On Sat, 7 Mar 2015, Rafael J. Wysocki wrote:
> But this is part of a bigger picture. Namely, if a separete wakeup interrupt
> is required for a device, the device's power.can_wakeup flag cannot be set
> until that interrupt has been successfully requested. Also for devices that
> can signal wakeup via their own IO interrupts, it would make sense to allow
> those interrupts to be registered somehow as "wakeup interrupts".
>
> So I wonder if we can define a new struct along the lines of your
> struct wakeirq_source, but call it struct wake_irq and make it look
> something like this:
>
> struct wake_irq {
> struct device *dev;
> int irq;
> irq_handler_t handler;
> };
>
> Then, add a struct wake_irq pointer to struct dev_pm_info *and* to
> struct wakeup_source. Next, make dev_pm_request_wake_irq() allocate the
> structure and request the interrupt and only set the pointer to it from
> struct dev_pm_info *along* *with* power.can_wakeup if all that was
> successful.
>
> For devices that use their own IO IRQ for wakeup, we can add something
> like dev_pm_set_wake_irq() that will work analogously, but without requesting
> the interrupt. It will just set the dev and irq members of struct wake_irq
> and point struct dev_pm_info to it and set its power.can_wakeup flag.
>
> Then, device_wakeup_enable() will be able to see that the device has a
> wakeup IRQ and it may then point its own struct wake_irq pointer to that.
> The core may then use that pointer to trigger enable_irq_wake() for the
> IRQ in question and it will cover the devices that don't need separate
> wakeup interrupts too.
>
> Does that make sense to you?
Can we back up a little? What is the basic problem the two of you are
trying to solve?
Alan Stern
--
To unsubscribe from this list: send the line "unsubscribe linux-serial" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
