-
由 Jeff Layton 提交于
Currently, we could have an open-by-handle (or NFS server) call into the filesystem and start working with an inode before it's properly filled out. Don't clear I_NEW until we have filled out the inode, and discard it properly if that fails. Note that we occasionally take an extra reference to the inode to ensure that we don't put the last reference in discard_new_inode, but rather leave it for ceph_async_iput. Signed-off-by: NJeff Layton <jlayton@kernel.org> Signed-off-by: NIlya Dryomov <idryomov@gmail.com>
893e456b