int kassert_warnings = 0;
440 SYSCTL_NODE(_debug, OID_AUTO, kassert, CTLFLAG_RW, NULL, "kassert options");
442 SYSCTL_INT(_debug_kassert, OID_AUTO, warn_only, CTLFLAG_RWTUN,
443 &kassert_warn_only, 0,
444 "KASSERT triggers a panic (1) or just a warning (0)");
447 SYSCTL_INT(_debug_kassert, OID_AUTO, do_kdb, CTLFLAG_RWTUN,
448 &kassert_do_kdb, 0, "KASSERT will enter the debugger");
452 SYSCTL_UINT(_debug_kassert, OID_AUTO, do_ktr, CTLFLAG_RWTUN,
454 "KASSERT does a KTR, set this to the KTRMASK you want");
457 SYSCTL_INT(_debug_kassert, OID_AUTO, do_log, CTLFLAG_RWTUN,
458 &kassert_do_log, 0, "KASSERT triggers a panic (1) or just a warning (0)");
460 SYSCTL_INT(_debug_kassert, OID_AUTO, warnings, CTLFLAG_RWTUN,
461 &kassert_warnings, 0, "number of KASSERTs that have been triggered");
463 SYSCTL_INT(_debug_kassert, OID_AUTO, log_panic_at, CTLFLAG_RWTUN,
464 &kassert_log_panic_at, 0, "max number of KASSERTS before we will panic");
466 SYSCTL_INT(_debug_kassert, OID_AUTO, log_pps_limit, CTLFLAG_RWTUN,
467 &kassert_log_pps_limit, 0, "limit number of log messages per second");
469 SYSCTL_INT(_debug_kassert, OID_AUTO, log_mute_at, CTLFLAG_RWTUN,
470 &kassert_log_mute_at, 0, "max number of KASSERTS to log");
472 static int kassert_sysctl_kassert(SYSCTL_HANDLER_ARGS);
474 SYSCTL_PROC(_debug_kassert, OID_AUTO, kassert,
475 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE, NULL, 0,
476 kassert_sysctl_kassert, "I", "set to trigger a test kassert");
479 kassert_sysctl_kassert(SYSCTL_HANDLER_ARGS)
483 error = sysctl_wire_old_buffer(req, sizeof(int));
486 error = sysctl_handle_int(oidp, &i, 0, req);
488 if (error != 0 || req->newptr == NULL)
490 KASSERT(0, ("kassert_sysctl_kassert triggered kassert %d", i));
495 * Called by KASSERT, this decides if we will panic
496 * or if we will log via printf and/or ktr.
499 kassert_panic(const char *fmt, ...)
501 static char buf[256];
505 (void)vsnprintf(buf, sizeof(buf), fmt, ap);
509 * panic if we're not just warning, or if we've exceeded
510 * kassert_log_panic_at warnings.
512 if (!kassert_warn_only ||
513 (kassert_log_panic_at > 0 &&
514 kassert_warnings >= kassert_log_panic_at)) {
524 * log if we've not yet met the mute limit.
526 if (kassert_do_log &&
527 (kassert_log_mute_at == 0 ||
528 kassert_warnings < kassert_log_mute_at)) {
529 static struct timeval lasterr;
532 if (ppsratecheck(&lasterr, &curerr, kassert_log_pps_limit)) {
533 printf("KASSERT failed: %s\n", buf);
538 if (kassert_do_kdb) {
539 kdb_enter(KDB_WHY_KASSERT, buf);
542 atomic_add_int(&kassert_warnings, 1);
547 * Panic is called on unresolvable fatal errors. It prints "panic: mesg",
548 * and then reboots. If we are called twice, then we avoid trying to sync
549 * the disks as this often leads to recursive panics.
552 panic(const char *fmt, ...)
561 vpanic(const char *fmt, va_list ap)
566 struct thread *td = curthread;
567 int bootopt, newpanic;
568 static char buf[256];
574 * stop_cpus_hard(other_cpus) should prevent multiple CPUs from
575 * concurrently entering panic. Only the winner will proceed
578 if (panicstr == NULL && !kdb_active) {
579 other_cpus = all_cpus;
580 CPU_CLR(PCPU_GET(cpuid), &other_cpus);
581 stop_cpus_hard(other_cpus);
585 * Ensure that the scheduler is stopped while panicking, even if panic
586 * has been entered from kdb.
588 td->td_stopsched = 1;
591 bootopt = RB_AUTOBOOT;
594 bootopt |= RB_NOSYNC;
602 (void)vsnprintf(buf, sizeof(buf), fmt, ap);
605 printf("panic: %s\n", buf);
612 printf("cpuid = %d\n", PCPU_GET(cpuid));
616 if (newpanic && trace_on_panic)
618 if (debugger_on_panic)
619 kdb_enter(KDB_WHY_PANIC, "panic");
621 /*thread_lock(td); */
622 td->td_flags |= TDF_INPANIC;
623 /* thread_unlock(td); */
625 bootopt |= RB_NOSYNC;
626 kern_reboot(bootopt);
630 * Support for poweroff delay.
632 * Please note that setting this delay too short might power off your machine
633 * before the write cache on your hard disk has been flushed, leading to
634 * soft-updates inconsistencies.
636 #ifndef POWEROFF_DELAY
637 # define POWEROFF_DELAY 5000
639 static int poweroff_delay = POWEROFF_DELAY;
641 SYSCTL_INT(_kern_shutdown, OID_AUTO, poweroff_delay, CTLFLAG_RW,
642 &poweroff_delay, 0, "Delay before poweroff to write disk caches (msec)");
645 poweroff_wait(void *junk, int howto)
648 if (!(howto & RB_POWEROFF) || poweroff_delay <= 0)
650 DELAY(poweroff_delay * 1000);
654 * Some system processes (e.g. syncer) need to be stopped at appropriate
655 * points in their main loops prior to a system shutdown, so that they
656 * won't interfere with the shutdown process (e.g. by holding a disk buf
657 * to cause sync to fail). For each of these system processes, register
658 * shutdown_kproc() as a handler for one of shutdown events.
660 static int kproc_shutdown_wait = 60;
661 SYSCTL_INT(_kern_shutdown, OID_AUTO, kproc_shutdown_wait, CTLFLAG_RW,
662 &kproc_shutdown_wait, 0, "Max wait time (sec) to stop for each process");
665 kproc_shutdown(void *arg, int howto)
673 p = (struct proc *)arg;
674 printf("Waiting (max %d seconds) for system process `%s' to stop...",
675 kproc_shutdown_wait, p->p_comm);
676 error = kproc_suspend(p, kproc_shutdown_wait * hz);
678 if (error == EWOULDBLOCK)
679 printf("timed out\n");
685 kthread_shutdown(void *arg, int howto)
693 td = (struct thread *)arg;
694 printf("Waiting (max %d seconds) for system thread `%s' to stop...",
695 kproc_shutdown_wait, td->td_name);
696 error = kthread_suspend(td, kproc_shutdown_wait * hz);
698 if (error == EWOULDBLOCK)
699 printf("timed out\n");
704 static char dumpdevname[sizeof(((struct cdev*)NULL)->si_name)];
705 SYSCTL_STRING(_kern_shutdown, OID_AUTO, dumpdevname, CTLFLAG_RD,
706 dumpdevname, 0, "Device for kernel dumps");
708 /* Registration of dumpers */
710 set_dumper(struct dumperinfo *di, const char *devname, struct thread *td)
715 error = priv_check(td, PRIV_SETDUMPER);
720 bzero(&dumper, sizeof dumper);
721 dumpdevname[0] = '\0';
724 if (dumper.dumper != NULL)
727 wantcopy = strlcpy(dumpdevname, devname, sizeof(dumpdevname));
728 if (wantcopy >= sizeof(dumpdevname)) {
729 printf("set_dumper: device name truncated from '%s' -> '%s'\n",
730 devname, dumpdevname);
735 /* Call dumper with bounds checking. */
737 dump_write(struct dumperinfo *di, void *virtual, vm_offset_t physical,
738 off_t offset, size_t length)
741 if (length != 0 && (offset < di->mediaoffset ||
742 offset - di->mediaoffset + length > di->mediasize)) {
743 printf("Attempt to write outside dump device boundaries.\n"
744 "offset(%jd), mediaoffset(%jd), length(%ju), mediasize(%jd).\n",
745 (intmax_t)offset, (intmax_t)di->mediaoffset,
746 (uintmax_t)length, (intmax_t)di->mediasize);
749 return (di->dumper(di->priv, virtual, physical, offset, length));
753 mkdumpheader(struct kerneldumpheader *kdh, char *magic, uint32_t archver,
754 uint64_t dumplen, uint32_t blksz)
757 bzero(kdh, sizeof(*kdh));
758 strlcpy(kdh->magic, magic, sizeof(kdh->magic));
759 strlcpy(kdh->architecture, MACHINE_ARCH, sizeof(kdh->architecture));
760 kdh->version = htod32(KERNELDUMPVERSION);
761 kdh->architectureversion = htod32(archver);
762 kdh->dumplength = htod64(dumplen);
763 kdh->dumptime = htod64(time_second);
764 kdh->blocksize = htod32(blksz);
765 strlcpy(kdh->hostname, prison0.pr_hostname, sizeof(kdh->hostname));
766 strlcpy(kdh->versionstring, version, sizeof(kdh->versionstring));
767 if (panicstr != NULL)
768 strlcpy(kdh->panicstring, panicstr, sizeof(kdh->panicstring));
769 kdh->parity = kerneldump_parity(kdh);