Re: [ACPI] S3 resume and Wake events

[Pavel M. <pa...@uc...>]

Hi!

> I am attempting to resume from S3 using a magic packet from an NIC. The NIC
> asserts the PME# and the status bit gets set in the I/O controller, but it
> turns out that the ACPI suspend code in the kernel disables *all* GPEs before
> sleeping and so I can't wake the system up after it goes to sleep.
> 
> Tracing through the code (on the 2.6.4 stock kernel), at one point, the
> function acpi_enter_sleep_state() calls acpi_disable_non_wakeup_gpes() which
> disables all GPEs, even the ones that are potentially the wake-up events like
> the PME_EN bit.

This needs some work. API should be invented to tell the kernel what
GPEs may wake us up.
									Pavel
-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]