On Wed, 2019-12-04 at 12:47 +0000, Mark Brown wrote:
> On Mon, Dec 02, 2019 at 02:02:41PM +0000, Vaittinen, Matti wrote:
> > On Mon, 2019-12-02 at 13:11 +0000, Mark Brown wrote:
> > > No, look at the bindings - we support a bunch of different
> > > suspend states matching the different suspend states that the
> > > kernel as a whole supports. We don't assume that the device will
> > > know this but you can always use the current suspend we're going
> > > for to decide where to update.
> > Hm. So if I understand this correctly, you mean user should set the
> > suspend 'target' - and then call the set_suspend_voltage for this
> > state. To set voltages for all states one should do loop
>
> The general idea is that we set the suspend state during the
> process of suspending rather than in advance - that way when the
> hardware doesn't understand different types of suspsend things
> work fine.
Ok. So voltage for the specific suspend state is set just before going
to suspend - when the target suspend state is already selected. Makes
sense. Thanks.
> > get_current_mode()
> > for_all_modes() {
> > set_mode()
> > set_voltage()
> > }
> > restore_original_mode()
> > am I on a right track? I'll try to see if I can find some examples
> > of
> > this - thanks.
>
> I don't understand the save and restore of mode?
I was thinking that there is some 'cahced suspend target state' in
framework. This get mode is pseudo-code for getting the suspend mode
from framework - "restore mode" is returning back the "original mode"
after voltages for all modes are set. But please ignore this - I think
I captured your message already :)
> If setting the
> suspend configuration affects the runtime state then the hardware
> doesn't support suspend configuration.
No, it does not. I probably explained this badly.
>
> > > The framework doesn't care how a device is controlled, that's up
> > > to the device. Like I said I recommend figuring out what
> > > voltages are useful to have quick access to at runtime, for
> > > example it's likely that it's good to have quick access to the
> > > highest voltage that's been set (and/or the top of the
> > > constraints).
> > Problem is that the run-level controlled regulator can't be
> > individually controlled (unless it is only regulator in the group).
> > I
>
> Regulators that have to be controlled en masse aren't really
> supported by the API, it only understands regulators that are
> individually controllable.
Thanks. This was the piece of information I wanted and assumed. Thus
the patch 12 did introduce new in-kernel APIs - but as I said, we don't
have any in-tree socs using BD71828 RUN0, ..., RUN3 states for now so
I'll drop the patch 12 and leave only the basic support for
individually controllable regulators.
> > But just to confirm, I meant for example assigning bucks 1,2,6 and
> > 7
> > into a group which 'state' is changed via GPIO line. Say 'states'
> > are
> > RUN0, RUN1. For each of these bucks we can define a voltage and
> > enable/disable status which is to be used on RUN0, and another
> > voltage/state tuple for RUN1.
>
> So you could also just create a group consisting of a single
> regulator? That would be fine for the API.
I think I once again explained myself badly. There can be only one
group with 4 RUN states selected by combination of 2 GPIO lines. bucks
1,2,6 and 7 can each either be assigned into this one group or
controlled individually via I2C. But I doubt assigning only one of the
bucks in this group is the typical use-case. What we would need would
really be the 'en-masse' control - either via GPIO or I2C - but I don't
want to suggest any framework expansion as I don't have proper in-tree
use-case for this as of now :) So I'll forget this for now (and thanks
for all the help!) and drop the patch 12 from series. I'll see if I can
send updated series tomorrow - or latest early next week :)
Br,
Matti Vaittinen
