X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libbcachefs%2Fbtree_iter.c;h=f430ca83720673043b135128aa75b272a73dfaff;hb=46d51608693bddac162232133516e975f3b1e835;hp=4cee5e6cd7f4aa5c7270d2f766ada56a6cef6ba3;hpb=a053ebfb8c89e023a44c365e369f4053cfc53376;p=bcachefs-tools-debian

diff --git a/libbcachefs/btree_iter.c b/libbcachefs/btree_iter.c
index 4cee5e6..f430ca8 100644
--- a/libbcachefs/btree_iter.c
+++ b/libbcachefs/btree_iter.c
@@ -257,7 +257,7 @@ static void bch2_btree_iter_verify(struct btree_iter *iter)
 
 	BUG_ON(!(iter->flags & __BTREE_ITER_ALL_SNAPSHOTS) &&
 	       (iter->flags & BTREE_ITER_ALL_SNAPSHOTS) &&
-	       !btree_type_has_snapshots(iter->btree_id));
+	       !btree_type_has_snapshot_field(iter->btree_id));
 
 	if (iter->update_path)
 		bch2_btree_path_verify(trans, iter->update_path);
@@ -362,7 +362,7 @@ void bch2_assert_pos_locked(struct btree_trans *trans, enum btree_id id,
 	bch2_bpos_to_text(&buf, pos);
 
 	panic("not locked: %s %s%s\n",
-	      bch2_btree_ids[id], buf.buf,
+	      bch2_btree_id_str(id), buf.buf,
 	      key_cache ? " cached" : "");
 }
 
@@ -781,7 +781,7 @@ static int btree_path_prefetch(struct btree_trans *trans, struct btree_path *pat
 	struct btree_node_iter node_iter = l->iter;
 	struct bkey_packed *k;
 	struct bkey_buf tmp;
-	unsigned nr = test_bit(BCH_FS_STARTED, &c->flags)
+	unsigned nr = test_bit(BCH_FS_started, &c->flags)
 		? (path->level > 1 ? 0 :  2)
 		: (path->level > 1 ? 1 : 16);
 	bool was_locked = btree_node_locked(path, path->level);
@@ -816,7 +816,7 @@ static int btree_path_prefetch_j(struct btree_trans *trans, struct btree_path *p
 	struct bch_fs *c = trans->c;
 	struct bkey_s_c k;
 	struct bkey_buf tmp;
-	unsigned nr = test_bit(BCH_FS_STARTED, &c->flags)
+	unsigned nr = test_bit(BCH_FS_started, &c->flags)
 		? (path->level > 1 ? 0 :  2)
 		: (path->level > 1 ? 1 : 16);
 	bool was_locked = btree_node_locked(path, path->level);
@@ -1109,6 +1109,9 @@ int bch2_btree_path_traverse_one(struct btree_trans *trans,
 	if (unlikely(ret))
 		goto out;
 
+	if (unlikely(!trans->srcu_held))
+		bch2_trans_srcu_lock(trans);
+
 	/*
 	 * Ensure we obey path->should_be_locked: if it's set, we can't unlock
 	 * and re-traverse the path without a transaction restart:
@@ -1211,8 +1214,6 @@ __bch2_btree_path_set_pos(struct btree_trans *trans,
 		   struct btree_path *path, struct bpos new_pos,
 		   bool intent, unsigned long ip, int cmp)
 {
-	unsigned level = path->level;
-
 	bch2_trans_verify_not_in_restart(trans);
 	EBUG_ON(!path->ref);
 
@@ -1228,7 +1229,7 @@ __bch2_btree_path_set_pos(struct btree_trans *trans,
 		goto out;
 	}
 
-	level = btree_path_up_until_good_node(trans, path, cmp);
+	unsigned level = btree_path_up_until_good_node(trans, path, cmp);
 
 	if (btree_path_node(path, level)) {
 		struct btree_path_level *l = &path->l[level];
@@ -1371,7 +1372,7 @@ void bch2_trans_updates_to_text(struct printbuf *buf, struct btree_trans *trans)
 		struct bkey_s_c old = { &i->old_k, i->old_v };
 
 		prt_printf(buf, "update: btree=%s cached=%u %pS",
-		       bch2_btree_ids[i->btree_id],
+		       bch2_btree_id_str(i->btree_id),
 		       i->cached,
 		       (void *) i->ip_allocated);
 		prt_newline(buf);
@@ -1387,7 +1388,7 @@ void bch2_trans_updates_to_text(struct printbuf *buf, struct btree_trans *trans)
 
 	trans_for_each_wb_update(trans, wb) {
 		prt_printf(buf, "update: btree=%s wb=1 %pS",
-		       bch2_btree_ids[wb->btree],
+		       bch2_btree_id_str(wb->btree),
 		       (void *) i->ip_allocated);
 		prt_newline(buf);
 
@@ -1416,7 +1417,7 @@ void bch2_btree_path_to_text(struct printbuf *out, struct btree_path *path)
 		   path->idx, path->ref, path->intent_ref,
 		   path->preserve ? 'P' : ' ',
 		   path->should_be_locked ? 'S' : ' ',
-		   bch2_btree_ids[path->btree_id],
+		   bch2_btree_id_str(path->btree_id),
 		   path->level);
 	bch2_bpos_to_text(out, path->pos);
 
@@ -1491,6 +1492,22 @@ static void bch2_trans_update_max_paths(struct btree_trans *trans)
 	trans->nr_max_paths = hweight64(trans->paths_allocated);
 }
 
+noinline __cold
+int __bch2_btree_trans_too_many_iters(struct btree_trans *trans)
+{
+	if (trace_trans_restart_too_many_iters_enabled()) {
+		struct printbuf buf = PRINTBUF;
+
+		bch2_trans_paths_to_text(&buf, trans);
+		trace_trans_restart_too_many_iters(trans, _THIS_IP_, buf.buf);
+		printbuf_exit(&buf);
+	}
+
+	count_event(trans->c, trans_restart_too_many_iters);
+
+	return btree_trans_restart(trans, BCH_ERR_transaction_restart_too_many_iters);
+}
+
 static noinline void btree_path_overflow(struct btree_trans *trans)
 {
 	bch2_dump_trans_paths_updates(trans);
@@ -1523,6 +1540,7 @@ static inline struct btree_path *btree_path_alloc(struct btree_trans *trans,
 	path->ref		= 0;
 	path->intent_ref	= 0;
 	path->nodes_locked	= 0;
+	path->alloc_seq++;
 
 	btree_path_list_add(trans, pos, path);
 	trans->paths_sorted = false;
@@ -1598,7 +1616,7 @@ struct btree_path *bch2_path_get(struct btree_trans *trans,
 
 	locks_want = min(locks_want, BTREE_MAX_DEPTH);
 	if (locks_want > path->locks_want)
-		bch2_btree_path_upgrade_noupgrade_sibs(trans, path, locks_want);
+		bch2_btree_path_upgrade_noupgrade_sibs(trans, path, locks_want, NULL);
 
 	return path;
 }
@@ -1795,23 +1813,15 @@ err:
 
 inline bool bch2_btree_iter_advance(struct btree_iter *iter)
 {
-	if (likely(!(iter->flags & BTREE_ITER_ALL_LEVELS))) {
-		struct bpos pos = iter->k.p;
-		bool ret = !(iter->flags & BTREE_ITER_ALL_SNAPSHOTS
-			     ? bpos_eq(pos, SPOS_MAX)
-			     : bkey_eq(pos, SPOS_MAX));
-
-		if (ret && !(iter->flags & BTREE_ITER_IS_EXTENTS))
-			pos = bkey_successor(iter, pos);
-		bch2_btree_iter_set_pos(iter, pos);
-		return ret;
-	} else {
-		if (!btree_path_node(iter->path, iter->path->level))
-			return true;
+	struct bpos pos = iter->k.p;
+	bool ret = !(iter->flags & BTREE_ITER_ALL_SNAPSHOTS
+		     ? bpos_eq(pos, SPOS_MAX)
+		     : bkey_eq(pos, SPOS_MAX));
 
-		iter->advanced = true;
-		return false;
-	}
+	if (ret && !(iter->flags & BTREE_ITER_IS_EXTENTS))
+		pos = bkey_successor(iter, pos);
+	bch2_btree_iter_set_pos(iter, pos);
+	return ret;
 }
 
 inline bool bch2_btree_iter_rewind(struct btree_iter *iter)
@@ -1860,19 +1870,11 @@ static struct bkey_i *bch2_btree_journal_peek(struct btree_trans *trans,
 					      struct btree_iter *iter,
 					      struct bpos end_pos)
 {
-	struct bkey_i *k;
-
-	if (bpos_lt(iter->path->pos, iter->journal_pos))
-		iter->journal_idx = 0;
-
-	k = bch2_journal_keys_peek_upto(trans->c, iter->btree_id,
-					iter->path->level,
-					iter->path->pos,
-					end_pos,
-					&iter->journal_idx);
-
-	iter->journal_pos = k ? k->k.p : end_pos;
-	return k;
+	return bch2_journal_keys_peek_upto(trans->c, iter->btree_id,
+					   iter->path->level,
+					   iter->path->pos,
+					   end_pos,
+					   &iter->journal_idx);
 }
 
 static noinline
@@ -2062,7 +2064,6 @@ struct bkey_s_c bch2_btree_iter_peek_upto(struct btree_iter *iter, struct bpos e
 	struct bpos iter_pos;
 	int ret;
 
-	EBUG_ON(iter->flags & BTREE_ITER_ALL_LEVELS);
 	EBUG_ON((iter->flags & BTREE_ITER_FILTER_SNAPSHOTS) && bkey_eq(end, POS_MAX));
 
 	if (iter->update_path) {
@@ -2188,103 +2189,6 @@ end:
 	goto out_no_locked;
 }
 
-/**
- * bch2_btree_iter_peek_all_levels() - returns the first key greater than or
- * equal to iterator's current position, returning keys from every level of the
- * btree. For keys at different levels of the btree that compare equal, the key
- * from the lower level (leaf) is returned first.
- * @iter:	iterator to peek from
- *
- * Returns:	key if found, or an error extractable with bkey_err().
- */
-struct bkey_s_c bch2_btree_iter_peek_all_levels(struct btree_iter *iter)
-{
-	struct btree_trans *trans = iter->trans;
-	struct bkey_s_c k;
-	int ret;
-
-	EBUG_ON(iter->path->cached);
-	bch2_btree_iter_verify(iter);
-	BUG_ON(iter->path->level < iter->min_depth);
-	BUG_ON(!(iter->flags & BTREE_ITER_ALL_SNAPSHOTS));
-	EBUG_ON(!(iter->flags & BTREE_ITER_ALL_LEVELS));
-
-	while (1) {
-		iter->path = bch2_btree_path_set_pos(trans, iter->path, iter->pos,
-					iter->flags & BTREE_ITER_INTENT,
-					btree_iter_ip_allocated(iter));
-
-		ret = bch2_btree_path_traverse(trans, iter->path, iter->flags);
-		if (unlikely(ret)) {
-			/* ensure that iter->k is consistent with iter->pos: */
-			bch2_btree_iter_set_pos(iter, iter->pos);
-			k = bkey_s_c_err(ret);
-			goto out_no_locked;
-		}
-
-		/* Already at end? */
-		if (!btree_path_node(iter->path, iter->path->level)) {
-			k = bkey_s_c_null;
-			goto out_no_locked;
-		}
-
-		k = btree_path_level_peek_all(trans->c,
-				&iter->path->l[iter->path->level], &iter->k);
-
-		/* Check if we should go up to the parent node: */
-		if (!k.k ||
-		    (iter->advanced &&
-		     bpos_eq(path_l(iter->path)->b->key.k.p, iter->pos))) {
-			iter->pos = path_l(iter->path)->b->key.k.p;
-			btree_path_set_level_up(trans, iter->path);
-			iter->advanced = false;
-			continue;
-		}
-
-		/*
-		 * Check if we should go back down to a leaf:
-		 * If we're not in a leaf node, we only return the current key
-		 * if it exactly matches iter->pos - otherwise we first have to
-		 * go back to the leaf:
-		 */
-		if (iter->path->level != iter->min_depth &&
-		    (iter->advanced ||
-		     !k.k ||
-		     !bpos_eq(iter->pos, k.k->p))) {
-			btree_path_set_level_down(trans, iter->path, iter->min_depth);
-			iter->pos = bpos_successor(iter->pos);
-			iter->advanced = false;
-			continue;
-		}
-
-		/* Check if we should go to the next key: */
-		if (iter->path->level == iter->min_depth &&
-		    iter->advanced &&
-		    k.k &&
-		    bpos_eq(iter->pos, k.k->p)) {
-			iter->pos = bpos_successor(iter->pos);
-			iter->advanced = false;
-			continue;
-		}
-
-		if (iter->advanced &&
-		    iter->path->level == iter->min_depth &&
-		    !bpos_eq(k.k->p, iter->pos))
-			iter->advanced = false;
-
-		BUG_ON(iter->advanced);
-		BUG_ON(!k.k);
-		break;
-	}
-
-	iter->pos = k.k->p;
-	btree_path_set_should_be_locked(iter->path);
-out_no_locked:
-	bch2_btree_iter_verify(iter);
-
-	return k;
-}
-
 /**
  * bch2_btree_iter_next() - returns first key greater than iterator's current
  * position
@@ -2451,7 +2355,6 @@ struct bkey_s_c bch2_btree_iter_peek_slot(struct btree_iter *iter)
 
 	bch2_btree_iter_verify(iter);
 	bch2_btree_iter_verify_entry_exit(iter);
-	EBUG_ON(iter->flags & BTREE_ITER_ALL_LEVELS);
 	EBUG_ON(iter->path->level && (iter->flags & BTREE_ITER_WITH_KEY_CACHE));
 
 	/* extents can't span inode numbers: */
@@ -2829,18 +2732,36 @@ void *__bch2_trans_kmalloc(struct btree_trans *trans, size_t size)
 	return p;
 }
 
-static noinline void bch2_trans_reset_srcu_lock(struct btree_trans *trans)
+static inline void check_srcu_held_too_long(struct btree_trans *trans)
 {
-	struct bch_fs *c = trans->c;
-	struct btree_path *path;
+	WARN(trans->srcu_held && time_after(jiffies, trans->srcu_lock_time + HZ * 10),
+	     "btree trans held srcu lock (delaying memory reclaim) for %lu seconds",
+	     (jiffies - trans->srcu_lock_time) / HZ);
+}
 
-	trans_for_each_path(trans, path)
-		if (path->cached && !btree_node_locked(path, 0))
-			path->l[0].b = ERR_PTR(-BCH_ERR_no_btree_node_srcu_reset);
+void bch2_trans_srcu_unlock(struct btree_trans *trans)
+{
+	if (trans->srcu_held) {
+		struct bch_fs *c = trans->c;
+		struct btree_path *path;
 
-	srcu_read_unlock(&c->btree_trans_barrier, trans->srcu_idx);
-	trans->srcu_idx = srcu_read_lock(&c->btree_trans_barrier);
-	trans->srcu_lock_time	= jiffies;
+		trans_for_each_path(trans, path)
+			if (path->cached && !btree_node_locked(path, 0))
+				path->l[0].b = ERR_PTR(-BCH_ERR_no_btree_node_srcu_reset);
+
+		check_srcu_held_too_long(trans);
+		srcu_read_unlock(&c->btree_trans_barrier, trans->srcu_idx);
+		trans->srcu_held = false;
+	}
+}
+
+void bch2_trans_srcu_lock(struct btree_trans *trans)
+{
+	if (!trans->srcu_held) {
+		trans->srcu_idx = srcu_read_lock(&trans->c->btree_trans_barrier);
+		trans->srcu_lock_time	= jiffies;
+		trans->srcu_held = true;
+	}
 }
 
 /**
@@ -2894,8 +2815,9 @@ u32 bch2_trans_begin(struct btree_trans *trans)
 	}
 	trans->last_begin_time = now;
 
-	if (unlikely(time_after(jiffies, trans->srcu_lock_time + msecs_to_jiffies(10))))
-		bch2_trans_reset_srcu_lock(trans);
+	if (unlikely(trans->srcu_held &&
+		     time_after(jiffies, trans->srcu_lock_time + msecs_to_jiffies(10))))
+		bch2_trans_srcu_unlock(trans);
 
 	trans->last_begin_ip = _RET_IP_;
 	if (trans->restarted) {
@@ -2960,7 +2882,8 @@ struct btree_trans *__bch2_trans_get(struct bch_fs *c, unsigned fn_idx)
 	trans->fn_idx		= fn_idx;
 	trans->locking_wait.task = current;
 	trans->journal_replay_not_finished =
-		!test_bit(JOURNAL_REPLAY_DONE, &c->journal.flags);
+		unlikely(!test_bit(JOURNAL_REPLAY_DONE, &c->journal.flags)) &&
+		atomic_inc_not_zero(&c->journal_keys.ref);
 	closure_init_stack(&trans->ref);
 
 	s = btree_trans_stats(trans);
@@ -2982,8 +2905,9 @@ struct btree_trans *__bch2_trans_get(struct bch_fs *c, unsigned fn_idx)
 		trans->wb_updates_size = s->wb_updates_size;
 	}
 
-	trans->srcu_idx = srcu_read_lock(&c->btree_trans_barrier);
+	trans->srcu_idx		= srcu_read_lock(&c->btree_trans_barrier);
 	trans->srcu_lock_time	= jiffies;
+	trans->srcu_held	= true;
 
 	if (IS_ENABLED(CONFIG_BCACHEFS_DEBUG_TRANSACTIONS)) {
 		struct btree_trans *pos;
@@ -3027,7 +2951,7 @@ leaked:
 	trans_for_each_path(trans, path)
 		if (path->ref)
 			printk(KERN_ERR "  btree %s %pS\n",
-			       bch2_btree_ids[path->btree_id],
+			       bch2_btree_id_str(path->btree_id),
 			       (void *) path->ip_allocated);
 	/* Be noisy about this: */
 	bch2_fatal_error(c);
@@ -3060,9 +2984,10 @@ void bch2_trans_put(struct btree_trans *trans)
 
 	check_btree_paths_leaked(trans);
 
-	srcu_read_unlock(&c->btree_trans_barrier, trans->srcu_idx);
-
-	bch2_journal_preres_put(&c->journal, &trans->journal_preres);
+	if (trans->srcu_held) {
+		check_srcu_held_too_long(trans);
+		srcu_read_unlock(&c->btree_trans_barrier, trans->srcu_idx);
+	}
 
 	kfree(trans->extra_journal_entries.data);
 
@@ -3075,6 +3000,9 @@ void bch2_trans_put(struct btree_trans *trans)
 			kfree(trans->fs_usage_deltas);
 	}
 
+	if (unlikely(trans->journal_replay_not_finished))
+		bch2_journal_keys_put(c);
+
 	if (trans->mem_bytes == BTREE_TRANS_MEM_MAX)
 		mempool_free(trans->mem, &c->btree_trans_mem_pool);
 	else
@@ -3102,7 +3030,7 @@ bch2_btree_bkey_cached_common_to_text(struct printbuf *out,
 
 	prt_tab(out);
 	prt_printf(out, "%px %c l=%u %s:", b, b->cached ? 'c' : 'b',
-		   b->level, bch2_btree_ids[b->btree_id]);
+		   b->level, bch2_btree_id_str(b->btree_id));
 	bch2_bpos_to_text(out, btree_node_pos(b));
 
 	prt_tab(out);
@@ -3115,6 +3043,7 @@ void bch2_btree_trans_to_text(struct printbuf *out, struct btree_trans *trans)
 	struct btree_path *path;
 	struct btree_bkey_cached_common *b;
 	static char lock_types[] = { 'r', 'i', 'w' };
+	struct task_struct *task = READ_ONCE(trans->locking_wait.task);
 	unsigned l, idx;
 
 	if (!out->nr_tabstops) {
@@ -3122,7 +3051,7 @@ void bch2_btree_trans_to_text(struct printbuf *out, struct btree_trans *trans)
 		printbuf_tabstop_push(out, 32);
 	}
 
-	prt_printf(out, "%i %s\n", trans->locking_wait.task->pid, trans->fn);
+	prt_printf(out, "%i %s\n", task ? task->pid : 0, trans->fn);
 
 	trans_for_each_path_safe(trans, path, idx) {
 		if (!path->nodes_locked)
@@ -3132,7 +3061,7 @@ void bch2_btree_trans_to_text(struct printbuf *out, struct btree_trans *trans)
 		       path->idx,
 		       path->cached ? 'c' : 'b',
 		       path->level,
-		       bch2_btree_ids[path->btree_id]);
+		       bch2_btree_id_str(path->btree_id));
 		bch2_bpos_to_text(out, path->pos);
 		prt_newline(out);