adu.c

   1 #include "adu.h"
   2 #include <dirent.h> /* readdir() */
   3 #include <pwd.h>
   4
   5 #include "gcc-compat.h"
   6 #include "cmdline.h"
   7 #include "fd.h"
   8 #include "string.h"
   9 #include "error.h"
  10 #include "portable_io.h"
  11
  12 DEFINE_ERRLIST;
  13 int osl_errno;
  14
  15 /** Command line and config file options. */
  16 static struct gengetopt_args_info conf;
  17
  18 enum uid_info_flags {
  19         /** whether this slot of the hash table is used. */
  20         UI_FL_SLOT_USED = 1,
  21         /** whether this uid should be taken into account. */
  22         UI_FL_ADMISSIBLE = 2,
  23 };
  24
  25 struct user_info {
  26         uint32_t uid;
  27         uint32_t flags;
  28         char *pw_name;
  29         struct osl_table *table;
  30         uint64_t files;
  31         uint64_t bytes;
  32         uint64_t dirs;
  33         struct osl_table_description *desc;
  34 };
  35
  36 /** The decimal representation of an uint64_t never exceeds that size. */
  37 #define FORMATED_VALUE_SIZE 25
  38
  39 #define FOR_EACH_USER(ui) for (ui = uid_hash_table; ui && ui < uid_hash_table \
  40                 + uid_hash_table_size; ui++)
  41
  42
  43 /**
  44  * Contains info for each user that owns at least one regular file.
  45  *
  46  * Even users that are not taken into account because of the --uid
  47  * option occupy a slot in this hash table. This allows to find out
  48  * quicky whether a uid is admissible. And yes, this has to be fast.
  49  */
  50 static struct user_info *uid_hash_table;
  51
  52 /* these get filled in by the select command. */
  53 static char count_unit_buf[4] = "( )", size_unit_buf[4] = "( )";
  54
  55 static inline int ui_used(struct user_info *ui)
  56 {
  57         return ui->flags & UI_FL_SLOT_USED;
  58 }
  59
  60 static inline int ui_admissible(struct user_info *ui)
  61 {
  62         return ui->flags & UI_FL_ADMISSIBLE;
  63 }
  64
  65 struct uid_range {
  66         uint32_t low;
  67         uint32_t high;
  68 };
  69
  70 static struct uid_range *admissible_uids;
  71
  72 static inline int check_uid_arg(const char *arg, uint32_t *uid)
  73 {
  74         const uint32_t max = ~0U;
  75         /*
  76          * we need an 64-bit int for string -> uid conversion because strtoll()
  77          * returns a signed value.
  78          */
  79         int64_t val;
  80         int ret = atoi64(arg, &val);
  81
  82         if (ret < 0)
  83                 return ret;
  84         if (val < 0 || val > max)
  85                 return -ERRNO_TO_ERROR(EINVAL);
  86         *uid = val;
  87         return 1;
  88 }
  89
  90 static int parse_uid_range(const char *orig_arg, struct uid_range *ur)
  91 {
  92         int ret;
  93         char *arg = adu_strdup(orig_arg), *p = strchr(arg, '-');
  94
  95         if (!p || p == arg) { /* -42 or 42 */
  96                 ret = check_uid_arg(p? p + 1 : arg, &ur->high);
  97                 if (ret < 0)
  98                         goto out;
  99                 ur->low = p? 0 : ur->high;
 100                 ret = 1;
 101                 goto out;
 102         }
 103         /* 42- or 42-4711 */
 104         *p = '\0';
 105         p++;
 106         ret = check_uid_arg(arg, &ur->low);
 107         if (ret < 0)
 108                 goto out;
 109         ur->high = ~0U;
 110         if (*p) { /* 42-4711 */
 111                 ret = check_uid_arg(p, &ur->high);
 112                 if (ret < 0)
 113                         goto out;
 114         }
 115         if (ur->low > ur->high)
 116                 ret = -ERRNO_TO_ERROR(EINVAL);
 117 out:
 118         if (ret < 0)
 119                 ERROR_LOG("bad uid option: %s\n", orig_arg);
 120         else
 121                 INFO_LOG("admissible uid range: %u - %u\n", ur->low,
 122                         ur->high);
 123         free(arg);
 124         return ret;
 125 }
 126
 127 /** evaluates to 1 if x < y, to -1 if x > y and to 0 if x == y */
 128 #define NUM_COMPARE(x, y) ((int)((x) < (y)) - (int)((x) > (y)))
 129
 130 /**
 131  * The log function.
 132  *
 133  * \param ll Loglevel.
 134  * \param fml Usual format string.
 135  *
 136  * All XXX_LOG() macros use this function.
 137  */
 138 __printf_2_3 void __log(int ll, const char* fmt,...)
 139 {
 140         va_list argp;
 141         FILE *outfd;
 142         struct tm *tm;
 143         time_t t1;
 144         char str[255] = "";
 145
 146         if (ll < conf.loglevel_arg)
 147                 return;
 148         outfd = stderr;
 149         time(&t1);
 150         tm = localtime(&t1);
 151         strftime(str, sizeof(str), "%b %d %H:%M:%S", tm);
 152         fprintf(outfd, "%s ", str);
 153         va_start(argp, fmt);
 154         vfprintf(outfd, fmt, argp);
 155         va_end(argp);
 156 }
 157
 158 /**
 159  * Compare the size of two directories
 160  *
 161  * \param obj1 Pointer to the first object.
 162  * \param obj2 Pointer to the second object.
 163  *
 164  * This function first compares the size values as usual integers. If they compare as
 165  * equal, the address of \a obj1 and \a obj2 are compared. So this compare function
 166  * returns zero if and only if \a obj1 and \a obj2 point to the same memory area.
 167  */
 168 static int size_compare(const struct osl_object *obj1, const struct osl_object *obj2)
 169 {
 170         uint64_t d1 = *(uint64_t *)obj1->data;
 171         uint64_t d2 = *(uint64_t *)obj2->data;
 172         int ret = NUM_COMPARE(d2, d1);
 173
 174         if (ret)
 175                 return ret;
 176         //INFO_LOG("addresses: %p, %p\n", obj1->data, obj2->data);
 177         return NUM_COMPARE(obj2->data, obj1->data);
 178 }
 179
 180 /**
 181  * Compare two osl objects pointing to unsigned integers of 64 bit size.
 182  *
 183  * \param obj1 Pointer to the first integer.
 184  * \param obj2 Pointer to the second integer.
 185  *
 186  * \return The values required for an osl compare function.
 187  *
 188  * \sa osl_compare_func, osl_hash_compare().
 189  */
 190 static int uint64_compare(const struct osl_object *obj1,
 191                 const struct osl_object *obj2)
 192 {
 193         uint64_t d1 = read_u64((const char *)obj1->data);
 194         uint64_t d2 = read_u64((const char *)obj2->data);
 195
 196         if (d1 < d2)
 197                 return 1;
 198         if (d1 > d2)
 199                 return -1;
 200         return 0;
 201 }
 202
 203 /** The columns of the directory table. */
 204 enum dir_table_columns {
 205         /** The name of the directory. */
 206         DT_NAME,
 207         /** The dir count number. */
 208         DT_NUM,
 209         /** The number of the parent directory. */
 210         DT_PARENT_NUM,
 211         /** The number of bytes of all regular files. */
 212         DT_BYTES,
 213         /** The number of all regular files. */
 214         DT_FILES,
 215         /** Number of columns in this table. */
 216         NUM_DT_COLUMNS
 217 };
 218
 219 static struct osl_column_description dir_table_cols[] = {
 220         [DT_NAME] = {
 221                 .storage_type = OSL_MAPPED_STORAGE,
 222                 .storage_flags = 0,
 223                 .name = "dir",
 224         },
 225         [DT_NUM] = {
 226                 .storage_type = OSL_MAPPED_STORAGE,
 227                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
 228                 .name = "num",
 229                 .compare_function = uint64_compare,
 230                 .data_size = sizeof(uint64_t)
 231         },
 232         [DT_PARENT_NUM] = {
 233                 .storage_type = OSL_MAPPED_STORAGE,
 234                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
 235                 .name = "parent_num",
 236                 .compare_function = size_compare,
 237                 .data_size = sizeof(uint64_t)
 238         },
 239         [DT_BYTES] = {
 240                 .storage_type = OSL_MAPPED_STORAGE,
 241                 .storage_flags =  OSL_RBTREE | OSL_FIXED_SIZE,
 242                 .compare_function = size_compare,
 243                 .name = "num_bytes",
 244                 .data_size = sizeof(uint64_t)
 245         },
 246         [DT_FILES] = {
 247                 .storage_type = OSL_MAPPED_STORAGE,
 248                 .storage_flags =  OSL_RBTREE | OSL_FIXED_SIZE,
 249                 .compare_function = size_compare,
 250                 .name = "num_files",
 251                 .data_size = sizeof(uint64_t)
 252         }
 253 };
 254
 255 static struct osl_table_description dir_table_desc = {
 256         .name = "dir_table",
 257         .num_columns = NUM_DT_COLUMNS,
 258         .flags = 0,
 259         .column_descriptions = dir_table_cols,
 260 };
 261
 262 /** The columns of the id table. */
 263 enum user_table_columns {
 264         /** The numer of the directory. */
 265         UT_DIR_NUM,
 266         /** The number of bytes of all regular files in this dir owned by this id. */
 267         UT_BYTES,
 268         /** The number of files in this dir owned by this id. */
 269         UT_FILES,
 270         /** Number of columns in this table. */
 271         NUM_UT_COLUMNS
 272 };
 273
 274 static struct osl_column_description user_table_cols[] = {
 275         [UT_DIR_NUM] = {
 276                 .storage_type = OSL_MAPPED_STORAGE,
 277                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
 278                 .name = "dir_num",
 279                 .compare_function = uint64_compare,
 280                 .data_size = sizeof(uint64_t)
 281         },
 282         [UT_BYTES] = {
 283                 .storage_type = OSL_MAPPED_STORAGE,
 284                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
 285                 .compare_function = size_compare,
 286                 .name = "num_bytes",
 287                 .data_size = sizeof(uint64_t)
 288         },
 289         [UT_FILES] = {
 290                 .storage_type = OSL_MAPPED_STORAGE,
 291                 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
 292                 .compare_function = size_compare,
 293                 .name = "num_files",
 294                 .data_size = sizeof(uint64_t)
 295         },
 296 };
 297
 298 static struct osl_table *dir_table;
 299
 300 static int add_directory(char *dirname, uint64_t *dir_num, uint64_t *parent_dir_num,
 301                 uint64_t *dir_size, uint64_t *dir_files)
 302 {
 303         struct osl_object dir_objects[NUM_DT_COLUMNS];
 304
 305         INFO_LOG("adding #%llu: %s\n", (long long unsigned)*dir_num, dirname);
 306         dir_objects[DT_NAME].data = dirname;
 307         dir_objects[DT_NAME].size = strlen(dirname) + 1;
 308         dir_objects[DT_NUM].data = dir_num;
 309         dir_objects[DT_NUM].size = sizeof(*dir_num);
 310         dir_objects[DT_PARENT_NUM].data = parent_dir_num;
 311         dir_objects[DT_PARENT_NUM].size = sizeof(*parent_dir_num);
 312         dir_objects[DT_BYTES].data = dir_size;
 313         dir_objects[DT_BYTES].size = sizeof(*dir_size);
 314         dir_objects[DT_FILES].data = dir_files;
 315         dir_objects[DT_FILES].size = sizeof(*dir_files);
 316         return osl(osl_add_row(dir_table, dir_objects));
 317 }
 318
 319 static uint32_t num_uids;
 320
 321 static int open_user_table(struct user_info *ui, int create)
 322 {
 323         int ret;
 324         struct passwd *pw;
 325
 326         ui->desc = adu_malloc(sizeof(*ui->desc));
 327         ui->desc->num_columns = NUM_UT_COLUMNS;
 328         ui->desc->flags = 0;
 329         ui->desc->column_descriptions = user_table_cols;
 330         ui->desc->dir = adu_strdup(conf.database_dir_arg);
 331         ui->desc->name = make_message("%u", (unsigned)ui->uid);
 332         pw = getpwuid(ui->uid);
 333         if (pw && pw->pw_name)
 334                 ui->pw_name = adu_strdup(pw->pw_name);
 335
 336         INFO_LOG(".............................uid #%u: %u\n",
 337                 (unsigned)num_uids, (unsigned)ui->uid);
 338         if (create) {
 339                 ret = osl(osl_create_table(ui->desc));
 340                 if (ret < 0)
 341                         goto err;
 342                 num_uids++;
 343         }
 344         ret = osl(osl_open_table(ui->desc, &ui->table));
 345         if (ret < 0)
 346                 goto err;
 347         return 1;
 348 err:
 349         free((char *)ui->desc->name);
 350         free((char *)ui->desc->dir);
 351         free(ui->pw_name);
 352         free(ui->desc);
 353         ui->desc->name = NULL;
 354         ui->desc->dir = NULL;
 355         ui->desc = NULL;
 356         ui->table = NULL;
 357         ui->flags = 0;
 358         return ret;
 359 }
 360
 361 #define uid_hash_bits 8
 362 static uint32_t uid_hash_table_size = 1 << uid_hash_bits;
 363 #define PRIME1 0x811c9dc5
 364 #define PRIME2 0x01000193
 365
 366 static void create_hash_table(void)
 367 {
 368         uid_hash_table = adu_calloc(uid_hash_table_size
 369                 * sizeof(struct user_info));
 370 }
 371
 372 static void free_hash_table(void)
 373 {
 374         free(uid_hash_table);
 375         uid_hash_table = NULL;
 376 }
 377
 378 static int create_tables(void)
 379 {
 380         int ret;
 381
 382         dir_table_desc.dir = adu_strdup(conf.database_dir_arg);
 383         ret = osl(osl_create_table(&dir_table_desc));
 384         if (ret < 0)
 385                 return ret;
 386         create_hash_table();
 387         return 1;
 388 }
 389
 390 static void close_dir_table(void)
 391 {
 392         int ret;
 393
 394         if (!dir_table)
 395                 return;
 396         ret = osl(osl_close_table(dir_table, OSL_MARK_CLEAN));
 397         if (ret < 0)
 398                 ERROR_LOG("failed to close dir table: %s\n", adu_strerror(-ret));
 399         free((char *)dir_table_desc.dir);
 400         dir_table = NULL;
 401 }
 402
 403 static void close_user_table(struct user_info *ui)
 404 {
 405         int ret;
 406
 407         if (!ui || !ui_used(ui) || !ui_admissible(ui))
 408                 return;
 409         ret = osl(osl_close_table(ui->table, OSL_MARK_CLEAN));
 410         if (ret < 0)
 411                 ERROR_LOG("failed to close user table %u: %s\n",
 412                         (unsigned) ui->uid, adu_strerror(-ret));
 413         free((char *)ui->desc->name);
 414         ui->desc->name = NULL;
 415         free((char *)ui->desc->dir);
 416         ui->desc->dir = NULL;
 417         free(ui->pw_name);
 418         ui->pw_name = NULL;
 419         free(ui->desc);
 420         ui->desc = NULL;
 421         ui->table = NULL;
 422         ui->flags = 0;
 423 }
 424
 425 static void close_user_tables(void)
 426 {
 427         struct user_info *ui;
 428
 429         FOR_EACH_USER(ui)
 430                 close_user_table(ui);
 431 }
 432
 433 static void close_all_tables(void)
 434 {
 435         close_dir_table();
 436         close_user_tables();
 437         free_hash_table();
 438 }
 439
 440 static int signum;
 441
 442 static void signal_handler(int s)
 443 {
 444         signum = s;
 445 }
 446
 447 static void check_signals(void)
 448 {
 449         if (likely(!signum))
 450                 return;
 451         EMERG_LOG("caught signal %d\n", signum);
 452         close_all_tables();
 453         exit(EXIT_FAILURE);
 454 }
 455
 456 static int init_signals(void)
 457 {
 458         if (signal(SIGINT, &signal_handler) == SIG_ERR)
 459                 return -E_SIGNAL_SIG_ERR;
 460         if (signal(SIGTERM, &signal_handler) == SIG_ERR)
 461                 return -E_SIGNAL_SIG_ERR;
 462         return 1;
 463 }
 464
 465 /*
 466  * We use a hash table of size s=2^uid_hash_bits to map the uids into the
 467  * interval [0..s]. Hash collisions are treated by open addressing, i.e.
 468  * unused slots in the table are used to store different uids that hash to the
 469  * same slot.
 470  *
 471  * If a hash collision occurs, different slots are successively probed in order
 472  * to find an unused slot for the new uid. Probing is implemented via a second
 473  * hash function that maps the uid to h=(uid * PRIME2) | 1, which is always an
 474  * odd number.
 475  *
 476  * An odd number is sufficient to make sure each entry of the hash table gets
 477  * probed for probe_num between 0 and s-1 because s is a power of two, hence
 478  * the second hash value has never a common divisor with the hash table size.
 479  * IOW: h is invertible in the ring [0..s].
 480  */
 481 static uint32_t double_hash(uint32_t uid, uint32_t probe_num)
 482 {
 483         return (uid * PRIME1 + ((uid * PRIME2) | 1) * probe_num)
 484                 % uid_hash_table_size;
 485 }
 486
 487 enum search_uid_flags {
 488         OPEN_USER_TABLE = 1,
 489         CREATE_USER_TABLE = 2,
 490 };
 491
 492 static int uid_is_admissible(uint32_t uid)
 493 {
 494         int i;
 495
 496         for (i = 0; i < conf.uid_given; i++) {
 497                 struct uid_range *ur = admissible_uids + i;
 498
 499                 if (ur->low <= uid && ur->high >= uid)
 500                         break;
 501         }
 502         i = !conf.uid_given || i < conf.uid_given;
 503         DEBUG_LOG("uid %u is %sadmissible\n", (unsigned)uid,
 504                 i? "" : "not ");
 505         return i;
 506 }
 507
 508 static int search_uid(uint32_t uid, enum search_uid_flags flags,
 509                 struct user_info **ui_ptr)
 510 {
 511         uint32_t p;
 512
 513         for (p = 0; p < uid_hash_table_size; p++) {
 514                 struct user_info *ui = uid_hash_table + double_hash(uid, p);
 515
 516                 if (!ui_used(ui)) {
 517                         int ret;
 518                         if (!flags)
 519                                 return -E_BAD_UID;
 520                         ui->uid = uid;
 521                         ui->flags |= UI_FL_SLOT_USED;
 522                         if (!uid_is_admissible(uid))
 523                                 return 0;
 524                         ui->flags |= UI_FL_ADMISSIBLE;
 525                         ret = open_user_table(ui, flags & CREATE_USER_TABLE);
 526                         if (ret < 0)
 527                                 return ret;
 528
 529                         if (ui_ptr)
 530                                 *ui_ptr = ui;
 531                         return 1;
 532                 }
 533                 if (ui->uid != uid)
 534                         continue;
 535                 if (ui_ptr)
 536                         *ui_ptr = ui;
 537                 return 0;
 538         }
 539         return flags? -E_HASH_TABLE_OVERFLOW : -E_BAD_UID;
 540 }
 541
 542 static int update_user_row(struct osl_table *t, uint64_t dir_num,
 543                 uint64_t *add)
 544 {
 545         struct osl_row *row;
 546         struct osl_object obj = {.data = &dir_num, .size = sizeof(dir_num)};
 547
 548         int ret = osl(osl_get_row(t, UT_DIR_NUM, &obj, &row));
 549
 550         if (ret == -E_OSL && osl_errno != E_OSL_RB_KEY_NOT_FOUND)
 551                 return ret;
 552         if (ret < 0) { /* this is the first file we add */
 553                 struct osl_object objects[NUM_UT_COLUMNS];
 554                 uint64_t num_files = 1;
 555
 556                 objects[UT_DIR_NUM].data = &dir_num;
 557                 objects[UT_DIR_NUM].size = sizeof(dir_num);
 558                 objects[UT_BYTES].data = add;
 559                 objects[UT_BYTES].size = sizeof(*add);
 560                 objects[UT_FILES].data = &num_files;
 561                 objects[UT_FILES].size = sizeof(num_files);
 562                 INFO_LOG("######################### ret: %d\n", ret);
 563                 ret = osl(osl_add_row(t, objects));
 564                 INFO_LOG("######################### ret: %d\n", ret);
 565                 return ret;
 566         } else { /* add size and increment file count */
 567                 uint64_t num;
 568                 struct osl_object obj1, obj2 = {.data = &num, .size = sizeof(num)};
 569
 570                 ret = osl(osl_get_object(t, row, UT_BYTES, &obj1));
 571                 if (ret < 0)
 572                         return ret;
 573                 num = *(uint64_t *)obj1.data + *add;
 574                 ret = osl(osl_update_object(t, row, UT_BYTES, &obj2));
 575                 if (ret < 0)
 576                         return ret;
 577                 ret = osl(osl_get_object(t, row, UT_FILES, &obj1));
 578                 if (ret < 0)
 579                         return ret;
 580                 num = *(uint64_t *)obj1.data + 1;
 581                 return osl(osl_update_object(t, row, UT_FILES, &obj2));
 582         }
 583 }
 584
 585 static uint64_t num_dirs;
 586 static uint64_t num_files;
 587 static uint64_t num_bytes;
 588
 589 /* id of the device containing the base dir. */
 590 static dev_t device_id;
 591
 592 static int scan_dir(char *dirname, uint64_t *parent_dir_num)
 593 {
 594         DIR *dir;
 595         struct dirent *entry;
 596         int ret, cwd_fd, ret2;
 597         uint64_t dir_size = 0, dir_files = 0;
 598         uint64_t this_dir_num = ++num_dirs;
 599
 600         check_signals();
 601         DEBUG_LOG("----------------- %llu: %s\n", (long long unsigned)num_dirs, dirname);
 602         ret = adu_opendir(dirname, &dir, &cwd_fd);
 603         if (ret < 0) {
 604                 if (ret != -ERRNO_TO_ERROR(EACCES))
 605                         return ret;
 606                 WARNING_LOG("permission denied for %s\n", dirname);
 607                 return 1;
 608         }
 609         while ((entry = readdir(dir))) {
 610                 mode_t m;
 611                 struct stat s;
 612                 uint32_t uid;
 613                 uint64_t size;
 614                 struct user_info *ui;
 615
 616                 if (!strcmp(entry->d_name, "."))
 617                         continue;
 618                 if (!strcmp(entry->d_name, ".."))
 619                         continue;
 620                 if (lstat(entry->d_name, &s) == -1) {
 621                         WARNING_LOG("lstat error for %s/%s\n", dirname,
 622                                 entry->d_name);
 623                         continue;
 624                 }
 625                 m = s.st_mode;
 626                 if (!S_ISREG(m) && !S_ISDIR(m))
 627                         continue;
 628                 if (S_ISDIR(m)) {
 629                         if (conf.one_file_system_given && s.st_dev != device_id)
 630                                 continue;
 631                         ret = scan_dir(entry->d_name, &this_dir_num);
 632                         if (ret < 0)
 633                                 goto out;
 634                         continue;
 635                 }
 636                 /* regular file */
 637                 size = s.st_size;
 638                 dir_size += size;
 639                 num_bytes += size;
 640                 dir_files++;
 641                 num_files++;
 642                 uid = s.st_uid;
 643                 ret = search_uid(uid, CREATE_USER_TABLE | OPEN_USER_TABLE, &ui);
 644                 if (ret < 0)
 645                         goto out;
 646                 ui->bytes += size;
 647                 ui->files++;
 648                 ret = update_user_row(ui->table, this_dir_num, &size);
 649                 if (ret < 0)
 650                         goto out;
 651         }
 652         ret = add_directory(dirname, &this_dir_num, parent_dir_num,
 653                         &dir_size, &dir_files);
 654 out:
 655         closedir(dir);
 656         ret2 = adu_fchdir(cwd_fd);
 657         if (ret2 < 0 && ret >= 0)
 658                 ret = ret2;
 659         close(cwd_fd);
 660         return ret;
 661 }
 662
 663 static int get_dir_name_by_number(uint64_t *dirnum, char **name)
 664 {
 665         char *result = NULL, *tmp;
 666         struct osl_row *row;
 667         uint64_t val = *dirnum;
 668         struct osl_object obj = {.data = &val, .size = sizeof(val)};
 669         int ret;
 670
 671 again:
 672         ret = osl(osl_get_row(dir_table, DT_NUM, &obj, &row));
 673         if (ret < 0)
 674                 goto out;
 675         ret = osl(osl_get_object(dir_table, row, DT_NAME, &obj));
 676         if (ret < 0)
 677                 goto out;
 678         if (result) {
 679                 tmp = make_message("%s/%s", (char *)obj.data, result);
 680                 free(result);
 681                 result = tmp;
 682         } else
 683                 result = adu_strdup((char *)obj.data);
 684         ret = osl(osl_get_object(dir_table, row, DT_PARENT_NUM, &obj));
 685         if (ret < 0)
 686                 goto out;
 687         val = *(uint64_t *)obj.data;
 688         if (val)
 689                 goto again;
 690 out:
 691         if (ret < 0) {
 692                 free(result);
 693                 *name = NULL;
 694         } else
 695                 *name = result;
 696         return ret;
 697 }
 698
 699 static int get_dir_name_of_row(struct osl_row *dir_table_row, char **name)
 700 {
 701         struct osl_object obj;
 702         int ret;
 703         char *this_dir, *prefix = NULL;
 704
 705         *name = NULL;
 706         ret = osl(osl_get_object(dir_table, dir_table_row, DT_NAME, &obj));
 707         if (ret < 0)
 708                 return ret;
 709         this_dir = adu_strdup((char *)obj.data);
 710         ret = osl(osl_get_object(dir_table, dir_table_row, DT_PARENT_NUM, &obj));
 711         if (ret < 0)
 712                 goto out;
 713         if (!*(uint64_t *)obj.data) {
 714                 *name = this_dir;
 715                 return 1;
 716         }
 717         ret = get_dir_name_by_number((uint64_t *)obj.data, &prefix);
 718         if (ret < 0)
 719                 goto out;
 720         *name = make_message("%s/%s", prefix, this_dir);
 721         free(prefix);
 722         ret = 1;
 723 out:
 724         free(this_dir);
 725         return ret;
 726 }
 727
 728 const uint64_t size_unit_divisors[] = {
 729         [size_unit_arg_b] = 1ULL,
 730         [size_unit_arg_k] = 1024ULL,
 731         [size_unit_arg_m] = 1024ULL * 1024ULL,
 732         [size_unit_arg_g] = 1024ULL * 1024ULL * 1024ULL,
 733         [size_unit_arg_t] = 1024ULL * 1024ULL * 1024ULL * 1024ULL,
 734 };
 735
 736 const uint64_t count_unit_divisors[] = {
 737
 738         [count_unit_arg_n] = 1ULL,
 739         [count_unit_arg_k] = 1000ULL,
 740         [count_unit_arg_m] = 1000ULL * 1000ULL,
 741         [count_unit_arg_g] = 1000ULL * 1000ULL * 1000ULL,
 742         [count_unit_arg_t] = 1000ULL * 1000ULL * 1000ULL * 1000ULL,
 743 };
 744
 745 const char size_unit_abbrevs[] = " BKMGT";
 746 const char count_unit_abbrevs[] = "  kmgt";
 747
 748 static enum enum_size_unit format_size_value(enum enum_size_unit unit,
 749                 uint64_t value, int print_unit, char *result)
 750 {
 751         enum enum_size_unit u = unit;
 752         char unit_buf[2] = "\0\0";
 753
 754         if (unit == size_unit_arg_h) /* human readable */
 755                 for (u = size_unit_arg_b; u < size_unit_arg_t &&
 756                                 value > size_unit_divisors[u + 1]; u++)
 757                         ; /* nothing */
 758         if (print_unit)
 759                 unit_buf[0] = size_unit_abbrevs[u];
 760         sprintf(result, "%llu%s",
 761                 (long long unsigned)value / size_unit_divisors[u], unit_buf);
 762         return u;
 763 }
 764
 765 static enum enum_count_unit format_count_value(enum enum_count_unit unit,
 766                 uint64_t value, int print_unit, char *result)
 767 {
 768         enum enum_count_unit u = unit;
 769         char unit_buf[2] = "\0\0";
 770
 771         if (unit == count_unit_arg_h) /* human readable */
 772                 for (u = count_unit_arg_n; u < count_unit_arg_t &&
 773                                 value > count_unit_divisors[u + 1]; u++)
 774                         ; /* nothing */
 775         if (print_unit)
 776                 unit_buf[0] = count_unit_abbrevs[u];
 777         sprintf(result, "%llu%s",
 778                 (long long unsigned)value / count_unit_divisors[u], unit_buf);
 779         return u;
 780 }
 781
 782 enum global_stats_flags {
 783         GSF_PRINT_DIRNAME = 1,
 784         GSF_PRINT_BYTES = 2,
 785         GSF_PRINT_FILES = 4,
 786         GSF_COMPUTE_SUMMARY = 8,
 787 };
 788
 789 struct global_stats_info {
 790         uint32_t count;
 791         int ret;
 792         int osl_errno;
 793         enum global_stats_flags flags;
 794 };
 795
 796 static int global_stats_loop_function(struct osl_row *row, void *data)
 797 {
 798         struct global_stats_info *gsi = data;
 799         struct osl_object obj;
 800         char *dirname, formated_value[FORMATED_VALUE_SIZE];
 801         int ret, summary = gsi->flags & GSF_COMPUTE_SUMMARY;
 802
 803         check_signals();
 804         if (!gsi->count && !summary) {
 805                 ret = -E_LOOP_COMPLETE;
 806                 goto err;
 807         }
 808         if (summary || (gsi->count && (gsi->flags & GSF_PRINT_FILES))) {
 809                 uint64_t files;
 810                 ret = osl(osl_get_object(dir_table, row, DT_FILES, &obj));
 811                 if (ret < 0)
 812                         goto err;
 813                 files = *(uint64_t *)obj.data;
 814                 if (gsi->count && (gsi->flags & GSF_PRINT_FILES)) {
 815                         format_count_value(conf.count_unit_arg, files,
 816                                 conf.count_unit_arg == count_unit_arg_h,
 817                                 formated_value);
 818                         printf("\t%s%s", formated_value,
 819                                 (gsi->flags & (GSF_PRINT_BYTES | GSF_PRINT_DIRNAME))?
 820                                 "\t" : "\n");
 821                 }
 822                 if (summary)
 823                         num_files += files;
 824         }
 825         if (summary || (gsi->count && (gsi->flags & GSF_PRINT_BYTES))) {
 826                 uint64_t bytes;
 827                 ret = osl(osl_get_object(dir_table, row, DT_BYTES, &obj));
 828                 if (ret < 0)
 829                         goto err;
 830                 bytes = *(uint64_t *)obj.data;
 831                 if (gsi->count && (gsi->flags & GSF_PRINT_BYTES)) {
 832                         format_size_value(conf.size_unit_arg, bytes,
 833                                 conf.size_unit_arg == size_unit_arg_h,
 834                                 formated_value);
 835                         printf("%s%s%s",
 836                                 (gsi->flags & GSF_PRINT_FILES)? "" : "\t",
 837                                 formated_value,
 838                                 (gsi->flags & GSF_PRINT_DIRNAME)? "\t" : "\n"
 839                         );
 840                 }
 841                 if (summary) {
 842                         num_bytes += bytes;
 843                         num_dirs++;
 844                 }
 845         }
 846         if (gsi->count && (gsi->flags & GSF_PRINT_DIRNAME)) {
 847                 ret = get_dir_name_of_row(row, &dirname);
 848                 if (ret < 0)
 849                         goto err;
 850                 printf("%s%s\n",
 851                         (gsi->flags & (GSF_PRINT_BYTES | GSF_PRINT_FILES))? "" : "\t",
 852                         dirname);
 853                 free(dirname);
 854         }
 855         if (gsi->count > 0)
 856                 gsi->count--;
 857         return 1;
 858 err:
 859         gsi->ret = ret;
 860         gsi->osl_errno = (ret == -E_OSL)? osl_errno : 0;
 861         return -1;
 862 }
 863
 864 static void print_id_stats(void)
 865 {
 866         struct user_info *ui;
 867
 868         printf("User summary "
 869                 "(pw_name/uid/dirs%s/files%s/size%s):\n",
 870                 count_unit_buf, count_unit_buf, size_unit_buf);
 871         FOR_EACH_USER(ui) {
 872                 char formated_dir_count[FORMATED_VALUE_SIZE],
 873                         formated_file_count[FORMATED_VALUE_SIZE],
 874                         formated_bytes[FORMATED_VALUE_SIZE ];
 875                 if (!ui_used(ui) || !ui_admissible(ui))
 876                         continue;
 877                 format_count_value(conf.count_unit_arg, ui->dirs,
 878                         conf.count_unit_arg == count_unit_arg_h,
 879                         formated_dir_count);
 880                 format_count_value(conf.count_unit_arg, ui->files,
 881                         conf.count_unit_arg == count_unit_arg_h,
 882                         formated_file_count);
 883                 format_size_value(conf.size_unit_arg, ui->bytes,
 884                         conf.size_unit_arg == size_unit_arg_h,
 885                         formated_bytes);
 886                 printf("\t%s\t%u\t%s\t%s\t%s\n",
 887                         ui->pw_name? ui->pw_name : "?",
 888                         (unsigned)ui->uid,
 889                         formated_dir_count,
 890                         formated_file_count,
 891                         formated_bytes
 892                 );
 893         }
 894 }
 895
 896 enum user_stats_flags {
 897         USF_PRINT_DIRNAME = 1,
 898         USF_PRINT_BYTES = 2,
 899         USF_PRINT_FILES = 4,
 900         USF_COMPUTE_SUMMARY = 8,
 901 };
 902
 903 struct user_stats_info {
 904         uint32_t count;
 905         enum user_stats_flags flags;
 906         int ret;
 907         int osl_errno;
 908         struct user_info *ui;
 909 };
 910
 911 static int user_stats_loop_function(struct osl_row *row, void *data)
 912 {
 913         struct user_stats_info *usi = data;
 914         struct osl_object obj;
 915         int ret, summary = usi->flags & GSF_COMPUTE_SUMMARY;
 916         char formated_value[FORMATED_VALUE_SIZE];
 917
 918         check_signals();
 919         if (!usi->count && !summary) {
 920                 ret = -E_LOOP_COMPLETE;
 921                 goto err;
 922         }
 923         if (summary || (usi->count && (usi->flags & USF_PRINT_FILES))) {
 924                 uint64_t files;
 925                 ret = osl(osl_get_object(usi->ui->table, row, UT_FILES, &obj));
 926                 if (ret < 0)
 927                         goto err;
 928                 files = *(uint64_t *)obj.data;
 929                 if (usi->count && (usi->flags & USF_PRINT_FILES)) {
 930                         format_count_value(conf.count_unit_arg, files,
 931                                 conf.count_unit_arg == count_unit_arg_h,
 932                                 formated_value);
 933                         printf("\t%s%s", formated_value,
 934                                 (usi->flags & (USF_PRINT_BYTES | USF_PRINT_DIRNAME))?
 935                                         "\t" : "\n"
 936                         );
 937                 }
 938                 if (summary)
 939                         usi->ui->files += files;
 940         }
 941         if (summary || (usi->count && (usi->flags & USF_PRINT_BYTES))) {
 942                 uint64_t bytes;
 943                 ret = osl(osl_get_object(usi->ui->table, row, UT_BYTES, &obj));
 944                 if (ret < 0)
 945                         goto err;
 946                 bytes = *(uint64_t *)obj.data;
 947                 if (usi->count && (usi->flags & USF_PRINT_BYTES)) {
 948                         format_size_value(conf.size_unit_arg, bytes,
 949                                 conf.size_unit_arg == size_unit_arg_h,
 950                                 formated_value);
 951                         printf("%s%s%s",
 952                                 (usi->flags & USF_PRINT_FILES)? "" : "\t",
 953                                 formated_value,
 954                                 usi->flags & USF_PRINT_DIRNAME?  "\t" : "\n"
 955                         );
 956                 }
 957                 if (summary) {
 958                         usi->ui->bytes += bytes;
 959                         usi->ui->dirs++;
 960                 }
 961
 962         }
 963         if (usi->count && (usi->flags & USF_PRINT_DIRNAME)) {
 964                 char *dirname;
 965                 ret = osl(osl_get_object(usi->ui->table, row, UT_DIR_NUM, &obj));
 966                 if (ret < 0)
 967                         goto err;
 968                 ret = get_dir_name_by_number((uint64_t *)obj.data, &dirname);
 969                 if (ret < 0)
 970                         goto err;
 971                 printf("%s%s\n",
 972                         (usi->flags & (USF_PRINT_BYTES | USF_PRINT_FILES))? "" : "\t",
 973                         dirname);
 974                 free(dirname);
 975         }
 976         if (usi->count > 0)
 977                 usi->count--;
 978         return 1;
 979 err:
 980         usi->ret = ret;
 981         usi->osl_errno = (ret == -E_OSL)? osl_errno : 0;
 982         return -1;
 983 }
 984
 985 static int check_loop_return(int ret, int loop_ret, int loop_osl_errno)
 986 {
 987         if (ret >= 0)
 988                 return ret;
 989         assert(ret == -E_OSL);
 990         if (osl_errno != E_OSL_LOOP)
 991                 /* error not caused by loop function returning negative. */
 992                 return ret;
 993         assert(loop_ret < 0);
 994         if (loop_ret == -E_LOOP_COMPLETE) /* no error */
 995                 return 1;
 996         if (loop_ret == -E_OSL) { /* osl error in loop function */
 997                 assert(loop_osl_errno);
 998                 osl_errno = loop_osl_errno;
 999         }
1000         return loop_ret;
1001 }
1002
1003 static int adu_loop_reverse(struct osl_table *t, unsigned col_num, void *private_data,
1004                 osl_rbtree_loop_func *func, int *loop_ret, int *loop_osl_errno)
1005 {
1006         int ret = osl(osl_rbtree_loop_reverse(t, col_num, private_data, func));
1007         return check_loop_return(ret, *loop_ret, *loop_osl_errno);
1008 }
1009
1010 static int print_user_stats(void)
1011 {
1012         struct user_info *ui;
1013         int ret;
1014
1015         FOR_EACH_USER(ui) {
1016                 struct user_stats_info usi = {
1017                         .count = conf.limit_arg,
1018                         .ui = ui
1019                 };
1020                 if (!ui_used(ui) || !ui_admissible(ui))
1021                         continue;
1022                 usi.flags = USF_PRINT_DIRNAME | USF_PRINT_BYTES | USF_COMPUTE_SUMMARY;
1023                 printf("%s (uid %u), by size%s:\n",
1024                         ui->pw_name? ui->pw_name : "?", (unsigned)ui->uid,
1025                         size_unit_buf);
1026                 ret = adu_loop_reverse(ui->table, UT_BYTES, &usi, user_stats_loop_function,
1027                         &usi.ret, &usi.osl_errno);
1028                 if (ret < 0)
1029                         return ret;
1030                 printf("\n%s (uid %u), by file count%s:\n",
1031                         ui->pw_name? ui->pw_name : "?", (unsigned)ui->uid,
1032                         count_unit_buf);
1033                 usi.count = conf.limit_arg,
1034                 usi.flags = USF_PRINT_DIRNAME | USF_PRINT_FILES;
1035                 ret = adu_loop_reverse(ui->table, UT_FILES, &usi, user_stats_loop_function,
1036                         &usi.ret, &usi.osl_errno);
1037                 if (ret < 0)
1038                         return ret;
1039                 printf("\n");
1040         }
1041         return 1;
1042 }
1043
1044 static void print_global_summary(void)
1045 {
1046         char d[FORMATED_VALUE_SIZE], f[FORMATED_VALUE_SIZE],
1047                 s[FORMATED_VALUE_SIZE];
1048         enum enum_count_unit ud, uf;
1049         enum enum_size_unit us;
1050
1051         ud = format_count_value(conf.count_unit_arg, num_dirs, 0, d);
1052         uf = format_count_value(conf.count_unit_arg, num_files, 0, f);
1053         us = format_size_value(conf.size_unit_arg, num_bytes, 0, s);
1054
1055         printf("Global summary "
1056                 "(dirs(%c)/files(%c)/size(%c))\n"
1057                 "\t%s\t%s\t%s\n\n",
1058                 count_unit_abbrevs[ud],
1059                 count_unit_abbrevs[uf],
1060                 size_unit_abbrevs[us],
1061                 d, f, s
1062         );
1063
1064 }
1065
1066 static int print_statistics(void)
1067 {
1068         int ret;
1069         struct global_stats_info gsi = {
1070                 .count = conf.limit_arg,
1071                 .flags = GSF_PRINT_DIRNAME | GSF_PRINT_BYTES | GSF_COMPUTE_SUMMARY
1072         };
1073
1074         printf("By size%s:\n",
1075                 size_unit_buf);
1076         ret = adu_loop_reverse(dir_table, DT_BYTES, &gsi,
1077                 global_stats_loop_function, &gsi.ret, &gsi.osl_errno);
1078         if (ret < 0)
1079                 return ret;
1080         printf("\n");
1081
1082         gsi.count = conf.limit_arg;
1083         gsi.flags = GSF_PRINT_DIRNAME | GSF_PRINT_FILES;
1084         printf("By file count%s:\n",
1085                 count_unit_buf);
1086         ret = adu_loop_reverse(dir_table, DT_FILES, &gsi,
1087                 global_stats_loop_function, &gsi.ret, &gsi.osl_errno);
1088         if (ret < 0)
1089                 return ret;
1090         printf("\n");
1091         print_global_summary();
1092         print_user_stats();
1093         print_id_stats();
1094         return 1;
1095 }
1096
1097 static char *get_uid_list_name(void)
1098 {
1099         return make_message("%s/uid_list", conf.database_dir_arg);
1100 }
1101
1102 static int write_uid_list(void)
1103 {
1104         char *buf, *filename;
1105         uint32_t count = 0;
1106         struct user_info *ui;
1107         size_t size = num_uids * sizeof(uint32_t);
1108         int ret;
1109
1110         if (!num_uids)
1111                 return 0;
1112         buf = adu_malloc(size);
1113         FOR_EACH_USER(ui) {
1114                 if (!ui_used(ui) || !ui_admissible(ui))
1115                         continue;
1116                 DEBUG_LOG("saving uid %u\n", (unsigned) ui->uid);
1117                 write_u32(buf + count++ * sizeof(uint32_t), ui->uid);
1118         }
1119         filename = get_uid_list_name();
1120         ret = adu_write_file(filename, buf, size);
1121         free(filename);
1122         free(buf);
1123         return ret;
1124 }
1125
1126 static int open_dir_table(void)
1127 {
1128         if (!dir_table_desc.dir) /* we did not create the table */
1129                 dir_table_desc.dir = adu_strdup(conf.database_dir_arg);
1130         return osl(osl_open_table(&dir_table_desc, &dir_table));
1131 }
1132 static int com_create()
1133 {
1134         uint64_t zero = 0ULL;
1135         int ret;
1136         struct stat statbuf;
1137
1138         if (lstat(conf.base_dir_arg, &statbuf) == -1)
1139                 return -ERRNO_TO_ERROR(errno);
1140         if (!S_ISDIR(statbuf.st_mode))
1141                 return -ERRNO_TO_ERROR(ENOTDIR);
1142         device_id = statbuf.st_dev;
1143         ret = create_tables();
1144         if (ret < 0)
1145                 return ret;
1146         check_signals();
1147         ret = open_dir_table();
1148         if (ret < 0)
1149                 return ret;
1150         check_signals();
1151         ret = scan_dir(conf.base_dir_arg, &zero);
1152         if (ret < 0)
1153                 goto out;
1154         ret = write_uid_list();
1155 out:
1156         close_all_tables();
1157         return ret;
1158 }
1159
1160 static int read_uid_file(void)
1161 {
1162         size_t size;
1163         uint32_t n;
1164         char *filename = get_uid_list_name(), *map;
1165         int ret = mmap_full_file(filename, O_RDONLY, (void **)&map, &size, NULL);
1166
1167         if (ret < 0) {
1168                 INFO_LOG("failed to map %s\n", filename);
1169                 free(filename);
1170                 return ret;
1171         }
1172         num_uids = size / 4;
1173         INFO_LOG("found %u uids in %s\n", (unsigned)num_uids, filename);
1174         free(filename);
1175         /* hash table size should be a power of two and larger than the number of uids */
1176         uid_hash_table_size = 4;
1177         while (uid_hash_table_size < num_uids)
1178                 uid_hash_table_size *= 2;
1179         create_hash_table();
1180         for (n = 0; n < num_uids; n++) {
1181                 uint32_t uid = read_u32(map + n * sizeof(uid));
1182                 ret = search_uid(uid, OPEN_USER_TABLE, NULL);
1183                 if (ret < 0)
1184                         goto out;
1185         }
1186 out:
1187         adu_munmap(map, size);
1188         return ret;
1189 }
1190
1191 static int com_select(void)
1192 {
1193         int ret;
1194
1195         if (conf.count_unit_arg != count_unit_arg_h)
1196                 count_unit_buf[1] = count_unit_abbrevs[conf.count_unit_arg];
1197         else
1198                 count_unit_buf[0] = '\0';
1199         if (conf.size_unit_arg != size_unit_arg_h)
1200                 size_unit_buf[1] = size_unit_abbrevs[conf.size_unit_arg];
1201         else
1202                 size_unit_buf[0] = '\0';
1203
1204         ret = open_dir_table();
1205         if (ret < 0)
1206                 return ret;
1207         check_signals();
1208         ret = read_uid_file();
1209         if (ret < 0)
1210                 return ret;
1211         check_signals();
1212         ret = print_statistics();
1213         close_all_tables();
1214         return ret;
1215 }
1216
1217 static int check_args(void)
1218 {
1219         int i, ret;
1220
1221         /* remove trailing slashes from base-dir arg */
1222         if (conf.base_dir_given) {
1223                 size_t len = strlen(conf.base_dir_arg);
1224                 for (;;) {
1225                         if (!len) /* empty string */
1226                                 return -ERRNO_TO_ERROR(EINVAL);
1227                         if (!--len) /* length 1 is always OK */
1228                                 break;
1229                         if (conf.base_dir_arg[len] != '/')
1230                                 break; /* no trailing slash, also OK */
1231                         conf.base_dir_arg[len] = '\0';
1232                 }
1233         }
1234         if (!conf.uid_given)
1235                 return 0;
1236         admissible_uids = adu_malloc(conf.uid_given * sizeof(*admissible_uids));
1237         for (i = 0; i < conf.uid_given; i++) {
1238                 ret = parse_uid_range(conf.uid_arg[i], admissible_uids + i);
1239                 if (ret < 0)
1240                         goto err;
1241         }
1242         return 1;
1243 err:
1244         free(admissible_uids);
1245         admissible_uids = NULL;
1246         return ret;
1247 }
1248
1249 int main(int argc, char **argv)
1250 {
1251         int ret;
1252         struct cmdline_parser_params params = {
1253                 .override = 0,
1254                 .initialize = 1,
1255                 .check_required = 0,
1256                 .check_ambiguity = 0,
1257                 .print_errors = 1
1258         };
1259
1260         cmdline_parser_ext(argc, argv, &conf, &params); /* aborts on errors */
1261         ret = init_signals();
1262         if (ret < 0)
1263                 goto out;
1264         ret = check_args();
1265         if (ret < 0)
1266                 goto out;
1267         ret = -E_SYNTAX;
1268         if (conf.select_given)
1269                 ret = com_select();
1270         else
1271                 ret = com_create();
1272         if (ret < 0)
1273                 goto out;
1274 out:
1275         free(admissible_uids);
1276         if (ret < 0) {
1277                 ERROR_LOG("%s\n", adu_strerror(-ret));
1278                 return -EXIT_FAILURE;
1279         }
1280         return EXIT_SUCCESS;
1281 }