Re: [PATCH] implement pm_ops.valid for everybody

["Rafael J. Wysocki" <rjw@sisk.pl>]

On Saturday, 24 March 2007 01:01, Pavel Machek wrote:
> Hi!
> 
> (I do not want to get into this flamewar).

I'd rather call it a misunderstanding.

> > > That's a false choice, when you "mean" anything more than
> > > fairly broad behavioral expectations:  STR saves more power
> > > than "standby", and transitions to/from STR take more
> > > time than to/from "standby".
> > 
> > So be it.
> > 
> > Assume that the user does 'echo standby > /sys/power/state'.  I think he can
> > expect that in such a case we'll freeze tasks and put devices into low-power
> > states and when he wakes up the system (BTW, I think the method of waking
> > up can be treated as a differentiating factor) he should be able to continue
> > from where he stopped after a little time.  Fine.
> > 
> > Now, we have to make that happen.  After we have frozen tasks, we need to
> > call something like device_suspend(some_argument) where the argument should
> > tell drivers what to do.  Say we use something like PMSG_STANDBY and now
> 
> We would add another field to that struct, distingushing "mem" and
> "standby". And meaning for the drivers would be "try to save a lot of
> power, but keep the latency low" for standby, vs. "save as much power
> as possible" for mem.

Well, I think we can add more PMSG_* constants just fine.  Still, for a driver,
the meaning of PMSG_WHATEVER should be clear.

I agree that whatever is done in pm_ops->enter(some_state) has to depend on
the platform for quite obvious reasons, but the PMSG_* have some meaning
beyond the platform-specific code.

Greetings,
Rafael