Re: [PATCH -v2] RTC: Add an alarm disable quirk

[John Stultz <john.stultz@linaro.org>]

On 07/22/2013 02:19 PM, Borislav Petkov wrote:
> On Mon, Jul 22, 2013 at 02:00:42PM -0700, John Stultz wrote:
>> Also, just to clarify, on the affected machines, with this patch,
>> wake-up alarm's will in effect be disabled, right?
> See below.
>
>>> @@ -385,6 +418,9 @@ static int cmos_alarm_irq_enable(struct device *dev, unsigned int enabled)
>>>   	if (!is_valid_irq(cmos->irq))
>>>   		return -EINVAL;
>>> +	if (!disable_alarm)
>>> +		return 0;
>>> +
>> Did you want this in cmos_alarm_irq_enable? Or cmos_irq_disable?
> You're right - the first version did call it only in rtc_alarm_disable()
> so I should move it to cmos_alarm_disable(). Will fix.
>
> Btw, I did some more runs on the weekend. It seems, the setting of the
> alarm interrupt bit in RTC_CONTROL doesn't matter. I dumped its contents
> on shutdown and I had cases where it was 0x22 (bit 5 set) and 0x2.

It doesn't matter in what way?  I get you're saying you had cases where 
the alarm irq was set and non-set on shutdown, but I'm confused as to 
what the result was. Did it boot up immediately on shutdown in both cases?

> So my hunch currently is us *not* disabling the alarm, doesn't make it
> reboot the box.
Just to avoid the double negatives, you're saying:
* Disabling the alarm seems to cause it power on immediately on shutdown.
* Leaving the alarm alone (even if its set?) doesn't seem to cause the 
immediate power-on

Does leaving it alone, cause eventual power-on if the wakeup alarm was 
set for some time (say 5 mins) in the future?

thanks
-john