savecore.c revision 47095
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 const 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#if 0 42static char sccsid[] = "@(#)savecore.c 8.3 (Berkeley) 1/2/94"; 43#endif 44static const char rcsid[] = 45 "$Id: savecore.c,v 1.22 1999/03/12 14:46:00 gallatin Exp $"; 46#endif /* not lint */ 47 48#include <sys/param.h> 49#include <sys/stat.h> 50#include <sys/mount.h> 51#include <sys/syslog.h> 52#include <sys/sysctl.h> 53 54#include <vm/vm.h> 55#include <vm/vm_param.h> 56#include <vm/pmap.h> 57 58#include <dirent.h> 59#include <errno.h> 60#include <fcntl.h> 61#include <nlist.h> 62#include <paths.h> 63#include <stdio.h> 64#include <stdlib.h> 65#include <string.h> 66#include <unistd.h> 67#include "zopen.h" 68 69#ifdef __alpha__ 70#define ok(number) ALPHA_K0SEG_TO_PHYS(number) 71#endif 72 73#ifdef __i386__ 74#define ok(number) ((number) - KERNBASE) 75#endif 76 77struct nlist current_nl[] = { /* Namelist for currently running system. */ 78#define X_DUMPLO 0 79 { "_dumplo" }, 80#define X_TIME 1 81 { "_time_second" }, 82#define X_DUMPSIZE 2 83 { "_dumpsize" }, 84#define X_VERSION 3 85 { "_version" }, 86#define X_PANICSTR 4 87 { "_panicstr" }, 88#define X_DUMPMAG 5 89 { "_dumpmag" }, 90 { "" }, 91}; 92int cursyms[] = { X_DUMPLO, X_VERSION, X_DUMPMAG, -1 }; 93int dumpsyms[] = { X_TIME, X_DUMPSIZE, X_VERSION, X_PANICSTR, X_DUMPMAG, -1 }; 94 95struct nlist dump_nl[] = { /* Name list for dumped system. */ 96 { "_dumplo" }, /* Entries MUST be the same as */ 97 { "_time_second" }, /* those in current_nl[]. */ 98 { "_dumpsize" }, 99 { "_version" }, 100 { "_panicstr" }, 101 { "_dumpmag" }, 102 { "" }, 103}; 104 105/* Types match kernel declarations. */ 106long dumplo; /* where dump starts on dumpdev */ 107int dumpmag; /* magic number in dump */ 108int dumpsize; /* amount of memory dumped */ 109 110char *kernel; 111char *dirname; /* directory to save dumps in */ 112char *ddname; /* name of dump device */ 113dev_t dumpdev; /* dump device */ 114int dumpfd; /* read/write descriptor on block dev */ 115time_t now; /* current date */ 116char panic_mesg[1024]; 117int panicstr; 118char vers[1024]; 119 120int clear, compress, force, verbose; /* flags */ 121 122void check_kmem __P((void)); 123int check_space __P((void)); 124void clear_dump __P((void)); 125int Create __P((char *, int)); 126int dump_exists __P((void)); 127char *find_dev __P((dev_t, int)); 128int get_crashtime __P((void)); 129void get_dumpsize __P((void)); 130void kmem_setup __P((void)); 131void log __P((int, char *, ...)); 132void Lseek __P((int, off_t, int)); 133int Open __P((const char *, int rw)); 134int Read __P((int, void *, int)); 135char *rawname __P((char *s)); 136void save_core __P((void)); 137void usage __P((void)); 138void Write __P((int, void *, int)); 139 140int 141main(argc, argv) 142 int argc; 143 char *argv[]; 144{ 145 int ch; 146 147 openlog("savecore", LOG_PERROR, LOG_DAEMON); 148 149 while ((ch = getopt(argc, argv, "cdfN:vz")) != -1) 150 switch(ch) { 151 case 'c': 152 clear = 1; 153 break; 154 case 'd': /* Not documented. */ 155 case 'v': 156 verbose = 1; 157 break; 158 case 'f': 159 force = 1; 160 break; 161 case 'N': 162 kernel = optarg; 163 break; 164 case 'z': 165 compress = 1; 166 break; 167 case '?': 168 default: 169 usage(); 170 } 171 argc -= optind; 172 argv += optind; 173 174 if (!clear) { 175 if (argc != 1 && argc != 2) 176 usage(); 177 dirname = argv[0]; 178 } 179 if (argc == 2) 180 kernel = argv[1]; 181 182 (void)time(&now); 183 kmem_setup(); 184 185 if (clear) { 186 clear_dump(); 187 exit(0); 188 } 189 190 if (!dump_exists() && !force) 191 exit(1); 192 193 check_kmem(); 194 195 if (panicstr) 196 syslog(LOG_ALERT, "reboot after panic: %s", panic_mesg); 197 else 198 syslog(LOG_ALERT, "reboot"); 199 200 get_dumpsize(); 201 202 if ((!get_crashtime() || !check_space()) && !force) 203 exit(1); 204 205 save_core(); 206 207 clear_dump(); 208 exit(0); 209} 210 211void 212kmem_setup() 213{ 214 FILE *fp; 215 int kmem, i; 216 const char *dump_sys; 217 int mib[2]; 218 size_t len; 219 220 /* 221 * Some names we need for the currently running system, others for 222 * the system that was running when the dump was made. The values 223 * obtained from the current system are used to look for things in 224 * /dev/kmem that cannot be found in the dump_sys namelist, but are 225 * presumed to be the same (since the disk partitions are probably 226 * the same!) 227 */ 228 if ((nlist(getbootfile(), current_nl)) == -1) 229 syslog(LOG_ERR, "%s: nlist: %s", getbootfile(), 230 strerror(errno)); 231 for (i = 0; cursyms[i] != -1; i++) 232 if (current_nl[cursyms[i]].n_value == 0) { 233 syslog(LOG_ERR, "%s: %s not in namelist", 234 getbootfile(), current_nl[cursyms[i]].n_name); 235 exit(1); 236 } 237 238 dump_sys = kernel ? kernel : getbootfile(); 239 if ((nlist(dump_sys, dump_nl)) == -1) 240 syslog(LOG_ERR, "%s: nlist: %s", dump_sys, strerror(errno)); 241 for (i = 0; dumpsyms[i] != -1; i++) 242 if (dump_nl[dumpsyms[i]].n_value == 0) { 243 syslog(LOG_ERR, "%s: %s not in namelist", 244 dump_sys, dump_nl[dumpsyms[i]].n_name); 245 exit(1); 246 } 247 248 mib[0] = CTL_KERN; 249 mib[1] = KERN_DUMPDEV; 250 len = sizeof dumpdev; 251 if (sysctl(mib, 2, &dumpdev, &len, NULL, 0) == -1) { 252 syslog(LOG_ERR, "sysctl: kern.dumpdev: %m"); 253 exit(1); 254 } 255 if (dumpdev == NODEV) { 256 syslog(LOG_WARNING, "no core dump (no dumpdev)"); 257 exit(1); 258 } 259 260 kmem = Open(_PATH_KMEM, O_RDONLY); 261 Lseek(kmem, (off_t)current_nl[X_DUMPLO].n_value, L_SET); 262 (void)Read(kmem, &dumplo, sizeof(dumplo)); 263 if (verbose) 264 (void)printf("dumplo = %ld (%ld * %d)\n", 265 dumplo, dumplo/DEV_BSIZE, DEV_BSIZE); 266 Lseek(kmem, (off_t)current_nl[X_DUMPMAG].n_value, L_SET); 267 (void)Read(kmem, &dumpmag, sizeof(dumpmag)); 268 dumplo *= DEV_BSIZE; 269 ddname = find_dev(dumpdev, S_IFBLK); 270 dumpfd = Open(ddname, O_RDWR); 271 fp = fdopen(kmem, "r"); 272 if (fp == NULL) { 273 syslog(LOG_ERR, "%s: fdopen: %m", _PATH_KMEM); 274 exit(1); 275 } 276 if (kernel) 277 return; 278 (void)fseek(fp, (off_t)current_nl[X_VERSION].n_value, L_SET); 279 (void)fgets(vers, sizeof(vers), fp); 280 281 /* Don't fclose(fp), we use dumpfd later. */ 282} 283 284void 285check_kmem() 286{ 287 register char *cp; 288 FILE *fp; 289 char core_vers[1024]; 290 291 fp = fdopen(dumpfd, "r"); 292 if (fp == NULL) { 293 syslog(LOG_ERR, "%s: fdopen: %m", ddname); 294 exit(1); 295 } 296 fseek(fp, (off_t)(dumplo + ok(dump_nl[X_VERSION].n_value)), L_SET); 297 fgets(core_vers, sizeof(core_vers), fp); 298 if (strcmp(vers, core_vers) && kernel == 0) 299 syslog(LOG_WARNING, 300 "warning: %s version mismatch:\n\t%s\nand\t%s\n", 301 getbootfile(), vers, core_vers); 302 (void)fseek(fp, 303 (off_t)(dumplo + ok(dump_nl[X_PANICSTR].n_value)), L_SET); 304 (void)fread(&panicstr, sizeof(panicstr), 1, fp); 305 if (panicstr) { 306 (void)fseek(fp, dumplo + ok(panicstr), L_SET); 307 cp = panic_mesg; 308 do 309 *cp = getc(fp); 310 while (*cp++ && cp < &panic_mesg[sizeof(panic_mesg)]); 311 } 312 /* Don't fclose(fp), we use dumpfd later. */ 313} 314 315void 316clear_dump() 317{ 318 long newdumplo; 319 320 newdumplo = 0; 321 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_DUMPMAG].n_value)), L_SET); 322 Write(dumpfd, &newdumplo, sizeof(newdumplo)); 323} 324 325int 326dump_exists() 327{ 328 int newdumpmag; 329 330 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_DUMPMAG].n_value)), L_SET); 331 (void)Read(dumpfd, &newdumpmag, sizeof(newdumpmag)); 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 342char buf[1024 * 1024]; 343 344void 345save_core() 346{ 347 register FILE *fp; 348 register int bounds, ifd, nr, nw, ofd; 349 char *rawp, path[MAXPATHLEN]; 350 mode_t oumask; 351 352 /* 353 * Get the current number and update the bounds file. Do the update 354 * now, because may fail later and don't want to overwrite anything. 355 */ 356 (void)snprintf(path, sizeof(path), "%s/bounds", dirname); 357 if ((fp = fopen(path, "r")) == NULL) 358 goto err1; 359 if (fgets(buf, sizeof(buf), fp) == NULL) { 360 if (ferror(fp)) 361err1: syslog(LOG_WARNING, "%s: %s", path, strerror(errno)); 362 bounds = 0; 363 } else 364 bounds = atoi(buf); 365 if (fp != NULL) 366 (void)fclose(fp); 367 if ((fp = fopen(path, "w")) == NULL) 368 syslog(LOG_ERR, "%s: %m", path); 369 else { 370 (void)fprintf(fp, "%d\n", bounds + 1); 371 (void)fclose(fp); 372 } 373 374 /* Create the core file. */ 375 oumask = umask(S_IRWXG|S_IRWXO); /* Restrict access to the core file.*/ 376 (void)snprintf(path, sizeof(path), "%s/vmcore.%d%s", 377 dirname, bounds, compress ? ".Z" : ""); 378 if (compress) { 379 if ((fp = zopen(path, "w", 0)) == NULL) { 380 syslog(LOG_ERR, "%s: %s", path, strerror(errno)); 381 exit(1); 382 } 383 } else 384 ofd = Create(path, S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH); 385 (void)umask(oumask); 386 387 /* Open the raw device. */ 388 rawp = rawname(ddname); 389 if ((ifd = open(rawp, O_RDONLY)) == -1) { 390 syslog(LOG_WARNING, "%s: %m; using block device", rawp); 391 ifd = dumpfd; 392 } 393 394 /* Seek to the start of the core. */ 395 Lseek(ifd, (off_t)dumplo, L_SET); 396 397 /* Copy the core file. */ 398 syslog(LOG_NOTICE, "writing %score to %s", 399 compress ? "compressed " : "", path); 400 for (; dumpsize > 0; dumpsize -= nr) { 401 (void)printf("%6dK\r", dumpsize / 1024); 402 (void)fflush(stdout); 403 nr = read(ifd, buf, MIN(dumpsize, sizeof(buf))); 404 if (nr <= 0) { 405 if (nr == 0) 406 syslog(LOG_WARNING, 407 "WARNING: EOF on dump device"); 408 else 409 syslog(LOG_ERR, "%s: %m", rawp); 410 goto err2; 411 } 412 if (compress) 413 nw = fwrite(buf, 1, nr, fp); 414 else 415 nw = write(ofd, buf, nr); 416 if (nw != nr) { 417 syslog(LOG_ERR, "%s: %s", 418 path, strerror(nw == 0 ? EIO : errno)); 419err2: syslog(LOG_WARNING, 420 "WARNING: vmcore may be incomplete"); 421 (void)printf("\n"); 422 exit(1); 423 } 424 } 425 (void)close(ifd); 426 if (compress) 427 (void)fclose(fp); 428 else 429 (void)close(ofd); 430 431 /* Copy the kernel. */ 432 ifd = Open(kernel ? kernel : getbootfile(), O_RDONLY); 433 (void)snprintf(path, sizeof(path), "%s/kernel.%d%s", 434 dirname, bounds, compress ? ".Z" : ""); 435 if (compress) { 436 if ((fp = zopen(path, "w", 0)) == NULL) { 437 syslog(LOG_ERR, "%s: %s", path, strerror(errno)); 438 exit(1); 439 } 440 } else 441 ofd = Create(path, S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH); 442 syslog(LOG_NOTICE, "writing %skernel to %s", 443 compress ? "compressed " : "", path); 444 while ((nr = read(ifd, buf, sizeof(buf))) > 0) { 445 if (compress) 446 nw = fwrite(buf, 1, nr, fp); 447 else 448 nw = write(ofd, buf, nr); 449 if (nw != nr) { 450 syslog(LOG_ERR, "%s: %s", 451 path, strerror(nw == 0 ? EIO : errno)); 452 syslog(LOG_WARNING, 453 "WARNING: kernel may be incomplete"); 454 exit(1); 455 } 456 } 457 if (nr < 0) { 458 syslog(LOG_ERR, "%s: %s", 459 kernel ? kernel : getbootfile(), strerror(errno)); 460 syslog(LOG_WARNING, 461 "WARNING: kernel may be incomplete"); 462 exit(1); 463 } 464 if (compress) 465 (void)fclose(fp); 466 else 467 (void)close(ofd); 468} 469 470char * 471find_dev(dev, type) 472 register dev_t dev; 473 register int type; 474{ 475 register DIR *dfd; 476 struct dirent *dir; 477 struct stat sb; 478 char *dp, devname[MAXPATHLEN + 1]; 479 480 if ((dfd = opendir(_PATH_DEV)) == NULL) { 481 syslog(LOG_ERR, "%s: %s", _PATH_DEV, strerror(errno)); 482 exit(1); 483 } 484 (void)strcpy(devname, _PATH_DEV); 485 while ((dir = readdir(dfd))) { 486 (void)strcpy(devname + sizeof(_PATH_DEV) - 1, dir->d_name); 487 if (lstat(devname, &sb)) { 488 syslog(LOG_ERR, "%s: %s", devname, strerror(errno)); 489 continue; 490 } 491 if ((sb.st_mode & S_IFMT) != type) 492 continue; 493 if (dev == sb.st_rdev) { 494 closedir(dfd); 495 if ((dp = strdup(devname)) == NULL) { 496 syslog(LOG_ERR, "%s", strerror(errno)); 497 exit(1); 498 } 499 return (dp); 500 } 501 } 502 closedir(dfd); 503 syslog(LOG_ERR, "can't find device %d/%d", major(dev), minor(dev)); 504 exit(1); 505} 506 507char * 508rawname(s) 509 char *s; 510{ 511 char *sl, name[MAXPATHLEN]; 512 513 if ((sl = rindex(s, '/')) == NULL || sl[1] == '0') { 514 syslog(LOG_ERR, 515 "can't make raw dump device name from %s", s); 516 return (s); 517 } 518 (void)snprintf(name, sizeof(name), "%.*s/r%s", sl - s, s, sl + 1); 519 if ((sl = strdup(name)) == NULL) { 520 syslog(LOG_ERR, "%s", strerror(errno)); 521 exit(1); 522 } 523 return (sl); 524} 525 526int 527get_crashtime() 528{ 529 time_t dumptime; /* Time the dump was taken. */ 530 531 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_TIME].n_value)), L_SET); 532 (void)Read(dumpfd, &dumptime, sizeof(dumptime)); 533 if (dumptime == 0) { 534 if (verbose) 535 syslog(LOG_ERR, "dump time is zero"); 536 return (0); 537 } 538 (void)printf("savecore: system went down at %s", ctime(&dumptime)); 539#define LEEWAY (7 * 86400) 540 if (dumptime < now - LEEWAY || dumptime > now + LEEWAY) { 541 (void)printf("dump time is unreasonable\n"); 542 return (0); 543 } 544 return (1); 545} 546 547void 548get_dumpsize() 549{ 550 /* Read the dump size. */ 551 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_DUMPSIZE].n_value)), L_SET); 552 (void)Read(dumpfd, &dumpsize, sizeof(dumpsize)); 553 dumpsize *= getpagesize(); 554} 555 556int 557check_space() 558{ 559 register FILE *fp; 560 const char *tkernel; 561 off_t minfree, spacefree, totfree, kernelsize, needed; 562 struct stat st; 563 struct statfs fsbuf; 564 char buf[100], path[MAXPATHLEN]; 565 566 tkernel = kernel ? kernel : getbootfile(); 567 if (stat(tkernel, &st) < 0) { 568 syslog(LOG_ERR, "%s: %m", tkernel); 569 exit(1); 570 } 571 kernelsize = st.st_blocks * S_BLKSIZE; 572 573 if (statfs(dirname, &fsbuf) < 0) { 574 syslog(LOG_ERR, "%s: %m", dirname); 575 exit(1); 576 } 577 spacefree = ((off_t) fsbuf.f_bavail * fsbuf.f_bsize) / 1024; 578 totfree = ((off_t) fsbuf.f_bfree * fsbuf.f_bsize) / 1024; 579 580 (void)snprintf(path, sizeof(path), "%s/minfree", dirname); 581 if ((fp = fopen(path, "r")) == NULL) 582 minfree = 0; 583 else { 584 if (fgets(buf, sizeof(buf), fp) == NULL) 585 minfree = 0; 586 else 587 minfree = atoi(buf); 588 (void)fclose(fp); 589 } 590 591 needed = (dumpsize + kernelsize) / 1024; 592 if (((minfree > 0) ? spacefree : totfree) - needed < minfree) { 593 syslog(LOG_WARNING, 594 "no dump, not enough free space on device"); 595 return (0); 596 } 597 if (spacefree - needed < 0) 598 syslog(LOG_WARNING, 599 "dump performed, but free space threshold crossed"); 600 return (1); 601} 602 603int 604Open(name, rw) 605 const char *name; 606 int rw; 607{ 608 int fd; 609 610 if ((fd = open(name, rw, 0)) < 0) { 611 syslog(LOG_ERR, "%s: %m", name); 612 exit(1); 613 } 614 return (fd); 615} 616 617int 618Read(fd, bp, size) 619 int fd, size; 620 void *bp; 621{ 622 int nr; 623 624 nr = read(fd, bp, size); 625 if (nr != size) { 626 syslog(LOG_ERR, "read: %m"); 627 exit(1); 628 } 629 return (nr); 630} 631 632void 633Lseek(fd, off, flag) 634 int fd, flag; 635 off_t off; 636{ 637 off_t ret; 638 639 ret = lseek(fd, off, flag); 640 if (ret == -1) { 641 syslog(LOG_ERR, "lseek: %m"); 642 exit(1); 643 } 644} 645 646int 647Create(file, mode) 648 char *file; 649 int mode; 650{ 651 register int fd; 652 653 fd = creat(file, mode); 654 if (fd < 0) { 655 syslog(LOG_ERR, "%s: %m", file); 656 exit(1); 657 } 658 return (fd); 659} 660 661void 662Write(fd, bp, size) 663 int fd, size; 664 void *bp; 665{ 666 int n; 667 668 if ((n = write(fd, bp, size)) < size) { 669 syslog(LOG_ERR, "write: %s", strerror(n == -1 ? errno : EIO)); 670 exit(1); 671 } 672} 673 674void 675usage() 676{ 677 (void)syslog(LOG_ERR, "usage: savecore [-cfvz] [-N system] directory"); 678 exit(1); 679} 680