Commit dd908648 authored by Kent Overstreet's avatar Kent Overstreet
Browse files

bcachefs: Fix I_NEW warning in race path in bch2_inode_insert() 

discard_new_inode() is the correct interface for tearing down an indoe
that was fully created but not made visible to other threads, but it
expects I_NEW to be set, which we don't use.

Reported-by: https://github.com/koverstreet/bcachefs/issues/690


Fixes: bcachefs: Fix race path in bch2_inode_insert()
Signed-off-by: default avatarKent Overstreet <kent.overstreet@linux.dev>
parent 50479406

fs/bcachefs/fs.c

+10 −2
Original line number Diff line number Diff line
@@ -188,6 +188,12 @@ static struct bch_inode_info *bch2_inode_insert(struct bch_fs *c, struct bch_ino 
	BUG_ON(!old);



	if (unlikely(old != inode)) {

		/*

		 * bcachefs doesn't use I_NEW; we have no use for it since we

		 * only insert fully created inodes in the inode hash table. But

		 * discard_new_inode() expects it to be set...

		 */

		inode->v.i_flags |= I_NEW;

		discard_new_inode(&inode->v);

		inode = old;

	} else {
@@ -195,8 +201,10 @@ static struct bch_inode_info *bch2_inode_insert(struct bch_fs *c, struct bch_ino 
		list_add(&inode->ei_vfs_inode_list, &c->vfs_inodes_list);

		mutex_unlock(&c->vfs_inodes_lock);

		/*

		 * we really don't want insert_inode_locked2() to be setting

		 * I_NEW...

		 * Again, I_NEW makes no sense for bcachefs. This is only needed

		 * for clearing I_NEW, but since the inode was already fully

		 * created and initialized we didn't actually want

		 * inode_insert5() to set it for us.

		 */

		unlock_new_inode(&inode->v);

	}