On Fri, 2023-07-14 at 15:06 +0200, Markus Elfring wrote: > > The device_for_each_child_node() loop in bgx_init_of_phy() > > function should have fwnode_handle_put() before break which could > > avoid resource leaks. This patch could fix this bug. > > Are imperative change descriptions still preferred? Yes. The commit message should be re-phrased. More importantly, it looks like the relevant reference is already released by of_node_put() and the additional fwnode_handle_put() will cause a reference underflow. This patch does not look correct to me. Cheers, Paolo
