AMD64 - Fix many compile-time warnings. int/ptr type mismatches, %llx, etc.
[dragonfly.git] / sys / kern / kern_shutdown.c
CommitLineData
984263bc
MD
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 * $FreeBSD: src/sys/kern/kern_shutdown.c,v 1.72.2.12 2002/02/21 19:15:10 dillon Exp $
ffe71962 40 * $DragonFly: src/sys/kern/kern_shutdown.c,v 1.62 2008/01/05 13:23:48 corecode Exp $
984263bc
MD
41 */
42
43#include "opt_ddb.h"
b1e04573 44#include "opt_ddb_trace.h"
984263bc
MD
45#include "opt_hw_wdog.h"
46#include "opt_panic.h"
47#include "opt_show_busybufs.h"
48
49#include <sys/param.h>
50#include <sys/systm.h>
51#include <sys/eventhandler.h>
52#include <sys/buf.h>
a6c0f342 53#include <sys/diskslice.h>
984263bc
MD
54#include <sys/reboot.h>
55#include <sys/proc.h>
895c1f85 56#include <sys/priv.h>
faa575e7
MD
57#include <sys/fcntl.h> /* FREAD */
58#include <sys/stat.h> /* S_IFCHR */
984263bc
MD
59#include <sys/vnode.h>
60#include <sys/kernel.h>
61#include <sys/kthread.h>
62#include <sys/malloc.h>
63#include <sys/mount.h>
64#include <sys/queue.h>
65#include <sys/sysctl.h>
f6a22062 66#include <sys/vkernel.h>
984263bc
MD
67#include <sys/conf.h>
68#include <sys/sysproto.h>
335dda38 69#include <sys/device.h>
984263bc 70#include <sys/cons.h>
f6a22062
MD
71#include <sys/shm.h>
72#include <sys/kern_syscall.h>
73#include <vm/vm_map.h>
74#include <vm/pmap.h>
75
9ec81f83 76#include <sys/thread2.h>
3020e3be 77#include <sys/buf2.h>
984263bc
MD
78
79#include <machine/pcb.h>
80#include <machine/clock.h>
81#include <machine/md_var.h>
0f7a3396 82#include <machine/smp.h> /* smp_active_mask, cpuid */
c439ad8f 83#include <machine/vmparam.h>
984263bc
MD
84
85#include <sys/signalvar.h>
86
87#ifndef PANIC_REBOOT_WAIT_TIME
88#define PANIC_REBOOT_WAIT_TIME 15 /* default to 15 seconds */
89#endif
90
91/*
92 * Note that stdarg.h and the ANSI style va_start macro is used for both
e2565a42
MD
93 * ANSI and traditional C compilers. We use the machine version to stay
94 * within the confines of the kernel header files.
984263bc
MD
95 */
96#include <machine/stdarg.h>
97
98#ifdef DDB
1e5fb84b 99#include <ddb/ddb.h>
984263bc
MD
100#ifdef DDB_UNATTENDED
101int debugger_on_panic = 0;
102#else
103int debugger_on_panic = 1;
104#endif
105SYSCTL_INT(_debug, OID_AUTO, debugger_on_panic, CTLFLAG_RW,
106 &debugger_on_panic, 0, "Run debugger on kernel panic");
b1e04573 107
b1e04573
JH
108#ifdef DDB_TRACE
109int trace_on_panic = 1;
110#else
111int trace_on_panic = 0;
112#endif
113SYSCTL_INT(_debug, OID_AUTO, trace_on_panic, CTLFLAG_RW,
114 &trace_on_panic, 0, "Print stack trace on kernel panic");
984263bc
MD
115#endif
116
baf88b6b
MD
117static int sync_on_panic = 1;
118SYSCTL_INT(_kern, OID_AUTO, sync_on_panic, CTLFLAG_RW,
119 &sync_on_panic, 0, "Do a sync before rebooting from a panic");
120
984263bc
MD
121SYSCTL_NODE(_kern, OID_AUTO, shutdown, CTLFLAG_RW, 0, "Shutdown environment");
122
123#ifdef HW_WDOG
124/*
125 * If there is a hardware watchdog, point this at the function needed to
126 * hold it off.
127 * It's needed when the kernel needs to do some lengthy operations.
128 * e.g. in wd.c when dumping core.. It's most annoying to have
129 * your precious core-dump only half written because the wdog kicked in.
130 */
131watchdog_tickle_fn wdog_tickler = NULL;
132#endif /* HW_WDOG */
133
134/*
135 * Variable panicstr contains argument to first call to panic; used as flag
136 * to indicate that the kernel has already called panic.
137 */
138const char *panicstr;
139
140int dumping; /* system is dumping */
ddcc5e10
MD
141#ifdef SMP
142u_int panic_cpu_interlock; /* panic interlock */
143globaldata_t panic_cpu_gd; /* which cpu took the panic */
144#endif
984263bc 145
a8a86718 146int bootverbose = 0; /* note: assignment to force non-bss */
ffe71962
SS
147SYSCTL_INT(_debug, OID_AUTO, bootverbose, CTLFLAG_RW,
148 &bootverbose, 0, "Verbose kernel messages");
149
a8a86718 150int cold = 1; /* note: assignment to force non-bss */
e0fc5693
MD
151int dumplo; /* OBSOLETE - savecore compat */
152u_int64_t dumplo64;
a8a86718 153
402ed7e1
RG
154static void boot (int) __dead2;
155static void dumpsys (void);
b13267a5 156static int setdumpdev (cdev_t dev);
402ed7e1
RG
157static void poweroff_wait (void *, int);
158static void print_uptime (void);
159static void shutdown_halt (void *junk, int howto);
160static void shutdown_panic (void *junk, int howto);
161static void shutdown_reset (void *junk, int howto);
41cbf46c
MD
162static int shutdown_busycount1(struct buf *bp, void *info);
163static int shutdown_busycount2(struct buf *bp, void *info);
f6a22062 164static void shutdown_cleanup_proc(struct proc *p);
984263bc
MD
165
166/* register various local shutdown events */
167static void
168shutdown_conf(void *unused)
169{
170 EVENTHANDLER_REGISTER(shutdown_final, poweroff_wait, NULL, SHUTDOWN_PRI_FIRST);
171 EVENTHANDLER_REGISTER(shutdown_final, shutdown_halt, NULL, SHUTDOWN_PRI_LAST + 100);
172 EVENTHANDLER_REGISTER(shutdown_final, shutdown_panic, NULL, SHUTDOWN_PRI_LAST + 100);
173 EVENTHANDLER_REGISTER(shutdown_final, shutdown_reset, NULL, SHUTDOWN_PRI_LAST + 200);
174}
175
ba39e2e0 176SYSINIT(shutdown_conf, SI_BOOT2_MACHDEP, SI_ORDER_ANY, shutdown_conf, NULL)
984263bc
MD
177
178/* ARGSUSED */
179
180/*
181 * The system call that results in a reboot
182 */
183int
753fd850 184sys_reboot(struct reboot_args *uap)
984263bc 185{
dadab5e9 186 struct thread *td = curthread;
984263bc
MD
187 int error;
188
3b1d99e9 189 if ((error = priv_check(td, PRIV_REBOOT)))
984263bc
MD
190 return (error);
191
192 boot(uap->opt);
193 return (0);
194}
195
196/*
197 * Called by events that want to shut down.. e.g <CTL><ALT><DEL> on a PC
198 */
199static int shutdown_howto = 0;
200
201void
202shutdown_nice(int howto)
203{
204 shutdown_howto = howto;
205
206 /* Send a signal to init(8) and have it shutdown the world */
207 if (initproc != NULL) {
84204577 208 ksignal(initproc, SIGINT);
984263bc
MD
209 } else {
210 /* No init(8) running, so simply reboot */
211 boot(RB_NOSYNC);
212 }
213 return;
214}
215static int waittime = -1;
8b6a428f 216static struct thread *dumpthread;
984263bc
MD
217static struct pcb dumppcb;
218
219static void
c972a82f 220print_uptime(void)
984263bc
MD
221{
222 int f;
223 struct timespec ts;
224
225 getnanouptime(&ts);
6ea70f76 226 kprintf("Uptime: ");
984263bc
MD
227 f = 0;
228 if (ts.tv_sec >= 86400) {
6ea70f76 229 kprintf("%ldd", ts.tv_sec / 86400);
984263bc
MD
230 ts.tv_sec %= 86400;
231 f = 1;
232 }
233 if (f || ts.tv_sec >= 3600) {
6ea70f76 234 kprintf("%ldh", ts.tv_sec / 3600);
984263bc
MD
235 ts.tv_sec %= 3600;
236 f = 1;
237 }
238 if (f || ts.tv_sec >= 60) {
6ea70f76 239 kprintf("%ldm", ts.tv_sec / 60);
984263bc
MD
240 ts.tv_sec %= 60;
241 f = 1;
242 }
6ea70f76 243 kprintf("%lds\n", ts.tv_sec);
984263bc
MD
244}
245
246/*
247 * Go through the rigmarole of shutting down..
248 * this used to be in machdep.c but I'll be dammned if I could see
249 * anything machine dependant in it.
250 */
251static void
41c20dac 252boot(int howto)
984263bc 253{
3ad8cc28
MD
254 /*
255 * Get rid of any user scheduler baggage and then give
256 * us a high priority.
257 */
258 if (curthread->td_release)
259 curthread->td_release(curthread);
260 lwkt_setpri_self(TDPRI_MAX);
984263bc
MD
261
262 /* collect extra flags that shutdown_nice might have set */
263 howto |= shutdown_howto;
264
265#ifdef SMP
7710b6e4
MD
266 /*
267 * We really want to shutdown on the BSP. Subsystems such as ACPI
268 * can't power-down the box otherwise.
269 */
0f7a3396 270 if (smp_active_mask > 1) {
6ea70f76 271 kprintf("boot() called on cpu#%d\n", mycpu->gd_cpuid);
984263bc 272 }
7710b6e4 273 if (panicstr == NULL && mycpu->gd_cpuid != 0) {
6ea70f76 274 kprintf("Switching to cpu #0 for shutdown\n");
7710b6e4
MD
275 lwkt_setcpu_self(globaldata_find(0));
276 }
984263bc
MD
277#endif
278 /*
279 * Do any callouts that should be done BEFORE syncing the filesystems.
280 */
281 EVENTHANDLER_INVOKE(shutdown_pre_sync, howto);
282
f6a22062
MD
283 /*
284 * Try to get rid of any remaining FS references. The calling
285 * process, proc0, and init may still hold references. The
286 * VFS cache subsystem may still hold a root reference to root.
ee92d6d6
MD
287 *
288 * XXX this needs work. We really need to SIGSTOP all remaining
289 * processes in order to avoid blowups due to proc0's filesystem
290 * references going away. For now just make sure that the init
291 * process is stopped.
f6a22062
MD
292 */
293 if (panicstr == NULL) {
294 shutdown_cleanup_proc(curproc);
295 shutdown_cleanup_proc(&proc0);
ee92d6d6
MD
296 if (initproc) {
297 if (initproc != curproc) {
298 ksignal(initproc, SIGSTOP);
299 tsleep(boot, 0, "shutdn", hz / 20);
300 }
f6a22062 301 shutdown_cleanup_proc(initproc);
ee92d6d6 302 }
f6a22062
MD
303 vfs_cache_setroot(NULL, NULL);
304 }
305
984263bc
MD
306 /*
307 * Now sync filesystems
308 */
309 if (!cold && (howto & RB_NOSYNC) == 0 && waittime < 0) {
984263bc
MD
310 int iter, nbusy, pbusy;
311
312 waittime = 0;
6ea70f76 313 kprintf("\nsyncing disks... ");
984263bc 314
753fd850 315 sys_sync(NULL); /* YYY was sync(&proc0, NULL). why proc0 ? */
984263bc
MD
316
317 /*
318 * With soft updates, some buffers that are
319 * written will be remarked as dirty until other
320 * buffers are written.
321 */
322 for (iter = pbusy = 0; iter < 20; iter++) {
41cbf46c 323 nbusy = scan_all_buffers(shutdown_busycount1, NULL);
984263bc
MD
324 if (nbusy == 0)
325 break;
6ea70f76 326 kprintf("%d ", nbusy);
984263bc
MD
327 if (nbusy < pbusy)
328 iter = 0;
329 pbusy = nbusy;
e200c40d
DR
330 /*
331 * XXX:
332 * Process soft update work queue if buffers don't sync
333 * after 6 iterations by permitting the syncer to run.
334 */
408357d8
MD
335 if (iter > 5)
336 bio_ops_sync(NULL);
337
753fd850 338 sys_sync(NULL); /* YYY was sync(&proc0, NULL). why proc0 ? */
3ad8cc28 339 tsleep(boot, 0, "shutdn", hz * iter / 20 + 1);
984263bc 340 }
6ea70f76 341 kprintf("\n");
984263bc
MD
342 /*
343 * Count only busy local buffers to prevent forcing
344 * a fsck if we're just a client of a wedged NFS server
345 */
41cbf46c 346 nbusy = scan_all_buffers(shutdown_busycount2, NULL);
984263bc
MD
347 if (nbusy) {
348 /*
349 * Failed to sync all blocks. Indicate this and don't
350 * unmount filesystems (thus forcing an fsck on reboot).
351 */
6ea70f76 352 kprintf("giving up on %d buffers\n", nbusy);
a88429e8
DR
353#ifdef DDB
354 Debugger("busy buffer problem");
355#endif /* DDB */
3ad8cc28 356 tsleep(boot, 0, "shutdn", hz * 5 + 1);
984263bc 357 } else {
6ea70f76 358 kprintf("done\n");
984263bc
MD
359 /*
360 * Unmount filesystems
361 */
f6a22062 362 if (panicstr == NULL)
984263bc
MD
363 vfs_unmountall();
364 }
3ad8cc28 365 tsleep(boot, 0, "shutdn", hz / 10 + 1);
984263bc
MD
366 }
367
368 print_uptime();
369
370 /*
87e2fa7e 371 * Dump before doing post_sync shutdown ops
984263bc 372 */
9ec81f83 373 crit_enter();
984263bc
MD
374 if ((howto & (RB_HALT|RB_DUMP)) == RB_DUMP && !cold)
375 dumpsys();
376
87e2fa7e
MD
377 /*
378 * Ok, now do things that assume all filesystem activity has
379 * been completed. This will also call the device shutdown
380 * methods.
381 */
382 EVENTHANDLER_INVOKE(shutdown_post_sync, howto);
383
984263bc
MD
384 /* Now that we're going to really halt the system... */
385 EVENTHANDLER_INVOKE(shutdown_final, howto);
386
387 for(;;) ; /* safety against shutdown_reset not working */
388 /* NOTREACHED */
389}
390
41cbf46c
MD
391static int
392shutdown_busycount1(struct buf *bp, void *info)
393{
394 if ((bp->b_flags & B_INVAL) == 0 && BUF_REFCNT(bp) > 0)
395 return(1);
396 if ((bp->b_flags & (B_DELWRI | B_INVAL)) == B_DELWRI)
397 return (1);
398 return (0);
399}
400
401static int
402shutdown_busycount2(struct buf *bp, void *info)
403{
404 if (((bp->b_flags & B_INVAL) == 0 && BUF_REFCNT(bp)) ||
405 ((bp->b_flags & (B_DELWRI|B_INVAL)) == B_DELWRI)) {
406 /*
407 * Only count buffers undergoing write I/O
408 * on the related vnode.
409 */
410 if (bp->b_vp == NULL ||
411 bp->b_vp->v_track_write.bk_active == 0) {
412 return (0);
413 }
414#if defined(SHOW_BUSYBUFS) || defined(DIAGNOSTIC)
6ea70f76 415 kprintf(
ce54c037 416 "%p dev:?, flags:%08x, loffset:%lld, doffset:%lld\n",
2dc6c129 417 bp,
41cbf46c
MD
418 bp->b_flags, bp->b_loffset,
419 bp->b_bio2.bio_offset);
420#endif
421 return(1);
422 }
423 return(0);
424}
425
984263bc
MD
426/*
427 * If the shutdown was a clean halt, behave accordingly.
428 */
429static void
430shutdown_halt(void *junk, int howto)
431{
432 if (howto & RB_HALT) {
6ea70f76
SW
433 kprintf("\n");
434 kprintf("The operating system has halted.\n");
40d0276c
MD
435#ifdef _KERNEL_VIRTUAL
436 cpu_halt();
437#else
6ea70f76 438 kprintf("Please press any key to reboot.\n\n");
984263bc
MD
439 switch (cngetc()) {
440 case -1: /* No console, just die */
441 cpu_halt();
442 /* NOTREACHED */
443 default:
444 howto &= ~RB_HALT;
445 break;
446 }
40d0276c 447#endif
984263bc
MD
448 }
449}
450
451/*
452 * Check to see if the system paniced, pause and then reboot
453 * according to the specified delay.
454 */
455static void
456shutdown_panic(void *junk, int howto)
457{
458 int loop;
459
460 if (howto & RB_DUMP) {
461 if (PANIC_REBOOT_WAIT_TIME != 0) {
462 if (PANIC_REBOOT_WAIT_TIME != -1) {
6ea70f76 463 kprintf("Automatic reboot in %d seconds - "
984263bc
MD
464 "press a key on the console to abort\n",
465 PANIC_REBOOT_WAIT_TIME);
466 for (loop = PANIC_REBOOT_WAIT_TIME * 10;
467 loop > 0; --loop) {
468 DELAY(1000 * 100); /* 1/10th second */
469 /* Did user type a key? */
470 if (cncheckc() != -1)
471 break;
472 }
473 if (!loop)
474 return;
475 }
476 } else { /* zero time specified - reboot NOW */
477 return;
478 }
6ea70f76
SW
479 kprintf("--> Press a key on the console to reboot,\n");
480 kprintf("--> or switch off the system now.\n");
984263bc
MD
481 cngetc();
482 }
483}
484
485/*
486 * Everything done, now reset
487 */
488static void
489shutdown_reset(void *junk, int howto)
490{
6ea70f76
SW
491 kprintf("Rebooting...\n");
492 DELAY(1000000); /* wait 1 sec for kprintf's to complete and be read */
984263bc
MD
493 /* cpu_boot(howto); */ /* doesn't do anything at the moment */
494 cpu_reset();
495 /* NOTREACHED */ /* assuming reset worked */
496}
497
498/*
f6a22062
MD
499 * Try to remove FS references in the specified process. This function
500 * is used during shutdown
501 */
502static
503void
504shutdown_cleanup_proc(struct proc *p)
505{
506 struct filedesc *fdp;
507 struct vmspace *vm;
508
509 if (p == NULL)
510 return;
511 if ((fdp = p->p_fd) != NULL) {
512 kern_closefrom(0);
513 if (fdp->fd_cdir) {
28623bf9 514 cache_drop(&fdp->fd_ncdir);
f6a22062
MD
515 vrele(fdp->fd_cdir);
516 fdp->fd_cdir = NULL;
f6a22062
MD
517 }
518 if (fdp->fd_rdir) {
28623bf9 519 cache_drop(&fdp->fd_nrdir);
f6a22062
MD
520 vrele(fdp->fd_rdir);
521 fdp->fd_rdir = NULL;
f6a22062
MD
522 }
523 if (fdp->fd_jdir) {
28623bf9 524 cache_drop(&fdp->fd_njdir);
f6a22062
MD
525 vrele(fdp->fd_jdir);
526 fdp->fd_jdir = NULL;
f6a22062
MD
527 }
528 }
4a22e893
MD
529 if (p->p_vkernel)
530 vkernel_exit(p);
f6a22062
MD
531 if (p->p_textvp) {
532 vrele(p->p_textvp);
533 p->p_textvp = NULL;
534 }
535 vm = p->p_vmspace;
536 if (vm != NULL) {
537 pmap_remove_pages(vmspace_pmap(vm),
88181b08
MD
538 VM_MIN_USER_ADDRESS,
539 VM_MAX_USER_ADDRESS);
f6a22062 540 vm_map_remove(&vm->vm_map,
88181b08
MD
541 VM_MIN_USER_ADDRESS,
542 VM_MAX_USER_ADDRESS);
f6a22062
MD
543 }
544}
545
546/*
984263bc
MD
547 * Magic number for savecore
548 *
549 * exported (symorder) and used at least by savecore(8)
550 *
04a77147 551 * Mark it as used so that gcc doesn't optimize it away.
984263bc 552 */
04a77147
SS
553__attribute__((__used__))
554 static u_long const dumpmag = 0x8fca0101UL;
984263bc
MD
555
556static int dumpsize = 0; /* also for savecore */
557
558static int dodump = 1;
559
560SYSCTL_INT(_machdep, OID_AUTO, do_dump, CTLFLAG_RW, &dodump, 0,
561 "Try to perform coredump on kernel panic");
562
563static int
c972a82f 564setdumpdev(cdev_t dev)
984263bc 565{
a6c0f342 566 struct partinfo pinfo;
e0fc5693 567 u_int64_t newdumplo;
a6c0f342 568 int error;
faa575e7 569 int doopen;
984263bc 570
028066b1 571 if (dev == NULL) {
984263bc
MD
572 dumpdev = dev;
573 return (0);
574 }
a6c0f342 575 bzero(&pinfo, sizeof(pinfo));
faa575e7
MD
576
577 /*
578 * We have to open the device before we can perform ioctls on it,
579 * or the slice/label data may not be present. Device opens are
580 * usually tracked by specfs, but the dump device can be set in
581 * early boot and may not be open so this is somewhat of a hack.
582 */
583 doopen = (dev->si_sysref.refcnt == 1);
584 if (doopen) {
585 error = dev_dopen(dev, FREAD, S_IFCHR, proc0.p_ucred);
586 if (error)
587 return (error);
588 }
a6c0f342 589 error = dev_dioctl(dev, DIOCGPART, (void *)&pinfo, 0, proc0.p_ucred);
faa575e7
MD
590 if (doopen)
591 dev_dclose(dev, FREAD, S_IFCHR);
a6c0f342 592 if (error || pinfo.media_blocks == 0 || pinfo.media_blksize == 0)
335dda38 593 return (ENXIO);
a6c0f342
MD
594
595 newdumplo = pinfo.media_blocks -
596 ((u_int64_t)Maxmem * PAGE_SIZE / DEV_BSIZE);
1c3c151b 597 if ((int64_t)newdumplo < (int64_t)pinfo.reserved_blocks)
984263bc
MD
598 return (ENOSPC);
599 dumpdev = dev;
e0fc5693 600 dumplo64 = newdumplo;
984263bc
MD
601 return (0);
602}
603
604
605/* ARGSUSED */
402ed7e1 606static void dump_conf (void *dummy);
984263bc 607static void
c972a82f 608dump_conf(void *dummy)
984263bc
MD
609{
610 char *path;
b13267a5 611 cdev_t dev;
984263bc 612
efda3bd0 613 path = kmalloc(MNAMELEN, M_TEMP, M_WAITOK);
984263bc 614 if (TUNABLE_STR_FETCH("dumpdev", path, MNAMELEN) != 0) {
bc01a404 615 dev = kgetdiskbyname(path);
028066b1 616 if (dev != NULL)
984263bc
MD
617 dumpdev = dev;
618 }
efda3bd0 619 kfree(path, M_TEMP);
984263bc 620 if (setdumpdev(dumpdev) != 0)
028066b1 621 dumpdev = NULL;
984263bc
MD
622}
623
624SYSINIT(dump_conf, SI_SUB_DUMP_CONF, SI_ORDER_FIRST, dump_conf, NULL)
625
626static int
627sysctl_kern_dumpdev(SYSCTL_HANDLER_ARGS)
628{
629 int error;
630 udev_t ndumpdev;
631
632 ndumpdev = dev2udev(dumpdev);
633 error = sysctl_handle_opaque(oidp, &ndumpdev, sizeof ndumpdev, req);
634 if (error == 0 && req->newptr != NULL)
635 error = setdumpdev(udev2dev(ndumpdev, 0));
636 return (error);
637}
638
639SYSCTL_PROC(_kern, KERN_DUMPDEV, dumpdev, CTLTYPE_OPAQUE|CTLFLAG_RW,
936c90c4 640 0, sizeof dumpdev, sysctl_kern_dumpdev, "T,udev_t", "");
984263bc
MD
641
642/*
643 * Doadump comes here after turning off memory management and
644 * getting on the dump stack, either when called above, or by
645 * the auto-restart code.
646 */
647static void
648dumpsys(void)
649{
650 int error;
651
652 savectx(&dumppcb);
8b6a428f 653 dumpthread = curthread;
984263bc 654 if (dumping++) {
6ea70f76 655 kprintf("Dump already in progress, bailing...\n");
984263bc
MD
656 return;
657 }
658 if (!dodump)
659 return;
028066b1 660 if (dumpdev == NULL)
984263bc 661 return;
984263bc 662 dumpsize = Maxmem;
e0fc5693 663 kprintf("\ndumping to dev %s, blockno %lld\n",
973c11b9
MD
664 devtoname(dumpdev),
665 (long long)dumplo64);
6ea70f76 666 kprintf("dump ");
335dda38 667 error = dev_ddump(dumpdev);
984263bc 668 if (error == 0) {
6ea70f76 669 kprintf("succeeded\n");
984263bc
MD
670 return;
671 }
6ea70f76 672 kprintf("failed, reason: ");
984263bc 673 switch (error) {
335dda38 674 case ENOSYS:
984263bc 675 case ENODEV:
6ea70f76 676 kprintf("device doesn't support a dump routine\n");
984263bc
MD
677 break;
678
679 case ENXIO:
6ea70f76 680 kprintf("device bad\n");
984263bc
MD
681 break;
682
683 case EFAULT:
6ea70f76 684 kprintf("device not ready\n");
984263bc
MD
685 break;
686
687 case EINVAL:
6ea70f76 688 kprintf("area improper\n");
984263bc
MD
689 break;
690
691 case EIO:
6ea70f76 692 kprintf("i/o error\n");
984263bc
MD
693 break;
694
695 case EINTR:
6ea70f76 696 kprintf("aborted from console\n");
984263bc
MD
697 break;
698
699 default:
6ea70f76 700 kprintf("unknown, error = %d\n", error);
984263bc
MD
701 break;
702 }
703}
704
705int
706dumpstatus(vm_offset_t addr, off_t count)
707{
708 int c;
709
710 if (addr % (1024 * 1024) == 0) {
711#ifdef HW_WDOG
712 if (wdog_tickler)
713 (*wdog_tickler)();
714#endif
6ea70f76 715 kprintf("%ld ", (long)(count / (1024 * 1024)));
984263bc
MD
716 }
717
718 if ((c = cncheckc()) == 0x03)
719 return -1;
720 else if (c != -1)
6ea70f76 721 kprintf("[CTRL-C to abort] ");
984263bc
MD
722
723 return 0;
724}
725
726/*
727 * Panic is called on unresolvable fatal errors. It prints "panic: mesg",
728 * and then reboots. If we are called twice, then we avoid trying to sync
729 * the disks as this often leads to recursive panics.
730 */
731void
732panic(const char *fmt, ...)
733{
b1e04573 734 int bootopt, newpanic;
e2565a42 735 __va_list ap;
984263bc
MD
736 static char buf[256];
737
ddcc5e10
MD
738#ifdef SMP
739 /*
740 * If a panic occurs on multiple cpus before the first is able to
741 * halt the other cpus, only one cpu is allowed to take the panic.
6ea70f76 742 * Attempt to be verbose about this situation but if the kprintf()
ddcc5e10
MD
743 * itself panics don't let us overrun the kernel stack.
744 *
745 * Be very nasty about descheduling our thread at the lowest
746 * level possible in an attempt to freeze the thread without
747 * inducing further panics.
748 *
749 * Bumping gd_trap_nesting_level will also bypass assertions in
750 * lwkt_switch() and allow us to switch away even if we are a
751 * FAST interrupt or IPI.
752 */
753 if (atomic_poll_acquire_int(&panic_cpu_interlock)) {
754 panic_cpu_gd = mycpu;
755 } else if (panic_cpu_gd != mycpu) {
756 crit_enter();
757 ++mycpu->gd_trap_nesting_level;
758 if (mycpu->gd_trap_nesting_level < 25) {
6ea70f76 759 kprintf("SECONDARY PANIC ON CPU %d THREAD %p\n",
ddcc5e10
MD
760 mycpu->gd_cpuid, curthread);
761 }
762 curthread->td_release = NULL; /* be a grinch */
763 for (;;) {
764 lwkt_deschedule_self(curthread);
765 lwkt_switch();
766 }
767 /* NOT REACHED */
768 /* --mycpu->gd_trap_nesting_level */
769 /* crit_exit() */
770 }
771#endif
984263bc 772 bootopt = RB_AUTOBOOT | RB_DUMP;
baf88b6b
MD
773 if (sync_on_panic == 0)
774 bootopt |= RB_NOSYNC;
b1e04573 775 newpanic = 0;
984263bc
MD
776 if (panicstr)
777 bootopt |= RB_NOSYNC;
b1e04573 778 else {
984263bc 779 panicstr = fmt;
b1e04573
JH
780 newpanic = 1;
781 }
984263bc 782
e2565a42 783 __va_start(ap, fmt);
379210cb 784 kvsnprintf(buf, sizeof(buf), fmt, ap);
984263bc
MD
785 if (panicstr == fmt)
786 panicstr = buf;
e2565a42 787 __va_end(ap);
6ea70f76 788 kprintf("panic: %s\n", buf);
984263bc 789#ifdef SMP
6a8aa90e 790 /* two separate prints in case of an unmapped page and trap */
6ea70f76 791 kprintf("mp_lock = %08x; ", mp_lock);
d1a8af1f 792 kprintf("cpuid = %d\n", mycpu->gd_cpuid);
984263bc
MD
793#endif
794
795#if defined(DDB)
b1e04573 796 if (newpanic && trace_on_panic)
1e5fb84b 797 print_backtrace();
984263bc 798 if (debugger_on_panic)
3cc1a5a9 799 Debugger("panic");
984263bc
MD
800#endif
801 boot(bootopt);
802}
803
804/*
805 * Support for poweroff delay.
806 */
807#ifndef POWEROFF_DELAY
808# define POWEROFF_DELAY 5000
809#endif
810static int poweroff_delay = POWEROFF_DELAY;
811
812SYSCTL_INT(_kern_shutdown, OID_AUTO, poweroff_delay, CTLFLAG_RW,
813 &poweroff_delay, 0, "");
814
815static void
816poweroff_wait(void *junk, int howto)
817{
818 if(!(howto & RB_POWEROFF) || poweroff_delay <= 0)
819 return;
820 DELAY(poweroff_delay * 1000);
821}
822
823/*
824 * Some system processes (e.g. syncer) need to be stopped at appropriate
825 * points in their main loops prior to a system shutdown, so that they
826 * won't interfere with the shutdown process (e.g. by holding a disk buf
827 * to cause sync to fail). For each of these system processes, register
828 * shutdown_kproc() as a handler for one of shutdown events.
829 */
830static int kproc_shutdown_wait = 60;
831SYSCTL_INT(_kern_shutdown, OID_AUTO, kproc_shutdown_wait, CTLFLAG_RW,
832 &kproc_shutdown_wait, 0, "");
833
834void
835shutdown_kproc(void *arg, int howto)
836{
bc6dffab 837 struct thread *td;
984263bc
MD
838 struct proc *p;
839 int error;
840
841 if (panicstr)
842 return;
843
bc6dffab
MD
844 td = (struct thread *)arg;
845 if ((p = td->td_proc) != NULL) {
6ea70f76 846 kprintf("Waiting (max %d seconds) for system process `%s' to stop...",
bc6dffab
MD
847 kproc_shutdown_wait, p->p_comm);
848 } else {
6ea70f76 849 kprintf("Waiting (max %d seconds) for system thread %s to stop...",
0cfcada1 850 kproc_shutdown_wait, td->td_comm);
bc6dffab
MD
851 }
852 error = suspend_kproc(td, kproc_shutdown_wait * hz);
984263bc
MD
853
854 if (error == EWOULDBLOCK)
6ea70f76 855 kprintf("timed out\n");
984263bc 856 else
6ea70f76 857 kprintf("stopped\n");
984263bc 858}