Re: [PATCH v2 4/5] rtc: tps6586x: Fix initial enable_irq/disable_irq balance

Re: [PATCH v2 4/5] rtc: tps6586x: Fix initial enable_irq/disable_irq balance

[Jon Hunter]

Hi Esben,

On 16/05/2025 08:23, Esben Haabendal wrote:
> Interrupts are automatically enabled when requested, so we need to
> initialize irq_en accordingly to avoid causing an unbalanced enable
> warning.
> 
> Signed-off-by: Esben Haabendal <esben@geanix.com>
> ---
>   drivers/rtc/rtc-tps6586x.c | 1 +
>   1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/rtc/rtc-tps6586x.c b/drivers/rtc/rtc-tps6586x.c
> index 54c8429b16bfcc692b1f4d5404f0c42f720e93b4..76ecf7b798f0de22aa89a552a263b473ab3065ef 100644
> --- a/drivers/rtc/rtc-tps6586x.c
> +++ b/drivers/rtc/rtc-tps6586x.c
> @@ -258,6 +258,7 @@ static int tps6586x_rtc_probe(struct platform_device *pdev)
>   
>   	irq_set_status_flags(rtc->irq, IRQ_NOAUTOEN);
>   
> +	rtc->irq_en = true;
>   	ret = devm_request_threaded_irq(&pdev->dev, rtc->irq, NULL,
>   				tps6586x_rtc_irq,
>   				IRQF_ONESHOT,
> 

I have bisected a suspend regression on one of our Tegra20 boards (that 
uses this driver) and bisect is pointing to this commit. Reverting this 
commit fixes the problem.

Looking at the above I see that the flag IRQ_NOAUTOEN is being set and 
so now with your change we never enable the interrupt. Hence, the 
wake-ups are now broken and suspend testing fails. So it would seem best 
to revert this.

BTW, I looked at the change to the CPCAP driver and that driver actively 
disables the IRQ after requesting it and so I am wondering if that will 
also have alarm issues? I don't have a board with CPCAP to test.

Thanks
Jon

-- 
nvpublic

Re: [PATCH v2 4/5] rtc: tps6586x: Fix initial enable_irq/disable_irq balance

[Alexandre Belloni]

On 29/10/2025 09:20:17+0000, Jon Hunter wrote:
> Hi Esben,
> 
> On 16/05/2025 08:23, Esben Haabendal wrote:
> > Interrupts are automatically enabled when requested, so we need to
> > initialize irq_en accordingly to avoid causing an unbalanced enable
> > warning.
> > 
> > Signed-off-by: Esben Haabendal <esben@geanix.com>
> > ---
> >   drivers/rtc/rtc-tps6586x.c | 1 +
> >   1 file changed, 1 insertion(+)
> > 
> > diff --git a/drivers/rtc/rtc-tps6586x.c b/drivers/rtc/rtc-tps6586x.c
> > index 54c8429b16bfcc692b1f4d5404f0c42f720e93b4..76ecf7b798f0de22aa89a552a263b473ab3065ef 100644
> > --- a/drivers/rtc/rtc-tps6586x.c
> > +++ b/drivers/rtc/rtc-tps6586x.c
> > @@ -258,6 +258,7 @@ static int tps6586x_rtc_probe(struct platform_device *pdev)
> >   	irq_set_status_flags(rtc->irq, IRQ_NOAUTOEN);
> > +	rtc->irq_en = true;
> >   	ret = devm_request_threaded_irq(&pdev->dev, rtc->irq, NULL,
> >   				tps6586x_rtc_irq,
> >   				IRQF_ONESHOT,
> > 
> 
> I have bisected a suspend regression on one of our Tegra20 boards (that uses
> this driver) and bisect is pointing to this commit. Reverting this commit
> fixes the problem.
> 
> Looking at the above I see that the flag IRQ_NOAUTOEN is being set and so
> now with your change we never enable the interrupt. Hence, the wake-ups are
> now broken and suspend testing fails. So it would seem best to revert this.
> 
> BTW, I looked at the change to the CPCAP driver and that driver actively
> disables the IRQ after requesting it and so I am wondering if that will also
> have alarm issues? I don't have a board with CPCAP to test.
> 

That's right, I guess you can send reverts.


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

Re: [PATCH v2 4/5] rtc: tps6586x: Fix initial enable_irq/disable_irq balance

[Jon Hunter]

On 29/10/2025 10:39, Alexandre Belloni wrote:
> On 29/10/2025 09:20:17+0000, Jon Hunter wrote:
>> Hi Esben,
>>
>> On 16/05/2025 08:23, Esben Haabendal wrote:
>>> Interrupts are automatically enabled when requested, so we need to
>>> initialize irq_en accordingly to avoid causing an unbalanced enable
>>> warning.
>>>
>>> Signed-off-by: Esben Haabendal <esben@geanix.com>
>>> ---
>>>    drivers/rtc/rtc-tps6586x.c | 1 +
>>>    1 file changed, 1 insertion(+)
>>>
>>> diff --git a/drivers/rtc/rtc-tps6586x.c b/drivers/rtc/rtc-tps6586x.c
>>> index 54c8429b16bfcc692b1f4d5404f0c42f720e93b4..76ecf7b798f0de22aa89a552a263b473ab3065ef 100644
>>> --- a/drivers/rtc/rtc-tps6586x.c
>>> +++ b/drivers/rtc/rtc-tps6586x.c
>>> @@ -258,6 +258,7 @@ static int tps6586x_rtc_probe(struct platform_device *pdev)
>>>    	irq_set_status_flags(rtc->irq, IRQ_NOAUTOEN);
>>> +	rtc->irq_en = true;
>>>    	ret = devm_request_threaded_irq(&pdev->dev, rtc->irq, NULL,
>>>    				tps6586x_rtc_irq,
>>>    				IRQF_ONESHOT,
>>>
>>
>> I have bisected a suspend regression on one of our Tegra20 boards (that uses
>> this driver) and bisect is pointing to this commit. Reverting this commit
>> fixes the problem.
>>
>> Looking at the above I see that the flag IRQ_NOAUTOEN is being set and so
>> now with your change we never enable the interrupt. Hence, the wake-ups are
>> now broken and suspend testing fails. So it would seem best to revert this.
>>
>> BTW, I looked at the change to the CPCAP driver and that driver actively
>> disables the IRQ after requesting it and so I am wondering if that will also
>> have alarm issues? I don't have a board with CPCAP to test.
>>
> 
> That's right, I guess you can send reverts.

You guessed right! I certainly can. I am happy to send a revert for this 
because I can clearly test and verify this. I am not sure if you and 
Esben want to review the rest of the series.

Jon

-- 
nvpublic