Re: [PATCH] mfd: tps6586x: Handle interrupts on suspend

From: Dmitry Osipenko
Date: Thu Oct 25 2018 - 09:01:28 EST

On 10/24/18 5:19 PM, Jon Hunter wrote:
> On 24/10/2018 13:44, Dmitry Osipenko wrote:
>> On 10/24/18 1:49 PM, Jon Hunter wrote:
>>> On 22/10/2018 12:19, Dmitry Osipenko wrote:
>>>> On 10/22/18 12:52 PM, Thierry Reding wrote:
>>>>> On Fri, Oct 19, 2018 at 02:22:53PM +0100, Jon Hunter wrote:
>>>>>> From: Jonathan Hunter <jonathanh@xxxxxxxxxx>
>>>>>> The tps6586x driver creates an irqchip that is used by its various child
>>>>>> devices for managing interrupts. The tps6586x-rtc device is one of its
>>>>>> children that uses the tps6586x irqchip. When using the tps6586x-rtc as
>>>>>> a wake-up device from suspend, the following is seen:
>>>>>> PM: Syncing filesystems ... done.
>>>>>> Freezing user space processes ... (elapsed 0.001 seconds) done.
>>>>>> OOM killer disabled.
>>>>>> Freezing remaining freezable tasks ... (elapsed 0.000 seconds) done.
>>>>>> Disabling non-boot CPUs ...
>>>>>> Entering suspend state LP1
>>>>>> Enabling non-boot CPUs ...
>>>>>> CPU1 is up
>>>>>> tps6586x 3-0034: failed to read interrupt status
>>>>>> tps6586x 3-0034: failed to read interrupt status
>>>>>> The reason why the tps6586x interrupt status cannot be read is because
>>>>>> the tps6586x interrupt is not masked during suspend and when the
>>>>>> tps6586x-rtc interrupt occurs, to wake-up the device, the interrupt is
>>>>>> seen before the i2c controller has been resumed in order to read the
>>>>>> tps6586x interrupt status.
>>>>>> The tps6586x-rtc driver sets it's interrupt as a wake-up source during
>>>>>> suspend, which gets propagated to the parent tps6586x interrupt.
>>>>>> However, the tps6586x-rtc driver cannot disable it's interrupt during
>>>>>> suspend otherwise we would never be woken up and so the tps6586x must
>>>>>> disable it's interrupt instead.
>>>>>> Prevent the tps6586x interrupt handler from executing on exiting suspend
>>>>>> before the i2c controller has been resumed by disabling the tps6586x
>>>>>> interrupt on entering suspend and re-enabling it on resuming from
>>>>>> suspend.
>>>>>> Cc: stable@xxxxxxxxxxxxxxx
>>>>>> Signed-off-by: Jon Hunter <jonathanh@xxxxxxxxxx>
>>>>>> ---
>>>>>> drivers/mfd/tps6586x.c | 24 ++++++++++++++++++++++++
>>>>>> 1 file changed, 24 insertions(+)
>>>>> So does this mean that the SPI interrupt for the PMIC can still be a
>>>>> wakeup source even if it is masked? This is slightly odd because now
>>>>> you're saying that this does work while it doesn't work for the RTC
>>>>> interrupt. So is this an implementation quirk of the LIC/GIC on Tegra
>>>>> which doesn't extend to the TPS6586x? Or am I missing something?
>>>> What is the expected behaviour of IRQ disabling? Should it disable wakeup ability or only mask IRQ handling?
>>> I believe only mask the interrupt. However, the caveat here could be if
>>> the parent interrupt controller actually supports wake-up. For Tegra it
>>> is the LIC that handles the wake-up.
>>>> Couple months ago disabling of IRQ was disabling the wakeup, now something has been changed in kernel and wakeup isn't getting disabled. So either there was a bug before that was fixed or there is a bug now.
>>> Are you sure you were disabling the PMIC host interrupt? If you disable
>>> the RTC interrupt in the PMIC's RTC driver, then this will prevent the
>>> wake-up from occurring because you are masking the interrupt within the
>>> PMIC and so it will never generate an interrupt to the host.
>> I'm pretty sure (but not 100%) that was trying the same change as in your patch and it didn't work sometime before. If disable_irq() shouldn't disable wakeup, then everything is perfect now.
> Please note that this is very similar to the following fix where I
> experienced the same problem with another PMIC a couple years back ...
> 35deff7eb212 ("mfd: as3722: Handle interrupts on suspend")
> I did not bother setting the enable/disable_irq_wake() for the tps6586x
> host irq during resume/suspend because the irqchip for the tps6586x has
> an irq_set_wake function that propagates the wake enable/disable.

Okay, thanks.