[PATCH 3/5] serial: imx: avoid spinlock recursion deadlock

[Huang Shijie]

于 2014年05月09日 23:19, dean_jenkins at mentor.com 写道:
> From: Andy Lowe<andy_lowe at mentor.com>
>
> The following deadlock has been observed:
>
> imx_int() {
>    imx_txint() {
>      spin_lock_irqsave(&sport->port.lock,flags);
>      /* ^^^uart_port spinlock taken in imx_txint */
>      imx_transmit_buffer() {
>        uart_write_wakeup(&sport->port) {
>          tty_wakeup() {
>            hci_uart_tty_wakeup() {
>              hci_uart_tx_wakeup() {
>                uart_write() {
>                  spin_lock_irqsave(&port->lock, flags);
>                  /* ^^^deadlock here when spinlock is taken again */
>                    .
>                    .
>                    .
>                  spin_unlock_irqrestore(&port->lock, flags);
>                }
>              }
>            }
>          }
>        }
>      }
>      spin_unlock_irqrestore(&sport->port.lock,flags);
>    }
> }
>
> To correct this call uart_write_wakeup() at the end of imx_txint() after
> the uart_port spinlock is unlocked.
>
> Signed-off-by: Andy Lowe<andy_lowe at mentor.com>
> Signed-off-by: Dirk Behme<dirk.behme at de.bosch.com>
> ---
>   drivers/tty/serial/imx.c |    7 ++++++-
>   1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c
> index abe31ad..cc79706 100644
> --- a/drivers/tty/serial/imx.c
> +++ b/drivers/tty/serial/imx.c
> @@ -636,8 +636,13 @@ static irqreturn_t imx_txint(int irq, void *dev_id)
>
>   	imx_transmit_buffer(sport);
>
> -	if (uart_circ_chars_pending(xmit)<  WAKEUP_CHARS)
> +	if (uart_circ_chars_pending(xmit)<  WAKEUP_CHARS) {
> +		spin_unlock_irqrestore(&sport->port.lock, flags);
>   		uart_write_wakeup(&sport->port);
> +	} else
> +		spin_unlock_irqrestore(&sport->port.lock, flags);
> +
> +	return IRQ_HANDLED;
>
>   out:
>   	spin_unlock_irqrestore(&sport->port.lock, flags);
I think this patch :

https://lkml.org/lkml/2014/3/20/623

has fixed this deadlock.

We can ignore this patch now.

thanks
Huang Shijie