adu.c

   1 #include "adu.h"
   2 #include <dirent.h> /* readdir() */
   3
   4 #include "gcc-compat.h"
   5 #include "cmdline.h"
   6 #include "fd.h"
   7 #include "string.h"
   8 #include "error.h"
   9 #include "portable_io.h"
  10
  11 DEFINE_ERRLIST;
  12
  13 /** Command line and config file options. */
  14 static struct gengetopt_args_info conf;
  15
  16 struct user_info {
  17         uint32_t uid;
  18         struct osl_table *table;
  19         uint64_t files;
  20         uint64_t bytes;
  21         uint64_t dirs;
  22         struct osl_table_description *desc;
  23 };
  24
  25 static struct user_info *uid_hash_table;
  26
  27 /** evaluates to 1 if x < y, to -1 if x > y and to 0 if x == y */
  28 #define NUM_COMPARE(x, y) ((int)((x) < (y)) - (int)((x) > (y)))
  29
  30
  31 /**
  32  * The log function.
  33  *
  34  * \param ll Loglevel.
  35  * \param fml Usual format string.
  36  *
  37  * All XXX_LOG() macros use this function.
  38  */
  39 __printf_2_3 void __log(int ll, const char* fmt,...)
  40 {
  41         va_list argp;
  42         FILE *outfd;
  43         struct tm *tm;
  44         time_t t1;
  45         char str[255] = "";
  46
  47         if (ll < conf.loglevel_arg)
  48                 return;
  49         outfd = stderr;
  50         time(&t1);
  51         tm = localtime(&t1);
  52         strftime(str, sizeof(str), "%b %d %H:%M:%S", tm);
  53         fprintf(outfd, "%s ", str);
  54         va_start(argp, fmt);
  55         vfprintf(outfd, fmt, argp);
  56         va_end(argp);
  57 }
  58
  59 /**
  60  * Compare the size of two directories
  61  *
  62  * \param obj1 Pointer to the first object.
  63  * \param obj2 Pointer to the second object.
  64  *
  65  * This function first compares the size values as usual integers. If they compare as
  66  * equal, the address of \a obj1 and \a obj2 are compared. So this compare function
  67  * returns zero if and only if \a obj1 and \a obj2 point to the same memory area.
  68  */
  69 static int size_compare(const struct osl_object *obj1, const struct osl_object *obj2)
  70 {
  71         uint64_t d1 = *(uint64_t *)obj1->data;
  72         uint64_t d2 = *(uint64_t *)obj2->data;
  73         int ret = NUM_COMPARE(d2, d1);
  74
  75         if (ret)
  76                 return ret;
  77         //INFO_LOG("addresses: %p, %p\n", obj1->data, obj2->data);
  78         return NUM_COMPARE(obj2->data, obj1->data);
  79 }
  80
  81 /**
  82  * Compare two osl objects of string type.
  83  *
  84  * \param obj1 Pointer to the first object.
  85  * \param obj2 Pointer to the second object.
  86  *
  87  * In any case, only \p MIN(obj1->size, obj2->size) characters of each string
  88  * are taken into account.
  89  *
  90  * \return It returns an integer less than, equal to, or greater than zero if
  91  * \a obj1 is found, respectively, to be less than, to match, or be greater
  92  * than obj2.
  93  *
  94  * \sa strcmp(3), strncmp(3), osl_compare_func.
  95  */
  96 static int string_compare(const struct osl_object *obj1,
  97                 const struct osl_object *obj2)
  98 {
  99         const char *str1 = (const char *)obj1->data;
 100         const char *str2 = (const char *)obj2->data;
 101         return strncmp(str1, str2, MIN(obj1->size, obj2->size));
 102 }
 103
 104 /**
 105  * Compare two osl objects pointing to unsigned integers of 64 bit size.
 106  *
 107  * \param obj1 Pointer to the first integer.
 108  * \param obj2 Pointer to the second integer.
 109  *
 110  * \return The values required for an osl compare function.
 111  *
 112  * \sa osl_compare_func, osl_hash_compare().
 113  */
 114 static int uint64_compare(const struct osl_object *obj1,
 115                 const struct osl_object *obj2)
 116 {
 117         uint64_t d1 = read_u64((const char *)obj1->data);
 118         uint64_t d2 = read_u64((const char *)obj2->data);
 119
 120         if (d1 < d2)
 121                 return 1;
 122         if (d1 > d2)
 123                 return -1;
 124         return 0;
 125 }
 126
 127 /** The columns of the directory table. */
 128 enum dir_table_columns {
 129         /** The name of the directory. */
 130         DT_NAME,
 131         /** The dir count number. */
 132         DT_NUM,
 133         /** The number of bytes of all regular files. */
 134         DT_BYTES,
 135         /** The number of all regular files. */
 136         DT_FILES,
 137         /** Number of columns in this table. */
 138         NUM_DT_COLUMNS
 139 };
 140
 141 static struct osl_column_description dir_table_cols[] = {
 142         [DT_NAME] = {
 143                 .storage_type = OSL_MAPPED_STORAGE,
 144                 .storage_flags = OSL_RBTREE | OSL_UNIQUE,
 145                 .name = "dir",
 146                 .compare_function = string_compare,
 147         },
 148         [DT_NUM] = {
 149                 .storage_type = OSL_MAPPED_STORAGE,
 150                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
 151                 .name = "num",
 152                 .compare_function = uint64_compare,
 153                 .data_size = sizeof(uint64_t)
 154         },
 155         [DT_BYTES] = {
 156                 .storage_type = OSL_MAPPED_STORAGE,
 157                 .storage_flags =  OSL_RBTREE | OSL_FIXED_SIZE,
 158                 .compare_function = size_compare,
 159                 .name = "num_bytes",
 160                 .data_size = sizeof(uint64_t)
 161         },
 162         [DT_FILES] = {
 163                 .storage_type = OSL_MAPPED_STORAGE,
 164                 .storage_flags =  OSL_RBTREE | OSL_FIXED_SIZE,
 165                 .compare_function = size_compare,
 166                 .name = "num_files",
 167                 .data_size = sizeof(uint64_t)
 168         }
 169 };
 170
 171 static struct osl_table_description dir_table_desc = {
 172         .name = "dir_table",
 173         .num_columns = NUM_DT_COLUMNS,
 174         .flags = 0,
 175         .column_descriptions = dir_table_cols,
 176 };
 177
 178 /** The columns of the id table. */
 179 enum user_table_columns {
 180         /** The numer of the directory. */
 181         UT_DIR_NUM,
 182         /** The number of bytes of all regular files in this dir owned by this id. */
 183         UT_BYTES,
 184         /** The number of files in this dir owned by this id. */
 185         UT_FILES,
 186         /** Number of columns in this table. */
 187         NUM_UT_COLUMNS
 188 };
 189
 190 static struct osl_column_description user_table_cols[] = {
 191         [UT_DIR_NUM] = {
 192                 .storage_type = OSL_MAPPED_STORAGE,
 193                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
 194                 .name = "dir_num",
 195                 .compare_function = uint64_compare,
 196                 .data_size = sizeof(uint64_t)
 197         },
 198         [UT_BYTES] = {
 199                 .storage_type = OSL_MAPPED_STORAGE,
 200                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
 201                 .compare_function = size_compare,
 202                 .name = "num_bytes",
 203                 .data_size = sizeof(uint64_t)
 204         },
 205         [UT_FILES] = {
 206                 .storage_type = OSL_MAPPED_STORAGE,
 207                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
 208                 .compare_function = size_compare,
 209                 .name = "num_files",
 210                 .data_size = sizeof(uint64_t)
 211         },
 212 };
 213
 214 static struct osl_table *dir_table;
 215
 216 static int add_directory(char *dirname, uint64_t dir_num, uint64_t *dir_size,
 217                 uint64_t *dir_files)
 218 {
 219         struct osl_object dir_objects[NUM_DT_COLUMNS];
 220
 221         INFO_LOG("adding #%llu: %s\n", (long long unsigned)dir_num, dirname);
 222         dir_objects[DT_NAME].data = dirname;
 223         dir_objects[DT_NAME].size = strlen(dirname) + 1;
 224         dir_objects[DT_NUM].data = &dir_num;
 225         dir_objects[DT_NUM].size = sizeof(dir_num);
 226         dir_objects[DT_BYTES].data = dir_size;
 227         dir_objects[DT_BYTES].size = sizeof(*dir_size);
 228         dir_objects[DT_FILES].data = dir_files;
 229         dir_objects[DT_FILES].size = sizeof(*dir_files);
 230
 231         return osl_add_row(dir_table, dir_objects);
 232 }
 233
 234 static uint32_t num_uids;
 235
 236 static int open_user_table(struct user_info *ui, int create)
 237 {
 238         int ret;
 239
 240         ui->desc = para_malloc(sizeof(*ui->desc));
 241         ui->desc->num_columns = NUM_UT_COLUMNS;
 242         ui->desc->flags = 0;
 243         ui->desc->column_descriptions = user_table_cols;
 244         ui->desc->dir = para_strdup(conf.database_dir_arg);
 245         ui->desc->name = make_message("%u", (unsigned)ui->uid);
 246         num_uids++;
 247         INFO_LOG(".............................uid #%u: %u\n",
 248                 (unsigned)num_uids, (unsigned)ui->uid);
 249         if (create) {
 250                 ret = osl_create_table(ui->desc);
 251                 if (ret < 0)
 252                         goto err;
 253         }
 254         ret = osl_open_table(ui->desc, &ui->table);
 255         if (ret < 0)
 256                 goto err;
 257         return 1;
 258 err:
 259         free((char *)ui->desc->name);
 260         free((char *)ui->desc->dir);
 261         free(ui->desc);
 262         ui->desc->name = NULL;
 263         ui->desc->dir = NULL;
 264         ui->desc = NULL;
 265         ui->table = NULL;
 266         return ret;
 267 }
 268
 269 #define uid_hash_bits 8
 270 static uint32_t uid_hash_table_size = 1 << uid_hash_bits;
 271 #define PRIME1 0x811c9dc5
 272 #define PRIME2 0x01000193
 273
 274 static void create_hash_table(void)
 275 {
 276         uid_hash_table = para_calloc(uid_hash_table_size
 277                 * sizeof(struct user_info));
 278 }
 279
 280 static void free_hash_table(void)
 281 {
 282         free(uid_hash_table);
 283         uid_hash_table = NULL;
 284 }
 285
 286 static int create_tables(void)
 287 {
 288         int ret;
 289
 290         dir_table_desc.dir = para_strdup(conf.database_dir_arg);
 291         ret = osl_create_table(&dir_table_desc);
 292         if (ret < 0)
 293                 return ret;
 294         create_hash_table();
 295         return 1;
 296 }
 297
 298 /*
 299  * We use a hash table of size s=2^uid_hash_bits to map the uids into the
 300  * interval [0..s]. Hash collisions are treated by open addressing, i.e.
 301  * unused slots in the table are used to store different uids that hash to the
 302  * same slot.
 303  *
 304  * If a hash collision occurs, different slots are successively probed in order
 305  * to find an unused slot for the new uid. Probing is implemented via a second
 306  * hash function that maps the uid to h=(uid * PRIME2) | 1, which is always an
 307  * odd number.
 308  *
 309  * An odd number is sufficient to make sure each entry of the hash table gets
 310  * probed for probe_num between 0 and s-1 because s is a power of two, hence
 311  * the second hash value never hash a common divisor with the hash table size.
 312  * IOW: h is invertible in the ring [0..s].
 313  */
 314 static uint32_t double_hash(uint32_t uid, uint32_t probe_num)
 315 {
 316         return (uid * PRIME1 + ((uid * PRIME2) | 1) * probe_num)
 317                 % uid_hash_table_size;
 318 }
 319
 320 #define FOR_EACH_USER(ui) for (ui = uid_hash_table; ui && ui < uid_hash_table \
 321                 + uid_hash_table_size; ui++)
 322
 323 enum search_uid_flags {
 324         OPEN_USER_TABLE = 1,
 325         CREATE_USER_TABLE = 2,
 326 };
 327
 328 static int search_uid(uint32_t uid, enum search_uid_flags flags,
 329                 struct user_info **ui_ptr)
 330 {
 331         uint32_t p;
 332
 333         for (p = 0; p < uid_hash_table_size; p++) {
 334                 struct user_info *ui = uid_hash_table + double_hash(uid, p);
 335
 336                 if (!ui->table) {
 337                         int ret;
 338
 339                         if (!flags)
 340                                 return -E_BAD_UID;
 341                         ui->uid = uid;
 342                         ret = open_user_table(ui, flags & CREATE_USER_TABLE);
 343                         if (ret < 0)
 344                                 return ret;
 345                         if (ui_ptr)
 346                                 *ui_ptr = ui;
 347                         return 1;
 348                 }
 349                 if (ui->uid != uid)
 350                         continue;
 351                 if (ui_ptr)
 352                         *ui_ptr = ui;
 353                 return 0;
 354         }
 355         return flags? -E_HASH_TABLE_OVERFLOW : -E_BAD_UID;
 356 }
 357
 358 static int update_user_row(struct osl_table *t, uint64_t dir_num,
 359                 uint64_t *add)
 360 {
 361         struct osl_row *row;
 362         struct osl_object obj = {.data = &dir_num, .size = sizeof(dir_num)};
 363
 364         int ret = osl_get_row(t, UT_DIR_NUM, &obj, &row);
 365
 366         if (ret < 0 && ret != -E_RB_KEY_NOT_FOUND)
 367                 return ret;
 368         if (ret < 0) { /* this is the first file we add */
 369                 struct osl_object objects[NUM_UT_COLUMNS];
 370                 uint64_t num_files = 1;
 371
 372                 objects[UT_DIR_NUM].data = &dir_num;
 373                 objects[UT_DIR_NUM].size = sizeof(dir_num);
 374                 objects[UT_BYTES].data = add;
 375                 objects[UT_BYTES].size = sizeof(*add);
 376                 objects[UT_FILES].data = &num_files;
 377                 objects[UT_FILES].size = sizeof(num_files);
 378                 INFO_LOG("######################### ret: %d\n", ret);
 379                 ret = osl_add_row(t, objects);
 380                 INFO_LOG("######################### ret: %d\n", ret);
 381                 return ret;
 382         } else { /* add size and increment file count */
 383                 uint64_t num;
 384                 struct osl_object obj1, obj2 = {.data = &num, .size = sizeof(num)};
 385
 386                 ret = osl_get_object(t, row, UT_BYTES, &obj1);
 387                 if (ret < 0)
 388                         return ret;
 389                 num = *(uint64_t *)obj1.data + *add;
 390                 ret = osl_update_object(t, row, UT_BYTES, &obj2);
 391                 if (ret < 0)
 392                         return ret;
 393                 ret = osl_get_object(t, row, UT_FILES, &obj1);
 394                 if (ret < 0)
 395                         return ret;
 396                 num = *(uint64_t *)obj1.data + 1;
 397                 return osl_update_object(t, row, UT_FILES, &obj2);
 398         }
 399 }
 400
 401 static uint64_t num_dirs;
 402 static uint64_t num_files;
 403 static uint64_t num_bytes;
 404
 405 int scan_dir(char *dirname)
 406 {
 407         DIR *dir;
 408         struct dirent *entry;
 409         int ret, cwd_fd, ret2;
 410         uint64_t dir_size = 0, dir_files = 0;
 411         uint64_t this_dir_num = num_dirs++;
 412
 413         DEBUG_LOG("----------------- %llu: %s\n", (long long unsigned)num_dirs, dirname);
 414         ret = para_opendir(dirname, &dir, &cwd_fd);
 415         if (ret < 0) {
 416                 if (ret != -ERRNO_TO_ERROR(EACCES))
 417                         return ret;
 418                 WARNING_LOG("permission denied for %s\n", dirname);
 419                 return 1;
 420         }
 421         while ((entry = readdir(dir))) {
 422                 mode_t m;
 423                 char *tmp;
 424                 struct stat s;
 425                 uint32_t uid;
 426                 uint64_t size;
 427                 struct user_info *ui;
 428
 429                 if (!strcmp(entry->d_name, "."))
 430                         continue;
 431                 if (!strcmp(entry->d_name, ".."))
 432                         continue;
 433                 if (lstat(entry->d_name, &s) == -1) {
 434                         WARNING_LOG("lstat error for %s/%s\n", dirname,
 435                                 entry->d_name);
 436                         continue;
 437                 }
 438                 m = s.st_mode;
 439                 if (!S_ISREG(m) && !S_ISDIR(m))
 440                         continue;
 441                 if (S_ISDIR(m)) {
 442                         tmp = make_message("%s/%s", dirname, entry->d_name);
 443                         ret = scan_dir(tmp);
 444                         free(tmp);
 445                         if (ret < 0)
 446                                 goto out;
 447                         continue;
 448                 }
 449                 /* regular file */
 450                 size = s.st_size;
 451                 dir_size += size;
 452                 num_bytes += size;
 453                 dir_files++;
 454                 num_files++;
 455                 uid = s.st_uid;
 456                 ret = search_uid(uid, CREATE_USER_TABLE | OPEN_USER_TABLE, &ui);
 457                 if (ret < 0)
 458                         goto out;
 459                 ui->bytes += size;
 460                 ui->files++;
 461                 ret = update_user_row(ui->table, this_dir_num, &size);
 462                 if (ret < 0)
 463                         goto out;
 464         }
 465         ret = add_directory(dirname, this_dir_num, &dir_size, &dir_files);
 466 out:
 467         closedir(dir);
 468         ret2 = para_fchdir(cwd_fd);
 469         if (ret2 < 0 && ret >= 0)
 470                 ret = ret2;
 471         close(cwd_fd);
 472         return ret;
 473 }
 474
 475 static int get_dir_name(struct osl_row *row, char **name)
 476 {
 477         struct osl_object obj;
 478         int ret = osl_get_object(dir_table, row, DT_NAME, &obj);
 479
 480         if (ret < 0)
 481                 return ret;
 482         *name = obj.data;
 483         return 1;
 484 }
 485
 486 enum global_stats_flags {
 487         GSF_PRINT_DIRNAME = 1,
 488         GSF_PRINT_BYTES = 2,
 489         GSF_PRINT_FILES = 4,
 490         GSF_COMPUTE_SUMMARY = 8,
 491 };
 492
 493 struct global_stats_info {
 494         uint32_t count;
 495         enum global_stats_flags flags;
 496 };
 497
 498 static int global_stats_loop_function(struct osl_row *row, void *data)
 499 {
 500         struct global_stats_info *gsi = data;
 501         struct osl_object obj;
 502         char *dirname;
 503         int ret, summary = gsi->flags & GSF_COMPUTE_SUMMARY;
 504
 505         if (!gsi->count && !summary)
 506                 return -E_LOOP_COMPLETE;
 507         if (gsi->count && (gsi->flags & GSF_PRINT_DIRNAME)) {
 508                 ret = get_dir_name(row, &dirname);
 509                 if (ret < 0)
 510                         return ret;
 511                 printf("%s%s", dirname,
 512                         (gsi->flags & (GSF_PRINT_FILES | GSF_PRINT_BYTES))?
 513                                 "\t" : "\n"
 514                 );
 515         }
 516         if (summary || (gsi->count && (gsi->flags & GSF_PRINT_FILES))) {
 517                 uint64_t files;
 518                 ret = osl_get_object(dir_table, row, DT_FILES, &obj);
 519                 if (ret < 0)
 520                         return ret;
 521                 files = *(uint64_t *)obj.data;
 522                 if (gsi->count && (gsi->flags & GSF_PRINT_FILES))
 523                         printf("%llu%s", (long long unsigned)files,
 524                                 (gsi->flags & GSF_PRINT_BYTES)? "\t" : "\n");
 525                 if (summary)
 526                         num_files += files;
 527         }
 528         if (summary || (gsi->count && (gsi->flags & GSF_PRINT_BYTES))) {
 529                 uint64_t bytes;
 530                 ret = osl_get_object(dir_table, row, DT_BYTES, &obj);
 531                 if (ret < 0)
 532                         return ret;
 533                 bytes = *(uint64_t *)obj.data;
 534                 if (gsi->count && (gsi->flags & GSF_PRINT_BYTES))
 535                         printf("%llu\n", (long long unsigned)bytes);
 536                 if (summary) {
 537                         num_bytes += bytes;
 538                         num_dirs++;
 539                 }
 540         }
 541         if (gsi->count > 0)
 542                 gsi->count--;
 543         return 1;
 544 }
 545
 546 static void print_id_stats(void)
 547 {
 548         struct user_info *ui;
 549
 550         printf("--------------------- user summary (uid/dirs/files/bytes):\n");
 551         FOR_EACH_USER(ui) {
 552                 if (!ui->table)
 553                         continue;
 554                 printf("%u\t%llu\t%llu\t%llu\n", (unsigned)ui->uid,
 555                         (long long unsigned)ui->dirs,
 556                         (long long unsigned)ui->files,
 557                         (long long unsigned)ui->bytes);
 558         }
 559 }
 560
 561 enum user_stats_flags {
 562         USF_PRINT_DIRNAME = 1,
 563         USF_PRINT_BYTES = 2,
 564         USF_PRINT_FILES = 4,
 565         USF_COMPUTE_SUMMARY = 8,
 566 };
 567
 568 struct user_stats_info {
 569         uint32_t count;
 570         enum user_stats_flags flags;
 571         struct user_info *ui;
 572 };
 573
 574 static int user_stats_loop_function(struct osl_row *row, void *data)
 575 {
 576         struct user_stats_info *usi = data;
 577         struct osl_row *dir_row;
 578         struct osl_object obj;
 579         int ret, summary = usi->flags & GSF_COMPUTE_SUMMARY;
 580
 581         if (!usi->count && !summary)
 582                 return -E_LOOP_COMPLETE;
 583         if (usi->count && (usi->flags & USF_PRINT_DIRNAME)) {
 584                 char *dirname;
 585                 ret = osl_get_object(usi->ui->table, row, UT_DIR_NUM, &obj);
 586                 if (ret < 0)
 587                         return ret;
 588                 ret = osl_get_row(dir_table, DT_NUM, &obj, &dir_row);
 589                 if (ret < 0)
 590                         return ret;
 591                 ret = osl_get_object(dir_table, dir_row, DT_NAME, &obj);
 592                 if (ret < 0)
 593                         return ret;
 594                 dirname = obj.data;
 595                 printf("%s%s",
 596                         dirname,
 597                         (usi->flags & (USF_PRINT_FILES | USF_PRINT_BYTES))?
 598                                 "\t" : "\n"
 599                 );
 600         }
 601         if (summary || (usi->count && (usi->flags & USF_PRINT_FILES))) {
 602                 uint64_t files;
 603                 ret = osl_get_object(usi->ui->table, row, UT_FILES, &obj);
 604                 if (ret < 0)
 605                         return ret;
 606                 files = *(uint64_t *)obj.data;
 607                 if (usi->count && (usi->flags & USF_PRINT_FILES))
 608                         printf("%llu%s",
 609                                 (long long unsigned)files,
 610                                 (usi->flags & USF_PRINT_BYTES)? "\t" : "\n"
 611                         );
 612                 if (summary)
 613                         usi->ui->files += files;
 614         }
 615         if (summary || (usi->count && (usi->flags & USF_PRINT_BYTES))) {
 616                 uint64_t bytes;
 617                 ret = osl_get_object(usi->ui->table, row, UT_BYTES, &obj);
 618                 if (ret < 0)
 619                         return ret;
 620                 bytes = *(uint64_t *)obj.data;
 621                 if (usi->count && (usi->flags & USF_PRINT_BYTES))
 622                         printf("%llu\n", (long long unsigned)bytes);
 623                 if (summary) {
 624                         usi->ui->bytes += bytes;
 625                         usi->ui->dirs++;
 626                 }
 627
 628         }
 629         if (usi->count > 0)
 630                 usi->count--;
 631         return 1;
 632 }
 633
 634 static void print_user_stats(void)
 635 {
 636         struct user_info *ui;
 637
 638         FOR_EACH_USER(ui) {
 639                 struct user_stats_info usi = {
 640                         .count = conf.limit_arg,
 641                         .ui = ui
 642                 };
 643                 if (!ui->table)
 644                         continue;
 645                 usi.flags = USF_PRINT_DIRNAME | USF_PRINT_BYTES | USF_COMPUTE_SUMMARY;
 646                 printf("************************************************ uid %u\n",
 647                         (unsigned) ui->uid);
 648                 if (!ui->table)
 649                         continue;
 650                 printf("----------------- Largest dirs -------------------\n");
 651                 osl_rbtree_loop_reverse(ui->table, UT_BYTES, &usi,
 652                         user_stats_loop_function);
 653                 printf("---------- dirs containing most files ------------\n");
 654                 usi.count = conf.limit_arg,
 655                 usi.flags = USF_PRINT_DIRNAME | USF_PRINT_FILES;
 656                 osl_rbtree_loop_reverse(ui->table, UT_FILES, &usi,
 657                         user_stats_loop_function);
 658         }
 659 }
 660
 661 static int print_statistics(void)
 662 {
 663         int ret;
 664         struct global_stats_info gsi = {
 665                 .count = conf.limit_arg,
 666                 .flags = GSF_PRINT_DIRNAME | GSF_PRINT_BYTES | GSF_COMPUTE_SUMMARY
 667         };
 668
 669         printf("----------------- Largest dirs -------------------\n");
 670         ret = osl_rbtree_loop_reverse(dir_table, DT_BYTES, &gsi,
 671                 global_stats_loop_function);
 672         if (ret < 0 && ret != -E_LOOP_COMPLETE)
 673                 return ret;
 674         gsi.count = conf.limit_arg;
 675
 676         gsi.flags = GSF_PRINT_DIRNAME | GSF_PRINT_FILES;
 677         printf("---------- dirs containing most files ------------\n");
 678         ret = osl_rbtree_loop_reverse(dir_table, DT_FILES, &gsi,
 679                 global_stats_loop_function);
 680         if (ret < 0 && ret != -E_LOOP_COMPLETE)
 681                 return ret;
 682
 683         printf("------------------ Global summary (dirs/files/bytes)\n"
 684                 "%llu\t%llu\t%llu\n",
 685                 (long long unsigned)num_dirs, (long long unsigned)num_files,
 686                 (long long unsigned)num_bytes);
 687         print_user_stats();
 688         print_id_stats();
 689         return 1;
 690 }
 691
 692 static char *get_uid_list_name(void)
 693 {
 694         return make_message("%s/uid_list", conf.database_dir_arg);
 695 }
 696
 697 static int write_uid_list(void)
 698 {
 699         char *buf, *filename;
 700         uint32_t count = 0;
 701         struct user_info *ui;
 702         size_t size = num_uids * sizeof(uint32_t);
 703         int ret;
 704
 705         if (!num_uids)
 706                 return 0;
 707         buf = para_malloc(size);
 708         FOR_EACH_USER(ui) {
 709                 if (!ui->table)
 710                         continue;
 711                 write_u32(buf + count++ * sizeof(uint32_t), ui->uid);
 712         }
 713         filename = get_uid_list_name();
 714         ret = para_write_file(filename, buf, size);
 715         free(filename);
 716         free(buf);
 717         return ret;
 718 }
 719
 720 static int open_dir_table(void)
 721 {
 722         if (!dir_table_desc.dir) /* we did not create the table */
 723                 dir_table_desc.dir = para_strdup(conf.database_dir_arg);
 724         return osl_open_table(&dir_table_desc, &dir_table);
 725 }
 726
 727 static void close_dir_table(void)
 728 {
 729         int ret;
 730
 731         if (!dir_table)
 732                 return;
 733         ret = osl_close_table(dir_table, OSL_MARK_CLEAN);
 734         if (ret < 0)
 735                 ERROR_LOG("failed to close dir table: %s\n", error_txt(-ret));
 736         free((char *)dir_table_desc.dir);
 737         dir_table = NULL;
 738 }
 739
 740 static void close_user_table(struct user_info *ui)
 741 {
 742         int ret;
 743
 744         if (!ui || !ui->table)
 745                 return;
 746         ret = osl_close_table(ui->table, OSL_MARK_CLEAN);
 747         if (ret < 0)
 748                 ERROR_LOG("failed to close user table %u: %s\n",
 749                         (unsigned) ui->uid, error_txt(-ret));
 750         free((char *)ui->desc->name);
 751         ui->desc->name = NULL;
 752         free((char *)ui->desc->dir);
 753         ui->desc->dir = NULL;
 754         free(ui->desc);
 755         ui->desc = NULL;
 756         ui->table = NULL;
 757 }
 758
 759 static void close_user_tables(void)
 760 {
 761         struct user_info *ui;
 762
 763         FOR_EACH_USER(ui)
 764                 close_user_table(ui);
 765 }
 766
 767 static void close_all_tables(void)
 768 {
 769         close_dir_table();
 770         close_user_tables();
 771         free_hash_table();
 772 }
 773
 774 static int com_create()
 775 {
 776         int ret = create_tables();
 777         if (ret < 0)
 778                 return ret;
 779         ret = open_dir_table();
 780         if (ret < 0)
 781                 return ret;
 782         ret = scan_dir(conf.base_dir_arg);
 783         if (ret < 0)
 784                 goto out;
 785         ret = write_uid_list();
 786 out:
 787         close_all_tables();
 788         return ret;
 789 }
 790
 791 static int read_uid_file(void)
 792 {
 793         size_t size;
 794         uint32_t n;
 795         char *filename = get_uid_list_name(), *map;
 796         int ret = mmap_full_file(filename, O_RDONLY, (void **)&map, &size, NULL);
 797
 798         free(filename);
 799         if (ret < 0)
 800                 return ret;
 801         num_uids = size / 4;
 802         /* hash table size should be a power of two and larger than the number of uids */
 803         uid_hash_table_size = 4;
 804         while (uid_hash_table_size < num_uids)
 805                 uid_hash_table_size *= 2;
 806         create_hash_table();
 807         for (n = 0; n < num_uids; n++) {
 808                 uint32_t uid = read_u32(map + n * sizeof(uid));
 809                 ret = search_uid(uid, OPEN_USER_TABLE, NULL);
 810                 if (ret < 0)
 811                         goto out;
 812         }
 813 out:
 814         para_munmap(map, size);
 815         return ret;
 816 }
 817
 818 static int com_select(void)
 819 {
 820         int ret;
 821
 822         ret = open_dir_table();
 823         if (ret < 0)
 824                 return ret;
 825         ret = read_uid_file();
 826         if (ret < 0)
 827                 return ret;
 828         print_statistics();
 829         close_all_tables();
 830         return 1;
 831 }
 832
 833 int main(int argc, char **argv)
 834 {
 835         int ret;
 836         struct cmdline_parser_params params = {
 837                 .override = 0,
 838                 .initialize = 1,
 839                 .check_required = 0,
 840                 .check_ambiguity = 0,
 841                 .print_errors = 1
 842         };
 843
 844         cmdline_parser_ext(argc, argv, &conf, &params); /* aborts on errors */
 845         ret = -E_SYNTAX;
 846         if (conf.select_given)
 847                 ret = com_select();
 848         else
 849                 ret = com_create();
 850         if (ret < 0)
 851                 goto out;
 852 out:
 853         if (ret < 0) {
 854                 ERROR_LOG("%s\n", error_txt(-ret));
 855                 return -EXIT_FAILURE;
 856         }
 857         return EXIT_SUCCESS;
 858 }