Re: [PATCH 1/4] mfd: arizona: Export function to control subsystem DVFS

[Charles Keepax <ckeepax@opensource.wolfsonmicro.com>]

On Mon, Jun 16, 2014 at 05:42:42PM +0100, Lee Jones wrote:
> On Mon, 09 Jun 2014, Richard Fitzgerald wrote:
> 
> > Moving this control from being a side-effect of the LDO1
> > regulator driver to a specific exported function.
> > 
> > Signed-off-by: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>
> > ---
> >  drivers/mfd/arizona-core.c       |   84 ++++++++++++++++++++++++++++++++++++++
> >  include/linux/mfd/arizona/core.h |   12 +++++
> >  2 files changed, 96 insertions(+), 0 deletions(-)
> > 
> > diff --git a/drivers/mfd/arizona-core.c b/drivers/mfd/arizona-core.c
> > index 58e1fe5..a1b4fe6 100644
> > --- a/drivers/mfd/arizona-core.c
> > +++ b/drivers/mfd/arizona-core.c
> > @@ -94,6 +94,89 @@ int arizona_clk32k_disable(struct arizona *arizona)
> >  }
> >  EXPORT_SYMBOL_GPL(arizona_clk32k_disable);
> >  
> > +int arizona_dvfs_up(struct arizona *arizona, unsigned int mask)
> > +{
> > +	unsigned int new_flags;
> > +	int ret = 0;
> > +
> > +	mutex_lock(&arizona->subsys_max_lock);
> > +
> > +	new_flags = arizona->subsys_max_rq | mask;
> 
> This doesn't look like a mask to me.  It looks like you're setting
> flags rather than masking out bits?

Richard is on holiday so I will fill in for him. Yeah these are
flags I think mask is just a poorly chosen variable name.

> 
> > +	if (arizona->subsys_max_rq != new_flags) {
> > +		switch (arizona->type) {
> > +		case WM5102:
> > +		case WM8997:
> > +			ret = regulator_set_voltage(arizona->dcvdd,
> > +						    1800000, 1800000);
> > +			if (ret != 0) {
> > +				dev_err(arizona->dev,
> > +					"Failed to raise dcvdd (%u)\n", ret);
> > +				goto err;
> > +			}
> > +
> > +			ret = regmap_update_bits(arizona->regmap,
> > +					ARIZONA_DYNAMIC_FREQUENCY_SCALING_1,
> > +					ARIZONA_SUBSYS_MAX_FREQ, 1);
> > +			if (ret != 0) {
> > +				dev_err(arizona->dev,
> > +					"Failed to enable subsys max (%u)\n",
> > +					ret);
> > +				regulator_set_voltage(arizona->dcvdd,
> > +						      1200000, 1800000);
> > +				goto err;
> > +			}
> > +			break;
> > +
> > +		default:
> > +			break;
> > +		}
> 
> I don't really get this.  What's the point in passing the mask
> parameter - I don't see it being used for anything in this routine? No
> matter what is passed in you always just turn on the same regulator.
> 
> What am I missing?

As Mark said each bit represents something that can require the
higher clock rate. Any of the causes being active requires the
higher clock rate.

> 
> > +	arizona->subsys_max_rq = new_flags;
> 
> This tabbing is incorrect.

Will get fixed with the other comments.

Thanks,
Charles