Re: [PATCH] arm: dts: fix rk3066a based boards vdd_log voltage initialization

[Boris Brezillon <boris.brezillon@free-electrons.com>]

+Mark

I realize Mark has been out of the discussion, and what started as a DT
problem actually turned into a PWM regulator discussion.
Maybe we should start a new thread.

On Mon, 19 Sep 2016 14:15:06 -0700
Doug Anderson <dianders@chromium.org> wrote:

> Hi,
> 
> On Mon, Sep 19, 2016 at 1:43 PM, Boris Brezillon
> <boris.brezillon@free-electrons.com> wrote:
> > On Mon, 19 Sep 2016 11:12:12 -0700
> > Doug Anderson <dianders@chromium.org> wrote:
> >  
> >> Hi,
> >>
> >> On Mon, Sep 19, 2016 at 11:06 AM, Boris Brezillon
> >> <boris.brezillon@free-electrons.com> wrote:  
> >> > On Mon, 19 Sep 2016 10:52:51 -0700
> >> > Doug Anderson <dianders@chromium.org> wrote:
> >> >  
> >> >> Hi,
> >> >>
> >> >> On Mon, Sep 19, 2016 at 10:48 AM, Boris Brezillon
> >> >> <boris.brezillon@free-electrons.com> wrote:  
> >> >> > The PWM chip has always claimed the pins and muxed them to the PWM IP.
> >> >> > So, this means it's broken from the beginning, and my patch is only
> >> >> > uncovering the problem (unless the pins stay configured as input until
> >> >> > the PWM is enabled, which I'm not sure is the case).  
> >> >>
> >> >> Such a solution is achievable with the pinctrl APIs pretty easily.
> >> >> You might not be able to use the automatic "init" state but you can do
> >> >> something similar and switch to an "active" state once the PWM is
> >> >> actually turned on the first time.  
> >> >
> >> > But is it really the case here (I don't see any code requesting a
> >> > specific pinmux depending on the PWM state)?  
> >>
> >> It is not happening right now as far as I know.  ...but that's a bug.
> >>  
> >> > Anyway, we really need to handle this case, we should define the
> >> > typical voltage when the PWM is disabled. Same as what you suggested
> >> > with voltage-when-input, but with a different naming (since the concept
> >> > of pinmux is PWM hardware/driver specific).
> >> >
> >> >         voltage-when-pwm-disabled = <...>;  
> >>
> >> Voltage when disabled and voltage when input are two different states.
> >> A disabled PWM will typically either drive high or low (depending on
> >> where it was when you turned it off).  Not all "disabled" states will
> >> mean that the pin is configured as an input.  
> >
> > Okay, after reading again your first answer, I think I understand why
> > you want to differentiate the when-disabled and when-input cases. You
> > want to use the "init" and "default" pinctrl states. The "init" state
> > (applied at probe time) would keep the PWM pin in gpio+input mode and
> > the "default" state (applied when the PWM device is enabled for the
> > first time) would mux the pin to the PWM device.
> > Your solution requires that the pwm-regulator device knows in which
> > state the pin attached to the PWM is, which IMO is breaking the
> > layering we have right now: a PWM-regulator is assigned a PWM device
> > which is assigned a pin and a pinmux config.
> > Another solution would be to expose an additional information in the
> > pwm_state: whether the PWM is in the INIT state (probed, but not yet
> > configured by its user) or DEFAULT state (probed and already
> > configured by its user). But again, by doing that we also expose
> > internal PWM details to its user, which I'm not sure will help keep
> > the PWM API simple.  
> 
> One note is that probably using the "init" state would not be a good
> idea because it would make assumptions about what state the firmware
> left things.  Possibly a firmware update could cause a change from a
> PWM being left as "input" to the PWM actually being initted.
> ...really we should be able to detect if the PWM was left on at
> bootup.

Yep.

> 
> 
> > Actually, I had something slightly different in mind. I thought about
> > having two new pinctrl states ("enabled" and "disabled"). The "enabled"
> > state (pin muxed to the PWM device) would be applied each time the PWM
> > is enabled, and "disabled" state (gpio+input mode) would be applied each
> > time the PWM is disabled.  
> 
> Adding "enabled" and "disabled" state is sane IMHO.  At the moment the
> PWM regulator never actually puts things in "disabled" state, but I
> suppose it could in the future.
> 
> > This way we can guarantee that even when the PWM is disabled, the
> > PWM-driven regulator is configured to output a non-destructive voltage.
> > Hence my suggestion to name the property 'voltage-when-pwm-disabled'.  
> 
> That's fine with me, as long as the PWM starts up as "enabled" if the
> pinctrl happened to be left initted by the BIOS.

Yes, that's already the case in the pwm-rockchip driver.

Okay, I can try to implement what is described above, but, in the
meantime, I think we should find a solution for Andy's initial problem.

As I said, the problem you're describing (pins muxed to the PWM device
when it should actually stay in gpio+input mode) is not new, and the old
pwm-regulator and pwm-rockchip implementation (before my atomic PWM
changes) were behaving the same way.

What is new though, is the pwm_regulator_init_state() function [1], and
it seems it's now preventing the probe of a pwm-regulator device if the
initial PWM state is not described in the voltage-table.

The question is, what should we do?

1/ Force users to put an entry matching this state (which means
   breaking DT compat)
2/ Put a valid value in drvdata->state even if it's not reflecting the
   real state
3/ Patch regulator core to support an "unknown-selector" return code.

Mark, any opinion?

Thanks,

Boris

[1]https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/tree/drivers/regulator/pwm-regulator.c?id=refs/tags/next-20160922#n60