Re: [PATCH 2/3] regmap: implement irq chip suspend/resume operations

[Stephen Warren <swarren@wwwdotorg.org>]

On 07/30/2012 11:38 AM, Mark Brown wrote:
> On Mon, Jul 30, 2012 at 11:10:30AM -0600, Stephen Warren wrote:
> 
>> hence exit sleep. If we are to port that code into the regmap-irq core,
>> it seems to make sense to have enable_base==wake_base, since the same
>> register truly is being used for both enable/wakeup-enable, just
>> time-multiplexed.
> 
> This would mean that we have to go round every single driver that
> doesn't have physical wake support and add a setting for the wake
> registers (which seems pointless given that the core can just as well
> figure this out from the fact that it's not had any wake registers
> specified) and we then have to add special cases for this in the core
> code.  This doesn't seem like great API design, it's not conveneint for
> either side of the interface and it's error prone.
> 
>> Or, perhaps the IRQ core already disables all non-wake interrupts for
>> us, so the driver doesn't have to do this, and we can just drop that
>> code completely?
> 
> IIRC it does actually do this, I'd need to check though.

It looks like the answer here is to set irq_chip flags
IRQCHIP_MASK_ON_SUSPEND and IRQCHIP_SKIP_SET_WAKE. Perhaps regmap-irq
should do this automatically if (!regmap_irq_chip.wake_base)?

For reference, these flags are implemented in
kernel/irq/pm.c:check_wakeup_irqs() and
kernel/irq/manage.c:set_irq_wake_real().