xref: /dragonfly/usr.bin/du/du.c (revision f746689a)
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 1.12 2008/08/12 03:35:35 y0netan1 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 <libutil.h>
51 #include <stdio.h>
52 #include <stdlib.h>
53 #include <string.h>
54 #include <sysexits.h>
55 #include <unistd.h>
56 
57 #define	HASHSIZE	256		/* power of 2 only */
58 #define HASHMASK	(HASHSIZE - 1)
59 
60 SLIST_HEAD(ignhead, ignentry) ignores;
61 struct ignentry {
62 	char			*mask;
63 	SLIST_ENTRY(ignentry)	next;
64 };
65 
66 static int	linkchk(FTSENT *);
67 static void	usage(void);
68 void		prthumanval(int64_t);
69 void		ignoreadd(const char *);
70 void		ignoreclean(void);
71 int		ignorep(FTSENT *);
72 
73 static char period[] = ".";
74 
75 typedef long long	du_number_t;
76 
77 int
78 main(int argc, char **argv)
79 {
80 	FTS		*fts;
81 	FTSENT		*p;
82 	long		blocksize;
83 	du_number_t	savednumber = 0;
84 	int		ftsoptions;
85 	int		listall;
86 	int		depth;
87 	int		Hflag, Lflag, Pflag, aflag, sflag, dflag, cflag, hflag, ch, notused, rval;
88 	char 		**save;
89 
90 	Hflag = Lflag = Pflag = aflag = sflag = dflag = cflag = hflag = 0;
91 
92 	save = argv;
93 	ftsoptions = 0;
94 	depth = INT_MAX;
95 	SLIST_INIT(&ignores);
96 
97 	while ((ch = getopt(argc, argv, "HI:LPasd:chkrx")) != -1)
98 		switch (ch) {
99 			case 'H':
100 				Hflag = 1;
101 				break;
102 			case 'I':
103 				ignoreadd(optarg);
104 				break;
105 			case 'L':
106 				if (Pflag)
107 					usage();
108 				Lflag = 1;
109 				break;
110 			case 'P':
111 				if (Lflag)
112 					usage();
113 				Pflag = 1;
114 				break;
115 			case 'a':
116 				aflag = 1;
117 				break;
118 			case 's':
119 				sflag = 1;
120 				break;
121 			case 'd':
122 				dflag = 1;
123 				errno = 0;
124 				depth = atoi(optarg);
125 				if (errno == ERANGE || depth < 0) {
126 					warnx("invalid argument to option d: %s", optarg);
127 					usage();
128 				}
129 				break;
130 			case 'c':
131 				cflag = 1;
132 				break;
133 			case 'h':
134 				if (setenv("BLOCKSIZE", "512", 1) == -1)
135 					warn("setenv: cannot set BLOCKSIZE=512");
136 				hflag = 1;
137 				break;
138 			case 'k':
139 				hflag = 0;
140 				if (setenv("BLOCKSIZE", "1024", 1) == -1)
141 					warn("setenv: cannot set BLOCKSIZE=1024");
142 				break;
143 			case 'r':		 /* Compatibility. */
144 				break;
145 			case 'x':
146 				ftsoptions |= FTS_XDEV;
147 				break;
148 			case '?':
149 			default:
150 				usage();
151 		}
152 
153 	argc -= optind;
154 	argv += optind;
155 
156 	/*
157 	 * XXX
158 	 * Because of the way that fts(3) works, logical walks will not count
159 	 * the blocks actually used by symbolic links.  We rationalize this by
160 	 * noting that users computing logical sizes are likely to do logical
161 	 * copies, so not counting the links is correct.  The real reason is
162 	 * that we'd have to re-implement the kernel's symbolic link traversing
163 	 * algorithm to get this right.  If, for example, you have relative
164 	 * symbolic links referencing other relative symbolic links, it gets
165 	 * very nasty, very fast.  The bottom line is that it's documented in
166 	 * the man page, so it's a feature.
167 	 */
168 
169 	if (Hflag + Lflag + Pflag > 1)
170 		usage();
171 
172 	if (Hflag + Lflag + Pflag == 0)
173 		Pflag = 1;			/* -P (physical) is default */
174 
175 	if (Hflag)
176 		ftsoptions |= FTS_COMFOLLOW;
177 
178 	if (Lflag)
179 		ftsoptions |= FTS_LOGICAL;
180 
181 	if (Pflag)
182 		ftsoptions |= FTS_PHYSICAL;
183 
184 	listall = 0;
185 
186 	if (aflag) {
187 		if (sflag || dflag)
188 			usage();
189 		listall = 1;
190 	} else if (sflag) {
191 		if (dflag)
192 			usage();
193 		depth = 0;
194 	}
195 
196 	if (!*argv) {
197 		argv = save;
198 		argv[0] = period;
199 		argv[1] = NULL;
200 	}
201 
202 	(void) getbsize(&notused, &blocksize);
203 	blocksize /= 512;
204 
205 	rval = 0;
206 
207 	if ((fts = fts_open(argv, ftsoptions, NULL)) == NULL)
208 		err(1, "fts_open");
209 
210 	while ((p = fts_read(fts)) != NULL) {
211 		switch (p->fts_info) {
212 			case FTS_D:			/* Ignore. */
213 				if (ignorep(p))
214 					fts_set(fts, p, FTS_SKIP);
215 				break;
216 			case FTS_DP:
217 				if (ignorep(p))
218 					break;
219 
220 				if (p->fts_pointer == NULL) {
221 					p->fts_pointer = malloc(sizeof(du_number_t));
222 					*(du_number_t *)p->fts_pointer = 0;
223 				}
224 				*(du_number_t *)p->fts_pointer += p->fts_statp->st_blocks;
225 
226 				if (p->fts_parent->fts_pointer == NULL) {
227 					p->fts_parent->fts_pointer = malloc(sizeof(du_number_t));
228 					*(du_number_t *)p->fts_parent->fts_pointer = 0;
229 				}
230 				*(du_number_t *)p->fts_parent->fts_pointer += *(du_number_t *)p->fts_pointer += p->fts_statp->st_blocks;
231 
232 				if (p->fts_level <= depth) {
233 					if (hflag) {
234 						(void) prthumanval(howmany(*(du_number_t *)p->fts_pointer, blocksize));
235 						(void) printf("\t%s\n", p->fts_path);
236 					} else {
237 					(void) printf("%lld\t%s\n",
238 					    howmany(*(du_number_t *)p->fts_pointer, blocksize),
239 					    p->fts_path);
240 					}
241 				}
242 				if (p->fts_pointer) {
243 					free(p->fts_pointer);
244 					p->fts_pointer = NULL;
245 				}
246 				break;
247 			case FTS_DC:			/* Ignore. */
248 				break;
249 			case FTS_DNR:			/* Warn, continue. */
250 			case FTS_ERR:
251 			case FTS_NS:
252 				warnx("%s: %s", p->fts_path, strerror(p->fts_errno));
253 				rval = 1;
254 				break;
255 			default:
256 				if (ignorep(p))
257 					break;
258 
259 				if (p->fts_statp->st_nlink > 1 && linkchk(p))
260 					break;
261 
262 				if (listall || p->fts_level == 0) {
263 					if (hflag) {
264 						(void) prthumanval(howmany(p->fts_statp->st_blocks,
265 							blocksize));
266 						(void) printf("\t%s\n", p->fts_path);
267 					} else {
268 						(void) printf("%lld\t%s\n",
269 							howmany((long long)p->fts_statp->st_blocks, blocksize),
270 							p->fts_path);
271 					}
272 				}
273 				if (p->fts_parent->fts_pointer == NULL) {
274 					p->fts_parent->fts_pointer = malloc(sizeof(du_number_t));
275 					*(du_number_t *)p->fts_parent->fts_pointer = 0;
276 				}
277 				*(du_number_t *)p->fts_parent->fts_pointer += p->fts_statp->st_blocks;
278 		}
279 		if (p->fts_parent->fts_pointer)
280 			savednumber = *(du_number_t *)p->fts_parent->fts_pointer;
281 	}
282 
283 	if (errno)
284 		err(1, "fts_read");
285 
286 	if (cflag) {
287 		if (hflag) {
288 			(void) prthumanval(howmany(savednumber, blocksize));
289 			(void) printf("\ttotal\n");
290 		} else {
291 			(void) printf("%lld\ttotal\n", howmany(savednumber, blocksize));
292 		}
293 	}
294 
295 	ignoreclean();
296 	exit(rval);
297 }
298 
299 static int
300 linkchk(FTSENT *p)
301 {
302 	struct links_entry {
303 		struct links_entry *next;
304 		struct links_entry *previous;
305 		int		links;
306 		dev_t		dev;
307 		ino_t		ino;
308 	};
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, new_size * sizeof(struct links_entry *));
353 			for (i = 0; i < number_buckets; i++) {
354 				while (buckets[i] != NULL) {
355 					/* Remove entry from old bucket. */
356 					le = buckets[i];
357 					buckets[i] = le->next;
358 
359 					/* Add entry to new bucket. */
360 					hash = (le->dev ^ le->ino) % new_size;
361 
362 					if (new_buckets[hash] != NULL)
363 						new_buckets[hash]->previous = le;
364 					le->next = new_buckets[hash];
365 					le->previous = NULL;
366 					new_buckets[hash] = le;
367 				}
368 			}
369 			free(buckets);
370 			buckets = new_buckets;
371 			number_buckets = new_size;
372 		}
373 	}
374 
375 	/* Try to locate this entry in the hash table. */
376 	hash = ( st->st_dev ^ st->st_ino ) % number_buckets;
377 	for (le = buckets[hash]; le != NULL; le = le->next) {
378 		if (le->dev == st->st_dev && le->ino == st->st_ino) {
379 			/*
380 			 * Save memory by releasing an entry when we've seen
381 			 * all of it's links.
382 			 */
383 			if (--le->links <= 0) {
384 				if (le->previous != NULL)
385 					le->previous->next = le->next;
386 				if (le->next != NULL)
387 					le->next->previous = le->previous;
388 				if (buckets[hash] == le)
389 					buckets[hash] = le->next;
390 				number_entries--;
391 				/* Recycle this node through the free list */
392 				if (stop_allocating) {
393 					free(le);
394 				} else {
395 					le->next = free_list;
396 					free_list = le;
397 				}
398 			}
399 			return (1);
400 		}
401 	}
402 
403 	if (stop_allocating)
404 		return (0);
405 
406 	/* Add this entry to the links cache. */
407 	if (free_list != NULL) {
408 		/* Pull a node from the free list if we can. */
409 		le = free_list;
410 		free_list = le->next;
411 	} else
412 		/* Malloc one if we have to. */
413 		le = malloc(sizeof(struct links_entry));
414 	if (le == NULL) {
415 		stop_allocating = 1;
416 		warnx("No more memory for tracking hard links");
417 		return (0);
418 	}
419 	le->dev = st->st_dev;
420 	le->ino = st->st_ino;
421 	le->links = st->st_nlink - 1;
422 	number_entries++;
423 	le->next = buckets[hash];
424 	le->previous = NULL;
425 	if (buckets[hash] != NULL)
426 		buckets[hash]->previous = le;
427 	buckets[hash] = le;
428 	return (0);
429 }
430 
431 void
432 prthumanval(int64_t bytes)
433 {
434 	char buf[sizeof("999M")];
435 
436 	bytes *= DEV_BSIZE;
437 
438 	humanize_number(buf, sizeof(buf), bytes, "", HN_AUTOSCALE,
439 			HN_B | HN_NOSPACE | HN_DECIMAL);
440 
441 	(void) printf("%4s", buf);
442 }
443 
444 static void
445 usage(void)
446 {
447 	(void)fprintf(stderr,
448 		"usage: du [-H | -L | -P] [-a | -s | -d depth] [-c] [-h | -k] [-x] [-I mask] [file ...]\n");
449 	exit(EX_USAGE);
450 }
451 
452 void
453 ignoreadd(const char *mask)
454 {
455 	struct ignentry *ign;
456 
457 	ign = calloc(1, sizeof(*ign));
458 	if (ign == NULL)
459 		errx(1, "cannot allocate memory");
460 	ign->mask = strdup(mask);
461 	if (ign->mask == NULL)
462 		errx(1, "cannot allocate memory");
463 	SLIST_INSERT_HEAD(&ignores, ign, next);
464 }
465 
466 void
467 ignoreclean(void)
468 {
469 	struct ignentry *ign;
470 
471 	while (!SLIST_EMPTY(&ignores)) {
472 		ign = SLIST_FIRST(&ignores);
473 		SLIST_REMOVE_HEAD(&ignores, next);
474 		free(ign->mask);
475 		free(ign);
476 	}
477 }
478 
479 int
480 ignorep(FTSENT *ent)
481 {
482 	struct ignentry *ign;
483 
484 	SLIST_FOREACH(ign, &ignores, next)
485 		if (fnmatch(ign->mask, ent->fts_name, 0) != FNM_NOMATCH)
486 			return 1;
487 	return 0;
488 }
489