Re: [PATCH] hwmon: mp2891: add MP2891 driver

[Guenter Roeck]

On Fri, May 26, 2023 at 05:42:07PM +0800, Noah Wang wrote:
> This driver is designed for MPS VR controller mp2891. The sum
> of input voltage, output voltage, output current, input power,
> output power and temperature of per rail can be obtained from
> hwmon interfaces that the driver provided.
> 
> Firstly, the driver get the vid_step of output voltage and
> current scale(iout_scale) from the mps specific register
> MFR_VOUT_LOOP_CTRL(0xBD) and MFR_SVI3_IOUT_PRT(0x65). These
> two parameters are used to calculate the output voltage and
> output current. Then, if the hwmon interfaces that the driver
> provided are read by the userspace, the driver will obtain
> corresponding data from mp2891 and the data will be submitted
> to the pmbus-core. Finally, userspace can obtain the result.
> 

That really should be a comment in the driver and not be part of the patch
description.

> Signed-off-by: Noah Wang <noahwang.wang@xxxxxxxxxxx>
> ---
>  drivers/hwmon/pmbus/Kconfig  |  10 +
>  drivers/hwmon/pmbus/Makefile |   1 +
>  drivers/hwmon/pmbus/mp2891.c | 371 +++++++++++++++++++++++++++++++++++

Documentation/hwmon/mp2891.rst is missing.

>  3 files changed, 382 insertions(+)
>  create mode 100644 drivers/hwmon/pmbus/mp2891.c
> 
> diff --git a/drivers/hwmon/pmbus/Kconfig b/drivers/hwmon/pmbus/Kconfig
> index 270b6336b76d..8e930c15aa89 100644
> --- a/drivers/hwmon/pmbus/Kconfig
> +++ b/drivers/hwmon/pmbus/Kconfig
> @@ -326,6 +326,16 @@ config SENSORS_MP5023
>  	  This driver can also be built as a module. If so, the module will
>  	  be called mp5023.
>  
> +config SENSORS_MP2891
> +	tristate "MPS MP2891"
> +	help
> +	  If you say yes here you get hardware monitoring support for MPS
> +	  MP2891 Dual Loop Digital Multi-Phase Controller.
> +
> +	  This driver can also be built as a module. If so, the module will
> +	  be called mp2891.
> +
> +
>  config SENSORS_MPQ7932_REGULATOR
>  	bool "Regulator support for MPQ7932"
>  	depends on SENSORS_MPQ7932 && REGULATOR
> diff --git a/drivers/hwmon/pmbus/Makefile b/drivers/hwmon/pmbus/Makefile
> index 84ee960a6c2d..1a3884fd4b1a 100644
> --- a/drivers/hwmon/pmbus/Makefile
> +++ b/drivers/hwmon/pmbus/Makefile
> @@ -35,6 +35,7 @@ obj-$(CONFIG_SENSORS_MAX8688)	+= max8688.o
>  obj-$(CONFIG_SENSORS_MP2888)	+= mp2888.o
>  obj-$(CONFIG_SENSORS_MP2975)	+= mp2975.o
>  obj-$(CONFIG_SENSORS_MP5023)	+= mp5023.o
> +obj-$(CONFIG_SENSORS_MP2891)	+= mp2891.o
>  obj-$(CONFIG_SENSORS_MPQ7932)	+= mpq7932.o
>  obj-$(CONFIG_SENSORS_PLI1209BC)	+= pli1209bc.o
>  obj-$(CONFIG_SENSORS_PM6764TR)	+= pm6764tr.o
> diff --git a/drivers/hwmon/pmbus/mp2891.c b/drivers/hwmon/pmbus/mp2891.c
> new file mode 100644
> index 000000000000..07adf5f9871f
> --- /dev/null
> +++ b/drivers/hwmon/pmbus/mp2891.c
> @@ -0,0 +1,371 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * Hardware monitoring driver for MPS Multi-phase Digital VR Controllers(MP2891)
> + *
> + * Copyright (C) 2023 MPS
> + */
> +
> +#include <linux/err.h>
> +#include <linux/i2c.h>
> +#include <linux/init.h>
> +#include <linux/kernel.h>
> +#include <linux/module.h>
> +#include "pmbus.h"
> +
> +/* Vendor specific registers. */
> +#define MFR_VOUT_LOOP_CTRL      0xBD
> +#define MFR_SVI3_IOUT_PRT       0x65
> +#define READ_PIN_EST            0x94

Standard PMBus register/command 0x94 is "READ_DUTY_CYCLE".  0x65 is
"TOFF_FAIL". I find it difficult to believe that the chip would
redefine those. If it does, that will need extra explanation here.

Also, the use of READ_PIN_EST (0x94) instead of READ_PIN will need
to be explained.

> +
> +#define VID_STEP_POS            14
> +#define VID_STEP_MSK            (0x3 << VID_STEP_POS)
> +#define DAC_2P5MV_POS           13
> +#define DAC_2P5MV_MSK           (0x1 << DAC_2P5MV_POS)
> +#define IOUT_SCALE_POS          0
> +#define IOUT_SCALE_MSK          (0x7 << IOUT_SCALE_POS)
> +
> +#define MP2891_PAGE_NUM			2
> +
> +#define MP2891_RAIL1_FUNC	(PMBUS_HAVE_VIN | PMBUS_HAVE_VOUT | \
> +							PMBUS_HAVE_IOUT | PMBUS_HAVE_TEMP | \
> +							PMBUS_HAVE_POUT | PMBUS_HAVE_PIN | \
> +							PMBUS_PHASE_VIRTUAL)
> +
> +#define MP2891_RAIL2_FUNC	(PMBUS_HAVE_VOUT | PMBUS_HAVE_IOUT | \
> +							PMBUS_HAVE_TEMP | PMBUS_HAVE_POUT | \
> +							PMBUS_PHASE_VIRTUAL)
> +
> +struct mp2891_data {
> +	struct pmbus_driver_info info;
> +	int vid_step[MP2891_PAGE_NUM];
> +	int iout_scale[MP2891_PAGE_NUM];
> +};
> +
> +#define to_mp2891_data(x) container_of(x, struct mp2891_data, info)
> +
> +static int
> +mp2891_read_pout(struct i2c_client *client, int page, int phase, int reg)
> +{
> +	int ret;
> +
> +	ret = pmbus_read_word_data(client, page, phase, reg);
> +
> +	return ret;
> +}
> +
> +static int
> +mp2891_read_pin(struct i2c_client *client, int page, int phase, int reg)
> +{
> +	int ret;
> +
> +	ret = pmbus_read_word_data(client, page, phase, reg);
> +
> +	return ret;
> +}
> +

The above functions really do not make any sense whatsoever, meaning I do
not trust that any of the remaining functions are needed either. Please
only provide necessary private functions. For each function which is
actually needed please explain why the core function can not be used.

Note that access to the datasheet would help me to determine which of
those functions are really needed. Requiring me to create an account
just to read a non-descriptive one-page "datasheet" doesn't help at all
(and, FWIW, is really pointless).

> +static int
> +mp2891_read_vin(struct i2c_client *client, int page, int phase, int reg)
> +{
> +	int ret;
> +
> +	ret = pmbus_read_word_data(client, page, phase, reg);
> +	ret = ((ret & 0x7FF) * 1000) >> 5;

All those functions ignore errors. Please fix. Also, as with other
functions, the use of this function instead of the common function
will need to be explained.

> +
> +	return ret;
> +}
> +
> +static int
> +mp2891_read_vout(struct i2c_client *client, int page, int phase, int reg)
> +{
> +	int ret;
> +
> +	const struct pmbus_driver_info *info = pmbus_get_driver_info(client);
> +	struct mp2891_data *data = to_mp2891_data(info);
> +
> +	ret = pmbus_read_word_data(client, page, phase, reg);
> +	ret = (ret & 0x7FF) * data->vid_step[page] / 100;
> +
> +	return ret;
> +}
> +
> +static int
> +mp2891_read_iout(struct i2c_client *client, int page, int phase, int reg)
> +{
> +	int ret;
> +
> +	const struct pmbus_driver_info *info = pmbus_get_driver_info(client);
> +	struct mp2891_data *data = to_mp2891_data(info);
> +
> +	ret = pmbus_read_word_data(client, page, phase, reg);
> +
> +	if (((ret & 0x8000) >> 15) == 0)
> +		ret = (ret & 0x7FF) * (((ret & 0x7800) >> 11) + 1);
> +	else
> +		ret = (ret & 0x7FF) >> (32 - ((ret & 0xF800) >> 11));
> +
> +	ret = ret * data->iout_scale[page];
> +
> +	return ret;
> +}
> +
> +static int
> +mp2891_read_temperature(struct i2c_client *client, int page, int phase, int reg)
> +{
> +	int ret;
> +
> +	ret = pmbus_read_word_data(client, page, phase, reg);
> +
> +	if (((ret & 0x400) >> 10) == 0)
> +		ret = ret & 0x7FF;
> +	else
> +		ret = ~(ret & 0x7FF) + 1;
> +

This most definitely needs explanation.

> +	return ret;
> +}
> +
> +static int mp2891_read_byte_data(struct i2c_client *client, int page, int reg)
> +{
> +	int ret;
> +
> +	switch (reg) {
> +	case PMBUS_VOUT_MODE:
> +		/*
> +		 * Enforce VOUT direct format, since device allows to set the
> +		 * different formats for the different rails. Conversion from
> +		 * VID to direct provided by driver internally, in case it is
> +		 * necessary.

"in case it is necessary" - please explain what that means. There is no
conditional in the vout handling code.

> +		 */
> +		ret = PB_VOUT_MODE_DIRECT;
> +		break;
> +	default:
> +		ret = -EINVAL;
> +		break;
> +	}
> +
> +	return ret;
> +}
> +
> +static int mp2891_read_word_data(struct i2c_client *client, int page, int phase,
> +			      int reg)
> +{
> +	int ret;
> +
> +	switch (reg) {
> +	case PMBUS_READ_VIN:
> +		ret = mp2891_read_vin(client, page, phase, reg);
> +		break;
> +	case PMBUS_READ_VOUT:
> +		ret = mp2891_read_vout(client, page, phase, reg);
> +		break;
> +	case PMBUS_READ_IOUT:
> +		ret = mp2891_read_iout(client, page, phase, reg);
> +		break;
> +	case PMBUS_READ_TEMPERATURE_1:
> +		ret = mp2891_read_temperature(client, page, phase, reg);
> +		break;
> +	case PMBUS_READ_PIN:
> +		ret = mp2891_read_pin(client, page, phase, READ_PIN_EST);
> +		break;
> +	case PMBUS_READ_POUT:
> +		ret = mp2891_read_pout(client, page, phase, reg);
> +		break;
> +	default:
> +		ret = -EINVAL;
> +		break;
> +	}
> +
> +	return ret;
> +}
> +
> +static int
> +mp2891_identify_vid(struct i2c_client *client, struct mp2891_data *data,
> +						u32 reg, int page)
> +{
> +	int ret;
> +
> +	ret = i2c_smbus_write_byte_data(client, PMBUS_PAGE, page);
> +	if (ret < 0)
> +		return ret;
> +
> +	ret = i2c_smbus_read_word_data(client, reg);
> +	if (ret < 0)
> +		return ret;
> +
> +	/*
> +	 * Obtain vid_step from the register MFR_VOUT_LOOP_CTRL, bits 15-14,bit 13.
> +	 * If MFR_VOUT_LOOP_CTRL[13] = 1, the vid_step is below:
> +	 * 2.5mV/LSB
> +	 * If MFR_VOUT_LOOP_CTRL[13] = 0, the vid_step is decided by
> +	 * MFR_VOUT_LOOP_CTRL[15:14]:
> +	 * 00b - 6.25mV/LSB, 01b - 5mV/LSB, 10b - 2mV/LSB, 11b - 1mV
> +	 */
> +	if (((ret & DAC_2P5MV_MSK) >> VID_STEP_POS))

Too many ()

> +		data->vid_step[page] = 250;

Coding style violation. Please run checkpatch --strict on your patch(es).

> +	else {
> +		ret = ((ret & VID_STEP_MSK) >> VID_STEP_POS);

Too many ()

> +		if (ret == 0)
> +			data->vid_step[page] = 625;
> +		else if (ret == 1)
> +			data->vid_step[page] = 500;
> +		else if (ret == 2)
> +			data->vid_step[page] = 200;
> +		else
> +			data->vid_step[page] = 100;
> +	}

This needs explanation why the per-page vrm version (and thus the use of
the standard conversion function) is insufficient.

> +
> +	return 0;
> +}
> +
> +static int
> +mp2891_identify_rails_vid(struct i2c_client *client, struct mp2891_data *data)
> +{
> +	int ret;
> +
> +	/* Identify vid_step from register  MFR_VOUT_LOOP_CTRL. */
> +	/* Identify vid_step for rail 1. */
> +	ret = mp2891_identify_vid(client, data, MFR_VOUT_LOOP_CTRL, 0);
> +	if (ret < 0)
> +		return ret;
> +
> +	/* Identify vid_step for rail 2. */
> +	ret = mp2891_identify_vid(client, data, MFR_VOUT_LOOP_CTRL, 1);
> +
> +	return ret;
> +}
> +
> +static int
> +mp2891_iout_scale_get(struct i2c_client *client, struct mp2891_data *data,
> +						u32 reg, int page)
> +{
> +	int ret;
> +
> +	ret = i2c_smbus_write_byte_data(client, PMBUS_PAGE, page);
> +	if (ret < 0)
> +		return ret;
> +
> +	ret = i2c_smbus_read_word_data(client, reg);
> +
> +	/*
> +	 * Obtain iout_scale from the register MFR_SVI3_IOUT_PRT, bits 2-0.
> +	 * The value is selected as below:
> +	 * 000b - 1A/LSB, 001b - (1/32)A/LSB, 010b - (1/16)A/LSB,
> +	 * 011b - (1/8)A/LSB, 100b - (1/4)A/LSB, 101b - (1/2)A/LSB
> +	 * 110b - 1A/LSB, 111b - 2A/LSB
> +	 */
> +	ret = (ret & IOUT_SCALE_MSK) >> IOUT_SCALE_POS;
> +	if ((ret == 0) | (ret == 6))
> +		data->iout_scale[page] = 32;
> +	else if (ret == 1)
> +		data->iout_scale[page] = 1;
> +	else if (ret == 2)
> +		data->iout_scale[page] = 2;
> +	else if (ret == 3)
> +		data->iout_scale[page] = 4;
> +	else if (ret == 4)
> +		data->iout_scale[page] = 8;
> +	else if (ret == 5)
> +		data->iout_scale[page] = 16;
> +	else
> +		data->iout_scale[page] = 64;

This needs extra explanation: Usually the scale commands (IOUC_CAL_GAIN
and IOUT_CAL_OFFSET) enable the chip to provide correct values in the
READ_IOUT register. According to the above, the chip would do the opposite
and provide _unscaled_ values which need to be corrected by software. From
a functionality perspective this doesn't really make sense. Therefore,
especially with the datasheet not being available, it needs to be
explained in detail here that the chip does the opposite of what it is
supposed to be doing, which needs to be corrected in software.

> +
> +	return 0;
> +}
> +
> +static int
> +mp2891_rails_iout_scale_get(struct i2c_client *client, struct mp2891_data *data)
> +{
> +	int ret;
> +
> +	/* Get iout_scale from register MFR_SVI3_IOUT_PRT. */
> +	/* Get iout_scale for rail 1. */
> +	ret = mp2891_iout_scale_get(client, data, MFR_SVI3_IOUT_PRT, 0);
> +	if (ret < 0)
> +		return ret;
> +
> +	/* Get iout_scale for rail 2. */
> +	ret = mp2891_iout_scale_get(client, data, MFR_SVI3_IOUT_PRT, 1);
> +
> +	return ret;
> +}
> +
> +static struct pmbus_driver_info mp2891_info = {
> +	.pages = MP2891_PAGE_NUM,
> +	.format[PSC_VOLTAGE_IN] = direct,
> +	.format[PSC_VOLTAGE_OUT] = direct,
> +	.format[PSC_CURRENT_OUT] = direct,
> +	.format[PSC_TEMPERATURE] = direct,
> +	.format[PSC_POWER] = linear,
> +
> +	.m[PSC_VOLTAGE_IN] = 1,
> +	.m[PSC_VOLTAGE_OUT] = 1,
> +	.m[PSC_CURRENT_OUT] = 32,
> +	.m[PSC_TEMPERATURE] = 1,
> +	.R[PSC_VOLTAGE_IN] = 3,
> +	.R[PSC_VOLTAGE_OUT] = 3,
> +	.R[PSC_CURRENT_OUT] = 0,
> +	.R[PSC_TEMPERATURE] = 0,
> +	.b[PSC_VOLTAGE_IN] = 0,
> +	.b[PSC_VOLTAGE_OUT] = 0,
> +	.b[PSC_CURRENT_OUT] = 0,
> +	.b[PSC_TEMPERATURE] = 0,
> +
> +	.func[0] = MP2891_RAIL1_FUNC,
> +	.func[1] = MP2891_RAIL2_FUNC,
> +	.read_word_data = mp2891_read_word_data,
> +	.read_byte_data = mp2891_read_byte_data,
> +};
> +
> +static int mp2891_probe(struct i2c_client *client, const struct i2c_device_id *id)

The second parameter no longer exists. Please build your patch(es) against
the latest kernel.

> +{
> +	struct pmbus_driver_info *info;
> +	struct mp2891_data *data;
> +	int ret;
> +
> +	data = devm_kzalloc(&client->dev, sizeof(struct mp2891_data), GFP_KERNEL);
> +
Unnecessary empty line

> +	if (!data)
> +		return -ENOMEM;
> +
> +	memcpy(&data->info, &mp2891_info, sizeof(*info));
> +	info = &data->info;
> +
> +	/* Identify VID setting per rail. */
> +	ret = mp2891_identify_rails_vid(client, data);
> +	if (ret < 0)
> +		return ret;
> +
> +	/* Get iout scale per rail. */
> +	ret = mp2891_rails_iout_scale_get(client, data);
> +	if (ret < 0)
> +		return ret;
> +
> +	return pmbus_do_probe(client, info);
> +}
> +
> +static const struct i2c_device_id mp2891_id[] = {
> +	{"mp2891", 0},
> +	{}
> +};
> +MODULE_DEVICE_TABLE(i2c, mp2891_id);
> +
> +static const struct of_device_id __maybe_unused mp2891_of_match[] = {
> +	{.compatible = "mps,mp2891"},
> +	{}
> +};
> +MODULE_DEVICE_TABLE(of, mp2891_of_match);

Needs devicetree documentation, probably as trivial device.

> +
> +static struct i2c_driver mp2891_driver = {
> +	.driver = {
> +		.name = "mp2891",
> +		.of_match_table = mp2891_of_match,
> +	},
> +	.probe = mp2891_probe,
> +	.id_table = mp2891_id,
> +};
> +
> +module_i2c_driver(mp2891_driver);
> +
> +MODULE_AUTHOR("Noah Wang <noahwang.wang@xxxxxxxxxxx>");
> +MODULE_DESCRIPTION("PMBus driver for MPS MP2891 device");
> +MODULE_LICENSE("GPL");
> +MODULE_IMPORT_NS(PMBUS);