Re: [PATCH] USB: g_serial: fix deadlock with PREEMPT_RT enabled

[Michael Thalmeier <michael.thalmeier@hale.at>]

On 2011-10-28 14:16, Michael Thalmeier wrote:

> In commit 44a0c0190b500ee6bcfc0976fe540f65dee2cd67 unlocking of spin_locks has
> been removed. This patch party restores this functionality.
> 
> Signed-off-by: Michael Thalmeier <michael.thalmeier@hale.at>
> ---
>  drivers/usb/gadget/u_serial.c |    9 +++++++++
>  1 files changed, 9 insertions(+), 0 deletions(-)
> 
> diff --git a/drivers/usb/gadget/u_serial.c b/drivers/usb/gadget/u_serial.c
> index 40f7716..a925b9e 100644
> --- a/drivers/usb/gadget/u_serial.c
> +++ b/drivers/usb/gadget/u_serial.c
> @@ -553,7 +553,16 @@ recycle:
>  	 * a workqueue, so we won't get callbacks and can hold port_lock
>  	 */
>  	if (tty && do_push) {
> +		/*
> +		 * Drop the lock here since it might end up calling
> +		 * gs_flush_chars, which takes the lock.
> +		 */
> +		spin_unlock_irq(&port->port_lock);
>  		tty_flip_buffer_push(tty);
> +		spin_lock_irq(&port->port_lock);
> +
> +		/* tty may have been closed */
> +		tty = port->port_tty;
>  	}
>  
>  

In the function tty_flip_buffer_push the PRREMPT_RT patch changes the
behaviour. Most serial drivers (those that don't set low_latency) assume
that work is only scheduled and not executed immediately in current context.
This patch only fixes this driver but other serial drivers likely have
similar problems.

Michael



--
Scanned by MailScanner.