On 3 June 2016 at 11:04, Sudeep Holla <sudeep.holla@xxxxxxx> wrote:
>
>
> On 02/06/16 14:53, Linus Walleij wrote:
>>
>> commit 93a6f9168f2f ("rtc: ab8500: remove misuse of IRQF_NO_SUSPEND flag")
>> introduced the dev_pm_set_wake_irq() call to register a wake
>> IRQ for the AB8500 driver.
>>
>> However this causes a regression since device_init_wakeup() must be
>> called *after* dev_pm_set_wake_irq() not *before* it. Before this
>> patch we get an error message like this during system resume from
>> sleep:
I think it's actually the opposite. device_init_wakeup() must be
called *before* dev_pm_set_wake_irq(), as otherwise
dev_pm_set_wake_irq() will fail.
Indeed this probably "solves" the problem for you, although only by
hiding it, as there is no error check of the return code from
dev_pm_set_wake_irq().
>>
>
> I am unable to understand this. Because it's the exact same sequence in
> rtc-pl031.c which works fine for me on ARM64 Juno platform. So I am
> struggle to understand the linkage of the backtrace to this calling
> sequence as I can't see any dependency. Let me know if you have already
> figured out where exactly does it go wrong.
>
> Prior to commit 93a6f9168f2f, set_irq_wake was never tested and hence
> one possible reason I can think of is set_irq_wake_real may be failing
> on suspend but the error value is ignored.
Yes, something else is wrong here!
Kind regards
Uffe
--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html