1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (c) 1989, 1993, 1994 5 * The Regents of the University of California. All rights reserved. 6 * 7 * This code is derived from software contributed to Berkeley by 8 * Chris Newcomb. 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. Neither the name of the University nor the names of its contributors 19 * may be used to endorse or promote products derived from this software 20 * without specific prior written permission. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 32 * SUCH DAMAGE. 33 */ 34 35 #include <sys/param.h> 36 #include <sys/queue.h> 37 #include <sys/stat.h> 38 #include <err.h> 39 #include <errno.h> 40 #include <fnmatch.h> 41 #include <fts.h> 42 #include <getopt.h> 43 #include <libutil.h> 44 #include <locale.h> 45 #include <stdint.h> 46 #include <stdio.h> 47 #include <stdlib.h> 48 #include <string.h> 49 #include <sysexits.h> 50 #include <unistd.h> 51 52 #define SI_OPT (CHAR_MAX + 1) 53 54 #define UNITS_2 1 55 #define UNITS_SI 2 56 57 static SLIST_HEAD(ignhead, ignentry) ignores; 58 struct ignentry { 59 char *mask; 60 SLIST_ENTRY(ignentry) next; 61 }; 62 63 static int linkchk(FTSENT *); 64 static void usage(void); 65 static void prthumanval(int64_t); 66 static void ignoreadd(const char *); 67 static void ignoreclean(void); 68 static int ignorep(FTSENT *); 69 static void siginfo(int __unused); 70 71 static int nodumpflag = 0; 72 static int Aflag, hflag; 73 static long blocksize, cblocksize; 74 static volatile sig_atomic_t info; 75 76 static const struct option long_options[] = 77 { 78 { "si", no_argument, NULL, SI_OPT }, 79 { NULL, no_argument, NULL, 0 }, 80 }; 81 82 int 83 main(int argc, char *argv[]) 84 { 85 FTS *fts; 86 FTSENT *p; 87 off_t savednumber, curblocks; 88 off_t threshold, threshold_sign; 89 int ftsoptions; 90 int depth; 91 int Hflag, Lflag, aflag, sflag, dflag, cflag; 92 int lflag, ch, notused, rval; 93 char **save; 94 static char dot[] = "."; 95 96 setlocale(LC_ALL, ""); 97 98 Hflag = Lflag = aflag = sflag = dflag = cflag = lflag = Aflag = 0; 99 100 save = argv; 101 ftsoptions = FTS_PHYSICAL; 102 savednumber = 0; 103 threshold = 0; 104 threshold_sign = 1; 105 cblocksize = DEV_BSIZE; 106 blocksize = 0; 107 depth = INT_MAX; 108 SLIST_INIT(&ignores); 109 110 while ((ch = getopt_long(argc, argv, "+AB:HI:LPasd:cghklmnrt:x", 111 long_options, NULL)) != -1) 112 switch (ch) { 113 case 'A': 114 Aflag = 1; 115 break; 116 case 'B': 117 errno = 0; 118 cblocksize = atoi(optarg); 119 if (errno == ERANGE || cblocksize <= 0) { 120 warnx("invalid argument to option B: %s", 121 optarg); 122 usage(); 123 } 124 break; 125 case 'H': 126 Hflag = 1; 127 Lflag = 0; 128 break; 129 case 'I': 130 ignoreadd(optarg); 131 break; 132 case 'L': 133 Lflag = 1; 134 Hflag = 0; 135 break; 136 case 'P': 137 Hflag = Lflag = 0; 138 break; 139 case 'a': 140 aflag = 1; 141 break; 142 case 's': 143 sflag = 1; 144 break; 145 case 'd': 146 dflag = 1; 147 errno = 0; 148 depth = atoi(optarg); 149 if (errno == ERANGE || depth < 0) { 150 warnx("invalid argument to option d: %s", 151 optarg); 152 usage(); 153 } 154 break; 155 case 'c': 156 cflag = 1; 157 break; 158 case 'g': 159 hflag = 0; 160 blocksize = 1073741824; 161 break; 162 case 'h': 163 hflag = UNITS_2; 164 break; 165 case 'k': 166 hflag = 0; 167 blocksize = 1024; 168 break; 169 case 'l': 170 lflag = 1; 171 break; 172 case 'm': 173 hflag = 0; 174 blocksize = 1048576; 175 break; 176 case 'n': 177 nodumpflag = 1; 178 break; 179 case 'r': /* Compatibility. */ 180 break; 181 case 't' : 182 if (expand_number(optarg, &threshold) != 0 || 183 threshold == 0) { 184 warnx("invalid threshold: %s", optarg); 185 usage(); 186 } else if (threshold < 0) 187 threshold_sign = -1; 188 break; 189 case 'x': 190 ftsoptions |= FTS_XDEV; 191 break; 192 case SI_OPT: 193 hflag = UNITS_SI; 194 break; 195 case '?': 196 default: 197 usage(); 198 /* NOTREACHED */ 199 } 200 201 argc -= optind; 202 argv += optind; 203 204 /* 205 * XXX 206 * Because of the way that fts(3) works, logical walks will not count 207 * the blocks actually used by symbolic links. We rationalize this by 208 * noting that users computing logical sizes are likely to do logical 209 * copies, so not counting the links is correct. The real reason is 210 * that we'd have to re-implement the kernel's symbolic link traversing 211 * algorithm to get this right. If, for example, you have relative 212 * symbolic links referencing other relative symbolic links, it gets 213 * very nasty, very fast. The bottom line is that it's documented in 214 * the man page, so it's a feature. 215 */ 216 217 if (Hflag) 218 ftsoptions |= FTS_COMFOLLOW; 219 if (Lflag) { 220 ftsoptions &= ~FTS_PHYSICAL; 221 ftsoptions |= FTS_LOGICAL; 222 } 223 224 if (!Aflag && (cblocksize % DEV_BSIZE) != 0) 225 cblocksize = howmany(cblocksize, DEV_BSIZE) * DEV_BSIZE; 226 227 if (aflag + dflag + sflag > 1) 228 usage(); 229 if (sflag) 230 depth = 0; 231 232 if (!*argv) { 233 argv = save; 234 argv[0] = dot; 235 argv[1] = NULL; 236 } 237 238 if (blocksize == 0) 239 (void)getbsize(¬used, &blocksize); 240 241 if (!Aflag) { 242 cblocksize /= DEV_BSIZE; 243 blocksize /= DEV_BSIZE; 244 } 245 246 if (threshold != 0) 247 threshold = howmany(threshold / DEV_BSIZE * cblocksize, 248 blocksize); 249 250 rval = 0; 251 252 (void)signal(SIGINFO, siginfo); 253 254 if ((fts = fts_open(argv, ftsoptions, NULL)) == NULL) 255 err(1, "fts_open"); 256 257 while (errno = 0, (p = fts_read(fts)) != NULL) { 258 switch (p->fts_info) { 259 case FTS_D: /* Ignore. */ 260 if (ignorep(p)) 261 fts_set(fts, p, FTS_SKIP); 262 break; 263 case FTS_DP: 264 if (ignorep(p)) 265 break; 266 267 curblocks = Aflag ? 268 howmany(p->fts_statp->st_size, cblocksize) : 269 howmany(p->fts_statp->st_blocks, cblocksize); 270 p->fts_parent->fts_bignum += p->fts_bignum += 271 curblocks; 272 273 if (p->fts_level <= depth && threshold <= 274 threshold_sign * howmany(p->fts_bignum * 275 cblocksize, blocksize)) { 276 if (hflag > 0) { 277 prthumanval(p->fts_bignum); 278 (void)printf("\t%s\n", p->fts_path); 279 } else { 280 (void)printf("%jd\t%s\n", 281 (intmax_t)howmany(p->fts_bignum * 282 cblocksize, blocksize), 283 p->fts_path); 284 } 285 } 286 if (info) { 287 info = 0; 288 (void)printf("\t%s\n", p->fts_path); 289 } 290 break; 291 case FTS_DC: /* Ignore. */ 292 break; 293 case FTS_DNR: /* Warn, continue. */ 294 case FTS_ERR: 295 case FTS_NS: 296 warnx("%s: %s", p->fts_path, strerror(p->fts_errno)); 297 rval = 1; 298 break; 299 default: 300 if (ignorep(p)) 301 break; 302 303 if (lflag == 0 && p->fts_statp->st_nlink > 1 && 304 linkchk(p)) 305 break; 306 307 curblocks = Aflag ? 308 howmany(p->fts_statp->st_size, cblocksize) : 309 howmany(p->fts_statp->st_blocks, cblocksize); 310 311 if (aflag || p->fts_level == 0) { 312 if (hflag > 0) { 313 prthumanval(curblocks); 314 (void)printf("\t%s\n", p->fts_path); 315 } else { 316 (void)printf("%jd\t%s\n", 317 (intmax_t)howmany(curblocks * 318 cblocksize, blocksize), 319 p->fts_path); 320 } 321 } 322 323 p->fts_parent->fts_bignum += curblocks; 324 } 325 savednumber = p->fts_parent->fts_bignum; 326 } 327 328 if (errno) 329 err(1, "fts_read"); 330 331 if (cflag) { 332 if (hflag > 0) { 333 prthumanval(savednumber); 334 (void)printf("\ttotal\n"); 335 } else { 336 (void)printf("%jd\ttotal\n", (intmax_t)howmany( 337 savednumber * cblocksize, blocksize)); 338 } 339 } 340 341 ignoreclean(); 342 exit(rval); 343 } 344 345 static int 346 linkchk(FTSENT *p) 347 { 348 struct links_entry { 349 struct links_entry *next; 350 struct links_entry *previous; 351 int links; 352 dev_t dev; 353 ino_t ino; 354 }; 355 static const size_t links_hash_initial_size = 8192; 356 static struct links_entry **buckets; 357 static struct links_entry *free_list; 358 static size_t number_buckets; 359 static unsigned long number_entries; 360 static char stop_allocating; 361 struct links_entry *le, **new_buckets; 362 struct stat *st; 363 size_t i, new_size; 364 int hash; 365 366 st = p->fts_statp; 367 368 /* If necessary, initialize the hash table. */ 369 if (buckets == NULL) { 370 number_buckets = links_hash_initial_size; 371 buckets = malloc(number_buckets * sizeof(buckets[0])); 372 if (buckets == NULL) 373 errx(1, "No memory for hardlink detection"); 374 for (i = 0; i < number_buckets; i++) 375 buckets[i] = NULL; 376 } 377 378 /* If the hash table is getting too full, enlarge it. */ 379 if (number_entries > number_buckets * 10 && !stop_allocating) { 380 new_size = number_buckets * 2; 381 new_buckets = calloc(new_size, sizeof(struct links_entry *)); 382 383 /* Try releasing the free list to see if that helps. */ 384 if (new_buckets == NULL && free_list != NULL) { 385 while (free_list != NULL) { 386 le = free_list; 387 free_list = le->next; 388 free(le); 389 } 390 new_buckets = calloc(new_size, sizeof(new_buckets[0])); 391 } 392 393 if (new_buckets == NULL) { 394 stop_allocating = 1; 395 warnx("No more memory for tracking hard links"); 396 } else { 397 for (i = 0; i < number_buckets; i++) { 398 while (buckets[i] != NULL) { 399 /* Remove entry from old bucket. */ 400 le = buckets[i]; 401 buckets[i] = le->next; 402 403 /* Add entry to new bucket. */ 404 hash = (le->dev ^ le->ino) % new_size; 405 406 if (new_buckets[hash] != NULL) 407 new_buckets[hash]->previous = 408 le; 409 le->next = new_buckets[hash]; 410 le->previous = NULL; 411 new_buckets[hash] = le; 412 } 413 } 414 free(buckets); 415 buckets = new_buckets; 416 number_buckets = new_size; 417 } 418 } 419 420 /* Try to locate this entry in the hash table. */ 421 hash = ( st->st_dev ^ st->st_ino ) % number_buckets; 422 for (le = buckets[hash]; le != NULL; le = le->next) { 423 if (le->dev == st->st_dev && le->ino == st->st_ino) { 424 /* 425 * Save memory by releasing an entry when we've seen 426 * all of its links. 427 */ 428 if (--le->links <= 0) { 429 if (le->previous != NULL) 430 le->previous->next = le->next; 431 if (le->next != NULL) 432 le->next->previous = le->previous; 433 if (buckets[hash] == le) 434 buckets[hash] = le->next; 435 number_entries--; 436 /* Recycle this node through the free list */ 437 if (stop_allocating) { 438 free(le); 439 } else { 440 le->next = free_list; 441 free_list = le; 442 } 443 } 444 return (1); 445 } 446 } 447 448 if (stop_allocating) 449 return (0); 450 451 /* Add this entry to the links cache. */ 452 if (free_list != NULL) { 453 /* Pull a node from the free list if we can. */ 454 le = free_list; 455 free_list = le->next; 456 } else 457 /* Malloc one if we have to. */ 458 le = malloc(sizeof(struct links_entry)); 459 if (le == NULL) { 460 stop_allocating = 1; 461 warnx("No more memory for tracking hard links"); 462 return (0); 463 } 464 le->dev = st->st_dev; 465 le->ino = st->st_ino; 466 le->links = st->st_nlink - 1; 467 number_entries++; 468 le->next = buckets[hash]; 469 le->previous = NULL; 470 if (buckets[hash] != NULL) 471 buckets[hash]->previous = le; 472 buckets[hash] = le; 473 return (0); 474 } 475 476 static void 477 prthumanval(int64_t bytes) 478 { 479 char buf[5]; 480 int flags; 481 482 bytes *= cblocksize; 483 flags = HN_B | HN_NOSPACE | HN_DECIMAL; 484 if (!Aflag) 485 bytes *= DEV_BSIZE; 486 if (hflag == UNITS_SI) 487 flags |= HN_DIVISOR_1000; 488 489 humanize_number(buf, sizeof(buf), bytes, "", HN_AUTOSCALE, flags); 490 491 (void)printf("%4s", buf); 492 } 493 494 static void 495 usage(void) 496 { 497 (void)fprintf(stderr, 498 "usage: du [-Aclnx] [-H | -L | -P] [-g | -h | -k | -m] " 499 "[-a | -s | -d depth] [-B blocksize] [-I mask] " 500 "[-t threshold] [file ...]\n"); 501 exit(EX_USAGE); 502 } 503 504 static void 505 ignoreadd(const char *mask) 506 { 507 struct ignentry *ign; 508 509 ign = calloc(1, sizeof(*ign)); 510 if (ign == NULL) 511 errx(1, "cannot allocate memory"); 512 ign->mask = strdup(mask); 513 if (ign->mask == NULL) 514 errx(1, "cannot allocate memory"); 515 SLIST_INSERT_HEAD(&ignores, ign, next); 516 } 517 518 static void 519 ignoreclean(void) 520 { 521 struct ignentry *ign; 522 523 while (!SLIST_EMPTY(&ignores)) { 524 ign = SLIST_FIRST(&ignores); 525 SLIST_REMOVE_HEAD(&ignores, next); 526 free(ign->mask); 527 free(ign); 528 } 529 } 530 531 static int 532 ignorep(FTSENT *ent) 533 { 534 struct ignentry *ign; 535 536 if (nodumpflag && (ent->fts_statp->st_flags & UF_NODUMP)) 537 return 1; 538 SLIST_FOREACH(ign, &ignores, next) 539 if (fnmatch(ign->mask, ent->fts_name, 0) != FNM_NOMATCH) 540 return 1; 541 return 0; 542 } 543 544 static void 545 siginfo(int sig __unused) 546 { 547 548 info = 1; 549 } 550