kern_shutdown.c revision 41514
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 1.42 1998/11/13 22:40:37 msmith 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
83static int debugger_on_panic = 0;
84#else
85static int debugger_on_panic = 1;
86#endif
87SYSCTL_INT(_debug, OID_AUTO, debugger_on_panic, CTLFLAG_RW,
88	&debugger_on_panic, 0, "");
89#endif
90
91#ifdef	HW_WDOG
92/*
93 * If there is a hardware watchdog, point this at the function needed to
94 * hold it off.
95 * It's needed when the kernel needs to do some lengthy operations.
96 * e.g. in wd.c when dumping core.. It's most annoying to have
97 * your precious core-dump only half written because the wdog kicked in.
98 */
99watchdog_tickle_fn wdog_tickler = NULL;
100#endif	/* HW_WDOG */
101
102/*
103 * Variable panicstr contains argument to first call to panic; used as flag
104 * to indicate that the kernel has already called panic.
105 */
106const char *panicstr;
107
108/*
109 * callout list for things to do a shutdown
110 */
111typedef struct shutdown_list_element {
112	LIST_ENTRY(shutdown_list_element) links;
113	bootlist_fn function;
114	void *arg;
115	int priority;
116} *sle_p;
117
118/*
119 * There are three shutdown lists. Some things need to be shut down
120 * earlier than others.
121 */
122LIST_HEAD(shutdown_list, shutdown_list_element);
123
124static struct shutdown_list shutdown_lists[SHUTDOWN_FINAL + 1];
125
126static void boot __P((int)) __dead2;
127static void dumpsys __P((void));
128
129#ifndef _SYS_SYSPROTO_H_
130struct reboot_args {
131	int	opt;
132};
133#endif
134/* ARGSUSED */
135
136/*
137 * The system call that results in a reboot
138 */
139int
140reboot(p, uap)
141	struct proc *p;
142	struct reboot_args *uap;
143{
144	int error;
145
146	if ((error = suser(p->p_ucred, &p->p_acflag)))
147		return (error);
148
149	boot(uap->opt);
150	return (0);
151}
152
153/*
154 * Called by events that want to shut down.. e.g  <CTL><ALT><DEL> on a PC
155 */
156void
157shutdown_nice()
158{
159	/* Send a signal to init(8) and have it shutdown the world */
160	if (initproc != NULL) {
161		psignal(initproc, SIGINT);
162	} else {
163		/* No init(8) running, so simply reboot */
164		boot(RB_NOSYNC);
165	}
166	return;
167}
168static int	waittime = -1;
169static struct pcb dumppcb;
170
171/*
172 *  Go through the rigmarole of shutting down..
173 * this used to be in machdep.c but I'll be dammned if I could see
174 * anything machine dependant in it.
175 */
176static void
177boot(howto)
178	int howto;
179{
180	sle_p ep;
181
182#ifdef SMP
183	if (smp_active) {
184		printf("boot() called on cpu#%d\n", cpuid);
185	}
186#endif
187	/*
188	 * Do any callouts that should be done BEFORE syncing the filesystems.
189	 */
190	LIST_FOREACH(ep, &shutdown_lists[SHUTDOWN_PRE_SYNC], links)
191		(*ep->function)(howto, ep->arg);
192
193	/*
194	 * Now sync filesystems
195	 */
196	if (!cold && (howto & RB_NOSYNC) == 0 && waittime < 0) {
197		register struct buf *bp;
198		int iter, nbusy;
199
200		waittime = 0;
201		printf("\nsyncing disks... ");
202
203		sync(&proc0, NULL);
204
205		/*
206		 * With soft updates, some buffers that are
207		 * written will be remarked as dirty until other
208		 * buffers are written.
209		 */
210		for (iter = 0; iter < 20; iter++) {
211			nbusy = 0;
212			for (bp = &buf[nbuf]; --bp >= buf; ) {
213				if ((bp->b_flags & (B_BUSY | B_INVAL))
214						== B_BUSY) {
215					nbusy++;
216				} else if ((bp->b_flags & (B_DELWRI | B_INVAL))
217						== B_DELWRI) {
218					/* bawrite(bp);*/
219					nbusy++;
220				}
221			}
222			if (nbusy == 0)
223				break;
224			printf("%d ", nbusy);
225			sync(&proc0, NULL);
226			DELAY(50000 * iter);
227		}
228		/*
229		 * Count only busy local buffers to prevent forcing
230		 * a fsck if we're just a client of a wedged NFS server
231		 */
232		nbusy = 0;
233		for (bp = &buf[nbuf]; --bp >= buf; ) {
234			if (((bp->b_flags & (B_BUSY | B_INVAL)) == B_BUSY)
235			    ||((bp->b_flags & (B_DELWRI | B_INVAL))== B_DELWRI))
236				if(bp->b_dev == NODEV)
237					CIRCLEQ_REMOVE(&mountlist, bp->b_vp->v_mount, mnt_list);
238				else
239					nbusy++;
240
241
242		}
243		if (nbusy) {
244			/*
245			 * Failed to sync all blocks. Indicate this and don't
246			 * unmount filesystems (thus forcing an fsck on reboot).
247			 */
248			printf("giving up\n");
249#ifdef SHOW_BUSYBUFS
250			nbusy = 0;
251			for (bp = &buf[nbuf]; --bp >= buf; ) {
252				if ((bp->b_flags & (B_BUSY | B_INVAL))
253						== B_BUSY) {
254					nbusy++;
255					printf(
256			"%d: dev:%08lx, flags:%08lx, blkno:%ld, lblkno:%ld\n",
257					    nbusy, (u_long)bp->b_dev,
258					    bp->b_flags, (long)bp->b_blkno,
259					    (long)bp->b_lblkno);
260				}
261			}
262			DELAY(5000000);	/* 5 seconds */
263#endif
264		} else {
265			printf("done\n");
266			/*
267			 * Unmount filesystems
268			 */
269			if (panicstr == 0)
270				vfs_unmountall();
271		}
272		DELAY(100000);		/* wait for console output to finish */
273	}
274
275	/*
276	 * Ok, now do things that assume all filesystem activity has
277	 * been completed.
278	 */
279	LIST_FOREACH(ep, &shutdown_lists[SHUTDOWN_POST_SYNC], links)
280		(*ep->function)(howto, ep->arg);
281	splhigh();
282	if ((howto & (RB_HALT|RB_DUMP)) == RB_DUMP && !cold) {
283		savectx(&dumppcb);
284#ifdef __i386__
285		dumppcb.pcb_cr3 = rcr3();
286#endif
287		dumpsys();
288	}
289
290	/* Now that we're going to really halt the system... */
291	LIST_FOREACH(ep, &shutdown_lists[SHUTDOWN_FINAL], links)
292		(*ep->function)(howto, ep->arg);
293
294	if (howto & RB_HALT) {
295		printf("\n");
296		printf("The operating system has halted.\n");
297		printf("Please press any key to reboot.\n\n");
298		switch (cngetc()) {
299		case -1:		/* No console, just die */
300			cpu_halt();
301			/* NOTREACHED */
302		default:
303			howto &= ~RB_HALT;
304			break;
305		}
306	} else if (howto & RB_DUMP) {
307		/* System Paniced */
308
309		if (PANIC_REBOOT_WAIT_TIME != 0) {
310			if (PANIC_REBOOT_WAIT_TIME != -1) {
311				int loop;
312				printf("Automatic reboot in %d seconds - "
313				       "press a key on the console to abort\n",
314					PANIC_REBOOT_WAIT_TIME);
315				for (loop = PANIC_REBOOT_WAIT_TIME * 10;
316				     loop > 0; --loop) {
317					DELAY(1000 * 100); /* 1/10th second */
318					/* Did user type a key? */
319					if (cncheckc() != -1)
320						break;
321				}
322				if (!loop)
323					goto die;
324			}
325		} else { /* zero time specified - reboot NOW */
326			goto die;
327		}
328		printf("--> Press a key on the console to reboot <--\n");
329		cngetc();
330	}
331die:
332	printf("Rebooting...\n");
333	DELAY(1000000);	/* wait 1 sec for printf's to complete and be read */
334	/* cpu_boot(howto); */ /* doesn't do anything at the moment */
335	cpu_reset();
336	for(;;) ;
337	/* NOTREACHED */
338}
339
340/*
341 * Magic number for savecore
342 *
343 * exported (symorder) and used at least by savecore(8)
344 *
345 */
346static u_long const	dumpmag = 0x8fca0101UL;
347
348static int	dumpsize = 0;		/* also for savecore */
349
350static int	dodump = 1;
351SYSCTL_INT(_machdep, OID_AUTO, do_dump, CTLFLAG_RW, &dodump, 0, "");
352
353/* ARGSUSED */
354static void dump_conf __P((void *dummy));
355static void
356dump_conf(dummy)
357	void *dummy;
358{
359	cpu_dumpconf();
360}
361SYSINIT(dump_conf, SI_SUB_DUMP_CONF, SI_ORDER_FIRST, dump_conf, NULL)
362
363/*
364 * Doadump comes here after turning off memory management and
365 * getting on the dump stack, either when called above, or by
366 * the auto-restart code.
367 */
368static void
369dumpsys(void)
370{
371
372	if (!dodump)
373		return;
374	if (dumpdev == NODEV)
375		return;
376	if (!(bdevsw[major(dumpdev)]))
377		return;
378	if (!(bdevsw[major(dumpdev)]->d_dump))
379		return;
380	dumpsize = Maxmem;
381	printf("\ndumping to dev %lx, offset %ld\n", (u_long)dumpdev, dumplo);
382	printf("dump ");
383	switch ((*bdevsw[major(dumpdev)]->d_dump)(dumpdev)) {
384
385	case ENXIO:
386		printf("device bad\n");
387		break;
388
389	case EFAULT:
390		printf("device not ready\n");
391		break;
392
393	case EINVAL:
394		printf("area improper\n");
395		break;
396
397	case EIO:
398		printf("i/o error\n");
399		break;
400
401	case EINTR:
402		printf("aborted from console\n");
403		break;
404
405	default:
406		printf("succeeded\n");
407		break;
408	}
409}
410
411/*
412 * Panic is called on unresolvable fatal errors.  It prints "panic: mesg",
413 * and then reboots.  If we are called twice, then we avoid trying to sync
414 * the disks as this often leads to recursive panics.
415 */
416void
417panic(const char *fmt, ...)
418{
419	int bootopt;
420	va_list ap;
421	static char buf[256];
422
423	bootopt = RB_AUTOBOOT | RB_DUMP;
424	if (panicstr)
425		bootopt |= RB_NOSYNC;
426	else
427		panicstr = fmt;
428
429	va_start(ap, fmt);
430	(void)vsnprintf(buf, sizeof(buf), fmt, ap);
431	if (panicstr == fmt)
432		panicstr = buf;
433	va_end(ap);
434	printf("panic: %s\n", buf);
435#ifdef SMP
436	/* three seperate prints in case of an unmapped page and trap */
437	printf("mp_lock = %08x; ", mp_lock);
438	printf("cpuid = %d; ", cpuid);
439	printf("lapic.id = %08x\n", lapic.id);
440#endif
441
442#if defined(DDB)
443	if (debugger_on_panic)
444		Debugger ("panic");
445#endif
446	boot(bootopt);
447}
448
449/*
450 * Three routines to handle adding/deleting items on the
451 * shutdown callout lists
452 *
453 * at_shutdown():
454 * Take the arguments given and put them onto the shutdown callout list.
455 * However first make sure that it's not already there.
456 * returns 0 on success.
457 */
458int
459at_shutdown(bootlist_fn function, void *arg, int queue)
460{
461	return(at_shutdown_pri(function, arg, queue, SHUTDOWN_PRI_DEFAULT));
462}
463
464/*
465 * at_shutdown_pri():
466 * Take the arguments given and put them onto the shutdown callout list
467 * with the given execution priority.
468 * returns 0 on success.
469 */
470int
471at_shutdown_pri(bootlist_fn function, void *arg, int queue, int pri)
472{
473	sle_p ep, ip;
474
475	if (queue < SHUTDOWN_PRE_SYNC
476	 || queue > SHUTDOWN_FINAL) {
477		printf("at_shutdown: bad exit callout queue %d specified\n",
478		       queue);
479		return (EINVAL);
480	}
481	if (rm_at_shutdown(function, arg))
482		printf("at_shutdown: exit callout entry was already present\n");
483	ep = malloc(sizeof(*ep), M_TEMP, M_NOWAIT);
484	if (ep == NULL)
485		return (ENOMEM);
486	ep->function = function;
487	ep->arg = arg;
488	ep->priority = pri;
489
490	/* Sort into list of items on this queue */
491	ip = LIST_FIRST(&shutdown_lists[queue]);
492	if (ip == NULL) {
493		LIST_INSERT_HEAD(&shutdown_lists[queue], ep, links);
494	} else {
495		for (; LIST_NEXT(ip, links) != NULL; ip = LIST_NEXT(ip, links)) {
496			if (ep->priority < ip->priority) {
497				LIST_INSERT_BEFORE(ip, ep, links);
498				ep = NULL;
499				break;
500			}
501		}
502		if (ep != NULL)
503			LIST_INSERT_AFTER(ip, ep, links);
504	}
505	return (0);
506}
507
508/*
509 * Scan the exit callout lists for the given items and remove them.
510 * Returns the number of items removed.
511 */
512int
513rm_at_shutdown(bootlist_fn function, void *arg)
514{
515	sle_p ep;
516	int   count;
517	int   queue;
518
519	count = 0;
520	for (queue = SHUTDOWN_PRE_SYNC; queue < SHUTDOWN_FINAL; queue++) {
521		LIST_FOREACH(ep, &shutdown_lists[queue], links) {
522			if ((ep->function == function) && (ep->arg == arg)) {
523				LIST_REMOVE(ep, links);
524				free(ep, M_TEMP);
525				count++;
526			}
527		}
528	}
529	return (count);
530}
531