savecore.c revision 1.17
1/*- 2 * Copyright (c) 1986, 1992, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 */ 33 34#ifndef lint 35static char copyright[] = 36"@(#) Copyright (c) 1986, 1992, 1993\n\ 37 The Regents of the University of California. All rights reserved.\n"; 38#endif /* not lint */ 39 40#ifndef lint 41/*static char sccsid[] = "from: @(#)savecore.c 8.3 (Berkeley) 1/2/94";*/ 42static char *rcsid = "$Id: savecore.c,v 1.17 1994/09/23 14:28:15 mycroft Exp $"; 43#endif /* not lint */ 44 45#include <sys/param.h> 46#include <sys/stat.h> 47#include <sys/mount.h> 48#include <sys/syslog.h> 49#include <sys/time.h> 50 51#include <dirent.h> 52#include <errno.h> 53#include <fcntl.h> 54#include <nlist.h> 55#include <paths.h> 56#include <stdio.h> 57#include <stdlib.h> 58#include <string.h> 59#include <tzfile.h> 60#include <unistd.h> 61 62#define ok(number) ((number) - KERNBASE) 63 64struct nlist current_nl[] = { /* Namelist for currently running system. */ 65#define X_DUMPDEV 0 66 { "_dumpdev" }, 67#define X_DUMPLO 1 68 { "_dumplo" }, 69#define X_TIME 2 70 { "_time" }, 71#define X_DUMPSIZE 3 72 { "_dumpsize" }, 73#define X_VERSION 4 74 { "_version" }, 75#define X_PANICSTR 5 76 { "_panicstr" }, 77#define X_DUMPMAG 6 78 { "_dumpmag" }, 79 { "" }, 80}; 81int cursyms[] = { X_DUMPDEV, X_DUMPLO, X_VERSION, X_DUMPMAG, -1 }; 82int dumpsyms[] = { X_TIME, X_DUMPSIZE, X_VERSION, X_PANICSTR, X_DUMPMAG, -1 }; 83 84struct nlist dump_nl[] = { /* Name list for dumped system. */ 85 { "_dumpdev" }, /* Entries MUST be the same as */ 86 { "_dumplo" }, /* those in current_nl[]. */ 87 { "_time" }, 88 { "_dumpsize" }, 89 { "_version" }, 90 { "_panicstr" }, 91 { "_dumpmag" }, 92 { "" }, 93}; 94 95/* Types match kernel declarations. */ 96long dumplo; /* where dump starts on dumpdev */ 97int dumpmag; /* magic number in dump */ 98int dumpsize; /* amount of memory dumped */ 99 100char *kernel; 101char *dirname; /* directory to save dumps in */ 102char *ddname; /* name of dump device */ 103dev_t dumpdev; /* dump device */ 104int dumpfd; /* read/write descriptor on block dev */ 105time_t now; /* current date */ 106char panic_mesg[1024]; 107int panicstr; 108char vers[1024]; 109 110int clear, compress, force, verbose; /* flags */ 111 112void check_kmem __P((void)); 113int check_space __P((void)); 114void clear_dump __P((void)); 115int Create __P((char *, int)); 116int dump_exists __P((void)); 117char *find_dev __P((dev_t, int)); 118int get_crashtime __P((void)); 119void kmem_setup __P((void)); 120void log __P((int, char *, ...)); 121void Lseek __P((int, off_t, int)); 122int Open __P((char *, int rw)); 123int Read __P((int, void *, int)); 124char *rawname __P((char *s)); 125void save_core __P((void)); 126void usage __P((void)); 127void Write __P((int, void *, int)); 128 129int 130main(argc, argv) 131 int argc; 132 char *argv[]; 133{ 134 int ch; 135 136 openlog("savecore", LOG_PERROR, LOG_DAEMON); 137 138 while ((ch = getopt(argc, argv, "cdfNvz")) != EOF) 139 switch(ch) { 140 case 'c': 141 clear = 1; 142 break; 143 case 'd': /* Not documented. */ 144 case 'v': 145 verbose = 1; 146 break; 147 case 'f': 148 force = 1; 149 break; 150 case 'N': 151 kernel = optarg; 152 break; 153 case 'z': 154 compress = 1; 155 break; 156 case '?': 157 default: 158 usage(); 159 } 160 argc -= optind; 161 argv += optind; 162 163 if (!clear) { 164 if (argc != 1 && argc != 2) 165 usage(); 166 dirname = argv[0]; 167 } 168 if (argc == 2) 169 kernel = argv[1]; 170 171 (void)time(&now); 172 kmem_setup(); 173 174 if (clear) { 175 clear_dump(); 176 exit(0); 177 } 178 179 if (!dump_exists() && !force) 180 exit(1); 181 182 check_kmem(); 183 184 if (panicstr) 185 syslog(LOG_ALERT, "reboot after panic: %s", panic_mesg); 186 else 187 syslog(LOG_ALERT, "reboot"); 188 189 if ((!get_crashtime() || !check_space()) && !force) 190 exit(1); 191 192 save_core(); 193 194 clear_dump(); 195 exit(0); 196} 197 198void 199kmem_setup() 200{ 201 FILE *fp; 202 int kmem, i; 203 char *dump_sys; 204 205 /* 206 * Some names we need for the currently running system, others for 207 * the system that was running when the dump was made. The values 208 * obtained from the current system are used to look for things in 209 * /dev/kmem that cannot be found in the dump_sys namelist, but are 210 * presumed to be the same (since the disk partitions are probably 211 * the same!) 212 */ 213 if ((nlist(_PATH_UNIX, current_nl)) == -1) 214 syslog(LOG_ERR, "%s: nlist: %s", _PATH_UNIX, strerror(errno)); 215 for (i = 0; cursyms[i] != -1; i++) 216 if (current_nl[cursyms[i]].n_value == 0) { 217 syslog(LOG_ERR, "%s: %s not in namelist", 218 _PATH_UNIX, current_nl[cursyms[i]].n_name); 219 exit(1); 220 } 221 222 dump_sys = kernel ? kernel : _PATH_UNIX; 223 if ((nlist(dump_sys, dump_nl)) == -1) 224 syslog(LOG_ERR, "%s: nlist: %s", dump_sys, strerror(errno)); 225 for (i = 0; dumpsyms[i] != -1; i++) 226 if (dump_nl[dumpsyms[i]].n_value == 0) { 227 syslog(LOG_ERR, "%s: %s not in namelist", 228 dump_sys, dump_nl[dumpsyms[i]].n_name); 229 exit(1); 230 } 231 232 kmem = Open(_PATH_KMEM, O_RDONLY); 233 Lseek(kmem, (off_t)current_nl[X_DUMPDEV].n_value, L_SET); 234 (void)Read(kmem, &dumpdev, sizeof(dumpdev)); 235 if (dumpdev == NODEV) { 236 syslog(LOG_WARNING, "no core dump (no dumpdev)"); 237 exit(1); 238 } 239 Lseek(kmem, (off_t)current_nl[X_DUMPLO].n_value, L_SET); 240 (void)Read(kmem, &dumplo, sizeof(dumplo)); 241 if (verbose) 242 (void)printf("dumplo = %d (%d * %d)\n", 243 dumplo, dumplo/DEV_BSIZE, DEV_BSIZE); 244 Lseek(kmem, (off_t)current_nl[X_DUMPMAG].n_value, L_SET); 245 (void)Read(kmem, &dumpmag, sizeof(dumpmag)); 246 dumplo *= DEV_BSIZE; 247 ddname = find_dev(dumpdev, S_IFBLK); 248 dumpfd = Open(ddname, O_RDWR); 249 fp = fdopen(kmem, "r"); 250 if (fp == NULL) { 251 syslog(LOG_ERR, "%s: fdopen: %m", _PATH_KMEM); 252 exit(1); 253 } 254 if (kernel) 255 return; 256 (void)fseek(fp, (off_t)current_nl[X_VERSION].n_value, L_SET); 257 (void)fgets(vers, sizeof(vers), fp); 258 259 /* Don't fclose(fp), we use dumpfd later. */ 260} 261 262void 263check_kmem() 264{ 265 register char *cp; 266 FILE *fp; 267 char core_vers[1024]; 268 269 fp = fdopen(dumpfd, "r"); 270 if (fp == NULL) { 271 syslog(LOG_ERR, "%s: fdopen: %m", ddname); 272 exit(1); 273 } 274 fseek(fp, (off_t)(dumplo + ok(dump_nl[X_VERSION].n_value)), L_SET); 275 fgets(core_vers, sizeof(core_vers), fp); 276 if (strcmp(vers, core_vers) && kernel == 0) 277 syslog(LOG_WARNING, 278 "warning: %s version mismatch:\n\t%s\nand\t%s\n", 279 _PATH_UNIX, vers, core_vers); 280 (void)fseek(fp, 281 (off_t)(dumplo + ok(dump_nl[X_PANICSTR].n_value)), L_SET); 282 (void)fread(&panicstr, sizeof(panicstr), 1, fp); 283 if (panicstr) { 284 (void)fseek(fp, dumplo + ok(panicstr), L_SET); 285 cp = panic_mesg; 286 do 287 *cp = getc(fp); 288 while (*cp++ && cp < &panic_mesg[sizeof(panic_mesg)]); 289 } 290 /* Don't fclose(fp), we use dumpfd later. */ 291} 292 293void 294clear_dump() 295{ 296 long newdumplo; 297 298 newdumplo = 0; 299 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_DUMPMAG].n_value)), L_SET); 300 Write(dumpfd, &newdumplo, sizeof(newdumplo)); 301} 302 303int 304dump_exists() 305{ 306 int newdumpmag; 307 308 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_DUMPMAG].n_value)), L_SET); 309 (void)Read(dumpfd, &newdumpmag, sizeof(newdumpmag)); 310 if (newdumpmag != dumpmag) { 311 if (verbose) 312 syslog(LOG_WARNING, "magic number mismatch (%x != %x)", 313 newdumpmag, dumpmag); 314 syslog(LOG_WARNING, "no core dump"); 315 return (0); 316 } 317 return (1); 318} 319 320char buf[1024 * 1024]; 321 322FILE *zopen __P((char *, char *, int)); 323 324void 325save_core() 326{ 327 register FILE *fp; 328 register int bounds, ifd, nr, nw, ofd; 329 char *rawp, path[MAXPATHLEN]; 330 331 /* 332 * Get the current number and update the bounds file. Do the update 333 * now, because may fail later and don't want to overwrite anything. 334 */ 335 (void)snprintf(path, sizeof(path), "%s/bounds", dirname); 336 if ((fp = fopen(path, "r")) == NULL) 337 goto err1; 338 if (fgets(buf, sizeof(buf), fp) == NULL) { 339 if (ferror(fp)) 340err1: syslog(LOG_WARNING, "%s: %s", path, strerror(errno)); 341 bounds = 0; 342 } else 343 bounds = atoi(buf); 344 if (fp != NULL) 345 (void)fclose(fp); 346 if ((fp = fopen(path, "w")) == NULL) 347 syslog(LOG_ERR, "%s: %m", path); 348 else { 349 (void)fprintf(fp, "%d\n", bounds + 1); 350 (void)fclose(fp); 351 } 352 (void)fclose(fp); 353 354 /* Create the core file. */ 355 (void)snprintf(path, sizeof(path), "%s/netbsd.%d%s.core", 356 dirname, bounds, compress ? ".Z" : ""); 357 if (compress) { 358 if ((fp = zopen(path, "w", 0)) == NULL) { 359 syslog(LOG_ERR, "%s: %s", path, strerror(errno)); 360 exit(1); 361 } 362 } else 363 ofd = Create(path, S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH); 364 365 /* Open the raw device. */ 366 rawp = rawname(ddname); 367 if ((ifd = open(rawp, O_RDONLY)) == -1) { 368 syslog(LOG_WARNING, "%s: %m; using block device", rawp); 369 ifd = dumpfd; 370 } 371 372 /* Read the dump size. */ 373 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_DUMPSIZE].n_value)), L_SET); 374 (void)Read(dumpfd, &dumpsize, sizeof(dumpsize)); 375 376 /* Seek to the start of the core. */ 377 Lseek(ifd, (off_t)dumplo, L_SET); 378 379 /* Copy the core file. */ 380 dumpsize *= getpagesize(); 381 syslog(LOG_NOTICE, "writing %score to %s", 382 compress ? "compressed " : "", path); 383 for (; dumpsize > 0; dumpsize -= nr) { 384 (void)printf("%6dK\r", dumpsize / 1024); 385 (void)fflush(stdout); 386 nr = read(ifd, buf, MIN(dumpsize, sizeof(buf))); 387 if (nr <= 0) { 388 if (nr == 0) 389 syslog(LOG_WARNING, 390 "WARNING: EOF on dump device"); 391 else 392 syslog(LOG_ERR, "%s: %m", rawp); 393 goto err2; 394 } 395 if (compress) 396 nw = fwrite(buf, 1, nr, fp); 397 else 398 nw = write(ofd, buf, nr); 399 if (nw != nr) { 400 syslog(LOG_ERR, "%s: %s", 401 path, strerror(nw == 0 ? EIO : errno)); 402err2: syslog(LOG_WARNING, 403 "WARNING: core may be incomplete"); 404 (void)printf("\n"); 405 exit(1); 406 } 407 } 408 (void)printf("\n"); 409 (void)close(ifd); 410 if (compress) 411 (void)fclose(fp); 412 else 413 (void)close(ofd); 414 415 /* Copy the kernel. */ 416 ifd = Open(kernel ? kernel : _PATH_UNIX, O_RDONLY); 417 (void)snprintf(path, sizeof(path), "%s/netbsd.%d%s", 418 dirname, bounds, compress ? ".Z" : ""); 419 if (compress) { 420 if ((fp = zopen(path, "w", 0)) == NULL) { 421 syslog(LOG_ERR, "%s: %s", path, strerror(errno)); 422 exit(1); 423 } 424 } else 425 ofd = Create(path, S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH); 426 syslog(LOG_NOTICE, "writing %skernel to %s", 427 compress ? "compressed " : "", path); 428 while ((nr = read(ifd, buf, sizeof(buf))) > 0) { 429 if (compress) 430 nw = fwrite(buf, 1, nr, fp); 431 else 432 nw = write(ofd, buf, nr); 433 if (nw != nr) { 434 syslog(LOG_ERR, "%s: %s", 435 path, strerror(nw == 0 ? EIO : errno)); 436 syslog(LOG_WARNING, 437 "WARNING: kernel may be incomplete"); 438 exit(1); 439 } 440 } 441 if (nr < 0) { 442 syslog(LOG_ERR, "%s: %s", 443 kernel ? kernel : _PATH_UNIX, strerror(errno)); 444 syslog(LOG_WARNING, 445 "WARNING: kernel may be incomplete"); 446 exit(1); 447 } 448 if (compress) 449 (void)fclose(fp); 450 else 451 (void)close(ofd); 452} 453 454char * 455find_dev(dev, type) 456 register dev_t dev; 457 register int type; 458{ 459 register DIR *dfd; 460 struct dirent *dir; 461 struct stat sb; 462 char *dp, devname[MAXPATHLEN + 1]; 463 464 if ((dfd = opendir(_PATH_DEV)) == NULL) { 465 syslog(LOG_ERR, "%s: %s", _PATH_DEV, strerror(errno)); 466 exit(1); 467 } 468 (void)strcpy(devname, _PATH_DEV); 469 while ((dir = readdir(dfd))) { 470 (void)strcpy(devname + sizeof(_PATH_DEV) - 1, dir->d_name); 471 if (lstat(devname, &sb)) { 472 syslog(LOG_ERR, "%s: %s", devname, strerror(errno)); 473 continue; 474 } 475 if ((sb.st_mode & S_IFMT) != type) 476 continue; 477 if (dev == sb.st_rdev) { 478 closedir(dfd); 479 if ((dp = strdup(devname)) == NULL) { 480 syslog(LOG_ERR, "%s", strerror(errno)); 481 exit(1); 482 } 483 return (dp); 484 } 485 } 486 closedir(dfd); 487 syslog(LOG_ERR, "can't find device %d/%d", major(dev), minor(dev)); 488 exit(1); 489} 490 491char * 492rawname(s) 493 char *s; 494{ 495 char *sl, name[MAXPATHLEN]; 496 497 if ((sl = strrchr(s, '/')) == NULL || sl[1] == '0') { 498 syslog(LOG_ERR, 499 "can't make raw dump device name from %s", s); 500 return (s); 501 } 502 (void)snprintf(name, sizeof(name), "%.*s/r%s", sl - s, s, sl + 1); 503 if ((sl = strdup(name)) == NULL) { 504 syslog(LOG_ERR, "%s", strerror(errno)); 505 exit(1); 506 } 507 return (sl); 508} 509 510int 511get_crashtime() 512{ 513 time_t dumptime; /* Time the dump was taken. */ 514 515 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_TIME].n_value)), L_SET); 516 (void)Read(dumpfd, &dumptime, sizeof(dumptime)); 517 if (dumptime == 0) { 518 if (verbose) 519 syslog(LOG_ERR, "dump time is zero"); 520 return (0); 521 } 522 (void)printf("savecore: system went down at %s", ctime(&dumptime)); 523#define LEEWAY (7 * SECSPERDAY) 524 if (dumptime < now - LEEWAY || dumptime > now + LEEWAY) { 525 (void)printf("dump time is unreasonable\n"); 526 return (0); 527 } 528 return (1); 529} 530 531int 532check_space() 533{ 534 register FILE *fp; 535 char *tkernel; 536 off_t minfree, spacefree, kernelsize, needed; 537 struct stat st; 538 struct statfs fsbuf; 539 char buf[100], path[MAXPATHLEN]; 540 541 tkernel = kernel ? kernel : _PATH_UNIX; 542 if (stat(tkernel, &st) < 0) { 543 syslog(LOG_ERR, "%s: %m", tkernel); 544 exit(1); 545 } 546 kernelsize = st.st_blocks * S_BLKSIZE; 547 if (statfs(dirname, &fsbuf) < 0) { 548 syslog(LOG_ERR, "%s: %m", dirname); 549 exit(1); 550 } 551 spacefree = (fsbuf.f_bavail * fsbuf.f_bsize) / 1024; 552 553 (void)snprintf(path, sizeof(path), "%s/minfree", dirname); 554 if ((fp = fopen(path, "r")) == NULL) 555 minfree = 0; 556 else { 557 if (fgets(buf, sizeof(buf), fp) == NULL) 558 minfree = 0; 559 else 560 minfree = atoi(buf); 561 (void)fclose(fp); 562 } 563 564 needed = (dumpsize + kernelsize) / 1024; 565 if (minfree > 0 && spacefree - needed < minfree) { 566 syslog(LOG_WARNING, 567 "no dump, not enough free space on device"); 568 return (0); 569 } 570 if (spacefree - needed < minfree) 571 syslog(LOG_WARNING, 572 "dump performed, but free space threshold crossed"); 573 return (1); 574} 575 576int 577Open(name, rw) 578 char *name; 579 int rw; 580{ 581 int fd; 582 583 if ((fd = open(name, rw, 0)) < 0) { 584 syslog(LOG_ERR, "%s: %m", name); 585 exit(1); 586 } 587 return (fd); 588} 589 590int 591Read(fd, bp, size) 592 int fd, size; 593 void *bp; 594{ 595 int nr; 596 597 nr = read(fd, bp, size); 598 if (nr != size) { 599 syslog(LOG_ERR, "read: %m"); 600 exit(1); 601 } 602 return (nr); 603} 604 605void 606Lseek(fd, off, flag) 607 int fd, flag; 608 off_t off; 609{ 610 off_t ret; 611 612 ret = lseek(fd, off, flag); 613 if (ret == -1) { 614 syslog(LOG_ERR, "lseek: %m"); 615 exit(1); 616 } 617} 618 619int 620Create(file, mode) 621 char *file; 622 int mode; 623{ 624 register int fd; 625 626 fd = open(file, O_WRONLY | O_CREAT | O_TRUNC, mode); 627 if (fd < 0) { 628 syslog(LOG_ERR, "%s: %m", file); 629 exit(1); 630 } 631 return (fd); 632} 633 634void 635Write(fd, bp, size) 636 int fd, size; 637 void *bp; 638{ 639 int n; 640 641 if ((n = write(fd, bp, size)) < size) { 642 syslog(LOG_ERR, "write: %s", strerror(n == -1 ? errno : EIO)); 643 exit(1); 644 } 645} 646 647void 648usage() 649{ 650 (void)syslog(LOG_ERR, "usage: savecore [-cfvz] [-N system] directory"); 651 exit(1); 652} 653