Date:   Thu, 27 Apr 2023 15:20:09 +0530
From:   Naresh Solanki <naresh.solanki@...ements.com>
To:     Guenter Roeck <linux@...ck-us.net>
Cc:     Jean Delvare <jdelvare@...e.com>,
        Patrick Rudolph <patrick.rudolph@...ements.com>,
        linux-kernel@...r.kernel.org, linux-hwmon@...r.kernel.org
Subject: Re: [PATCH] hwmon: (max597x) Add Maxim Max597x

Hi Guenter,

On 26-04-2023 08:05 pm, Guenter Roeck wrote:
> On Wed, Apr 26, 2023 at 11:03:55AM +0200, Naresh Solanki wrote:
>> From: Patrick Rudolph <patrick.rudolph@...ements.com>
>>
>> Add support for the Maxim Max59x power switch with current/voltage
>> monitor.
>>
>> Signed-off-by: Patrick Rudolph <patrick.rudolph@...ements.com>
>> Signed-off-by: Naresh Solanki <Naresh.Solanki@...ements.com>
> 
> Please no wildcards in file names or descriptions for hwmon drivers,
> much less dual-digit wildcards.
> 
> This patch does not add support for MAX597{0..9} (and much less for
> MAX59{00..99}), only for MAX5970 and MAX5978. For example, it does not
> and never will support MAX5977 because that chip does not have an I2C
> interface, or MAX5974 which happens to be a PWM controller.
> 
> Pick one of {5970,5978} for the file name, and mention both in
> descriptions. Yes, I understand, this was accepted with wildcard
> in mfd and regulator subsystems, but that doesn't make it acceptable
> here.
> 
> Either case, what does this driver provide that isn't already
> available through drivers/regulator/max597x-regulator.c ?
> As written, the driver doesn't support any of the limit or alarm
> registers, and there seems to be quite some overlap with the regulator
> driver in terms of functionality. Please explore if that is acceptable
> for the regulator subsystem. If it is not, come back and we can continue
> discussing feasibility as separate hwmon driver.Thanks for your feedback.
I agree that file name wildcard doesn't makes sense here & there are 
other chips that aren't related to this driver. Hence will work on it to 
get it fixed for sure.
Will check with regulator subsystem maintainers & based on feedback will 
proceed.
Thanks :)
> 
>> ---
>>   drivers/hwmon/Kconfig   |   9 ++
>>   drivers/hwmon/Makefile  |   1 +
>>   drivers/hwmon/max597x.c | 212 ++++++++++++++++++++++++++++++++++++++++
> 
> Documentation missing.
> 
>>   3 files changed, 222 insertions(+)
>>   create mode 100644 drivers/hwmon/max597x.c
>>
>> diff --git a/drivers/hwmon/Kconfig b/drivers/hwmon/Kconfig
>> index 5b3b76477b0e..164d980d9de2 100644
>> --- a/drivers/hwmon/Kconfig
>> +++ b/drivers/hwmon/Kconfig
>> @@ -1097,6 +1097,15 @@ config SENSORS_MAX31760
>>   	  This driver can also be built as a module. If so, the module
>>   	  will be called max31760.
>>   
>> +config SENSORS_MAX597X
>> +	tristate "Maxim 597x power switch and monitor"
>> +	depends on I2C
>> +	depends on OF
>> +	select MFD_MAX597X
> 
> That should be "depends on".
Ack
> 
>> +	help
>> +	  This driver exposes Maxim 5970/5978 voltage/current monitoring
>> +	  interface.
>> +
>>   config SENSORS_MAX6620
>>   	tristate "Maxim MAX6620 fan controller"
>>   	depends on I2C
>> diff --git a/drivers/hwmon/Makefile b/drivers/hwmon/Makefile
>> index 88712b5031c8..720eb7d5fe46 100644
>> --- a/drivers/hwmon/Makefile
>> +++ b/drivers/hwmon/Makefile
>> @@ -142,6 +142,7 @@ obj-$(CONFIG_SENSORS_MAX197)	+= max197.o
>>   obj-$(CONFIG_SENSORS_MAX31722)	+= max31722.o
>>   obj-$(CONFIG_SENSORS_MAX31730)	+= max31730.o
>>   obj-$(CONFIG_SENSORS_MAX31760)  += max31760.o
>> +obj-$(CONFIG_SENSORS_MAX597X)	+= max597x.o
>>   obj-$(CONFIG_SENSORS_MAX6620)	+= max6620.o
>>   obj-$(CONFIG_SENSORS_MAX6621)	+= max6621.o
>>   obj-$(CONFIG_SENSORS_MAX6639)	+= max6639.o
>> diff --git a/drivers/hwmon/max597x.c b/drivers/hwmon/max597x.c
>> new file mode 100644
>> index 000000000000..d4d8c2faf55c
>> --- /dev/null
>> +++ b/drivers/hwmon/max597x.c
>> @@ -0,0 +1,212 @@
>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + * Device driver for regulators in MAX5970 and MAX5978 IC
>> + *
>> + * Copyright (c) 2022 9elements GmbH
>> + *
>> + * Author: Patrick Rudolph <patrick.rudolph@...ements.com>
>> + */
>> +
>> +#include <linux/hwmon.h>
>> +#include <linux/i2c.h>
>> +#include <linux/mfd/max597x.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/regmap.h>
>> +
>> +struct max597x_hwmon {
>> +	int num_switches, irng[MAX5970_NUM_SWITCHES], mon_rng[MAX5970_NUM_SWITCHES];
>> +	struct regmap *regmap;
>> +};
>> +
>> +static int max597x_read_reg(struct max597x_hwmon *ddata, int reg, int range, long *val)
>> +{
>> +	u8 reg_data[2];
>> +	int ret;
>> +
>> +	ret = regmap_bulk_read(ddata->regmap, reg, &reg_data[0], 2);
>> +	if (ret < 0)
>> +		return ret;
>> +
>> +	*val = (reg_data[0] << 2) | (reg_data[1] & 3);
>> +	*val = *val * range;
>> +	/*
>> +	 * From datasheet, the range is fractionally less.
>> +	 * To compensate that, divide with 1033 number.
>> +	 */
>> +	*val = *val / 1033;
> 
> Where in the datasheet, and why is this conversion not needed in the
> regulator driver ?
This is taken from the typical value mentioned in table containing MON_ 
LSB Voltage in datasheet. i.e., divide typical value with corresponding 
range.

This is applicable for use in hwmon driver to accurate report readings 
from device.
> 
>> +
>> +	return 0;
>> +}
>> +
>> +static int max597x_read(struct device *dev, enum hwmon_sensor_types type,
>> +			u32 attr, int channel, long *val)
>> +{
>> +	struct max597x_hwmon *ddata = dev_get_drvdata(dev);
>> +	int ret;
>> +
>> +	switch (type) {
>> +	case hwmon_curr:
>> +		switch (attr) {
>> +		case hwmon_curr_input:
>> +			ret = max597x_read_reg(ddata, MAX5970_REG_CURRENT_H(channel),
>> +					       ddata->irng[channel], val);
>> +			if (ret < 0)
>> +				return ret;
>> +
>> +			return 0;
> 
> max597x_read_reg() returns a negative value or 0. Why not just return it ?
Ack.
> 
> The regulator driver takes shunt resistor values into account.
Agree. The shunt resistor value should be accounted for correct current 
measurement. This driver lacks that. This needs to be fixed. Will post 
new patch revision to address that.
> 
> Again, I think it would make much more sense to add hwmon support
> to the regulator driver than having a separate driver since there
> is lots of overlap. For example, the regulator driver already
> sets and monitors limits.
I agree. But the chip also has led functionality. I got review feedback 
to make it MFD. So when rewriting as MFD driver made separate driver 
based on subsystem. I'm not sure if it is ok to use hwmon subsytem in 
regulator driver. Will once check with maintainer on this.
> 
>> +		default:
>> +			return -EOPNOTSUPP;
>> +		}
>> +
>> +	case hwmon_in:
>> +		switch (attr) {
>> +		case hwmon_in_input:
>> +			ret = max597x_read_reg(ddata, MAX5970_REG_VOLTAGE_H(channel),
>> +					       ddata->mon_rng[channel], val);
>> +			if (ret < 0)
>> +				return ret;
>> +			return 0;
>> +		default:
>> +			return -EOPNOTSUPP;
>> +		}
>> +	}
> 
> Please add missing default: cases as reported by 0-day.
Ack
> 
>> +	return -EOPNOTSUPP;
>> +}
>> +
>> +static umode_t max597x_is_visible(const void *data,
>> +				  enum hwmon_sensor_types type,
>> +				  u32 attr, int channel)
>> +{
>> +	struct max597x_hwmon *ddata = (struct max597x_hwmon *)data;
>> +
>> +	if (channel >= ddata->num_switches)
>> +		return 0;
>> +
>> +	switch (type) {
>> +	case hwmon_in:
>> +		switch (attr) {
>> +		case hwmon_in_input:
>> +			return 0444;
>> +		}
>> +		break;
>> +	case hwmon_curr:
>> +		switch (attr) {
>> +		case hwmon_curr_input:
>> +			return 0444;
>> +		}
>> +		break;
>> +	default:
>> +		break;
>> +	}
>> +	return 0;
>> +}
>> +
>> +static const struct hwmon_ops max597x_hwmon_ops = {
>> +	.is_visible = max597x_is_visible,
>> +	.read = max597x_read,
>> +};
>> +
>> +#define HWMON_CURRENT	HWMON_C_INPUT
>> +#define HWMON_VOLTAGE	HWMON_I_INPUT
> 
> Please drop and use HWMON_C_INPUT as well as HWMON_I_INPUT
> directly.
Ack.
> 
>> +
>> +static const struct hwmon_channel_info *max597x_info[] = {
>> +	HWMON_CHANNEL_INFO(in, HWMON_VOLTAGE, HWMON_VOLTAGE),
>> +	HWMON_CHANNEL_INFO(curr, HWMON_CURRENT, HWMON_CURRENT),
>> +	NULL
>> +};
>> +
>> +static const struct hwmon_chip_info max597x_chip_info = {
>> +	.ops = &max597x_hwmon_ops,
>> +	.info = max597x_info,
>> +};
>> +
>> +static int max597x_adc_range(struct regmap *regmap, const int ch,
>> +			     u32 *irng, u32 *mon_rng)
>> +{
> 
> That function also exists on the regulator driver.
Yes. This is to read setting from chip. Added this to avoid dependency & 
race condition with regulator driver.
> 
>> +	unsigned int reg;
>> +	int ret;
>> +
>> +	/* Decode current ADC range */
>> +	ret = regmap_read(regmap, MAX5970_REG_STATUS2, &reg);
>> +	if (ret)
>> +		return ret;
>> +	switch (MAX5970_IRNG(reg, ch)) {
>> +	case 0:
>> +		*irng = 100000;	/* 100 mV */
>> +		break;
>> +	case 1:
>> +		*irng = 50000;	/* 50 mV */
>> +		break;
>> +	case 2:
>> +		*irng = 25000;	/* 25 mV */
>> +		break;
>> +	default:
>> +		return -EINVAL;
>> +	}
>> +
>> +	/* Decode current voltage monitor range */
>> +	ret = regmap_read(regmap, MAX5970_REG_MON_RANGE, &reg);
>> +	if (ret)
>> +		return ret;
>> +
>> +	*mon_rng = MAX5970_MON_MAX_RANGE_UV >> MAX5970_MON(reg, ch);
>> +	*mon_rng /= 1000; /* uV to mV */
>> +
>> +	return 0;
>> +}
>> +
>> +static int max597x_sensor_probe(struct platform_device *pdev)
>> +{
>> +	struct device *dev = &pdev->dev;
>> +	struct i2c_client *i2c = to_i2c_client(pdev->dev.parent);
>> +	struct max597x_hwmon *ddata;
>> +	struct regmap *regmap = dev_get_regmap(pdev->dev.parent, NULL);
>> +	struct device *hwmon_dev;
>> +	int err;
>> +
>> +	if (!regmap)
>> +		return -EPROBE_DEFER;
> 
> Since the parent driver is a mfd driver, I'd assume that this driver
> should be instantiated from there, and I don't see why probe would ever
> have to be deferred. Please explain how that can happen.
I agree with you but this implementation is referenced from other MFD 
leaf driver.
Example: "drivers/hwmon/sy7636a-hwmon.c"
> 
>> +
>> +	ddata = devm_kzalloc(dev, sizeof(struct max597x_hwmon), GFP_KERNEL);
>> +	if (!ddata)
>> +		return -ENOMEM;
>> +
>> +	if (of_device_is_compatible(i2c->dev.of_node, "maxim,max5978"))
>> +		ddata->num_switches = MAX597x_TYPE_MAX5978;
>> +	else if (of_device_is_compatible(i2c->dev.of_node, "maxim,max5970"))
>> +		ddata->num_switches = MAX597x_TYPE_MAX5970;
> 
> Uuh, no. That is just accidentally correct, based on
> 
> enum max597x_chip_type {
>          MAX597x_TYPE_MAX5978 = 1,
>          MAX597x_TYPE_MAX5970,
> };
> 
> The assumption that the numeric chip type matches the number of channels
> it supports is as wrong as it can be. There are defines for the number
> of channels. I am pretty much completely at loss why they are not used.
> Tha same applies for the regulator driver, of course.
Ack. Will use defines.
> 
>> +	else
>> +		return -ENODEV;
>> +
>> +	ddata->regmap = regmap;
>> +
>> +	for (int i = 0; i < ddata->num_switches; i++) {
>> +		err = max597x_adc_range(regmap, i, &ddata->irng[i], &ddata->mon_rng[i]);
>> +		if (err < 0)
>> +			return err;
> 
> Why no error message here but one below ? I am not in favor of error
> messages in the probe function, but if you use them please be consistent.
Ack.
> 
>> +	}
>> +
>> +	hwmon_dev = devm_hwmon_device_register_with_info(dev,
>> +							 "max597x_hwmon", ddata,
>> +							 &max597x_chip_info, NULL);
>> +
>> +	if (IS_ERR(hwmon_dev)) {
>> +		err = PTR_ERR(hwmon_dev);
>> +		dev_err(dev, "Unable to register hwmon device, returned %d\n", err);
> 
> Use dev_err_probe()
Ack
> 
>> +		return err;
>> +	}
>> +
>> +	return 0;
>> +}
>> +
>> +static struct platform_driver max597x_sensor_driver = {
>> +	.probe = max597x_sensor_probe,
>> +	.driver = {
>> +		.name = "max597x-hwmon",
>> +	},
>> +};
>> +module_platform_driver(max597x_sensor_driver);
>> +
>> +MODULE_AUTHOR("Patrick Rudolph <patrick.rudolph@...ements.com>");
>> +MODULE_DESCRIPTION("MAX5970_hot-swap controller driver");
>> +MODULE_LICENSE("GPL v2");
>>
>> base-commit: b4c288cfd2f84c44994330c408e14645d45dee5b
>> -- 
>> 2.39.1
>>
Regards,
Naresh

Powered by blists - more mailing lists