bcachefs: Fix failure to allocate btree node in cache

The error code when we fail to allocate a node in the btree node cache
doesn't make it to bch2_btree_path_traverse_all(). Instead, we need to
stash a flag in btree_trans so we know we have to take the cannibalize
lock.

Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

4 files changed, 23 insertions, 17 deletions

diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c
index 36b82df79fc2..c17db1d07187 100644
--- a/fs/bcachefs/btree_cache.c
+++ b/fs/bcachefs/btree_cache.c
@@ -672,6 +672,15 @@ static noinline struct btree *bch2_btree_node_fill(struct bch_fs *c,
 	}
 
 	b = bch2_btree_node_mem_alloc(c);
+
+	if (trans && b == ERR_PTR(-ENOMEM)) {
+		trans->memory_allocation_failure = true;
+		trace_trans_restart_memory_allocation_failure(trans->fn,
+				_THIS_IP_, btree_id, &path->pos);
+		btree_trans_restart(trans);
+		return ERR_PTR(-EINTR);
+	}
+
 	if (IS_ERR(b))
 		return b;
 
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index c6c1c9da45f1..1015e89d2d68 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -1407,12 +1407,12 @@ err:
 static int btree_path_traverse_one(struct btree_trans *, struct btree_path *,
 				   unsigned, unsigned long);
 
-static int __btree_path_traverse_all(struct btree_trans *trans, int ret,
-				     unsigned long trace_ip)
+static int bch2_btree_path_traverse_all(struct btree_trans *trans)
 {
 	struct bch_fs *c = trans->c;
 	struct btree_path *path, *prev = NULL;
-	int i;
+	unsigned long trace_ip = _RET_IP_;
+	int i, ret = 0;
 
 	if (trans->in_traverse_all)
 		return -EINTR;
@@ -1441,7 +1441,7 @@ retry_all:
 	bch2_trans_unlock(trans);
 	cond_resched();
 
-	if (unlikely(ret == -ENOMEM)) {
+	if (unlikely(trans->memory_allocation_failure)) {
 		struct closure cl;
 
 		closure_init_stack(&cl);
@@ -1452,11 +1452,6 @@ retry_all:
 		} while (ret);
 	}
 
-	if (unlikely(ret == -EIO))
-		goto out;
-
-	BUG_ON(ret && ret != -EINTR);
-
 	/* Now, redo traversals in correct order: */
 	i = 0;
 	while (i < trans->nr_sorted) {
@@ -1482,7 +1477,7 @@ retry_all:
 	 */
 	trans_for_each_path(trans, path)
 		BUG_ON(path->uptodate >= BTREE_ITER_NEED_TRAVERSE);
-out:
+
 	bch2_btree_cache_cannibalize_unlock(c);
 
 	trans->in_traverse_all = false;
@@ -1491,11 +1486,6 @@ out:
 	return ret;
 }
 
-static int bch2_btree_path_traverse_all(struct btree_trans *trans)
-{
-	return __btree_path_traverse_all(trans, 0, _RET_IP_);
-}
-
 static inline bool btree_path_good_node(struct btree_trans *trans,
 					struct btree_path *path,
 					unsigned l, int check_pos)
@@ -1619,8 +1609,6 @@ out:
 	return ret;
 }
 
-static int __btree_path_traverse_all(struct btree_trans *, int, unsigned long);
-
 int __must_check bch2_btree_path_traverse(struct btree_trans *trans,
 					  struct btree_path *path, unsigned flags)
 {
diff --git a/fs/bcachefs/btree_types.h b/fs/bcachefs/btree_types.h
index 7e5b70f60444..89c0d2272d91 100644
--- a/fs/bcachefs/btree_types.h
+++ b/fs/bcachefs/btree_types.h
@@ -393,6 +393,7 @@ struct btree_trans {
 	bool			in_traverse_all:1;
 	bool			restarted:1;
 	bool			paths_sorted:1;
+	bool			memory_allocation_failure:1;
 	bool			journal_transaction_names:1;
 	bool			journal_replay_not_finished:1;
 	/*
diff --git a/fs/bcachefs/trace.h b/fs/bcachefs/trace.h
index 64b7d9364fd9..b35022dc66c2 100644
--- a/fs/bcachefs/trace.h
+++ b/fs/bcachefs/trace.h
@@ -802,6 +802,14 @@ DEFINE_EVENT(transaction_restart_iter,	trans_restart_traverse,
 	TP_ARGS(trans_fn, caller_ip, btree_id, pos)
 );
 
+DEFINE_EVENT(transaction_restart_iter,	trans_restart_memory_allocation_failure,
+	TP_PROTO(const char *trans_fn,
+		 unsigned long caller_ip,
+		 enum btree_id btree_id,
+		 struct bpos *pos),
+	TP_ARGS(trans_fn, caller_ip, btree_id, pos)
+);
+
 TRACE_EVENT(trans_restart_would_deadlock,
 	TP_PROTO(const char *trans_fn,
 		 unsigned long	caller_ip,