1/*-
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.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
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 * 3. All advertising materials mentioning features or use of this software
19 * must display the following acknowledgement:
20 * This product includes software developed by the University of
21 * California, Berkeley and its contributors.
22 * 4. Neither the name of the University nor the names of its contributors
23 * may be used to endorse or promote products derived from this software
24 * without specific prior written permission.
25 *
26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * SUCH DAMAGE.
37 *
38 * @(#)kern_shutdown.c 8.3 (Berkeley) 1/21/94
39 * $Id: kern_shutdown.c,v 1.48 1999/05/03 23:57:22 billf Exp $
40 */
41
42#include "opt_ddb.h"
43#include "opt_hw_wdog.h"
44#include "opt_panic.h"
45#include "opt_show_busybufs.h"
46
47#include <sys/param.h>
48#include <sys/systm.h>
49#include <sys/buf.h>
50#include <sys/reboot.h>
51#include <sys/proc.h>
52#include <sys/vnode.h>
53#include <sys/malloc.h>
54#include <sys/kernel.h>
55#include <sys/mount.h>
56#include <sys/queue.h>
57#include <sys/sysctl.h>
58#include <sys/conf.h>
59#include <sys/sysproto.h>
60
61#include <machine/pcb.h>
62#include <machine/clock.h>
63#include <machine/cons.h>
64#include <machine/md_var.h>
65#ifdef SMP
66#include <machine/smp.h> /* smp_active, cpuid */
67#endif
68
69#include <sys/signalvar.h>
70
71#ifndef PANIC_REBOOT_WAIT_TIME
72#define PANIC_REBOOT_WAIT_TIME 15 /* default to 15 seconds */
73#endif
74
75/*
76 * Note that stdarg.h and the ANSI style va_start macro is used for both
77 * ANSI and traditional C compilers.
78 */
79#include <machine/stdarg.h>
80
81#ifdef DDB
82#ifdef DDB_UNATTENDED
83int debugger_on_panic = 0;
84#else
85int debugger_on_panic = 1;
86#endif
87SYSCTL_INT(_debug, OID_AUTO, debugger_on_panic, CTLFLAG_RW,
88 &debugger_on_panic, 0, "Run debugger on kernel panic");
89#endif
90
91SYSCTL_NODE(_kern, OID_AUTO, shutdown, CTLFLAG_RW, 0, "Shutdown environment");
92
93#ifdef HW_WDOG
94/*
95 * If there is a hardware watchdog, point this at the function needed to
96 * hold it off.
97 * It's needed when the kernel needs to do some lengthy operations.
98 * e.g. in wd.c when dumping core.. It's most annoying to have
99 * your precious core-dump only half written because the wdog kicked in.
100 */
101watchdog_tickle_fn wdog_tickler = NULL;
102#endif /* HW_WDOG */
103
104/*
105 * Variable panicstr contains argument to first call to panic; used as flag
106 * to indicate that the kernel has already called panic.
107 */
108const char *panicstr;
109
110/*
111 * callout list for things to do a shutdown
112 */
113typedef struct shutdown_list_element {
114 LIST_ENTRY(shutdown_list_element) links;
115 bootlist_fn function;
116 void *arg;
117 int priority;
118} *sle_p;
119
120/*
121 * There are three shutdown lists. Some things need to be shut down
122 * earlier than others.
123 */
124LIST_HEAD(shutdown_list, shutdown_list_element);
125
126static struct shutdown_list shutdown_lists[SHUTDOWN_FINAL + 1];
127
128static void boot __P((int)) __dead2;
129static void dumpsys __P((void));
130
131#ifndef _SYS_SYSPROTO_H_
132struct reboot_args {
133 int opt;
134};
135#endif
136/* ARGSUSED */
137
138/*
139 * The system call that results in a reboot
140 */
141int
142reboot(p, uap)
143 struct proc *p;
144 struct reboot_args *uap;
145{
146 int error;
147
148 if ((error = suser(p)))
149 return (error);
150
151 boot(uap->opt);
152 return (0);
153}
154
155/*
156 * Called by events that want to shut down.. e.g <CTL><ALT><DEL> on a PC
157 */
158void
159shutdown_nice()
160{
161 /* Send a signal to init(8) and have it shutdown the world */
162 if (initproc != NULL) {
163 psignal(initproc, SIGINT);
164 } else {
165 /* No init(8) running, so simply reboot */
166 boot(RB_NOSYNC);
167 }
168 return;
169}
170static int waittime = -1;
171static struct pcb dumppcb;
172
173/*
174 * Go through the rigmarole of shutting down..
175 * this used to be in machdep.c but I'll be dammned if I could see
176 * anything machine dependant in it.
177 */
178static void
179boot(howto)
180 int howto;
181{
182 sle_p ep;
183
184#ifdef SMP
185 if (smp_active) {
186 printf("boot() called on cpu#%d\n", cpuid);
187 }
188#endif
189 /*
190 * Do any callouts that should be done BEFORE syncing the filesystems.
191 */
192 LIST_FOREACH(ep, &shutdown_lists[SHUTDOWN_PRE_SYNC], links)
193 (*ep->function)(howto, ep->arg);
194
195 /*
196 * Now sync filesystems
197 */
198 if (!cold && (howto & RB_NOSYNC) == 0 && waittime < 0) {
199 register struct buf *bp;
200 int iter, nbusy;
201
202 waittime = 0;
203 printf("\nsyncing disks... ");
204
205 sync(&proc0, NULL);
206
207 /*
208 * With soft updates, some buffers that are
209 * written will be remarked as dirty until other
210 * buffers are written.
211 */
212 for (iter = 0; iter < 20; iter++) {
213 nbusy = 0;
214 for (bp = &buf[nbuf]; --bp >= buf; ) {
215 if ((bp->b_flags & (B_BUSY | B_INVAL))
216 == B_BUSY) {
217 nbusy++;
218 } else if ((bp->b_flags & (B_DELWRI | B_INVAL))
219 == B_DELWRI) {
220 /* bawrite(bp);*/
221 nbusy++;
222 }
223 }
224 if (nbusy == 0)
225 break;
226 printf("%d ", nbusy);
227 sync(&proc0, NULL);
228 DELAY(50000 * iter);
229 }
230 /*
231 * Count only busy local buffers to prevent forcing
232 * a fsck if we're just a client of a wedged NFS server
233 */
234 nbusy = 0;
235 for (bp = &buf[nbuf]; --bp >= buf; ) {
236 if (((bp->b_flags & (B_BUSY | B_INVAL)) == B_BUSY) ||
237 ((bp->b_flags & (B_DELWRI | B_INVAL))== B_DELWRI)) {
238 if(bp->b_dev == NODEV)
239 CIRCLEQ_REMOVE(&mountlist, bp->b_vp->v_mount, mnt_list);
240 else
241 nbusy++;
242 }
243
244
245 }
246 if (nbusy) {
247 /*
248 * Failed to sync all blocks. Indicate this and don't
249 * unmount filesystems (thus forcing an fsck on reboot).
250 */
251 printf("giving up\n");
252#ifdef SHOW_BUSYBUFS
253 nbusy = 0;
254 for (bp = &buf[nbuf]; --bp >= buf; ) {
255 if ((bp->b_flags & (B_BUSY | B_INVAL))
256 == B_BUSY) {
257 nbusy++;
258 printf(
259 "%d: dev:%08lx, flags:%08lx, blkno:%ld, lblkno:%ld\n",
260 nbusy, (u_long)bp->b_dev,
261 bp->b_flags, (long)bp->b_blkno,
262 (long)bp->b_lblkno);
263 }
264 }
265 DELAY(5000000); /* 5 seconds */
266#endif
267 } else {
268 printf("done\n");
269 /*
270 * Unmount filesystems
271 */
272 if (panicstr == 0)
273 vfs_unmountall();
274 }
275 DELAY(100000); /* wait for console output to finish */
276 }
277
278 /*
279 * Ok, now do things that assume all filesystem activity has
280 * been completed.
281 */
282 LIST_FOREACH(ep, &shutdown_lists[SHUTDOWN_POST_SYNC], links)
283 (*ep->function)(howto, ep->arg);
284 splhigh();
285 if ((howto & (RB_HALT|RB_DUMP)) == RB_DUMP && !cold) {
286 savectx(&dumppcb);
287#ifdef __i386__
288 dumppcb.pcb_cr3 = rcr3();
289#endif
290 dumpsys();
291 }
292
293 /* Now that we're going to really halt the system... */
294 LIST_FOREACH(ep, &shutdown_lists[SHUTDOWN_FINAL], links)
295 (*ep->function)(howto, ep->arg);
296
297 if (howto & RB_HALT) {
298 printf("\n");
299 printf("The operating system has halted.\n");
300 printf("Please press any key to reboot.\n\n");
301 switch (cngetc()) {
302 case -1: /* No console, just die */
303 cpu_halt();
304 /* NOTREACHED */
305 default:
306 howto &= ~RB_HALT;
307 break;
308 }
309 } else if (howto & RB_DUMP) {
310 /* System Paniced */
311
312 if (PANIC_REBOOT_WAIT_TIME != 0) {
313 if (PANIC_REBOOT_WAIT_TIME != -1) {
314 int loop;
315 printf("Automatic reboot in %d seconds - "
316 "press a key on the console to abort\n",
317 PANIC_REBOOT_WAIT_TIME);
318 for (loop = PANIC_REBOOT_WAIT_TIME * 10;
319 loop > 0; --loop) {
320 DELAY(1000 * 100); /* 1/10th second */
321 /* Did user type a key? */
322 if (cncheckc() != -1)
323 break;
324 }
325 if (!loop)
326 goto die;
327 }
328 } else { /* zero time specified - reboot NOW */
329 goto die;
330 }
331 printf("--> Press a key on the console to reboot <--\n");
332 cngetc();
333 }
334die:
335 printf("Rebooting...\n");
336 DELAY(1000000); /* wait 1 sec for printf's to complete and be read */
337 /* cpu_boot(howto); */ /* doesn't do anything at the moment */
338 cpu_reset();
339 for(;;) ;
340 /* NOTREACHED */
341}
342
343/*
344 * Magic number for savecore
345 *
346 * exported (symorder) and used at least by savecore(8)
347 *
348 */
349static u_long const dumpmag = 0x8fca0101UL;
350
351static int dumpsize = 0; /* also for savecore */
352
353static int dodump = 1;
354SYSCTL_INT(_machdep, OID_AUTO, do_dump, CTLFLAG_RW,
355 &dodump, 0, "Do coredump on kernel panic");
356
357/* ARGSUSED */
358static void dump_conf __P((void *dummy));
359static void
360dump_conf(dummy)
361 void *dummy;
362{
363 cpu_dumpconf();
364}
365SYSINIT(dump_conf, SI_SUB_DUMP_CONF, SI_ORDER_FIRST, dump_conf, NULL)
366
367/*
368 * Doadump comes here after turning off memory management and
369 * getting on the dump stack, either when called above, or by
370 * the auto-restart code.
371 */
372static void
373dumpsys(void)
374{
375
376 if (!dodump)
377 return;
378 if (dumpdev == NODEV)
379 return;
380 if (!(bdevsw[major(dumpdev)]))
381 return;
382 if (!(bdevsw[major(dumpdev)]->d_dump))
383 return;
384 dumpsize = Maxmem;
385 printf("\ndumping to dev %lx, offset %ld\n", (u_long)dumpdev, dumplo);
386 printf("dump ");
387 switch ((*bdevsw[major(dumpdev)]->d_dump)(dumpdev)) {
388
389 case ENXIO:
390 printf("device bad\n");
391 break;
392
393 case EFAULT:
394 printf("device not ready\n");
395 break;
396
397 case EINVAL:
398 printf("area improper\n");
399 break;
400
401 case EIO:
402 printf("i/o error\n");
403 break;
404
405 case EINTR:
406 printf("aborted from console\n");
407 break;
408
409 default:
410 printf("succeeded\n");
411 break;
412 }
413}
414
415/*
416 * Panic is called on unresolvable fatal errors. It prints "panic: mesg",
417 * and then reboots. If we are called twice, then we avoid trying to sync
418 * the disks as this often leads to recursive panics.
419 */
420void
421panic(const char *fmt, ...)
422{
423 int bootopt;
424 va_list ap;
425 static char buf[256];
426
427 bootopt = RB_AUTOBOOT | RB_DUMP;
428 if (panicstr)
429 bootopt |= RB_NOSYNC;
430 else
431 panicstr = fmt;
432
433 va_start(ap, fmt);
434 (void)vsnprintf(buf, sizeof(buf), fmt, ap);
435 if (panicstr == fmt)
436 panicstr = buf;
437 va_end(ap);
438 printf("panic: %s\n", buf);
439#ifdef SMP
440 /* three seperate prints in case of an unmapped page and trap */
441 printf("mp_lock = %08x; ", mp_lock);
442 printf("cpuid = %d; ", cpuid);
443 printf("lapic.id = %08x\n", lapic.id);
444#endif
445
446#if defined(DDB)
447 if (debugger_on_panic)
448 Debugger ("panic");
449#endif
450 boot(bootopt);
451}
452
453/*
454 * Three routines to handle adding/deleting items on the
455 * shutdown callout lists
456 *
457 * at_shutdown():
458 * Take the arguments given and put them onto the shutdown callout list.
459 * However first make sure that it's not already there.
460 * returns 0 on success.
461 */
462int
463at_shutdown(bootlist_fn function, void *arg, int queue)
464{
465 return(at_shutdown_pri(function, arg, queue, SHUTDOWN_PRI_DEFAULT));
466}
467
468/*
469 * at_shutdown_pri():
470 * Take the arguments given and put them onto the shutdown callout list
471 * with the given execution priority.
472 * returns 0 on success.
473 */
474int
475at_shutdown_pri(bootlist_fn function, void *arg, int queue, int pri)
476{
477 sle_p op, ep, ip;
478
479 if (queue < SHUTDOWN_PRE_SYNC
480 || queue > SHUTDOWN_FINAL) {
481 printf("at_shutdown: bad exit callout queue %d specified\n",
482 queue);
483 return (EINVAL);
484 }
485 if (rm_at_shutdown(function, arg))
486 printf("at_shutdown: exit callout entry was already present\n");
487 ep = malloc(sizeof(*ep), M_TEMP, M_NOWAIT);
488 if (ep == NULL)
489 return (ENOMEM);
490 ep->function = function;
491 ep->arg = arg;
492 ep->priority = pri;
493
494 /* Sort into list of items on this queue */
495 ip = LIST_FIRST(&shutdown_lists[queue]);
496 if (ip == NULL) {
497 LIST_INSERT_HEAD(&shutdown_lists[queue], ep, links);
498 } else {
499 for (; ip != NULL; op = ip, ip = LIST_NEXT(ip, links)) {
500 if (ep->priority < ip->priority) {
501 LIST_INSERT_BEFORE(ip, ep, links);
502 ep = NULL;
503 break;
504 }
505 }
506 if (ep != NULL)
507 LIST_INSERT_AFTER(op, ep, links);
508 }
509 return (0);
510}
511
512/*
513 * Scan the exit callout lists for the given items and remove them.
514 * Returns the number of items removed.
515 */
516int
517rm_at_shutdown(bootlist_fn function, void *arg)
518{
519 sle_p ep;
520 int count;
521 int queue;
522
523 count = 0;
524 for (queue = SHUTDOWN_PRE_SYNC; queue < SHUTDOWN_FINAL; queue++) {
525 LIST_FOREACH(ep, &shutdown_lists[queue], links) {
526 if ((ep->function == function) && (ep->arg == arg)) {
527 LIST_REMOVE(ep, links);
528 free(ep, M_TEMP);
529 count++;
530 }
531 }
532 }
533 return (count);
534}
535
536/*
537 * Support for poweroff delay.
538 */
539static int poweroff_delay = 0;
540SYSCTL_INT(_kern_shutdown, OID_AUTO, poweroff_delay, CTLFLAG_RW,
541 &poweroff_delay, 0, "");
542
543static void poweroff_wait(int howto, void *unused)
544{
545 if(!(howto & RB_POWEROFF) || poweroff_delay <= 0)
546 return;
547 DELAY(poweroff_delay * 1000);
548}
549
550/*
551 * XXX OK? This implies I know SHUTDOWN_PRI_LAST > SHUTDOWN_PRI_FIRST
552 */
553static void poweroff_conf(void *unused)
554{
555 at_shutdown_pri(poweroff_wait, NULL, SHUTDOWN_FINAL, SHUTDOWN_PRI_FIRST);
556}
557
558SYSINIT(poweroff_conf, SI_SUB_INTRINSIC, SI_ORDER_ANY, poweroff_conf, NULL)