[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20120523095808.GD4085@opensource.wolfsonmicro.com>
Date: Wed, 23 May 2012 10:58:08 +0100
From: Mark Brown <broonie@...nsource.wolfsonmicro.com>
To: "Kim, Milo" <Milo.Kim@...com>
Cc: "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"Girdwood, Liam" <lrg@...com>
Subject: Re: [PATCH] regulator: add new regulator driver for lp872x
On Wed, May 23, 2012 at 07:50:59AM +0000, Kim, Milo wrote:
> +static int lp872x_read_byte(struct lp872x *lp, u8 addr, u8 *data)
> +{
> + int ret;
> + unsigned int val;
> +
> + mutex_lock(&lp->xfer_lock);
> + ret = regmap_read(lp->regmap, addr, &val);
> + if (ret < 0) {
> + mutex_unlock(&lp->xfer_lock);
> + dev_err(lp->dev, "failed to read 0x%.2x\n", addr);
> + return ret;
> + }
> + mutex_unlock(&lp->xfer_lock);
> +
> + *data = (u8)val;
> + return 0;
> +}
This should just be a trivial wrapper around regmap_read, the regmap API
handles concurrency for you. Similarly for all your other I/O
functions.
> +static int lp872x_ldo_list_voltage(struct regulator_dev *rdev,
> + unsigned selector)
regulator_list_voltage_table() will appear in -next just after the merge
window.
> +static int lp872x_ldo_set_voltage_sel(struct regulator_dev *rdev,
> + unsigned selector)
regulator_set_voltage_sel_regmap().
> +static int lp872x_ldo_get_voltage_sel(struct regulator_dev *rdev)
regulator_get_voltage_sel_regmap().
> +static int lp872x_regulator_enable(struct regulator_dev *rdev)
Your ENABLE define made the code a bit obscure but this is
regulator_enable_regmap()
> +static int lp872x_regualtor_disable(struct regulator_dev *rdev)
regulator_disable_regmap()
> +static int lp872x_regulator_is_enabled(struct regulator_dev *rdev)
regulator_is_enabled_regmap()
> +static int lp872x_buck_list_voltage(struct regulator_dev *rdev,
> + unsigned selector)
regulator_list_voltage_table()
> + case LP8720_ID_BUCK:
> + if (val & LP8720_EXT_DVS_M) {
> + pinstate = pdata->get_dvs_pin_state ?
> + pdata->get_dvs_pin_state() : DVS_LOW;
This is a really odd implementation. Why isn't the driver managing the
pin and what happens to the consumers that expected to set a voltage
when the pin gets changed?
> +static int lp8725_buck_enable(struct regulator_dev *rdev)
regulator_enable_regmap()
> +static int lp8725_buck_disable(struct regulator_dev *rdev)
regulator_disable_regmap()
> +static int lp8725_buck_is_enabled(struct regulator_dev *rdev)
regulator_is_enabled_regmap()
> + for (i = 0 ; i < num ; i++) {
> + regulator_data = lp->pdata->regulator_data + i;
> + desc = &lp872x_regulator_desc[regulator_data->id];
> + init_data = regulator_data->init_data;
> +
> + rdev = regulator_register(desc, lp->dev, init_data, lp, NULL);
> + if (IS_ERR(rdev)) {
> + dev_err(lp->dev, "regulator register err");
> + ret = PTR_ERR(rdev);
> + goto err;
> + }
You should unconditionally register all the regulators the chip has and
match the init_data to them rather than only registering regulators you
have init_data. This allows users to inspect unconfigured regulators
and allows the core to support things like powering off unused
regulators.
> +static const struct regmap_config lp872x_regmap_config = {
> + .reg_bits = 8,
> + .val_bits = 8,
> +};
Setting max_register would get you debugfs inspection of the regmap.
> + lp->regmap = regmap_init_i2c(cl, &lp872x_regmap_config);
devm_regmap_init_i2c().
> +/**
> + * lp872x_platform_data
> + * @general_config (mandatory) : the value of LP872X_GENERAL_CFG register
Is the chip default never valid?
> + * @regulator_data (mandatory) : platform regulator id and init data
> + * @num_regulators (mandatory) : total number of platform regulators
These should be optional.
Download attachment "signature.asc" of type "application/pgp-signature" (837 bytes)
Powered by blists - more mailing lists