pgf at laptop.org
Tue Jun 23 11:52:40 EDT 2009
> Paul Fox wrote:
> > but what thinking about this has really made me realize is that i
> > don't fully understand how lid wakeups work, even for the
> > lid-open case that we do support. it seems that EC involvement
> > is required, even though it's clear from the code in olpc-pm.c
> > (and the history in #5703) that the geode gpio connections are
> > key to the process. are there cases where the geode isn't awake
> > enough to force a wakeup by itself?
> > (i'm talking exclusively about gen1 here, btw.)
> The EC is involved because the EC controls the DCON power, its just not
> in charge. You have to drop power to the DCON because the DCON's "low
> power" mode does not save you much. So the kernel tells the EC to turn
> the DCON on or off.
> On a B1 the lid signal was connected to both the geode and the EC but
> when we moved things around in later revs it ended up only run to the
> geode. (We needed EC IO) The geode lid events turned out to be
> unreliable (for a never fully understood reason) and we ended up with
> the logic that ohm does now.
thank you. i was misled by the EC code that still looks at the
(now-disconnected) lid signal, and by the (now-depopulated) connection
to that gpio.
> in Gen 1.5 we expect the gpio wakeups to work correctly (which is
> someting we need to test soon) and lid wakeups won't be so messy.
paul fox, pgf at laptop.org
More information about the Devel