From: Kent Overstreet <kent.overstreet@gmail.com>
Date: Thu, 22 Jul 2021 16:39:11 +0000 (-0400)
Subject: bcachefs: Ensure btree_iter_traverse() obeys iter->should_be_locked
X-Git-Url: http://git.maquefel.me/?a=commitdiff_plain;h=e829b7175b54c9ccdf4412baea592bc495bc97f1;p=linux.git

bcachefs: Ensure btree_iter_traverse() obeys iter->should_be_locked

iter->should_be_locked means that if bch2_btree_iter_relock() fails, we
need to restart the transaction.

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

diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index ac7a7c18ebe24..b7292f9320002 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -1403,6 +1403,15 @@ static int btree_iter_traverse_one(struct btree_iter *iter,
 	unsigned l, depth_want = iter->level;
 	int ret = 0;
 
+	/*
+	 * Ensure we obey iter->should_be_locked: if it's set, we can't unlock
+	 * and re-traverse the iterator without a transaction restart:
+	 */
+	if (iter->should_be_locked) {
+		ret = bch2_btree_iter_relock(iter, trace_ip) ? 0 : -EINTR;
+		goto out;
+	}
+
 	if (btree_iter_type(iter) == BTREE_ITER_CACHED) {
 		ret = bch2_btree_iter_traverse_cached(iter);
 		goto out;