file_cmds-184.tar.gz
[apple/file_cmds.git] / du / du.c
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
37 #ifndef lint
38 static const char copyright[] =
39 "@(#) Copyright (c) 1989, 1993, 1994\n\
40 The Regents of the University of California. All rights reserved.\n";
41 #endif /* not lint */
42
43 #ifndef lint
44 #if 0
45 static const char sccsid[] = "@(#)du.c 8.5 (Berkeley) 5/4/95";
46 #endif
47 #endif /* not lint */
48 #include <sys/cdefs.h>
49 __FBSDID("$FreeBSD: src/usr.bin/du/du.c,v 1.38 2005/04/09 14:31:40 stefanf Exp $");
50
51 #include <sys/mount.h>
52 #include <sys/param.h>
53 #include <sys/queue.h>
54 #include <sys/stat.h>
55
56 #include <err.h>
57 #include <errno.h>
58 #include <fnmatch.h>
59 #include <fts.h>
60 #include <locale.h>
61 #include <math.h>
62 #include <stdint.h>
63 #include <stdio.h>
64 #include <stdlib.h>
65 #include <string.h>
66 #include <sysexits.h>
67 #include <unistd.h>
68
69 #ifdef __APPLE__
70 #include <get_compat.h>
71 #else
72 #define COMPAT_MODE(func, mode) (1)
73 #endif
74
75 #define KILO_SZ(n) (n)
76 #define MEGA_SZ(n) ((n) * (n))
77 #define GIGA_SZ(n) ((n) * (n) * (n))
78 #define TERA_SZ(n) ((n) * (n) * (n) * (n))
79 #define PETA_SZ(n) ((n) * (n) * (n) * (n) * (n))
80
81 #define KILO_2_SZ (KILO_SZ(1024ULL))
82 #define MEGA_2_SZ (MEGA_SZ(1024ULL))
83 #define GIGA_2_SZ (GIGA_SZ(1024ULL))
84 #define TERA_2_SZ (TERA_SZ(1024ULL))
85 #define PETA_2_SZ (PETA_SZ(1024ULL))
86
87 #define KILO_SI_SZ (KILO_SZ(1000ULL))
88 #define MEGA_SI_SZ (MEGA_SZ(1000ULL))
89 #define GIGA_SI_SZ (GIGA_SZ(1000ULL))
90 #define TERA_SI_SZ (TERA_SZ(1000ULL))
91 #define PETA_SI_SZ (PETA_SZ(1000ULL))
92
93 #define TWO_TB (2LL * 1024LL * 1024LL * 1024LL * 1024LL)
94
95 unsigned long long vals_si [] = {1, KILO_SI_SZ, MEGA_SI_SZ, GIGA_SI_SZ, TERA_SI_SZ, PETA_SI_SZ};
96 unsigned long long vals_base2[] = {1, KILO_2_SZ, MEGA_2_SZ, GIGA_2_SZ, TERA_2_SZ, PETA_2_SZ};
97 unsigned long long *valp;
98
99 typedef enum { NONE, KILO, MEGA, GIGA, TERA, PETA, UNIT_MAX } unit_t;
100
101 int unitp [] = { NONE, KILO, MEGA, GIGA, TERA, PETA };
102
103 SLIST_HEAD(ignhead, ignentry) ignores;
104 struct ignentry {
105 char *mask;
106 SLIST_ENTRY(ignentry) next;
107 };
108
109 static int linkchk(FTSENT *);
110 static void usage(void);
111 void prthumanval(double);
112 unit_t unit_adjust(double *);
113 void ignoreadd(const char *);
114 void ignoreclean(void);
115 int ignorep(FTSENT *);
116
117 int
118 main(int argc, char *argv[])
119 {
120 FTS *fts;
121 FTSENT *p;
122 off_t savednumber = 0;
123 long blocksize;
124 int ftsoptions;
125 int listall;
126 int depth;
127 int Hflag, Lflag, Pflag, aflag, sflag, dflag, cflag, hflag, ch, notused, rval;
128 char **save;
129 static char dot[] = ".";
130 off_t *ftsnum, *ftsparnum;
131
132 setlocale(LC_ALL, "");
133
134 Hflag = Lflag = Pflag = aflag = sflag = dflag = cflag = hflag = 0;
135
136 save = argv;
137 ftsoptions = 0;
138 depth = INT_MAX;
139 SLIST_INIT(&ignores);
140
141 while ((ch = getopt(argc, argv, "HI:LPasd:chkmrx")) != -1)
142 switch (ch) {
143 case 'H':
144 Lflag = Pflag = 0;
145 Hflag = 1;
146 break;
147 case 'I':
148 ignoreadd(optarg);
149 break;
150 case 'L':
151 Hflag = Pflag = 0;
152 Lflag = 1;
153 break;
154 case 'P':
155 Hflag = Lflag = 0;
156 Pflag = 1;
157 break;
158 case 'a':
159 aflag = 1;
160 break;
161 case 's':
162 sflag = 1;
163 break;
164 case 'd':
165 dflag = 1;
166 errno = 0;
167 depth = atoi(optarg);
168 if (errno == ERANGE || depth < 0) {
169 warnx("invalid argument to option d: %s", optarg);
170 usage();
171 }
172 break;
173 case 'c':
174 cflag = 1;
175 break;
176 case 'h':
177 putenv("BLOCKSIZE=512");
178 hflag = 1;
179 valp = vals_base2;
180 break;
181 case 'k':
182 hflag = 0;
183 putenv("BLOCKSIZE=1024");
184 break;
185 case 'm':
186 hflag = 0;
187 putenv("BLOCKSIZE=1048576");
188 break;
189 case 'r': /* Compatibility. */
190 break;
191 case 'x':
192 ftsoptions |= FTS_XDEV;
193 break;
194 case '?':
195 default:
196 usage();
197 }
198
199 argc -= optind;
200 argv += optind;
201
202 /*
203 * XXX
204 * Because of the way that fts(3) works, logical walks will not count
205 * the blocks actually used by symbolic links. We rationalize this by
206 * noting that users computing logical sizes are likely to do logical
207 * copies, so not counting the links is correct. The real reason is
208 * that we'd have to re-implement the kernel's symbolic link traversing
209 * algorithm to get this right. If, for example, you have relative
210 * symbolic links referencing other relative symbolic links, it gets
211 * very nasty, very fast. The bottom line is that it's documented in
212 * the man page, so it's a feature.
213 */
214
215 if (Hflag + Lflag + Pflag > 1)
216 usage();
217
218 if (Hflag + Lflag + Pflag == 0)
219 Pflag = 1; /* -P (physical) is default */
220
221 if (Hflag)
222 ftsoptions |= FTS_COMFOLLOW;
223
224 if (Lflag)
225 ftsoptions |= FTS_LOGICAL;
226
227 if (Pflag)
228 ftsoptions |= FTS_PHYSICAL;
229
230 listall = 0;
231
232 if (aflag) {
233 if (sflag || dflag)
234 usage();
235 listall = 1;
236 } else if (sflag) {
237 if (dflag)
238 usage();
239 depth = 0;
240 }
241
242 if (!*argv) {
243 argv = save;
244 argv[0] = dot;
245 argv[1] = NULL;
246 }
247
248 (void) getbsize(&notused, &blocksize);
249 blocksize /= 512;
250
251 rval = 0;
252
253 if ((fts = fts_open(argv, ftsoptions, NULL)) == NULL)
254 err(1, "fts_open");
255
256 while ((p = fts_read(fts)) != NULL) {
257 switch (p->fts_info) {
258 case FTS_D: /* Ignore. */
259 if (ignorep(p))
260 fts_set(fts, p, FTS_SKIP);
261 break;
262 case FTS_DP:
263 if (ignorep(p))
264 break;
265
266 ftsparnum = (off_t *)&p->fts_parent->fts_number;
267 ftsnum = (off_t *)&p->fts_number;
268 if (p->fts_statp->st_size < TWO_TB) {
269 ftsparnum[0] += ftsnum[0] += p->fts_statp->st_blocks;
270 } else {
271 ftsparnum[0] += ftsnum[0] += howmany(p->fts_statp->st_size, 512LL);
272 }
273
274 if (p->fts_level <= depth) {
275 if (hflag) {
276 (void) prthumanval(howmany(*ftsnum, blocksize));
277 (void) printf("\t%s\n", p->fts_path);
278 } else {
279 (void) printf("%jd\t%s\n",
280 (intmax_t)howmany(*ftsnum, blocksize),
281 p->fts_path);
282 }
283 }
284 break;
285 case FTS_DC: /* Ignore. */
286 if (COMPAT_MODE("bin/du", "unix2003")) {
287 errx(1, "Can't follow symlink cycle from %s to %s", p->fts_path, p->fts_cycle->fts_path);
288 }
289 break;
290 case FTS_DNR: /* Warn, continue. */
291 case FTS_ERR:
292 case FTS_NS:
293 warnx("%s: %s", p->fts_path, strerror(p->fts_errno));
294 rval = 1;
295 break;
296 case FTS_SLNONE:
297 if (COMPAT_MODE("bin/du", "unix2003")) {
298 struct stat sb;
299 int rc = stat(p->fts_path, &sb);
300 if (rc < 0 && errno == ELOOP) {
301 errx(1, "Too many symlinks at %s", p->fts_path);
302 }
303 }
304 default:
305 if (ignorep(p))
306 break;
307
308 if (p->fts_statp->st_nlink > 1 && linkchk(p))
309 break;
310
311 if (listall || p->fts_level == 0) {
312 if (hflag) {
313 if (p->fts_statp->st_size < TWO_TB) {
314 (void) prthumanval(howmany(p->fts_statp->st_blocks,
315 blocksize));
316 } else {
317 (void) prthumanval(howmany(howmany(p->fts_statp->st_size, 512LL),
318 blocksize));
319 }
320 (void) printf("\t%s\n", p->fts_path);
321 } else {
322 if (p->fts_statp->st_size < TWO_TB) {
323 (void) printf("%jd\t%s\n",
324 (intmax_t)howmany(p->fts_statp->st_blocks, blocksize),
325 p->fts_path);
326 } else {
327 (void) printf("%jd\t%s\n",
328 (intmax_t)howmany(howmany(p->fts_statp->st_size, 512LL), blocksize),
329 p->fts_path);
330 }
331 }
332 }
333
334 ftsparnum = (off_t *)&p->fts_parent->fts_number;
335 if (p->fts_statp->st_size < TWO_TB) {
336 ftsparnum[0] += p->fts_statp->st_blocks;
337 } else {
338 ftsparnum[0] += p->fts_statp->st_size / 512LL;
339 }
340 }
341 savednumber = ((off_t *)&p->fts_parent->fts_number)[0];
342 }
343
344 if (errno)
345 err(1, "fts_read");
346
347 if (cflag) {
348 if (hflag) {
349 (void) prthumanval(howmany(savednumber, blocksize));
350 (void) printf("\ttotal\n");
351 } else {
352 (void) printf("%jd\ttotal\n", (intmax_t)howmany(savednumber, blocksize));
353 }
354 }
355
356 ignoreclean();
357 exit(rval);
358 }
359
360 static int
361 linkchk(FTSENT *p)
362 {
363 struct links_entry {
364 struct links_entry *next;
365 struct links_entry *previous;
366 int links;
367 dev_t dev;
368 ino_t ino;
369 };
370 static const size_t links_hash_initial_size = 8192;
371 static struct links_entry **buckets;
372 static struct links_entry *free_list;
373 static size_t number_buckets;
374 static unsigned long number_entries;
375 static char stop_allocating;
376 struct links_entry *le, **new_buckets;
377 struct stat *st;
378 size_t i, new_size;
379 int hash;
380
381 st = p->fts_statp;
382
383 /* If necessary, initialize the hash table. */
384 if (buckets == NULL) {
385 number_buckets = links_hash_initial_size;
386 buckets = malloc(number_buckets * sizeof(buckets[0]));
387 if (buckets == NULL)
388 errx(1, "No memory for hardlink detection");
389 for (i = 0; i < number_buckets; i++)
390 buckets[i] = NULL;
391 }
392
393 /* If the hash table is getting too full, enlarge it. */
394 if (number_entries > number_buckets * 10 && !stop_allocating) {
395 new_size = number_buckets * 2;
396 new_buckets = malloc(new_size * sizeof(struct links_entry *));
397
398 /* Try releasing the free list to see if that helps. */
399 if (new_buckets == NULL && free_list != NULL) {
400 while (free_list != NULL) {
401 le = free_list;
402 free_list = le->next;
403 free(le);
404 }
405 new_buckets = malloc(new_size * sizeof(new_buckets[0]));
406 }
407
408 if (new_buckets == NULL) {
409 stop_allocating = 1;
410 warnx("No more memory for tracking hard links");
411 } else {
412 memset(new_buckets, 0,
413 new_size * sizeof(struct links_entry *));
414 for (i = 0; i < number_buckets; i++) {
415 while (buckets[i] != NULL) {
416 /* Remove entry from old bucket. */
417 le = buckets[i];
418 buckets[i] = le->next;
419
420 /* Add entry to new bucket. */
421 hash = (le->dev ^ le->ino) % new_size;
422
423 if (new_buckets[hash] != NULL)
424 new_buckets[hash]->previous =
425 le;
426 le->next = new_buckets[hash];
427 le->previous = NULL;
428 new_buckets[hash] = le;
429 }
430 }
431 free(buckets);
432 buckets = new_buckets;
433 number_buckets = new_size;
434 }
435 }
436
437 /* Try to locate this entry in the hash table. */
438 hash = ( st->st_dev ^ st->st_ino ) % number_buckets;
439 for (le = buckets[hash]; le != NULL; le = le->next) {
440 if (le->dev == st->st_dev && le->ino == st->st_ino) {
441 /*
442 * Save memory by releasing an entry when we've seen
443 * all of it's links.
444 */
445 if (--le->links <= 0) {
446 if (le->previous != NULL)
447 le->previous->next = le->next;
448 if (le->next != NULL)
449 le->next->previous = le->previous;
450 if (buckets[hash] == le)
451 buckets[hash] = le->next;
452 number_entries--;
453 /* Recycle this node through the free list */
454 if (stop_allocating) {
455 free(le);
456 } else {
457 le->next = free_list;
458 free_list = le;
459 }
460 }
461 return (1);
462 }
463 }
464
465 if (stop_allocating)
466 return (0);
467
468 /* Add this entry to the links cache. */
469 if (free_list != NULL) {
470 /* Pull a node from the free list if we can. */
471 le = free_list;
472 free_list = le->next;
473 } else
474 /* Malloc one if we have to. */
475 le = malloc(sizeof(struct links_entry));
476 if (le == NULL) {
477 stop_allocating = 1;
478 warnx("No more memory for tracking hard links");
479 return (0);
480 }
481 le->dev = st->st_dev;
482 le->ino = st->st_ino;
483 le->links = st->st_nlink - 1;
484 number_entries++;
485 le->next = buckets[hash];
486 le->previous = NULL;
487 if (buckets[hash] != NULL)
488 buckets[hash]->previous = le;
489 buckets[hash] = le;
490 return (0);
491 }
492
493 /*
494 * Output in "human-readable" format. Uses 3 digits max and puts
495 * unit suffixes at the end. Makes output compact and easy to read,
496 * especially on huge disks.
497 *
498 */
499 unit_t
500 unit_adjust(double *val)
501 {
502 double abval;
503 unit_t unit;
504 unsigned int unit_sz;
505
506 abval = fabs(*val);
507
508 unit_sz = abval ? ilogb(abval) / 10 : 0;
509
510 if (unit_sz >= UNIT_MAX) {
511 unit = NONE;
512 } else {
513 unit = unitp[unit_sz];
514 *val /= (double)valp[unit_sz];
515 }
516
517 return (unit);
518 }
519
520 void
521 prthumanval(double bytes)
522 {
523 unit_t unit;
524
525 bytes *= 512;
526 unit = unit_adjust(&bytes);
527
528 if (bytes == 0)
529 (void)printf(" 0B");
530 else if (bytes > 10)
531 (void)printf("%3.0f%c", bytes, "BKMGTPE"[unit]);
532 else
533 (void)printf("%3.1f%c", bytes, "BKMGTPE"[unit]);
534 }
535
536 static void
537 usage(void)
538 {
539 (void)fprintf(stderr,
540 "usage: du [-H | -L | -P] [-a | -s | -d depth] [-c] [-h | -k | -m] [-x] [-I mask] [file ...]\n");
541 exit(EX_USAGE);
542 }
543
544 void
545 ignoreadd(const char *mask)
546 {
547 struct ignentry *ign;
548
549 ign = calloc(1, sizeof(*ign));
550 if (ign == NULL)
551 errx(1, "cannot allocate memory");
552 ign->mask = strdup(mask);
553 if (ign->mask == NULL)
554 errx(1, "cannot allocate memory");
555 SLIST_INSERT_HEAD(&ignores, ign, next);
556 }
557
558 void
559 ignoreclean(void)
560 {
561 struct ignentry *ign;
562
563 while (!SLIST_EMPTY(&ignores)) {
564 ign = SLIST_FIRST(&ignores);
565 SLIST_REMOVE_HEAD(&ignores, next);
566 free(ign->mask);
567 free(ign);
568 }
569 }
570
571 int
572 ignorep(FTSENT *ent)
573 {
574 struct ignentry *ign;
575
576 #ifdef __APPLE__
577 if (S_ISDIR(ent->fts_statp->st_mode) && !strcmp("fd", ent->fts_name)) {
578 struct statfs sfsb;
579 int rc = statfs(ent->fts_accpath, &sfsb);
580 if (rc >= 0 && !strcmp("fdesc", sfsb.f_fstypename)) {
581 /* Don't cd into /dev/fd/N since one of those is likely to be
582 the cwd as of the start of du which causes all manner of
583 unpleasant surprises */
584 return 1;
585 }
586 }
587 #endif /* __APPLE__ */
588 SLIST_FOREACH(ign, &ignores, next)
589 if (fnmatch(ign->mask, ent->fts_name, 0) != FNM_NOMATCH)
590 return 1;
591 return 0;
592 }