1 // SPDX-License-Identifier: GPL-2.0 2 3 #include "bcachefs.h" 4 #include "bkey_buf.h" 5 #include "btree_key_cache.h" 6 #include "btree_update.h" 7 #include "buckets.h" 8 #include "errcode.h" 9 #include "error.h" 10 #include "fs.h" 11 #include "recovery_passes.h" 12 #include "snapshot.h" 13 14 #include <linux/random.h> 15 16 /* 17 * Snapshot trees: 18 * 19 * Keys in BTREE_ID_snapshot_trees identify a whole tree of snapshot nodes; they 20 * exist to provide a stable identifier for the whole lifetime of a snapshot 21 * tree. 22 */ 23 24 void bch2_snapshot_tree_to_text(struct printbuf *out, struct bch_fs *c, 25 struct bkey_s_c k) 26 { 27 struct bkey_s_c_snapshot_tree t = bkey_s_c_to_snapshot_tree(k); 28 29 prt_printf(out, "subvol %u root snapshot %u", 30 le32_to_cpu(t.v->master_subvol), 31 le32_to_cpu(t.v->root_snapshot)); 32 } 33 34 int bch2_snapshot_tree_validate(struct bch_fs *c, struct bkey_s_c k, 35 enum bch_validate_flags flags) 36 { 37 int ret = 0; 38 39 bkey_fsck_err_on(bkey_gt(k.k->p, POS(0, U32_MAX)) || 40 bkey_lt(k.k->p, POS(0, 1)), 41 c, snapshot_tree_pos_bad, 42 "bad pos"); 43 fsck_err: 44 return ret; 45 } 46 47 int bch2_snapshot_tree_lookup(struct btree_trans *trans, u32 id, 48 struct bch_snapshot_tree *s) 49 { 50 int ret = bch2_bkey_get_val_typed(trans, BTREE_ID_snapshot_trees, POS(0, id), 51 BTREE_ITER_with_updates, snapshot_tree, s); 52 53 if (bch2_err_matches(ret, ENOENT)) 54 ret = -BCH_ERR_ENOENT_snapshot_tree; 55 return ret; 56 } 57 58 struct bkey_i_snapshot_tree * 59 __bch2_snapshot_tree_create(struct btree_trans *trans) 60 { 61 struct btree_iter iter; 62 int ret = bch2_bkey_get_empty_slot(trans, &iter, 63 BTREE_ID_snapshot_trees, POS(0, U32_MAX)); 64 struct bkey_i_snapshot_tree *s_t; 65 66 if (ret == -BCH_ERR_ENOSPC_btree_slot) 67 ret = -BCH_ERR_ENOSPC_snapshot_tree; 68 if (ret) 69 return ERR_PTR(ret); 70 71 s_t = bch2_bkey_alloc(trans, &iter, 0, snapshot_tree); 72 ret = PTR_ERR_OR_ZERO(s_t); 73 bch2_trans_iter_exit(trans, &iter); 74 return ret ? ERR_PTR(ret) : s_t; 75 } 76 77 static int bch2_snapshot_tree_create(struct btree_trans *trans, 78 u32 root_id, u32 subvol_id, u32 *tree_id) 79 { 80 struct bkey_i_snapshot_tree *n_tree = 81 __bch2_snapshot_tree_create(trans); 82 83 if (IS_ERR(n_tree)) 84 return PTR_ERR(n_tree); 85 86 n_tree->v.master_subvol = cpu_to_le32(subvol_id); 87 n_tree->v.root_snapshot = cpu_to_le32(root_id); 88 *tree_id = n_tree->k.p.offset; 89 return 0; 90 } 91 92 /* Snapshot nodes: */ 93 94 static bool __bch2_snapshot_is_ancestor_early(struct snapshot_table *t, u32 id, u32 ancestor) 95 { 96 while (id && id < ancestor) { 97 const struct snapshot_t *s = __snapshot_t(t, id); 98 id = s ? s->parent : 0; 99 } 100 return id == ancestor; 101 } 102 103 static bool bch2_snapshot_is_ancestor_early(struct bch_fs *c, u32 id, u32 ancestor) 104 { 105 rcu_read_lock(); 106 bool ret = __bch2_snapshot_is_ancestor_early(rcu_dereference(c->snapshots), id, ancestor); 107 rcu_read_unlock(); 108 109 return ret; 110 } 111 112 static inline u32 get_ancestor_below(struct snapshot_table *t, u32 id, u32 ancestor) 113 { 114 const struct snapshot_t *s = __snapshot_t(t, id); 115 if (!s) 116 return 0; 117 118 if (s->skip[2] <= ancestor) 119 return s->skip[2]; 120 if (s->skip[1] <= ancestor) 121 return s->skip[1]; 122 if (s->skip[0] <= ancestor) 123 return s->skip[0]; 124 return s->parent; 125 } 126 127 static bool test_ancestor_bitmap(struct snapshot_table *t, u32 id, u32 ancestor) 128 { 129 const struct snapshot_t *s = __snapshot_t(t, id); 130 if (!s) 131 return false; 132 133 return test_bit(ancestor - id - 1, s->is_ancestor); 134 } 135 136 bool __bch2_snapshot_is_ancestor(struct bch_fs *c, u32 id, u32 ancestor) 137 { 138 bool ret; 139 140 rcu_read_lock(); 141 struct snapshot_table *t = rcu_dereference(c->snapshots); 142 143 if (unlikely(c->recovery_pass_done < BCH_RECOVERY_PASS_check_snapshots)) { 144 ret = __bch2_snapshot_is_ancestor_early(t, id, ancestor); 145 goto out; 146 } 147 148 while (id && id < ancestor - IS_ANCESTOR_BITMAP) 149 id = get_ancestor_below(t, id, ancestor); 150 151 ret = id && id < ancestor 152 ? test_ancestor_bitmap(t, id, ancestor) 153 : id == ancestor; 154 155 EBUG_ON(ret != __bch2_snapshot_is_ancestor_early(t, id, ancestor)); 156 out: 157 rcu_read_unlock(); 158 159 return ret; 160 } 161 162 static noinline struct snapshot_t *__snapshot_t_mut(struct bch_fs *c, u32 id) 163 { 164 size_t idx = U32_MAX - id; 165 struct snapshot_table *new, *old; 166 167 size_t new_bytes = kmalloc_size_roundup(struct_size(new, s, idx + 1)); 168 size_t new_size = (new_bytes - sizeof(*new)) / sizeof(new->s[0]); 169 170 if (unlikely(new_bytes > INT_MAX)) 171 return NULL; 172 173 new = kvzalloc(new_bytes, GFP_KERNEL); 174 if (!new) 175 return NULL; 176 177 new->nr = new_size; 178 179 old = rcu_dereference_protected(c->snapshots, true); 180 if (old) 181 memcpy(new->s, old->s, sizeof(old->s[0]) * old->nr); 182 183 rcu_assign_pointer(c->snapshots, new); 184 kvfree_rcu(old, rcu); 185 186 return &rcu_dereference_protected(c->snapshots, 187 lockdep_is_held(&c->snapshot_table_lock))->s[idx]; 188 } 189 190 static inline struct snapshot_t *snapshot_t_mut(struct bch_fs *c, u32 id) 191 { 192 size_t idx = U32_MAX - id; 193 struct snapshot_table *table = 194 rcu_dereference_protected(c->snapshots, 195 lockdep_is_held(&c->snapshot_table_lock)); 196 197 lockdep_assert_held(&c->snapshot_table_lock); 198 199 if (likely(table && idx < table->nr)) 200 return &table->s[idx]; 201 202 return __snapshot_t_mut(c, id); 203 } 204 205 void bch2_snapshot_to_text(struct printbuf *out, struct bch_fs *c, 206 struct bkey_s_c k) 207 { 208 struct bkey_s_c_snapshot s = bkey_s_c_to_snapshot(k); 209 210 prt_printf(out, "is_subvol %llu deleted %llu parent %10u children %10u %10u subvol %u tree %u", 211 BCH_SNAPSHOT_SUBVOL(s.v), 212 BCH_SNAPSHOT_DELETED(s.v), 213 le32_to_cpu(s.v->parent), 214 le32_to_cpu(s.v->children[0]), 215 le32_to_cpu(s.v->children[1]), 216 le32_to_cpu(s.v->subvol), 217 le32_to_cpu(s.v->tree)); 218 219 if (bkey_val_bytes(k.k) > offsetof(struct bch_snapshot, depth)) 220 prt_printf(out, " depth %u skiplist %u %u %u", 221 le32_to_cpu(s.v->depth), 222 le32_to_cpu(s.v->skip[0]), 223 le32_to_cpu(s.v->skip[1]), 224 le32_to_cpu(s.v->skip[2])); 225 } 226 227 int bch2_snapshot_validate(struct bch_fs *c, struct bkey_s_c k, 228 enum bch_validate_flags flags) 229 { 230 struct bkey_s_c_snapshot s; 231 u32 i, id; 232 int ret = 0; 233 234 bkey_fsck_err_on(bkey_gt(k.k->p, POS(0, U32_MAX)) || 235 bkey_lt(k.k->p, POS(0, 1)), 236 c, snapshot_pos_bad, 237 "bad pos"); 238 239 s = bkey_s_c_to_snapshot(k); 240 241 id = le32_to_cpu(s.v->parent); 242 bkey_fsck_err_on(id && id <= k.k->p.offset, 243 c, snapshot_parent_bad, 244 "bad parent node (%u <= %llu)", 245 id, k.k->p.offset); 246 247 bkey_fsck_err_on(le32_to_cpu(s.v->children[0]) < le32_to_cpu(s.v->children[1]), 248 c, snapshot_children_not_normalized, 249 "children not normalized"); 250 251 bkey_fsck_err_on(s.v->children[0] && s.v->children[0] == s.v->children[1], 252 c, snapshot_child_duplicate, 253 "duplicate child nodes"); 254 255 for (i = 0; i < 2; i++) { 256 id = le32_to_cpu(s.v->children[i]); 257 258 bkey_fsck_err_on(id >= k.k->p.offset, 259 c, snapshot_child_bad, 260 "bad child node (%u >= %llu)", 261 id, k.k->p.offset); 262 } 263 264 if (bkey_val_bytes(k.k) > offsetof(struct bch_snapshot, skip)) { 265 bkey_fsck_err_on(le32_to_cpu(s.v->skip[0]) > le32_to_cpu(s.v->skip[1]) || 266 le32_to_cpu(s.v->skip[1]) > le32_to_cpu(s.v->skip[2]), 267 c, snapshot_skiplist_not_normalized, 268 "skiplist not normalized"); 269 270 for (i = 0; i < ARRAY_SIZE(s.v->skip); i++) { 271 id = le32_to_cpu(s.v->skip[i]); 272 273 bkey_fsck_err_on(id && id < le32_to_cpu(s.v->parent), 274 c, snapshot_skiplist_bad, 275 "bad skiplist node %u", id); 276 } 277 } 278 fsck_err: 279 return ret; 280 } 281 282 static void __set_is_ancestor_bitmap(struct bch_fs *c, u32 id) 283 { 284 struct snapshot_t *t = snapshot_t_mut(c, id); 285 u32 parent = id; 286 287 while ((parent = bch2_snapshot_parent_early(c, parent)) && 288 parent - id - 1 < IS_ANCESTOR_BITMAP) 289 __set_bit(parent - id - 1, t->is_ancestor); 290 } 291 292 static void set_is_ancestor_bitmap(struct bch_fs *c, u32 id) 293 { 294 mutex_lock(&c->snapshot_table_lock); 295 __set_is_ancestor_bitmap(c, id); 296 mutex_unlock(&c->snapshot_table_lock); 297 } 298 299 static int __bch2_mark_snapshot(struct btree_trans *trans, 300 enum btree_id btree, unsigned level, 301 struct bkey_s_c old, struct bkey_s_c new, 302 enum btree_iter_update_trigger_flags flags) 303 { 304 struct bch_fs *c = trans->c; 305 struct snapshot_t *t; 306 u32 id = new.k->p.offset; 307 int ret = 0; 308 309 mutex_lock(&c->snapshot_table_lock); 310 311 t = snapshot_t_mut(c, id); 312 if (!t) { 313 ret = -BCH_ERR_ENOMEM_mark_snapshot; 314 goto err; 315 } 316 317 if (new.k->type == KEY_TYPE_snapshot) { 318 struct bkey_s_c_snapshot s = bkey_s_c_to_snapshot(new); 319 320 t->parent = le32_to_cpu(s.v->parent); 321 t->children[0] = le32_to_cpu(s.v->children[0]); 322 t->children[1] = le32_to_cpu(s.v->children[1]); 323 t->subvol = BCH_SNAPSHOT_SUBVOL(s.v) ? le32_to_cpu(s.v->subvol) : 0; 324 t->tree = le32_to_cpu(s.v->tree); 325 326 if (bkey_val_bytes(s.k) > offsetof(struct bch_snapshot, depth)) { 327 t->depth = le32_to_cpu(s.v->depth); 328 t->skip[0] = le32_to_cpu(s.v->skip[0]); 329 t->skip[1] = le32_to_cpu(s.v->skip[1]); 330 t->skip[2] = le32_to_cpu(s.v->skip[2]); 331 } else { 332 t->depth = 0; 333 t->skip[0] = 0; 334 t->skip[1] = 0; 335 t->skip[2] = 0; 336 } 337 338 __set_is_ancestor_bitmap(c, id); 339 340 if (BCH_SNAPSHOT_DELETED(s.v)) { 341 set_bit(BCH_FS_need_delete_dead_snapshots, &c->flags); 342 if (c->curr_recovery_pass > BCH_RECOVERY_PASS_delete_dead_snapshots) 343 bch2_delete_dead_snapshots_async(c); 344 } 345 } else { 346 memset(t, 0, sizeof(*t)); 347 } 348 err: 349 mutex_unlock(&c->snapshot_table_lock); 350 return ret; 351 } 352 353 int bch2_mark_snapshot(struct btree_trans *trans, 354 enum btree_id btree, unsigned level, 355 struct bkey_s_c old, struct bkey_s new, 356 enum btree_iter_update_trigger_flags flags) 357 { 358 return __bch2_mark_snapshot(trans, btree, level, old, new.s_c, flags); 359 } 360 361 int bch2_snapshot_lookup(struct btree_trans *trans, u32 id, 362 struct bch_snapshot *s) 363 { 364 return bch2_bkey_get_val_typed(trans, BTREE_ID_snapshots, POS(0, id), 365 BTREE_ITER_with_updates, snapshot, s); 366 } 367 368 static int bch2_snapshot_live(struct btree_trans *trans, u32 id) 369 { 370 struct bch_snapshot v; 371 int ret; 372 373 if (!id) 374 return 0; 375 376 ret = bch2_snapshot_lookup(trans, id, &v); 377 if (bch2_err_matches(ret, ENOENT)) 378 bch_err(trans->c, "snapshot node %u not found", id); 379 if (ret) 380 return ret; 381 382 return !BCH_SNAPSHOT_DELETED(&v); 383 } 384 385 /* 386 * If @k is a snapshot with just one live child, it's part of a linear chain, 387 * which we consider to be an equivalence class: and then after snapshot 388 * deletion cleanup, there should only be a single key at a given position in 389 * this equivalence class. 390 * 391 * This sets the equivalence class of @k to be the child's equivalence class, if 392 * it's part of such a linear chain: this correctly sets equivalence classes on 393 * startup if we run leaf to root (i.e. in natural key order). 394 */ 395 static int bch2_snapshot_set_equiv(struct btree_trans *trans, struct bkey_s_c k) 396 { 397 struct bch_fs *c = trans->c; 398 unsigned i, nr_live = 0, live_idx = 0; 399 struct bkey_s_c_snapshot snap; 400 u32 id = k.k->p.offset, child[2]; 401 402 if (k.k->type != KEY_TYPE_snapshot) 403 return 0; 404 405 snap = bkey_s_c_to_snapshot(k); 406 407 child[0] = le32_to_cpu(snap.v->children[0]); 408 child[1] = le32_to_cpu(snap.v->children[1]); 409 410 for (i = 0; i < 2; i++) { 411 int ret = bch2_snapshot_live(trans, child[i]); 412 413 if (ret < 0) 414 return ret; 415 416 if (ret) 417 live_idx = i; 418 nr_live += ret; 419 } 420 421 mutex_lock(&c->snapshot_table_lock); 422 423 snapshot_t_mut(c, id)->equiv = nr_live == 1 424 ? snapshot_t_mut(c, child[live_idx])->equiv 425 : id; 426 427 mutex_unlock(&c->snapshot_table_lock); 428 429 return 0; 430 } 431 432 /* fsck: */ 433 434 static u32 bch2_snapshot_child(struct bch_fs *c, u32 id, unsigned child) 435 { 436 return snapshot_t(c, id)->children[child]; 437 } 438 439 static u32 bch2_snapshot_left_child(struct bch_fs *c, u32 id) 440 { 441 return bch2_snapshot_child(c, id, 0); 442 } 443 444 static u32 bch2_snapshot_right_child(struct bch_fs *c, u32 id) 445 { 446 return bch2_snapshot_child(c, id, 1); 447 } 448 449 static u32 bch2_snapshot_tree_next(struct bch_fs *c, u32 id) 450 { 451 u32 n, parent; 452 453 n = bch2_snapshot_left_child(c, id); 454 if (n) 455 return n; 456 457 while ((parent = bch2_snapshot_parent(c, id))) { 458 n = bch2_snapshot_right_child(c, parent); 459 if (n && n != id) 460 return n; 461 id = parent; 462 } 463 464 return 0; 465 } 466 467 static u32 bch2_snapshot_tree_oldest_subvol(struct bch_fs *c, u32 snapshot_root) 468 { 469 u32 id = snapshot_root; 470 u32 subvol = 0, s; 471 472 rcu_read_lock(); 473 while (id) { 474 s = snapshot_t(c, id)->subvol; 475 476 if (s && (!subvol || s < subvol)) 477 subvol = s; 478 479 id = bch2_snapshot_tree_next(c, id); 480 } 481 rcu_read_unlock(); 482 483 return subvol; 484 } 485 486 static int bch2_snapshot_tree_master_subvol(struct btree_trans *trans, 487 u32 snapshot_root, u32 *subvol_id) 488 { 489 struct bch_fs *c = trans->c; 490 struct btree_iter iter; 491 struct bkey_s_c k; 492 bool found = false; 493 int ret; 494 495 for_each_btree_key_norestart(trans, iter, BTREE_ID_subvolumes, POS_MIN, 496 0, k, ret) { 497 if (k.k->type != KEY_TYPE_subvolume) 498 continue; 499 500 struct bkey_s_c_subvolume s = bkey_s_c_to_subvolume(k); 501 if (!bch2_snapshot_is_ancestor(c, le32_to_cpu(s.v->snapshot), snapshot_root)) 502 continue; 503 if (!BCH_SUBVOLUME_SNAP(s.v)) { 504 *subvol_id = s.k->p.offset; 505 found = true; 506 break; 507 } 508 } 509 510 bch2_trans_iter_exit(trans, &iter); 511 512 if (!ret && !found) { 513 struct bkey_i_subvolume *u; 514 515 *subvol_id = bch2_snapshot_tree_oldest_subvol(c, snapshot_root); 516 517 u = bch2_bkey_get_mut_typed(trans, &iter, 518 BTREE_ID_subvolumes, POS(0, *subvol_id), 519 0, subvolume); 520 ret = PTR_ERR_OR_ZERO(u); 521 if (ret) 522 return ret; 523 524 SET_BCH_SUBVOLUME_SNAP(&u->v, false); 525 } 526 527 return ret; 528 } 529 530 static int check_snapshot_tree(struct btree_trans *trans, 531 struct btree_iter *iter, 532 struct bkey_s_c k) 533 { 534 struct bch_fs *c = trans->c; 535 struct bkey_s_c_snapshot_tree st; 536 struct bch_snapshot s; 537 struct bch_subvolume subvol; 538 struct printbuf buf = PRINTBUF; 539 u32 root_id; 540 int ret; 541 542 if (k.k->type != KEY_TYPE_snapshot_tree) 543 return 0; 544 545 st = bkey_s_c_to_snapshot_tree(k); 546 root_id = le32_to_cpu(st.v->root_snapshot); 547 548 ret = bch2_snapshot_lookup(trans, root_id, &s); 549 if (ret && !bch2_err_matches(ret, ENOENT)) 550 goto err; 551 552 if (fsck_err_on(ret || 553 root_id != bch2_snapshot_root(c, root_id) || 554 st.k->p.offset != le32_to_cpu(s.tree), 555 trans, snapshot_tree_to_missing_snapshot, 556 "snapshot tree points to missing/incorrect snapshot:\n %s", 557 (bch2_bkey_val_to_text(&buf, c, st.s_c), buf.buf))) { 558 ret = bch2_btree_delete_at(trans, iter, 0); 559 goto err; 560 } 561 562 ret = bch2_subvolume_get(trans, le32_to_cpu(st.v->master_subvol), 563 false, 0, &subvol); 564 if (ret && !bch2_err_matches(ret, ENOENT)) 565 goto err; 566 567 if (fsck_err_on(ret, 568 trans, snapshot_tree_to_missing_subvol, 569 "snapshot tree points to missing subvolume:\n %s", 570 (printbuf_reset(&buf), 571 bch2_bkey_val_to_text(&buf, c, st.s_c), buf.buf)) || 572 fsck_err_on(!bch2_snapshot_is_ancestor(c, 573 le32_to_cpu(subvol.snapshot), 574 root_id), 575 trans, snapshot_tree_to_wrong_subvol, 576 "snapshot tree points to subvolume that does not point to snapshot in this tree:\n %s", 577 (printbuf_reset(&buf), 578 bch2_bkey_val_to_text(&buf, c, st.s_c), buf.buf)) || 579 fsck_err_on(BCH_SUBVOLUME_SNAP(&subvol), 580 trans, snapshot_tree_to_snapshot_subvol, 581 "snapshot tree points to snapshot subvolume:\n %s", 582 (printbuf_reset(&buf), 583 bch2_bkey_val_to_text(&buf, c, st.s_c), buf.buf))) { 584 struct bkey_i_snapshot_tree *u; 585 u32 subvol_id; 586 587 ret = bch2_snapshot_tree_master_subvol(trans, root_id, &subvol_id); 588 bch_err_fn(c, ret); 589 590 if (bch2_err_matches(ret, ENOENT)) { /* nothing to be done here */ 591 ret = 0; 592 goto err; 593 } 594 595 if (ret) 596 goto err; 597 598 u = bch2_bkey_make_mut_typed(trans, iter, &k, 0, snapshot_tree); 599 ret = PTR_ERR_OR_ZERO(u); 600 if (ret) 601 goto err; 602 603 u->v.master_subvol = cpu_to_le32(subvol_id); 604 st = snapshot_tree_i_to_s_c(u); 605 } 606 err: 607 fsck_err: 608 printbuf_exit(&buf); 609 return ret; 610 } 611 612 /* 613 * For each snapshot_tree, make sure it points to the root of a snapshot tree 614 * and that snapshot entry points back to it, or delete it. 615 * 616 * And, make sure it points to a subvolume within that snapshot tree, or correct 617 * it to point to the oldest subvolume within that snapshot tree. 618 */ 619 int bch2_check_snapshot_trees(struct bch_fs *c) 620 { 621 int ret = bch2_trans_run(c, 622 for_each_btree_key_commit(trans, iter, 623 BTREE_ID_snapshot_trees, POS_MIN, 624 BTREE_ITER_prefetch, k, 625 NULL, NULL, BCH_TRANS_COMMIT_no_enospc, 626 check_snapshot_tree(trans, &iter, k))); 627 bch_err_fn(c, ret); 628 return ret; 629 } 630 631 /* 632 * Look up snapshot tree for @tree_id and find root, 633 * make sure @snap_id is a descendent: 634 */ 635 static int snapshot_tree_ptr_good(struct btree_trans *trans, 636 u32 snap_id, u32 tree_id) 637 { 638 struct bch_snapshot_tree s_t; 639 int ret = bch2_snapshot_tree_lookup(trans, tree_id, &s_t); 640 641 if (bch2_err_matches(ret, ENOENT)) 642 return 0; 643 if (ret) 644 return ret; 645 646 return bch2_snapshot_is_ancestor_early(trans->c, snap_id, le32_to_cpu(s_t.root_snapshot)); 647 } 648 649 u32 bch2_snapshot_skiplist_get(struct bch_fs *c, u32 id) 650 { 651 const struct snapshot_t *s; 652 653 if (!id) 654 return 0; 655 656 rcu_read_lock(); 657 s = snapshot_t(c, id); 658 if (s->parent) 659 id = bch2_snapshot_nth_parent(c, id, get_random_u32_below(s->depth)); 660 rcu_read_unlock(); 661 662 return id; 663 } 664 665 static int snapshot_skiplist_good(struct btree_trans *trans, u32 id, struct bch_snapshot s) 666 { 667 unsigned i; 668 669 for (i = 0; i < 3; i++) 670 if (!s.parent) { 671 if (s.skip[i]) 672 return false; 673 } else { 674 if (!bch2_snapshot_is_ancestor_early(trans->c, id, le32_to_cpu(s.skip[i]))) 675 return false; 676 } 677 678 return true; 679 } 680 681 /* 682 * snapshot_tree pointer was incorrect: look up root snapshot node, make sure 683 * its snapshot_tree pointer is correct (allocate new one if necessary), then 684 * update this node's pointer to root node's pointer: 685 */ 686 static int snapshot_tree_ptr_repair(struct btree_trans *trans, 687 struct btree_iter *iter, 688 struct bkey_s_c k, 689 struct bch_snapshot *s) 690 { 691 struct bch_fs *c = trans->c; 692 struct btree_iter root_iter; 693 struct bch_snapshot_tree s_t; 694 struct bkey_s_c_snapshot root; 695 struct bkey_i_snapshot *u; 696 u32 root_id = bch2_snapshot_root(c, k.k->p.offset), tree_id; 697 int ret; 698 699 root = bch2_bkey_get_iter_typed(trans, &root_iter, 700 BTREE_ID_snapshots, POS(0, root_id), 701 BTREE_ITER_with_updates, snapshot); 702 ret = bkey_err(root); 703 if (ret) 704 goto err; 705 706 tree_id = le32_to_cpu(root.v->tree); 707 708 ret = bch2_snapshot_tree_lookup(trans, tree_id, &s_t); 709 if (ret && !bch2_err_matches(ret, ENOENT)) 710 return ret; 711 712 if (ret || le32_to_cpu(s_t.root_snapshot) != root_id) { 713 u = bch2_bkey_make_mut_typed(trans, &root_iter, &root.s_c, 0, snapshot); 714 ret = PTR_ERR_OR_ZERO(u) ?: 715 bch2_snapshot_tree_create(trans, root_id, 716 bch2_snapshot_tree_oldest_subvol(c, root_id), 717 &tree_id); 718 if (ret) 719 goto err; 720 721 u->v.tree = cpu_to_le32(tree_id); 722 if (k.k->p.offset == root_id) 723 *s = u->v; 724 } 725 726 if (k.k->p.offset != root_id) { 727 u = bch2_bkey_make_mut_typed(trans, iter, &k, 0, snapshot); 728 ret = PTR_ERR_OR_ZERO(u); 729 if (ret) 730 goto err; 731 732 u->v.tree = cpu_to_le32(tree_id); 733 *s = u->v; 734 } 735 err: 736 bch2_trans_iter_exit(trans, &root_iter); 737 return ret; 738 } 739 740 static int check_snapshot(struct btree_trans *trans, 741 struct btree_iter *iter, 742 struct bkey_s_c k) 743 { 744 struct bch_fs *c = trans->c; 745 struct bch_snapshot s; 746 struct bch_subvolume subvol; 747 struct bch_snapshot v; 748 struct bkey_i_snapshot *u; 749 u32 parent_id = bch2_snapshot_parent_early(c, k.k->p.offset); 750 u32 real_depth; 751 struct printbuf buf = PRINTBUF; 752 u32 i, id; 753 int ret = 0; 754 755 if (k.k->type != KEY_TYPE_snapshot) 756 return 0; 757 758 memset(&s, 0, sizeof(s)); 759 memcpy(&s, k.v, min(sizeof(s), bkey_val_bytes(k.k))); 760 761 id = le32_to_cpu(s.parent); 762 if (id) { 763 ret = bch2_snapshot_lookup(trans, id, &v); 764 if (bch2_err_matches(ret, ENOENT)) 765 bch_err(c, "snapshot with nonexistent parent:\n %s", 766 (bch2_bkey_val_to_text(&buf, c, k), buf.buf)); 767 if (ret) 768 goto err; 769 770 if (le32_to_cpu(v.children[0]) != k.k->p.offset && 771 le32_to_cpu(v.children[1]) != k.k->p.offset) { 772 bch_err(c, "snapshot parent %u missing pointer to child %llu", 773 id, k.k->p.offset); 774 ret = -EINVAL; 775 goto err; 776 } 777 } 778 779 for (i = 0; i < 2 && s.children[i]; i++) { 780 id = le32_to_cpu(s.children[i]); 781 782 ret = bch2_snapshot_lookup(trans, id, &v); 783 if (bch2_err_matches(ret, ENOENT)) 784 bch_err(c, "snapshot node %llu has nonexistent child %u", 785 k.k->p.offset, id); 786 if (ret) 787 goto err; 788 789 if (le32_to_cpu(v.parent) != k.k->p.offset) { 790 bch_err(c, "snapshot child %u has wrong parent (got %u should be %llu)", 791 id, le32_to_cpu(v.parent), k.k->p.offset); 792 ret = -EINVAL; 793 goto err; 794 } 795 } 796 797 bool should_have_subvol = BCH_SNAPSHOT_SUBVOL(&s) && 798 !BCH_SNAPSHOT_DELETED(&s); 799 800 if (should_have_subvol) { 801 id = le32_to_cpu(s.subvol); 802 ret = bch2_subvolume_get(trans, id, 0, false, &subvol); 803 if (bch2_err_matches(ret, ENOENT)) 804 bch_err(c, "snapshot points to nonexistent subvolume:\n %s", 805 (bch2_bkey_val_to_text(&buf, c, k), buf.buf)); 806 if (ret) 807 goto err; 808 809 if (BCH_SNAPSHOT_SUBVOL(&s) != (le32_to_cpu(subvol.snapshot) == k.k->p.offset)) { 810 bch_err(c, "snapshot node %llu has wrong BCH_SNAPSHOT_SUBVOL", 811 k.k->p.offset); 812 ret = -EINVAL; 813 goto err; 814 } 815 } else { 816 if (fsck_err_on(s.subvol, 817 trans, snapshot_should_not_have_subvol, 818 "snapshot should not point to subvol:\n %s", 819 (bch2_bkey_val_to_text(&buf, c, k), buf.buf))) { 820 u = bch2_bkey_make_mut_typed(trans, iter, &k, 0, snapshot); 821 ret = PTR_ERR_OR_ZERO(u); 822 if (ret) 823 goto err; 824 825 u->v.subvol = 0; 826 s = u->v; 827 } 828 } 829 830 ret = snapshot_tree_ptr_good(trans, k.k->p.offset, le32_to_cpu(s.tree)); 831 if (ret < 0) 832 goto err; 833 834 if (fsck_err_on(!ret, 835 trans, snapshot_to_bad_snapshot_tree, 836 "snapshot points to missing/incorrect tree:\n %s", 837 (bch2_bkey_val_to_text(&buf, c, k), buf.buf))) { 838 ret = snapshot_tree_ptr_repair(trans, iter, k, &s); 839 if (ret) 840 goto err; 841 } 842 ret = 0; 843 844 real_depth = bch2_snapshot_depth(c, parent_id); 845 846 if (fsck_err_on(le32_to_cpu(s.depth) != real_depth, 847 trans, snapshot_bad_depth, 848 "snapshot with incorrect depth field, should be %u:\n %s", 849 real_depth, (bch2_bkey_val_to_text(&buf, c, k), buf.buf))) { 850 u = bch2_bkey_make_mut_typed(trans, iter, &k, 0, snapshot); 851 ret = PTR_ERR_OR_ZERO(u); 852 if (ret) 853 goto err; 854 855 u->v.depth = cpu_to_le32(real_depth); 856 s = u->v; 857 } 858 859 ret = snapshot_skiplist_good(trans, k.k->p.offset, s); 860 if (ret < 0) 861 goto err; 862 863 if (fsck_err_on(!ret, 864 trans, snapshot_bad_skiplist, 865 "snapshot with bad skiplist field:\n %s", 866 (bch2_bkey_val_to_text(&buf, c, k), buf.buf))) { 867 u = bch2_bkey_make_mut_typed(trans, iter, &k, 0, snapshot); 868 ret = PTR_ERR_OR_ZERO(u); 869 if (ret) 870 goto err; 871 872 for (i = 0; i < ARRAY_SIZE(u->v.skip); i++) 873 u->v.skip[i] = cpu_to_le32(bch2_snapshot_skiplist_get(c, parent_id)); 874 875 bubble_sort(u->v.skip, ARRAY_SIZE(u->v.skip), cmp_le32); 876 s = u->v; 877 } 878 ret = 0; 879 err: 880 fsck_err: 881 printbuf_exit(&buf); 882 return ret; 883 } 884 885 int bch2_check_snapshots(struct bch_fs *c) 886 { 887 /* 888 * We iterate backwards as checking/fixing the depth field requires that 889 * the parent's depth already be correct: 890 */ 891 int ret = bch2_trans_run(c, 892 for_each_btree_key_reverse_commit(trans, iter, 893 BTREE_ID_snapshots, POS_MAX, 894 BTREE_ITER_prefetch, k, 895 NULL, NULL, BCH_TRANS_COMMIT_no_enospc, 896 check_snapshot(trans, &iter, k))); 897 bch_err_fn(c, ret); 898 return ret; 899 } 900 901 static int check_snapshot_exists(struct btree_trans *trans, u32 id) 902 { 903 struct bch_fs *c = trans->c; 904 905 if (bch2_snapshot_equiv(c, id)) 906 return 0; 907 908 /* Do we need to reconstruct the snapshot_tree entry as well? */ 909 struct btree_iter iter; 910 struct bkey_s_c k; 911 int ret = 0; 912 u32 tree_id = 0; 913 914 for_each_btree_key_norestart(trans, iter, BTREE_ID_snapshot_trees, POS_MIN, 915 0, k, ret) { 916 if (le32_to_cpu(bkey_s_c_to_snapshot_tree(k).v->root_snapshot) == id) { 917 tree_id = k.k->p.offset; 918 break; 919 } 920 } 921 bch2_trans_iter_exit(trans, &iter); 922 923 if (ret) 924 return ret; 925 926 if (!tree_id) { 927 ret = bch2_snapshot_tree_create(trans, id, 0, &tree_id); 928 if (ret) 929 return ret; 930 } 931 932 struct bkey_i_snapshot *snapshot = bch2_trans_kmalloc(trans, sizeof(*snapshot)); 933 ret = PTR_ERR_OR_ZERO(snapshot); 934 if (ret) 935 return ret; 936 937 bkey_snapshot_init(&snapshot->k_i); 938 snapshot->k.p = POS(0, id); 939 snapshot->v.tree = cpu_to_le32(tree_id); 940 snapshot->v.btime.lo = cpu_to_le64(bch2_current_time(c)); 941 942 for_each_btree_key_norestart(trans, iter, BTREE_ID_subvolumes, POS_MIN, 943 0, k, ret) { 944 if (le32_to_cpu(bkey_s_c_to_subvolume(k).v->snapshot) == id) { 945 snapshot->v.subvol = cpu_to_le32(k.k->p.offset); 946 SET_BCH_SNAPSHOT_SUBVOL(&snapshot->v, true); 947 break; 948 } 949 } 950 bch2_trans_iter_exit(trans, &iter); 951 952 return bch2_btree_insert_trans(trans, BTREE_ID_snapshots, &snapshot->k_i, 0) ?: 953 bch2_mark_snapshot(trans, BTREE_ID_snapshots, 0, 954 bkey_s_c_null, bkey_i_to_s(&snapshot->k_i), 0) ?: 955 bch2_snapshot_set_equiv(trans, bkey_i_to_s_c(&snapshot->k_i)); 956 } 957 958 /* Figure out which snapshot nodes belong in the same tree: */ 959 struct snapshot_tree_reconstruct { 960 enum btree_id btree; 961 struct bpos cur_pos; 962 snapshot_id_list cur_ids; 963 DARRAY(snapshot_id_list) trees; 964 }; 965 966 static void snapshot_tree_reconstruct_exit(struct snapshot_tree_reconstruct *r) 967 { 968 darray_for_each(r->trees, i) 969 darray_exit(i); 970 darray_exit(&r->trees); 971 darray_exit(&r->cur_ids); 972 } 973 974 static inline bool same_snapshot(struct snapshot_tree_reconstruct *r, struct bpos pos) 975 { 976 return r->btree == BTREE_ID_inodes 977 ? r->cur_pos.offset == pos.offset 978 : r->cur_pos.inode == pos.inode; 979 } 980 981 static inline bool snapshot_id_lists_have_common(snapshot_id_list *l, snapshot_id_list *r) 982 { 983 darray_for_each(*l, i) 984 if (snapshot_list_has_id(r, *i)) 985 return true; 986 return false; 987 } 988 989 static void snapshot_id_list_to_text(struct printbuf *out, snapshot_id_list *s) 990 { 991 bool first = true; 992 darray_for_each(*s, i) { 993 if (!first) 994 prt_char(out, ' '); 995 first = false; 996 prt_printf(out, "%u", *i); 997 } 998 } 999 1000 static int snapshot_tree_reconstruct_next(struct bch_fs *c, struct snapshot_tree_reconstruct *r) 1001 { 1002 if (r->cur_ids.nr) { 1003 darray_for_each(r->trees, i) 1004 if (snapshot_id_lists_have_common(i, &r->cur_ids)) { 1005 int ret = snapshot_list_merge(c, i, &r->cur_ids); 1006 if (ret) 1007 return ret; 1008 goto out; 1009 } 1010 darray_push(&r->trees, r->cur_ids); 1011 darray_init(&r->cur_ids); 1012 } 1013 out: 1014 r->cur_ids.nr = 0; 1015 return 0; 1016 } 1017 1018 static int get_snapshot_trees(struct bch_fs *c, struct snapshot_tree_reconstruct *r, struct bpos pos) 1019 { 1020 if (!same_snapshot(r, pos)) 1021 snapshot_tree_reconstruct_next(c, r); 1022 r->cur_pos = pos; 1023 return snapshot_list_add_nodup(c, &r->cur_ids, pos.snapshot); 1024 } 1025 1026 int bch2_reconstruct_snapshots(struct bch_fs *c) 1027 { 1028 struct btree_trans *trans = bch2_trans_get(c); 1029 struct printbuf buf = PRINTBUF; 1030 struct snapshot_tree_reconstruct r = {}; 1031 int ret = 0; 1032 1033 for (unsigned btree = 0; btree < BTREE_ID_NR; btree++) { 1034 if (btree_type_has_snapshots(btree)) { 1035 r.btree = btree; 1036 1037 ret = for_each_btree_key(trans, iter, btree, POS_MIN, 1038 BTREE_ITER_all_snapshots|BTREE_ITER_prefetch, k, ({ 1039 get_snapshot_trees(c, &r, k.k->p); 1040 })); 1041 if (ret) 1042 goto err; 1043 1044 snapshot_tree_reconstruct_next(c, &r); 1045 } 1046 } 1047 1048 darray_for_each(r.trees, t) { 1049 printbuf_reset(&buf); 1050 snapshot_id_list_to_text(&buf, t); 1051 1052 darray_for_each(*t, id) { 1053 if (fsck_err_on(!bch2_snapshot_equiv(c, *id), 1054 trans, snapshot_node_missing, 1055 "snapshot node %u from tree %s missing, recreate?", *id, buf.buf)) { 1056 if (t->nr > 1) { 1057 bch_err(c, "cannot reconstruct snapshot trees with multiple nodes"); 1058 ret = -BCH_ERR_fsck_repair_unimplemented; 1059 goto err; 1060 } 1061 1062 ret = commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc, 1063 check_snapshot_exists(trans, *id)); 1064 if (ret) 1065 goto err; 1066 } 1067 } 1068 } 1069 fsck_err: 1070 err: 1071 bch2_trans_put(trans); 1072 snapshot_tree_reconstruct_exit(&r); 1073 printbuf_exit(&buf); 1074 bch_err_fn(c, ret); 1075 return ret; 1076 } 1077 1078 int bch2_check_key_has_snapshot(struct btree_trans *trans, 1079 struct btree_iter *iter, 1080 struct bkey_s_c k) 1081 { 1082 struct bch_fs *c = trans->c; 1083 struct printbuf buf = PRINTBUF; 1084 int ret = 0; 1085 1086 if (fsck_err_on(!bch2_snapshot_equiv(c, k.k->p.snapshot), 1087 trans, bkey_in_missing_snapshot, 1088 "key in missing snapshot %s, delete?", 1089 (bch2_bkey_val_to_text(&buf, c, k), buf.buf))) 1090 ret = bch2_btree_delete_at(trans, iter, 1091 BTREE_UPDATE_internal_snapshot_node) ?: 1; 1092 fsck_err: 1093 printbuf_exit(&buf); 1094 return ret; 1095 } 1096 1097 /* 1098 * Mark a snapshot as deleted, for future cleanup: 1099 */ 1100 int bch2_snapshot_node_set_deleted(struct btree_trans *trans, u32 id) 1101 { 1102 struct btree_iter iter; 1103 struct bkey_i_snapshot *s; 1104 int ret = 0; 1105 1106 s = bch2_bkey_get_mut_typed(trans, &iter, 1107 BTREE_ID_snapshots, POS(0, id), 1108 0, snapshot); 1109 ret = PTR_ERR_OR_ZERO(s); 1110 if (unlikely(ret)) { 1111 bch2_fs_inconsistent_on(bch2_err_matches(ret, ENOENT), 1112 trans->c, "missing snapshot %u", id); 1113 return ret; 1114 } 1115 1116 /* already deleted? */ 1117 if (BCH_SNAPSHOT_DELETED(&s->v)) 1118 goto err; 1119 1120 SET_BCH_SNAPSHOT_DELETED(&s->v, true); 1121 SET_BCH_SNAPSHOT_SUBVOL(&s->v, false); 1122 s->v.subvol = 0; 1123 err: 1124 bch2_trans_iter_exit(trans, &iter); 1125 return ret; 1126 } 1127 1128 static inline void normalize_snapshot_child_pointers(struct bch_snapshot *s) 1129 { 1130 if (le32_to_cpu(s->children[0]) < le32_to_cpu(s->children[1])) 1131 swap(s->children[0], s->children[1]); 1132 } 1133 1134 static int bch2_snapshot_node_delete(struct btree_trans *trans, u32 id) 1135 { 1136 struct bch_fs *c = trans->c; 1137 struct btree_iter iter, p_iter = (struct btree_iter) { NULL }; 1138 struct btree_iter c_iter = (struct btree_iter) { NULL }; 1139 struct btree_iter tree_iter = (struct btree_iter) { NULL }; 1140 struct bkey_s_c_snapshot s; 1141 u32 parent_id, child_id; 1142 unsigned i; 1143 int ret = 0; 1144 1145 s = bch2_bkey_get_iter_typed(trans, &iter, BTREE_ID_snapshots, POS(0, id), 1146 BTREE_ITER_intent, snapshot); 1147 ret = bkey_err(s); 1148 bch2_fs_inconsistent_on(bch2_err_matches(ret, ENOENT), c, 1149 "missing snapshot %u", id); 1150 1151 if (ret) 1152 goto err; 1153 1154 BUG_ON(s.v->children[1]); 1155 1156 parent_id = le32_to_cpu(s.v->parent); 1157 child_id = le32_to_cpu(s.v->children[0]); 1158 1159 if (parent_id) { 1160 struct bkey_i_snapshot *parent; 1161 1162 parent = bch2_bkey_get_mut_typed(trans, &p_iter, 1163 BTREE_ID_snapshots, POS(0, parent_id), 1164 0, snapshot); 1165 ret = PTR_ERR_OR_ZERO(parent); 1166 bch2_fs_inconsistent_on(bch2_err_matches(ret, ENOENT), c, 1167 "missing snapshot %u", parent_id); 1168 if (unlikely(ret)) 1169 goto err; 1170 1171 /* find entry in parent->children for node being deleted */ 1172 for (i = 0; i < 2; i++) 1173 if (le32_to_cpu(parent->v.children[i]) == id) 1174 break; 1175 1176 if (bch2_fs_inconsistent_on(i == 2, c, 1177 "snapshot %u missing child pointer to %u", 1178 parent_id, id)) 1179 goto err; 1180 1181 parent->v.children[i] = cpu_to_le32(child_id); 1182 1183 normalize_snapshot_child_pointers(&parent->v); 1184 } 1185 1186 if (child_id) { 1187 struct bkey_i_snapshot *child; 1188 1189 child = bch2_bkey_get_mut_typed(trans, &c_iter, 1190 BTREE_ID_snapshots, POS(0, child_id), 1191 0, snapshot); 1192 ret = PTR_ERR_OR_ZERO(child); 1193 bch2_fs_inconsistent_on(bch2_err_matches(ret, ENOENT), c, 1194 "missing snapshot %u", child_id); 1195 if (unlikely(ret)) 1196 goto err; 1197 1198 child->v.parent = cpu_to_le32(parent_id); 1199 1200 if (!child->v.parent) { 1201 child->v.skip[0] = 0; 1202 child->v.skip[1] = 0; 1203 child->v.skip[2] = 0; 1204 } 1205 } 1206 1207 if (!parent_id) { 1208 /* 1209 * We're deleting the root of a snapshot tree: update the 1210 * snapshot_tree entry to point to the new root, or delete it if 1211 * this is the last snapshot ID in this tree: 1212 */ 1213 struct bkey_i_snapshot_tree *s_t; 1214 1215 BUG_ON(s.v->children[1]); 1216 1217 s_t = bch2_bkey_get_mut_typed(trans, &tree_iter, 1218 BTREE_ID_snapshot_trees, POS(0, le32_to_cpu(s.v->tree)), 1219 0, snapshot_tree); 1220 ret = PTR_ERR_OR_ZERO(s_t); 1221 if (ret) 1222 goto err; 1223 1224 if (s.v->children[0]) { 1225 s_t->v.root_snapshot = s.v->children[0]; 1226 } else { 1227 s_t->k.type = KEY_TYPE_deleted; 1228 set_bkey_val_u64s(&s_t->k, 0); 1229 } 1230 } 1231 1232 ret = bch2_btree_delete_at(trans, &iter, 0); 1233 err: 1234 bch2_trans_iter_exit(trans, &tree_iter); 1235 bch2_trans_iter_exit(trans, &p_iter); 1236 bch2_trans_iter_exit(trans, &c_iter); 1237 bch2_trans_iter_exit(trans, &iter); 1238 return ret; 1239 } 1240 1241 static int create_snapids(struct btree_trans *trans, u32 parent, u32 tree, 1242 u32 *new_snapids, 1243 u32 *snapshot_subvols, 1244 unsigned nr_snapids) 1245 { 1246 struct bch_fs *c = trans->c; 1247 struct btree_iter iter; 1248 struct bkey_i_snapshot *n; 1249 struct bkey_s_c k; 1250 unsigned i, j; 1251 u32 depth = bch2_snapshot_depth(c, parent); 1252 int ret; 1253 1254 bch2_trans_iter_init(trans, &iter, BTREE_ID_snapshots, 1255 POS_MIN, BTREE_ITER_intent); 1256 k = bch2_btree_iter_peek(&iter); 1257 ret = bkey_err(k); 1258 if (ret) 1259 goto err; 1260 1261 for (i = 0; i < nr_snapids; i++) { 1262 k = bch2_btree_iter_prev_slot(&iter); 1263 ret = bkey_err(k); 1264 if (ret) 1265 goto err; 1266 1267 if (!k.k || !k.k->p.offset) { 1268 ret = -BCH_ERR_ENOSPC_snapshot_create; 1269 goto err; 1270 } 1271 1272 n = bch2_bkey_alloc(trans, &iter, 0, snapshot); 1273 ret = PTR_ERR_OR_ZERO(n); 1274 if (ret) 1275 goto err; 1276 1277 n->v.flags = 0; 1278 n->v.parent = cpu_to_le32(parent); 1279 n->v.subvol = cpu_to_le32(snapshot_subvols[i]); 1280 n->v.tree = cpu_to_le32(tree); 1281 n->v.depth = cpu_to_le32(depth); 1282 n->v.btime.lo = cpu_to_le64(bch2_current_time(c)); 1283 n->v.btime.hi = 0; 1284 1285 for (j = 0; j < ARRAY_SIZE(n->v.skip); j++) 1286 n->v.skip[j] = cpu_to_le32(bch2_snapshot_skiplist_get(c, parent)); 1287 1288 bubble_sort(n->v.skip, ARRAY_SIZE(n->v.skip), cmp_le32); 1289 SET_BCH_SNAPSHOT_SUBVOL(&n->v, true); 1290 1291 ret = __bch2_mark_snapshot(trans, BTREE_ID_snapshots, 0, 1292 bkey_s_c_null, bkey_i_to_s_c(&n->k_i), 0); 1293 if (ret) 1294 goto err; 1295 1296 new_snapids[i] = iter.pos.offset; 1297 1298 mutex_lock(&c->snapshot_table_lock); 1299 snapshot_t_mut(c, new_snapids[i])->equiv = new_snapids[i]; 1300 mutex_unlock(&c->snapshot_table_lock); 1301 } 1302 err: 1303 bch2_trans_iter_exit(trans, &iter); 1304 return ret; 1305 } 1306 1307 /* 1308 * Create new snapshot IDs as children of an existing snapshot ID: 1309 */ 1310 static int bch2_snapshot_node_create_children(struct btree_trans *trans, u32 parent, 1311 u32 *new_snapids, 1312 u32 *snapshot_subvols, 1313 unsigned nr_snapids) 1314 { 1315 struct btree_iter iter; 1316 struct bkey_i_snapshot *n_parent; 1317 int ret = 0; 1318 1319 n_parent = bch2_bkey_get_mut_typed(trans, &iter, 1320 BTREE_ID_snapshots, POS(0, parent), 1321 0, snapshot); 1322 ret = PTR_ERR_OR_ZERO(n_parent); 1323 if (unlikely(ret)) { 1324 if (bch2_err_matches(ret, ENOENT)) 1325 bch_err(trans->c, "snapshot %u not found", parent); 1326 return ret; 1327 } 1328 1329 if (n_parent->v.children[0] || n_parent->v.children[1]) { 1330 bch_err(trans->c, "Trying to add child snapshot nodes to parent that already has children"); 1331 ret = -EINVAL; 1332 goto err; 1333 } 1334 1335 ret = create_snapids(trans, parent, le32_to_cpu(n_parent->v.tree), 1336 new_snapids, snapshot_subvols, nr_snapids); 1337 if (ret) 1338 goto err; 1339 1340 n_parent->v.children[0] = cpu_to_le32(new_snapids[0]); 1341 n_parent->v.children[1] = cpu_to_le32(new_snapids[1]); 1342 n_parent->v.subvol = 0; 1343 SET_BCH_SNAPSHOT_SUBVOL(&n_parent->v, false); 1344 err: 1345 bch2_trans_iter_exit(trans, &iter); 1346 return ret; 1347 } 1348 1349 /* 1350 * Create a snapshot node that is the root of a new tree: 1351 */ 1352 static int bch2_snapshot_node_create_tree(struct btree_trans *trans, 1353 u32 *new_snapids, 1354 u32 *snapshot_subvols, 1355 unsigned nr_snapids) 1356 { 1357 struct bkey_i_snapshot_tree *n_tree; 1358 int ret; 1359 1360 n_tree = __bch2_snapshot_tree_create(trans); 1361 ret = PTR_ERR_OR_ZERO(n_tree) ?: 1362 create_snapids(trans, 0, n_tree->k.p.offset, 1363 new_snapids, snapshot_subvols, nr_snapids); 1364 if (ret) 1365 return ret; 1366 1367 n_tree->v.master_subvol = cpu_to_le32(snapshot_subvols[0]); 1368 n_tree->v.root_snapshot = cpu_to_le32(new_snapids[0]); 1369 return 0; 1370 } 1371 1372 int bch2_snapshot_node_create(struct btree_trans *trans, u32 parent, 1373 u32 *new_snapids, 1374 u32 *snapshot_subvols, 1375 unsigned nr_snapids) 1376 { 1377 BUG_ON((parent == 0) != (nr_snapids == 1)); 1378 BUG_ON((parent != 0) != (nr_snapids == 2)); 1379 1380 return parent 1381 ? bch2_snapshot_node_create_children(trans, parent, 1382 new_snapids, snapshot_subvols, nr_snapids) 1383 : bch2_snapshot_node_create_tree(trans, 1384 new_snapids, snapshot_subvols, nr_snapids); 1385 1386 } 1387 1388 /* 1389 * If we have an unlinked inode in an internal snapshot node, and the inode 1390 * really has been deleted in all child snapshots, how does this get cleaned up? 1391 * 1392 * first there is the problem of how keys that have been overwritten in all 1393 * child snapshots get deleted (unimplemented?), but inodes may perhaps be 1394 * special? 1395 * 1396 * also: unlinked inode in internal snapshot appears to not be getting deleted 1397 * correctly if inode doesn't exist in leaf snapshots 1398 * 1399 * solution: 1400 * 1401 * for a key in an interior snapshot node that needs work to be done that 1402 * requires it to be mutated: iterate over all descendent leaf nodes and copy 1403 * that key to snapshot leaf nodes, where we can mutate it 1404 */ 1405 1406 static int delete_dead_snapshots_process_key(struct btree_trans *trans, 1407 struct btree_iter *iter, 1408 struct bkey_s_c k, 1409 snapshot_id_list *deleted, 1410 snapshot_id_list *equiv_seen, 1411 struct bpos *last_pos) 1412 { 1413 int ret = bch2_check_key_has_snapshot(trans, iter, k); 1414 if (ret) 1415 return ret < 0 ? ret : 0; 1416 1417 struct bch_fs *c = trans->c; 1418 u32 equiv = bch2_snapshot_equiv(c, k.k->p.snapshot); 1419 if (!equiv) /* key for invalid snapshot node, but we chose not to delete */ 1420 return 0; 1421 1422 if (!bkey_eq(k.k->p, *last_pos)) 1423 equiv_seen->nr = 0; 1424 1425 if (snapshot_list_has_id(deleted, k.k->p.snapshot)) 1426 return bch2_btree_delete_at(trans, iter, 1427 BTREE_UPDATE_internal_snapshot_node); 1428 1429 if (!bpos_eq(*last_pos, k.k->p) && 1430 snapshot_list_has_id(equiv_seen, equiv)) 1431 return bch2_btree_delete_at(trans, iter, 1432 BTREE_UPDATE_internal_snapshot_node); 1433 1434 *last_pos = k.k->p; 1435 1436 ret = snapshot_list_add_nodup(c, equiv_seen, equiv); 1437 if (ret) 1438 return ret; 1439 1440 /* 1441 * When we have a linear chain of snapshot nodes, we consider 1442 * those to form an equivalence class: we're going to collapse 1443 * them all down to a single node, and keep the leaf-most node - 1444 * which has the same id as the equivalence class id. 1445 * 1446 * If there are multiple keys in different snapshots at the same 1447 * position, we're only going to keep the one in the newest 1448 * snapshot (we delete the others above) - the rest have been 1449 * overwritten and are redundant, and for the key we're going to keep we 1450 * need to move it to the equivalance class ID if it's not there 1451 * already. 1452 */ 1453 if (equiv != k.k->p.snapshot) { 1454 struct bkey_i *new = bch2_bkey_make_mut_noupdate(trans, k); 1455 int ret = PTR_ERR_OR_ZERO(new); 1456 if (ret) 1457 return ret; 1458 1459 new->k.p.snapshot = equiv; 1460 1461 struct btree_iter new_iter; 1462 bch2_trans_iter_init(trans, &new_iter, iter->btree_id, new->k.p, 1463 BTREE_ITER_all_snapshots| 1464 BTREE_ITER_cached| 1465 BTREE_ITER_intent); 1466 1467 ret = bch2_btree_iter_traverse(&new_iter) ?: 1468 bch2_trans_update(trans, &new_iter, new, 1469 BTREE_UPDATE_internal_snapshot_node) ?: 1470 bch2_btree_delete_at(trans, iter, 1471 BTREE_UPDATE_internal_snapshot_node); 1472 bch2_trans_iter_exit(trans, &new_iter); 1473 if (ret) 1474 return ret; 1475 } 1476 1477 return 0; 1478 } 1479 1480 static int bch2_snapshot_needs_delete(struct btree_trans *trans, struct bkey_s_c k) 1481 { 1482 struct bkey_s_c_snapshot snap; 1483 u32 children[2]; 1484 int ret; 1485 1486 if (k.k->type != KEY_TYPE_snapshot) 1487 return 0; 1488 1489 snap = bkey_s_c_to_snapshot(k); 1490 if (BCH_SNAPSHOT_DELETED(snap.v) || 1491 BCH_SNAPSHOT_SUBVOL(snap.v)) 1492 return 0; 1493 1494 children[0] = le32_to_cpu(snap.v->children[0]); 1495 children[1] = le32_to_cpu(snap.v->children[1]); 1496 1497 ret = bch2_snapshot_live(trans, children[0]) ?: 1498 bch2_snapshot_live(trans, children[1]); 1499 if (ret < 0) 1500 return ret; 1501 return !ret; 1502 } 1503 1504 /* 1505 * For a given snapshot, if it doesn't have a subvolume that points to it, and 1506 * it doesn't have child snapshot nodes - it's now redundant and we can mark it 1507 * as deleted. 1508 */ 1509 static int bch2_delete_redundant_snapshot(struct btree_trans *trans, struct bkey_s_c k) 1510 { 1511 int ret = bch2_snapshot_needs_delete(trans, k); 1512 1513 return ret <= 0 1514 ? ret 1515 : bch2_snapshot_node_set_deleted(trans, k.k->p.offset); 1516 } 1517 1518 static inline u32 bch2_snapshot_nth_parent_skip(struct bch_fs *c, u32 id, u32 n, 1519 snapshot_id_list *skip) 1520 { 1521 rcu_read_lock(); 1522 while (snapshot_list_has_id(skip, id)) 1523 id = __bch2_snapshot_parent(c, id); 1524 1525 while (n--) { 1526 do { 1527 id = __bch2_snapshot_parent(c, id); 1528 } while (snapshot_list_has_id(skip, id)); 1529 } 1530 rcu_read_unlock(); 1531 1532 return id; 1533 } 1534 1535 static int bch2_fix_child_of_deleted_snapshot(struct btree_trans *trans, 1536 struct btree_iter *iter, struct bkey_s_c k, 1537 snapshot_id_list *deleted) 1538 { 1539 struct bch_fs *c = trans->c; 1540 u32 nr_deleted_ancestors = 0; 1541 struct bkey_i_snapshot *s; 1542 int ret; 1543 1544 if (k.k->type != KEY_TYPE_snapshot) 1545 return 0; 1546 1547 if (snapshot_list_has_id(deleted, k.k->p.offset)) 1548 return 0; 1549 1550 s = bch2_bkey_make_mut_noupdate_typed(trans, k, snapshot); 1551 ret = PTR_ERR_OR_ZERO(s); 1552 if (ret) 1553 return ret; 1554 1555 darray_for_each(*deleted, i) 1556 nr_deleted_ancestors += bch2_snapshot_is_ancestor(c, s->k.p.offset, *i); 1557 1558 if (!nr_deleted_ancestors) 1559 return 0; 1560 1561 le32_add_cpu(&s->v.depth, -nr_deleted_ancestors); 1562 1563 if (!s->v.depth) { 1564 s->v.skip[0] = 0; 1565 s->v.skip[1] = 0; 1566 s->v.skip[2] = 0; 1567 } else { 1568 u32 depth = le32_to_cpu(s->v.depth); 1569 u32 parent = bch2_snapshot_parent(c, s->k.p.offset); 1570 1571 for (unsigned j = 0; j < ARRAY_SIZE(s->v.skip); j++) { 1572 u32 id = le32_to_cpu(s->v.skip[j]); 1573 1574 if (snapshot_list_has_id(deleted, id)) { 1575 id = bch2_snapshot_nth_parent_skip(c, 1576 parent, 1577 depth > 1 1578 ? get_random_u32_below(depth - 1) 1579 : 0, 1580 deleted); 1581 s->v.skip[j] = cpu_to_le32(id); 1582 } 1583 } 1584 1585 bubble_sort(s->v.skip, ARRAY_SIZE(s->v.skip), cmp_le32); 1586 } 1587 1588 return bch2_trans_update(trans, iter, &s->k_i, 0); 1589 } 1590 1591 int bch2_delete_dead_snapshots(struct bch_fs *c) 1592 { 1593 struct btree_trans *trans; 1594 snapshot_id_list deleted = { 0 }; 1595 snapshot_id_list deleted_interior = { 0 }; 1596 int ret = 0; 1597 1598 if (!test_and_clear_bit(BCH_FS_need_delete_dead_snapshots, &c->flags)) 1599 return 0; 1600 1601 trans = bch2_trans_get(c); 1602 1603 /* 1604 * For every snapshot node: If we have no live children and it's not 1605 * pointed to by a subvolume, delete it: 1606 */ 1607 ret = for_each_btree_key_commit(trans, iter, BTREE_ID_snapshots, 1608 POS_MIN, 0, k, 1609 NULL, NULL, 0, 1610 bch2_delete_redundant_snapshot(trans, k)); 1611 bch_err_msg(c, ret, "deleting redundant snapshots"); 1612 if (ret) 1613 goto err; 1614 1615 ret = for_each_btree_key(trans, iter, BTREE_ID_snapshots, 1616 POS_MIN, 0, k, 1617 bch2_snapshot_set_equiv(trans, k)); 1618 bch_err_msg(c, ret, "in bch2_snapshots_set_equiv"); 1619 if (ret) 1620 goto err; 1621 1622 ret = for_each_btree_key(trans, iter, BTREE_ID_snapshots, 1623 POS_MIN, 0, k, ({ 1624 if (k.k->type != KEY_TYPE_snapshot) 1625 continue; 1626 1627 BCH_SNAPSHOT_DELETED(bkey_s_c_to_snapshot(k).v) 1628 ? snapshot_list_add(c, &deleted, k.k->p.offset) 1629 : 0; 1630 })); 1631 bch_err_msg(c, ret, "walking snapshots"); 1632 if (ret) 1633 goto err; 1634 1635 for (unsigned btree = 0; btree < BTREE_ID_NR; btree++) { 1636 struct bpos last_pos = POS_MIN; 1637 snapshot_id_list equiv_seen = { 0 }; 1638 struct disk_reservation res = { 0 }; 1639 1640 if (!btree_type_has_snapshots(btree)) 1641 continue; 1642 1643 ret = for_each_btree_key_commit(trans, iter, 1644 btree, POS_MIN, 1645 BTREE_ITER_prefetch|BTREE_ITER_all_snapshots, k, 1646 &res, NULL, BCH_TRANS_COMMIT_no_enospc, 1647 delete_dead_snapshots_process_key(trans, &iter, k, &deleted, 1648 &equiv_seen, &last_pos)); 1649 1650 bch2_disk_reservation_put(c, &res); 1651 darray_exit(&equiv_seen); 1652 1653 bch_err_msg(c, ret, "deleting keys from dying snapshots"); 1654 if (ret) 1655 goto err; 1656 } 1657 1658 bch2_trans_unlock(trans); 1659 down_write(&c->snapshot_create_lock); 1660 1661 ret = for_each_btree_key(trans, iter, BTREE_ID_snapshots, 1662 POS_MIN, 0, k, ({ 1663 u32 snapshot = k.k->p.offset; 1664 u32 equiv = bch2_snapshot_equiv(c, snapshot); 1665 1666 equiv != snapshot 1667 ? snapshot_list_add(c, &deleted_interior, snapshot) 1668 : 0; 1669 })); 1670 1671 bch_err_msg(c, ret, "walking snapshots"); 1672 if (ret) 1673 goto err_create_lock; 1674 1675 /* 1676 * Fixing children of deleted snapshots can't be done completely 1677 * atomically, if we crash between here and when we delete the interior 1678 * nodes some depth fields will be off: 1679 */ 1680 ret = for_each_btree_key_commit(trans, iter, BTREE_ID_snapshots, POS_MIN, 1681 BTREE_ITER_intent, k, 1682 NULL, NULL, BCH_TRANS_COMMIT_no_enospc, 1683 bch2_fix_child_of_deleted_snapshot(trans, &iter, k, &deleted_interior)); 1684 if (ret) 1685 goto err_create_lock; 1686 1687 darray_for_each(deleted, i) { 1688 ret = commit_do(trans, NULL, NULL, 0, 1689 bch2_snapshot_node_delete(trans, *i)); 1690 bch_err_msg(c, ret, "deleting snapshot %u", *i); 1691 if (ret) 1692 goto err_create_lock; 1693 } 1694 1695 darray_for_each(deleted_interior, i) { 1696 ret = commit_do(trans, NULL, NULL, 0, 1697 bch2_snapshot_node_delete(trans, *i)); 1698 bch_err_msg(c, ret, "deleting snapshot %u", *i); 1699 if (ret) 1700 goto err_create_lock; 1701 } 1702 err_create_lock: 1703 up_write(&c->snapshot_create_lock); 1704 err: 1705 darray_exit(&deleted_interior); 1706 darray_exit(&deleted); 1707 bch2_trans_put(trans); 1708 bch_err_fn(c, ret); 1709 return ret; 1710 } 1711 1712 void bch2_delete_dead_snapshots_work(struct work_struct *work) 1713 { 1714 struct bch_fs *c = container_of(work, struct bch_fs, snapshot_delete_work); 1715 1716 set_worker_desc("bcachefs-delete-dead-snapshots/%s", c->name); 1717 1718 bch2_delete_dead_snapshots(c); 1719 bch2_write_ref_put(c, BCH_WRITE_REF_delete_dead_snapshots); 1720 } 1721 1722 void bch2_delete_dead_snapshots_async(struct bch_fs *c) 1723 { 1724 if (bch2_write_ref_tryget(c, BCH_WRITE_REF_delete_dead_snapshots) && 1725 !queue_work(c->write_ref_wq, &c->snapshot_delete_work)) 1726 bch2_write_ref_put(c, BCH_WRITE_REF_delete_dead_snapshots); 1727 } 1728 1729 int __bch2_key_has_snapshot_overwrites(struct btree_trans *trans, 1730 enum btree_id id, 1731 struct bpos pos) 1732 { 1733 struct bch_fs *c = trans->c; 1734 struct btree_iter iter; 1735 struct bkey_s_c k; 1736 int ret; 1737 1738 bch2_trans_iter_init(trans, &iter, id, pos, 1739 BTREE_ITER_not_extents| 1740 BTREE_ITER_all_snapshots); 1741 while (1) { 1742 k = bch2_btree_iter_prev(&iter); 1743 ret = bkey_err(k); 1744 if (ret) 1745 break; 1746 1747 if (!k.k) 1748 break; 1749 1750 if (!bkey_eq(pos, k.k->p)) 1751 break; 1752 1753 if (bch2_snapshot_is_ancestor(c, k.k->p.snapshot, pos.snapshot)) { 1754 ret = 1; 1755 break; 1756 } 1757 } 1758 bch2_trans_iter_exit(trans, &iter); 1759 1760 return ret; 1761 } 1762 1763 static int bch2_check_snapshot_needs_deletion(struct btree_trans *trans, struct bkey_s_c k) 1764 { 1765 struct bch_fs *c = trans->c; 1766 struct bkey_s_c_snapshot snap; 1767 int ret = 0; 1768 1769 if (k.k->type != KEY_TYPE_snapshot) 1770 return 0; 1771 1772 snap = bkey_s_c_to_snapshot(k); 1773 if (BCH_SNAPSHOT_DELETED(snap.v) || 1774 bch2_snapshot_equiv(c, k.k->p.offset) != k.k->p.offset || 1775 (ret = bch2_snapshot_needs_delete(trans, k)) > 0) { 1776 set_bit(BCH_FS_need_delete_dead_snapshots, &c->flags); 1777 return 0; 1778 } 1779 1780 return ret; 1781 } 1782 1783 int bch2_snapshots_read(struct bch_fs *c) 1784 { 1785 int ret = bch2_trans_run(c, 1786 for_each_btree_key(trans, iter, BTREE_ID_snapshots, 1787 POS_MIN, 0, k, 1788 __bch2_mark_snapshot(trans, BTREE_ID_snapshots, 0, bkey_s_c_null, k, 0) ?: 1789 bch2_snapshot_set_equiv(trans, k) ?: 1790 bch2_check_snapshot_needs_deletion(trans, k)) ?: 1791 for_each_btree_key(trans, iter, BTREE_ID_snapshots, 1792 POS_MIN, 0, k, 1793 (set_is_ancestor_bitmap(c, k.k->p.offset), 0))); 1794 bch_err_fn(c, ret); 1795 1796 /* 1797 * It's important that we check if we need to reconstruct snapshots 1798 * before going RW, so we mark that pass as required in the superblock - 1799 * otherwise, we could end up deleting keys with missing snapshot nodes 1800 * instead 1801 */ 1802 BUG_ON(!test_bit(BCH_FS_new_fs, &c->flags) && 1803 test_bit(BCH_FS_may_go_rw, &c->flags)); 1804 1805 if (bch2_err_matches(ret, EIO) || 1806 (c->sb.btrees_lost_data & BIT_ULL(BTREE_ID_snapshots))) 1807 ret = bch2_run_explicit_recovery_pass_persistent(c, BCH_RECOVERY_PASS_reconstruct_snapshots); 1808 1809 return ret; 1810 } 1811 1812 void bch2_fs_snapshots_exit(struct bch_fs *c) 1813 { 1814 kvfree(rcu_dereference_protected(c->snapshots, true)); 1815 } 1816