2 * Amanda, The Advanced Maryland Automatic Network Disk Archiver
3 * Copyright (c) 1991-1998 University of Maryland at College Park
6 * Permission to use, copy, modify, distribute, and sell this software and its
7 * documentation for any purpose is hereby granted without fee, provided that
8 * the above copyright notice appear in all copies and that both that
9 * copyright notice and this permission notice appear in supporting
10 * documentation, and that the name of U.M. not be used in advertising or
11 * publicity pertaining to distribution of the software without specific,
12 * written prior permission. U.M. makes no representations about the
13 * suitability of this software for any purpose. It is provided "as is"
14 * without express or implied warranty.
16 * U.M. DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL
17 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL U.M.
18 * BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
19 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION
20 * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN
21 * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
23 * Authors: the Amanda Development Team. Its members are listed in a
24 * file named AUTHORS, in the root directory of this distribution.
27 * $Id: calcsize.c,v 1.24.2.3.6.1.2.5 2005/09/20 21:31:52 jrjackson Exp $
29 * traverse directory tree to get backup size estimates
35 #define ROUND(n,x) ((x) + (n) - 1 - (((x) + (n) - 1) % (n)))
38 static unsigned long round_function(n, x)
41 unsigned long remainder = x % n;
48 #define ST_BLOCKS(s) ((((s).st_blocks * 512) <= (s).st_size) ? (s).st_blocks+1 : ((s).st_size / 512 + (((s).st_size % 512) ? 1 : 0)))
50 #define FILETYPES (S_IFREG|S_IFLNK|S_IFDIR)
52 typedef struct name_s {
67 } dumpstats[MAXDUMPS];
69 time_t dumpdate[MAXDUMPS];
70 int dumplevel[MAXDUMPS];
73 void (*add_file_name) P((int, char *));
74 void (*add_file) P((int, struct stat *));
75 long (*final_size) P((int, char *));
78 int main P((int, char **));
79 void traverse_dirs P((char *, char *));
82 void add_file_name_dump P((int, char *));
83 void add_file_dump P((int, struct stat *));
84 long final_size_dump P((int, char *));
86 void add_file_name_gnutar P((int, char *));
87 void add_file_gnutar P((int, struct stat *));
88 long final_size_gnutar P((int, char *));
90 void add_file_name_unknown P((int, char *));
91 void add_file_unknown P((int, struct stat *));
92 long final_size_unknown P((int, char *));
94 sl_t *calc_load_file P((char *filename));
95 int calc_check_exclude P((char *filename));
97 int use_gtar_excl = 0;
98 sl_t *include_sl=NULL, *exclude_sl=NULL;
105 /* standalone test to ckeck wether the calculated file size is ok */
108 unsigned long dump_total=0, gtar_total=0;
114 set_pname("calcsize");
117 fprintf(stderr,"Usage: %s file[s]\n",argv[0]);
120 for(i=1; i<argc; i++) {
121 if(lstat(argv[i], &finfo) == -1) {
122 fprintf(stderr, "%s: %s\n", argv[i], strerror(errno));
125 printf("%s: st_size=%lu", argv[i],(unsigned long)finfo.st_size);
126 printf(": blocks=%lu\n", (unsigned long)ST_BLOCKS(finfo));
127 dump_total += (ST_BLOCKS(finfo) + 1)/2 + 1;
128 gtar_total += ROUND(4,(ST_BLOCKS(finfo) + 1));
130 printf(" gtar dump\n");
131 printf("total %-9lu %-9lu\n",gtar_total,dump_total);
135 char *dirname=NULL, *amname=NULL, *filename=NULL;
136 unsigned long malloc_hist_1, malloc_size_1;
137 unsigned long malloc_hist_2, malloc_size_2;
142 set_pname("calcsize");
144 malloc_size_1 = malloc_inuse(&malloc_hist_1);
147 erroutput_type = (ERR_INTERACTIVE|ERR_SYSLOG);
150 argc--, argv++; /* skip program name */
152 /* need at least program, amname, and directory name */
155 error("Usage: %s [DUMP|GNUTAR] name dir [-X exclude-file] [-I include-file] [level date]*",
160 /* parse backup program name */
162 if(strcmp(*argv, "DUMP") == 0) {
163 #if !defined(DUMP) && !defined(XFSDUMP)
164 error("dump not available on this system");
167 add_file_name = add_file_name_dump;
168 add_file = add_file_dump;
169 final_size = final_size_dump;
172 else if(strcmp(*argv, "GNUTAR") == 0) {
174 error("gnutar not available on this system");
177 add_file_name = add_file_name_gnutar;
178 add_file = add_file_gnutar;
179 final_size = final_size_gnutar;
184 add_file_name = add_file_name_unknown;
185 add_file = add_file_unknown;
186 final_size = final_size_unknown;
190 /* the amanda name can be different from the directory name */
196 error("missing <name>");
198 /* the toplevel directory name to search from */
203 error("missing <dir>");
205 if ((argc > 1) && strcmp(*argv,"-X") == 0) {
208 if (!use_gtar_excl) {
209 error("exclusion specification not supported");
213 filename = stralloc(*argv);
214 if (access(filename, R_OK) != 0) {
215 fprintf(stderr,"Cannot open exclude file %s\n",filename);
218 exclude_sl = calc_load_file(filename);
226 if ((argc > 1) && strcmp(*argv,"-I") == 0) {
229 filename = stralloc(*argv);
230 if (access(filename, R_OK) != 0) {
231 fprintf(stderr,"Cannot open include file %s\n",filename);
234 include_sl = calc_load_file(filename);
241 /* the dump levels to calculate sizes for */
245 if(ndumps < MAXDUMPS) {
246 dumplevel[ndumps] = atoi(argv[0]);
247 dumpdate [ndumps] = (time_t) atol(argv[1]);
249 argc -= 2, argv += 2;
254 error("leftover arg \"%s\", expected <level> and <date>", *argv);
256 if(is_empty_sl(include_sl)) {
257 traverse_dirs(dirname,".");
260 sle_t *an_include = include_sl->first;
261 while(an_include != NULL) {
263 char *adirname = stralloc2(dirname, an_include->name+1);
264 traverse_dirs(adirname);
267 traverse_dirs(dirname, an_include->name);
268 an_include = an_include->next;
271 for(i = 0; i < ndumps; i++) {
275 lseek(1, (off_t)0, SEEK_END);
277 fprintf(stderr, "%s %d SIZE %ld\n",
278 amname, dumplevel[i], final_size(i, dirname));
281 amfunlock(1, "size");
284 malloc_size_2 = malloc_inuse(&malloc_hist_2);
286 if(malloc_size_1 != malloc_size_2) {
287 malloc_list(fileno(stderr), malloc_hist_1, malloc_hist_2);
295 * =========================================================================
298 #ifndef HAVE_BASENAME
304 if ( (cp = strrchr(file,'/')) )
310 void push_name P((char *str));
311 char *pop_name P((void));
313 void traverse_dirs(parent_dir, include)
320 char *dirname, *newname = NULL;
321 char *newbase = NULL;
322 dev_t parent_dev = 0;
326 int has_exclude = !is_empty_sl(exclude_sl) && use_gtar_excl;
328 char *aparent = vstralloc(parent_dir, "/", include, NULL);
330 if(parent_dir && stat(parent_dir, &finfo) != -1)
331 parent_dev = finfo.st_dev;
333 parent_len = strlen(parent_dir);
337 for(dirname = pop_name(); dirname; free(dirname), dirname = pop_name()) {
338 if(has_exclude && calc_check_exclude(dirname+parent_len+1)) {
341 if((d = opendir(dirname)) == NULL) {
347 if(l > 0 && dirname[l - 1] != '/') {
348 newbase = newstralloc2(newbase, dirname, "/");
350 newbase = newstralloc(newbase, dirname);
353 while((f = readdir(d)) != NULL) {
357 if(is_dot_or_dotdot(f->d_name)) {
361 newname = newstralloc2(newname, newbase, f->d_name);
362 if(lstat(newname, &finfo) == -1) {
363 fprintf(stderr, "%s/%s: %s\n",
364 dirname, f->d_name, strerror(errno));
368 if(finfo.st_dev != parent_dev) {
373 is_symlink = ((finfo.st_mode & S_IFMT) == S_IFLNK);
375 is_dir = ((finfo.st_mode & S_IFMT) == S_IFDIR);
376 is_file = ((finfo.st_mode & S_IFMT) == S_IFREG);
378 if (!(is_file || is_dir || is_symlink)) {
383 int is_excluded = -1;
384 for(i = 0; i < ndumps; i++) {
385 add_file_name(i, newname);
386 if(is_file && finfo.st_ctime >= dumpdate[i]) {
389 if(is_excluded == -1)
391 calc_check_exclude(newname+parent_len+1);
392 if(is_excluded == 1) {
401 if(has_exclude && calc_check_exclude(newname+parent_len+1))
411 if(closedir(d) == -1)
425 newp = alloc(sizeof(*newp));
426 newp->str = stralloc(str);
428 newp->next = name_stack;
434 Name *newp = name_stack;
437 if(!newp) return NULL;
439 name_stack = newp->next;
447 * =========================================================================
448 * Backup size calculations for DUMP program
450 * Given the system-dependent nature of dump, it's impossible to pin this
451 * down accurately. Luckily, that's not necessary.
453 * Dump rounds each file up to TP_BSIZE bytes, which is 1k in the BSD dump,
454 * others are unknown. In addition, dump stores three bitmaps at the
455 * beginning of the dump: a used inode map, a dumped dir map, and a dumped
456 * inode map. These are sized by the number of inodes in the filesystem.
458 * We don't take into account the complexities of BSD dump's indirect block
459 * requirements for files with holes, nor the dumping of directories that
460 * are not themselves modified.
462 void add_file_name_dump(level, name)
469 void add_file_dump(level, sp)
473 /* keep the size in kbytes, rounded up, plus a 1k header block */
474 if((sp->st_mode & S_IFMT) == S_IFREG || (sp->st_mode & S_IFMT) == S_IFDIR)
475 dumpstats[level].total_size += (ST_BLOCKS(*sp) + 1)/2 + 1;
478 long final_size_dump(level, topdir)
482 generic_fs_stats_t stats;
486 /* calculate the map sizes */
488 s = stralloc2(topdir, "/.");
489 if(get_fs_stats(s, &stats) == -1) {
490 error("statfs %s: %s", s, strerror(errno));
494 mapsize = (stats.files + 7) / 8; /* in bytes */
495 mapsize = (mapsize + 1023) / 1024; /* in kbytes */
497 /* the dump contains three maps plus the files */
499 return 3*mapsize + dumpstats[level].total_size;
503 * =========================================================================
504 * Backup size calculations for GNUTAR program
506 * Gnutar's basic blocksize is 512 bytes. Each file is rounded up to that
507 * size, plus one header block. Gnutar stores directories' file lists in
508 * incremental dumps - we'll pick up size of the modified dirs here. These
509 * will be larger than a simple filelist of their contents, but that's ok.
511 * As with DUMP, we only need a reasonable estimate, not an exact figure.
513 void add_file_name_gnutar(level, name)
517 /* dumpstats[level].total_size_name += strlen(name) + 64;*/
518 dumpstats[level].total_size += 1;
521 void add_file_gnutar(level, sp)
525 /* the header takes one additional block */
526 dumpstats[level].total_size += ST_BLOCKS(*sp);
529 long final_size_gnutar(level, topdir)
533 /* divide by two to get kbytes, rounded up */
534 /* + 4 blocks for security */
535 return (dumpstats[level].total_size + 5 + (dumpstats[level].total_size_name/512)) / 2;
539 * =========================================================================
540 * Backup size calculations for unknown backup programs.
542 * Here we'll just add up the file sizes and output that.
545 void add_file_name_unknown(level, name)
552 void add_file_unknown(level, sp)
556 /* just add up the block counts */
557 if((sp->st_mode & S_IFMT) == S_IFREG || (sp->st_mode & S_IFMT) == S_IFDIR)
558 dumpstats[level].total_size += ST_BLOCKS(*sp);
561 long final_size_unknown(level, topdir)
565 /* divide by two to get kbytes, rounded up */
566 return (dumpstats[level].total_size + 1) / 2;
570 * =========================================================================
572 sl_t *calc_load_file(filename)
577 sl_t *sl_list = new_sl();
579 FILE *file = fopen(filename, "r");
581 while(fgets(pattern, 1025, file)) {
582 if(strlen(pattern)>0 && pattern[strlen(pattern)-1] == '\n')
583 pattern[strlen(pattern)-1] = '\0';
584 sl_list = append_sl(sl_list, pattern);
591 int calc_check_exclude(filename)
595 if(is_empty_sl(exclude_sl)) return 0;
597 an_exclude=exclude_sl->first;
598 while(an_exclude != NULL) {
599 if(match_tar(an_exclude->name, filename)) {
602 an_exclude=an_exclude->next;