]> git.tuebingen.mpg.de Git - adu.git/blob - adu.c
8068d205f2274dded3074158a105243bb4de1b11
[adu.git] / adu.c
1 #include "adu.h"
2 #include <dirent.h> /* readdir() */
3
4 #include "gcc-compat.h"
5 #include "cmdline.h"
6 #include "osl.h"
7 #include "fd.h"
8 #include "hash.h"
9 #include "string.h"
10 #include "error.h"
11
12 DEFINE_ERRLIST;
13
14 #define DATABASE_DIR "/tmp/adu"
15 #define UID_LIST DATABASE_DIR "/" "uid_list"
16
17 /** Command line and config file options. */
18 static struct gengetopt_args_info conf;
19
20 struct user_info {
21         uint32_t uid;
22         struct osl_table *table;
23         uint64_t files;
24         uint64_t bytes;
25         uint64_t dirs;
26         struct osl_table_description *desc;
27 };
28
29 static struct user_info *uid_hash_table;
30
31 /** evaluates to 1 if x < y, to -1 if x > y and to 0 if x == y */
32 #define NUM_COMPARE(x, y) ((int)((x) < (y)) - (int)((x) > (y)))
33
34
35 /**
36  * The log function.
37  *
38  * \param ll Loglevel.
39  * \param fml Usual format string.
40  *
41  * All XXX_LOG() macros use this function.
42  */
43 __printf_2_3 void __log(int ll, const char* fmt,...)
44 {
45         va_list argp;
46         FILE *outfd;
47         struct tm *tm;
48         time_t t1;
49         char str[255] = "";
50
51         if (ll < conf.loglevel_arg)
52                 return;
53         outfd = stderr;
54         time(&t1);
55         tm = localtime(&t1);
56         strftime(str, sizeof(str), "%b %d %H:%M:%S", tm);
57         fprintf(outfd, "%s ", str);
58         va_start(argp, fmt);
59         vfprintf(outfd, fmt, argp);
60         va_end(argp);
61 }
62
63 /**
64  * Compare the size of two directories
65  *
66  * \param obj1 Pointer to the first object.
67  * \param obj2 Pointer to the second object.
68  *
69  * This function first compares the size values as usual integers. If they compare as
70  * equal, the address of \a obj1 and \a obj2 are compared. So this compare function
71  * returns zero if and only if \a obj1 and \a obj2 point to the same memory area.
72  */
73 static int size_compare(const struct osl_object *obj1, const struct osl_object *obj2)
74 {
75         uint64_t d1 = *(uint64_t *)obj1->data;
76         uint64_t d2 = *(uint64_t *)obj2->data;
77         int ret = NUM_COMPARE(d2, d1);
78
79         if (ret)
80                 return ret;
81         //INFO_LOG("addresses: %p, %p\n", obj1->data, obj2->data);
82         return NUM_COMPARE(obj2->data, obj1->data);
83 }
84
85 /**
86  * Compare two osl objects of string type.
87  *
88  * \param obj1 Pointer to the first object.
89  * \param obj2 Pointer to the second object.
90  *
91  * In any case, only \p MIN(obj1->size, obj2->size) characters of each string
92  * are taken into account.
93  *
94  * \return It returns an integer less than, equal to, or greater than zero if
95  * \a obj1 is found, respectively, to be less than, to match, or be greater
96  * than obj2.
97  *
98  * \sa strcmp(3), strncmp(3), osl_compare_func.
99  */
100 static int string_compare(const struct osl_object *obj1,
101                 const struct osl_object *obj2)
102 {
103         const char *str1 = (const char *)obj1->data;
104         const char *str2 = (const char *)obj2->data;
105         return strncmp(str1, str2, MIN(obj1->size, obj2->size));
106 }
107
108 /**
109  * Compare two osl objects pointing to unsigned integers of 64 bit size.
110  *
111  * \param obj1 Pointer to the first integer.
112  * \param obj2 Pointer to the second integer.
113  *
114  * \return The values required for an osl compare function.
115  *
116  * \sa osl_compare_func, osl_hash_compare().
117  */
118 static int uint64_compare(const struct osl_object *obj1,
119                 const struct osl_object *obj2)
120 {
121         uint64_t d1 = read_u64((const char *)obj1->data);
122         uint64_t d2 = read_u64((const char *)obj2->data);
123
124         if (d1 < d2)
125                 return 1;
126         if (d1 > d2)
127                 return -1;
128         return 0;
129 }
130
131 /** The columns of the directory table. */
132 enum dir_table_columns {
133         /** The name of the directory. */
134         DT_NAME,
135         /** The dir count number. */
136         DT_NUM,
137         /** The number of bytes of all regular files. */
138         DT_BYTES,
139         /** The number of all regular files. */
140         DT_FILES,
141         /** Number of columns in this table. */
142         NUM_DT_COLUMNS
143 };
144
145 static struct osl_column_description dir_table_cols[] = {
146         [DT_NAME] = {
147                 .storage_type = OSL_MAPPED_STORAGE,
148                 .storage_flags = OSL_RBTREE | OSL_UNIQUE,
149                 .name = "dir",
150                 .compare_function = string_compare,
151         },
152         [DT_NUM] = {
153                 .storage_type = OSL_MAPPED_STORAGE,
154                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
155                 .name = "num",
156                 .compare_function = uint64_compare,
157                 .data_size = sizeof(uint64_t)
158         },
159         [DT_BYTES] = {
160                 .storage_type = OSL_MAPPED_STORAGE,
161                 .storage_flags =  OSL_RBTREE | OSL_FIXED_SIZE,
162                 .compare_function = size_compare,
163                 .name = "num_bytes",
164                 .data_size = sizeof(uint64_t)
165         },
166         [DT_FILES] = {
167                 .storage_type = OSL_MAPPED_STORAGE,
168                 .storage_flags =  OSL_RBTREE | OSL_FIXED_SIZE,
169                 .compare_function = size_compare,
170                 .name = "num_files",
171                 .data_size = sizeof(uint64_t)
172         }
173 };
174
175 static struct osl_table_description dir_table_desc = {
176         .name = "dir_table",
177         .num_columns = NUM_DT_COLUMNS,
178         .flags = 0,
179         .column_descriptions = dir_table_cols,
180         .dir = DATABASE_DIR
181 };
182
183 /** The columns of the id table. */
184 enum user_table_columns {
185         /** The numer of the directory. */
186         UT_DIR_NUM,
187         /** The number of bytes of all regular files in this dir owned by this id. */
188         UT_BYTES,
189         /** The number of files in this dir owned by this id. */
190         UT_FILES,
191         /** Number of columns in this table. */
192         NUM_UT_COLUMNS
193 };
194
195 static struct osl_column_description user_table_cols[] = {
196         [UT_DIR_NUM] = {
197                 .storage_type = OSL_MAPPED_STORAGE,
198                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
199                 .name = "dir_num",
200                 .compare_function = uint64_compare,
201                 .data_size = sizeof(uint64_t)
202         },
203         [UT_BYTES] = {
204                 .storage_type = OSL_MAPPED_STORAGE,
205                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
206                 .compare_function = size_compare,
207                 .name = "num_bytes",
208                 .data_size = sizeof(uint64_t)
209         },
210         [UT_FILES] = {
211                 .storage_type = OSL_MAPPED_STORAGE,
212                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
213                 .compare_function = size_compare,
214                 .name = "num_files",
215                 .data_size = sizeof(uint64_t)
216         },
217 };
218
219 static struct osl_table *dir_table;
220
221 static int add_directory(char *dirname, uint64_t dir_num, uint64_t *dir_size,
222                 uint64_t *dir_files)
223 {
224         struct osl_object dir_objects[NUM_DT_COLUMNS];
225
226         INFO_LOG("adding #%llu: %s\n", (long long unsigned)dir_num, dirname);
227         dir_objects[DT_NAME].data = dirname;
228         dir_objects[DT_NAME].size = strlen(dirname) + 1;
229         dir_objects[DT_NUM].data = &dir_num;
230         dir_objects[DT_NUM].size = sizeof(dir_num);
231         dir_objects[DT_BYTES].data = dir_size;
232         dir_objects[DT_BYTES].size = sizeof(*dir_size);
233         dir_objects[DT_FILES].data = dir_files;
234         dir_objects[DT_FILES].size = sizeof(*dir_files);
235
236         return osl_add_row(dir_table, dir_objects);
237 }
238
239 static uint32_t num_uids;
240
241 static int open_user_table(struct user_info *ui, int create)
242 {
243         int ret;
244
245         ui->desc = para_malloc(sizeof(*ui->desc));
246         ui->desc->num_columns = NUM_UT_COLUMNS;
247         ui->desc->flags = 0;
248         ui->desc->column_descriptions = user_table_cols;
249         ui->desc->dir = para_strdup(DATABASE_DIR);
250         ui->desc->name = make_message("%u", (unsigned)ui->uid);
251         num_uids++;
252         INFO_LOG(".............................uid #%u: %u\n",
253                 (unsigned)num_uids, (unsigned)ui->uid);
254         if (create) {
255                 ret = osl_create_table(ui->desc);
256                 if (ret < 0)
257                         goto err;
258         }
259         ret = osl_open_table(ui->desc, &ui->table);
260         if (ret < 0)
261                 goto err;
262         return 1;
263 err:
264         free((char *)ui->desc->name);
265         free((char *)ui->desc->dir);
266         free(ui->desc);
267         ui->desc->name = NULL;
268         ui->desc->dir = NULL;
269         ui->desc = NULL;
270         ui->table = NULL;
271         return ret;
272 }
273
274 #define uid_hash_bits 8
275 static uint32_t uid_hash_table_size = 1 << uid_hash_bits;
276 #define PRIME1 0x811c9dc5
277 #define PRIME2 0x01000193
278
279 static void create_hash_table(void)
280 {
281         uid_hash_table = para_calloc(uid_hash_table_size
282                 * sizeof(struct user_info));
283 }
284
285 static void free_hash_table(void)
286 {
287         free(uid_hash_table);
288         uid_hash_table = NULL;
289 }
290
291 static int create_tables(void)
292 {
293         int ret;
294
295         ret = osl_create_table(&dir_table_desc);
296         if (ret < 0)
297                 return ret;
298         create_hash_table();
299         return 1;
300 }
301
302 /*
303  * We use a hash table of size s=2^uid_hash_bits to map the uids into the
304  * interval [0..s]. Hash collisions are treated by open addressing, i.e.
305  * unused slots in the table are used to store different uids that hash to the
306  * same slot.
307  *
308  * If a hash collision occurs, different slots are successively probed in order
309  * to find an unused slot for the new uid. Probing is implemented via a second
310  * hash function that maps the uid to h=(uid * PRIME2) | 1, which is always an
311  * odd number.
312  *
313  * An odd number is sufficient to make sure each entry of the hash table gets
314  * probed for probe_num between 0 and s-1 because s is a power of two, hence
315  * the second hash value never hash a common divisor with the hash table size.
316  * IOW: h is invertible in the ring [0..s].
317  */
318 static uint32_t double_hash(uint32_t uid, uint32_t probe_num)
319 {
320         return (uid * PRIME1 + ((uid * PRIME2) | 1) * probe_num)
321                 % uid_hash_table_size;
322 }
323
324 #define FOR_EACH_USER(ui) for (ui = uid_hash_table; ui && ui < uid_hash_table \
325                 + uid_hash_table_size; ui++)
326
327 enum search_uid_flags {
328         OPEN_USER_TABLE = 1,
329         CREATE_USER_TABLE = 2,
330 };
331
332 static int search_uid(uint32_t uid, enum search_uid_flags flags,
333                 struct user_info **ui_ptr)
334 {
335         uint32_t p;
336
337         for (p = 0; p < uid_hash_table_size; p++) {
338                 struct user_info *ui = uid_hash_table + double_hash(uid, p);
339
340                 if (!ui->table) {
341                         int ret;
342
343                         if (!flags)
344                                 return -E_BAD_UID;
345                         ui->uid = uid;
346                         ret = open_user_table(ui, flags & CREATE_USER_TABLE);
347                         if (ret < 0)
348                                 return ret;
349                         if (ui_ptr)
350                                 *ui_ptr = ui;
351                         return 1;
352                 }
353                 if (ui->uid != uid)
354                         continue;
355                 if (ui_ptr)
356                         *ui_ptr = ui;
357                 return 0;
358         }
359         return flags? -E_HASH_TABLE_OVERFLOW : -E_BAD_UID;
360 }
361
362 static int update_user_row(struct osl_table *t, uint64_t dir_num,
363                 uint64_t *add)
364 {
365         struct osl_row *row;
366         struct osl_object obj = {.data = &dir_num, .size = sizeof(dir_num)};
367
368         int ret = osl_get_row(t, UT_DIR_NUM, &obj, &row);
369
370         if (ret < 0 && ret != -E_RB_KEY_NOT_FOUND)
371                 return ret;
372         if (ret < 0) { /* this is the first file we add */
373                 struct osl_object objects[NUM_UT_COLUMNS];
374                 uint64_t num_files = 1;
375
376                 objects[UT_DIR_NUM].data = &dir_num;
377                 objects[UT_DIR_NUM].size = sizeof(dir_num);
378                 objects[UT_BYTES].data = add;
379                 objects[UT_BYTES].size = sizeof(*add);
380                 objects[UT_FILES].data = &num_files;
381                 objects[UT_FILES].size = sizeof(num_files);
382                 INFO_LOG("######################### ret: %d\n", ret);
383                 ret = osl_add_row(t, objects);
384                 INFO_LOG("######################### ret: %d\n", ret);
385                 return ret;
386         } else { /* add size and increment file count */
387                 uint64_t num;
388                 struct osl_object obj1, obj2 = {.data = &num, .size = sizeof(num)};
389
390                 ret = osl_get_object(t, row, UT_BYTES, &obj1);
391                 if (ret < 0)
392                         return ret;
393                 num = *(uint64_t *)obj1.data + *add;
394                 ret = osl_update_object(t, row, UT_BYTES, &obj2);
395                 if (ret < 0)
396                         return ret;
397                 ret = osl_get_object(t, row, UT_FILES, &obj1);
398                 if (ret < 0)
399                         return ret;
400                 num = *(uint64_t *)obj1.data + 1;
401                 return osl_update_object(t, row, UT_FILES, &obj2);
402         }
403 }
404
405 static uint64_t num_dirs;
406 static uint64_t num_files;
407 static uint64_t num_bytes;
408
409 int scan_dir(char *dirname)
410 {
411         DIR *dir;
412         struct dirent *entry;
413         int ret, cwd_fd, ret2;
414         uint64_t dir_size = 0, dir_files = 0;
415         uint64_t this_dir_num = num_dirs++;
416
417         DEBUG_LOG("----------------- %llu: %s\n", (long long unsigned)num_dirs, dirname);
418         ret = para_opendir(dirname, &dir, &cwd_fd);
419         if (ret < 0) {
420                 if (ret != -ERRNO_TO_ERROR(EACCES))
421                         return ret;
422                 WARNING_LOG("permission denied for %s\n", dirname);
423                 return 1;
424         }
425         while ((entry = readdir(dir))) {
426                 mode_t m;
427                 char *tmp;
428                 struct stat s;
429                 uint32_t uid;
430                 uint64_t size;
431                 struct user_info *ui;
432
433                 if (!strcmp(entry->d_name, "."))
434                         continue;
435                 if (!strcmp(entry->d_name, ".."))
436                         continue;
437                 if (lstat(entry->d_name, &s) == -1) {
438                         WARNING_LOG("lstat error for %s/%s\n", dirname,
439                                 entry->d_name);
440                         continue;
441                 }
442                 m = s.st_mode;
443                 if (!S_ISREG(m) && !S_ISDIR(m))
444                         continue;
445                 if (S_ISDIR(m)) {
446                         tmp = make_message("%s/%s", dirname, entry->d_name);
447                         ret = scan_dir(tmp);
448                         free(tmp);
449                         if (ret < 0)
450                                 goto out;
451                         continue;
452                 }
453                 /* regular file */
454                 size = s.st_size;
455                 dir_size += size;
456                 num_bytes += size;
457                 dir_files++;
458                 num_files++;
459                 uid = s.st_uid;
460                 ret = search_uid(uid, CREATE_USER_TABLE | OPEN_USER_TABLE, &ui);
461                 if (ret < 0)
462                         goto out;
463                 ui->bytes += size;
464                 ui->files++;
465                 ret = update_user_row(ui->table, this_dir_num, &size);
466                 if (ret < 0)
467                         goto out;
468         }
469         ret = add_directory(dirname, this_dir_num, &dir_size, &dir_files);
470 out:
471         closedir(dir);
472         ret2 = para_fchdir(cwd_fd);
473         if (ret2 < 0 && ret >= 0)
474                 ret = ret2;
475         close(cwd_fd);
476         return ret;
477 }
478
479 static int get_dir_name(struct osl_row *row, char **name)
480 {
481         struct osl_object obj;
482         int ret = osl_get_object(dir_table, row, DT_NAME, &obj);
483
484         if (ret < 0)
485                 return ret;
486         *name = obj.data;
487         return 1;
488 }
489
490 enum global_stats_flags {
491         GSF_PRINT_DIRNAME = 1,
492         GSF_PRINT_BYTES = 2,
493         GSF_PRINT_FILES = 4,
494         GSF_COMPUTE_SUMMARY = 8,
495 };
496
497 struct global_stats_info {
498         uint32_t count;
499         enum global_stats_flags flags;
500 };
501
502 static int global_stats_loop_function(struct osl_row *row, void *data)
503 {
504         struct global_stats_info *gsi = data;
505         struct osl_object obj;
506         char *dirname;
507         int ret, summary = gsi->flags & GSF_COMPUTE_SUMMARY;
508
509         if (!gsi->count && !summary)
510                 return -E_LOOP_COMPLETE;
511         if (gsi->count && (gsi->flags & GSF_PRINT_DIRNAME)) {
512                 ret = get_dir_name(row, &dirname);
513                 if (ret < 0)
514                         return ret;
515                 printf("%s%s", dirname,
516                         (gsi->flags & (GSF_PRINT_FILES | GSF_PRINT_BYTES))?
517                                 "\t" : "\n"
518                 );
519         }
520         if (summary || (gsi->count && (gsi->flags & GSF_PRINT_FILES))) {
521                 uint64_t files;
522                 ret = osl_get_object(dir_table, row, DT_FILES, &obj);
523                 if (ret < 0)
524                         return ret;
525                 files = *(uint64_t *)obj.data;
526                 if (gsi->count && (gsi->flags & GSF_PRINT_FILES))
527                         printf("%llu%s", (long long unsigned)files,
528                                 (gsi->flags & GSF_PRINT_BYTES)? "\t" : "\n");
529                 if (summary)
530                         num_files += files;
531         }
532         if (summary || (gsi->count && (gsi->flags & GSF_PRINT_BYTES))) {
533                 uint64_t bytes;
534                 ret = osl_get_object(dir_table, row, DT_BYTES, &obj);
535                 if (ret < 0)
536                         return ret;
537                 bytes = *(uint64_t *)obj.data;
538                 if (gsi->count && (gsi->flags & GSF_PRINT_BYTES))
539                         printf("%llu\n", (long long unsigned)bytes);
540                 if (summary) {
541                         num_bytes += bytes;
542                         num_dirs++;
543                 }
544         }
545         if (gsi->count)
546                 gsi->count--;
547         return 1;
548 }
549
550 static void print_id_stats(void)
551 {
552         struct user_info *ui;
553
554         printf("--------------------- user summary (uid/dirs/files/bytes):\n");
555         FOR_EACH_USER(ui) {
556                 if (!ui->table)
557                         continue;
558                 printf("%u\t%llu\t%llu\t%llu\n", (unsigned)ui->uid,
559                         (long long unsigned)ui->dirs,
560                         (long long unsigned)ui->files,
561                         (long long unsigned)ui->bytes);
562         }
563 }
564
565 enum user_stats_flags {
566         USF_PRINT_DIRNAME = 1,
567         USF_PRINT_BYTES = 2,
568         USF_PRINT_FILES = 4,
569         USF_COMPUTE_SUMMARY = 8,
570 };
571
572 struct user_stats_info {
573         uint32_t count;
574         enum user_stats_flags flags;
575         struct user_info *ui;
576 };
577
578 static int user_stats_loop_function(struct osl_row *row, void *data)
579 {
580         struct user_stats_info *usi = data;
581         struct osl_row *dir_row;
582         struct osl_object obj;
583         int ret, summary = usi->flags & GSF_COMPUTE_SUMMARY;
584
585         if (!usi->count && !summary)
586                 return -E_LOOP_COMPLETE;
587         if (usi->count && (usi->flags & USF_PRINT_DIRNAME)) {
588                 char *dirname;
589                 ret = osl_get_object(usi->ui->table, row, UT_DIR_NUM, &obj);
590                 if (ret < 0)
591                         return ret;
592                 ret = osl_get_row(dir_table, DT_NUM, &obj, &dir_row);
593                 if (ret < 0)
594                         return ret;
595                 ret = osl_get_object(dir_table, dir_row, DT_NAME, &obj);
596                 if (ret < 0)
597                         return ret;
598                 dirname = obj.data;
599                 printf("%s%s",
600                         dirname,
601                         (usi->flags & (USF_PRINT_FILES | USF_PRINT_BYTES))?
602                                 "\t" : "\n"
603                 );
604         }
605         if (summary || (usi->count && (usi->flags & USF_PRINT_FILES))) {
606                 uint64_t files;
607                 ret = osl_get_object(usi->ui->table, row, UT_FILES, &obj);
608                 if (ret < 0)
609                         return ret;
610                 files = *(uint64_t *)obj.data;
611                 if (usi->count && (usi->flags & USF_PRINT_FILES))
612                         printf("%llu%s",
613                                 (long long unsigned)files,
614                                 (usi->flags & USF_PRINT_BYTES)? "\t" : "\n"
615                         );
616                 if (summary)
617                         usi->ui->files += files;
618         }
619         if (summary || (usi->count && (usi->flags & USF_PRINT_BYTES))) {
620                 uint64_t bytes;
621                 ret = osl_get_object(usi->ui->table, row, UT_BYTES, &obj);
622                 if (ret < 0)
623                         return ret;
624                 bytes = *(uint64_t *)obj.data;
625                 if (usi->count && (usi->flags & USF_PRINT_BYTES))
626                         printf("%llu\n", (long long unsigned)bytes);
627                 if (summary) {
628                         usi->ui->bytes += bytes;
629                         usi->ui->dirs++;
630                 }
631
632         }
633         if (usi->count)
634                 usi->count--;
635         return 1;
636 }
637
638 static void print_user_stats(void)
639 {
640         struct user_info *ui;
641
642         FOR_EACH_USER(ui) {
643                 struct user_stats_info usi = {
644                         .count = 10,
645                         .ui = ui
646                 };
647                 if (!ui->table)
648                         continue;
649                 usi.flags = USF_PRINT_DIRNAME | USF_PRINT_BYTES | USF_COMPUTE_SUMMARY;
650                 printf("************************************************ uid %u\n",
651                         (unsigned) ui->uid);
652                 if (!ui->table)
653                         continue;
654                 printf("----------------- Largest dirs -------------------\n");
655                 osl_rbtree_loop_reverse(ui->table, UT_BYTES, &usi,
656                         user_stats_loop_function);
657                 printf("---------- dirs containing most files ------------\n");
658                 usi.count = 10;
659                 usi.flags = USF_PRINT_DIRNAME | USF_PRINT_FILES;
660                 osl_rbtree_loop_reverse(ui->table, UT_FILES, &usi,
661                         user_stats_loop_function);
662         }
663 }
664
665 static int print_statistics(void)
666 {
667         int ret;
668         struct global_stats_info gsi = {
669                 .count = 10,
670                 .flags = GSF_PRINT_DIRNAME | GSF_PRINT_BYTES | GSF_COMPUTE_SUMMARY
671         };
672
673         printf("----------------- Largest dirs -------------------\n");
674         ret = osl_rbtree_loop_reverse(dir_table, DT_BYTES, &gsi,
675                 global_stats_loop_function);
676         if (ret < 0 && ret != -E_LOOP_COMPLETE)
677                 return ret;
678         gsi.count = 10;
679
680         gsi.flags = GSF_PRINT_DIRNAME | GSF_PRINT_FILES;
681         printf("---------- dirs containing most files ------------\n");
682         ret = osl_rbtree_loop_reverse(dir_table, DT_FILES, &gsi,
683                 global_stats_loop_function);
684         if (ret < 0 && ret != -E_LOOP_COMPLETE)
685                 return ret;
686
687         printf("------------------ Global summary (dirs/files/bytes)\n"
688                 "%llu\t%llu\t%llu\n",
689                 (long long unsigned)num_dirs, (long long unsigned)num_files,
690                 (long long unsigned)num_bytes);
691         print_user_stats();
692         print_id_stats();
693         return 1;
694 }
695
696 static int write_uid_list(void)
697 {
698         char *buf;
699         uint32_t count = 0;
700         struct user_info *ui;
701         size_t size = num_uids * sizeof(uint32_t);
702         int ret;
703
704         if (!num_uids)
705                 return 0;
706         buf = para_malloc(size);
707         FOR_EACH_USER(ui) {
708                 if (!ui->table)
709                         continue;
710                 write_u32(buf + count++ * sizeof(uint32_t), ui->uid);
711         }
712         ret = para_write_file(UID_LIST, buf, size);
713         free(buf);
714         return ret;
715 }
716
717 static int open_dir_table(void)
718 {
719         return osl_open_table(&dir_table_desc, &dir_table);
720 }
721
722 static void close_dir_table(void)
723 {
724         int ret;
725
726         if (!dir_table)
727                 return;
728         ret = osl_close_table(dir_table, OSL_MARK_CLEAN);
729         if (ret < 0)
730                 ERROR_LOG("failed to close dir table: %s\n", error_txt(-ret));
731         dir_table = NULL;
732 }
733
734 static void close_user_table(struct user_info *ui)
735 {
736         int ret;
737
738         if (!ui || !ui->table)
739                 return;
740         ret = osl_close_table(ui->table, OSL_MARK_CLEAN);
741         if (ret < 0)
742                 ERROR_LOG("failed to close user table %u: %s\n",
743                         (unsigned) ui->uid, error_txt(-ret));
744         free((char *)ui->desc->name);
745         ui->desc->name = NULL;
746         free((char *)ui->desc->dir);
747         ui->desc->dir = NULL;
748         free(ui->desc);
749         ui->desc = NULL;
750         ui->table = NULL;
751 }
752
753 static void close_user_tables(void)
754 {
755         struct user_info *ui;
756
757         FOR_EACH_USER(ui)
758                 close_user_table(ui);
759 }
760
761 static void close_all_tables(void)
762 {
763         close_dir_table();
764         close_user_tables();
765         free_hash_table();
766 }
767
768 static int com_create()
769 {
770         int ret = create_tables();
771         if (ret < 0)
772                 return ret;
773         ret = open_dir_table();
774         if (ret < 0)
775                 return ret;
776         ret = scan_dir(conf.base_dir_arg);
777         if (ret < 0)
778                 goto out;
779         ret = write_uid_list();
780 out:
781         close_all_tables();
782         return ret;
783 }
784
785 static int read_uid_file(void)
786 {
787         char *map;
788         size_t size;
789         int ret = mmap_full_file(UID_LIST, O_RDONLY, (void **)&map, &size, NULL);
790         uint32_t n;
791
792         if (ret < 0)
793                 return ret;
794         num_uids = size / 4;
795         /* hash table size should be a power of two and larger than the number of uids */
796         uid_hash_table_size = 4;
797         while (uid_hash_table_size < num_uids)
798                 uid_hash_table_size *= 2;
799         create_hash_table();
800         for (n = 0; n < num_uids; n++) {
801                 uint32_t uid = read_u32(map + n * sizeof(uid));
802                 ret = search_uid(uid, OPEN_USER_TABLE, NULL);
803                 if (ret < 0)
804                         goto out;
805         }
806 out:
807         para_munmap(map, size);
808         return ret;
809 }
810
811 static int com_select(void)
812 {
813         int ret;
814
815         ret = open_dir_table();
816         if (ret < 0)
817                 return ret;
818         ret = read_uid_file();
819         if (ret < 0)
820                 return ret;
821         print_statistics();
822         close_all_tables();
823         return 1;
824 }
825
826 int main(int argc, char **argv)
827 {
828         int ret;
829         struct cmdline_parser_params params = {
830                 .override = 0,
831                 .initialize = 1,
832                 .check_required = 0,
833                 .check_ambiguity = 0,
834                 .print_errors = 1
835         };
836
837         cmdline_parser_ext(argc, argv, &conf, &params); /* aborts on errors */
838         ret = -E_SYNTAX;
839         if (conf.select_given)
840                 ret = com_select();
841         else
842                 ret = com_create();
843         if (ret < 0)
844                 goto out;
845 out:
846         if (ret < 0) {
847                 ERROR_LOG("%s\n", error_txt(-ret));
848                 return -EXIT_FAILURE;
849         }
850         return EXIT_SUCCESS;
851 }