Re: [PATCH] ACPI: evaluate _PS3 when entering D3 Cold

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

On Thursday, April 05, 2012, Lin Ming wrote:
> On Sun, 2012-04-01 at 09:23 +0200, Rafael J. Wysocki wrote:
> > Hi,
> > 
> > Sorry for the delayed response, I've been travelling recently.
> > 
> > On Sunday, April 01, 2012, Lin Ming wrote:
> > > On Sun, 2012-04-01 at 13:56 +0800, Aaron Lu wrote:
> > > > Hi,
> > > > 
> > > > On Sun, Apr 01, 2012 at 01:27:33PM +0800, Lin Ming wrote:
> > > > > > -		if (device->power.states[state].flags.explicit_set) {
> > > > > > +		/* If state is D3 Cold, try to evaluate _PS3 first */
> > > > > > +		if (state == ACPI_STATE_D3_COLD) {
> > > > > > +			explicit_set = (ps - 1)->flags.explicit_set;
> > > > > > +			object_name[3] -= 1;
> > > > > > +		}
> > > > > 
> > > > > I'm not sure whether this works or not.
> > > > > 
> > > > > From ACPI spec,
> > > > > 
> > > > > _PS3 "is used to put the specific device into its D3hot or D3 state"
> > > > > 
> > > > > D3 neither means D3hot nor D3cold. It's an old term before D3hot and
> > > > > D3cold were introduced.
> > > > I guess D3 has to mean something, right? :-)
> > 
> > Well, not necessarily.
> > 
> > The problem is what state the _PS3 method puts the device into: D3_hot or
> > D3_cold.
> > 
> > Unfortunately, as far as I can say, ACPI 4.0 didn't specify any "official"
> > mapping between the "old" D3 and the "new" D3_{hod|cold} states, so we need to
> > figure out something.  In my opinion, the only reasonable approach is to
> > assume that the state _PS3 puts the device into is always D3_cold, becuase
> > _PS3 may remove power completely from the device.  It may not do that, but
> > we _must_ assume it does that in general.
> > 
> > > > Here is the problem, there is no _PR3 in AMD's implementation, just _PS3.
> > > > And since _S0W evaluates 4, I've to put this device into D3 cold state
> > > > with _PS3.
> > > > 
> > > > And the ACPI does have some words like:
> > > > 
> > > > ------
> > > > Platform/drivers must assume that the device will have power completely
> > > > removed when the device is place into “D3” via _PS3
> > 
> > Exactly.  What it means is basically "always reinitialize the device from
> > scratch if you have run _PS3 on it".  And that's what we should do.
> > 
> > > > ------
> > > > 
> > > > This is in section 7.2.11: _PR3.
> > > > 
> > > > > 
> > > > > Another problem:
> > > > > 
> > > > > With your patch, both D3hot and D3cold will evaluate _PS3, right?
> > > > > 
> > > > Yes.
> > > > 
> > > > > Will it have problem on AMD platform if you try to put ODD into D3hot
> > > > > state? _PS3 is evaluated, so it actually enters D3Cold state.
> > > > 
> > > > There is no D3 hot support for this device(from the firmware's
> > > > perspective), either it is at D0(via _PS0), or it will be at D3 cold(via
> > > > _PS3).
> > > 
> > > But this is the generic code. We can't only consider some special
> > > device.
> > > 
> > > Maybe we need some flag to tell which D3 state _PS3 is used for.
> > 
> > No, please.  As I said above, we need to reinitialize devices that _PS3 was
> > executed on, which is equivalent to saying that those devices were put into
> > D3_cold.
> > 
> > The only situation where a device can be put into ACPI D3_hot (which is not
> > the same as PCI D3_hot, mind you) is when:
> > 
> > (1) There is _PR3 listing some of the device's power resources as "on".
> > (2) The power resources listed by the _PR3 as "off" are turned off and the
> >     power resources listed by the _PR3 as "on" are left in the "on" state.
> 
> I don't understand item (2):
> 
> If the power resource is listed as "off", which means it's already
> turned off. Then why should it be turned off again?

Sorry, there are two lists.  One of them is returned by _PR0 (these
are needed to put the device into D0) and the second is returned by
_PR3 (if that exists) and these need to be "on" for the device to stay
in D3_hot (if they are "off", the device will go into D3_cold instead).

Actually, section 7.2.11 of ACPI 5.0 is quite clear in that respect.

> Let's see an example
> 
> Assume a device "dev0" depends on 5 power resources:
> 
> pr1, pr2, pr3, pr4, pr5
> 
> _PR3 lists 3 power resources: pr3, pr4, pr5
> 
> Device(dev0)
> {
> 	Name(_PR3, Package (0x03)
> 	{
> 		pr3,
> 		pr4,
> 		pr5
> 	})
> }
> 
> If dev0 is put into ACPI D3_hot and pr1 and pr2 are not referenced by
> other devices, then it requires:
> 
> - pr1 and pr2 are off
> - pr3, pr4 and pr5 are on
> 
> right?

That's correct.

Thanks,
Rafael