Re: [PATCH] rtc: zynqmp: Clear alarm interrupt status before interrupt enable

[Michal Simek <michal.simek@xilinx.com>]

On 11. 02. 20 11:39, Alexandre Belloni wrote:
> On 10/02/2020 12:48:14+0100, Michal Simek wrote:
>> Hi,
>>
>> čt 12. 12. 2019 v 14:01 odesílatel Srinivas Neeli
>> <srinivas.neeli@xilinx.com> napsal:
>>>
>>> Fix multiple occurring interrupts for alarm interrupt. RTC module doesn't
>>> clear the alarm interrupt status bit immediately after the interrupt is
>>> triggered.This is due to the sticky nature of the alarm interrupt status
>>> register. The alarm interrupt status register can be cleared only after
>>> the second counter outruns the set alarm value. To fix multiple spurious
>>> interrupts, disable alarm interrupt in the handler and clear the status
>>> bit before enabling the alarm interrupt.
>>>
>>> Fixes: 11143c19eb57 ("rtc: add xilinx zynqmp rtc driver")
>>> Signed-off-by: Srinivas Neeli <srinivas.neeli@xilinx.com>
>>> ---
>>>  drivers/rtc/rtc-zynqmp.c | 29 ++++++++++++++++++++++++-----
>>>  1 file changed, 24 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/drivers/rtc/rtc-zynqmp.c b/drivers/rtc/rtc-zynqmp.c
>>> index 5786866c09e9..d311e3ef1f21 100644
>>> --- a/drivers/rtc/rtc-zynqmp.c
>>> +++ b/drivers/rtc/rtc-zynqmp.c
>>> @@ -38,6 +38,8 @@
>>>
>>>  #define RTC_CALIB_DEF          0x198233
>>>  #define RTC_CALIB_MASK         0x1FFFFF
>>> +#define RTC_ALRM_MASK          BIT(1)
>>> +#define RTC_MSEC               1000
>>>
>>>  struct xlnx_rtc_dev {
>>>         struct rtc_device       *rtc;
>>> @@ -124,11 +126,28 @@ static int xlnx_rtc_alarm_irq_enable(struct device *dev, u32 enabled)
>>>  {
>>>         struct xlnx_rtc_dev *xrtcdev = dev_get_drvdata(dev);
>>>
>>
>> here shouldn't be empty line.
>>
>>> -       if (enabled)
>>> +       unsigned int status;
>>> +       ulong timeout;
>>> +
>>> +       timeout = jiffies + msecs_to_jiffies(RTC_MSEC);
>>> +
>>> +       if (enabled) {
>>> +               while (1) {
>>> +                       status = readl(xrtcdev->reg_base + RTC_INT_STS);
>>> +                       if (!((status & RTC_ALRM_MASK) == RTC_ALRM_MASK))
>>> +                               break;
>>> +
>>> +                       if (time_after_eq(jiffies, timeout)) {
>>> +                               dev_err(dev, "Time out occur, while clearing alarm status bit\n");
>>> +                               return -ETIMEDOUT;
>>> +                       }
>>> +                       writel(RTC_INT_ALRM, xrtcdev->reg_base + RTC_INT_STS);
>>> +               }
>>> +
>>>                 writel(RTC_INT_ALRM, xrtcdev->reg_base + RTC_INT_EN);
>>> -       else
>>> +       } else {
>>>                 writel(RTC_INT_ALRM, xrtcdev->reg_base + RTC_INT_DIS);
>>> -
>>> +       }
>>
>> And here it was good to have empty line.
>>
>>>         return 0;
>>>  }
>>>
>>> @@ -183,8 +202,8 @@ static irqreturn_t xlnx_rtc_interrupt(int irq, void *id)
>>>         if (!(status & (RTC_INT_SEC | RTC_INT_ALRM)))
>>>                 return IRQ_NONE;
>>>
>>> -       /* Clear RTC_INT_ALRM interrupt only */
>>> -       writel(RTC_INT_ALRM, xrtcdev->reg_base + RTC_INT_STS);
>>> +       /* Disable RTC_INT_ALRM interrupt only */
>>> +       writel(RTC_INT_ALRM, xrtcdev->reg_base + RTC_INT_DIS);
>>>
>>>         if (status & RTC_INT_ALRM)
>>>                 rtc_update_irq(xrtcdev->rtc, 1, RTC_IRQF | RTC_AF);
>>> --
>>> 2.7.4
>>
>> Other then these two above things look good.
>>
>> Alexandre: Any issue with this patch?
>>
> 
> No issue, I was kind of waiting for your review. I'll take the patch
> once your comments are addressed.

Ok. Thanks,
Michal