SMP suspend/resume.

[Takanori Watanabe]

>I've tried to build patch with recent CURRENT and get 
>SMP redefined for acpi.c and acpi_wakeup.c
>
>Was it expected ? 

Expected. I tested with kernel module, so I define directly for now.
I want to make it work without SMP defined by cleaning up and moving 
the smp dependent code to mp_machdep.c .

>On Wed, 2008-05-21 at 18:21 +0400, Vladimir Grebenschikov wrote:
>> On Tue, 2008-05-13 at 20:25 +0900, takawata at init-main.com wrote:
>> > Hi, I managed to make suspend and resume work on SMP system.
>> > The patch following is a bit crude patch, but it begin=20
>> > to work on my ThinkPad X61 (core2duo system).
>>=20
>> Tried patch on T60, from text console,=20
>> on acpiconf -s3
>> it shows multiple times "forward_wakeup: Idle processor not found" on
>> console and then friezed hard (even DDB does not works). But nothing was
>> suspended actually ( before system goes to sleep but did not awakes)
>>=20
>> Any hints ?
>
>In some conditions (single-user, no additional drivers) it shows:
>
>forward_wakeup: Idle processor not found
>PREP_WALK
>LEAVE_SLEEP_STATE
>DEVICE_RESUME
>=EF=BB=BFforward_wakeup: Idle processor not found
>ENABLE_FIXED_EVENT
>=EF=BB=BFforward_wakeup: Idle processor not found
>=EF=BB=BFforward_wakeup: Idle processor not found
>=EF=BB=BFforward_wakeup: Idle processor not found
>=EF=BB=BFforward_wakeup: Idle processor not found
>=EF=BB=BFforward_wakeup: Idle processor not found
>... etc ...
>
>sometimes it shows just many
>=EF=BB=BFforward_wakeup: Idle processor not found
>
>
>and in any case locks hard and did not enter S3
>
>But RELENG_7 (without patch) normally enters to S3, but failed to wake.

I tested it on sched_ule. So I know littele about sched_4bsd.
The message may caused on forward_wakeup on sched_4bsd, so
how about kern.sched.ipiwakeup.forward_wakeup set to 0.