savecore.c revision 1.26 1 /* $NetBSD: savecore.c,v 1.26 1996/03/18 21:16:05 leo Exp $ */
2
3 /*-
4 * Copyright (c) 1986, 1992, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. All advertising materials mentioning features or use of this software
16 * must display the following acknowledgement:
17 * This product includes software developed by the University of
18 * California, Berkeley and its contributors.
19 * 4. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 */
35
36 #ifndef lint
37 static char copyright[] =
38 "@(#) Copyright (c) 1986, 1992, 1993\n\
39 The Regents of the University of California. All rights reserved.\n";
40 #endif /* not lint */
41
42 #ifndef lint
43 #if 0
44 static char sccsid[] = "@(#)savecore.c 8.3 (Berkeley) 1/2/94";
45 #else
46 static char rcsid[] = "$NetBSD: savecore.c,v 1.26 1996/03/18 21:16:05 leo Exp $";
47 #endif
48 #endif /* not lint */
49
50 #include <sys/param.h>
51 #include <sys/stat.h>
52 #include <sys/mount.h>
53 #include <sys/syslog.h>
54 #include <sys/time.h>
55
56 #include <dirent.h>
57 #include <errno.h>
58 #include <fcntl.h>
59 #include <nlist.h>
60 #include <paths.h>
61 #include <stdio.h>
62 #include <stdlib.h>
63 #include <string.h>
64 #include <tzfile.h>
65 #include <unistd.h>
66 #include <limits.h>
67 #include <kvm.h>
68
69 extern FILE *zopen __P((const char *fname, const char *mode, int bits));
70
71 #define KREAD(kd, addr, p)\
72 (kvm_read(kd, addr, (char *)(p), sizeof(*(p))) != sizeof(*(p)))
73
74 struct nlist current_nl[] = { /* Namelist for currently running system. */
75 #define X_DUMPDEV 0
76 { "_dumpdev" },
77 #define X_DUMPLO 1
78 { "_dumplo" },
79 #define X_TIME 2
80 { "_time" },
81 #define X_DUMPSIZE 3
82 { "_dumpsize" },
83 #define X_VERSION 4
84 { "_version" },
85 #define X_PANICSTR 5
86 { "_panicstr" },
87 #define X_DUMPMAG 6
88 { "_dumpmag" },
89 { NULL },
90 };
91 int cursyms[] = { X_DUMPDEV, X_DUMPLO, X_VERSION, X_DUMPMAG, -1 };
92 int dumpsyms[] = { X_TIME, X_DUMPSIZE, X_VERSION, X_PANICSTR, X_DUMPMAG, -1 };
93
94 struct nlist dump_nl[] = { /* Name list for dumped system. */
95 { "_dumpdev" }, /* Entries MUST be the same as */
96 { "_dumplo" }, /* those in current_nl[]. */
97 { "_time" },
98 { "_dumpsize" },
99 { "_version" },
100 { "_panicstr" },
101 { "_dumpmag" },
102 { NULL },
103 };
104
105 /* Types match kernel declarations. */
106 long dumplo; /* where dump starts on dumpdev */
107 int dumpmag; /* magic number in dump */
108 int dumpsize; /* amount of memory dumped */
109
110 char *kernel;
111 char *dirname; /* directory to save dumps in */
112 char *ddname; /* name of dump device */
113 dev_t dumpdev; /* dump device */
114 int dumpfd; /* read/write descriptor on block dev */
115 kvm_t *kd_dump; /* kvm descriptor on block dev */
116 time_t now; /* current date */
117 char panic_mesg[1024];
118 int panicstr;
119 char vers[1024];
120
121 int clear, compress, force, verbose; /* flags */
122
123 void check_kmem __P((void));
124 int check_space __P((void));
125 void clear_dump __P((void));
126 int Create __P((char *, int));
127 int dump_exists __P((void));
128 char *find_dev __P((dev_t, int));
129 int get_crashtime __P((void));
130 void kmem_setup __P((void));
131 void log __P((int, char *, ...));
132 void Lseek __P((int, off_t, int));
133 int Open __P((char *, int rw));
134 char *rawname __P((char *s));
135 void save_core __P((void));
136 void usage __P((void));
137 void Write __P((int, void *, int));
138
139 int
140 main(argc, argv)
141 int argc;
142 char *argv[];
143 {
144 int ch;
145
146 openlog("savecore", LOG_PERROR, LOG_DAEMON);
147
148 while ((ch = getopt(argc, argv, "cdfN:vz")) != -1)
149 switch(ch) {
150 case 'c':
151 clear = 1;
152 break;
153 case 'd': /* Not documented. */
154 case 'v':
155 verbose = 1;
156 break;
157 case 'f':
158 force = 1;
159 break;
160 case 'N':
161 kernel = optarg;
162 break;
163 case 'z':
164 compress = 1;
165 break;
166 case '?':
167 default:
168 usage();
169 }
170 argc -= optind;
171 argv += optind;
172
173 if (!clear) {
174 if (argc != 1 && argc != 2)
175 usage();
176 dirname = argv[0];
177 }
178 if (argc == 2)
179 kernel = argv[1];
180
181 (void)time(&now);
182 kmem_setup();
183
184 if (clear) {
185 clear_dump();
186 exit(0);
187 }
188
189 if (!dump_exists() && !force)
190 exit(1);
191
192 check_kmem();
193
194 if (panicstr)
195 syslog(LOG_ALERT, "reboot after panic: %s", panic_mesg);
196 else
197 syslog(LOG_ALERT, "reboot");
198
199 if ((!get_crashtime() || !check_space()) && !force)
200 exit(1);
201
202 save_core();
203
204 clear_dump();
205 exit(0);
206 }
207
208 void
209 kmem_setup()
210 {
211 kvm_t *kd_kern;
212 char errbuf[_POSIX2_LINE_MAX];
213 int i, hdrsz;
214 char *dump_sys;
215
216 /*
217 * Some names we need for the currently running system, others for
218 * the system that was running when the dump was made. The values
219 * obtained from the current system are used to look for things in
220 * /dev/kmem that cannot be found in the dump_sys namelist, but are
221 * presumed to be the same (since the disk partitions are probably
222 * the same!)
223 */
224 kd_kern = kvm_openfiles(NULL, NULL, NULL, O_RDONLY, errbuf);
225 if (kd_kern == NULL) {
226 syslog(LOG_ERR, "%s: kvm_openfiles: %s", _PATH_UNIX, errbuf);
227 exit(1);
228 }
229 if (kvm_nlist(kd_kern, current_nl) == -1)
230 syslog(LOG_ERR, "%s: kvm_nlist: %s", _PATH_UNIX,
231 kvm_geterr(kd_kern));
232
233 for (i = 0; cursyms[i] != -1; i++)
234 if (current_nl[cursyms[i]].n_value == 0) {
235 syslog(LOG_ERR, "%s: %s not in namelist",
236 _PATH_UNIX, current_nl[cursyms[i]].n_name);
237 exit(1);
238 }
239
240 KREAD(kd_kern, current_nl[X_DUMPDEV].n_value, &dumpdev);
241 if (dumpdev == NODEV) {
242 syslog(LOG_WARNING, "no core dump (no dumpdev)");
243 exit(1);
244 }
245 KREAD(kd_kern, current_nl[X_DUMPLO].n_value, &dumplo);
246 dumplo *= DEV_BSIZE;
247 if (verbose)
248 (void)printf("dumplo = %d (%d * %d)\n",
249 dumplo, dumplo / DEV_BSIZE, DEV_BSIZE);
250 KREAD(kd_kern, current_nl[X_DUMPMAG].n_value, &dumpmag);
251
252 if (kernel == NULL) {
253 (void)kvm_read(kd_kern, current_nl[X_VERSION].n_value,
254 vers, sizeof(vers));
255 vers[sizeof(vers) - 1] = '\0';
256 }
257
258 ddname = find_dev(dumpdev, S_IFBLK);
259 dumpfd = Open(ddname, O_RDWR);
260
261 dump_sys = kernel ? kernel : _PATH_UNIX;
262
263 kd_dump = kvm_openfiles(dump_sys, ddname, NULL, O_RDWR, errbuf);
264 if (kd_dump == NULL) {
265 syslog(LOG_ERR, "%s: kvm_openfiles: %s", dump_sys, errbuf);
266 exit(1);
267 }
268
269 if (kvm_nlist(kd_dump, dump_nl) == -1)
270 syslog(LOG_ERR, "%s: kvm_nlist: %s", dump_sys,
271 kvm_geterr(kd_dump));
272
273 for (i = 0; dumpsyms[i] != -1; i++)
274 if (dump_nl[dumpsyms[i]].n_value == 0) {
275 syslog(LOG_ERR, "%s: %s not in namelist",
276 dump_sys, dump_nl[dumpsyms[i]].n_name);
277 exit(1);
278 }
279 hdrsz = kvm_dump_mkheader(kd_dump, (off_t)dumplo);
280 if (hdrsz == -1) {
281 syslog(LOG_ERR, "%s: kvm_dump_mkheader: %s", dump_sys,
282 kvm_geterr(kd_dump));
283 exit(1);
284 }
285 dumplo += hdrsz;
286 kvm_close(kd_kern);
287 }
288
289 void
290 check_kmem()
291 {
292 register char *cp;
293 register int panicloc;
294 char core_vers[1024];
295
296 (void)kvm_read(kd_dump, dump_nl[X_VERSION].n_value, core_vers,
297 sizeof(core_vers));
298 core_vers[sizeof(core_vers) - 1] = '\0';
299
300 if (strcmp(vers, core_vers) && kernel == 0)
301 syslog(LOG_WARNING,
302 "warning: %s version mismatch:\n\t%s\nand\t%s\n",
303 _PATH_UNIX, vers, core_vers);
304
305 KREAD(kd_dump, dump_nl[X_PANICSTR].n_value, &panicstr);
306 if (panicstr) {
307 cp = panic_mesg;
308 panicloc = panicstr;
309 do {
310 KREAD(kd_dump, panicloc, cp);
311 panicloc++;
312 } while (*cp++ && cp < &panic_mesg[sizeof(panic_mesg)]);
313 }
314 }
315
316 int
317 dump_exists()
318 {
319 int newdumpmag;
320
321 KREAD(kd_dump, dump_nl[X_DUMPMAG].n_value, &newdumpmag);
322
323 /* Read the dump size. */
324 KREAD(kd_dump, dump_nl[X_DUMPSIZE].n_value, &dumpsize);
325 dumpsize *= getpagesize();
326
327 /*
328 * Return zero if core dump doesn't seem to be there, and note
329 * it for syslog. This check and return happens after the dump size
330 * is read, so dumpsize is whether or not the core is valid (for -f).
331 */
332 if (newdumpmag != dumpmag) {
333 if (verbose)
334 syslog(LOG_WARNING, "magic number mismatch (%x != %x)",
335 newdumpmag, dumpmag);
336 syslog(LOG_WARNING, "no core dump");
337 return (0);
338 }
339 return (1);
340 }
341
342 void
343 clear_dump()
344 {
345 if (kvm_dump_inval(kd_dump) == -1)
346 syslog(LOG_ERR, "%s: kvm_clear_dump: %s", ddname,
347 kvm_geterr(kd_dump));
348
349 }
350
351 char buf[1024 * 1024];
352
353 void
354 save_core()
355 {
356 register FILE *fp;
357 register int bounds, ifd, nr, nw, ofd;
358 char *rawp, path[MAXPATHLEN];
359
360 /*
361 * Get the current number and update the bounds file. Do the update
362 * now, because may fail later and don't want to overwrite anything.
363 */
364 (void)snprintf(path, sizeof(path), "%s/bounds", dirname);
365 if ((fp = fopen(path, "r")) == NULL)
366 goto err1;
367 if (fgets(buf, sizeof(buf), fp) == NULL) {
368 if (ferror(fp))
369 err1: syslog(LOG_WARNING, "%s: %s", path, strerror(errno));
370 bounds = 0;
371 } else
372 bounds = atoi(buf);
373 if (fp != NULL)
374 (void)fclose(fp);
375 if ((fp = fopen(path, "w")) == NULL)
376 syslog(LOG_ERR, "%s: %m", path);
377 else {
378 (void)fprintf(fp, "%d\n", bounds + 1);
379 (void)fclose(fp);
380 }
381 (void)fclose(fp);
382
383 /* Create the core file. */
384 (void)snprintf(path, sizeof(path), "%s/netbsd.%d.core%s",
385 dirname, bounds, compress ? ".Z" : "");
386 if (compress) {
387 if ((fp = zopen(path, "w", 0)) == NULL) {
388 syslog(LOG_ERR, "%s: %s", path, strerror(errno));
389 exit(1);
390 }
391 } else {
392 ofd = Create(path, S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH);
393 fp = fdopen(ofd, "w");
394 if (fp == NULL) {
395 syslog(LOG_ERR, "%s: fdopen: %s", path);
396 exit(1);
397 }
398 }
399
400 /* Open the raw device. */
401 rawp = rawname(ddname);
402 if ((ifd = open(rawp, O_RDONLY)) == -1) {
403 syslog(LOG_WARNING, "%s: %m; using block device", rawp);
404 ifd = dumpfd;
405 }
406
407 /* Seek to the start of the core. */
408 Lseek(ifd, (off_t)dumplo, L_SET);
409
410 if (kvm_dump_wrtheader(kd_dump, fp, dumpsize) == -1) {
411 syslog(LOG_ERR, "kvm_dump_wrtheader: %s : %s", path,
412 kvm_geterr(kd_dump));
413 exit(1);
414 }
415
416 /* Copy the core file. */
417 syslog(LOG_NOTICE, "writing %score to %s",
418 compress ? "compressed " : "", path);
419 for (; dumpsize > 0; dumpsize -= nr) {
420 (void)printf("%6dK\r", dumpsize / 1024);
421 (void)fflush(stdout);
422 nr = read(ifd, buf, MIN(dumpsize, sizeof(buf)));
423 if (nr <= 0) {
424 if (nr == 0)
425 syslog(LOG_WARNING,
426 "WARNING: EOF on dump device");
427 else
428 syslog(LOG_ERR, "%s: %m", rawp);
429 goto err2;
430 }
431 nw = fwrite(buf, 1, nr, fp);
432 if (nw != nr) {
433 syslog(LOG_ERR, "%s: %s",
434 path, strerror(nw == 0 ? EIO : errno));
435 err2: syslog(LOG_WARNING,
436 "WARNING: core may be incomplete");
437 (void)printf("\n");
438 exit(1);
439 }
440 }
441 (void)close(ifd);
442 (void)fclose(fp);
443
444 /* Copy the kernel. */
445 ifd = Open(kernel ? kernel : _PATH_UNIX, O_RDONLY);
446 (void)snprintf(path, sizeof(path), "%s/netbsd.%d%s",
447 dirname, bounds, compress ? ".Z" : "");
448 if (compress) {
449 if ((fp = zopen(path, "w", 0)) == NULL) {
450 syslog(LOG_ERR, "%s: %s", path, strerror(errno));
451 exit(1);
452 }
453 } else
454 ofd = Create(path, S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH);
455 syslog(LOG_NOTICE, "writing %skernel to %s",
456 compress ? "compressed " : "", path);
457 while ((nr = read(ifd, buf, sizeof(buf))) > 0) {
458 if (compress)
459 nw = fwrite(buf, 1, nr, fp);
460 else
461 nw = write(ofd, buf, nr);
462 if (nw != nr) {
463 syslog(LOG_ERR, "%s: %s",
464 path, strerror(nw == 0 ? EIO : errno));
465 syslog(LOG_WARNING,
466 "WARNING: kernel may be incomplete");
467 exit(1);
468 }
469 }
470 if (nr < 0) {
471 syslog(LOG_ERR, "%s: %s",
472 kernel ? kernel : _PATH_UNIX, strerror(errno));
473 syslog(LOG_WARNING,
474 "WARNING: kernel may be incomplete");
475 exit(1);
476 }
477 if (compress)
478 (void)fclose(fp);
479 else
480 (void)close(ofd);
481 }
482
483 char *
484 find_dev(dev, type)
485 register dev_t dev;
486 register int type;
487 {
488 register DIR *dfd;
489 struct dirent *dir;
490 struct stat sb;
491 char *dp, devname[MAXPATHLEN + 1];
492
493 if ((dfd = opendir(_PATH_DEV)) == NULL) {
494 syslog(LOG_ERR, "%s: %s", _PATH_DEV, strerror(errno));
495 exit(1);
496 }
497 (void)strcpy(devname, _PATH_DEV);
498 while ((dir = readdir(dfd))) {
499 (void)strcpy(devname + sizeof(_PATH_DEV) - 1, dir->d_name);
500 if (lstat(devname, &sb)) {
501 syslog(LOG_ERR, "%s: %s", devname, strerror(errno));
502 continue;
503 }
504 if ((sb.st_mode & S_IFMT) != type)
505 continue;
506 if (dev == sb.st_rdev) {
507 closedir(dfd);
508 if ((dp = strdup(devname)) == NULL) {
509 syslog(LOG_ERR, "%s", strerror(errno));
510 exit(1);
511 }
512 return (dp);
513 }
514 }
515 closedir(dfd);
516 syslog(LOG_ERR, "can't find device %d/%d", major(dev), minor(dev));
517 exit(1);
518 }
519
520 char *
521 rawname(s)
522 char *s;
523 {
524 char *sl, name[MAXPATHLEN];
525
526 if ((sl = strrchr(s, '/')) == NULL || sl[1] == '0') {
527 syslog(LOG_ERR,
528 "can't make raw dump device name from %s", s);
529 return (s);
530 }
531 (void)snprintf(name, sizeof(name), "%.*s/r%s", sl - s, s, sl + 1);
532 if ((sl = strdup(name)) == NULL) {
533 syslog(LOG_ERR, "%s", strerror(errno));
534 exit(1);
535 }
536 return (sl);
537 }
538
539 int
540 get_crashtime()
541 {
542 time_t dumptime; /* Time the dump was taken. */
543
544 KREAD(kd_dump, dump_nl[X_TIME].n_value, &dumptime);
545 if (dumptime == 0) {
546 if (verbose)
547 syslog(LOG_ERR, "dump time is zero");
548 return (0);
549 }
550 (void)printf("savecore: system went down at %s", ctime(&dumptime));
551 #define LEEWAY (7 * SECSPERDAY)
552 if (dumptime < now - LEEWAY || dumptime > now + LEEWAY) {
553 (void)printf("dump time is unreasonable\n");
554 return (0);
555 }
556 return (1);
557 }
558
559 int
560 check_space()
561 {
562 register FILE *fp;
563 char *tkernel;
564 off_t minfree, spacefree, kernelsize, needed;
565 struct stat st;
566 struct statfs fsbuf;
567 char buf[100], path[MAXPATHLEN];
568
569 tkernel = kernel ? kernel : _PATH_UNIX;
570 if (stat(tkernel, &st) < 0) {
571 syslog(LOG_ERR, "%s: %m", tkernel);
572 exit(1);
573 }
574 kernelsize = st.st_blocks * S_BLKSIZE;
575 if (statfs(dirname, &fsbuf) < 0) {
576 syslog(LOG_ERR, "%s: %m", dirname);
577 exit(1);
578 }
579 spacefree = (fsbuf.f_bavail * fsbuf.f_bsize) / 1024;
580
581 (void)snprintf(path, sizeof(path), "%s/minfree", dirname);
582 if ((fp = fopen(path, "r")) == NULL)
583 minfree = 0;
584 else {
585 if (fgets(buf, sizeof(buf), fp) == NULL)
586 minfree = 0;
587 else
588 minfree = atoi(buf);
589 (void)fclose(fp);
590 }
591
592 needed = (dumpsize + kernelsize) / 1024;
593 if (minfree > 0 && spacefree - needed < minfree) {
594 syslog(LOG_WARNING,
595 "no dump, not enough free space on device");
596 return (0);
597 }
598 if (spacefree - needed < minfree)
599 syslog(LOG_WARNING,
600 "dump performed, but free space threshold crossed");
601 return (1);
602 }
603
604 int
605 Open(name, rw)
606 char *name;
607 int rw;
608 {
609 int fd;
610
611 if ((fd = open(name, rw, 0)) < 0) {
612 syslog(LOG_ERR, "%s: %m", name);
613 exit(1);
614 }
615 return (fd);
616 }
617
618 void
619 Lseek(fd, off, flag)
620 int fd, flag;
621 off_t off;
622 {
623 off_t ret;
624
625 ret = lseek(fd, off, flag);
626 if (ret == -1) {
627 syslog(LOG_ERR, "lseek: %m");
628 exit(1);
629 }
630 }
631
632 int
633 Create(file, mode)
634 char *file;
635 int mode;
636 {
637 register int fd;
638
639 fd = open(file, O_WRONLY | O_CREAT | O_TRUNC, mode);
640 if (fd < 0) {
641 syslog(LOG_ERR, "%s: %m", file);
642 exit(1);
643 }
644 return (fd);
645 }
646
647 void
648 Write(fd, bp, size)
649 int fd, size;
650 void *bp;
651 {
652 int n;
653
654 if ((n = write(fd, bp, size)) < size) {
655 syslog(LOG_ERR, "write: %s", strerror(n == -1 ? errno : EIO));
656 exit(1);
657 }
658 }
659
660 void
661 usage()
662 {
663 (void)syslog(LOG_ERR, "usage: savecore [-cfvz] [-N system] directory");
664 exit(1);
665 }
666