joda@pdc.kth.se said:
> If I add a printf before and after the call to setpowstate, it prints
> as it should with interrupts

The set_powstate() shouldn't return until resume time - is this
what you see?

> but only the one before gets printed
> without interrupts

...because it doesn't awake from the get_powstate(), as was to
expect after your previous report - or am I misunderstanding
something?

In Linux mailing lists I've found that some Thinkpads issue
multiple SUSPEND requests. The NetBSD code seems to [try to]
catch this case, involving the "damn_fool_bios" variable.
Can you try to find out whether this happens?
("apmdebug" should be helpful)

best regards
Matthias