diff options
Diffstat (limited to 'fs')
39 files changed, 370 insertions, 205 deletions
diff --git a/fs/bcachefs/acl.c b/fs/bcachefs/acl.c index 307824d6eccb..8f970dc19dea 100644 --- a/fs/bcachefs/acl.c +++ b/fs/bcachefs/acl.c @@ -138,8 +138,8 @@ static struct posix_acl *bch2_acl_from_disk(struct btree_trans *trans, acl = allocate_dropping_locks(trans, ret, posix_acl_alloc(count, _gfp)); - if (!acl) - return ERR_PTR(-ENOMEM); + if (!acl && !ret) + ret = bch_err_throw(trans->c, ENOMEM_acl); if (ret) { kfree(acl); return ERR_PTR(ret); diff --git a/fs/bcachefs/alloc_background.c b/fs/bcachefs/alloc_background.c index 4c1604fd80f9..55c21e9b4c52 100644 --- a/fs/bcachefs/alloc_background.c +++ b/fs/bcachefs/alloc_background.c @@ -20,6 +20,7 @@ #include "enumerated_ref.h" #include "error.h" #include "lru.h" +#include "progress.h" #include "recovery.h" #include "varint.h" @@ -337,9 +338,10 @@ void bch2_alloc_v4_swab(struct bkey_s k) } static inline void __bch2_alloc_v4_to_text(struct printbuf *out, struct bch_fs *c, - unsigned dev, const struct bch_alloc_v4 *a) + struct bkey_s_c k, + const struct bch_alloc_v4 *a) { - struct bch_dev *ca = c ? bch2_dev_tryget_noerror(c, dev) : NULL; + struct bch_dev *ca = c ? bch2_dev_tryget_noerror(c, k.k->p.inode) : NULL; prt_newline(out); printbuf_indent_add(out, 2); @@ -348,11 +350,14 @@ static inline void __bch2_alloc_v4_to_text(struct printbuf *out, struct bch_fs * bch2_prt_data_type(out, a->data_type); prt_newline(out); prt_printf(out, "journal_seq_nonempty %llu\n", a->journal_seq_nonempty); - prt_printf(out, "journal_seq_empty %llu\n", a->journal_seq_empty); + if (bkey_val_bytes(k.k) > offsetof(struct bch_alloc_v4, journal_seq_empty)) + prt_printf(out, "journal_seq_empty %llu\n", a->journal_seq_empty); + prt_printf(out, "need_discard %llu\n", BCH_ALLOC_V4_NEED_DISCARD(a)); prt_printf(out, "need_inc_gen %llu\n", BCH_ALLOC_V4_NEED_INC_GEN(a)); prt_printf(out, "dirty_sectors %u\n", a->dirty_sectors); - prt_printf(out, "stripe_sectors %u\n", a->stripe_sectors); + if (bkey_val_bytes(k.k) > offsetof(struct bch_alloc_v4, stripe_sectors)) + prt_printf(out, "stripe_sectors %u\n", a->stripe_sectors); prt_printf(out, "cached_sectors %u\n", a->cached_sectors); prt_printf(out, "stripe %u\n", a->stripe); prt_printf(out, "stripe_redundancy %u\n", a->stripe_redundancy); @@ -372,12 +377,12 @@ void bch2_alloc_to_text(struct printbuf *out, struct bch_fs *c, struct bkey_s_c struct bch_alloc_v4 _a; const struct bch_alloc_v4 *a = bch2_alloc_to_v4(k, &_a); - __bch2_alloc_v4_to_text(out, c, k.k->p.inode, a); + __bch2_alloc_v4_to_text(out, c, k, a); } void bch2_alloc_v4_to_text(struct printbuf *out, struct bch_fs *c, struct bkey_s_c k) { - __bch2_alloc_v4_to_text(out, c, k.k->p.inode, bkey_s_c_to_alloc_v4(k).v); + __bch2_alloc_v4_to_text(out, c, k, bkey_s_c_to_alloc_v4(k).v); } void __bch2_alloc_to_v4(struct bkey_s_c k, struct bch_alloc_v4 *out) @@ -385,7 +390,7 @@ void __bch2_alloc_to_v4(struct bkey_s_c k, struct bch_alloc_v4 *out) if (k.k->type == KEY_TYPE_alloc_v4) { void *src, *dst; - *out = *bkey_s_c_to_alloc_v4(k).v; + bkey_val_copy(out, bkey_s_c_to_alloc_v4(k)); src = alloc_v4_backpointers(out); SET_BCH_ALLOC_V4_BACKPOINTERS_START(out, BCH_ALLOC_V4_U64s); @@ -1549,6 +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); @@ -1572,6 +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); @@ -1732,12 +1742,16 @@ int bch2_check_alloc_to_lru_refs(struct bch_fs *c) bch2_bkey_buf_init(&last_flushed); bkey_init(&last_flushed.k->k); + struct progress_indicator_state progress; + bch2_progress_init(&progress, c, BIT_ULL(BTREE_ID_alloc)); + CLASS(btree_trans, trans)(c); int ret = for_each_btree_key_commit(trans, iter, BTREE_ID_alloc, POS_MIN, BTREE_ITER_prefetch, k, - NULL, NULL, BCH_TRANS_COMMIT_no_enospc, - bch2_check_alloc_to_lru_ref(trans, &iter, &last_flushed)) ?: - bch2_check_stripe_to_lru_refs(trans); + NULL, NULL, BCH_TRANS_COMMIT_no_enospc, ({ + progress_update_iter(trans, &progress, &iter); + bch2_check_alloc_to_lru_ref(trans, &iter, &last_flushed); + }))?: bch2_check_stripe_to_lru_refs(trans); bch2_bkey_buf_exit(&last_flushed, c); return ret; diff --git a/fs/bcachefs/bcachefs.h b/fs/bcachefs/bcachefs.h index 8a6f886b5bf2..45c15bdaa6f4 100644 --- a/fs/bcachefs/bcachefs.h +++ b/fs/bcachefs/bcachefs.h @@ -1277,4 +1277,11 @@ static inline int bch2_fs_casefold_enabled(struct bch_fs *c) return 0; } +static inline const char *strip_bch2(const char *msg) +{ + if (!strncmp("bch2_", msg, 5)) + return msg + 5; + return msg; +} + #endif /* _BCACHEFS_H */ 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_gc.c b/fs/bcachefs/btree_gc.c index 34cb8a4324dc..e95bb6849aef 100644 --- a/fs/bcachefs/btree_gc.c +++ b/fs/bcachefs/btree_gc.c @@ -44,10 +44,6 @@ #include <linux/rcupdate.h> #include <linux/sched/task.h> -#define DROP_THIS_NODE 10 -#define DROP_PREV_NODE 11 -#define DID_FILL_FROM_SCAN 12 - /* * Returns true if it's a btree we can easily reconstruct, or otherwise won't * cause data loss if it's missing: @@ -252,7 +248,7 @@ static int btree_check_node_boundaries(struct btree_trans *trans, struct btree * return ret; *pulled_from_scan = cur->data->min_key; - ret = DID_FILL_FROM_SCAN; + ret = bch_err_throw(c, topology_repair_did_fill_from_scan); } else { if (mustfix_fsck_err(trans, btree_node_topology_bad_min_key, "btree node with incorrect min_key%s", buf.buf)) @@ -263,7 +259,7 @@ static int btree_check_node_boundaries(struct btree_trans *trans, struct btree * if (bpos_ge(prev->data->min_key, cur->data->min_key)) { /* fully? */ if (mustfix_fsck_err(trans, btree_node_topology_overwritten_by_next_node, "btree node overwritten by next node%s", buf.buf)) - ret = DROP_PREV_NODE; + ret = bch_err_throw(c, topology_repair_drop_prev_node); } else { if (mustfix_fsck_err(trans, btree_node_topology_bad_max_key, "btree node with incorrect max_key%s", buf.buf)) @@ -274,7 +270,7 @@ static int btree_check_node_boundaries(struct btree_trans *trans, struct btree * if (bpos_ge(expected_start, cur->data->max_key)) { /* fully? */ if (mustfix_fsck_err(trans, btree_node_topology_overwritten_by_prev_node, "btree node overwritten by prev node%s", buf.buf)) - ret = DROP_THIS_NODE; + ret = bch_err_throw(c, topology_repair_drop_this_node); } else { if (mustfix_fsck_err(trans, btree_node_topology_bad_min_key, "btree node with incorrect min_key%s", buf.buf)) @@ -314,7 +310,7 @@ static int btree_repair_node_end(struct btree_trans *trans, struct btree *b, return ret; *pulled_from_scan = b->key.k.p; - ret = DID_FILL_FROM_SCAN; + ret = bch_err_throw(c, topology_repair_did_fill_from_scan); } else { ret = set_node_max(c, child, b->key.k.p); } @@ -391,15 +387,15 @@ again: ret = lockrestart_do(trans, btree_check_node_boundaries(trans, b, prev, cur, pulled_from_scan)); - if (ret < 0) + if (ret && !bch2_err_matches(ret, BCH_ERR_topology_repair)) goto err; - if (ret == DID_FILL_FROM_SCAN) { + if (bch2_err_matches(ret, BCH_ERR_topology_repair_did_fill_from_scan)) { new_pass = true; ret = 0; } - if (ret == DROP_THIS_NODE) { + if (bch2_err_matches(ret, BCH_ERR_topology_repair_drop_this_node)) { six_unlock_read(&cur->c.lock); bch2_btree_node_evict(trans, cur_k.k); ret = bch2_journal_key_delete(c, b->c.btree_id, @@ -414,7 +410,7 @@ again: six_unlock_read(&prev->c.lock); prev = NULL; - if (ret == DROP_PREV_NODE) { + if (bch2_err_matches(ret, BCH_ERR_topology_repair_drop_prev_node)) { bch_info(c, "dropped prev node"); bch2_btree_node_evict(trans, prev_k.k); ret = bch2_journal_key_delete(c, b->c.btree_id, @@ -436,7 +432,7 @@ again: BUG_ON(cur); ret = lockrestart_do(trans, btree_repair_node_end(trans, b, prev, pulled_from_scan)); - if (ret == DID_FILL_FROM_SCAN) { + if (bch2_err_matches(ret, BCH_ERR_topology_repair_did_fill_from_scan)) { new_pass = true; ret = 0; } @@ -477,7 +473,7 @@ again: six_unlock_read(&cur->c.lock); cur = NULL; - if (ret == DROP_THIS_NODE) { + if (bch2_err_matches(ret, BCH_ERR_topology_repair_drop_this_node)) { bch2_btree_node_evict(trans, cur_k.k); ret = bch2_journal_key_delete(c, b->c.btree_id, b->c.level, cur_k.k->k.p); @@ -504,7 +500,7 @@ again: if (mustfix_fsck_err_on(!have_child, c, btree_node_topology_interior_node_empty, "empty interior btree node at %s", buf.buf)) - ret = DROP_THIS_NODE; + ret = bch_err_throw(c, topology_repair_drop_this_node); err: fsck_err: if (!IS_ERR_OR_NULL(prev)) @@ -521,7 +517,8 @@ fsck_err: bch2_bkey_buf_exit(&prev_k, c); bch2_bkey_buf_exit(&cur_k, c); - bch_err_fn(c, ret); + if (!bch2_err_matches(ret, BCH_ERR_topology_repair)) + bch_err_fn(c, ret); return ret; } @@ -592,7 +589,7 @@ recover: ret = bch2_btree_repair_topology_recurse(trans, b, &pulled_from_scan); six_unlock_read(&b->c.lock); - if (ret == DROP_THIS_NODE) { + if (bch2_err_matches(ret, BCH_ERR_topology_repair_drop_this_node)) { scoped_guard(mutex, &c->btree_cache.lock) bch2_btree_node_hash_remove(&c->btree_cache, b); diff --git a/fs/bcachefs/btree_io.c b/fs/bcachefs/btree_io.c index bd86dd7151a1..8a03cd75a64f 100644 --- a/fs/bcachefs/btree_io.c +++ b/fs/bcachefs/btree_io.c @@ -1405,10 +1405,8 @@ static void btree_node_read_work(struct work_struct *work) ret = bch2_bkey_pick_read_device(c, bkey_i_to_s_c(&b->key), &failed, &rb->pick, -1); - if (ret <= 0) { - set_btree_node_read_error(b); + if (ret <= 0) break; - } ca = bch2_dev_get_ioref(c, rb->pick.ptr.dev, READ, BCH_DEV_READ_REF_btree_node_read); rb->have_ioref = ca != NULL; @@ -1442,27 +1440,21 @@ start: bch2_maybe_corrupt_bio(bio, bch2_btree_read_corrupt_ratio); ret = bch2_btree_node_read_done(c, ca, b, &failed, &buf); - if (ret == -BCH_ERR_btree_node_read_err_want_retry || - ret == -BCH_ERR_btree_node_read_err_must_retry) - continue; - - if (ret) - set_btree_node_read_error(b); - - break; + if (ret != -BCH_ERR_btree_node_read_err_want_retry && + ret != -BCH_ERR_btree_node_read_err_must_retry) + break; } bch2_io_failures_to_text(&buf, c, &failed); - if (btree_node_read_error(b)) - bch2_btree_lost_data(c, &buf, b->c.btree_id); - /* * only print retry success if we read from a replica with no errors */ - if (btree_node_read_error(b)) + if (ret) { + set_btree_node_read_error(b); + bch2_btree_lost_data(c, &buf, b->c.btree_id); prt_printf(&buf, "ret %s", bch2_err_str(ret)); - else if (failed.nr) { + } else if (failed.nr) { if (!bch2_dev_io_failures(&failed, rb->pick.ptr.dev)) prt_printf(&buf, "retry success"); else @@ -2019,7 +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 cc771affa511..dd716b35a11d 100644 --- a/fs/bcachefs/btree_iter.c +++ b/fs/bcachefs/btree_iter.c @@ -2744,7 +2744,7 @@ 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; @@ -2860,8 +2860,9 @@ struct bkey_s_c bch2_btree_iter_peek_slot(struct btree_trans *trans, struct btre !bkey_deleted(k.k) && (k2 = btree_trans_peek_key_cache(trans, iter, iter->pos)).k) { k = k2; - if (!bkey_err(k)) - iter->k = *k.k; + if (bkey_err(k)) + goto out; + iter->k = *k.k; } if (unlikely(k.k->type == KEY_TYPE_whiteout && diff --git a/fs/bcachefs/btree_key_cache.c b/fs/bcachefs/btree_key_cache.c index ebba14da92b4..d61b782087ce 100644 --- a/fs/bcachefs/btree_key_cache.c +++ b/fs/bcachefs/btree_key_cache.c @@ -254,11 +254,13 @@ static int btree_key_cache_create(struct btree_trans *trans, struct bkey_i *new_k = allocate_dropping_locks(trans, ret, kmalloc(key_u64s * sizeof(u64), _gfp)); - if (unlikely(!new_k)) { + if (unlikely(!new_k && !ret)) { bch_err(trans->c, "error allocating memory for key cache key, btree %s u64s %u", bch2_btree_id_str(ck->key.btree_id), key_u64s); ret = bch_err_throw(c, ENOMEM_btree_key_cache_fill); - } else if (ret) { + } + + if (unlikely(ret)) { kfree(new_k); goto err; } @@ -407,7 +409,7 @@ int bch2_btree_path_traverse_cached(struct btree_trans *trans, btree_node_unlock(trans, path, 0); path->l[0].b = ERR_PTR(ret); } - } else { + } else if (!(flags & BTREE_ITER_cached_nofill)) { BUG_ON(path->uptodate); BUG_ON(!path->nodes_locked); } diff --git a/fs/bcachefs/btree_node_scan.c b/fs/bcachefs/btree_node_scan.c index d997e3818c30..4b7b5ca74ba1 100644 --- a/fs/bcachefs/btree_node_scan.c +++ b/fs/bcachefs/btree_node_scan.c @@ -158,14 +158,6 @@ static void try_read_btree_node(struct find_btree_nodes *f, struct bch_dev *ca, if (BTREE_NODE_ID(bn) >= BTREE_ID_NR_MAX) return; - bio_reset(bio, ca->disk_sb.bdev, REQ_OP_READ); - bio->bi_iter.bi_sector = offset; - bch2_bio_map(bio, b->data, c->opts.btree_node_size); - - submit_time = local_clock(); - submit_bio_wait(bio); - bch2_account_io_completion(ca, BCH_MEMBER_ERROR_read, submit_time, !bio->bi_status); - rcu_read_lock(); struct found_btree_node n = { .btree_id = BTREE_NODE_ID(bn), @@ -182,6 +174,14 @@ static void try_read_btree_node(struct find_btree_nodes *f, struct bch_dev *ca, }; rcu_read_unlock(); + bio_reset(bio, ca->disk_sb.bdev, REQ_OP_READ); + bio->bi_iter.bi_sector = offset; + bch2_bio_map(bio, b->data, c->opts.btree_node_size); + + submit_time = local_clock(); + submit_bio_wait(bio); + bch2_account_io_completion(ca, BCH_MEMBER_ERROR_read, submit_time, !bio->bi_status); + found_btree_node_to_key(&b->key, &n); CLASS(printbuf, buf)(); @@ -270,6 +270,9 @@ static int read_btree_nodes(struct find_btree_nodes *f) int ret = 0; closure_init_stack(&cl); + CLASS(printbuf, buf)(); + + prt_printf(&buf, "scanning for btree nodes on"); for_each_online_member(c, ca, BCH_DEV_READ_REF_btree_node_scan) { if (!(ca->mi.data_allowed & BIT(BCH_DATA_btree))) @@ -295,10 +298,14 @@ static int read_btree_nodes(struct find_btree_nodes *f) break; } + prt_printf(&buf, " %s", ca->name); + closure_get(&cl); enumerated_ref_get(&ca->io_ref[READ], BCH_DEV_READ_REF_btree_node_scan); wake_up_process(t); } + + bch_notice(c, "%s", buf.buf); err: while (closure_sync_timeout(&cl, sysctl_hung_task_timeout_secs * HZ / 2)) ; diff --git a/fs/bcachefs/btree_trans_commit.c b/fs/bcachefs/btree_trans_commit.c index 1f9965ae610c..8b94a8156fbf 100644 --- a/fs/bcachefs/btree_trans_commit.c +++ b/fs/bcachefs/btree_trans_commit.c @@ -772,12 +772,13 @@ bch2_trans_commit_write_locked(struct btree_trans *trans, trans->journal_res.offset += trans->journal_entries.u64s; trans->journal_res.u64s -= trans->journal_entries.u64s; - memcpy_u64s_small(bch2_journal_add_entry(j, &trans->journal_res, - BCH_JSET_ENTRY_write_buffer_keys, - BTREE_ID_accounting, 0, - trans->accounting.u64s)->_data, - btree_trans_subbuf_base(trans, &trans->accounting), - trans->accounting.u64s); + if (trans->accounting.u64s) + memcpy_u64s_small(bch2_journal_add_entry(j, &trans->journal_res, + BCH_JSET_ENTRY_write_buffer_keys, + BTREE_ID_accounting, 0, + trans->accounting.u64s)->_data, + btree_trans_subbuf_base(trans, &trans->accounting), + trans->accounting.u64s); if (trans->journal_seq) *trans->journal_seq = trans->journal_res.seq; @@ -968,7 +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); @@ -1024,13 +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; } @@ -1065,11 +1070,15 @@ int __bch2_trans_commit(struct btree_trans *trans, enum bch_trans_commit_flags f EBUG_ON(test_bit(BCH_FS_clean_shutdown, &c->flags)); - journal_u64s = jset_u64s(trans->accounting.u64s); + journal_u64s = 0; + trans->journal_transaction_names = READ_ONCE(c->opts.journal_transaction_names); if (trans->journal_transaction_names) journal_u64s += jset_u64s(JSET_ENTRY_LOG_U64s); + if (trans->accounting.u64s) + journal_u64s += jset_u64s(trans->accounting.u64s); + trans_for_each_update(trans, i) { struct btree_path *path = trans->paths + i->path; diff --git a/fs/bcachefs/btree_types.h b/fs/bcachefs/btree_types.h index 76adf75617aa..e4870fbc11d0 100644 --- a/fs/bcachefs/btree_types.h +++ b/fs/bcachefs/btree_types.h @@ -485,7 +485,7 @@ 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 { @@ -854,15 +854,15 @@ 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_type_has_snapshot_field(enum btree_id btree) diff --git a/fs/bcachefs/btree_update_interior.c b/fs/bcachefs/btree_update_interior.c index 312ef203b27b..e4aa4fa749bc 100644 --- a/fs/bcachefs/btree_update_interior.c +++ b/fs/bcachefs/btree_update_interior.c @@ -14,6 +14,7 @@ #include "btree_locking.h" #include "buckets.h" #include "clock.h" +#include "disk_groups.h" #include "enumerated_ref.h" #include "error.h" #include "extents.h" @@ -277,6 +278,36 @@ static void bch2_btree_node_free_never_used(struct btree_update *as, bch2_trans_node_drop(trans, b); } +static bool can_use_btree_node(struct bch_fs *c, + struct disk_reservation *res, + unsigned target, + struct bkey_s_c k) +{ + if (!bch2_bkey_devs_rw(c, k)) + return false; + + if (target && !bch2_bkey_in_target(c, k, target)) + return false; + + unsigned durability = bch2_bkey_durability(c, k); + + if (durability >= res->nr_replicas) + return true; + + struct bch_devs_mask devs = target_rw_devs(c, BCH_DATA_btree, target); + + guard(rcu)(); + + unsigned durability_available = 0, i; + for_each_set_bit(i, devs.d, BCH_SB_MEMBERS_MAX) { + struct bch_dev *ca = bch2_dev_rcu_noerror(c, i); + if (ca) + durability_available += ca->mi.durability; + } + + return durability >= durability_available; +} + static struct btree *__bch2_btree_node_alloc(struct btree_trans *trans, struct disk_reservation *res, struct closure *cl, @@ -303,10 +334,14 @@ static struct btree *__bch2_btree_node_alloc(struct btree_trans *trans, mutex_lock(&c->btree_reserve_cache_lock); if (c->btree_reserve_cache_nr > nr_reserve) { for (struct btree_alloc *a = c->btree_reserve_cache; - a < c->btree_reserve_cache + c->btree_reserve_cache_nr; - a++) { - if (target && !bch2_bkey_in_target(c, bkey_i_to_s_c(&a->k), target)) + a < c->btree_reserve_cache + c->btree_reserve_cache_nr;) { + /* check if it has sufficient durability */ + + if (!can_use_btree_node(c, res, target, bkey_i_to_s_c(&a->k))) { + bch2_open_buckets_put(c, &a->ob); + *a = c->btree_reserve_cache[--c->btree_reserve_cache_nr]; continue; + } bkey_copy(&b->key, &a->k); b->ob = a->ob; 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/ec.c b/fs/bcachefs/ec.c index 62dda821247e..bea14f02114f 100644 --- a/fs/bcachefs/ec.c +++ b/fs/bcachefs/ec.c @@ -2060,6 +2060,9 @@ allocated: BUG_ON(trans->restarted); return h; err: + if (waiting && + !bch2_err_matches(ret, BCH_ERR_operation_blocked)) + closure_wake_up(&c->freelist_wait); bch2_ec_stripe_head_put(c, h); return ERR_PTR(ret); } diff --git a/fs/bcachefs/errcode.h b/fs/bcachefs/errcode.h index 2de0dc91a69e..cec8b0f47d3d 100644 --- a/fs/bcachefs/errcode.h +++ b/fs/bcachefs/errcode.h @@ -90,6 +90,8 @@ x(ENOMEM, ENOMEM_disk_accounting) \ x(ENOMEM, ENOMEM_stripe_head_alloc) \ x(ENOMEM, ENOMEM_journal_read_bucket) \ + x(ENOMEM, ENOMEM_acl) \ + x(ENOMEM, ENOMEM_move_extent) \ x(ENOSPC, ENOSPC_disk_reservation) \ x(ENOSPC, ENOSPC_bucket_alloc) \ x(ENOSPC, ENOSPC_disk_label_add) \ @@ -216,9 +218,13 @@ x(EINVAL, varint_decode_error) \ x(EINVAL, erasure_coding_found_btree_node) \ x(EINVAL, option_negative) \ + x(EINVAL, topology_repair) \ + x(BCH_ERR_topology_repair, topology_repair_drop_this_node) \ + x(BCH_ERR_topology_repair, topology_repair_drop_prev_node) \ + x(BCH_ERR_topology_repair, topology_repair_did_fill_from_scan) \ x(EOPNOTSUPP, may_not_use_incompat_feature) \ x(EOPNOTSUPP, no_casefolding_without_utf8) \ - x(EOPNOTSUPP, casefolding_disabled) \ + x(EOPNOTSUPP, casefolding_disabled) \ x(EOPNOTSUPP, casefold_opt_is_dir_only) \ x(EOPNOTSUPP, unsupported_fsx_flag) \ x(EOPNOTSUPP, unsupported_fa_flag) \ diff --git a/fs/bcachefs/extents.c b/fs/bcachefs/extents.c index b36ecfc0ab9d..b879a586b7f6 100644 --- a/fs/bcachefs/extents.c +++ b/fs/bcachefs/extents.c @@ -282,9 +282,9 @@ int bch2_bkey_pick_read_device(struct bch_fs *c, struct bkey_s_c k, if (have_pick) return 1; - if (!have_dirty_ptrs) + if (!have_dirty_ptrs && !bkey_is_btree_ptr(k.k)) return 0; - if (have_missing_devs) + if (have_missing_devs || !have_dirty_ptrs) return bch_err_throw(c, no_device_to_read_from); if (have_csum_errors) return bch_err_throw(c, data_read_csum_err); @@ -1006,6 +1006,20 @@ const struct bch_extent_ptr *bch2_bkey_has_device_c(struct bkey_s_c k, unsigned return NULL; } +bool bch2_bkey_devs_rw(struct bch_fs *c, struct bkey_s_c k) +{ + struct bkey_ptrs_c ptrs = bch2_bkey_ptrs_c(k); + + guard(rcu)(); + bkey_for_each_ptr(ptrs, ptr) { + CLASS(bch2_dev_tryget, ca)(c, ptr->dev); + if (!ca || ca->mi.state != BCH_MEMBER_STATE_rw) + return false; + } + + return true; +} + bool bch2_bkey_has_target(struct bch_fs *c, struct bkey_s_c k, unsigned target) { struct bkey_ptrs_c ptrs = bch2_bkey_ptrs_c(k); diff --git a/fs/bcachefs/extents.h b/fs/bcachefs/extents.h index f212f91c278d..35ee03cd5065 100644 --- a/fs/bcachefs/extents.h +++ b/fs/bcachefs/extents.h @@ -614,6 +614,8 @@ static inline struct bch_extent_ptr *bch2_bkey_has_device(struct bkey_s k, unsig return (void *) bch2_bkey_has_device_c(k.s_c, dev); } +bool bch2_bkey_devs_rw(struct bch_fs *, struct bkey_s_c); + bool bch2_bkey_has_target(struct bch_fs *, struct bkey_s_c, unsigned); bool bch2_bkey_in_target(struct bch_fs *, struct bkey_s_c, unsigned); 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-direct.c b/fs/bcachefs/fs-io-direct.c index 73d44875faf2..e53fee0513fd 100644 --- a/fs/bcachefs/fs-io-direct.c +++ b/fs/bcachefs/fs-io-direct.c @@ -127,7 +127,7 @@ static int bch2_direct_IO_read(struct kiocb *req, struct iov_iter *iter) * the dirtying of requests that are internal from the kernel (i.e. from * loopback), because we'll deadlock on page_lock. */ - dio->should_dirty = iter_is_iovec(iter); + dio->should_dirty = user_backed_iter(iter); blk_start_plug(&plug); 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/fs.c b/fs/bcachefs/fs.c index 2789b30add10..56b7126bc31d 100644 --- a/fs/bcachefs/fs.c +++ b/fs/bcachefs/fs.c @@ -1295,8 +1295,14 @@ static int bch2_fill_extent(struct bch_fs *c, flags| FIEMAP_EXTENT_DELALLOC| FIEMAP_EXTENT_UNWRITTEN); + } else if (k.k->type == KEY_TYPE_error) { + return 0; } else { - BUG(); + WARN_ONCE(1, "unhandled key type %s", + k.k->type < KEY_TYPE_MAX + ? bch2_bkey_types[k.k->type] + : "(unknown)"); + return 0; } } diff --git a/fs/bcachefs/fsck.c b/fs/bcachefs/fsck.c index df0aa2522b18..f5a9da40c647 100644 --- a/fs/bcachefs/fsck.c +++ b/fs/bcachefs/fsck.c @@ -15,6 +15,7 @@ #include "io_misc.h" #include "keylist.h" #include "namei.h" +#include "progress.h" #include "recovery_passes.h" #include "snapshot.h" #include "super.h" @@ -1331,11 +1332,16 @@ int bch2_check_inodes(struct bch_fs *c) CLASS(btree_trans, trans)(c); CLASS(snapshots_seen, s)(); + struct progress_indicator_state progress; + bch2_progress_init(&progress, c, BIT_ULL(BTREE_ID_inodes)); + return for_each_btree_key_commit(trans, iter, BTREE_ID_inodes, POS_MIN, BTREE_ITER_prefetch|BTREE_ITER_all_snapshots, k, - NULL, NULL, BCH_TRANS_COMMIT_no_enospc, - check_inode(trans, &iter, k, &snapshot_root, &s)); + NULL, NULL, BCH_TRANS_COMMIT_no_enospc, ({ + progress_update_iter(trans, &progress, &iter); + check_inode(trans, &iter, k, &snapshot_root, &s); + })); } static int find_oldest_inode_needs_reattach(struct btree_trans *trans, @@ -1422,12 +1428,17 @@ fsck_err: */ int bch2_check_unreachable_inodes(struct bch_fs *c) { + struct progress_indicator_state progress; + bch2_progress_init(&progress, c, BIT_ULL(BTREE_ID_inodes)); + CLASS(btree_trans, trans)(c); return for_each_btree_key_commit(trans, iter, BTREE_ID_inodes, POS_MIN, BTREE_ITER_prefetch|BTREE_ITER_all_snapshots, k, - NULL, NULL, BCH_TRANS_COMMIT_no_enospc, - check_unreachable_inode(trans, &iter, k)); + NULL, NULL, BCH_TRANS_COMMIT_no_enospc, ({ + progress_update_iter(trans, &progress, &iter); + check_unreachable_inode(trans, &iter, k); + })); } static inline bool btree_matches_i_mode(enum btree_id btree, unsigned mode) @@ -1961,11 +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; @@ -1975,6 +1987,10 @@ static int check_extent(struct btree_trans *trans, struct btree_iter *iter, } } + ret = check_extent_overbig(trans, iter, k); + if (ret) + goto err; + ret = bch2_trans_commit(trans, res, NULL, BCH_TRANS_COMMIT_no_enospc); if (ret) goto err; @@ -2017,12 +2033,15 @@ int bch2_check_extents(struct bch_fs *c) CLASS(inode_walker, w)(); CLASS(extent_ends, extent_ends)(); + struct progress_indicator_state progress; + bch2_progress_init(&progress, c, BIT_ULL(BTREE_ID_extents)); + int ret = for_each_btree_key(trans, iter, BTREE_ID_extents, POS(BCACHEFS_ROOT_INO, 0), BTREE_ITER_prefetch|BTREE_ITER_all_snapshots, k, ({ + progress_update_iter(trans, &progress, &iter); bch2_disk_reservation_put(c, &res); - check_extent(trans, &iter, k, &w, &s, &extent_ends, &res) ?: - check_extent_overbig(trans, &iter, k); + check_extent(trans, &iter, k, &w, &s, &extent_ends, &res); })) ?: check_i_sectors_notnested(trans, &w); @@ -2035,11 +2054,15 @@ int bch2_check_indirect_extents(struct bch_fs *c) CLASS(btree_trans, trans)(c); struct disk_reservation res = { 0 }; + struct progress_indicator_state progress; + bch2_progress_init(&progress, c, BIT_ULL(BTREE_ID_reflink)); + int ret = for_each_btree_key_commit(trans, iter, BTREE_ID_reflink, POS_MIN, BTREE_ITER_prefetch, k, &res, NULL, BCH_TRANS_COMMIT_no_enospc, ({ + progress_update_iter(trans, &progress, &iter); bch2_disk_reservation_put(c, &res); check_extent_overbig(trans, &iter, k); })); @@ -2448,15 +2471,20 @@ int bch2_check_dirents(struct bch_fs *c) CLASS(snapshots_seen, s)(); CLASS(inode_walker, dir)(); CLASS(inode_walker, target)(); + struct progress_indicator_state progress; bool need_second_pass = false, did_second_pass = false; int ret; again: + bch2_progress_init(&progress, c, BIT_ULL(BTREE_ID_dirents)); + ret = for_each_btree_key_commit(trans, iter, BTREE_ID_dirents, POS(BCACHEFS_ROOT_INO, 0), BTREE_ITER_prefetch|BTREE_ITER_all_snapshots, k, - NULL, NULL, BCH_TRANS_COMMIT_no_enospc, + NULL, NULL, BCH_TRANS_COMMIT_no_enospc, ({ + progress_update_iter(trans, &progress, &iter); check_dirent(trans, &iter, k, &hash_info, &dir, &target, &s, - &need_second_pass)) ?: + &need_second_pass); + })) ?: check_subdir_count_notnested(trans, &dir); if (!ret && need_second_pass && !did_second_pass) { @@ -2516,13 +2544,18 @@ int bch2_check_xattrs(struct bch_fs *c) CLASS(btree_trans, trans)(c); CLASS(inode_walker, inode)(); + struct progress_indicator_state progress; + bch2_progress_init(&progress, c, BIT_ULL(BTREE_ID_xattrs)); + int ret = for_each_btree_key_commit(trans, iter, BTREE_ID_xattrs, POS(BCACHEFS_ROOT_INO, 0), BTREE_ITER_prefetch|BTREE_ITER_all_snapshots, k, NULL, NULL, - BCH_TRANS_COMMIT_no_enospc, - check_xattr(trans, &iter, k, &hash_info, &inode)); + BCH_TRANS_COMMIT_no_enospc, ({ + progress_update_iter(trans, &progress, &iter); + check_xattr(trans, &iter, k, &hash_info, &inode); + })); return ret; } @@ -2664,10 +2697,16 @@ err: int bch2_check_subvolume_structure(struct bch_fs *c) { CLASS(btree_trans, trans)(c); + + struct progress_indicator_state progress; + bch2_progress_init(&progress, c, BIT_ULL(BTREE_ID_subvolumes)); + return for_each_btree_key_commit(trans, iter, BTREE_ID_subvolumes, POS_MIN, BTREE_ITER_prefetch, k, - NULL, NULL, BCH_TRANS_COMMIT_no_enospc, - check_subvol_path(trans, &iter, k)); + NULL, NULL, BCH_TRANS_COMMIT_no_enospc, ({ + progress_update_iter(trans, &progress, &iter); + check_subvol_path(trans, &iter, k); + })); } static int bch2_bi_depth_renumber_one(struct btree_trans *trans, diff --git a/fs/bcachefs/io_write.c b/fs/bcachefs/io_write.c index d7620138e038..44b02d4b6502 100644 --- a/fs/bcachefs/io_write.c +++ b/fs/bcachefs/io_write.c @@ -89,7 +89,12 @@ void bch2_latency_acct(struct bch_dev *ca, u64 submit_time, int rw) new = ewma_add(old, io_latency, 5); } while (!atomic64_try_cmpxchg(latency, &old, new)); - bch2_congested_acct(ca, io_latency, now, rw); + /* + * Only track read latency for congestion accounting: writes are subject + * to heavy queuing delays from page cache writeback: + */ + if (rw == READ) + bch2_congested_acct(ca, io_latency, now, rw); __bch2_time_stats_update(&ca->io_latency[rw].stats, submit_time, now); } diff --git a/fs/bcachefs/journal.c b/fs/bcachefs/journal.c index 3ba1f9fd3402..07869436a964 100644 --- a/fs/bcachefs/journal.c +++ b/fs/bcachefs/journal.c @@ -182,6 +182,8 @@ 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++) { @@ -196,6 +198,7 @@ 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); } @@ -1057,6 +1060,7 @@ 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; } @@ -1292,7 +1296,7 @@ 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); @@ -1467,6 +1471,10 @@ int bch2_fs_journal_start(struct journal *j, u64 last_seq, u64 cur_seq) last_seq = cur_seq; u64 nr = cur_seq - last_seq; + if (nr * sizeof(struct journal_entry_pin_list) > 1U << 30) { + bch_err(c, "too many ntjournal fifo (%llu open entries)", nr); + return bch_err_throw(c, ENOMEM_journal_pin_fifo); + } /* * Extra fudge factor, in case we crashed when the journal pin fifo was @@ -1479,7 +1487,7 @@ int bch2_fs_journal_start(struct journal *j, u64 last_seq, u64 cur_seq) nr = max(nr, JOURNAL_PIN); init_fifo(&j->pin, roundup_pow_of_two(nr), GFP_KERNEL); if (!j->pin.data) { - bch_err(c, "error reallocating journal fifo (%llu open entries)", nr); + bch_err(c, "error allocating journal fifo (%llu open entries)", nr); return bch_err_throw(c, ENOMEM_journal_pin_fifo); } diff --git a/fs/bcachefs/journal_io.c b/fs/bcachefs/journal_io.c index 2835250a14c4..093e4acad085 100644 --- a/fs/bcachefs/journal_io.c +++ b/fs/bcachefs/journal_io.c @@ -428,15 +428,22 @@ static void journal_entry_btree_keys_to_text(struct printbuf *out, struct bch_fs bool first = true; jset_entry_for_each_key(entry, k) { - /* We may be called on entries that haven't been validated: */ - if (!k->k.u64s) - break; - if (!first) { prt_newline(out); bch2_prt_jset_entry_type(out, entry->type); prt_str(out, ": "); } + /* We may be called on entries that haven't been validated: */ + if (!k->k.u64s) { + prt_str(out, "(invalid, k->u64s 0)"); + break; + } + + if (bkey_next(k) > vstruct_last(entry)) { + prt_str(out, "(invalid, bkey overruns jset_entry)"); + break; + } + bch2_btree_id_level_to_text(out, entry->btree_id, entry->level); prt_char(out, ' '); bch2_bkey_val_to_text(out, c, bkey_i_to_s_c(k)); @@ -1820,6 +1827,8 @@ 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/lru.c b/fs/bcachefs/lru.c index ee14656c3fdd..76109b37d681 100644 --- a/fs/bcachefs/lru.c +++ b/fs/bcachefs/lru.c @@ -9,6 +9,7 @@ #include "ec.h" #include "error.h" #include "lru.h" +#include "progress.h" #include "recovery.h" /* KEY_TYPE_lru is obsolete: */ @@ -207,11 +208,16 @@ int bch2_check_lrus(struct bch_fs *c) bch2_bkey_buf_init(&last_flushed); bkey_init(&last_flushed.k->k); + struct progress_indicator_state progress; + bch2_progress_init(&progress, c, BIT_ULL(BTREE_ID_lru)); + CLASS(btree_trans, trans)(c); int ret = for_each_btree_key_commit(trans, iter, BTREE_ID_lru, POS_MIN, BTREE_ITER_prefetch, k, - NULL, NULL, BCH_TRANS_COMMIT_no_enospc, - bch2_check_lru_key(trans, &iter, k, &last_flushed)); + NULL, NULL, BCH_TRANS_COMMIT_no_enospc, ({ + progress_update_iter(trans, &progress, &iter); + bch2_check_lru_key(trans, &iter, k, &last_flushed); + })); bch2_bkey_buf_exit(&last_flushed, c); return ret; diff --git a/fs/bcachefs/move.c b/fs/bcachefs/move.c index 3f44bb54f91a..84a228c42f06 100644 --- a/fs/bcachefs/move.c +++ b/fs/bcachefs/move.c @@ -330,7 +330,7 @@ int bch2_move_extent(struct moving_context *ctxt, { struct btree_trans *trans = ctxt->trans; struct bch_fs *c = trans->c; - int ret = -ENOMEM; + int ret = 0; if (trace_io_move_enabled()) trace_io_move2(c, k, &io_opts, &data_opts); @@ -351,11 +351,10 @@ int bch2_move_extent(struct moving_context *ctxt, struct moving_io *io = allocate_dropping_locks(trans, ret, kzalloc(sizeof(struct moving_io), _gfp)); - if (!io) - goto err; - + if (!io && !ret) + ret = bch_err_throw(c, ENOMEM_move_extent); if (ret) - goto err_free; + goto err; INIT_LIST_HEAD(&io->io_list); io->write.ctxt = ctxt; @@ -366,7 +365,7 @@ int bch2_move_extent(struct moving_context *ctxt, ret = bch2_data_update_init(trans, iter, ctxt, &io->write, ctxt->wp, &io_opts, data_opts, iter->btree_id, k); if (ret) - goto err_free; + goto err; io->write.op.end_io = move_write_done; } else { @@ -380,7 +379,7 @@ int bch2_move_extent(struct moving_context *ctxt, ret = bch2_data_update_bios_init(&io->write, c, &io_opts); if (ret) - goto err_free; + goto err; } io->write.rbio.bio.bi_end_io = move_read_endio; @@ -423,9 +422,8 @@ int bch2_move_extent(struct moving_context *ctxt, BCH_READ_last_fragment, data_opts.scrub ? data_opts.read_dev : -1); return 0; -err_free: - kfree(io); err: + kfree(io); if (bch2_err_matches(ret, EROFS) || bch2_err_matches(ret, BCH_ERR_transaction_restart)) return ret; @@ -795,50 +793,50 @@ out: return ret; } -int __bch2_move_data(struct moving_context *ctxt, - struct bbpos start, - struct bbpos end, - move_pred_fn pred, void *arg) +static int bch2_move_data(struct bch_fs *c, + struct bbpos start, + struct bbpos end, + unsigned min_depth, + struct bch_ratelimit *rate, + struct bch_move_stats *stats, + struct write_point_specifier wp, + bool wait_on_copygc, + move_pred_fn pred, void *arg) { - struct bch_fs *c = ctxt->trans->c; - enum btree_id id; int ret = 0; - for (id = start.btree; + struct moving_context ctxt; + bch2_moving_ctxt_init(&ctxt, c, rate, stats, wp, wait_on_copygc); + + for (enum btree_id id = start.btree; id <= min_t(unsigned, end.btree, btree_id_nr_alive(c) - 1); id++) { - ctxt->stats->pos = BBPOS(id, POS_MIN); + ctxt.stats->pos = BBPOS(id, POS_MIN); - if (!btree_type_has_ptrs(id) || - !bch2_btree_id_root(c, id)->b) + if (!bch2_btree_id_root(c, id)->b) continue; - ret = bch2_move_data_btree(ctxt, - id == start.btree ? start.pos : POS_MIN, - id == end.btree ? end.pos : POS_MAX, - pred, arg, id, 0); + unsigned min_depth_this_btree = min_depth; + + if (!btree_type_has_ptrs(id)) + min_depth_this_btree = max(min_depth_this_btree, 1); + + for (unsigned level = min_depth_this_btree; + level < BTREE_MAX_DEPTH; + level++) { + ret = bch2_move_data_btree(&ctxt, + id == start.btree ? start.pos : POS_MIN, + id == end.btree ? end.pos : POS_MAX, + pred, arg, id, level); + if (ret) + break; + } + if (ret) break; } - return ret; -} - -int bch2_move_data(struct bch_fs *c, - struct bbpos start, - struct bbpos end, - struct bch_ratelimit *rate, - struct bch_move_stats *stats, - struct write_point_specifier wp, - bool wait_on_copygc, - move_pred_fn pred, void *arg) -{ - struct moving_context ctxt; - - bch2_moving_ctxt_init(&ctxt, c, rate, stats, wp, wait_on_copygc); - int ret = __bch2_move_data(&ctxt, start, end, pred, arg); bch2_moving_ctxt_exit(&ctxt); - return ret; } @@ -1206,14 +1204,6 @@ static bool migrate_pred(struct bch_fs *c, void *arg, return data_opts->rewrite_ptrs != 0; } -static bool rereplicate_btree_pred(struct bch_fs *c, void *arg, - struct btree *b, - struct bch_io_opts *io_opts, - struct data_update_opts *data_opts) -{ - return rereplicate_pred(c, arg, b->c.btree_id, bkey_i_to_s_c(&b->key), io_opts, data_opts); -} - /* * Ancient versions of bcachefs produced packed formats which could represent * keys that the in memory format cannot represent; this checks for those @@ -1293,15 +1283,6 @@ static bool drop_extra_replicas_pred(struct bch_fs *c, void *arg, return data_opts->kill_ptrs != 0; } -static bool drop_extra_replicas_btree_pred(struct bch_fs *c, void *arg, - struct btree *b, - struct bch_io_opts *io_opts, - struct data_update_opts *data_opts) -{ - return drop_extra_replicas_pred(c, arg, b->c.btree_id, bkey_i_to_s_c(&b->key), - io_opts, data_opts); -} - static bool scrub_pred(struct bch_fs *c, void *_arg, enum btree_id btree, struct bkey_s_c k, struct bch_io_opts *io_opts, @@ -1359,14 +1340,11 @@ int bch2_data_job(struct bch_fs *c, case BCH_DATA_OP_rereplicate: stats->data_type = BCH_DATA_journal; ret = bch2_journal_flush_device_pins(&c->journal, -1); - ret = bch2_move_btree(c, start, end, - rereplicate_btree_pred, c, stats) ?: ret; - ret = bch2_move_data(c, start, end, - NULL, - stats, + ret = bch2_move_data(c, start, end, 0, NULL, stats, writepoint_hashed((unsigned long) current), true, rereplicate_pred, c) ?: ret; + bch2_btree_interior_updates_flush(c); ret = bch2_replicas_gc2(c) ?: ret; break; case BCH_DATA_OP_migrate: @@ -1389,12 +1367,10 @@ int bch2_data_job(struct bch_fs *c, ret = bch2_scan_old_btree_nodes(c, stats); break; case BCH_DATA_OP_drop_extra_replicas: - ret = bch2_move_btree(c, start, end, - drop_extra_replicas_btree_pred, c, stats) ?: ret; - ret = bch2_move_data(c, start, end, NULL, stats, - writepoint_hashed((unsigned long) current), - true, - drop_extra_replicas_pred, c) ?: ret; + ret = bch2_move_data(c, start, end, 0, NULL, stats, + writepoint_hashed((unsigned long) current), + true, + drop_extra_replicas_pred, c) ?: ret; ret = bch2_replicas_gc2(c) ?: ret; break; default: diff --git a/fs/bcachefs/move.h b/fs/bcachefs/move.h index fe92ca6d418d..481026ff99ab 100644 --- a/fs/bcachefs/move.h +++ b/fs/bcachefs/move.h @@ -128,18 +128,6 @@ struct bch_io_opts *bch2_move_get_io_opts(struct btree_trans *, int bch2_move_data_btree(struct moving_context *, struct bpos, struct bpos, move_pred_fn, void *, enum btree_id, unsigned); -int __bch2_move_data(struct moving_context *, - struct bbpos, - struct bbpos, - move_pred_fn, void *); -int bch2_move_data(struct bch_fs *, - struct bbpos start, - struct bbpos end, - struct bch_ratelimit *, - struct bch_move_stats *, - struct write_point_specifier, - bool, - move_pred_fn, void *); int bch2_move_data_phys(struct bch_fs *, unsigned, u64, u64, unsigned, struct bch_ratelimit *, struct bch_move_stats *, diff --git a/fs/bcachefs/progress.c b/fs/bcachefs/progress.c index 42353067ba28..792fc6fef270 100644 --- a/fs/bcachefs/progress.c +++ b/fs/bcachefs/progress.c @@ -52,7 +52,8 @@ void bch2_progress_update_iter(struct btree_trans *trans, : 0; prt_printf(&buf, "%s: %d%%, done %llu/%llu nodes, at ", - msg, percent, s->nodes_seen, s->nodes_total); + strip_bch2(msg), + percent, s->nodes_seen, s->nodes_total); bch2_bbpos_to_text(&buf, BBPOS(iter->btree_id, iter->pos)); bch_info(c, "%s", buf.buf); diff --git a/fs/bcachefs/progress.h b/fs/bcachefs/progress.h index 23fb1811f943..972a73087ffe 100644 --- a/fs/bcachefs/progress.h +++ b/fs/bcachefs/progress.h @@ -26,4 +26,7 @@ void bch2_progress_update_iter(struct btree_trans *, struct btree_iter *, const char *); +#define progress_update_iter(trans, p, iter) \ + bch2_progress_update_iter(trans, p, iter, __func__) + #endif /* _BCACHEFS_PROGRESS_H */ diff --git a/fs/bcachefs/rebalance.c b/fs/bcachefs/rebalance.c index 32fa7cf90b63..c7e7f508fd0b 100644 --- a/fs/bcachefs/rebalance.c +++ b/fs/bcachefs/rebalance.c @@ -15,6 +15,7 @@ #include "inode.h" #include "io_write.h" #include "move.h" +#include "progress.h" #include "rebalance.h" #include "subvolume.h" #include "super-io.h" @@ -858,7 +859,12 @@ int bch2_check_rebalance_work(struct bch_fs *c) bch2_bkey_buf_init(&last_flushed); bkey_init(&last_flushed.k->k); + struct progress_indicator_state progress; + bch2_progress_init(&progress, c, BIT_ULL(BTREE_ID_rebalance_work)); + while (!ret) { + progress_update_iter(trans, &progress, &rebalance_iter); + bch2_trans_begin(trans); ret = check_rebalance_work_one(trans, &extent_iter, &rebalance_iter, &last_flushed); diff --git a/fs/bcachefs/recovery.c b/fs/bcachefs/recovery.c index 58c159e5f10d..304473dac268 100644 --- a/fs/bcachefs/recovery.c +++ b/fs/bcachefs/recovery.c @@ -67,13 +67,16 @@ int bch2_btree_lost_data(struct bch_fs *c, ret = __bch2_run_explicit_recovery_pass(c, msg, BCH_RECOVERY_PASS_check_backpointers_to_extents, 0, &write_sb) ?: ret; #endif + write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_lru_entry_bad, ext->errors_silent); + write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_backpointer_to_missing_ptr, ext->errors_silent); + write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_data_type_wrong, ext->errors_silent); + write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_dirty_sectors_wrong, ext->errors_silent); + switch (btree) { case BTREE_ID_alloc: ret = __bch2_run_explicit_recovery_pass(c, msg, BCH_RECOVERY_PASS_check_alloc_info, 0, &write_sb) ?: ret; - write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_data_type_wrong, ext->errors_silent); write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_gen_wrong, ext->errors_silent); - write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_dirty_sectors_wrong, ext->errors_silent); write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_cached_sectors_wrong, ext->errors_silent); write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_stripe_wrong, ext->errors_silent); write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_stripe_redundancy_wrong, ext->errors_silent); diff --git a/fs/bcachefs/recovery_passes.c b/fs/bcachefs/recovery_passes.c index b2cdd111fd0e..bd442652d0f5 100644 --- a/fs/bcachefs/recovery_passes.c +++ b/fs/bcachefs/recovery_passes.c @@ -639,6 +639,8 @@ void bch2_recovery_pass_status_to_text(struct printbuf *out, struct bch_fs *c) prt_printf(out, "Current pass:\t%s\n", bch2_recovery_passes[r->curr_pass]); prt_passes(out, "Current passes", r->passes_to_run); } + + prt_printf(out, "Pass done:\t%s\n", bch2_recovery_passes[r->pass_done]); } void bch2_fs_recovery_passes_init(struct bch_fs *c) diff --git a/fs/bcachefs/recovery_passes.h b/fs/bcachefs/recovery_passes.h index 4f2c2f811d5e..95e3612bb96c 100644 --- a/fs/bcachefs/recovery_passes.h +++ b/fs/bcachefs/recovery_passes.h @@ -26,6 +26,12 @@ static inline bool go_rw_in_recovery(struct bch_fs *c) (c->opts.fsck && !(c->sb.features & BIT_ULL(BCH_FEATURE_no_alloc_info)))); } +static inline bool recovery_pass_will_run(struct bch_fs *c, enum bch_recovery_pass pass) +{ + return unlikely(test_bit(BCH_FS_in_recovery, &c->flags) && + c->recovery.passes_to_run & BIT_ULL(pass)); +} + int bch2_run_print_explicit_recovery_pass(struct bch_fs *, enum bch_recovery_pass); int __bch2_run_explicit_recovery_pass(struct bch_fs *, struct printbuf *, diff --git a/fs/bcachefs/sb-members_format.h b/fs/bcachefs/sb-members_format.h index fb72ad730518..b2b892687cdd 100644 --- a/fs/bcachefs/sb-members_format.h +++ b/fs/bcachefs/sb-members_format.h @@ -17,7 +17,7 @@ UUID_INIT(0xffffffff, 0xffff, 0xffff, \ 0xd9, 0x6a, 0x60, 0xcf, 0x80, 0x3d, 0xf7, 0xef) -#define BCH_MIN_NR_NBUCKETS (1 << 6) +#define BCH_MIN_NR_NBUCKETS (1 << 9) #define BCH_IOPS_MEASUREMENTS() \ x(seqread, 0) \ diff --git a/fs/bcachefs/snapshot.c b/fs/bcachefs/snapshot.c index 7a801513b134..c1419afe4239 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" @@ -142,7 +143,7 @@ bool __bch2_snapshot_is_ancestor(struct bch_fs *c, u32 id, u32 ancestor) guard(rcu)(); struct snapshot_table *t = rcu_dereference(c->snapshots); - if (unlikely(c->recovery.pass_done < BCH_RECOVERY_PASS_check_snapshots)) + if (unlikely(recovery_pass_will_run(c, BCH_RECOVERY_PASS_check_snapshots))) return __bch2_snapshot_is_ancestor_early(t, id, ancestor); if (likely(ancestor >= IS_ANCESTOR_BITMAP)) @@ -973,12 +974,16 @@ 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) diff --git a/fs/bcachefs/str_hash.c b/fs/bcachefs/str_hash.c index dfe4b6ae0733..3e08e55d2dc1 100644 --- a/fs/bcachefs/str_hash.c +++ b/fs/bcachefs/str_hash.c @@ -329,7 +329,6 @@ duplicate_entries: out: fsck_err: bch2_trans_iter_exit(trans, dup_iter); - printbuf_exit(&buf); if (free_snapshots_seen) darray_exit(&s->ids); return ret; diff --git a/fs/bcachefs/super.c b/fs/bcachefs/super.c index 4e038f655f83..b3b2d8353a36 100644 --- a/fs/bcachefs/super.c +++ b/fs/bcachefs/super.c @@ -514,6 +514,10 @@ static int __bch2_fs_read_write(struct bch_fs *c, bool early) if (ret) return ret; + ret = bch2_fs_mark_dirty(c); + if (ret) + return ret; + clear_bit(BCH_FS_clean_shutdown, &c->flags); scoped_guard(rcu) @@ -537,10 +541,6 @@ static int __bch2_fs_read_write(struct bch_fs *c, bool early) bch2_journal_space_available(&c->journal); } - ret = bch2_fs_mark_dirty(c); - if (ret) - return ret; - /* * Don't jump to our error path, and call bch2_fs_read_only(), unless we * successfully marked the filesystem dirty @@ -729,6 +729,8 @@ void __bch2_fs_stop(struct bch_fs *c) cancel_work_sync(&ca->io_error_work); cancel_work_sync(&c->read_only_work); + + flush_work(&c->btree_interior_update_work); } void bch2_fs_free(struct bch_fs *c) diff --git a/fs/bcachefs/sysfs.c b/fs/bcachefs/sysfs.c index 158f526e3dcc..bd3fa9c3372d 100644 --- a/fs/bcachefs/sysfs.c +++ b/fs/bcachefs/sysfs.c @@ -18,6 +18,7 @@ #include "btree_key_cache.h" #include "btree_update.h" #include "btree_update_interior.h" +#include "btree_write_buffer.h" #include "btree_gc.h" #include "buckets.h" #include "clock.h" @@ -150,6 +151,7 @@ write_attribute(trigger_journal_flush); write_attribute(trigger_journal_writes); write_attribute(trigger_btree_cache_shrink); write_attribute(trigger_btree_key_cache_shrink); +write_attribute(trigger_btree_write_buffer_flush); write_attribute(trigger_btree_updates); write_attribute(trigger_freelist_wakeup); write_attribute(trigger_recalc_capacity); @@ -539,6 +541,11 @@ STORE(bch2_fs) c->btree_key_cache.shrink->scan_objects(c->btree_key_cache.shrink, &sc); } + if (attr == &sysfs_trigger_btree_write_buffer_flush) + bch2_trans_do(c, + (bch2_btree_write_buffer_flush_sync(trans), + bch2_trans_begin(trans))); + if (attr == &sysfs_trigger_gc) bch2_gc_gens(c); @@ -709,6 +716,7 @@ struct attribute *bch2_fs_internal_files[] = { &sysfs_trigger_journal_writes, &sysfs_trigger_btree_cache_shrink, &sysfs_trigger_btree_key_cache_shrink, + &sysfs_trigger_btree_write_buffer_flush, &sysfs_trigger_btree_updates, &sysfs_trigger_freelist_wakeup, &sysfs_trigger_recalc_capacity, |