On Fri, Aug 07, 2020 at 09:28:31AM +0200, Michael Walle wrote:
> Hi Uwe, Hi Lee,
> 
> Am 2020-08-06 10:40, schrieb Uwe Kleine-König:
> > On Mon, Aug 03, 2020 at 11:35:52AM +0200, Michael Walle wrote:
> > > diff --git a/drivers/pwm/Kconfig b/drivers/pwm/Kconfig
> > > index 7dbcf6973d33..a0d50d70c3b9 100644
> > > --- a/drivers/pwm/Kconfig
> > > +++ b/drivers/pwm/Kconfig
> > > @@ -428,6 +428,16 @@ config PWM_SIFIVE
> > >  	  To compile this driver as a module, choose M here: the module
> > >  	  will be called pwm-sifive.
> > > 
> > > +config PWM_SL28CPLD
> > > +	tristate "Kontron sl28cpld PWM support"
> > > +	select MFD_SIMPLE_MFD_I2C
> > 
> > Is it sensible to present this option to everyone? Maybe
> > 
> > 	depends on SOME_SYMBOL_ONLY_TRUE_ON_SL28CPLD || COMPILE_TEST
> 
> Because there is now no real MFD driver anymore, there is also
> no symbol for that. The closest would be ARCH_ARM64 but I don't
> think that is a good idea.
> 
> Lee, what do you think about adding a symbol to the MFD, which
> selects MFD_SIMPLE_MFD_I2C but doesn't enable any C modules?
> 
> I.e.
> config MFD_SL28CPLD
>     tristate "Kontron sl28cpld"
>     select MFD_SIMPLE_MFD_I2C
>     help
>       Say yes here to add support for the Kontron sl28cpld board
>       management controller.
> 
> Then all the other device driver could depend on the MFD_SL28CPLD
> symbol.
> 
> [..]
> 
> > > +static void sl28cpld_pwm_get_state(struct pwm_chip *chip,
> > > +				   struct pwm_device *pwm,
> > > +				   struct pwm_state *state)
> > > +{
> > > +	struct sl28cpld_pwm *priv = dev_get_drvdata(chip->dev);
> > > +	unsigned int reg;
> > > +	int prescaler;
> > > +
> > > +	sl28cpld_pwm_read(priv, SL28CPLD_PWM_CTRL, &reg);
> > > +
> > > +	state->enabled = reg & SL28CPLD_PWM_CTRL_ENABLE;
> > > +
> > > +	prescaler = FIELD_GET(SL28CPLD_PWM_CTRL_PRESCALER_MASK, reg);
> > > +	state->period = SL28CPLD_PWM_PERIOD(prescaler);
> > > +
> > > +	sl28cpld_pwm_read(priv, SL28CPLD_PWM_CYCLE, &reg);
> > > +	state->duty_cycle = SL28CPLD_PWM_TO_DUTY_CYCLE(reg);
> > 
> > Should reg be masked to SL28CPLD_PWM_CYCLE_MAX, or is it guaranteed that
> > the upper bits are zero?
> 
> Mh, the hardware guarantees that bit7 is zero. So masking with
> SL28CPLD_PWM_CYCLE_MAX won't buy us much. But what I could think
> could go wrong is this: someone set the prescaler to != 0 and the
> duty cycle to a value greater than the max value for this particular
> prescaler mode. For the above calculations this would result in a
> duty_cycle greater than the period, if I'm not mistaken.
> 
> The behavior of the hardware is undefined in that case (at the moment
> it will be always on, I guess). So this isn't a valid setting.
> Nevertheless it might happen. So what about the following:
> 
> state->duty_cycle = min(state->duty_cycle, state->period);

If you care about this: This can also happen (at least shortly) in
sl28cpld_pwm_apply() as you write SL28CPLD_PWM_CTRL before
SL28CPLD_PWM_CYCLE there.

I wonder if we want to sanitize the values returned from driver's
.get_state in the core; or scream loud (maybe only if PWM_DEBUG is on).

Something like:

	if (state->enabled && state->duty_cycle > state->period) {
		if (IS_ENABLED(CONFIG_PWM_DEBUG))
			dev_warn(chip->dev, ".get_state() returned invalid setting.\n");

		state->duty_cycle = state->period;
	}

Do we want to catch state->period = 0, too? Do we interpret this as
disabled?

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |