savecore.c revision 18914
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 41static char sccsid[] = "@(#)savecore.c 8.3 (Berkeley) 1/2/94"; 42#endif /* not lint */ 43 44#include <sys/param.h> 45#include <sys/stat.h> 46#include <sys/mount.h> 47#include <sys/syslog.h> 48#include <sys/time.h> 49 50#include <vm/vm.h> 51#include <vm/vm_param.h> 52#include <vm/pmap.h> 53 54#include <dirent.h> 55#include <errno.h> 56#include <fcntl.h> 57#include <nlist.h> 58#include <paths.h> 59#include <stdio.h> 60#include <stdlib.h> 61#include <string.h> 62#include <unistd.h> 63#include "zopen.h" 64 65#define ok(number) ((number) - KERNBASE) 66 67struct nlist current_nl[] = { /* Namelist for currently running system. */ 68#define X_DUMPDEV 0 69 { "_dumpdev" }, 70#define X_DUMPLO 1 71 { "_dumplo" }, 72#define X_TIME 2 73 { "_time" }, 74#define X_DUMPSIZE 3 75 { "_dumpsize" }, 76#define X_VERSION 4 77 { "_version" }, 78#define X_PANICSTR 5 79 { "_panicstr" }, 80#define X_DUMPMAG 6 81 { "_dumpmag" }, 82 { "" }, 83}; 84int cursyms[] = { X_DUMPDEV, X_DUMPLO, X_VERSION, X_DUMPMAG, -1 }; 85int dumpsyms[] = { X_TIME, X_DUMPSIZE, X_VERSION, X_PANICSTR, X_DUMPMAG, -1 }; 86 87struct nlist dump_nl[] = { /* Name list for dumped system. */ 88 { "_dumpdev" }, /* Entries MUST be the same as */ 89 { "_dumplo" }, /* those in current_nl[]. */ 90 { "_time" }, 91 { "_dumpsize" }, 92 { "_version" }, 93 { "_panicstr" }, 94 { "_dumpmag" }, 95 { "" }, 96}; 97 98/* Types match kernel declarations. */ 99long dumplo; /* where dump starts on dumpdev */ 100int dumpmag; /* magic number in dump */ 101int dumpsize; /* amount of memory dumped */ 102 103char *kernel; 104char *dirname; /* directory to save dumps in */ 105char *ddname; /* name of dump device */ 106dev_t dumpdev; /* dump device */ 107int dumpfd; /* read/write descriptor on block dev */ 108time_t now; /* current date */ 109char panic_mesg[1024]; 110int panicstr; 111char vers[1024]; 112 113int clear, compress, force, verbose; /* flags */ 114 115void check_kmem __P((void)); 116int check_space __P((void)); 117void clear_dump __P((void)); 118int Create __P((char *, int)); 119int dump_exists __P((void)); 120char *find_dev __P((dev_t, int)); 121int get_crashtime __P((void)); 122int get_dumpsize __P((void)); 123void kmem_setup __P((void)); 124void log __P((int, char *, ...)); 125void Lseek __P((int, off_t, int)); 126int Open __P((const char *, int rw)); 127int Read __P((int, void *, int)); 128char *rawname __P((char *s)); 129void save_core __P((void)); 130void usage __P((void)); 131void Write __P((int, void *, int)); 132 133int 134main(argc, argv) 135 int argc; 136 char *argv[]; 137{ 138 int ch; 139 140 openlog("savecore", LOG_PERROR, LOG_DAEMON); 141 142 while ((ch = getopt(argc, argv, "cdfN:vz")) != EOF) 143 switch(ch) { 144 case 'c': 145 clear = 1; 146 break; 147 case 'd': /* Not documented. */ 148 case 'v': 149 verbose = 1; 150 break; 151 case 'f': 152 force = 1; 153 break; 154 case 'N': 155 kernel = optarg; 156 break; 157 case 'z': 158 compress = 1; 159 break; 160 case '?': 161 default: 162 usage(); 163 } 164 argc -= optind; 165 argv += optind; 166 167 if (!clear) { 168 if (argc != 1 && argc != 2) 169 usage(); 170 dirname = argv[0]; 171 } 172 if (argc == 2) 173 kernel = argv[1]; 174 175 (void)time(&now); 176 kmem_setup(); 177 178 if (clear) { 179 clear_dump(); 180 exit(0); 181 } 182 183 if (!dump_exists() && !force) 184 exit(1); 185 186 check_kmem(); 187 188 if (panicstr) 189 syslog(LOG_ALERT, "reboot after panic: %s", panic_mesg); 190 else 191 syslog(LOG_ALERT, "reboot"); 192 193 get_dumpsize(); 194 195 if ((!get_crashtime() || !check_space()) && !force) 196 exit(1); 197 198 save_core(); 199 200 clear_dump(); 201 exit(0); 202} 203 204void 205kmem_setup() 206{ 207 FILE *fp; 208 int kmem, i; 209 const char *dump_sys; 210 211 /* 212 * Some names we need for the currently running system, others for 213 * the system that was running when the dump was made. The values 214 * obtained from the current system are used to look for things in 215 * /dev/kmem that cannot be found in the dump_sys namelist, but are 216 * presumed to be the same (since the disk partitions are probably 217 * the same!) 218 */ 219 if ((nlist(getbootfile(), current_nl)) == -1) 220 syslog(LOG_ERR, "%s: nlist: %s", getbootfile(), 221 strerror(errno)); 222 for (i = 0; cursyms[i] != -1; i++) 223 if (current_nl[cursyms[i]].n_value == 0) { 224 syslog(LOG_ERR, "%s: %s not in namelist", 225 getbootfile(), current_nl[cursyms[i]].n_name); 226 exit(1); 227 } 228 229 dump_sys = kernel ? kernel : getbootfile(); 230 if ((nlist(dump_sys, dump_nl)) == -1) 231 syslog(LOG_ERR, "%s: nlist: %s", dump_sys, strerror(errno)); 232 for (i = 0; dumpsyms[i] != -1; i++) 233 if (dump_nl[dumpsyms[i]].n_value == 0) { 234 syslog(LOG_ERR, "%s: %s not in namelist", 235 dump_sys, dump_nl[dumpsyms[i]].n_name); 236 exit(1); 237 } 238 239 kmem = Open(_PATH_KMEM, O_RDONLY); 240 Lseek(kmem, (off_t)current_nl[X_DUMPDEV].n_value, L_SET); 241 (void)Read(kmem, &dumpdev, sizeof(dumpdev)); 242 if (dumpdev == NODEV) { 243 syslog(LOG_WARNING, "no core dump (no dumpdev)"); 244 exit(1); 245 } 246 Lseek(kmem, (off_t)current_nl[X_DUMPLO].n_value, L_SET); 247 (void)Read(kmem, &dumplo, sizeof(dumplo)); 248 if (verbose) 249 (void)printf("dumplo = %d (%d * %d)\n", 250 dumplo, dumplo/DEV_BSIZE, DEV_BSIZE); 251 Lseek(kmem, (off_t)current_nl[X_DUMPMAG].n_value, L_SET); 252 (void)Read(kmem, &dumpmag, sizeof(dumpmag)); 253 dumplo *= DEV_BSIZE; 254 ddname = find_dev(dumpdev, S_IFBLK); 255 dumpfd = Open(ddname, O_RDWR); 256 fp = fdopen(kmem, "r"); 257 if (fp == NULL) { 258 syslog(LOG_ERR, "%s: fdopen: %m", _PATH_KMEM); 259 exit(1); 260 } 261 if (kernel) 262 return; 263 (void)fseek(fp, (off_t)current_nl[X_VERSION].n_value, L_SET); 264 (void)fgets(vers, sizeof(vers), fp); 265 266 /* Don't fclose(fp), we use dumpfd later. */ 267} 268 269void 270check_kmem() 271{ 272 register char *cp; 273 FILE *fp; 274 char core_vers[1024]; 275 276 fp = fdopen(dumpfd, "r"); 277 if (fp == NULL) { 278 syslog(LOG_ERR, "%s: fdopen: %m", ddname); 279 exit(1); 280 } 281 fseek(fp, (off_t)(dumplo + ok(dump_nl[X_VERSION].n_value)), L_SET); 282 fgets(core_vers, sizeof(core_vers), fp); 283 if (strcmp(vers, core_vers) && kernel == 0) 284 syslog(LOG_WARNING, 285 "warning: %s version mismatch:\n\t%s\nand\t%s\n", 286 getbootfile(), vers, core_vers); 287 (void)fseek(fp, 288 (off_t)(dumplo + ok(dump_nl[X_PANICSTR].n_value)), L_SET); 289 (void)fread(&panicstr, sizeof(panicstr), 1, fp); 290 if (panicstr) { 291 (void)fseek(fp, dumplo + ok(panicstr), L_SET); 292 cp = panic_mesg; 293 do 294 *cp = getc(fp); 295 while (*cp++ && cp < &panic_mesg[sizeof(panic_mesg)]); 296 } 297 /* Don't fclose(fp), we use dumpfd later. */ 298} 299 300void 301clear_dump() 302{ 303 long newdumplo; 304 305 newdumplo = 0; 306 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_DUMPMAG].n_value)), L_SET); 307 Write(dumpfd, &newdumplo, sizeof(newdumplo)); 308} 309 310int 311dump_exists() 312{ 313 int newdumpmag; 314 315 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_DUMPMAG].n_value)), L_SET); 316 (void)Read(dumpfd, &newdumpmag, sizeof(newdumpmag)); 317 if (newdumpmag != dumpmag) { 318 if (verbose) 319 syslog(LOG_WARNING, "magic number mismatch (%x != %x)", 320 newdumpmag, dumpmag); 321 syslog(LOG_WARNING, "no core dump"); 322 return (0); 323 } 324 return (1); 325} 326 327char buf[1024 * 1024]; 328 329void 330save_core() 331{ 332 register FILE *fp; 333 register int bounds, ifd, nr, nw, ofd; 334 char *rawp, path[MAXPATHLEN]; 335 336 /* 337 * Get the current number and update the bounds file. Do the update 338 * now, because may fail later and don't want to overwrite anything. 339 */ 340 (void)snprintf(path, sizeof(path), "%s/bounds", dirname); 341 if ((fp = fopen(path, "r")) == NULL) 342 goto err1; 343 if (fgets(buf, sizeof(buf), fp) == NULL) { 344 if (ferror(fp)) 345err1: syslog(LOG_WARNING, "%s: %s", path, strerror(errno)); 346 bounds = 0; 347 } else 348 bounds = atoi(buf); 349 if (fp != NULL) 350 (void)fclose(fp); 351 if ((fp = fopen(path, "w")) == NULL) 352 syslog(LOG_ERR, "%s: %m", path); 353 else { 354 (void)fprintf(fp, "%d\n", bounds + 1); 355 (void)fclose(fp); 356 } 357 358 /* Create the core file. */ 359 (void)snprintf(path, sizeof(path), "%s/vmcore.%d%s", 360 dirname, bounds, compress ? ".Z" : ""); 361 if (compress) { 362 if ((fp = zopen(path, "w", 0)) == NULL) { 363 syslog(LOG_ERR, "%s: %s", path, strerror(errno)); 364 exit(1); 365 } 366 } else 367 ofd = Create(path, S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH); 368 369 /* Open the raw device. */ 370 rawp = rawname(ddname); 371 if ((ifd = open(rawp, O_RDONLY)) == -1) { 372 syslog(LOG_WARNING, "%s: %m; using block device", rawp); 373 ifd = dumpfd; 374 } 375 376 /* Seek to the start of the core. */ 377 Lseek(ifd, (off_t)dumplo, L_SET); 378 379 /* Copy the core file. */ 380 syslog(LOG_NOTICE, "writing %score to %s", 381 compress ? "compressed " : "", path); 382 for (; dumpsize > 0; dumpsize -= nr) { 383 (void)printf("%6dK\r", dumpsize / 1024); 384 (void)fflush(stdout); 385 nr = read(ifd, buf, MIN(dumpsize, sizeof(buf))); 386 if (nr <= 0) { 387 if (nr == 0) 388 syslog(LOG_WARNING, 389 "WARNING: EOF on dump device"); 390 else 391 syslog(LOG_ERR, "%s: %m", rawp); 392 goto err2; 393 } 394 if (compress) 395 nw = fwrite(buf, 1, nr, fp); 396 else 397 nw = write(ofd, buf, nr); 398 if (nw != nr) { 399 syslog(LOG_ERR, "%s: %s", 400 path, strerror(nw == 0 ? EIO : errno)); 401err2: syslog(LOG_WARNING, 402 "WARNING: vmcore may be incomplete"); 403 (void)printf("\n"); 404 exit(1); 405 } 406 } 407 (void)close(ifd); 408 if (compress) 409 (void)fclose(fp); 410 else 411 (void)close(ofd); 412 413 /* Copy the kernel. */ 414 ifd = Open(kernel ? kernel : getbootfile(), O_RDONLY); 415 (void)snprintf(path, sizeof(path), "%s/kernel.%d%s", 416 dirname, bounds, compress ? ".Z" : ""); 417 if (compress) { 418 if ((fp = zopen(path, "w", 0)) == NULL) { 419 syslog(LOG_ERR, "%s: %s", path, strerror(errno)); 420 exit(1); 421 } 422 } else 423 ofd = Create(path, S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH); 424 syslog(LOG_NOTICE, "writing %skernel to %s", 425 compress ? "compressed " : "", path); 426 while ((nr = read(ifd, buf, sizeof(buf))) > 0) { 427 if (compress) 428 nw = fwrite(buf, 1, nr, fp); 429 else 430 nw = write(ofd, buf, nr); 431 if (nw != nr) { 432 syslog(LOG_ERR, "%s: %s", 433 path, strerror(nw == 0 ? EIO : errno)); 434 syslog(LOG_WARNING, 435 "WARNING: kernel may be incomplete"); 436 exit(1); 437 } 438 } 439 if (nr < 0) { 440 syslog(LOG_ERR, "%s: %s", 441 kernel ? kernel : getbootfile(), strerror(errno)); 442 syslog(LOG_WARNING, 443 "WARNING: kernel may be incomplete"); 444 exit(1); 445 } 446 if (compress) 447 (void)fclose(fp); 448 else 449 (void)close(ofd); 450} 451 452char * 453find_dev(dev, type) 454 register dev_t dev; 455 register int type; 456{ 457 register DIR *dfd; 458 struct dirent *dir; 459 struct stat sb; 460 char *dp, devname[MAXPATHLEN + 1]; 461 462 if ((dfd = opendir(_PATH_DEV)) == NULL) { 463 syslog(LOG_ERR, "%s: %s", _PATH_DEV, strerror(errno)); 464 exit(1); 465 } 466 (void)strcpy(devname, _PATH_DEV); 467 while ((dir = readdir(dfd))) { 468 (void)strcpy(devname + sizeof(_PATH_DEV) - 1, dir->d_name); 469 if (lstat(devname, &sb)) { 470 syslog(LOG_ERR, "%s: %s", devname, strerror(errno)); 471 continue; 472 } 473 if ((sb.st_mode & S_IFMT) != type) 474 continue; 475 if (dev == sb.st_rdev) { 476 closedir(dfd); 477 if ((dp = strdup(devname)) == NULL) { 478 syslog(LOG_ERR, "%s", strerror(errno)); 479 exit(1); 480 } 481 return (dp); 482 } 483 } 484 closedir(dfd); 485 syslog(LOG_ERR, "can't find device %d/%d", major(dev), minor(dev)); 486 exit(1); 487} 488 489char * 490rawname(s) 491 char *s; 492{ 493 char *sl, name[MAXPATHLEN]; 494 495 if ((sl = rindex(s, '/')) == NULL || sl[1] == '0') { 496 syslog(LOG_ERR, 497 "can't make raw dump device name from %s", s); 498 return (s); 499 } 500 (void)snprintf(name, sizeof(name), "%.*s/r%s", sl - s, s, sl + 1); 501 if ((sl = strdup(name)) == NULL) { 502 syslog(LOG_ERR, "%s", strerror(errno)); 503 exit(1); 504 } 505 return (sl); 506} 507 508int 509get_crashtime() 510{ 511 time_t dumptime; /* Time the dump was taken. */ 512 513 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_TIME].n_value)), L_SET); 514 (void)Read(dumpfd, &dumptime, sizeof(dumptime)); 515 if (dumptime == 0) { 516 if (verbose) 517 syslog(LOG_ERR, "dump time is zero"); 518 return (0); 519 } 520 (void)printf("savecore: system went down at %s", ctime(&dumptime)); 521#define LEEWAY (7 * 86400) 522 if (dumptime < now - LEEWAY || dumptime > now + LEEWAY) { 523 (void)printf("dump time is unreasonable\n"); 524 return (0); 525 } 526 return (1); 527} 528 529int 530get_dumpsize() 531{ 532 /* Read the dump size. */ 533 Lseek(dumpfd, (off_t)(dumplo + ok(dump_nl[X_DUMPSIZE].n_value)), L_SET); 534 (void)Read(dumpfd, &dumpsize, sizeof(dumpsize)); 535 dumpsize *= getpagesize(); 536 537 return(1); 538} 539 540int 541check_space() 542{ 543 register FILE *fp; 544 const char *tkernel; 545 off_t minfree, spacefree, totfree, kernelsize, needed; 546 struct stat st; 547 struct statfs fsbuf; 548 char buf[100], path[MAXPATHLEN]; 549 550 tkernel = kernel ? kernel : getbootfile(); 551 if (stat(tkernel, &st) < 0) { 552 syslog(LOG_ERR, "%s: %m", tkernel); 553 exit(1); 554 } 555 kernelsize = st.st_blocks * S_BLKSIZE; 556 557 if (statfs(dirname, &fsbuf) < 0) { 558 syslog(LOG_ERR, "%s: %m", dirname); 559 exit(1); 560 } 561 spacefree = ((off_t) fsbuf.f_bavail * fsbuf.f_bsize) / 1024; 562 totfree = ((off_t) fsbuf.f_bfree * fsbuf.f_bsize) / 1024; 563 564 (void)snprintf(path, sizeof(path), "%s/minfree", dirname); 565 if ((fp = fopen(path, "r")) == NULL) 566 minfree = 0; 567 else { 568 if (fgets(buf, sizeof(buf), fp) == NULL) 569 minfree = 0; 570 else 571 minfree = atoi(buf); 572 (void)fclose(fp); 573 } 574 575 needed = (dumpsize + kernelsize) / 1024; 576 if (((minfree > 0) ? spacefree : totfree) - needed < minfree) { 577 syslog(LOG_WARNING, 578 "no dump, not enough free space on device"); 579 return (0); 580 } 581 if (spacefree - needed < 0) 582 syslog(LOG_WARNING, 583 "dump performed, but free space threshold crossed"); 584 return (1); 585} 586 587int 588Open(name, rw) 589 const char *name; 590 int rw; 591{ 592 int fd; 593 594 if ((fd = open(name, rw, 0)) < 0) { 595 syslog(LOG_ERR, "%s: %m", name); 596 exit(1); 597 } 598 return (fd); 599} 600 601int 602Read(fd, bp, size) 603 int fd, size; 604 void *bp; 605{ 606 int nr; 607 608 nr = read(fd, bp, size); 609 if (nr != size) { 610 syslog(LOG_ERR, "read: %m"); 611 exit(1); 612 } 613 return (nr); 614} 615 616void 617Lseek(fd, off, flag) 618 int fd, flag; 619 off_t off; 620{ 621 off_t ret; 622 623 ret = lseek(fd, off, flag); 624 if (ret == -1) { 625 syslog(LOG_ERR, "lseek: %m"); 626 exit(1); 627 } 628} 629 630int 631Create(file, mode) 632 char *file; 633 int mode; 634{ 635 register int fd; 636 637 fd = creat(file, mode); 638 if (fd < 0) { 639 syslog(LOG_ERR, "%s: %m", file); 640 exit(1); 641 } 642 return (fd); 643} 644 645void 646Write(fd, bp, size) 647 int fd, size; 648 void *bp; 649{ 650 int n; 651 652 if ((n = write(fd, bp, size)) < size) { 653 syslog(LOG_ERR, "write: %s", strerror(n == -1 ? errno : EIO)); 654 exit(1); 655 } 656} 657 658void 659usage() 660{ 661 (void)syslog(LOG_ERR, "usage: savecore [-cfvz] [-N system] directory"); 662 exit(1); 663} 664