[PATCH] spi: don't lock the bus in __spi_pump_messages if it is already locked

[Vignesh R]

On 03/18/2016 04:48 AM, Heiko Stübner wrote:
> External users may use spi_bus_lock to get exclusive access. This will
> also grab the bus_lock_mutex and may therefore result in a deadlock if
> __spi_pump_messages also tries to get the mutex.

I suggest bus_locked flag passed to __spi_pump_messages() by
__spi_sync() be updated to take care of the fact that spi_bus_lock() may
have acquired the mutex, so that below changes in __spi_pump_messages()
are not needed.

> 
> Fixes: 49023d2e4ead ("spi: core: Fix deadlock when sending messages")
> Signed-off-by: Heiko Stuebner <heiko at sntech.de>
> ---
>  drivers/spi/spi.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
> index de2f2f9..62f962b 100644
> --- a/drivers/spi/spi.c
> +++ b/drivers/spi/spi.c
> @@ -1160,7 +1160,7 @@ static void __spi_pump_messages(struct spi_master *master, bool in_kthread,
>  		}
>  	}
>  
> -	if (!bus_locked)
> +	if (!bus_locked && !master->bus_lock_flag)
>  		mutex_lock(&master->bus_lock_mutex);
>  
>  	trace_spi_message_start(master->cur_msg);
> @@ -1192,7 +1192,7 @@ static void __spi_pump_messages(struct spi_master *master, bool in_kthread,
>  	}
>  
>  out:
> -	if (!bus_locked)
> +	if (!bus_locked && !master->bus_lock_flag)
>  		mutex_unlock(&master->bus_lock_mutex);
>  
>  	/* Prod the scheduler in case transfer_one() was busy waiting */
> 

-- 
Regards
Vignesh