[PATCH] rtc: stm32: Handle single EXTI IRQ as wake up source

[Amelie Delaunay]

Hi Marek,

On 5/18/23 02:33, Marek Vasut wrote:
> The arch/arm/boot/dts/stm32mp151.dtsi specifies one interrupt for the
> RTC node, while the expectation of the RTC driver is two interrupts on
> STM32MP15xx SoC, one connected to GIC interrupt controller and another
> one to EXTI. Per STM32MP15xx reference manual, the two interrupts serve
> the same purpose, except the EXTI one can also wake the system up. The
> EXTI driver already internally handles this GIC and EXTI duality and
> maps the EXTI interrupt onto GIC during runtime, and only uses the EXTI
> for wake up functionality.
> 
> Therefore, fix the driver such that if on STM32MP15xx there is only one
> interrupt specified in the DT, use that interrupt as EXTI interrupt and
> set it as wake up source.
> 
> This fixes the following warning in the kernel log on STM32MP15xx:
> "
> stm32_rtc 5c004000.rtc: error -ENXIO: IRQ index 1 not found
> stm32_rtc 5c004000.rtc: alarm can't wake up the system: -6
> "
> 
> This also fixes the system wake up via built-in RTC using e.g.:
> $ rtcwake -s 5 -m mem
> 
> Fixes: b72252b6580c ("rtc: stm32: add stm32mp1 rtc support")
> Signed-off-by: Marek Vasut <marex at denx.de>
> ---
> Cc: Alessandro Zummo <a.zummo at towertech.it>
> Cc: Alexandre Belloni <alexandre.belloni at bootlin.com>
> Cc: Alexandre Torgue <alexandre.torgue at foss.st.com>
> Cc: Amelie DELAUNAY <amelie.delaunay at foss.st.com>
> Cc: Maxime Coquelin <mcoquelin.stm32 at gmail.com>
> Cc: linux-arm-kernel at lists.infradead.org
> Cc: linux-rtc at vger.kernel.org
> Cc: linux-stm32 at st-md-mailman.stormreply.com
> ---
>   drivers/rtc/rtc-stm32.c | 9 +++++----
>   1 file changed, 5 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/rtc/rtc-stm32.c b/drivers/rtc/rtc-stm32.c
> index 229cb2847cc48..72994b9f95319 100644
> --- a/drivers/rtc/rtc-stm32.c
> +++ b/drivers/rtc/rtc-stm32.c
> @@ -780,14 +780,15 @@ static int stm32_rtc_probe(struct platform_device *pdev)
>   
>   	ret = device_init_wakeup(&pdev->dev, true);
>   	if (rtc->data->has_wakeirq) {
> -		rtc->wakeirq_alarm = platform_get_irq(pdev, 1);
> +		rtc->wakeirq_alarm = platform_get_irq_optional(pdev, 1);
>   		if (rtc->wakeirq_alarm > 0) {
>   			ret = dev_pm_set_dedicated_wake_irq(&pdev->dev,
>   							    rtc->wakeirq_alarm);
> -		} else {
> +		} else if (rtc->wakeirq_alarm == -EPROBE_DEFER) {
>   			ret = rtc->wakeirq_alarm;
> -			if (rtc->wakeirq_alarm == -EPROBE_DEFER)
> -				goto err;
> +			goto err;
> +		} else {
> +			ret = dev_pm_set_wake_irq(&pdev->dev, rtc->irq_alarm);
>   		}
>   	}
>   	if (ret)

In our downstream [1], dedicated wakeup irq management is dropped: it is 
neither described in st,stm32-rtc bindings nor used in STM32Fx, STM32Hx, 
STM32MP1x device trees.
The pointed patch is going to be upstreamed.

[1] 
https://github.com/STMicroelectronics/linux/commit/5a45e1f100d59c33b6b50fe98c0f62862bd6f3d2

Regards,
Amelie