On 01/06/2023 11:44, Tony Lindgren wrote: > Hi, > > * Steven Price <steven.price@xxxxxxx> [230601 10:04]: >> I haven't studied this change in detail, but I assume the bug is that >> serial_base_port_device_remove() shouldn't be dropping port_mutex. The >> below hack gets my board booting again. > > You're right. I wonder how I managed to miss that.. Care to post a proper > fix for this or do you want me to post it? I'll post a proper fix shortly. Thanks for the confirmation of the fix. >> Thanks, >> >> Steve >> >> Hack fix: >> ----8<---- >> diff --git a/drivers/tty/serial/serial_core.c b/drivers/tty/serial/serial_core.c >> index 29bd5ede0b25..044e4853341a 100644 >> --- a/drivers/tty/serial/serial_core.c >> +++ b/drivers/tty/serial/serial_core.c >> @@ -3234,8 +3234,7 @@ static void serial_core_remove_one_port(struct uart_driver *drv, >> wait_event(state->remove_wait, !atomic_read(&state->refcount)); >> state->uart_port = NULL; >> mutex_unlock(&port->mutex); >> -out: >> - mutex_unlock(&port_mutex); >> +out:; >> } > > Seems you can remove out here and just do a return earlier instead of goto. Yes, this was just the smallest change. I'll do it properly with an early return in the proper patch. Thanks, Steve > Regards, > > Tony
