xref: /freebsd/usr.bin/du/du.c (revision 1669d8afc64812c8d2d1d147ae1fd42ff441e1b1)
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("$FreeBSD$");
50 
51 #include <sys/param.h>
52 #include <sys/queue.h>
53 #include <sys/stat.h>
54 
55 #include <err.h>
56 #include <errno.h>
57 #include <fnmatch.h>
58 #include <fts.h>
59 #include <libutil.h>
60 #include <locale.h>
61 #include <stdint.h>
62 #include <stdio.h>
63 #include <stdlib.h>
64 #include <string.h>
65 #include <sysexits.h>
66 #include <unistd.h>
67 
68 SLIST_HEAD(ignhead, ignentry) ignores;
69 struct ignentry {
70 	char			*mask;
71 	SLIST_ENTRY(ignentry)	next;
72 };
73 
74 static int	linkchk(FTSENT *);
75 static void	usage(void);
76 void		prthumanval(int64_t);
77 void		ignoreadd(const char *);
78 void		ignoreclean(void);
79 int		ignorep(FTSENT *);
80 
81 int		nodumpflag = 0;
82 
83 int
84 main(int argc, char *argv[])
85 {
86 	FTS		*fts;
87 	FTSENT		*p;
88 	off_t		savednumber = 0;
89 	long		blocksize;
90 	int		ftsoptions;
91 	int		listall;
92 	int		depth;
93 	int		Hflag, Lflag, Pflag, aflag, sflag, dflag, cflag, hflag, ch, notused, rval;
94 	char 		**save;
95 	static char	dot[] = ".";
96 
97 	setlocale(LC_ALL, "");
98 
99 	Hflag = Lflag = Pflag = aflag = sflag = dflag = cflag = hflag = 0;
100 
101 	save = argv;
102 	ftsoptions = 0;
103 	depth = INT_MAX;
104 	SLIST_INIT(&ignores);
105 
106 	while ((ch = getopt(argc, argv, "HI:LPasd:chkmnrx")) != -1)
107 		switch (ch) {
108 			case 'H':
109 				Hflag = 1;
110 				break;
111 			case 'I':
112 				ignoreadd(optarg);
113 				break;
114 			case 'L':
115 				if (Pflag)
116 					usage();
117 				Lflag = 1;
118 				break;
119 			case 'P':
120 				if (Lflag)
121 					usage();
122 				Pflag = 1;
123 				break;
124 			case 'a':
125 				aflag = 1;
126 				break;
127 			case 's':
128 				sflag = 1;
129 				break;
130 			case 'd':
131 				dflag = 1;
132 				errno = 0;
133 				depth = atoi(optarg);
134 				if (errno == ERANGE || depth < 0) {
135 					warnx("invalid argument to option d: %s", optarg);
136 					usage();
137 				}
138 				break;
139 			case 'c':
140 				cflag = 1;
141 				break;
142 			case 'h':
143 				if (setenv("BLOCKSIZE", "512", 1) == -1)
144 					warn(
145 					    "setenv: cannot set BLOCKSIZE=512");
146 				hflag = 1;
147 				break;
148 			case 'k':
149 				hflag = 0;
150 				if (setenv("BLOCKSIZE", "1024", 1) == -1)
151 					warn("setenv: cannot set BLOCKSIZE=1024");
152 				break;
153 			case 'm':
154 				hflag = 0;
155 				if (setenv("BLOCKSIZE", "1048576", 1) == -1)
156 					warn("setenv: cannot set BLOCKSIZE=1048576");
157 				break;
158 			case 'n':
159 				nodumpflag = 1;
160 				break;
161 			case 'r':		 /* Compatibility. */
162 				break;
163 			case 'x':
164 				ftsoptions |= FTS_XDEV;
165 				break;
166 			case '?':
167 			default:
168 				usage();
169 		}
170 
171 	argc -= optind;
172 	argv += optind;
173 
174 	/*
175 	 * XXX
176 	 * Because of the way that fts(3) works, logical walks will not count
177 	 * the blocks actually used by symbolic links.  We rationalize this by
178 	 * noting that users computing logical sizes are likely to do logical
179 	 * copies, so not counting the links is correct.  The real reason is
180 	 * that we'd have to re-implement the kernel's symbolic link traversing
181 	 * algorithm to get this right.  If, for example, you have relative
182 	 * symbolic links referencing other relative symbolic links, it gets
183 	 * very nasty, very fast.  The bottom line is that it's documented in
184 	 * the man page, so it's a feature.
185 	 */
186 
187 	if (Hflag + Lflag + Pflag > 1)
188 		usage();
189 
190 	if (Hflag + Lflag + Pflag == 0)
191 		Pflag = 1;			/* -P (physical) is default */
192 
193 	if (Hflag)
194 		ftsoptions |= FTS_COMFOLLOW;
195 
196 	if (Lflag)
197 		ftsoptions |= FTS_LOGICAL;
198 
199 	if (Pflag)
200 		ftsoptions |= FTS_PHYSICAL;
201 
202 	listall = 0;
203 
204 	if (aflag) {
205 		if (sflag || dflag)
206 			usage();
207 		listall = 1;
208 	} else if (sflag) {
209 		if (dflag)
210 			usage();
211 		depth = 0;
212 	}
213 
214 	if (!*argv) {
215 		argv = save;
216 		argv[0] = dot;
217 		argv[1] = NULL;
218 	}
219 
220 	(void) getbsize(&notused, &blocksize);
221 	blocksize /= 512;
222 
223 	rval = 0;
224 
225 	if ((fts = fts_open(argv, ftsoptions, NULL)) == NULL)
226 		err(1, "fts_open");
227 
228 	while ((p = fts_read(fts)) != NULL) {
229 		switch (p->fts_info) {
230 			case FTS_D:			/* Ignore. */
231 				if (ignorep(p))
232 					fts_set(fts, p, FTS_SKIP);
233 				break;
234 			case FTS_DP:
235 				if (ignorep(p))
236 					break;
237 
238 				p->fts_parent->fts_bignum +=
239 				    p->fts_bignum += p->fts_statp->st_blocks;
240 
241 				if (p->fts_level <= depth) {
242 					if (hflag) {
243 						(void) prthumanval(howmany(p->fts_bignum, blocksize));
244 						(void) printf("\t%s\n", p->fts_path);
245 					} else {
246 					(void) printf("%jd\t%s\n",
247 					    (intmax_t)howmany(p->fts_bignum, blocksize),
248 					    p->fts_path);
249 					}
250 				}
251 				break;
252 			case FTS_DC:			/* Ignore. */
253 				break;
254 			case FTS_DNR:			/* Warn, continue. */
255 			case FTS_ERR:
256 			case FTS_NS:
257 				warnx("%s: %s", p->fts_path, strerror(p->fts_errno));
258 				rval = 1;
259 				break;
260 			default:
261 				if (ignorep(p))
262 					break;
263 
264 				if (p->fts_statp->st_nlink > 1 && linkchk(p))
265 					break;
266 
267 				if (listall || p->fts_level == 0) {
268 					if (hflag) {
269 						(void) prthumanval(howmany(p->fts_statp->st_blocks,
270 							blocksize));
271 						(void) printf("\t%s\n", p->fts_path);
272 					} else {
273 						(void) printf("%jd\t%s\n",
274 							(intmax_t)howmany(p->fts_statp->st_blocks, blocksize),
275 							p->fts_path);
276 					}
277 				}
278 
279 				p->fts_parent->fts_bignum += p->fts_statp->st_blocks;
280 		}
281 		savednumber = p->fts_parent->fts_bignum;
282 	}
283 
284 	if (errno)
285 		err(1, "fts_read");
286 
287 	if (cflag) {
288 		if (hflag) {
289 			(void) prthumanval(howmany(savednumber, blocksize));
290 			(void) printf("\ttotal\n");
291 		} else {
292 			(void) printf("%jd\ttotal\n", (intmax_t)howmany(savednumber, blocksize));
293 		}
294 	}
295 
296 	ignoreclean();
297 	exit(rval);
298 }
299 
300 static int
301 linkchk(FTSENT *p)
302 {
303 	struct links_entry {
304 		struct links_entry *next;
305 		struct links_entry *previous;
306 		int	 links;
307 		dev_t	 dev;
308 		ino_t	 ino;
309 	};
310 	static const size_t links_hash_initial_size = 8192;
311 	static struct links_entry **buckets;
312 	static struct links_entry *free_list;
313 	static size_t number_buckets;
314 	static unsigned long number_entries;
315 	static char stop_allocating;
316 	struct links_entry *le, **new_buckets;
317 	struct stat *st;
318 	size_t i, new_size;
319 	int hash;
320 
321 	st = p->fts_statp;
322 
323 	/* If necessary, initialize the hash table. */
324 	if (buckets == NULL) {
325 		number_buckets = links_hash_initial_size;
326 		buckets = malloc(number_buckets * sizeof(buckets[0]));
327 		if (buckets == NULL)
328 			errx(1, "No memory for hardlink detection");
329 		for (i = 0; i < number_buckets; i++)
330 			buckets[i] = NULL;
331 	}
332 
333 	/* If the hash table is getting too full, enlarge it. */
334 	if (number_entries > number_buckets * 10 && !stop_allocating) {
335 		new_size = number_buckets * 2;
336 		new_buckets = malloc(new_size * sizeof(struct links_entry *));
337 
338 		/* Try releasing the free list to see if that helps. */
339 		if (new_buckets == NULL && free_list != NULL) {
340 			while (free_list != NULL) {
341 				le = free_list;
342 				free_list = le->next;
343 				free(le);
344 			}
345 			new_buckets = malloc(new_size * sizeof(new_buckets[0]));
346 		}
347 
348 		if (new_buckets == NULL) {
349 			stop_allocating = 1;
350 			warnx("No more memory for tracking hard links");
351 		} else {
352 			memset(new_buckets, 0,
353 			    new_size * sizeof(struct links_entry *));
354 			for (i = 0; i < number_buckets; i++) {
355 				while (buckets[i] != NULL) {
356 					/* Remove entry from old bucket. */
357 					le = buckets[i];
358 					buckets[i] = le->next;
359 
360 					/* Add entry to new bucket. */
361 					hash = (le->dev ^ le->ino) % new_size;
362 
363 					if (new_buckets[hash] != NULL)
364 						new_buckets[hash]->previous =
365 						    le;
366 					le->next = new_buckets[hash];
367 					le->previous = NULL;
368 					new_buckets[hash] = le;
369 				}
370 			}
371 			free(buckets);
372 			buckets = new_buckets;
373 			number_buckets = new_size;
374 		}
375 	}
376 
377 	/* Try to locate this entry in the hash table. */
378 	hash = ( st->st_dev ^ st->st_ino ) % number_buckets;
379 	for (le = buckets[hash]; le != NULL; le = le->next) {
380 		if (le->dev == st->st_dev && le->ino == st->st_ino) {
381 			/*
382 			 * Save memory by releasing an entry when we've seen
383 			 * all of it's links.
384 			 */
385 			if (--le->links <= 0) {
386 				if (le->previous != NULL)
387 					le->previous->next = le->next;
388 				if (le->next != NULL)
389 					le->next->previous = le->previous;
390 				if (buckets[hash] == le)
391 					buckets[hash] = le->next;
392 				number_entries--;
393 				/* Recycle this node through the free list */
394 				if (stop_allocating) {
395 					free(le);
396 				} else {
397 					le->next = free_list;
398 					free_list = le;
399 				}
400 			}
401 			return (1);
402 		}
403 	}
404 
405 	if (stop_allocating)
406 		return (0);
407 
408 	/* Add this entry to the links cache. */
409 	if (free_list != NULL) {
410 		/* Pull a node from the free list if we can. */
411 		le = free_list;
412 		free_list = le->next;
413 	} else
414 		/* Malloc one if we have to. */
415 		le = malloc(sizeof(struct links_entry));
416 	if (le == NULL) {
417 		stop_allocating = 1;
418 		warnx("No more memory for tracking hard links");
419 		return (0);
420 	}
421 	le->dev = st->st_dev;
422 	le->ino = st->st_ino;
423 	le->links = st->st_nlink - 1;
424 	number_entries++;
425 	le->next = buckets[hash];
426 	le->previous = NULL;
427 	if (buckets[hash] != NULL)
428 		buckets[hash]->previous = le;
429 	buckets[hash] = le;
430 	return (0);
431 }
432 
433 void
434 prthumanval(int64_t bytes)
435 {
436 	char buf[5];
437 
438 	bytes *= DEV_BSIZE;
439 
440 	humanize_number(buf, sizeof(buf), bytes, "", HN_AUTOSCALE,
441 	    HN_B | HN_NOSPACE | HN_DECIMAL);
442 
443 	(void)printf("%4s", buf);
444 }
445 
446 static void
447 usage(void)
448 {
449 	(void)fprintf(stderr,
450 		"usage: du [-H | -L | -P] [-a | -s | -d depth] [-c] [-h | -k | -m] [-n] [-x] [-I mask] [file ...]\n");
451 	exit(EX_USAGE);
452 }
453 
454 void
455 ignoreadd(const char *mask)
456 {
457 	struct ignentry *ign;
458 
459 	ign = calloc(1, sizeof(*ign));
460 	if (ign == NULL)
461 		errx(1, "cannot allocate memory");
462 	ign->mask = strdup(mask);
463 	if (ign->mask == NULL)
464 		errx(1, "cannot allocate memory");
465 	SLIST_INSERT_HEAD(&ignores, ign, next);
466 }
467 
468 void
469 ignoreclean(void)
470 {
471 	struct ignentry *ign;
472 
473 	while (!SLIST_EMPTY(&ignores)) {
474 		ign = SLIST_FIRST(&ignores);
475 		SLIST_REMOVE_HEAD(&ignores, next);
476 		free(ign->mask);
477 		free(ign);
478 	}
479 }
480 
481 int
482 ignorep(FTSENT *ent)
483 {
484 	struct ignentry *ign;
485 
486 	if (nodumpflag && (ent->fts_statp->st_flags & UF_NODUMP))
487 		return 1;
488 	SLIST_FOREACH(ign, &ignores, next)
489 		if (fnmatch(ign->mask, ent->fts_name, 0) != FNM_NOMATCH)
490 			return 1;
491 	return 0;
492 }
493