Re: [PATCH 3/3] rtc: ds1685: disable interrupts when moving work to work queue

[Alexandre Belloni <alexandre.belloni@bootlin.com>]

On 11/04/2019 16:33:23+0200, Thomas Bogendoerfer wrote:
> Handling of extended interrupts (kickstart, wake-up, ram-clear) is
> moved off to a work queue, but the interrupts aren't acknowledged
> in the interrupt handler. This leads to a deadlock, if driver
> is used with interrupts. To fix this we now disable in irq handler
> and re-enable it after work queue is done.
> 

The correct fix to that seems to switch to a threaded interrupt handler.
Can you do that?

> Fixes: aaaf5fbf56f1 ("rtc: add driver for DS1685 family of real time clocks")
> Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
> ---
>  drivers/rtc/rtc-ds1685.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/rtc/rtc-ds1685.c b/drivers/rtc/rtc-ds1685.c
> index 929f28375b87..5dabfa57bd2a 100644
> --- a/drivers/rtc/rtc-ds1685.c
> +++ b/drivers/rtc/rtc-ds1685.c
> @@ -635,6 +635,7 @@ ds1685_rtc_irq_handler(int irq, void *dev_id)
>  			 * to be minimized.  Schedule them into a workqueue
>  			 * and inform the RTC core that the IRQs were handled.
>  			 */
> +			disable_irq_nosync(rtc->irq_num);
>  			spin_unlock(&rtc->lock);
>  			schedule_work(&rtc->work);
>  			rtc_update_irq(rtc->dev, 0, 0);
> @@ -741,6 +742,7 @@ ds1685_rtc_work_queue(struct work_struct *work)
>  	ds1685_rtc_switch_to_bank0(rtc);
>  
>  	mutex_unlock(rtc_mutex);
> +	enable_irq(rtc->irq_num);
>  }
>  /* ----------------------------------------------------------------------- */
>  
> -- 
> 2.13.7
> 

-- 
Alexandre Belloni, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com