Home | History | Annotate | Download | only in toolbox
      1 /*	$NetBSD: du.c,v 1.33 2008/07/30 22:03:40 dsl Exp $	*/
      2 
      3 /*
      4  * Copyright (c) 1989, 1993, 1994
      5  *	The Regents of the University of California.  All rights reserved.
      6  *
      7  * This code is derived from software contributed to Berkeley by
      8  * Chris Newcomb.
      9  *
     10  * Redistribution and use in source and binary forms, with or without
     11  * modification, are permitted provided that the following conditions
     12  * are met:
     13  * 1. Redistributions of source code must retain the above copyright
     14  *    notice, this list of conditions and the following disclaimer.
     15  * 2. Redistributions in binary form must reproduce the above copyright
     16  *    notice, this list of conditions and the following disclaimer in the
     17  *    documentation and/or other materials provided with the distribution.
     18  * 3. Neither the name of the University nor the names of its contributors
     19  *    may be used to endorse or promote products derived from this software
     20  *    without specific prior written permission.
     21  *
     22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     32  * SUCH DAMAGE.
     33  */
     34 
     35 #include <sys/cdefs.h>
     36 #ifndef lint
     37 __COPYRIGHT("@(#) Copyright (c) 1989, 1993, 1994\
     38  The Regents of the University of California.  All rights reserved.");
     39 #endif /* not lint */
     40 
     41 #ifndef lint
     42 #if 0
     43 static char sccsid[] = "@(#)du.c	8.5 (Berkeley) 5/4/95";
     44 #else
     45 __RCSID("$NetBSD: du.c,v 1.33 2008/07/30 22:03:40 dsl Exp $");
     46 #endif
     47 #endif /* not lint */
     48 
     49 #include <sys/types.h>
     50 #include <sys/stat.h>
     51 
     52 #include <dirent.h>
     53 #include <err.h>
     54 #include <errno.h>
     55 #include <fts.h>
     56 #include <util.h>
     57 #include <stdio.h>
     58 #include <stdlib.h>
     59 #include <string.h>
     60 #include <unistd.h>
     61 #include <limits.h>
     62 
     63 int	linkchk(dev_t, ino_t);
     64 void	prstat(const char *, int64_t);
     65 static void	usage(void);
     66 
     67 long blocksize;
     68 
     69 #define howmany(x, y)   (((x)+((y)-1))/(y))
     70 
     71 int
     72 du_main(int argc, char *argv[])
     73 {
     74 	FTS *fts;
     75 	FTSENT *p;
     76 	int64_t totalblocks;
     77 	int ftsoptions, listfiles;
     78 	int depth;
     79 	int Hflag, Lflag, aflag, ch, cflag, dflag, gkmflag, nflag, rval, sflag;
     80 	const char *noargv[2];
     81 
     82 	Hflag = Lflag = aflag = cflag = dflag = gkmflag = sflag = 0;
     83 	totalblocks = 0;
     84 	ftsoptions = FTS_PHYSICAL;
     85 	depth = INT_MAX;
     86 	while ((ch = getopt(argc, argv, "HLPacd:ghkmnrsx")) != -1)
     87 		switch (ch) {
     88 		case 'H':
     89 			Hflag = 1;
     90 			Lflag = 0;
     91 			break;
     92 		case 'L':
     93 			Lflag = 1;
     94 			Hflag = 0;
     95 			break;
     96 		case 'P':
     97 			Hflag = Lflag = 0;
     98 			break;
     99 		case 'a':
    100 			aflag = 1;
    101 			break;
    102 		case 'c':
    103 			cflag = 1;
    104 			break;
    105 		case 'd':
    106 			dflag = 1;
    107 			depth = atoi(optarg);
    108 			if (depth < 0 || depth > SHRT_MAX) {
    109 				warnx("invalid argument to option d: %s",
    110 					optarg);
    111 				usage();
    112 			}
    113 			break;
    114 		case 'g':
    115 			blocksize = 1024 * 1024 * 1024;
    116 			gkmflag = 1;
    117 			break;
    118 		case 'k':
    119 			blocksize = 1024;
    120 			gkmflag = 1;
    121 			break;
    122 		case 'm':
    123 			blocksize = 1024 * 1024;
    124 			gkmflag = 1;
    125 			break;
    126 		case 'r':
    127 			break;
    128 		case 's':
    129 			sflag = 1;
    130 			break;
    131 		case 'x':
    132 			ftsoptions |= FTS_XDEV;
    133 			break;
    134 		case '?':
    135 		default:
    136 			usage();
    137 		}
    138 	argc -= optind;
    139 	argv += optind;
    140 
    141 	/*
    142 	 * XXX
    143 	 * Because of the way that fts(3) works, logical walks will not count
    144 	 * the blocks actually used by symbolic links.  We rationalize this by
    145 	 * noting that users computing logical sizes are likely to do logical
    146 	 * copies, so not counting the links is correct.  The real reason is
    147 	 * that we'd have to re-implement the kernel's symbolic link traversing
    148 	 * algorithm to get this right.  If, for example, you have relative
    149 	 * symbolic links referencing other relative symbolic links, it gets
    150 	 * very nasty, very fast.  The bottom line is that it's documented in
    151 	 * the man page, so it's a feature.
    152 	 */
    153 	if (Hflag)
    154 		ftsoptions |= FTS_COMFOLLOW;
    155 	if (Lflag) {
    156 		ftsoptions &= ~FTS_PHYSICAL;
    157 		ftsoptions |= FTS_LOGICAL;
    158 	}
    159 
    160 	listfiles = 0;
    161 	if (aflag) {
    162 		if (sflag || dflag)
    163 			usage();
    164 		listfiles = 1;
    165 	} else if (sflag) {
    166 		if (dflag)
    167 			usage();
    168 		depth = 0;
    169 	}
    170 
    171 	if (!*argv) {
    172 		noargv[0] = ".";
    173 		noargv[1] = NULL;
    174 		argv = __UNCONST(noargv);
    175 	}
    176 
    177 	if (!gkmflag)
    178 		blocksize = 512;
    179 	blocksize /= 512;
    180 
    181 	if ((fts = fts_open(argv, ftsoptions, NULL)) == NULL)
    182 		err(1, "fts_open `%s'", *argv);
    183 
    184 	for (rval = 0; (p = fts_read(fts)) != NULL;) {
    185 		switch (p->fts_info) {
    186 		case FTS_D:			/* Ignore. */
    187 			break;
    188 		case FTS_DP:
    189 			p->fts_parent->fts_number +=
    190 			    p->fts_number += p->fts_statp->st_blocks;
    191 			if (cflag)
    192 				totalblocks += p->fts_statp->st_blocks;
    193 			/*
    194 			 * If listing each directory, or not listing files
    195 			 * or directories and this is post-order of the
    196 			 * root of a traversal, display the total.
    197 			 */
    198 			if (p->fts_level <= depth
    199 			    || (!listfiles && !p->fts_level))
    200 				prstat(p->fts_path, p->fts_number);
    201 			break;
    202 		case FTS_DC:			/* Ignore. */
    203 			break;
    204 		case FTS_DNR:			/* Warn, continue. */
    205 		case FTS_ERR:
    206 		case FTS_NS:
    207 			warnx("%s: %s", p->fts_path, strerror(p->fts_errno));
    208 			rval = 1;
    209 			break;
    210 		default:
    211 			if (p->fts_statp->st_nlink > 1 &&
    212 			    linkchk(p->fts_statp->st_dev, p->fts_statp->st_ino))
    213 				break;
    214 			/*
    215 			 * If listing each file, or a non-directory file was
    216 			 * the root of a traversal, display the total.
    217 			 */
    218 			if (listfiles || !p->fts_level)
    219 				prstat(p->fts_path, p->fts_statp->st_blocks);
    220 			p->fts_parent->fts_number += p->fts_statp->st_blocks;
    221 			if (cflag)
    222 				totalblocks += p->fts_statp->st_blocks;
    223 		}
    224 	}
    225 	if (errno)
    226 		err(1, "fts_read");
    227 	if (cflag)
    228 		prstat("total", totalblocks);
    229 	exit(rval);
    230 }
    231 
    232 void
    233 prstat(const char *fname, int64_t blocks)
    234 {
    235 	(void)printf("%lld\t%s\n",
    236 	    (long long)howmany(blocks, (int64_t)blocksize),
    237 	    fname);
    238 }
    239 
    240 int
    241 linkchk(dev_t dev, ino_t ino)
    242 {
    243 	static struct entry {
    244 		dev_t	dev;
    245 		ino_t	ino;
    246 	} *htable;
    247 	static int htshift;  /* log(allocated size) */
    248 	static int htmask;   /* allocated size - 1 */
    249 	static int htused;   /* 2*number of insertions */
    250 	static int sawzero;  /* Whether zero is in table or not */
    251 	int h, h2;
    252 	uint64_t tmp;
    253 	/* this constant is (1<<64)/((1+sqrt(5))/2)
    254 	 * aka (word size)/(golden ratio)
    255 	 */
    256 	const uint64_t HTCONST = 11400714819323198485ULL;
    257 	const int HTBITS = CHAR_BIT * sizeof(tmp);
    258 
    259 	/* Never store zero in hashtable */
    260 	if (dev == 0 && ino == 0) {
    261 		h = sawzero;
    262 		sawzero = 1;
    263 		return h;
    264 	}
    265 
    266 	/* Extend hash table if necessary, keep load under 0.5 */
    267 	if (htused<<1 >= htmask) {
    268 		struct entry *ohtable;
    269 
    270 		if (!htable)
    271 			htshift = 10;   /* starting hashtable size */
    272 		else
    273 			htshift++;   /* exponential hashtable growth */
    274 
    275 		htmask  = (1 << htshift) - 1;
    276 		htused = 0;
    277 
    278 		ohtable = htable;
    279 		htable = calloc(htmask+1, sizeof(*htable));
    280 		if (!htable)
    281 			err(1, "calloc");
    282 
    283 		/* populate newly allocated hashtable */
    284 		if (ohtable) {
    285 			int i;
    286 			for (i = 0; i <= htmask>>1; i++)
    287 				if (ohtable[i].ino || ohtable[i].dev)
    288 					linkchk(ohtable[i].dev, ohtable[i].ino);
    289 			free(ohtable);
    290 		}
    291 	}
    292 
    293 	/* multiplicative hashing */
    294 	tmp = dev;
    295 	tmp <<= HTBITS>>1;
    296 	tmp |=  ino;
    297 	tmp *= HTCONST;
    298 	h  = tmp >> (HTBITS - htshift);
    299 	h2 = 1 | ( tmp >> (HTBITS - (htshift<<1) - 1)); /* must be odd */
    300 
    301 	/* open address hashtable search with double hash probing */
    302 	while (htable[h].ino || htable[h].dev) {
    303 		if ((htable[h].ino == ino) && (htable[h].dev == dev))
    304 			return 1;
    305 		h = (h + h2) & htmask;
    306 	}
    307 
    308 	/* Insert the current entry into hashtable */
    309 	htable[h].dev = dev;
    310 	htable[h].ino = ino;
    311 	htused++;
    312 	return 0;
    313 }
    314 
    315 static void
    316 usage(void)
    317 {
    318 
    319 	(void)fprintf(stderr,
    320 		"usage: du [-H | -L | -P] [-a | -d depth | -s] [-cgkmrx] [file ...]\n");
    321 	exit(1);
    322 }
    323