Hi Brian,
On 08/23/2017 09:57 AM, Brian Norris wrote:
Hi Jeffy,
On Wed, Aug 23, 2017 at 09:32:39AM +0800, Jeffy Chen wrote:
and for eage irq, maybe we should enable it right after(or before)
the driver activate wake function(for example activate WOWLAN or
WOLAN), otherwise would it be possible to miss some irqs(triggered
before we actually enable the wake irq)?
I already mentioned this: for the PCI case, the specification explicitly
says that the WAKE# pin must remain asserted until the system wakes and
resets the link. So we don't have this problem.
Sorry, i means for other use cases of wakeirq, for example sdio wifi
But it is probably still useful to make sure there's a well-defined
point at which these interrupts are armed, so that if a device driver
does care, it can account for that. Just before suspend_noirq (as it is
today) is probably fine, so if there's some device-level handling that
needs to happen before we get to suspend (but after the wakeirq is
armed), it can go in the device or bus {suspend,resume}_noirq callbacks.
Yes, then we may need to handle "disable level irq" job in the irq
handler(or runtime resume callback as current wakeirq API suggested) for
irqs received before suspend devices irqs.
Brian
