1 /* 2 * Copyright (c) 1989, 1993, 1994 3 * The Regents of the University of California. All rights reserved. 4 * 5 * This code is derived from software contributed to Berkeley by 6 * Chris Newcomb. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by the University of 19 * California, Berkeley and its contributors. 20 * 4. Neither the name of the University nor the names of its contributors 21 * may be used to endorse or promote products derived from this software 22 * without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 34 * SUCH DAMAGE. 35 * 36 * @(#) Copyright (c) 1989, 1993, 1994 The Regents of the University of California. All rights reserved. 37 * @(#)du.c 8.5 (Berkeley) 5/4/95 38 * $FreeBSD: src/usr.bin/du/du.c,v 1.17.2.4 2002/12/12 16:29:39 trhodes Exp $ 39 * $DragonFly: src/usr.bin/du/du.c,v 1.6 2004/07/04 10:34:47 eirikn Exp $ 40 */ 41 42 #include <sys/param.h> 43 #include <sys/queue.h> 44 #include <sys/stat.h> 45 46 #include <err.h> 47 #include <errno.h> 48 #include <fnmatch.h> 49 #include <fts.h> 50 #include <math.h> 51 #include <stdio.h> 52 #include <stdlib.h> 53 #include <string.h> 54 #include <sysexits.h> 55 #include <unistd.h> 56 57 #define KILO_SZ(n) (n) 58 #define MEGA_SZ(n) ((n) * (n)) 59 #define GIGA_SZ(n) ((n) * (n) * (n)) 60 #define TERA_SZ(n) ((n) * (n) * (n) * (n)) 61 #define PETA_SZ(n) ((n) * (n) * (n) * (n) * (n)) 62 63 #define KILO_2_SZ (KILO_SZ(1024ULL)) 64 #define MEGA_2_SZ (MEGA_SZ(1024ULL)) 65 #define GIGA_2_SZ (GIGA_SZ(1024ULL)) 66 #define TERA_2_SZ (TERA_SZ(1024ULL)) 67 #define PETA_2_SZ (PETA_SZ(1024ULL)) 68 69 #define KILO_SI_SZ (KILO_SZ(1000ULL)) 70 #define MEGA_SI_SZ (MEGA_SZ(1000ULL)) 71 #define GIGA_SI_SZ (GIGA_SZ(1000ULL)) 72 #define TERA_SI_SZ (TERA_SZ(1000ULL)) 73 #define PETA_SI_SZ (PETA_SZ(1000ULL)) 74 75 #define HASHSIZE 256 /* power of 2 only */ 76 #define HASHMASK (HASHSIZE - 1) 77 78 unsigned long long vals_si [] = {1, KILO_SI_SZ, MEGA_SI_SZ, GIGA_SI_SZ, TERA_SI_SZ, PETA_SI_SZ}; 79 unsigned long long vals_base2[] = {1, KILO_2_SZ, MEGA_2_SZ, GIGA_2_SZ, TERA_2_SZ, PETA_2_SZ}; 80 unsigned long long *valp; 81 82 typedef enum { NONE, KILO, MEGA, GIGA, TERA, PETA, UNIT_MAX } unit_t; 83 84 int unitp [] = { NONE, KILO, MEGA, GIGA, TERA, PETA }; 85 86 SLIST_HEAD(ignhead, ignentry) ignores; 87 struct ignentry { 88 char *mask; 89 SLIST_ENTRY(ignentry) next; 90 }; 91 92 static int linkchk(FTSENT *); 93 static void usage(void); 94 void prthumanval(double); 95 unit_t unit_adjust(double *); 96 void ignoreadd(const char *); 97 void ignoreclean(void); 98 int ignorep(FTSENT *); 99 100 int 101 main(int argc, char **argv) 102 { 103 FTS *fts; 104 FTSENT *p; 105 long blocksize, savednumber = 0; 106 int ftsoptions; 107 int listall; 108 int depth; 109 int Hflag, Lflag, Pflag, aflag, sflag, dflag, cflag, hflag, ch, notused, rval; 110 char **save; 111 112 Hflag = Lflag = Pflag = aflag = sflag = dflag = cflag = hflag = 0; 113 114 save = argv; 115 ftsoptions = 0; 116 depth = INT_MAX; 117 SLIST_INIT(&ignores); 118 119 while ((ch = getopt(argc, argv, "HI:LPasd:chkrx")) != -1) 120 switch (ch) { 121 case 'H': 122 Hflag = 1; 123 break; 124 case 'I': 125 ignoreadd(optarg); 126 break; 127 case 'L': 128 if (Pflag) 129 usage(); 130 Lflag = 1; 131 break; 132 case 'P': 133 if (Lflag) 134 usage(); 135 Pflag = 1; 136 break; 137 case 'a': 138 aflag = 1; 139 break; 140 case 's': 141 sflag = 1; 142 break; 143 case 'd': 144 dflag = 1; 145 errno = 0; 146 depth = atoi(optarg); 147 if (errno == ERANGE || depth < 0) { 148 warnx("invalid argument to option d: %s", optarg); 149 usage(); 150 } 151 break; 152 case 'c': 153 cflag = 1; 154 break; 155 case 'h': 156 putenv("BLOCKSIZE=512"); 157 hflag = 1; 158 valp = vals_base2; 159 break; 160 case 'k': 161 hflag = 0; 162 putenv("BLOCKSIZE=1024"); 163 break; 164 case 'r': /* Compatibility. */ 165 break; 166 case 'x': 167 ftsoptions |= FTS_XDEV; 168 break; 169 case '?': 170 default: 171 usage(); 172 } 173 174 argc -= optind; 175 argv += optind; 176 177 /* 178 * XXX 179 * Because of the way that fts(3) works, logical walks will not count 180 * the blocks actually used by symbolic links. We rationalize this by 181 * noting that users computing logical sizes are likely to do logical 182 * copies, so not counting the links is correct. The real reason is 183 * that we'd have to re-implement the kernel's symbolic link traversing 184 * algorithm to get this right. If, for example, you have relative 185 * symbolic links referencing other relative symbolic links, it gets 186 * very nasty, very fast. The bottom line is that it's documented in 187 * the man page, so it's a feature. 188 */ 189 190 if (Hflag + Lflag + Pflag > 1) 191 usage(); 192 193 if (Hflag + Lflag + Pflag == 0) 194 Pflag = 1; /* -P (physical) is default */ 195 196 if (Hflag) 197 ftsoptions |= FTS_COMFOLLOW; 198 199 if (Lflag) 200 ftsoptions |= FTS_LOGICAL; 201 202 if (Pflag) 203 ftsoptions |= FTS_PHYSICAL; 204 205 listall = 0; 206 207 if (aflag) { 208 if (sflag || dflag) 209 usage(); 210 listall = 1; 211 } else if (sflag) { 212 if (dflag) 213 usage(); 214 depth = 0; 215 } 216 217 if (!*argv) { 218 argv = save; 219 argv[0] = "."; 220 argv[1] = NULL; 221 } 222 223 (void) getbsize(¬used, &blocksize); 224 blocksize /= 512; 225 226 rval = 0; 227 228 if ((fts = fts_open(argv, ftsoptions, NULL)) == NULL) 229 err(1, "fts_open"); 230 231 while ((p = fts_read(fts)) != NULL) { 232 switch (p->fts_info) { 233 case FTS_D: /* Ignore. */ 234 if (ignorep(p)) 235 fts_set(fts, p, FTS_SKIP); 236 break; 237 case FTS_DP: 238 if (ignorep(p)) 239 break; 240 241 p->fts_parent->fts_number += 242 p->fts_number += p->fts_statp->st_blocks; 243 244 if (p->fts_level <= depth) { 245 if (hflag) { 246 (void) prthumanval(howmany(p->fts_number, blocksize)); 247 (void) printf("\t%s\n", p->fts_path); 248 } else { 249 (void) printf("%ld\t%s\n", 250 howmany(p->fts_number, blocksize), 251 p->fts_path); 252 } 253 } 254 break; 255 case FTS_DC: /* Ignore. */ 256 break; 257 case FTS_DNR: /* Warn, continue. */ 258 case FTS_ERR: 259 case FTS_NS: 260 warnx("%s: %s", p->fts_path, strerror(p->fts_errno)); 261 rval = 1; 262 break; 263 default: 264 if (ignorep(p)) 265 break; 266 267 if (p->fts_statp->st_nlink > 1 && linkchk(p)) 268 break; 269 270 if (listall || p->fts_level == 0) { 271 if (hflag) { 272 (void) prthumanval(howmany(p->fts_statp->st_blocks, 273 blocksize)); 274 (void) printf("\t%s\n", p->fts_path); 275 } else { 276 (void) printf("%qd\t%s\n", 277 howmany(p->fts_statp->st_blocks, blocksize), 278 p->fts_path); 279 } 280 } 281 282 p->fts_parent->fts_number += p->fts_statp->st_blocks; 283 } 284 savednumber = p->fts_parent->fts_number; 285 } 286 287 if (errno) 288 err(1, "fts_read"); 289 290 if (cflag) { 291 if (hflag) { 292 (void) prthumanval(howmany(savednumber, blocksize)); 293 (void) printf("\ttotal\n"); 294 } else { 295 (void) printf("%ld\ttotal\n", howmany(savednumber, blocksize)); 296 } 297 } 298 299 ignoreclean(); 300 exit(rval); 301 } 302 303 static int 304 linkchk(FTSENT *p) 305 { 306 struct links_entry { 307 struct links_entry *next; 308 struct links_entry *previous; 309 int links; 310 dev_t dev; 311 ino_t ino; 312 }; 313 314 static const size_t links_hash_initial_size = 8192; 315 static struct links_entry **buckets; 316 static struct links_entry *free_list; 317 static size_t number_buckets; 318 static unsigned long number_entries; 319 static char stop_allocating; 320 struct links_entry *le, **new_buckets; 321 struct stat *st; 322 size_t i, new_size; 323 int count, hash; 324 325 st = p->fts_statp; 326 327 /* If necessary, initialize the hash table. */ 328 if (buckets == NULL) { 329 number_buckets = links_hash_initial_size; 330 buckets = malloc(number_buckets * sizeof(buckets[0])); 331 if (buckets == NULL) 332 errx(1, "No memory for hardlink detection"); 333 for (i = 0; i < number_buckets; i++) 334 buckets[i] = NULL; 335 } 336 337 /* If the hash table is getting too full, enlarge it. */ 338 if (number_entries > number_buckets * 10 && !stop_allocating) { 339 new_size = number_buckets * 2; 340 new_buckets = malloc(new_size * sizeof(struct links_entry *)); 341 count = 0; 342 343 /* Try releasing the free list to see if that helps. */ 344 if (new_buckets == NULL && free_list != NULL) { 345 while (free_list != NULL) { 346 le = free_list; 347 free_list = le->next; 348 free(le); 349 } 350 new_buckets = malloc(new_size * sizeof(new_buckets[0])); 351 } 352 353 if (new_buckets == NULL) { 354 stop_allocating = 1; 355 warnx("No more memory for tracking hard links"); 356 } else { 357 memset(new_buckets, 0, new_size * sizeof(struct links_entry *)); 358 for (i = 0; i < number_buckets; i++) { 359 while (buckets[i] != NULL) { 360 /* Remove entry from old bucket. */ 361 le = buckets[i]; 362 buckets[i] = le->next; 363 364 /* Add entry to new bucket. */ 365 hash = (le->dev ^ le->ino) % new_size; 366 367 if (new_buckets[hash] != NULL) 368 new_buckets[hash]->previous = le; 369 le->next = new_buckets[hash]; 370 le->previous = NULL; 371 new_buckets[hash] = le; 372 } 373 } 374 free(buckets); 375 buckets = new_buckets; 376 number_buckets = new_size; 377 } 378 } 379 380 /* Try to locate this entry in the hash table. */ 381 hash = ( st->st_dev ^ st->st_ino ) % number_buckets; 382 for (le = buckets[hash]; le != NULL; le = le->next) { 383 if (le->dev == st->st_dev && le->ino == st->st_ino) { 384 /* 385 * Save memory by releasing an entry when we've seen 386 * all of it's links. 387 */ 388 if (--le->links <= 0) { 389 if (le->previous != NULL) 390 le->previous->next = le->next; 391 if (le->next != NULL) 392 le->next->previous = le->previous; 393 if (buckets[hash] == le) 394 buckets[hash] = le->next; 395 number_entries--; 396 /* Recycle this node through the free list */ 397 if (stop_allocating) { 398 free(le); 399 } else { 400 le->next = free_list; 401 free_list = le; 402 } 403 } 404 return (1); 405 } 406 } 407 408 if (stop_allocating) 409 return (0); 410 411 /* Add this entry to the links cache. */ 412 if (free_list != NULL) { 413 /* Pull a node from the free list if we can. */ 414 le = free_list; 415 free_list = le->next; 416 } else 417 /* Malloc one if we have to. */ 418 le = malloc(sizeof(struct links_entry)); 419 if (le == NULL) { 420 stop_allocating = 1; 421 warnx("No more memory for tracking hard links"); 422 return (0); 423 } 424 le->dev = st->st_dev; 425 le->ino = st->st_ino; 426 le->links = st->st_nlink - 1; 427 number_entries++; 428 le->next = buckets[hash]; 429 le->previous = NULL; 430 if (buckets[hash] != NULL) 431 buckets[hash]->previous = le; 432 buckets[hash] = le; 433 return (0); 434 } 435 436 /* 437 * Output in "human-readable" format. Uses 3 digits max and puts 438 * unit suffixes at the end. Makes output compact and easy to read, 439 * especially on huge disks. 440 * 441 */ 442 unit_t 443 unit_adjust(double *val) 444 { 445 double abval; 446 unit_t unit; 447 unsigned int unit_sz; 448 449 abval = fabs(*val); 450 451 unit_sz = abval ? ilogb(abval) / 10 : 0; 452 453 if (unit_sz >= UNIT_MAX) { 454 unit = NONE; 455 } else { 456 unit = unitp[unit_sz]; 457 *val /= (double)valp[unit_sz]; 458 } 459 460 return (unit); 461 } 462 463 void 464 prthumanval(double bytes) 465 { 466 unit_t unit; 467 468 bytes *= 512; 469 unit = unit_adjust(&bytes); 470 471 if (bytes == 0) 472 (void)printf(" 0B"); 473 else if (bytes > 10) 474 (void)printf("%3.0f%c", bytes, "BKMGTPE"[unit]); 475 else 476 (void)printf("%3.1f%c", bytes, "BKMGTPE"[unit]); 477 } 478 479 static void 480 usage(void) 481 { 482 (void)fprintf(stderr, 483 "usage: du [-H | -L | -P] [-a | -s | -d depth] [-c] [-h | -k] [-x] [-I mask] [file ...]\n"); 484 exit(EX_USAGE); 485 } 486 487 void 488 ignoreadd(const char *mask) 489 { 490 struct ignentry *ign; 491 492 ign = calloc(1, sizeof(*ign)); 493 if (ign == NULL) 494 errx(1, "cannot allocate memory"); 495 ign->mask = strdup(mask); 496 if (ign->mask == NULL) 497 errx(1, "cannot allocate memory"); 498 SLIST_INSERT_HEAD(&ignores, ign, next); 499 } 500 501 void 502 ignoreclean(void) 503 { 504 struct ignentry *ign; 505 506 while (!SLIST_EMPTY(&ignores)) { 507 ign = SLIST_FIRST(&ignores); 508 SLIST_REMOVE_HEAD(&ignores, next); 509 free(ign->mask); 510 free(ign); 511 } 512 } 513 514 int 515 ignorep(FTSENT *ent) 516 { 517 struct ignentry *ign; 518 519 SLIST_FOREACH(ign, &ignores, next) 520 if (fnmatch(ign->mask, ent->fts_name, 0) != FNM_NOMATCH) 521 return 1; 522 return 0; 523 } 524