bcachefs: Avoiding dropping/retaking write locks in bch2_btree_write_buffer_flush_one()

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

diff --git a/fs/bcachefs/btree_write_buffer.c b/fs/bcachefs/btree_write_buffer.c
index 7f3147e..f40ac36 100644 (file)
--- a/fs/bcachefs/btree_write_buffer.c
+++ b/fs/bcachefs/btree_write_buffer.c
@@ -48,6 +48,13 @@ static int bch2_btree_write_buffer_flush_one(struct btree_trans *trans,
        if (ret)
                return ret;
 
+       /*
+        * We can't clone a path that has write locks: unshare it now, before
+        * set_pos and traverse():
+        */
+       if (iter->path->ref > 1)
+               iter->path = __bch2_btree_path_make_mut(trans, iter->path, true, _THIS_IP_);
+
        path = iter->path;
 
        if (!*write_locked) {
@@ -67,15 +74,6 @@ static int bch2_btree_write_buffer_flush_one(struct btree_trans *trans,
 
        bch2_btree_insert_key_leaf(trans, path, &wb->k, wb->journal_seq);
        (*fast)++;
-
-       if (path->ref > 1) {
-               /*
-                * We can't clone a path that has write locks: if the path is
-                * shared, unlock before set_pos(), traverse():
-                */
-               bch2_btree_node_unlock_write(trans, path, path->l[0].b);
-               *write_locked = false;
-       }
        return 0;
 trans_commit:
        trans->journal_res.seq = wb->journal_seq;