diff options
Diffstat (limited to 'fs/bcachefs/subvolume.c')
-rw-r--r-- | fs/bcachefs/subvolume.c | 260 |
1 files changed, 159 insertions, 101 deletions
diff --git a/fs/bcachefs/subvolume.c b/fs/bcachefs/subvolume.c index 88a79c823276..35c9f86a73c1 100644 --- a/fs/bcachefs/subvolume.c +++ b/fs/bcachefs/subvolume.c @@ -3,9 +3,11 @@ #include "bcachefs.h" #include "btree_key_cache.h" #include "btree_update.h" +#include "enumerated_ref.h" #include "errcode.h" #include "error.h" #include "fs.h" +#include "recovery_passes.h" #include "snapshot.h" #include "subvolume.h" @@ -13,6 +15,22 @@ static int bch2_subvolume_delete(struct btree_trans *, u32); +static int bch2_subvolume_missing(struct bch_fs *c, u32 subvolid) +{ + struct printbuf buf = PRINTBUF; + bch2_log_msg_start(c, &buf); + + prt_printf(&buf, "missing subvolume %u", subvolid); + bool print = bch2_count_fsck_err(c, subvol_missing, &buf); + + int ret = bch2_run_explicit_recovery_pass(c, &buf, + BCH_RECOVERY_PASS_check_inodes, 0); + if (print) + bch2_print_str(c, KERN_ERR, buf.buf); + printbuf_exit(&buf); + return ret; +} + static struct bpos subvolume_children_pos(struct bkey_s_c k) { if (k.k->type != KEY_TYPE_subvolume) @@ -44,8 +62,8 @@ static int check_subvol(struct btree_trans *trans, ret = bch2_snapshot_lookup(trans, snapid, &snapshot); if (bch2_err_matches(ret, ENOENT)) - bch_err(c, "subvolume %llu points to nonexistent snapshot %u", - k.k->p.offset, snapid); + return bch2_run_print_explicit_recovery_pass(c, + BCH_RECOVERY_PASS_reconstruct_snapshots) ?: ret; if (ret) return ret; @@ -57,7 +75,7 @@ static int check_subvol(struct btree_trans *trans, if (fsck_err_on(subvol.k->p.offset == BCACHEFS_ROOT_SUBVOL && subvol.v->fs_path_parent, - c, subvol_root_fs_path_parent_nonzero, + trans, subvol_root_fs_path_parent_nonzero, "root subvolume has nonzero fs_path_parent\n%s", (bch2_bkey_val_to_text(&buf, c, k), buf.buf))) { struct bkey_i_subvolume *n = @@ -80,7 +98,7 @@ static int check_subvol(struct btree_trans *trans, goto err; if (fsck_err_on(subvol_children_k.k->type != KEY_TYPE_set, - c, subvol_children_not_set, + trans, subvol_children_not_set, "subvolume not set in subvolume_children btree at %llu:%llu\n%s", pos.inode, pos.offset, (printbuf_reset(&buf), @@ -92,33 +110,33 @@ static int check_subvol(struct btree_trans *trans, } struct bch_inode_unpacked inode; - struct btree_iter inode_iter = {}; - ret = bch2_inode_peek_nowarn(trans, &inode_iter, &inode, + ret = bch2_inode_find_by_inum_nowarn_trans(trans, (subvol_inum) { k.k->p.offset, le64_to_cpu(subvol.v->inode) }, - 0); - bch2_trans_iter_exit(trans, &inode_iter); - - if (ret && !bch2_err_matches(ret, ENOENT)) - return ret; - - if (fsck_err_on(ret, c, subvol_to_missing_root, - "subvolume %llu points to missing subvolume root %llu:%u", - k.k->p.offset, le64_to_cpu(subvol.v->inode), - le32_to_cpu(subvol.v->snapshot))) { - ret = bch2_subvolume_delete(trans, iter->pos.offset); - bch_err_msg(c, ret, "deleting subvolume %llu", iter->pos.offset); - return ret ?: -BCH_ERR_transaction_restart_nested; - } - - if (fsck_err_on(inode.bi_subvol != subvol.k->p.offset, - c, subvol_root_wrong_bi_subvol, - "subvol root %llu:%u has wrong bi_subvol field: got %u, should be %llu", - inode.bi_inum, inode_iter.k.p.snapshot, - inode.bi_subvol, subvol.k->p.offset)) { - inode.bi_subvol = subvol.k->p.offset; - ret = __bch2_fsck_write_inode(trans, &inode, le32_to_cpu(subvol.v->snapshot)); - if (ret) + &inode); + if (!ret) { + if (fsck_err_on(inode.bi_subvol != subvol.k->p.offset, + trans, subvol_root_wrong_bi_subvol, + "subvol root %llu:%u has wrong bi_subvol field: got %u, should be %llu", + inode.bi_inum, inode.bi_snapshot, + inode.bi_subvol, subvol.k->p.offset)) { + inode.bi_subvol = subvol.k->p.offset; + inode.bi_snapshot = le32_to_cpu(subvol.v->snapshot); + ret = __bch2_fsck_write_inode(trans, &inode); + if (ret) + goto err; + } + } else if (bch2_err_matches(ret, ENOENT)) { + if (fsck_err(trans, subvol_to_missing_root, + "subvolume %llu points to missing subvolume root %llu:%u", + k.k->p.offset, le64_to_cpu(subvol.v->inode), + le32_to_cpu(subvol.v->snapshot))) { + ret = bch2_subvolume_delete(trans, iter->pos.offset); + bch_err_msg(c, ret, "deleting subvolume %llu", iter->pos.offset); + ret = ret ?: -BCH_ERR_transaction_restart_nested; goto err; + } + } else { + goto err; } if (!BCH_SUBVOLUME_SNAP(subvol.v)) { @@ -136,17 +154,17 @@ static int check_subvol(struct btree_trans *trans, "%s: snapshot tree %u not found", __func__, snapshot_tree); if (ret) - return ret; + goto err; if (fsck_err_on(le32_to_cpu(st.master_subvol) != subvol.k->p.offset, - c, subvol_not_master_and_not_snapshot, + trans, subvol_not_master_and_not_snapshot, "subvolume %llu is not set as snapshot but is not master subvolume", k.k->p.offset)) { struct bkey_i_subvolume *s = bch2_bkey_make_mut_typed(trans, iter, &subvol.s_c, 0, subvolume); ret = PTR_ERR_OR_ZERO(s); if (ret) - return ret; + goto err; SET_BCH_SUBVOLUME_SNAP(&s->v, true); } @@ -162,7 +180,7 @@ int bch2_check_subvols(struct bch_fs *c) { int ret = bch2_trans_run(c, for_each_btree_key_commit(trans, iter, - BTREE_ID_subvolumes, POS_MIN, BTREE_ITER_PREFETCH, k, + BTREE_ID_subvolumes, POS_MIN, BTREE_ITER_prefetch, k, NULL, NULL, BCH_TRANS_COMMIT_no_enospc, check_subvol(trans, &iter, k))); bch_err_fn(c, ret); @@ -173,7 +191,6 @@ static int check_subvol_child(struct btree_trans *trans, struct btree_iter *child_iter, struct bkey_s_c child_k) { - struct bch_fs *c = trans->c; struct bch_subvolume s; int ret = bch2_bkey_get_val_typed(trans, BTREE_ID_subvolumes, POS(0, child_k.k->p.offset), 0, subvolume, &s); @@ -182,7 +199,7 @@ static int check_subvol_child(struct btree_trans *trans, if (fsck_err_on(ret || le32_to_cpu(s.fs_path_parent) != child_k.k->p.inode, - c, subvol_children_bad, + trans, subvol_children_bad, "incorrect entry in subvolume_children btree %llu:%llu", child_k.k->p.inode, child_k.k->p.offset)) { ret = bch2_btree_delete_at(trans, child_iter, 0); @@ -198,7 +215,7 @@ int bch2_check_subvol_children(struct bch_fs *c) { int ret = bch2_trans_run(c, for_each_btree_key_commit(trans, iter, - BTREE_ID_subvolume_children, POS_MIN, BTREE_ITER_PREFETCH, k, + BTREE_ID_subvolume_children, POS_MIN, BTREE_ITER_prefetch, k, NULL, NULL, BCH_TRANS_COMMIT_no_enospc, check_subvol_child(trans, &iter, k))); bch_err_fn(c, ret); @@ -207,15 +224,24 @@ int bch2_check_subvol_children(struct bch_fs *c) /* Subvolumes: */ -int bch2_subvolume_invalid(struct bch_fs *c, struct bkey_s_c k, - enum bkey_invalid_flags flags, struct printbuf *err) +int bch2_subvolume_validate(struct bch_fs *c, struct bkey_s_c k, + struct bkey_validate_context from) { + struct bkey_s_c_subvolume subvol = bkey_s_c_to_subvolume(k); int ret = 0; bkey_fsck_err_on(bkey_lt(k.k->p, SUBVOL_POS_MIN) || - bkey_gt(k.k->p, SUBVOL_POS_MAX), c, err, - subvol_pos_bad, + bkey_gt(k.k->p, SUBVOL_POS_MAX), + c, subvol_pos_bad, "invalid pos"); + + bkey_fsck_err_on(!subvol.v->snapshot, + c, subvol_snapshot_bad, + "invalid snapshot"); + + bkey_fsck_err_on(!subvol.v->inode, + c, subvol_inode_bad, + "invalid inode"); fsck_err: return ret; } @@ -245,9 +271,9 @@ static int subvolume_children_mod(struct btree_trans *trans, struct bpos pos, bo int bch2_subvolume_trigger(struct btree_trans *trans, enum btree_id btree_id, unsigned level, struct bkey_s_c old, struct bkey_s new, - unsigned flags) + enum btree_iter_update_trigger_flags flags) { - if (flags & BTREE_TRIGGER_TRANSACTIONAL) { + if (flags & BTREE_TRIGGER_transactional) { struct bpos children_pos_old = subvolume_children_pos(old); struct bpos children_pos_new = subvolume_children_pos(new.s_c); @@ -267,7 +293,7 @@ int bch2_subvol_has_children(struct btree_trans *trans, u32 subvol) struct btree_iter iter; bch2_trans_iter_init(trans, &iter, BTREE_ID_subvolume_children, POS(subvol, 0), 0); - struct bkey_s_c k = bch2_btree_iter_peek(&iter); + struct bkey_s_c k = bch2_btree_iter_peek(trans, &iter); bch2_trans_iter_exit(trans, &iter); return bkey_err(k) ?: k.k && k.k->p.inode == subvol @@ -278,29 +304,27 @@ int bch2_subvol_has_children(struct btree_trans *trans, u32 subvol) static __always_inline int bch2_subvolume_get_inlined(struct btree_trans *trans, unsigned subvol, bool inconsistent_if_not_found, - int iter_flags, struct bch_subvolume *s) { int ret = bch2_bkey_get_val_typed(trans, BTREE_ID_subvolumes, POS(0, subvol), - iter_flags, subvolume, s); - bch2_fs_inconsistent_on(bch2_err_matches(ret, ENOENT) && - inconsistent_if_not_found, - trans->c, "missing subvolume %u", subvol); + BTREE_ITER_cached| + BTREE_ITER_with_updates, subvolume, s); + if (bch2_err_matches(ret, ENOENT) && inconsistent_if_not_found) + ret = bch2_subvolume_missing(trans->c, subvol) ?: ret; return ret; } int bch2_subvolume_get(struct btree_trans *trans, unsigned subvol, bool inconsistent_if_not_found, - int iter_flags, struct bch_subvolume *s) { - return bch2_subvolume_get_inlined(trans, subvol, inconsistent_if_not_found, iter_flags, s); + return bch2_subvolume_get_inlined(trans, subvol, inconsistent_if_not_found, s); } int bch2_subvol_is_ro_trans(struct btree_trans *trans, u32 subvol) { struct bch_subvolume s; - int ret = bch2_subvolume_get_inlined(trans, subvol, true, 0, &s); + int ret = bch2_subvolume_get_inlined(trans, subvol, true, &s); if (ret) return ret; @@ -311,8 +335,7 @@ int bch2_subvol_is_ro_trans(struct btree_trans *trans, u32 subvol) int bch2_subvol_is_ro(struct bch_fs *c, u32 subvol) { - return bch2_trans_do(c, NULL, NULL, 0, - bch2_subvol_is_ro_trans(trans, subvol)); + return bch2_trans_do(c, bch2_subvol_is_ro_trans(trans, subvol)); } int bch2_snapshot_get_subvol(struct btree_trans *trans, u32 snapshot, @@ -321,11 +344,11 @@ int bch2_snapshot_get_subvol(struct btree_trans *trans, u32 snapshot, struct bch_snapshot snap; return bch2_snapshot_lookup(trans, snapshot, &snap) ?: - bch2_subvolume_get(trans, le32_to_cpu(snap.subvol), true, 0, subvol); + bch2_subvolume_get(trans, le32_to_cpu(snap.subvol), true, subvol); } -int bch2_subvolume_get_snapshot(struct btree_trans *trans, u32 subvolid, - u32 *snapid) +int __bch2_subvolume_get_snapshot(struct btree_trans *trans, u32 subvolid, + u32 *snapid, bool warn) { struct btree_iter iter; struct bkey_s_c_subvolume subvol; @@ -333,11 +356,12 @@ int bch2_subvolume_get_snapshot(struct btree_trans *trans, u32 subvolid, subvol = bch2_bkey_get_iter_typed(trans, &iter, BTREE_ID_subvolumes, POS(0, subvolid), - BTREE_ITER_CACHED|BTREE_ITER_WITH_UPDATES, + BTREE_ITER_cached|BTREE_ITER_with_updates, subvolume); ret = bkey_err(subvol); - bch2_fs_inconsistent_on(bch2_err_matches(ret, ENOENT), trans->c, - "missing subvolume %u", subvolid); + + if (bch2_err_matches(ret, ENOENT)) + ret = bch2_subvolume_missing(trans->c, subvolid) ?: ret; if (likely(!ret)) *snapid = le32_to_cpu(subvol.v->snapshot); @@ -345,6 +369,12 @@ int bch2_subvolume_get_snapshot(struct btree_trans *trans, u32 subvolid, return ret; } +int bch2_subvolume_get_snapshot(struct btree_trans *trans, u32 subvolid, + u32 *snapid) +{ + return __bch2_subvolume_get_snapshot(trans, subvolid, snapid, true); +} + static int bch2_subvolume_reparent(struct btree_trans *trans, struct btree_iter *iter, struct bkey_s_c k, @@ -382,10 +412,9 @@ static int bch2_subvolumes_reparent(struct btree_trans *trans, u32 subvolid_to_d struct bch_subvolume s; return lockrestart_do(trans, - bch2_subvolume_get(trans, subvolid_to_delete, true, - BTREE_ITER_CACHED, &s)) ?: + bch2_subvolume_get(trans, subvolid_to_delete, true, &s)) ?: for_each_btree_key_commit(trans, iter, - BTREE_ID_subvolumes, POS_MIN, BTREE_ITER_PREFETCH, k, + BTREE_ID_subvolumes, POS_MIN, BTREE_ITER_prefetch, k, NULL, NULL, BCH_TRANS_COMMIT_no_enospc, bch2_subvolume_reparent(trans, &iter, k, subvolid_to_delete, le32_to_cpu(s.creation_parent))); @@ -397,26 +426,61 @@ static int bch2_subvolumes_reparent(struct btree_trans *trans, u32 subvolid_to_d */ static int __bch2_subvolume_delete(struct btree_trans *trans, u32 subvolid) { - struct btree_iter iter; - struct bkey_s_c_subvolume subvol; - u32 snapid; - int ret = 0; + struct btree_iter subvol_iter = {}, snapshot_iter = {}, snapshot_tree_iter = {}; - subvol = bch2_bkey_get_iter_typed(trans, &iter, + struct bkey_s_c_subvolume subvol = + bch2_bkey_get_iter_typed(trans, &subvol_iter, BTREE_ID_subvolumes, POS(0, subvolid), - BTREE_ITER_CACHED|BTREE_ITER_INTENT, + BTREE_ITER_cached|BTREE_ITER_intent, subvolume); - ret = bkey_err(subvol); + int ret = bkey_err(subvol); + if (bch2_err_matches(ret, ENOENT)) + ret = bch2_subvolume_missing(trans->c, subvolid) ?: ret; + if (ret) + goto err; + + u32 snapid = le32_to_cpu(subvol.v->snapshot); + + struct bkey_s_c_snapshot snapshot = + bch2_bkey_get_iter_typed(trans, &snapshot_iter, + BTREE_ID_snapshots, POS(0, snapid), + 0, snapshot); + ret = bkey_err(snapshot); bch2_fs_inconsistent_on(bch2_err_matches(ret, ENOENT), trans->c, - "missing subvolume %u", subvolid); + "missing snapshot %u", snapid); if (ret) - return ret; + goto err; - snapid = le32_to_cpu(subvol.v->snapshot); + u32 treeid = le32_to_cpu(snapshot.v->tree); + + struct bkey_s_c_snapshot_tree snapshot_tree = + bch2_bkey_get_iter_typed(trans, &snapshot_tree_iter, + BTREE_ID_snapshot_trees, POS(0, treeid), + 0, snapshot_tree); + ret = bkey_err(snapshot_tree); + bch2_fs_inconsistent_on(bch2_err_matches(ret, ENOENT), trans->c, + "missing snapshot tree %u", treeid); + if (ret) + goto err; + + if (le32_to_cpu(snapshot_tree.v->master_subvol) == subvolid) { + struct bkey_i_snapshot_tree *snapshot_tree_mut = + bch2_bkey_make_mut_typed(trans, &snapshot_tree_iter, + &snapshot_tree.s_c, + 0, snapshot_tree); + ret = PTR_ERR_OR_ZERO(snapshot_tree_mut); + if (ret) + goto err; - ret = bch2_btree_delete_at(trans, &iter, 0) ?: + snapshot_tree_mut->v.master_subvol = 0; + } + + ret = bch2_btree_delete_at(trans, &subvol_iter, 0) ?: bch2_snapshot_node_set_deleted(trans, snapid); - bch2_trans_iter_exit(trans, &iter); +err: + bch2_trans_iter_exit(trans, &snapshot_tree_iter); + bch2_trans_iter_exit(trans, &snapshot_iter); + bch2_trans_iter_exit(trans, &subvol_iter); return ret; } @@ -431,13 +495,11 @@ static void bch2_subvolume_wait_for_pagecache_and_delete(struct work_struct *wor { struct bch_fs *c = container_of(work, struct bch_fs, snapshot_wait_for_pagecache_and_delete_work); - snapshot_id_list s; - u32 *id; int ret = 0; while (!ret) { mutex_lock(&c->snapshots_unlinked_lock); - s = c->snapshots_unlinked; + snapshot_id_list s = c->snapshots_unlinked; darray_init(&c->snapshots_unlinked); mutex_unlock(&c->snapshots_unlinked_lock); @@ -446,7 +508,7 @@ static void bch2_subvolume_wait_for_pagecache_and_delete(struct work_struct *wor bch2_evict_subvolume_inodes(c, &s); - for (id = s.data; id < s.data + s.nr; id++) { + darray_for_each(s, id) { ret = bch2_trans_run(c, bch2_subvolume_delete(trans, *id)); bch_err_msg(c, ret, "deleting subvolume %u", *id); if (ret) @@ -456,7 +518,7 @@ static void bch2_subvolume_wait_for_pagecache_and_delete(struct work_struct *wor darray_exit(&s); } - bch2_write_ref_put(c, BCH_WRITE_REF_snapshot_delete_pagecache); + enumerated_ref_put(&c->writes, BCH_WRITE_REF_snapshot_delete_pagecache); } struct subvolume_unlink_hook { @@ -479,11 +541,11 @@ static int bch2_subvolume_wait_for_pagecache_and_delete_hook(struct btree_trans if (ret) return ret; - if (!bch2_write_ref_tryget(c, BCH_WRITE_REF_snapshot_delete_pagecache)) + if (!enumerated_ref_tryget(&c->writes, BCH_WRITE_REF_snapshot_delete_pagecache)) return -EROFS; if (!queue_work(c->write_ref_wq, &c->snapshot_wait_for_pagecache_and_delete_work)) - bch2_write_ref_put(c, BCH_WRITE_REF_snapshot_delete_pagecache); + enumerated_ref_put(&c->writes, BCH_WRITE_REF_snapshot_delete_pagecache); return 0; } @@ -505,15 +567,15 @@ int bch2_subvolume_unlink(struct btree_trans *trans, u32 subvolid) n = bch2_bkey_get_mut_typed(trans, &iter, BTREE_ID_subvolumes, POS(0, subvolid), - BTREE_ITER_CACHED, subvolume); + BTREE_ITER_cached, subvolume); ret = PTR_ERR_OR_ZERO(n); - if (unlikely(ret)) { - bch2_fs_inconsistent_on(bch2_err_matches(ret, ENOENT), trans->c, - "missing subvolume %u", subvolid); + if (bch2_err_matches(ret, ENOENT)) + ret = bch2_subvolume_missing(trans->c, subvolid) ?: ret; + if (unlikely(ret)) return ret; - } SET_BCH_SUBVOLUME_UNLINKED(&n->v, true); + n->v.fs_path_parent = 0; bch2_trans_iter_exit(trans, &iter); return ret; } @@ -526,7 +588,7 @@ int bch2_subvolume_create(struct btree_trans *trans, u64 inode, bool ro) { struct bch_fs *c = trans->c; - struct btree_iter dst_iter, src_iter = (struct btree_iter) { NULL }; + struct btree_iter dst_iter, src_iter = {}; struct bkey_i_subvolume *new_subvol = NULL; struct bkey_i_subvolume *src_subvol = NULL; u32 parent = 0, new_nodes[2], snapshot_subvols[2]; @@ -547,13 +609,12 @@ int bch2_subvolume_create(struct btree_trans *trans, u64 inode, src_subvol = bch2_bkey_get_mut_typed(trans, &src_iter, BTREE_ID_subvolumes, POS(0, src_subvolid), - BTREE_ITER_CACHED, subvolume); + BTREE_ITER_cached, subvolume); ret = PTR_ERR_OR_ZERO(src_subvol); - if (unlikely(ret)) { - bch2_fs_inconsistent_on(bch2_err_matches(ret, ENOENT), c, - "subvolume %u not found", src_subvolid); + if (bch2_err_matches(ret, ENOENT)) + ret = bch2_subvolume_missing(trans->c, src_subvolid) ?: ret; + if (unlikely(ret)) goto err; - } parent = le32_to_cpu(src_subvol->v.snapshot); } @@ -621,9 +682,9 @@ int bch2_initialize_subvolumes(struct bch_fs *c) root_volume.v.snapshot = cpu_to_le32(U32_MAX); root_volume.v.inode = cpu_to_le64(BCACHEFS_ROOT_INO); - ret = bch2_btree_insert(c, BTREE_ID_snapshot_trees, &root_tree.k_i, NULL, 0) ?: - bch2_btree_insert(c, BTREE_ID_snapshots, &root_snapshot.k_i, NULL, 0) ?: - bch2_btree_insert(c, BTREE_ID_subvolumes, &root_volume.k_i, NULL, 0); + ret = bch2_btree_insert(c, BTREE_ID_snapshot_trees, &root_tree.k_i, NULL, 0, 0) ?: + bch2_btree_insert(c, BTREE_ID_snapshots, &root_snapshot.k_i, NULL, 0, 0) ?: + bch2_btree_insert(c, BTREE_ID_subvolumes, &root_volume.k_i, NULL, 0, 0); bch_err_fn(c, ret); return ret; } @@ -661,17 +722,14 @@ err: /* set bi_subvol on root inode */ int bch2_fs_upgrade_for_subvolumes(struct bch_fs *c) { - int ret = bch2_trans_do(c, NULL, NULL, BCH_TRANS_COMMIT_lazy_rw, - __bch2_fs_upgrade_for_subvolumes(trans)); + int ret = bch2_trans_commit_do(c, NULL, NULL, BCH_TRANS_COMMIT_no_enospc, + __bch2_fs_upgrade_for_subvolumes(trans)); bch_err_fn(c, ret); return ret; } -int bch2_fs_subvolumes_init(struct bch_fs *c) +void bch2_fs_subvolumes_init_early(struct bch_fs *c) { - INIT_WORK(&c->snapshot_delete_work, bch2_delete_dead_snapshots_work); INIT_WORK(&c->snapshot_wait_for_pagecache_and_delete_work, bch2_subvolume_wait_for_pagecache_and_delete); - mutex_init(&c->snapshots_unlinked_lock); - return 0; } |