Re: [PATCH 4/4] regulator: max14577: Add regulator driver for Maxim 14577

[Mark Brown <broonie-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>]

[-- Attachment #1: Type: text/plain, Size: 2739 bytes --]

On Wed, Nov 13, 2013 at 08:40:57AM +0100, Krzysztof Kozlowski wrote:

> +static inline int max14577_reg_set_safeout(struct regmap *rmap, int on)
> +{
> +	u8 reg_data = (on ? 0x1 << CTRL2_SFOUTORD_SHIFT : 0x0);

The ternery operator is rarely helpful for legibility.

> +	case MAX14577_SAFEOUT:
> +		max14577_read_reg(rmap, MAX14577_REG_CONTROL2, &reg_data);
> +		return (reg_data & CTRL2_SFOUTORD_MASK) != 0;

regulator_is_enabled_regmap() and the other helpers should be used for
this regulator.

> +static int max14577_reg_disable(struct regulator_dev *rdev)
> +{
> +	int rid = rdev_get_id(rdev);
> +
> +	switch (rid) {
> +	case MAX14577_CHARGER:
> +		return max14577_reg_set_charging(rdev->regmap, 0);
> +	case MAX14577_SAFEOUT:
> +		return max14577_reg_set_safeout(rdev->regmap, 0);
> +	default:
> +		return -EINVAL;
> +	}
> +}

This sort of indirection isn't adding anything, just make the ops
directly callable (and in the case of SAFEOUT use the helpers).

> +static int max14577_reg_get_current_limit(struct regulator_dev *rdev)
> +{
> +	if (rdev_get_id(rdev) != MAX14577_CHARGER)
> +		return -EINVAL;
> +
> +	return max14577_reg_calc_charge_current(rdev->regmap);

Just inline the function, it's only called once.

> +	if (rdev_get_id(rdev) != MAX14577_CHARGER)
> +		return -EINVAL;
> +
> +	dev_info(max14577->dev, "Requested current <%d, %d>\n", min_uA, max_uA);

Don't spam the logs like this, there's problems like this throughout the
rest of the code.

> +static int max14577_regulator_dt_parse_pdata(struct platform_device *pdev,
> +					struct max14577_platform_data *pdata)

No DT binding document.

> +static int max14577_regulator_probe(struct platform_device *pdev)
> +{
> +	struct max14577_regulator *max14577_reg;
> +	struct max14577 *max14577 = dev_get_drvdata(pdev->dev.parent);
> +	struct max14577_platform_data *pdata = dev_get_platdata(max14577->dev);
> +	int i, ret, size;
> +	struct regulator_config config = {};
> +
> +	if (!pdata) {
> +		dev_err(&pdev->dev, "No platform data found.\n");
> +		return -ENODEV;
> +	}
> +
> +	if (max14577->dev->of_node) {
> +		ret = max14577_regulator_dt_parse_pdata(pdev, pdata);
> +		if (ret)
> +			return ret;
> +	}

The driver insists on platform data but then requires DT...

> +	size = sizeof(struct regulator_dev *) * pdata->num_regulators;
> +	max14577_reg->regulators = devm_kzalloc(&pdev->dev, size, GFP_KERNEL);

You shouldn't be getting the number of regulators from platform data,
the number of regulators physically present in the silicon isn't going
to vary.

> +		max14577_reg->regulators[i] =
> +			regulator_register(&supported_regulators[id], &config);

devm_regulator_register().

> +subsys_initcall(max14577_regulator_init);

Why is this subsys_initcall()?

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]