diff options
-rw-r--r-- | fs/bcachefs/alloc_background.c | 5 | ||||
-rw-r--r-- | fs/bcachefs/backpointers.c | 72 | ||||
-rw-r--r-- | fs/bcachefs/bcachefs_format.h | 11 | ||||
-rw-r--r-- | fs/bcachefs/bkey_methods.c | 6 | ||||
-rw-r--r-- | fs/bcachefs/bkey_types.h | 5 | ||||
-rw-r--r-- | fs/bcachefs/btree_cache.c | 2 | ||||
-rw-r--r-- | fs/bcachefs/btree_io.c | 2 | ||||
-rw-r--r-- | fs/bcachefs/btree_iter.c | 70 | ||||
-rw-r--r-- | fs/bcachefs/btree_key_cache.c | 1 | ||||
-rw-r--r-- | fs/bcachefs/btree_trans_commit.c | 10 | ||||
-rw-r--r-- | fs/bcachefs/btree_types.h | 32 | ||||
-rw-r--r-- | fs/bcachefs/btree_update.c | 45 | ||||
-rw-r--r-- | fs/bcachefs/dirent.c | 6 | ||||
-rw-r--r-- | fs/bcachefs/fast_list.h | 2 | ||||
-rw-r--r-- | fs/bcachefs/fs-io-pagecache.c | 2 | ||||
-rw-r--r-- | fs/bcachefs/fsck.c | 11 | ||||
-rw-r--r-- | fs/bcachefs/journal.c | 8 | ||||
-rw-r--r-- | fs/bcachefs/journal.h | 2 | ||||
-rw-r--r-- | fs/bcachefs/journal_io.c | 2 | ||||
-rw-r--r-- | fs/bcachefs/snapshot.c | 79 | ||||
-rw-r--r-- | fs/bcachefs/snapshot.h | 30 | ||||
-rw-r--r-- | fs/bcachefs/super-io.c | 4 |
22 files changed, 274 insertions, 133 deletions
diff --git a/fs/bcachefs/alloc_background.c b/fs/bcachefs/alloc_background.c index afc0ab75abf6..55c21e9b4c52 100644 --- a/fs/bcachefs/alloc_background.c +++ b/fs/bcachefs/alloc_background.c @@ -1554,6 +1554,9 @@ int bch2_check_alloc_info(struct bch_fs *c) struct bkey_s_c k; int ret = 0; + struct progress_indicator_state progress; + bch2_progress_init(&progress, c, BIT_ULL(BTREE_ID_alloc)); + CLASS(btree_trans, trans)(c); bch2_trans_iter_init(trans, &iter, BTREE_ID_alloc, POS_MIN, BTREE_ITER_prefetch); @@ -1577,6 +1580,8 @@ int bch2_check_alloc_info(struct bch_fs *c) if (!k.k) break; + progress_update_iter(trans, &progress, &iter); + if (k.k->type) { next = bpos_nosnap_successor(k.k->p); diff --git a/fs/bcachefs/backpointers.c b/fs/bcachefs/backpointers.c index bd26ab3e6812..afdfe991e4dd 100644 --- a/fs/bcachefs/backpointers.c +++ b/fs/bcachefs/backpointers.c @@ -872,13 +872,27 @@ static int check_bucket_backpointers_to_extents(struct btree_trans *, struct bch static int check_bucket_backpointer_mismatch(struct btree_trans *trans, struct bkey_s_c alloc_k, bool *had_mismatch, - struct bkey_buf *last_flushed) + struct bkey_buf *last_flushed, + struct bpos *last_pos, + unsigned *nr_iters) { struct bch_fs *c = trans->c; struct bch_alloc_v4 a_convert; const struct bch_alloc_v4 *a = bch2_alloc_to_v4(alloc_k, &a_convert); bool need_commit = false; + if (!bpos_eq(*last_pos, alloc_k.k->p)) + *nr_iters = 0; + + *last_pos = alloc_k.k->p; + *nr_iters += 1; + + if (*nr_iters > 2) { + bch_err(c, "%s: looping at %llu:%llu", __func__, + last_pos->inode, last_pos->offset); + return 0; + } + *had_mismatch = false; if (a->data_type == BCH_DATA_sb || @@ -937,19 +951,52 @@ static int check_bucket_backpointer_mismatch(struct btree_trans *trans, struct b if (sectors[ALLOC_dirty] != a->dirty_sectors || sectors[ALLOC_cached] != a->cached_sectors || sectors[ALLOC_stripe] != a->stripe_sectors) { - if (c->sb.version_upgrade_complete >= bcachefs_metadata_version_backpointer_bucket_gen) { - ret = bch2_backpointers_maybe_flush(trans, alloc_k, last_flushed); - if (ret) - return ret; - } - if (sectors[ALLOC_dirty] > a->dirty_sectors || sectors[ALLOC_cached] > a->cached_sectors || sectors[ALLOC_stripe] > a->stripe_sectors) { + if (*nr_iters > 1) { + ret = 0; + + CLASS(printbuf, buf)(); + bch2_log_msg_start(c, &buf); + + prt_printf(&buf, "backpointer sectors > bucket sectors, but found no bad backpointers\n" + "bucket %llu:%llu data type %s, counters\n", + alloc_k.k->p.inode, + alloc_k.k->p.offset, + __bch2_data_types[a->data_type]); + if (sectors[ALLOC_dirty] > a->dirty_sectors) + prt_printf(&buf, "dirty: %u > %u\n", + sectors[ALLOC_dirty], a->dirty_sectors); + if (sectors[ALLOC_cached] > a->cached_sectors) + prt_printf(&buf, "cached: %u > %u\n", + sectors[ALLOC_cached], a->cached_sectors); + if (sectors[ALLOC_stripe] > a->stripe_sectors) + prt_printf(&buf, "stripe: %u > %u\n", + sectors[ALLOC_stripe], a->stripe_sectors); + + for_each_btree_key_max_norestart(trans, iter, BTREE_ID_backpointers, + bucket_pos_to_bp_start(ca, alloc_k.k->p), + bucket_pos_to_bp_end(ca, alloc_k.k->p), 0, bp_k, ret) { + bch2_bkey_val_to_text(&buf, c, bp_k); + prt_newline(&buf); + } + bch2_trans_iter_exit(trans, &iter); + + bch2_print_str(c, KERN_ERR, buf.buf); + return ret; + } + return check_bucket_backpointers_to_extents(trans, ca, alloc_k.k->p) ?: bch_err_throw(c, transaction_restart_nested); } + if (c->sb.version_upgrade_complete >= bcachefs_metadata_version_backpointer_bucket_gen) { + ret = bch2_backpointers_maybe_flush(trans, alloc_k, last_flushed); + if (ret) + return ret; + } + bool empty = (sectors[ALLOC_dirty] + sectors[ALLOC_stripe] + sectors[ALLOC_cached]) == 0; @@ -1101,6 +1148,8 @@ int bch2_check_extents_to_backpointers(struct bch_fs *c) CLASS(btree_trans, trans)(c); struct extents_to_bp_state s = { .bp_start = POS_MIN }; + struct bpos last_pos = POS_MIN; + unsigned nr_iters = 0; bch2_bkey_buf_init(&s.last_flushed); bkey_init(&s.last_flushed.k->k); @@ -1109,7 +1158,8 @@ int bch2_check_extents_to_backpointers(struct bch_fs *c) POS_MIN, BTREE_ITER_prefetch, k, ({ bool had_mismatch; bch2_fs_going_ro(c) ?: - check_bucket_backpointer_mismatch(trans, k, &had_mismatch, &s.last_flushed); + check_bucket_backpointer_mismatch(trans, k, &had_mismatch, &s.last_flushed, + &last_pos, &nr_iters); })); if (ret) goto err; @@ -1178,7 +1228,11 @@ static int check_bucket_backpointer_pos_mismatch(struct btree_trans *trans, if (ret) return ret; - ret = check_bucket_backpointer_mismatch(trans, k, had_mismatch, last_flushed); + struct bpos last_pos = POS_MIN; + unsigned nr_iters = 0; + ret = check_bucket_backpointer_mismatch(trans, k, had_mismatch, + last_flushed, + &last_pos, &nr_iters); bch2_trans_iter_exit(trans, &alloc_iter); return ret; } diff --git a/fs/bcachefs/bcachefs_format.h b/fs/bcachefs/bcachefs_format.h index b4a04df5ea95..a8f59522e258 100644 --- a/fs/bcachefs/bcachefs_format.h +++ b/fs/bcachefs/bcachefs_format.h @@ -423,7 +423,8 @@ enum bch_bkey_type_flags { x(logged_op_truncate, 32, BKEY_TYPE_strict_btree_checks) \ x(logged_op_finsert, 33, BKEY_TYPE_strict_btree_checks) \ x(accounting, 34, BKEY_TYPE_strict_btree_checks) \ - x(inode_alloc_cursor, 35, BKEY_TYPE_strict_btree_checks) + x(inode_alloc_cursor, 35, BKEY_TYPE_strict_btree_checks) \ + x(extent_whiteout, 36, BKEY_TYPE_strict_btree_checks) enum bch_bkey_type { #define x(name, nr, ...) KEY_TYPE_##name = nr, @@ -440,6 +441,10 @@ struct bch_whiteout { struct bch_val v; }; +struct bch_extent_whiteout { + struct bch_val v; +}; + struct bch_error { struct bch_val v; }; @@ -700,7 +705,8 @@ struct bch_sb_field_ext { x(extent_flags, BCH_VERSION(1, 25)) \ x(snapshot_deletion_v2, BCH_VERSION(1, 26)) \ x(fast_device_removal, BCH_VERSION(1, 27)) \ - x(inode_has_case_insensitive, BCH_VERSION(1, 28)) + x(inode_has_case_insensitive, BCH_VERSION(1, 28)) \ + x(extent_snapshot_whiteouts, BCH_VERSION(1, 29)) enum bcachefs_metadata_version { bcachefs_metadata_version_min = 9, @@ -1340,6 +1346,7 @@ enum btree_id_flags { BTREE_IS_snapshots| \ BTREE_IS_data, \ BIT_ULL(KEY_TYPE_whiteout)| \ + BIT_ULL(KEY_TYPE_extent_whiteout)| \ BIT_ULL(KEY_TYPE_error)| \ BIT_ULL(KEY_TYPE_cookie)| \ BIT_ULL(KEY_TYPE_extent)| \ diff --git a/fs/bcachefs/bkey_methods.c b/fs/bcachefs/bkey_methods.c index fcd8c82cba4f..75d73677c4d8 100644 --- a/fs/bcachefs/bkey_methods.c +++ b/fs/bcachefs/bkey_methods.c @@ -41,6 +41,10 @@ static int deleted_key_validate(struct bch_fs *c, struct bkey_s_c k, .key_validate = deleted_key_validate, \ }) +#define bch2_bkey_ops_extent_whiteout ((struct bkey_ops) { \ + .key_validate = deleted_key_validate, \ +}) + static int empty_val_key_validate(struct bch_fs *c, struct bkey_s_c k, struct bkey_validate_context from) { @@ -203,7 +207,7 @@ int __bch2_bkey_validate(struct bch_fs *c, struct bkey_s_c k, ? bch2_bkey_types[k.k->type] : "(unknown)"); - if (btree_node_type_is_extents(type) && !bkey_whiteout(k.k)) { + if (btree_node_type_is_extents(type) && !bkey_extent_whiteout(k.k)) { bkey_fsck_err_on(k.k->size == 0, c, bkey_extent_size_zero, "size == 0"); diff --git a/fs/bcachefs/bkey_types.h b/fs/bcachefs/bkey_types.h index b4f328f9853c..88a48ce63656 100644 --- a/fs/bcachefs/bkey_types.h +++ b/fs/bcachefs/bkey_types.h @@ -44,6 +44,11 @@ static inline void set_bkey_val_bytes(struct bkey *k, unsigned bytes) #define bkey_whiteout(_k) \ ((_k)->type == KEY_TYPE_deleted || (_k)->type == KEY_TYPE_whiteout) +#define bkey_extent_whiteout(_k) \ + ((_k)->type == KEY_TYPE_deleted || \ + (_k)->type == KEY_TYPE_whiteout || \ + (_k)->type == KEY_TYPE_extent_whiteout) + /* bkey with split value, const */ struct bkey_s_c { const struct bkey *k; diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c index 23ed7393f07f..25b01e750880 100644 --- a/fs/bcachefs/btree_cache.c +++ b/fs/bcachefs/btree_cache.c @@ -511,7 +511,7 @@ restart: if (btree_node_accessed(b)) { clear_btree_node_accessed(b); bc->not_freed[BCH_BTREE_CACHE_NOT_FREED_access_bit]++; - --touched;; + --touched; } else if (!btree_node_reclaim(c, b)) { __bch2_btree_node_hash_remove(bc, b); __btree_node_data_free(b); diff --git a/fs/bcachefs/btree_io.c b/fs/bcachefs/btree_io.c index 83c836080ea2..8a03cd75a64f 100644 --- a/fs/bcachefs/btree_io.c +++ b/fs/bcachefs/btree_io.c @@ -2011,7 +2011,7 @@ static void btree_node_scrub_work(struct work_struct *work) bch_err_fn_ratelimited(c, ret); } - bch2_bkey_buf_exit(&scrub->key, c);; + bch2_bkey_buf_exit(&scrub->key, c); btree_bounce_free(c, c->opts.btree_node_size, scrub->used_mempool, scrub->buf); enumerated_ref_put(&scrub->ca->io_ref[READ], BCH_DEV_READ_REF_btree_node_scrub); kfree(scrub); diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c index a282c3886168..45ddfb54188a 100644 --- a/fs/bcachefs/btree_iter.c +++ b/fs/bcachefs/btree_iter.c @@ -2448,10 +2448,27 @@ struct bkey_s_c bch2_btree_iter_peek_max(struct btree_trans *trans, struct btree continue; } - if (bkey_whiteout(k.k) && - !(iter->flags & BTREE_ITER_key_cache_fill)) { - search_key = bkey_successor(iter, k.k->p); - continue; + if (!(iter->flags & BTREE_ITER_nofilter_whiteouts)) { + /* + * KEY_TYPE_extent_whiteout indicates that there + * are no extents that overlap with this + * whiteout - meaning bkey_start_pos() is + * monotonically increasing when including + * KEY_TYPE_extent_whiteout (not + * KEY_TYPE_whiteout). + * + * Without this @end wouldn't be able to + * terminate searches and we'd have to scan + * through tons of whiteouts: + */ + if (k.k->type == KEY_TYPE_extent_whiteout && + bkey_ge(k.k->p, end)) + goto end; + + if (bkey_extent_whiteout(k.k)) { + search_key = bkey_successor(iter, k.k->p); + continue; + } } } @@ -2711,7 +2728,7 @@ struct bkey_s_c bch2_btree_iter_peek_prev_min(struct btree_trans *trans, struct saved_path = 0; } - if (!bkey_whiteout(k.k)) { + if (!bkey_extent_whiteout(k.k)) { saved_path = btree_path_clone(trans, iter->path, iter->flags & BTREE_ITER_intent, _THIS_IP_); @@ -2724,7 +2741,7 @@ struct bkey_s_c bch2_btree_iter_peek_prev_min(struct btree_trans *trans, struct continue; } - if (bkey_whiteout(k.k)) { + if (bkey_extent_whiteout(k.k)) { search_key = bkey_predecessor(iter, k.k->p); search_key.snapshot = U32_MAX; continue; @@ -2744,7 +2761,7 @@ struct bkey_s_c bch2_btree_iter_peek_prev_min(struct btree_trans *trans, struct } /* Extents can straddle iter->pos: */ - iter->pos = bpos_min(iter->pos, k.k->p);; + iter->pos = bpos_min(iter->pos, k.k->p); if (iter->flags & BTREE_ITER_filter_snapshots) iter->pos.snapshot = iter->snapshot; @@ -2865,9 +2882,9 @@ struct bkey_s_c bch2_btree_iter_peek_slot(struct btree_trans *trans, struct btre iter->k = *k.k; } - if (unlikely(k.k->type == KEY_TYPE_whiteout && + if (unlikely(bkey_extent_whiteout(k.k) && (iter->flags & BTREE_ITER_filter_snapshots) && - !(iter->flags & BTREE_ITER_key_cache_fill))) + !(iter->flags & BTREE_ITER_nofilter_whiteouts))) iter->k.type = KEY_TYPE_deleted; } else { struct bpos next; @@ -2878,27 +2895,30 @@ struct bkey_s_c bch2_btree_iter_peek_slot(struct btree_trans *trans, struct btre EBUG_ON(btree_iter_path(trans, iter)->level); - if (iter->flags & BTREE_ITER_intent) { - struct btree_iter iter2; + struct btree_iter iter2; - bch2_trans_copy_iter(trans, &iter2, iter); - k = bch2_btree_iter_peek_max(trans, &iter2, end); + bch2_trans_copy_iter(trans, &iter2, iter); + iter2.flags |= BTREE_ITER_nofilter_whiteouts; - if (k.k && !bkey_err(k)) { - swap(iter->key_cache_path, iter2.key_cache_path); - iter->k = iter2.k; - k.k = &iter->k; + while (1) { + k = bch2_btree_iter_peek_max(trans, &iter2, end); + if ((iter2.flags & BTREE_ITER_is_extents) && + k.k && + !bkey_err(k) && + k.k->type == KEY_TYPE_whiteout) { + bch2_btree_iter_set_pos(trans, &iter2, k.k->p); + continue; } - bch2_trans_iter_exit(trans, &iter2); - } else { - struct bpos pos = iter->pos; - k = bch2_btree_iter_peek_max(trans, iter, end); - if (unlikely(bkey_err(k))) - bch2_btree_iter_set_pos(trans, iter, pos); - else - iter->pos = pos; + break; + } + + if (k.k && !bkey_err(k)) { + swap(iter->key_cache_path, iter2.key_cache_path); + iter->k = iter2.k; + k.k = &iter->k; } + bch2_trans_iter_exit(trans, &iter2); if (unlikely(bkey_err(k))) goto out; diff --git a/fs/bcachefs/btree_key_cache.c b/fs/bcachefs/btree_key_cache.c index d61b782087ce..2317fe5b3c23 100644 --- a/fs/bcachefs/btree_key_cache.c +++ b/fs/bcachefs/btree_key_cache.c @@ -329,6 +329,7 @@ static noinline int btree_key_cache_fill(struct btree_trans *trans, bch2_trans_iter_init(trans, &iter, ck_path->btree_id, ck_path->pos, BTREE_ITER_intent| + BTREE_ITER_nofilter_whiteouts| BTREE_ITER_key_cache_fill| BTREE_ITER_cached_nofill); iter.flags &= ~BTREE_ITER_with_journal; diff --git a/fs/bcachefs/btree_trans_commit.c b/fs/bcachefs/btree_trans_commit.c index 58590ccc26bd..8b94a8156fbf 100644 --- a/fs/bcachefs/btree_trans_commit.c +++ b/fs/bcachefs/btree_trans_commit.c @@ -969,7 +969,7 @@ do_bch2_trans_commit_to_journal_replay(struct btree_trans *trans, BUG_ON(current != c->recovery_task); struct bkey_i *accounting; - +retry: percpu_down_read(&c->mark_lock); for (accounting = btree_trans_subbuf_base(trans, &trans->accounting); accounting != btree_trans_subbuf_top(trans, &trans->accounting); @@ -1025,13 +1025,17 @@ fatal_err: bch2_fs_fatal_error(c, "fatal error in transaction commit: %s", bch2_err_str(ret)); percpu_down_read(&c->mark_lock); revert_fs_usage: - BUG(); - /* error path not handled by __bch2_trans_commit() */ for (struct bkey_i *i = btree_trans_subbuf_base(trans, &trans->accounting); i != accounting; i = bkey_next(i)) bch2_accounting_trans_commit_revert(trans, bkey_i_to_accounting(i), flags); percpu_up_read(&c->mark_lock); + + if (bch2_err_matches(ret, BCH_ERR_btree_insert_need_mark_replicas)) { + ret = drop_locks_do(trans, bch2_accounting_update_sb(trans)); + if (!ret) + goto retry; + } return ret; } diff --git a/fs/bcachefs/btree_types.h b/fs/bcachefs/btree_types.h index 76adf75617aa..1d5611bc9a50 100644 --- a/fs/bcachefs/btree_types.h +++ b/fs/bcachefs/btree_types.h @@ -229,6 +229,7 @@ struct btree_node_iter { x(snapshot_field) \ x(all_snapshots) \ x(filter_snapshots) \ + x(nofilter_whiteouts) \ x(nopreserve) \ x(cached_nofill) \ x(key_cache_fill) \ @@ -485,7 +486,7 @@ typedef DARRAY(struct trans_kmalloc_trace) darray_trans_kmalloc_trace; struct btree_trans_subbuf { u16 base; u16 u64s; - u16 size;; + u16 size; }; struct btree_trans { @@ -838,15 +839,15 @@ static inline bool btree_node_type_has_triggers(enum btree_node_type type) return BIT_ULL(type) & BTREE_NODE_TYPE_HAS_TRIGGERS; } -static inline bool btree_id_is_extents(enum btree_id btree) -{ - const u64 mask = 0 +static const u64 btree_is_extents_mask = 0 #define x(name, nr, flags, ...) |((!!((flags) & BTREE_IS_extents)) << nr) - BCH_BTREE_IDS() +BCH_BTREE_IDS() #undef x - ; +; - return BIT_ULL(btree) & mask; +static inline bool btree_id_is_extents(enum btree_id btree) +{ + return BIT_ULL(btree) & btree_is_extents_mask; } static inline bool btree_node_type_is_extents(enum btree_node_type type) @@ -854,15 +855,20 @@ static inline bool btree_node_type_is_extents(enum btree_node_type type) return type != BKEY_TYPE_btree && btree_id_is_extents(type - 1); } -static inline bool btree_type_has_snapshots(enum btree_id btree) -{ - const u64 mask = 0 +static const u64 btree_has_snapshots_mask = 0 #define x(name, nr, flags, ...) |((!!((flags) & BTREE_IS_snapshots)) << nr) - BCH_BTREE_IDS() +BCH_BTREE_IDS() #undef x - ; +; - return BIT_ULL(btree) & mask; +static inline bool btree_type_has_snapshots(enum btree_id btree) +{ + return BIT_ULL(btree) & btree_has_snapshots_mask; +} + +static inline bool btree_id_is_extents_snapshots(enum btree_id btree) +{ + return BIT_ULL(btree) & btree_has_snapshots_mask & btree_is_extents_mask; } static inline bool btree_type_has_snapshot_field(enum btree_id btree) diff --git a/fs/bcachefs/btree_update.c b/fs/bcachefs/btree_update.c index f514a8ad7a89..ff7b53c6b0fb 100644 --- a/fs/bcachefs/btree_update.c +++ b/fs/bcachefs/btree_update.c @@ -164,6 +164,19 @@ int __bch2_insert_snapshot_whiteouts(struct btree_trans *trans, return ret; } +static inline enum bch_bkey_type extent_whiteout_type(enum btree_id btree, const struct bkey *k) +{ + /* + * KEY_TYPE_extent_whiteout indicates that there isn't a real extent + * present at that position: key start positions inclusive of + * KEY_TYPE_extent_whiteout (but not KEY_TYPE_whiteout) are + * monotonically increasing + */ + return btree_id_is_extents_snapshots(btree) && bkey_deleted(k) + ? KEY_TYPE_extent_whiteout + : KEY_TYPE_whiteout; +} + int bch2_trans_update_extent_overwrite(struct btree_trans *trans, struct btree_iter *iter, enum btree_iter_update_trigger_flags flags, @@ -231,13 +244,13 @@ int bch2_trans_update_extent_overwrite(struct btree_trans *trans, update->k.p.snapshot = new.k->p.snapshot; if (new.k->p.snapshot != old.k->p.snapshot) { - update->k.type = KEY_TYPE_whiteout; + update->k.type = extent_whiteout_type(iter->btree_id, new.k); } else if (btree_type_has_snapshots(btree_id)) { ret = need_whiteout_for_snapshot(trans, btree_id, update->k.p); if (ret < 0) return ret; if (ret) - update->k.type = KEY_TYPE_whiteout; + update->k.type = extent_whiteout_type(iter->btree_id, new.k); } ret = bch2_btree_insert_nonextent(trans, btree_id, update, @@ -276,7 +289,8 @@ static int bch2_trans_update_extent(struct btree_trans *trans, bch2_trans_iter_init(trans, &iter, btree_id, bkey_start_pos(&insert->k), BTREE_ITER_intent| BTREE_ITER_with_updates| - BTREE_ITER_not_extents); + BTREE_ITER_not_extents| + BTREE_ITER_nofilter_whiteouts); k = bch2_btree_iter_peek_max(trans, &iter, POS(insert->k.p.inode, U64_MAX)); if ((ret = bkey_err(k))) goto err; @@ -293,7 +307,28 @@ static int bch2_trans_update_extent(struct btree_trans *trans, goto next; } - while (bkey_gt(insert->k.p, bkey_start_pos(k.k))) { + while (true) { + if (k.k->type == KEY_TYPE_whiteout && + bkey_lt(insert->k.p, k.k->p)) + continue; + + if (bkey_le(insert->k.p, bkey_start_pos(k.k))) + break; + + if (bkey_extent_whiteout(k.k)) { + enum bch_bkey_type whiteout_type = extent_whiteout_type(btree_id, &insert->k); + if (k.k->type != whiteout_type) { + struct bkey_i *update = bch2_bkey_make_mut(trans, &iter, &k, 0); + ret = PTR_ERR_OR_ZERO(update); + if (ret) + goto err; + + update->k.type = whiteout_type; + } + + goto next; + } + bool done = bkey_lt(insert->k.p, k.k->p); ret = bch2_trans_update_extent_overwrite(trans, &iter, flags, k, bkey_i_to_s_c(insert)); @@ -525,7 +560,7 @@ int __must_check bch2_trans_update_ip(struct btree_trans *trans, struct btree_it return ret; if (ret) - k->k.type = KEY_TYPE_whiteout; + k->k.type = extent_whiteout_type(iter->btree_id, &k->k); } /* diff --git a/fs/bcachefs/dirent.c b/fs/bcachefs/dirent.c index dd60c47528da..1b891ac43053 100644 --- a/fs/bcachefs/dirent.c +++ b/fs/bcachefs/dirent.c @@ -214,11 +214,13 @@ void bch2_dirent_to_text(struct printbuf *out, struct bch_fs *c, struct bkey_s_c struct bkey_s_c_dirent d = bkey_s_c_to_dirent(k); struct qstr d_name = bch2_dirent_get_name(d); - prt_printf(out, "%.*s", d_name.len, d_name.name); + prt_bytes(out, d_name.name, d_name.len); if (d.v->d_casefold) { + prt_str(out, " (casefold "); struct qstr d_name = bch2_dirent_get_lookup_name(d); - prt_printf(out, " (casefold %.*s)", d_name.len, d_name.name); + prt_bytes(out, d_name.name, d_name.len); + prt_char(out, ')'); } prt_str(out, " ->"); diff --git a/fs/bcachefs/fast_list.h b/fs/bcachefs/fast_list.h index 73c9bf591fd6..f67df3f72ee2 100644 --- a/fs/bcachefs/fast_list.h +++ b/fs/bcachefs/fast_list.h @@ -9,7 +9,7 @@ struct fast_list_pcpu; struct fast_list { GENRADIX(void *) items; - struct ida slots_allocated;; + struct ida slots_allocated; struct fast_list_pcpu __percpu *buffer; }; diff --git a/fs/bcachefs/fs-io-pagecache.c b/fs/bcachefs/fs-io-pagecache.c index 2a6705186c44..469492f6264a 100644 --- a/fs/bcachefs/fs-io-pagecache.c +++ b/fs/bcachefs/fs-io-pagecache.c @@ -635,6 +635,8 @@ vm_fault_t bch2_page_mkwrite(struct vm_fault *vmf) goto out; } + inode->ei_last_dirtied = (unsigned long) current; + bch2_set_folio_dirty(c, inode, folio, &res, offset, len); bch2_folio_reservation_put(c, inode, &res); diff --git a/fs/bcachefs/fsck.c b/fs/bcachefs/fsck.c index 40fc3c4e3c43..f5a9da40c647 100644 --- a/fs/bcachefs/fsck.c +++ b/fs/bcachefs/fsck.c @@ -1972,11 +1972,12 @@ static int check_extent(struct btree_trans *trans, struct btree_iter *iter, "extent type past end of inode %llu:%u, i_size %llu\n%s", i->inode.bi_inum, i->inode.bi_snapshot, i->inode.bi_size, (bch2_bkey_val_to_text(&buf, c, k), buf.buf))) { - ret = bch2_fpunch_snapshot(trans, - SPOS(i->inode.bi_inum, - last_block, - i->inode.bi_snapshot), - POS(i->inode.bi_inum, U64_MAX)); + ret = snapshots_seen_add_inorder(c, s, i->inode.bi_snapshot) ?: + bch2_fpunch_snapshot(trans, + SPOS(i->inode.bi_inum, + last_block, + i->inode.bi_snapshot), + POS(i->inode.bi_inum, U64_MAX)); if (ret) goto err; diff --git a/fs/bcachefs/journal.c b/fs/bcachefs/journal.c index f9e2e1a4e415..8c5b61b7bf4f 100644 --- a/fs/bcachefs/journal.c +++ b/fs/bcachefs/journal.c @@ -182,8 +182,6 @@ journal_error_check_stuck(struct journal *j, int error, unsigned flags) void bch2_journal_do_writes(struct journal *j) { - struct bch_fs *c = container_of(j, struct bch_fs, journal); - for (u64 seq = journal_last_unwritten_seq(j); seq <= journal_cur_seq(j); seq++) { @@ -198,7 +196,6 @@ void bch2_journal_do_writes(struct journal *j) if (!journal_state_seq_count(j, j->reservations, seq)) { j->seq_write_started = seq; w->write_started = true; - closure_get(&c->cl); closure_call(&w->io, bch2_journal_write, j->wq, NULL); } @@ -1060,14 +1057,13 @@ static struct journal_buf *__bch2_next_write_buffer_flush_journal_buf(struct jou if (open && !*blocked) { __bch2_journal_block(j); + s.v = atomic64_read_acquire(&j->reservations.counter); *blocked = true; } ret = journal_state_count(s, idx & JOURNAL_STATE_BUF_MASK) > open ? ERR_PTR(-EAGAIN) : buf; - if (!IS_ERR(ret)) - smp_mb(); break; } } @@ -1297,7 +1293,7 @@ int bch2_dev_journal_bucket_delete(struct bch_dev *ca, u64 b) return -EINVAL; } - u64 *new_buckets = kcalloc(ja->nr, sizeof(u64), GFP_KERNEL);; + u64 *new_buckets = kcalloc(ja->nr, sizeof(u64), GFP_KERNEL); if (!new_buckets) return bch_err_throw(c, ENOMEM_set_nr_journal_buckets); diff --git a/fs/bcachefs/journal.h b/fs/bcachefs/journal.h index c05aa94237f8..b46b9718d841 100644 --- a/fs/bcachefs/journal.h +++ b/fs/bcachefs/journal.h @@ -267,7 +267,7 @@ static inline union journal_res_state journal_state_buf_put(struct journal *j, u { union journal_res_state s; - s.v = atomic64_sub_return_release(((union journal_res_state) { + s.v = atomic64_sub_return(((union journal_res_state) { .buf0_count = idx == 0, .buf1_count = idx == 1, .buf2_count = idx == 2, diff --git a/fs/bcachefs/journal_io.c b/fs/bcachefs/journal_io.c index 093e4acad085..228701d9c745 100644 --- a/fs/bcachefs/journal_io.c +++ b/fs/bcachefs/journal_io.c @@ -1827,8 +1827,6 @@ static CLOSURE_CALLBACK(journal_write_done) if (do_discards) bch2_do_discards(c); - - closure_put(&c->cl); } static void journal_write_endio(struct bio *bio) diff --git a/fs/bcachefs/snapshot.c b/fs/bcachefs/snapshot.c index af3fae33b5ba..5370ccb85d2d 100644 --- a/fs/bcachefs/snapshot.c +++ b/fs/bcachefs/snapshot.c @@ -11,6 +11,7 @@ #include "errcode.h" #include "error.h" #include "fs.h" +#include "progress.h" #include "recovery_passes.h" #include "snapshot.h" @@ -364,31 +365,32 @@ int bch2_snapshot_lookup(struct btree_trans *trans, u32 id, /* fsck: */ -static u32 bch2_snapshot_child(struct bch_fs *c, u32 id, unsigned child) +static u32 bch2_snapshot_child(struct snapshot_table *t, + u32 id, unsigned child) { - return snapshot_t(c, id)->children[child]; + return __snapshot_t(t, id)->children[child]; } -static u32 bch2_snapshot_left_child(struct bch_fs *c, u32 id) +static u32 bch2_snapshot_left_child(struct snapshot_table *t, u32 id) { - return bch2_snapshot_child(c, id, 0); + return bch2_snapshot_child(t, id, 0); } -static u32 bch2_snapshot_right_child(struct bch_fs *c, u32 id) +static u32 bch2_snapshot_right_child(struct snapshot_table *t, u32 id) { - return bch2_snapshot_child(c, id, 1); + return bch2_snapshot_child(t, id, 1); } -static u32 bch2_snapshot_tree_next(struct bch_fs *c, u32 id) +static u32 bch2_snapshot_tree_next(struct snapshot_table *t, u32 id) { u32 n, parent; - n = bch2_snapshot_left_child(c, id); + n = bch2_snapshot_left_child(t, id); if (n) return n; - while ((parent = bch2_snapshot_parent(c, id))) { - n = bch2_snapshot_right_child(c, parent); + while ((parent = __bch2_snapshot_parent(t, id))) { + n = bch2_snapshot_right_child(t, parent); if (n && n != id) return n; id = parent; @@ -401,17 +403,18 @@ u32 bch2_snapshot_oldest_subvol(struct bch_fs *c, u32 snapshot_root, snapshot_id_list *skip) { guard(rcu)(); + struct snapshot_table *t = rcu_dereference(c->snapshots); u32 id, subvol = 0, s; retry: id = snapshot_root; - while (id && bch2_snapshot_exists(c, id)) { + while (id && __bch2_snapshot_exists(t, id)) { if (!(skip && snapshot_list_has_id(skip, id))) { - s = snapshot_t(c, id)->subvol; + s = __snapshot_t(t, id)->subvol; if (s && (!subvol || s < subvol)) subvol = s; } - id = bch2_snapshot_tree_next(c, id); + id = bch2_snapshot_tree_next(t, id); if (id == snapshot_root) break; } @@ -973,12 +976,16 @@ int bch2_reconstruct_snapshots(struct bch_fs *c) struct snapshot_tree_reconstruct r = {}; int ret = 0; + struct progress_indicator_state progress; + bch2_progress_init(&progress, c, btree_has_snapshots_mask); + for (unsigned btree = 0; btree < BTREE_ID_NR; btree++) { if (btree_type_has_snapshots(btree)) { r.btree = btree; ret = for_each_btree_key(trans, iter, btree, POS_MIN, BTREE_ITER_all_snapshots|BTREE_ITER_prefetch, k, ({ + progress_update_iter(trans, &progress, &iter); get_snapshot_trees(c, &r, k.k->p); })); if (ret) @@ -1424,38 +1431,22 @@ static inline u32 interior_delete_has_id(interior_delete_list *l, u32 id) return i ? i->live_child : 0; } -static unsigned __live_child(struct snapshot_table *t, u32 id, - snapshot_id_list *delete_leaves, - interior_delete_list *delete_interior) -{ - struct snapshot_t *s = __snapshot_t(t, id); - if (!s) - return 0; - - for (unsigned i = 0; i < ARRAY_SIZE(s->children); i++) - if (s->children[i] && - !snapshot_list_has_id(delete_leaves, s->children[i]) && - !interior_delete_has_id(delete_interior, s->children[i])) - return s->children[i]; - - for (unsigned i = 0; i < ARRAY_SIZE(s->children); i++) { - u32 live_child = s->children[i] - ? __live_child(t, s->children[i], delete_leaves, delete_interior) - : 0; - if (live_child) - return live_child; - } - - return 0; -} - -static unsigned live_child(struct bch_fs *c, u32 id) +static unsigned live_child(struct bch_fs *c, u32 start) { struct snapshot_delete *d = &c->snapshot_delete; guard(rcu)(); - return __live_child(rcu_dereference(c->snapshots), id, - &d->delete_leaves, &d->delete_interior); + struct snapshot_table *t = rcu_dereference(c->snapshots); + + for (u32 id = bch2_snapshot_tree_next(t, start); + id && id != start; + id = bch2_snapshot_tree_next(t, id)) + if (bch2_snapshot_is_leaf(c, id) && + !snapshot_list_has_id(&d->delete_leaves, id) && + !interior_delete_has_id(&d->delete_interior, id)) + return id; + + return 0; } static bool snapshot_id_dying(struct snapshot_delete *d, unsigned id) @@ -1712,12 +1703,14 @@ static inline u32 bch2_snapshot_nth_parent_skip(struct bch_fs *c, u32 id, u32 n, interior_delete_list *skip) { guard(rcu)(); + struct snapshot_table *t = rcu_dereference(c->snapshots); + while (interior_delete_has_id(skip, id)) - id = __bch2_snapshot_parent(c, id); + id = __bch2_snapshot_parent(t, id); while (n--) { do { - id = __bch2_snapshot_parent(c, id); + id = __bch2_snapshot_parent(t, id); } while (interior_delete_has_id(skip, id)); } diff --git a/fs/bcachefs/snapshot.h b/fs/bcachefs/snapshot.h index 6dcb118b0fbd..fef32a0118c4 100644 --- a/fs/bcachefs/snapshot.h +++ b/fs/bcachefs/snapshot.h @@ -63,19 +63,19 @@ static inline u32 bch2_snapshot_parent_early(struct bch_fs *c, u32 id) return __bch2_snapshot_parent_early(c, id); } -static inline u32 __bch2_snapshot_parent(struct bch_fs *c, u32 id) +static inline u32 __bch2_snapshot_parent(struct snapshot_table *t, u32 id) { - const struct snapshot_t *s = snapshot_t(c, id); + const struct snapshot_t *s = __snapshot_t(t, id); if (!s) return 0; u32 parent = s->parent; if (IS_ENABLED(CONFIG_BCACHEFS_DEBUG) && parent && - s->depth != snapshot_t(c, parent)->depth + 1) + s->depth != __snapshot_t(t, parent)->depth + 1) panic("id %u depth=%u parent %u depth=%u\n", - id, snapshot_t(c, id)->depth, - parent, snapshot_t(c, parent)->depth); + id, __snapshot_t(t, id)->depth, + parent, __snapshot_t(t, parent)->depth); return parent; } @@ -83,14 +83,16 @@ static inline u32 __bch2_snapshot_parent(struct bch_fs *c, u32 id) static inline u32 bch2_snapshot_parent(struct bch_fs *c, u32 id) { guard(rcu)(); - return __bch2_snapshot_parent(c, id); + return __bch2_snapshot_parent(rcu_dereference(c->snapshots), id); } static inline u32 bch2_snapshot_nth_parent(struct bch_fs *c, u32 id, u32 n) { guard(rcu)(); + struct snapshot_table *t = rcu_dereference(c->snapshots); + while (n--) - id = __bch2_snapshot_parent(c, id); + id = __bch2_snapshot_parent(t, id); return id; } @@ -100,23 +102,29 @@ u32 bch2_snapshot_skiplist_get(struct bch_fs *, u32); static inline u32 bch2_snapshot_root(struct bch_fs *c, u32 id) { guard(rcu)(); + struct snapshot_table *t = rcu_dereference(c->snapshots); u32 parent; - while ((parent = __bch2_snapshot_parent(c, id))) + while ((parent = __bch2_snapshot_parent(t, id))) id = parent; return id; } -static inline enum snapshot_id_state __bch2_snapshot_id_state(struct bch_fs *c, u32 id) +static inline enum snapshot_id_state __bch2_snapshot_id_state(struct snapshot_table *t, u32 id) { - const struct snapshot_t *s = snapshot_t(c, id); + const struct snapshot_t *s = __snapshot_t(t, id); return s ? s->state : SNAPSHOT_ID_empty; } static inline enum snapshot_id_state bch2_snapshot_id_state(struct bch_fs *c, u32 id) { guard(rcu)(); - return __bch2_snapshot_id_state(c, id); + return __bch2_snapshot_id_state(rcu_dereference(c->snapshots), id); +} + +static inline bool __bch2_snapshot_exists(struct snapshot_table *t, u32 id) +{ + return __bch2_snapshot_id_state(t, id) == SNAPSHOT_ID_live; } static inline bool bch2_snapshot_exists(struct bch_fs *c, u32 id) diff --git a/fs/bcachefs/super-io.c b/fs/bcachefs/super-io.c index 40fa87ce1d09..c88759964575 100644 --- a/fs/bcachefs/super-io.c +++ b/fs/bcachefs/super-io.c @@ -79,7 +79,7 @@ int bch2_set_version_incompat(struct bch_fs *c, enum bcachefs_metadata_version v } else { darray_for_each(c->incompat_versions_requested, i) if (version == *i) - return -BCH_ERR_may_not_use_incompat_feature; + return bch_err_throw(c, may_not_use_incompat_feature); darray_push(&c->incompat_versions_requested, version); CLASS(printbuf, buf)(); @@ -90,7 +90,7 @@ int bch2_set_version_incompat(struct bch_fs *c, enum bcachefs_metadata_version v prt_printf(&buf, "\n set version_upgrade=incompat to enable"); bch_notice(c, "%s", buf.buf); - return -BCH_ERR_may_not_use_incompat_feature; + return bch_err_throw(c, may_not_use_incompat_feature); } } |