1 // SPDX-License-Identifier: GPL-2.0-only 2 /* 3 * This file is part of UBIFS. 4 * 5 * Copyright (C) 2006-2008 Nokia Corporation. 6 * 7 * Author: Adrian Hunter 8 */ 9 10 #include "ubifs.h" 11 12 /* 13 * An orphan is an inode number whose inode node has been committed to the index 14 * with a link count of zero. That happens when an open file is deleted 15 * (unlinked) and then a commit is run. In the normal course of events the inode 16 * would be deleted when the file is closed. However in the case of an unclean 17 * unmount, orphans need to be accounted for. After an unclean unmount, the 18 * orphans' inodes must be deleted which means either scanning the entire index 19 * looking for them, or keeping a list on flash somewhere. This unit implements 20 * the latter approach. 21 * 22 * The orphan area is a fixed number of LEBs situated between the LPT area and 23 * the main area. The number of orphan area LEBs is specified when the file 24 * system is created. The minimum number is 1. The size of the orphan area 25 * should be so that it can hold the maximum number of orphans that are expected 26 * to ever exist at one time. 27 * 28 * The number of orphans that can fit in a LEB is: 29 * 30 * (c->leb_size - UBIFS_ORPH_NODE_SZ) / sizeof(__le64) 31 * 32 * For example: a 15872 byte LEB can fit 1980 orphans so 1 LEB may be enough. 33 * 34 * Orphans are accumulated in a rb-tree. When an inode's link count drops to 35 * zero, the inode number is added to the rb-tree. It is removed from the tree 36 * when the inode is deleted. Any new orphans that are in the orphan tree when 37 * the commit is run, are written to the orphan area in 1 or more orphan nodes. 38 * If the orphan area is full, it is consolidated to make space. There is 39 * always enough space because validation prevents the user from creating more 40 * than the maximum number of orphans allowed. 41 */ 42 43 static int dbg_check_orphans(struct ubifs_info *c); 44 45 static struct ubifs_orphan *orphan_add(struct ubifs_info *c, ino_t inum, 46 struct ubifs_orphan *parent_orphan) 47 { 48 struct ubifs_orphan *orphan, *o; 49 struct rb_node **p, *parent = NULL; 50 51 orphan = kzalloc(sizeof(struct ubifs_orphan), GFP_NOFS); 52 if (!orphan) 53 return ERR_PTR(-ENOMEM); 54 orphan->inum = inum; 55 orphan->new = 1; 56 INIT_LIST_HEAD(&orphan->child_list); 57 58 spin_lock(&c->orphan_lock); 59 if (c->tot_orphans >= c->max_orphans) { 60 spin_unlock(&c->orphan_lock); 61 kfree(orphan); 62 return ERR_PTR(-ENFILE); 63 } 64 p = &c->orph_tree.rb_node; 65 while (*p) { 66 parent = *p; 67 o = rb_entry(parent, struct ubifs_orphan, rb); 68 if (inum < o->inum) 69 p = &(*p)->rb_left; 70 else if (inum > o->inum) 71 p = &(*p)->rb_right; 72 else { 73 ubifs_err(c, "orphaned twice"); 74 spin_unlock(&c->orphan_lock); 75 kfree(orphan); 76 return ERR_PTR(-EINVAL); 77 } 78 } 79 c->tot_orphans += 1; 80 c->new_orphans += 1; 81 rb_link_node(&orphan->rb, parent, p); 82 rb_insert_color(&orphan->rb, &c->orph_tree); 83 list_add_tail(&orphan->list, &c->orph_list); 84 list_add_tail(&orphan->new_list, &c->orph_new); 85 86 if (parent_orphan) { 87 list_add_tail(&orphan->child_list, 88 &parent_orphan->child_list); 89 } 90 91 spin_unlock(&c->orphan_lock); 92 dbg_gen("ino %lu", (unsigned long)inum); 93 return orphan; 94 } 95 96 static struct ubifs_orphan *lookup_orphan(struct ubifs_info *c, ino_t inum) 97 { 98 struct ubifs_orphan *o; 99 struct rb_node *p; 100 101 p = c->orph_tree.rb_node; 102 while (p) { 103 o = rb_entry(p, struct ubifs_orphan, rb); 104 if (inum < o->inum) 105 p = p->rb_left; 106 else if (inum > o->inum) 107 p = p->rb_right; 108 else { 109 return o; 110 } 111 } 112 return NULL; 113 } 114 115 static void __orphan_drop(struct ubifs_info *c, struct ubifs_orphan *o) 116 { 117 rb_erase(&o->rb, &c->orph_tree); 118 list_del(&o->list); 119 c->tot_orphans -= 1; 120 121 if (o->new) { 122 list_del(&o->new_list); 123 c->new_orphans -= 1; 124 } 125 126 kfree(o); 127 } 128 129 static void orphan_delete(struct ubifs_info *c, ino_t inum) 130 { 131 struct ubifs_orphan *orph, *child_orph, *tmp_o; 132 133 spin_lock(&c->orphan_lock); 134 135 orph = lookup_orphan(c, inum); 136 if (!orph) { 137 spin_unlock(&c->orphan_lock); 138 ubifs_err(c, "missing orphan ino %lu", (unsigned long)inum); 139 dump_stack(); 140 141 return; 142 } 143 144 if (orph->del) { 145 spin_unlock(&c->orphan_lock); 146 dbg_gen("deleted twice ino %lu", 147 (unsigned long)inum); 148 return; 149 } 150 151 if (orph->cmt) { 152 orph->del = 1; 153 orph->dnext = c->orph_dnext; 154 c->orph_dnext = orph; 155 spin_unlock(&c->orphan_lock); 156 dbg_gen("delete later ino %lu", 157 (unsigned long)inum); 158 return; 159 } 160 161 list_for_each_entry_safe(child_orph, tmp_o, &orph->child_list, child_list) { 162 list_del(&child_orph->child_list); 163 __orphan_drop(c, child_orph); 164 } 165 166 __orphan_drop(c, orph); 167 168 spin_unlock(&c->orphan_lock); 169 } 170 171 /** 172 * ubifs_add_orphan - add an orphan. 173 * @c: UBIFS file-system description object 174 * @inum: orphan inode number 175 * 176 * Add an orphan. This function is called when an inodes link count drops to 177 * zero. 178 */ 179 int ubifs_add_orphan(struct ubifs_info *c, ino_t inum) 180 { 181 int err = 0; 182 ino_t xattr_inum; 183 union ubifs_key key; 184 struct ubifs_dent_node *xent; 185 struct fscrypt_name nm = {0}; 186 struct ubifs_orphan *xattr_orphan; 187 struct ubifs_orphan *orphan; 188 189 orphan = orphan_add(c, inum, NULL); 190 if (IS_ERR(orphan)) 191 return PTR_ERR(orphan); 192 193 lowest_xent_key(c, &key, inum); 194 while (1) { 195 xent = ubifs_tnc_next_ent(c, &key, &nm); 196 if (IS_ERR(xent)) { 197 err = PTR_ERR(xent); 198 if (err == -ENOENT) 199 break; 200 return err; 201 } 202 203 fname_name(&nm) = xent->name; 204 fname_len(&nm) = le16_to_cpu(xent->nlen); 205 xattr_inum = le64_to_cpu(xent->inum); 206 207 xattr_orphan = orphan_add(c, xattr_inum, orphan); 208 if (IS_ERR(xattr_orphan)) 209 return PTR_ERR(xattr_orphan); 210 211 key_read(c, &xent->key, &key); 212 } 213 214 return 0; 215 } 216 217 /** 218 * ubifs_delete_orphan - delete an orphan. 219 * @c: UBIFS file-system description object 220 * @inum: orphan inode number 221 * 222 * Delete an orphan. This function is called when an inode is deleted. 223 */ 224 void ubifs_delete_orphan(struct ubifs_info *c, ino_t inum) 225 { 226 orphan_delete(c, inum); 227 } 228 229 /** 230 * ubifs_orphan_start_commit - start commit of orphans. 231 * @c: UBIFS file-system description object 232 * 233 * Start commit of orphans. 234 */ 235 int ubifs_orphan_start_commit(struct ubifs_info *c) 236 { 237 struct ubifs_orphan *orphan, **last; 238 239 spin_lock(&c->orphan_lock); 240 last = &c->orph_cnext; 241 list_for_each_entry(orphan, &c->orph_new, new_list) { 242 ubifs_assert(c, orphan->new); 243 ubifs_assert(c, !orphan->cmt); 244 orphan->new = 0; 245 orphan->cmt = 1; 246 *last = orphan; 247 last = &orphan->cnext; 248 } 249 *last = NULL; 250 c->cmt_orphans = c->new_orphans; 251 c->new_orphans = 0; 252 dbg_cmt("%d orphans to commit", c->cmt_orphans); 253 INIT_LIST_HEAD(&c->orph_new); 254 if (c->tot_orphans == 0) 255 c->no_orphs = 1; 256 else 257 c->no_orphs = 0; 258 spin_unlock(&c->orphan_lock); 259 return 0; 260 } 261 262 /** 263 * avail_orphs - calculate available space. 264 * @c: UBIFS file-system description object 265 * 266 * This function returns the number of orphans that can be written in the 267 * available space. 268 */ 269 static int avail_orphs(struct ubifs_info *c) 270 { 271 int avail_lebs, avail, gap; 272 273 avail_lebs = c->orph_lebs - (c->ohead_lnum - c->orph_first) - 1; 274 avail = avail_lebs * 275 ((c->leb_size - UBIFS_ORPH_NODE_SZ) / sizeof(__le64)); 276 gap = c->leb_size - c->ohead_offs; 277 if (gap >= UBIFS_ORPH_NODE_SZ + sizeof(__le64)) 278 avail += (gap - UBIFS_ORPH_NODE_SZ) / sizeof(__le64); 279 return avail; 280 } 281 282 /** 283 * tot_avail_orphs - calculate total space. 284 * @c: UBIFS file-system description object 285 * 286 * This function returns the number of orphans that can be written in half 287 * the total space. That leaves half the space for adding new orphans. 288 */ 289 static int tot_avail_orphs(struct ubifs_info *c) 290 { 291 int avail_lebs, avail; 292 293 avail_lebs = c->orph_lebs; 294 avail = avail_lebs * 295 ((c->leb_size - UBIFS_ORPH_NODE_SZ) / sizeof(__le64)); 296 return avail / 2; 297 } 298 299 /** 300 * do_write_orph_node - write a node to the orphan head. 301 * @c: UBIFS file-system description object 302 * @len: length of node 303 * @atomic: write atomically 304 * 305 * This function writes a node to the orphan head from the orphan buffer. If 306 * %atomic is not zero, then the write is done atomically. On success, %0 is 307 * returned, otherwise a negative error code is returned. 308 */ 309 static int do_write_orph_node(struct ubifs_info *c, int len, int atomic) 310 { 311 int err = 0; 312 313 if (atomic) { 314 ubifs_assert(c, c->ohead_offs == 0); 315 ubifs_prepare_node(c, c->orph_buf, len, 1); 316 len = ALIGN(len, c->min_io_size); 317 err = ubifs_leb_change(c, c->ohead_lnum, c->orph_buf, len); 318 } else { 319 if (c->ohead_offs == 0) { 320 /* Ensure LEB has been unmapped */ 321 err = ubifs_leb_unmap(c, c->ohead_lnum); 322 if (err) 323 return err; 324 } 325 err = ubifs_write_node(c, c->orph_buf, len, c->ohead_lnum, 326 c->ohead_offs); 327 } 328 return err; 329 } 330 331 /** 332 * write_orph_node - write an orphan node. 333 * @c: UBIFS file-system description object 334 * @atomic: write atomically 335 * 336 * This function builds an orphan node from the cnext list and writes it to the 337 * orphan head. On success, %0 is returned, otherwise a negative error code 338 * is returned. 339 */ 340 static int write_orph_node(struct ubifs_info *c, int atomic) 341 { 342 struct ubifs_orphan *orphan, *cnext; 343 struct ubifs_orph_node *orph; 344 int gap, err, len, cnt, i; 345 346 ubifs_assert(c, c->cmt_orphans > 0); 347 gap = c->leb_size - c->ohead_offs; 348 if (gap < UBIFS_ORPH_NODE_SZ + sizeof(__le64)) { 349 c->ohead_lnum += 1; 350 c->ohead_offs = 0; 351 gap = c->leb_size; 352 if (c->ohead_lnum > c->orph_last) { 353 /* 354 * We limit the number of orphans so that this should 355 * never happen. 356 */ 357 ubifs_err(c, "out of space in orphan area"); 358 return -EINVAL; 359 } 360 } 361 cnt = (gap - UBIFS_ORPH_NODE_SZ) / sizeof(__le64); 362 if (cnt > c->cmt_orphans) 363 cnt = c->cmt_orphans; 364 len = UBIFS_ORPH_NODE_SZ + cnt * sizeof(__le64); 365 ubifs_assert(c, c->orph_buf); 366 orph = c->orph_buf; 367 orph->ch.node_type = UBIFS_ORPH_NODE; 368 spin_lock(&c->orphan_lock); 369 cnext = c->orph_cnext; 370 for (i = 0; i < cnt; i++) { 371 orphan = cnext; 372 ubifs_assert(c, orphan->cmt); 373 orph->inos[i] = cpu_to_le64(orphan->inum); 374 orphan->cmt = 0; 375 cnext = orphan->cnext; 376 orphan->cnext = NULL; 377 } 378 c->orph_cnext = cnext; 379 c->cmt_orphans -= cnt; 380 spin_unlock(&c->orphan_lock); 381 if (c->cmt_orphans) 382 orph->cmt_no = cpu_to_le64(c->cmt_no); 383 else 384 /* Mark the last node of the commit */ 385 orph->cmt_no = cpu_to_le64((c->cmt_no) | (1ULL << 63)); 386 ubifs_assert(c, c->ohead_offs + len <= c->leb_size); 387 ubifs_assert(c, c->ohead_lnum >= c->orph_first); 388 ubifs_assert(c, c->ohead_lnum <= c->orph_last); 389 err = do_write_orph_node(c, len, atomic); 390 c->ohead_offs += ALIGN(len, c->min_io_size); 391 c->ohead_offs = ALIGN(c->ohead_offs, 8); 392 return err; 393 } 394 395 /** 396 * write_orph_nodes - write orphan nodes until there are no more to commit. 397 * @c: UBIFS file-system description object 398 * @atomic: write atomically 399 * 400 * This function writes orphan nodes for all the orphans to commit. On success, 401 * %0 is returned, otherwise a negative error code is returned. 402 */ 403 static int write_orph_nodes(struct ubifs_info *c, int atomic) 404 { 405 int err; 406 407 while (c->cmt_orphans > 0) { 408 err = write_orph_node(c, atomic); 409 if (err) 410 return err; 411 } 412 if (atomic) { 413 int lnum; 414 415 /* Unmap any unused LEBs after consolidation */ 416 for (lnum = c->ohead_lnum + 1; lnum <= c->orph_last; lnum++) { 417 err = ubifs_leb_unmap(c, lnum); 418 if (err) 419 return err; 420 } 421 } 422 return 0; 423 } 424 425 /** 426 * consolidate - consolidate the orphan area. 427 * @c: UBIFS file-system description object 428 * 429 * This function enables consolidation by putting all the orphans into the list 430 * to commit. The list is in the order that the orphans were added, and the 431 * LEBs are written atomically in order, so at no time can orphans be lost by 432 * an unclean unmount. 433 * 434 * This function returns %0 on success and a negative error code on failure. 435 */ 436 static int consolidate(struct ubifs_info *c) 437 { 438 int tot_avail = tot_avail_orphs(c), err = 0; 439 440 spin_lock(&c->orphan_lock); 441 dbg_cmt("there is space for %d orphans and there are %d", 442 tot_avail, c->tot_orphans); 443 if (c->tot_orphans - c->new_orphans <= tot_avail) { 444 struct ubifs_orphan *orphan, **last; 445 int cnt = 0; 446 447 /* Change the cnext list to include all non-new orphans */ 448 last = &c->orph_cnext; 449 list_for_each_entry(orphan, &c->orph_list, list) { 450 if (orphan->new) 451 continue; 452 orphan->cmt = 1; 453 *last = orphan; 454 last = &orphan->cnext; 455 cnt += 1; 456 } 457 *last = NULL; 458 ubifs_assert(c, cnt == c->tot_orphans - c->new_orphans); 459 c->cmt_orphans = cnt; 460 c->ohead_lnum = c->orph_first; 461 c->ohead_offs = 0; 462 } else { 463 /* 464 * We limit the number of orphans so that this should 465 * never happen. 466 */ 467 ubifs_err(c, "out of space in orphan area"); 468 err = -EINVAL; 469 } 470 spin_unlock(&c->orphan_lock); 471 return err; 472 } 473 474 /** 475 * commit_orphans - commit orphans. 476 * @c: UBIFS file-system description object 477 * 478 * This function commits orphans to flash. On success, %0 is returned, 479 * otherwise a negative error code is returned. 480 */ 481 static int commit_orphans(struct ubifs_info *c) 482 { 483 int avail, atomic = 0, err; 484 485 ubifs_assert(c, c->cmt_orphans > 0); 486 avail = avail_orphs(c); 487 if (avail < c->cmt_orphans) { 488 /* Not enough space to write new orphans, so consolidate */ 489 err = consolidate(c); 490 if (err) 491 return err; 492 atomic = 1; 493 } 494 err = write_orph_nodes(c, atomic); 495 return err; 496 } 497 498 /** 499 * erase_deleted - erase the orphans marked for deletion. 500 * @c: UBIFS file-system description object 501 * 502 * During commit, the orphans being committed cannot be deleted, so they are 503 * marked for deletion and deleted by this function. Also, the recovery 504 * adds killed orphans to the deletion list, and therefore they are deleted 505 * here too. 506 */ 507 static void erase_deleted(struct ubifs_info *c) 508 { 509 struct ubifs_orphan *orphan, *dnext; 510 511 spin_lock(&c->orphan_lock); 512 dnext = c->orph_dnext; 513 while (dnext) { 514 orphan = dnext; 515 dnext = orphan->dnext; 516 ubifs_assert(c, !orphan->new); 517 ubifs_assert(c, orphan->del); 518 rb_erase(&orphan->rb, &c->orph_tree); 519 list_del(&orphan->list); 520 c->tot_orphans -= 1; 521 dbg_gen("deleting orphan ino %lu", (unsigned long)orphan->inum); 522 kfree(orphan); 523 } 524 c->orph_dnext = NULL; 525 spin_unlock(&c->orphan_lock); 526 } 527 528 /** 529 * ubifs_orphan_end_commit - end commit of orphans. 530 * @c: UBIFS file-system description object 531 * 532 * End commit of orphans. 533 */ 534 int ubifs_orphan_end_commit(struct ubifs_info *c) 535 { 536 int err; 537 538 if (c->cmt_orphans != 0) { 539 err = commit_orphans(c); 540 if (err) 541 return err; 542 } 543 erase_deleted(c); 544 err = dbg_check_orphans(c); 545 return err; 546 } 547 548 /** 549 * ubifs_clear_orphans - erase all LEBs used for orphans. 550 * @c: UBIFS file-system description object 551 * 552 * If recovery is not required, then the orphans from the previous session 553 * are not needed. This function locates the LEBs used to record 554 * orphans, and un-maps them. 555 */ 556 int ubifs_clear_orphans(struct ubifs_info *c) 557 { 558 int lnum, err; 559 560 for (lnum = c->orph_first; lnum <= c->orph_last; lnum++) { 561 err = ubifs_leb_unmap(c, lnum); 562 if (err) 563 return err; 564 } 565 c->ohead_lnum = c->orph_first; 566 c->ohead_offs = 0; 567 return 0; 568 } 569 570 /** 571 * insert_dead_orphan - insert an orphan. 572 * @c: UBIFS file-system description object 573 * @inum: orphan inode number 574 * 575 * This function is a helper to the 'do_kill_orphans()' function. The orphan 576 * must be kept until the next commit, so it is added to the rb-tree and the 577 * deletion list. 578 */ 579 static int insert_dead_orphan(struct ubifs_info *c, ino_t inum) 580 { 581 struct ubifs_orphan *orphan, *o; 582 struct rb_node **p, *parent = NULL; 583 584 orphan = kzalloc(sizeof(struct ubifs_orphan), GFP_KERNEL); 585 if (!orphan) 586 return -ENOMEM; 587 orphan->inum = inum; 588 589 p = &c->orph_tree.rb_node; 590 while (*p) { 591 parent = *p; 592 o = rb_entry(parent, struct ubifs_orphan, rb); 593 if (inum < o->inum) 594 p = &(*p)->rb_left; 595 else if (inum > o->inum) 596 p = &(*p)->rb_right; 597 else { 598 /* Already added - no problem */ 599 kfree(orphan); 600 return 0; 601 } 602 } 603 c->tot_orphans += 1; 604 rb_link_node(&orphan->rb, parent, p); 605 rb_insert_color(&orphan->rb, &c->orph_tree); 606 list_add_tail(&orphan->list, &c->orph_list); 607 orphan->del = 1; 608 orphan->dnext = c->orph_dnext; 609 c->orph_dnext = orphan; 610 dbg_mnt("ino %lu, new %d, tot %d", (unsigned long)inum, 611 c->new_orphans, c->tot_orphans); 612 return 0; 613 } 614 615 /** 616 * do_kill_orphans - remove orphan inodes from the index. 617 * @c: UBIFS file-system description object 618 * @sleb: scanned LEB 619 * @last_cmt_no: cmt_no of last orphan node read is passed and returned here 620 * @outofdate: whether the LEB is out of date is returned here 621 * @last_flagged: whether the end orphan node is encountered 622 * 623 * This function is a helper to the 'kill_orphans()' function. It goes through 624 * every orphan node in a LEB and for every inode number recorded, removes 625 * all keys for that inode from the TNC. 626 */ 627 static int do_kill_orphans(struct ubifs_info *c, struct ubifs_scan_leb *sleb, 628 unsigned long long *last_cmt_no, int *outofdate, 629 int *last_flagged) 630 { 631 struct ubifs_scan_node *snod; 632 struct ubifs_orph_node *orph; 633 unsigned long long cmt_no; 634 ino_t inum; 635 int i, n, err, first = 1; 636 637 list_for_each_entry(snod, &sleb->nodes, list) { 638 if (snod->type != UBIFS_ORPH_NODE) { 639 ubifs_err(c, "invalid node type %d in orphan area at %d:%d", 640 snod->type, sleb->lnum, snod->offs); 641 ubifs_dump_node(c, snod->node); 642 return -EINVAL; 643 } 644 645 orph = snod->node; 646 647 /* Check commit number */ 648 cmt_no = le64_to_cpu(orph->cmt_no) & LLONG_MAX; 649 /* 650 * The commit number on the master node may be less, because 651 * of a failed commit. If there are several failed commits in a 652 * row, the commit number written on orphan nodes will continue 653 * to increase (because the commit number is adjusted here) even 654 * though the commit number on the master node stays the same 655 * because the master node has not been re-written. 656 */ 657 if (cmt_no > c->cmt_no) 658 c->cmt_no = cmt_no; 659 if (cmt_no < *last_cmt_no && *last_flagged) { 660 /* 661 * The last orphan node had a higher commit number and 662 * was flagged as the last written for that commit 663 * number. That makes this orphan node, out of date. 664 */ 665 if (!first) { 666 ubifs_err(c, "out of order commit number %llu in orphan node at %d:%d", 667 cmt_no, sleb->lnum, snod->offs); 668 ubifs_dump_node(c, snod->node); 669 return -EINVAL; 670 } 671 dbg_rcvry("out of date LEB %d", sleb->lnum); 672 *outofdate = 1; 673 return 0; 674 } 675 676 if (first) 677 first = 0; 678 679 n = (le32_to_cpu(orph->ch.len) - UBIFS_ORPH_NODE_SZ) >> 3; 680 for (i = 0; i < n; i++) { 681 union ubifs_key key1, key2; 682 683 inum = le64_to_cpu(orph->inos[i]); 684 dbg_rcvry("deleting orphaned inode %lu", 685 (unsigned long)inum); 686 687 lowest_ino_key(c, &key1, inum); 688 highest_ino_key(c, &key2, inum); 689 690 err = ubifs_tnc_remove_range(c, &key1, &key2); 691 if (err) 692 return err; 693 err = insert_dead_orphan(c, inum); 694 if (err) 695 return err; 696 } 697 698 *last_cmt_no = cmt_no; 699 if (le64_to_cpu(orph->cmt_no) & (1ULL << 63)) { 700 dbg_rcvry("last orph node for commit %llu at %d:%d", 701 cmt_no, sleb->lnum, snod->offs); 702 *last_flagged = 1; 703 } else 704 *last_flagged = 0; 705 } 706 707 return 0; 708 } 709 710 /** 711 * kill_orphans - remove all orphan inodes from the index. 712 * @c: UBIFS file-system description object 713 * 714 * If recovery is required, then orphan inodes recorded during the previous 715 * session (which ended with an unclean unmount) must be deleted from the index. 716 * This is done by updating the TNC, but since the index is not updated until 717 * the next commit, the LEBs where the orphan information is recorded are not 718 * erased until the next commit. 719 */ 720 static int kill_orphans(struct ubifs_info *c) 721 { 722 unsigned long long last_cmt_no = 0; 723 int lnum, err = 0, outofdate = 0, last_flagged = 0; 724 725 c->ohead_lnum = c->orph_first; 726 c->ohead_offs = 0; 727 /* Check no-orphans flag and skip this if no orphans */ 728 if (c->no_orphs) { 729 dbg_rcvry("no orphans"); 730 return 0; 731 } 732 /* 733 * Orph nodes always start at c->orph_first and are written to each 734 * successive LEB in turn. Generally unused LEBs will have been unmapped 735 * but may contain out of date orphan nodes if the unmap didn't go 736 * through. In addition, the last orphan node written for each commit is 737 * marked (top bit of orph->cmt_no is set to 1). It is possible that 738 * there are orphan nodes from the next commit (i.e. the commit did not 739 * complete successfully). In that case, no orphans will have been lost 740 * due to the way that orphans are written, and any orphans added will 741 * be valid orphans anyway and so can be deleted. 742 */ 743 for (lnum = c->orph_first; lnum <= c->orph_last; lnum++) { 744 struct ubifs_scan_leb *sleb; 745 746 dbg_rcvry("LEB %d", lnum); 747 sleb = ubifs_scan(c, lnum, 0, c->sbuf, 1); 748 if (IS_ERR(sleb)) { 749 if (PTR_ERR(sleb) == -EUCLEAN) 750 sleb = ubifs_recover_leb(c, lnum, 0, 751 c->sbuf, -1); 752 if (IS_ERR(sleb)) { 753 err = PTR_ERR(sleb); 754 break; 755 } 756 } 757 err = do_kill_orphans(c, sleb, &last_cmt_no, &outofdate, 758 &last_flagged); 759 if (err || outofdate) { 760 ubifs_scan_destroy(sleb); 761 break; 762 } 763 if (sleb->endpt) { 764 c->ohead_lnum = lnum; 765 c->ohead_offs = sleb->endpt; 766 } 767 ubifs_scan_destroy(sleb); 768 } 769 return err; 770 } 771 772 /** 773 * ubifs_mount_orphans - delete orphan inodes and erase LEBs that recorded them. 774 * @c: UBIFS file-system description object 775 * @unclean: indicates recovery from unclean unmount 776 * @read_only: indicates read only mount 777 * 778 * This function is called when mounting to erase orphans from the previous 779 * session. If UBIFS was not unmounted cleanly, then the inodes recorded as 780 * orphans are deleted. 781 */ 782 int ubifs_mount_orphans(struct ubifs_info *c, int unclean, int read_only) 783 { 784 int err = 0; 785 786 c->max_orphans = tot_avail_orphs(c); 787 788 if (!read_only) { 789 c->orph_buf = vmalloc(c->leb_size); 790 if (!c->orph_buf) 791 return -ENOMEM; 792 } 793 794 if (unclean) 795 err = kill_orphans(c); 796 else if (!read_only) 797 err = ubifs_clear_orphans(c); 798 799 return err; 800 } 801 802 /* 803 * Everything below is related to debugging. 804 */ 805 806 struct check_orphan { 807 struct rb_node rb; 808 ino_t inum; 809 }; 810 811 struct check_info { 812 unsigned long last_ino; 813 unsigned long tot_inos; 814 unsigned long missing; 815 unsigned long long leaf_cnt; 816 struct ubifs_ino_node *node; 817 struct rb_root root; 818 }; 819 820 static bool dbg_find_orphan(struct ubifs_info *c, ino_t inum) 821 { 822 bool found = false; 823 824 spin_lock(&c->orphan_lock); 825 found = !!lookup_orphan(c, inum); 826 spin_unlock(&c->orphan_lock); 827 828 return found; 829 } 830 831 static int dbg_ins_check_orphan(struct rb_root *root, ino_t inum) 832 { 833 struct check_orphan *orphan, *o; 834 struct rb_node **p, *parent = NULL; 835 836 orphan = kzalloc(sizeof(struct check_orphan), GFP_NOFS); 837 if (!orphan) 838 return -ENOMEM; 839 orphan->inum = inum; 840 841 p = &root->rb_node; 842 while (*p) { 843 parent = *p; 844 o = rb_entry(parent, struct check_orphan, rb); 845 if (inum < o->inum) 846 p = &(*p)->rb_left; 847 else if (inum > o->inum) 848 p = &(*p)->rb_right; 849 else { 850 kfree(orphan); 851 return 0; 852 } 853 } 854 rb_link_node(&orphan->rb, parent, p); 855 rb_insert_color(&orphan->rb, root); 856 return 0; 857 } 858 859 static int dbg_find_check_orphan(struct rb_root *root, ino_t inum) 860 { 861 struct check_orphan *o; 862 struct rb_node *p; 863 864 p = root->rb_node; 865 while (p) { 866 o = rb_entry(p, struct check_orphan, rb); 867 if (inum < o->inum) 868 p = p->rb_left; 869 else if (inum > o->inum) 870 p = p->rb_right; 871 else 872 return 1; 873 } 874 return 0; 875 } 876 877 static void dbg_free_check_tree(struct rb_root *root) 878 { 879 struct check_orphan *o, *n; 880 881 rbtree_postorder_for_each_entry_safe(o, n, root, rb) 882 kfree(o); 883 } 884 885 static int dbg_orphan_check(struct ubifs_info *c, struct ubifs_zbranch *zbr, 886 void *priv) 887 { 888 struct check_info *ci = priv; 889 ino_t inum; 890 int err; 891 892 inum = key_inum(c, &zbr->key); 893 if (inum != ci->last_ino) { 894 /* Lowest node type is the inode node, so it comes first */ 895 if (key_type(c, &zbr->key) != UBIFS_INO_KEY) 896 ubifs_err(c, "found orphan node ino %lu, type %d", 897 (unsigned long)inum, key_type(c, &zbr->key)); 898 ci->last_ino = inum; 899 ci->tot_inos += 1; 900 err = ubifs_tnc_read_node(c, zbr, ci->node); 901 if (err) { 902 ubifs_err(c, "node read failed, error %d", err); 903 return err; 904 } 905 if (ci->node->nlink == 0) 906 /* Must be recorded as an orphan */ 907 if (!dbg_find_check_orphan(&ci->root, inum) && 908 !dbg_find_orphan(c, inum)) { 909 ubifs_err(c, "missing orphan, ino %lu", 910 (unsigned long)inum); 911 ci->missing += 1; 912 } 913 } 914 ci->leaf_cnt += 1; 915 return 0; 916 } 917 918 static int dbg_read_orphans(struct check_info *ci, struct ubifs_scan_leb *sleb) 919 { 920 struct ubifs_scan_node *snod; 921 struct ubifs_orph_node *orph; 922 ino_t inum; 923 int i, n, err; 924 925 list_for_each_entry(snod, &sleb->nodes, list) { 926 cond_resched(); 927 if (snod->type != UBIFS_ORPH_NODE) 928 continue; 929 orph = snod->node; 930 n = (le32_to_cpu(orph->ch.len) - UBIFS_ORPH_NODE_SZ) >> 3; 931 for (i = 0; i < n; i++) { 932 inum = le64_to_cpu(orph->inos[i]); 933 err = dbg_ins_check_orphan(&ci->root, inum); 934 if (err) 935 return err; 936 } 937 } 938 return 0; 939 } 940 941 static int dbg_scan_orphans(struct ubifs_info *c, struct check_info *ci) 942 { 943 int lnum, err = 0; 944 void *buf; 945 946 /* Check no-orphans flag and skip this if no orphans */ 947 if (c->no_orphs) 948 return 0; 949 950 buf = __vmalloc(c->leb_size, GFP_NOFS, PAGE_KERNEL); 951 if (!buf) { 952 ubifs_err(c, "cannot allocate memory to check orphans"); 953 return 0; 954 } 955 956 for (lnum = c->orph_first; lnum <= c->orph_last; lnum++) { 957 struct ubifs_scan_leb *sleb; 958 959 sleb = ubifs_scan(c, lnum, 0, buf, 0); 960 if (IS_ERR(sleb)) { 961 err = PTR_ERR(sleb); 962 break; 963 } 964 965 err = dbg_read_orphans(ci, sleb); 966 ubifs_scan_destroy(sleb); 967 if (err) 968 break; 969 } 970 971 vfree(buf); 972 return err; 973 } 974 975 static int dbg_check_orphans(struct ubifs_info *c) 976 { 977 struct check_info ci; 978 int err; 979 980 if (!dbg_is_chk_orph(c)) 981 return 0; 982 983 ci.last_ino = 0; 984 ci.tot_inos = 0; 985 ci.missing = 0; 986 ci.leaf_cnt = 0; 987 ci.root = RB_ROOT; 988 ci.node = kmalloc(UBIFS_MAX_INO_NODE_SZ, GFP_NOFS); 989 if (!ci.node) { 990 ubifs_err(c, "out of memory"); 991 return -ENOMEM; 992 } 993 994 err = dbg_scan_orphans(c, &ci); 995 if (err) 996 goto out; 997 998 err = dbg_walk_index(c, &dbg_orphan_check, NULL, &ci); 999 if (err) { 1000 ubifs_err(c, "cannot scan TNC, error %d", err); 1001 goto out; 1002 } 1003 1004 if (ci.missing) { 1005 ubifs_err(c, "%lu missing orphan(s)", ci.missing); 1006 err = -EINVAL; 1007 goto out; 1008 } 1009 1010 dbg_cmt("last inode number is %lu", ci.last_ino); 1011 dbg_cmt("total number of inodes is %lu", ci.tot_inos); 1012 dbg_cmt("total number of leaf nodes is %llu", ci.leaf_cnt); 1013 1014 out: 1015 dbg_free_check_tree(&ci.root); 1016 kfree(ci.node); 1017 return err; 1018 } 1019