On (12/12/18 12:42), Sergey Senozhatsky wrote: [..] > > >> [ 87.255156] CPU0 CPU1 > > >> [ 87.255813] ---- ---- > > >> [ 87.256460] lock(&port_lock_key); > > >> [ 87.256973] lock(console_owner); > > >> [ 87.257829] lock(&port_lock_key); > > >> [ 87.258680] lock(&obj_hash[i].lock); > > So it's like > > CPU0 CPU1 > > uart_shutdown() db->lock > uart_port->lock debug_print_object() > free_page() printk > debug_check_no_obj_freed uart_port->lock > db->lock > > > In this particular case we probably can just move free_page() > out of uart_port lock scope. Note that free_page()->MM can printk() > on its own. > [..] > > [1] https://lore.kernel.org/lkml/1542653726-5655-8-git-send-email-longman@xxxxxxxxxx/T/#u That said, I'd first try Waiman's patch. The one I suggested is more of a defense move - there are too many things happening under uart_port->lock. This is not the first time we see lockdep complaining about the way uart and the rest of the kernel interact. -ss
