Messages in this thread | | | Date | Thu, 1 Apr 2010 19:35:11 +0100 | From | Mark Brown <> | Subject | Re: [PATCH 04/19] regulator: add support for Dallas DS1803 dual digital potentiometer |
| |
On Thu, Apr 01, 2010 at 01:23:32PM +0200, Rodolfo Giometti wrote:
This looks good, a few nitpicky things below.
> +config REGULATOR_DS1803 > + tristate "Dallas Maxim DS1803 addressable dual digital potentiometer" > + depends on I2C > + help > + Say Y here to support the dual digital potentiometer on > + Dallas Maxim DS1803 > +
Oh, wow. A pot as a voltage regulator... :)
> +struct ds1803_data { > + struct i2c_client *client; > + struct regulator_dev *rdev; > + > + unsigned int min_uV; /* voltage for selector value 0 */ > + unsigned int max_uV; /* voltage for selector value 255 */ > + unsigned int init_uV; /* initial voltage */
Given that you don't support get_voltage() or otherwise reference init_uV it seems as well to just drop that for now.
> +static u8 ds1803_write_pot_lut[] = { 0xa9, 0xaa, /* both not supported */ };
Both what? Since you don't seem to reference this I guess it could just be dropped...
> +static int ds1803_list_voltage(struct regulator_dev *rdev, unsigned index) > +{ > + struct ds1803_data *ds1803 = rdev_get_drvdata(rdev); > + > + return (ds1803->max_uV - ds1803->min_uV) * index / 255 > + + ds1803->min_uV; > +}
I think I'd like more brackets in this calculation for clarity that the operator precedence is OK.
> + /* Set chip's name according to user supplied type */ > + ds1803_reg[i].name = ds1803_names[pdata->type];
Perhaps just let the user write in something they feel like, or use a constant string for the chip? The type doesn't seem to be used otherwise so I can see the data ending up wrong and misleading folks.
> +enum ds1803_type { > + DS1803_100K, > + DS1803_50K, > + DS1803_10K, > +};
If you are going to keep these assign a value to the first item so you don't end up with 0 as a valid type, or make the 0 type be "unspecified" or something. That way platform data that's left initialised to zero can be distinguished from something that someone deliberately set.
| |