e2fsck/dirinfo.c

   1 /*
   2  * dirinfo.c --- maintains the directory information table for e2fsck.
   3  *
   4  * Copyright (C) 1993 Theodore Ts'o.  This file may be redistributed
   5  * under the terms of the GNU Public License.
   6  */
   7
   8 #undef DIRINFO_DEBUG
   9
  10 #include <assert.h>
  11 #include "config.h"
  12 #include "e2fsck.h"
  13 #include <sys/stat.h>
  14 #include <fcntl.h>
  15 #include "uuid/uuid.h"
  16
  17 #include "ext2fs/ext2fs.h"
  18 #include <ext2fs/tdb.h>
  19
  20 struct dir_info_db {
  21         ext2_ino_t      count;
  22         ext2_ino_t      size;
  23         struct dir_info *array;
  24         struct dir_info *last_lookup;
  25 #ifdef CONFIG_TDB
  26         char            *tdb_fn;
  27         TDB_CONTEXT     *tdb;
  28 #endif
  29 };
  30
  31 struct dir_info_iter {
  32         ext2_ino_t      i;
  33 #ifdef CONFIG_TDB
  34         TDB_DATA        tdb_iter;
  35 #endif
  36 };
  37
  38 struct dir_info_ent {
  39         ext2_ino_t              dotdot; /* Parent according to '..' */
  40         ext2_ino_t              parent; /* Parent according to treewalk */
  41 };
  42
  43
  44 static void e2fsck_put_dir_info(e2fsck_t ctx, struct dir_info *dir);
  45
  46 #ifdef CONFIG_TDB
  47 static void setup_tdb(e2fsck_t ctx, ext2_ino_t num_dirs)
  48 {
  49         struct dir_info_db      *db = ctx->dir_info;
  50         ext2_ino_t              threshold;
  51         errcode_t               retval;
  52         mode_t                  save_umask;
  53         char                    *tdb_dir, uuid[40];
  54         int                     fd, enable;
  55
  56         profile_get_string(ctx->profile, "scratch_files", "directory", 0, 0,
  57                            &tdb_dir);
  58         profile_get_uint(ctx->profile, "scratch_files",
  59                          "numdirs_threshold", 0, 0, &threshold);
  60         profile_get_boolean(ctx->profile, "scratch_files",
  61                             "dirinfo", 0, 1, &enable);
  62
  63         if (!enable || !tdb_dir || access(tdb_dir, W_OK) ||
  64             (threshold && num_dirs <= threshold))
  65                 return;
  66
  67         retval = ext2fs_get_mem(strlen(tdb_dir) + 64, &db->tdb_fn);
  68         if (retval)
  69                 return;
  70
  71         uuid_unparse(ctx->fs->super->s_uuid, uuid);
  72         sprintf(db->tdb_fn, "%s/%s-dirinfo-XXXXXX", tdb_dir, uuid);
  73         save_umask = umask(077);
  74         fd = mkstemp(db->tdb_fn);
  75         umask(save_umask);
  76         if (fd < 0) {
  77                 db->tdb = NULL;
  78                 return;
  79         }
  80
  81         if (num_dirs < 99991)
  82                 num_dirs = 99991; /* largest 5 digit prime */
  83
  84         db->tdb = tdb_open(db->tdb_fn, num_dirs, TDB_NOLOCK | TDB_NOSYNC,
  85                            O_RDWR | O_CREAT | O_TRUNC, 0600);
  86         close(fd);
  87 }
  88 #endif
  89
  90 static void setup_db(e2fsck_t ctx)
  91 {
  92         struct dir_info_db      *db;
  93         ext2_ino_t              num_dirs;
  94         errcode_t               retval;
  95
  96         db = (struct dir_info_db *)
  97                 e2fsck_allocate_memory(ctx, sizeof(struct dir_info_db),
  98                                        "directory map db");
  99         db->count = db->size = 0;
 100         db->array = 0;
 101
 102         ctx->dir_info = db;
 103
 104         retval = ext2fs_get_num_dirs(ctx->fs, &num_dirs);
 105         if (retval)
 106                 num_dirs = 1024;        /* Guess */
 107
 108 #ifdef CONFIG_TDB
 109         setup_tdb(ctx, num_dirs);
 110
 111         if (db->tdb) {
 112 #ifdef DIRINFO_DEBUG
 113                 printf("Note: using tdb!\n");
 114 #endif
 115                 return;
 116         }
 117 #endif
 118
 119         db->size = num_dirs + 10;
 120         db->array  = (struct dir_info *)
 121                 e2fsck_allocate_memory(ctx, db->size
 122                                        * sizeof (struct dir_info),
 123                                        "directory map");
 124 }
 125
 126 /*
 127  * Return the min index that has ino larger or equal to @ino
 128  * If not found, return -ENOENT
 129  */
 130 static int
 131 e2fsck_dir_info_min_larger_equal(struct dir_info_db *dir_info,
 132                                  ext2_ino_t ino, ext2_ino_t *index)
 133 {
 134         ext2_ino_t low = 0;
 135         ext2_ino_t mid, high;
 136         ext2_ino_t tmp_ino;
 137         int found = 0;
 138
 139         if (dir_info->count == 0)
 140                 return -ENOENT;
 141
 142         high = dir_info->count - 1;
 143         while (low <= high) {
 144                 /* sum may overflow, but result will fit into mid again */
 145                 mid = (unsigned long long)(low + high) / 2;
 146                 tmp_ino = dir_info->array[mid].ino;
 147                 if (ino == tmp_ino) {
 148                         *index = mid;
 149                         found = 1;
 150                         return 0;
 151                 } else if (ino < tmp_ino) {
 152                         /*
 153                          * The mid ino is larger than @ino, remember the index
 154                          * here so we won't miss this ino
 155                          */
 156                         *index = mid;
 157                         found = 1;
 158                         if (mid == 0)
 159                                 break;
 160                         high = mid - 1;
 161                 } else {
 162                         low = mid + 1;
 163                 }
 164         }
 165
 166         if (found)
 167                 return 0;
 168
 169         return -ENOENT;
 170 }
 171
 172 /*
 173  * Merge two sorted dir info to @dest
 174  */
 175 void e2fsck_merge_dir_info(e2fsck_t ctx, struct dir_info_db *src,
 176                            struct dir_info_db *dest)
 177 {
 178         size_t           size_dir_info = sizeof(struct dir_info);
 179         ext2_ino_t       size = dest->size;
 180         struct dir_info  *src_array = src->array;
 181         struct dir_info  *dest_array = dest->array;
 182         ext2_ino_t       src_count = src->count;
 183         ext2_ino_t       dest_count = dest->count;
 184         ext2_ino_t       total_count = src_count + dest_count;
 185         struct dir_info *tmp_array;
 186         struct dir_info *array_ptr;
 187         ext2_ino_t       src_index = 0;
 188         ext2_ino_t       dest_index = 0;
 189
 190         if (src->count == 0)
 191                 return;
 192
 193         if (size < total_count)
 194                 size = total_count;
 195
 196         if (size < src->size)
 197                 size = src->size;
 198
 199         tmp_array = e2fsck_allocate_memory(ctx, size * size_dir_info,
 200                                             "directory map");
 201         array_ptr = tmp_array;
 202         /*
 203          * This can be improved by binary search and memcpy, but codes
 204          * would be more complex. And if the groups distributed to each
 205          * thread are strided, this implementation won't be too bad
 206          * comparing to the optimiztion.
 207          */
 208         while (src_index < src_count || dest_index < dest_count) {
 209                 if (src_index >= src_count) {
 210                         memcpy(array_ptr, &dest_array[dest_index],
 211                                (dest_count - dest_index) * size_dir_info);
 212                         break;
 213                 }
 214                 if (dest_index >= dest_count) {
 215                         memcpy(array_ptr, &src_array[src_index],
 216                                (src_count - src_index) * size_dir_info);
 217                         break;
 218                 }
 219                 if (src_array[src_index].ino < dest_array[dest_index].ino) {
 220                         *array_ptr = src_array[src_index];
 221                         src_index++;
 222                 } else {
 223                         assert(src_array[src_index].ino >
 224                                dest_array[dest_index].ino);
 225                         *array_ptr = dest_array[dest_index];
 226                         dest_index++;
 227                 }
 228                 array_ptr++;
 229         }
 230
 231         if (dest->array)
 232                 ext2fs_free_mem(&dest->array);
 233         dest->array = tmp_array;
 234         dest->size = size;
 235         dest->count = total_count;
 236 }
 237
 238 /*
 239  *
 240  * Insert an inode into the sorted array. The array should have at least one
 241  * free slot.
 242  *
 243  * Normally, add_dir_info is called with each inode in
 244  * sequential order; but once in a while (like when pass 3
 245  * needs to recreate the root directory or lost+found
 246  * directory) it is called out of order.  In those cases, we
 247  * need to move the dir_info entries down to make room, since
 248  * the dir_info array needs to be sorted by inode number for
 249  * get_dir_info()'s sake.
 250  */
 251 static void e2fsck_insert_dir_info(struct dir_info_db *dir_info, ext2_ino_t ino, ext2_ino_t parent)
 252 {
 253         ext2_ino_t              index;
 254         struct dir_info         *dir;
 255         size_t                  dir_size = sizeof(*dir);
 256         struct dir_info         *array = dir_info->array;
 257         ext2_ino_t              array_count = dir_info->count;
 258         int                     err;
 259
 260         /*
 261          * Removing this check won't break anything. But since seqential ino
 262          * inserting happens a lot, this check avoids binary search.
 263          */
 264         if (array_count == 0 || array[array_count - 1].ino < ino) {
 265                 dir = &array[array_count];
 266                 dir_info->count++;
 267                 goto out;
 268         }
 269
 270         err = e2fsck_dir_info_min_larger_equal(dir_info, ino, &index);
 271         if (err >= 0 && array[index].ino == ino) {
 272                 dir = &array[index];
 273                 goto out;
 274         }
 275         if (err < 0) {
 276                 dir = &array[array_count];
 277                 dir_info->count++;
 278                 goto out;
 279         }
 280
 281         dir = &array[index];
 282         memmove((char *)dir + dir_size, dir, dir_size * (array_count - index));
 283         dir_info->count++;
 284 out:
 285         dir->ino = ino;
 286         dir->dotdot = parent;
 287         dir->parent = parent;
 288 }
 289
 290 /*
 291  * This subroutine is called during pass1 to create a directory info
 292  * entry.  During pass1, the passed-in parent is 0; it will get filled
 293  * in during pass2.
 294  */
 295 void e2fsck_add_dir_info(e2fsck_t ctx, ext2_ino_t ino, ext2_ino_t parent)
 296 {
 297         struct dir_info         *old_array;
 298         errcode_t               retval;
 299         unsigned long           old_size;
 300
 301 #ifdef DIRINFO_DEBUG
 302         printf("add_dir_info for inode (%u, %u)...\n", ino, parent);
 303 #endif
 304         if (!ctx->dir_info)
 305                 setup_db(ctx);
 306
 307         if (ctx->dir_info->count >= ctx->dir_info->size) {
 308                 old_size = ctx->dir_info->size * sizeof(struct dir_info);
 309                 ctx->dir_info->size += 10;
 310                 old_array = ctx->dir_info->array;
 311                 retval = ext2fs_resize_mem(old_size, ctx->dir_info->size *
 312                                            sizeof(struct dir_info),
 313                                            &ctx->dir_info->array);
 314                 if (retval) {
 315                         fprintf(stderr, "Couldn't reallocate dir_info "
 316                                 "structure to %u entries\n",
 317                                 ctx->dir_info->size);
 318                         fatal_error(ctx, 0);
 319                         ctx->dir_info->size -= 10;
 320                         return;
 321                 }
 322                 if (old_array != ctx->dir_info->array)
 323                         ctx->dir_info->last_lookup = NULL;
 324         }
 325
 326 #ifdef CONFIG_TDB
 327         if (ctx->dir_info->tdb) {
 328                 struct dir_info ent;
 329
 330                 ent.ino = ino;
 331                 ent.parent = parent;
 332                 ent.dotdot = parent;
 333                 e2fsck_put_dir_info(ctx, &ent);
 334                 return;
 335         }
 336 #endif
 337
 338         e2fsck_insert_dir_info(ctx->dir_info, ino, parent);
 339 }
 340
 341 /*
 342  * get_dir_info() --- given an inode number, try to find the directory
 343  * information entry for it.
 344  */
 345 static struct dir_info *e2fsck_get_dir_info(e2fsck_t ctx, ext2_ino_t ino)
 346 {
 347         struct dir_info_db      *db = ctx->dir_info;
 348         ext2_ino_t              index;
 349         int                     err;
 350
 351         if (!db)
 352                 return 0;
 353
 354 #ifdef DIRINFO_DEBUG
 355         printf("e2fsck_get_dir_info %u...", ino);
 356 #endif
 357
 358 #ifdef CONFIG_TDB
 359         if (db->tdb) {
 360                 static struct dir_info  ret_dir_info;
 361                 TDB_DATA key, data;
 362                 struct dir_info_ent     *buf;
 363
 364                 key.dptr = (unsigned char *) &ino;
 365                 key.dsize = sizeof(ext2_ino_t);
 366
 367                 data = tdb_fetch(db->tdb, key);
 368                 if (!data.dptr) {
 369                         if (tdb_error(db->tdb) != TDB_ERR_NOEXIST)
 370                                 printf("fetch failed: %s\n",
 371                                        tdb_errorstr(db->tdb));
 372                         return 0;
 373                 }
 374
 375                 buf = (struct dir_info_ent *) data.dptr;
 376                 ret_dir_info.ino = ino;
 377                 ret_dir_info.dotdot = buf->dotdot;
 378                 ret_dir_info.parent = buf->parent;
 379 #ifdef DIRINFO_DEBUG
 380                 printf("(%u,%u,%u)\n", ino, buf->dotdot, buf->parent);
 381 #endif
 382                 free(data.dptr);
 383                 return &ret_dir_info;
 384         }
 385 #endif
 386
 387         if (db->last_lookup && db->last_lookup->ino == ino)
 388                 return db->last_lookup;
 389
 390         err = e2fsck_dir_info_min_larger_equal(ctx->dir_info, ino, &index);
 391         if (err < 0)
 392                 return NULL;
 393         assert(ino <= ctx->dir_info->array[index].ino);
 394         if (ino == ctx->dir_info->array[index].ino) {
 395 #ifdef DIRINFO_DEBUG
 396                 printf("(%d,%d,%d)\n", ino,
 397                        ctx->dir_info->array[index].dotdot,
 398                        ctx->dir_info->array[index].parent);
 399 #endif
 400                 return &ctx->dir_info->array[index];
 401         }
 402         return NULL;
 403 }
 404
 405 static void e2fsck_put_dir_info(e2fsck_t ctx EXT2FS_NO_TDB_UNUSED,
 406                                 struct dir_info *dir EXT2FS_NO_TDB_UNUSED)
 407 {
 408 #ifdef CONFIG_TDB
 409         struct dir_info_db      *db = ctx->dir_info;
 410         struct dir_info_ent     buf;
 411         TDB_DATA                key, data;
 412 #endif
 413
 414 #ifdef DIRINFO_DEBUG
 415         printf("e2fsck_put_dir_info (%u, %u, %u)...", dir->ino, dir->dotdot,
 416                dir->parent);
 417 #endif
 418
 419 #ifdef CONFIG_TDB
 420         if (!db->tdb)
 421                 return;
 422
 423         buf.parent = dir->parent;
 424         buf.dotdot = dir->dotdot;
 425
 426         key.dptr = (unsigned char *) &dir->ino;
 427         key.dsize = sizeof(ext2_ino_t);
 428         data.dptr = (unsigned char *) &buf;
 429         data.dsize = sizeof(buf);
 430
 431         if (tdb_store(db->tdb, key, data, TDB_REPLACE) == -1) {
 432                 printf("store failed: %s\n", tdb_errorstr(db->tdb));
 433         }
 434 #endif
 435 }
 436
 437 /*
 438  * Free the dir_info structure when it isn't needed any more.
 439  */
 440 void e2fsck_free_dir_info(e2fsck_t ctx)
 441 {
 442         if (ctx->dir_info) {
 443 #ifdef CONFIG_TDB
 444                 if (ctx->dir_info->tdb)
 445                         tdb_close(ctx->dir_info->tdb);
 446                 if (ctx->dir_info->tdb_fn) {
 447                         if (unlink(ctx->dir_info->tdb_fn) < 0)
 448                                 com_err("e2fsck_free_dir_info", errno,
 449                                         _("while freeing dir_info tdb file"));
 450                         ext2fs_free_mem(&ctx->dir_info->tdb_fn);
 451                 }
 452 #endif
 453                 if (ctx->dir_info->array)
 454                         ext2fs_free_mem(&ctx->dir_info->array);
 455                 ctx->dir_info->array = 0;
 456                 ctx->dir_info->size = 0;
 457                 ctx->dir_info->count = 0;
 458                 ext2fs_free_mem(&ctx->dir_info);
 459                 ctx->dir_info = 0;
 460         }
 461 }
 462
 463 /*
 464  * Return the count of number of directories in the dir_info structure
 465  */
 466 int e2fsck_get_num_dirinfo(e2fsck_t ctx)
 467 {
 468         return ctx->dir_info ? ctx->dir_info->count : 0;
 469 }
 470
 471 struct dir_info_iter *e2fsck_dir_info_iter_begin(e2fsck_t ctx)
 472 {
 473         struct dir_info_iter *iter;
 474
 475         iter = e2fsck_allocate_memory(ctx, sizeof(struct dir_info_iter),
 476                                       "dir_info iterator");
 477
 478 #ifdef CONFIG_TDB
 479         if (ctx->dir_info->tdb)
 480                 iter->tdb_iter = tdb_firstkey(ctx->dir_info->tdb);
 481 #endif
 482
 483         return iter;
 484 }
 485
 486 void e2fsck_dir_info_iter_end(e2fsck_t ctx EXT2FS_ATTR((unused)),
 487                               struct dir_info_iter *iter)
 488 {
 489 #ifdef CONFIG_TDB
 490         free(iter->tdb_iter.dptr);
 491 #endif
 492         ext2fs_free_mem(&iter);
 493 }
 494
 495 /*
 496  * A simple interator function
 497  */
 498 struct dir_info *e2fsck_dir_info_iter(e2fsck_t ctx, struct dir_info_iter *iter)
 499 {
 500         if (!ctx->dir_info || !iter)
 501                 return 0;
 502
 503 #ifdef CONFIG_TDB
 504         if (ctx->dir_info->tdb) {
 505                 static struct dir_info ret_dir_info;
 506                 struct dir_info_ent *buf;
 507                 TDB_DATA data, key;
 508
 509                 if (iter->tdb_iter.dptr == 0)
 510                         return 0;
 511                 key = iter->tdb_iter;
 512                 data = tdb_fetch(ctx->dir_info->tdb, key);
 513                 if (!data.dptr) {
 514                         printf("iter fetch failed: %s\n",
 515                                tdb_errorstr(ctx->dir_info->tdb));
 516                         return 0;
 517                 }
 518                 buf = (struct dir_info_ent *) data.dptr;
 519                 ret_dir_info.ino = *((ext2_ino_t *) iter->tdb_iter.dptr);
 520                 ret_dir_info.dotdot = buf->dotdot;
 521                 ret_dir_info.parent = buf->parent;
 522                 iter->tdb_iter = tdb_nextkey(ctx->dir_info->tdb, key);
 523                 free(key.dptr);
 524                 free(data.dptr);
 525                 return &ret_dir_info;
 526         }
 527 #endif
 528
 529         if (iter->i >= ctx->dir_info->count)
 530                 return 0;
 531
 532 #ifdef DIRINFO_DEBUG
 533         printf("iter(%u, %u, %u)...", ctx->dir_info->array[iter->i].ino,
 534                ctx->dir_info->array[iter->i].dotdot,
 535                ctx->dir_info->array[iter->i].parent);
 536 #endif
 537         ctx->dir_info->last_lookup = ctx->dir_info->array + iter->i++;
 538         return(ctx->dir_info->last_lookup);
 539 }
 540
 541 /*
 542  * This function only sets the parent pointer, and requires that
 543  * dirinfo structure has already been created.
 544  */
 545 int e2fsck_dir_info_set_parent(e2fsck_t ctx, ext2_ino_t ino,
 546                                ext2_ino_t parent)
 547 {
 548         struct dir_info *p;
 549
 550         p = e2fsck_get_dir_info(ctx, ino);
 551         if (!p)
 552                 return 1;
 553         p->parent = parent;
 554         e2fsck_put_dir_info(ctx, p);
 555         return 0;
 556 }
 557
 558 /*
 559  * This function only sets the dot dot pointer, and requires that
 560  * dirinfo structure has already been created.
 561  */
 562 int e2fsck_dir_info_set_dotdot(e2fsck_t ctx, ext2_ino_t ino,
 563                                ext2_ino_t dotdot)
 564 {
 565         struct dir_info *p;
 566
 567         p = e2fsck_get_dir_info(ctx, ino);
 568         if (!p)
 569                 return 1;
 570         p->dotdot = dotdot;
 571         e2fsck_put_dir_info(ctx, p);
 572         return 0;
 573 }
 574
 575 /*
 576  * This function only sets the parent pointer, and requires that
 577  * dirinfo structure has already been created.
 578  */
 579 int e2fsck_dir_info_get_parent(e2fsck_t ctx, ext2_ino_t ino,
 580                                ext2_ino_t *parent)
 581 {
 582         struct dir_info *p;
 583
 584         p = e2fsck_get_dir_info(ctx, ino);
 585         if (!p)
 586                 return 1;
 587         *parent = p->parent;
 588         return 0;
 589 }
 590
 591 /*
 592  * This function only sets the dot dot pointer, and requires that
 593  * dirinfo structure has already been created.
 594  */
 595 int e2fsck_dir_info_get_dotdot(e2fsck_t ctx, ext2_ino_t ino,
 596                                ext2_ino_t *dotdot)
 597 {
 598         struct dir_info *p;
 599
 600         p = e2fsck_get_dir_info(ctx, ino);
 601         if (!p)
 602                 return 1;
 603         *dotdot = p->dotdot;
 604         return 0;
 605 }
 606