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 1.7 2005/01/01 23:02:42 cpressey 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 static char period[] = "."; 101 102 int 103 main(int argc, char **argv) 104 { 105 FTS *fts; 106 FTSENT *p; 107 long blocksize, savednumber = 0; 108 int ftsoptions; 109 int listall; 110 int depth; 111 int Hflag, Lflag, Pflag, aflag, sflag, dflag, cflag, hflag, ch, notused, rval; 112 char **save; 113 114 Hflag = Lflag = Pflag = aflag = sflag = dflag = cflag = hflag = 0; 115 116 save = argv; 117 ftsoptions = 0; 118 depth = INT_MAX; 119 SLIST_INIT(&ignores); 120 121 while ((ch = getopt(argc, argv, "HI:LPasd:chkrx")) != -1) 122 switch (ch) { 123 case 'H': 124 Hflag = 1; 125 break; 126 case 'I': 127 ignoreadd(optarg); 128 break; 129 case 'L': 130 if (Pflag) 131 usage(); 132 Lflag = 1; 133 break; 134 case 'P': 135 if (Lflag) 136 usage(); 137 Pflag = 1; 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", optarg); 151 usage(); 152 } 153 break; 154 case 'c': 155 cflag = 1; 156 break; 157 case 'h': 158 putenv("BLOCKSIZE=512"); 159 hflag = 1; 160 valp = vals_base2; 161 break; 162 case 'k': 163 hflag = 0; 164 putenv("BLOCKSIZE=1024"); 165 break; 166 case 'r': /* Compatibility. */ 167 break; 168 case 'x': 169 ftsoptions |= FTS_XDEV; 170 break; 171 case '?': 172 default: 173 usage(); 174 } 175 176 argc -= optind; 177 argv += optind; 178 179 /* 180 * XXX 181 * Because of the way that fts(3) works, logical walks will not count 182 * the blocks actually used by symbolic links. We rationalize this by 183 * noting that users computing logical sizes are likely to do logical 184 * copies, so not counting the links is correct. The real reason is 185 * that we'd have to re-implement the kernel's symbolic link traversing 186 * algorithm to get this right. If, for example, you have relative 187 * symbolic links referencing other relative symbolic links, it gets 188 * very nasty, very fast. The bottom line is that it's documented in 189 * the man page, so it's a feature. 190 */ 191 192 if (Hflag + Lflag + Pflag > 1) 193 usage(); 194 195 if (Hflag + Lflag + Pflag == 0) 196 Pflag = 1; /* -P (physical) is default */ 197 198 if (Hflag) 199 ftsoptions |= FTS_COMFOLLOW; 200 201 if (Lflag) 202 ftsoptions |= FTS_LOGICAL; 203 204 if (Pflag) 205 ftsoptions |= FTS_PHYSICAL; 206 207 listall = 0; 208 209 if (aflag) { 210 if (sflag || dflag) 211 usage(); 212 listall = 1; 213 } else if (sflag) { 214 if (dflag) 215 usage(); 216 depth = 0; 217 } 218 219 if (!*argv) { 220 argv = save; 221 argv[0] = period; 222 argv[1] = NULL; 223 } 224 225 (void) getbsize(¬used, &blocksize); 226 blocksize /= 512; 227 228 rval = 0; 229 230 if ((fts = fts_open(argv, ftsoptions, NULL)) == NULL) 231 err(1, "fts_open"); 232 233 while ((p = fts_read(fts)) != NULL) { 234 switch (p->fts_info) { 235 case FTS_D: /* Ignore. */ 236 if (ignorep(p)) 237 fts_set(fts, p, FTS_SKIP); 238 break; 239 case FTS_DP: 240 if (ignorep(p)) 241 break; 242 243 p->fts_parent->fts_number += 244 p->fts_number += p->fts_statp->st_blocks; 245 246 if (p->fts_level <= depth) { 247 if (hflag) { 248 (void) prthumanval(howmany(p->fts_number, blocksize)); 249 (void) printf("\t%s\n", p->fts_path); 250 } else { 251 (void) printf("%ld\t%s\n", 252 howmany(p->fts_number, blocksize), 253 p->fts_path); 254 } 255 } 256 break; 257 case FTS_DC: /* Ignore. */ 258 break; 259 case FTS_DNR: /* Warn, continue. */ 260 case FTS_ERR: 261 case FTS_NS: 262 warnx("%s: %s", p->fts_path, strerror(p->fts_errno)); 263 rval = 1; 264 break; 265 default: 266 if (ignorep(p)) 267 break; 268 269 if (p->fts_statp->st_nlink > 1 && linkchk(p)) 270 break; 271 272 if (listall || p->fts_level == 0) { 273 if (hflag) { 274 (void) prthumanval(howmany(p->fts_statp->st_blocks, 275 blocksize)); 276 (void) printf("\t%s\n", p->fts_path); 277 } else { 278 (void) printf("%qd\t%s\n", 279 howmany(p->fts_statp->st_blocks, blocksize), 280 p->fts_path); 281 } 282 } 283 284 p->fts_parent->fts_number += p->fts_statp->st_blocks; 285 } 286 savednumber = p->fts_parent->fts_number; 287 } 288 289 if (errno) 290 err(1, "fts_read"); 291 292 if (cflag) { 293 if (hflag) { 294 (void) prthumanval(howmany(savednumber, blocksize)); 295 (void) printf("\ttotal\n"); 296 } else { 297 (void) printf("%ld\ttotal\n", howmany(savednumber, blocksize)); 298 } 299 } 300 301 ignoreclean(); 302 exit(rval); 303 } 304 305 static int 306 linkchk(FTSENT *p) 307 { 308 struct links_entry { 309 struct links_entry *next; 310 struct links_entry *previous; 311 int links; 312 dev_t dev; 313 ino_t ino; 314 }; 315 316 static const size_t links_hash_initial_size = 8192; 317 static struct links_entry **buckets; 318 static struct links_entry *free_list; 319 static size_t number_buckets; 320 static unsigned long number_entries; 321 static char stop_allocating; 322 struct links_entry *le, **new_buckets; 323 struct stat *st; 324 size_t i, new_size; 325 int count, hash; 326 327 st = p->fts_statp; 328 329 /* If necessary, initialize the hash table. */ 330 if (buckets == NULL) { 331 number_buckets = links_hash_initial_size; 332 buckets = malloc(number_buckets * sizeof(buckets[0])); 333 if (buckets == NULL) 334 errx(1, "No memory for hardlink detection"); 335 for (i = 0; i < number_buckets; i++) 336 buckets[i] = NULL; 337 } 338 339 /* If the hash table is getting too full, enlarge it. */ 340 if (number_entries > number_buckets * 10 && !stop_allocating) { 341 new_size = number_buckets * 2; 342 new_buckets = malloc(new_size * sizeof(struct links_entry *)); 343 count = 0; 344 345 /* Try releasing the free list to see if that helps. */ 346 if (new_buckets == NULL && free_list != NULL) { 347 while (free_list != NULL) { 348 le = free_list; 349 free_list = le->next; 350 free(le); 351 } 352 new_buckets = malloc(new_size * sizeof(new_buckets[0])); 353 } 354 355 if (new_buckets == NULL) { 356 stop_allocating = 1; 357 warnx("No more memory for tracking hard links"); 358 } else { 359 memset(new_buckets, 0, new_size * sizeof(struct links_entry *)); 360 for (i = 0; i < number_buckets; i++) { 361 while (buckets[i] != NULL) { 362 /* Remove entry from old bucket. */ 363 le = buckets[i]; 364 buckets[i] = le->next; 365 366 /* Add entry to new bucket. */ 367 hash = (le->dev ^ le->ino) % new_size; 368 369 if (new_buckets[hash] != NULL) 370 new_buckets[hash]->previous = le; 371 le->next = new_buckets[hash]; 372 le->previous = NULL; 373 new_buckets[hash] = le; 374 } 375 } 376 free(buckets); 377 buckets = new_buckets; 378 number_buckets = new_size; 379 } 380 } 381 382 /* Try to locate this entry in the hash table. */ 383 hash = ( st->st_dev ^ st->st_ino ) % number_buckets; 384 for (le = buckets[hash]; le != NULL; le = le->next) { 385 if (le->dev == st->st_dev && le->ino == st->st_ino) { 386 /* 387 * Save memory by releasing an entry when we've seen 388 * all of it's links. 389 */ 390 if (--le->links <= 0) { 391 if (le->previous != NULL) 392 le->previous->next = le->next; 393 if (le->next != NULL) 394 le->next->previous = le->previous; 395 if (buckets[hash] == le) 396 buckets[hash] = le->next; 397 number_entries--; 398 /* Recycle this node through the free list */ 399 if (stop_allocating) { 400 free(le); 401 } else { 402 le->next = free_list; 403 free_list = le; 404 } 405 } 406 return (1); 407 } 408 } 409 410 if (stop_allocating) 411 return (0); 412 413 /* Add this entry to the links cache. */ 414 if (free_list != NULL) { 415 /* Pull a node from the free list if we can. */ 416 le = free_list; 417 free_list = le->next; 418 } else 419 /* Malloc one if we have to. */ 420 le = malloc(sizeof(struct links_entry)); 421 if (le == NULL) { 422 stop_allocating = 1; 423 warnx("No more memory for tracking hard links"); 424 return (0); 425 } 426 le->dev = st->st_dev; 427 le->ino = st->st_ino; 428 le->links = st->st_nlink - 1; 429 number_entries++; 430 le->next = buckets[hash]; 431 le->previous = NULL; 432 if (buckets[hash] != NULL) 433 buckets[hash]->previous = le; 434 buckets[hash] = le; 435 return (0); 436 } 437 438 /* 439 * Output in "human-readable" format. Uses 3 digits max and puts 440 * unit suffixes at the end. Makes output compact and easy to read, 441 * especially on huge disks. 442 * 443 */ 444 unit_t 445 unit_adjust(double *val) 446 { 447 double abval; 448 unit_t unit; 449 unsigned int unit_sz; 450 451 abval = fabs(*val); 452 453 unit_sz = abval ? ilogb(abval) / 10 : 0; 454 455 if (unit_sz >= UNIT_MAX) { 456 unit = NONE; 457 } else { 458 unit = unitp[unit_sz]; 459 *val /= (double)valp[unit_sz]; 460 } 461 462 return (unit); 463 } 464 465 void 466 prthumanval(double bytes) 467 { 468 unit_t unit; 469 470 bytes *= 512; 471 unit = unit_adjust(&bytes); 472 473 if (bytes == 0) 474 (void)printf(" 0B"); 475 else if (bytes > 10) 476 (void)printf("%3.0f%c", bytes, "BKMGTPE"[unit]); 477 else 478 (void)printf("%3.1f%c", bytes, "BKMGTPE"[unit]); 479 } 480 481 static void 482 usage(void) 483 { 484 (void)fprintf(stderr, 485 "usage: du [-H | -L | -P] [-a | -s | -d depth] [-c] [-h | -k] [-x] [-I mask] [file ...]\n"); 486 exit(EX_USAGE); 487 } 488 489 void 490 ignoreadd(const char *mask) 491 { 492 struct ignentry *ign; 493 494 ign = calloc(1, sizeof(*ign)); 495 if (ign == NULL) 496 errx(1, "cannot allocate memory"); 497 ign->mask = strdup(mask); 498 if (ign->mask == NULL) 499 errx(1, "cannot allocate memory"); 500 SLIST_INSERT_HEAD(&ignores, ign, next); 501 } 502 503 void 504 ignoreclean(void) 505 { 506 struct ignentry *ign; 507 508 while (!SLIST_EMPTY(&ignores)) { 509 ign = SLIST_FIRST(&ignores); 510 SLIST_REMOVE_HEAD(&ignores, next); 511 free(ign->mask); 512 free(ign); 513 } 514 } 515 516 int 517 ignorep(FTSENT *ent) 518 { 519 struct ignentry *ign; 520 521 SLIST_FOREACH(ign, &ignores, next) 522 if (fnmatch(ign->mask, ent->fts_name, 0) != FNM_NOMATCH) 523 return 1; 524 return 0; 525 } 526