diff options
Diffstat (limited to 'fs/bcachefs/btree_io.c')
| -rw-r--r-- | fs/bcachefs/btree_io.c | 62 |
1 files changed, 49 insertions, 13 deletions
diff --git a/fs/bcachefs/btree_io.c b/fs/bcachefs/btree_io.c index 19fd951495ac..b30799e494eb 100644 --- a/fs/bcachefs/btree_io.c +++ b/fs/bcachefs/btree_io.c @@ -26,6 +26,12 @@ #include <linux/sched/mm.h> +#ifdef CONFIG_BCACHEFS_DEBUG +static unsigned bch2_btree_read_corrupt_ratio; +module_param_named(btree_read_corrupt_ratio, bch2_btree_read_corrupt_ratio, uint, 0644); +MODULE_PARM_DESC(btree_read_corrupt_ratio, ""); +#endif + static void bch2_btree_node_header_to_text(struct printbuf *out, struct btree_node *bn) { bch2_btree_id_level_to_text(out, BTREE_NODE_ID(bn), BTREE_NODE_LEVEL(bn)); @@ -568,9 +574,9 @@ static int __btree_err(int ret, bch2_mark_btree_validate_failure(failed, ca->dev_idx); struct extent_ptr_decoded pick; - have_retry = !bch2_bkey_pick_read_device(c, + have_retry = bch2_bkey_pick_read_device(c, bkey_i_to_s_c(&b->key), - failed, &pick, -1); + failed, &pick, -1) == 1; } if (!have_retry && ret == -BCH_ERR_btree_node_read_err_want_retry) @@ -615,7 +621,6 @@ static int __btree_err(int ret, goto out; case -BCH_ERR_btree_node_read_err_bad_node: prt_str(&out, ", "); - ret = __bch2_topology_error(c, &out); break; } @@ -644,7 +649,6 @@ static int __btree_err(int ret, goto out; case -BCH_ERR_btree_node_read_err_bad_node: prt_str(&out, ", "); - ret = __bch2_topology_error(c, &out); break; } print: @@ -1337,15 +1341,42 @@ int bch2_btree_node_read_done(struct bch_fs *c, struct bch_dev *ca, btree_node_reset_sib_u64s(b); - scoped_guard(rcu) - bkey_for_each_ptr(bch2_bkey_ptrs(bkey_i_to_s(&b->key)), ptr) { - struct bch_dev *ca2 = bch2_dev_rcu(c, ptr->dev); - - if (!ca2 || ca2->mi.state != BCH_MEMBER_STATE_rw) { - set_btree_node_need_rewrite(b); - set_btree_node_need_rewrite_degraded(b); + /* + * XXX: + * + * We deadlock if too many btree updates require node rewrites while + * we're still in journal replay. + * + * This is because btree node rewrites generate more updates for the + * interior updates (alloc, backpointers), and if those updates touch + * new nodes and generate more rewrites - well, you see the problem. + * + * The biggest cause is that we don't use the btree write buffer (for + * the backpointer updates - this needs some real thought on locking in + * order to fix. + * + * The problem with this workaround (not doing the rewrite for degraded + * nodes in journal replay) is that those degraded nodes persist, and we + * don't want that (this is a real bug when a btree node write completes + * with fewer replicas than we wanted and leaves a degraded node due to + * device _removal_, i.e. the device went away mid write). + * + * It's less of a bug here, but still a problem because we don't yet + * have a way of tracking degraded data - we another index (all + * extents/btree nodes, by replicas entry) in order to fix properly + * (re-replicate degraded data at the earliest possible time). + */ + if (c->recovery.passes_complete & BIT_ULL(BCH_RECOVERY_PASS_journal_replay)) { + scoped_guard(rcu) + bkey_for_each_ptr(bch2_bkey_ptrs(bkey_i_to_s(&b->key)), ptr) { + struct bch_dev *ca2 = bch2_dev_rcu(c, ptr->dev); + + if (!ca2 || ca2->mi.state != BCH_MEMBER_STATE_rw) { + set_btree_node_need_rewrite(b); + set_btree_node_need_rewrite_degraded(b); + } } - } + } if (!ptr_written) { set_btree_node_need_rewrite(b); @@ -1381,7 +1412,7 @@ 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) { + if (ret <= 0) { set_btree_node_read_error(b); break; } @@ -1412,6 +1443,11 @@ start: continue; } + memset(&bio->bi_iter, 0, sizeof(bio->bi_iter)); + bio->bi_iter.bi_size = btree_buf_bytes(b); + + 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) |