Re: [RFC] drivers/hwmon: Corsair Commander Pro driver

[Guenter Roeck]

On Wed, Jun 10, 2020 at 12:26:40AM +0200, Marius Zachmann wrote:
> This is a driver for the Corsair Commander Pro.
> Since this is my first addition to the kernel I would be happy, if you would take a look at it.
> I am using this driver on my computer at home and I had a few people test it.
> 
> The device:
> 
> The Corsair Commander Pro is a USB device, which is usually connected on the internal USB headers on the mainboard.
> It has 6 fan connectors, 4 thermal sensor connectors and 2 RGB connectors.
> It also reads the voltages on the SATA connector, which it needs for power.
> 
> The driver:
> 
> I hope hwmon is the right place.
> The driver is a HID driver, but it uses usb_bulk_msg for communication.
> The device registers as a HID device and there are userspace tools, which use hidraw to access it. I think, it would be good to maintain compatibility with these tools, but the driver can easily be rewritten to be a pure USB driver.
> For now it can read temp sensors, fan speeds, voltage values and set pwm values.
> It also reads the connection status on the fan headers.
> 
> There are a few more things, which I would like to add in the near future:
> * Fan curves (not yet sure about the nicest way to provide sysfs access)
> * Force 3pin or 4pin mode. (Sometimes the device doesn't detect the fans correctly)
> * Setting fixed RPM
> 
> I do not work for Corsair and I intend to keep the driver maintainted as long as I use the device privately.
> 
The above should be limited to the driver description. Discussion should be
below the '---'. Either case, make sure to split your lines.

> Signed-off-by: Marius Zachmann <mail@xxxxxxxxxxxxxxxxx>
> ---
>  Documentation/hwmon/corsair-cpro.rst |  42 +++

Needs to be added to Documentation/hwmon/index.rst.

>  MAINTAINERS                          |   6 +
>  drivers/hwmon/Kconfig                |  10 +
>  drivers/hwmon/Makefile               |   1 +
>  drivers/hwmon/corsair-cpro.c         | 481 +++++++++++++++++++++++++++
>  5 files changed, 540 insertions(+)
>  create mode 100644 Documentation/hwmon/corsair-cpro.rst
>  create mode 100644 drivers/hwmon/corsair-cpro.c
> 
> diff --git a/Documentation/hwmon/corsair-cpro.rst b/Documentation/hwmon/corsair-cpro.rst
> new file mode 100644
> index 000000000000..d4ea1b6b9336
> --- /dev/null
> +++ b/Documentation/hwmon/corsair-cpro.rst
> @@ -0,0 +1,42 @@
> +Kernel driver corsair-cpro
> +==========================
> +
> +Supported devices:
> +
> +  * Corsair Commander Pro
> +  * Corsair Commander Pro (1000D)
> +
> +Author: Marius Zachmann
> +
> +Description
> +-----------
> +
> +This driver implements the sysfs interface for the Corsair Commander Pro.
> +The Corsair Commander Pro is a USB device with 6 fan connectors,
> +4 temperature sensor connectors and 2 Corsair LED connectors.
> +It can read the voltage levels on the SATA power connector.
> +
> +Usage Notes
> +-----------
> +
> +Since it is a USB device, hotswapping is possible. The device is autodetected.
> +
> +Sysfs entries
> +-------------
> +
> +in0_input		Voltage on SATA 12v
> +in1_input		Voltage on SATA 5v
> +in2_input		Voltage on SATA 3.3v
> +
> +temp[0-3]_input		Connected temperature sensors
> +
Index starts with 1 for everything except inX.

> +fan[0-5]_input		Connected fan rpm.
> +fan[0-5]_label		Shows connection status of the fan as detected by the
> +			device.
> +			"fanX nc"   no connection
> +			"fanX 3pin" 3-pin fan detected
> +			"fanX 4pin" 4-pin fan detected
> +fan[0-5]_enable		the driver only reports fan speeds when 1
> +pwm[0-5]		Sets the fan speed. Values from 0-255.
> +			When reading, it reports the last value, which
> +			was set by the driver.
> diff --git a/MAINTAINERS b/MAINTAINERS
> index f08f290df174..169530c7eede 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -4386,6 +4386,12 @@ S:	Maintained
>  F:	Documentation/hwmon/coretemp.rst
>  F:	drivers/hwmon/coretemp.c
>  
> +CORSAIR-CPRO HARDWARE MONITOR DRIVER
> +M:	Marius Zachmann <mail@xxxxxxxxxxxxxxxxx>
> +L:	linux-hwmon@xxxxxxxxxxxxxxx
> +S:	Maintained
> +F:	drivers/hwmon/corsair-cpro.c
> +
>  COSA/SRP SYNC SERIAL DRIVER
>  M:	Jan "Yenya" Kasprzak <kas@xxxxxxxxxx>
>  S:	Maintained
> diff --git a/drivers/hwmon/Kconfig b/drivers/hwmon/Kconfig
> index 288ae9f63588..9f5a808768ca 100644
> --- a/drivers/hwmon/Kconfig
> +++ b/drivers/hwmon/Kconfig
> @@ -439,6 +439,16 @@ config SENSORS_BT1_PVT_ALARMS
>  	  the data conversion will be periodically performed and the data will be
>  	  saved in the internal driver cache.
>  
> +config SENSORS_CORSAIR_CPRO
> +	tristate "Corsair Commander Pro controller"
> +	depends on USB_HID
> +	help
> +	  If you say yes here you get support for the Corsair Commander Pro
> +	  controller.
> +
> +	  This driver can also be built as a module. If so, the module
> +	  will be called corsair-cpro.
> +
>  config SENSORS_DRIVETEMP
>  	tristate "Hard disk drives with temperature sensors"
>  	depends on SCSI && ATA
> diff --git a/drivers/hwmon/Makefile b/drivers/hwmon/Makefile
> index 3e32c21f5efe..18e1ef74ade7 100644
> --- a/drivers/hwmon/Makefile
> +++ b/drivers/hwmon/Makefile
> @@ -56,6 +56,7 @@ obj-$(CONFIG_SENSORS_ATXP1)	+= atxp1.o
>  obj-$(CONFIG_SENSORS_AXI_FAN_CONTROL) += axi-fan-control.o
>  obj-$(CONFIG_SENSORS_BT1_PVT)	+= bt1-pvt.o
>  obj-$(CONFIG_SENSORS_CORETEMP)	+= coretemp.o
> +obj-$(CONFIG_SENSORS_CORSAIR_CPRO) += corsair-cpro.o
>  obj-$(CONFIG_SENSORS_DA9052_ADC)+= da9052-hwmon.o
>  obj-$(CONFIG_SENSORS_DA9055)+= da9055-hwmon.o
>  obj-$(CONFIG_SENSORS_DELL_SMM)	+= dell-smm-hwmon.o
> diff --git a/drivers/hwmon/corsair-cpro.c b/drivers/hwmon/corsair-cpro.c
> new file mode 100644
> index 000000000000..8ba7fb061184
> --- /dev/null
> +++ b/drivers/hwmon/corsair-cpro.c
> @@ -0,0 +1,481 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later

This conflicts with MODULE_LICENSE which explicitly states GPL v2.

> +/*
> + * corsair-cpro.c - Linux driver for Corsair Commander Pro
> + * Copyright (C) 2020 Marius Zachmann <mail@xxxxxxxxxxxxxxxxx>
> + *
> + */
> +
> +#include <linux/init.h>
> +#include <linux/kernel.h>
> +#include <linux/hid.h>
> +#include <linux/hwmon.h>
> +#include <linux/mutex.h>
> +#include <linux/module.h>
> +#include <linux/slab.h>
> +#include <linux/usb.h>
> +
> +#define	hid_to_usb_dev(hid_dev) \
> +	to_usb_device(hid_dev->dev.parent->parent)

Unnecessary line break.

> +
> +#define USB_VENDOR_ID_CORSAIR               0x1b1c
> +#define USB_PRODUCT_ID_CORSAIR_COMMANDERPRO 0x0c10
> +#define USB_PRODUCT_ID_CORSAIR_1000D	    0x1d00
> +
> +#define OUT_BUFFER_SIZE 63
> +#define IN_BUFFER_SIZE 16
> +#define LABEL_LENGTH 10
> +
> +#define CTL_GET_TMP	 0x11  /* byte 1 is channel, rest zero              */
> +			       /* returns temp for channel in bytes 1 and 2 */
> +#define CTL_GET_VOLT	 0x12  /* byte 1 = rail number 12, 5, 3.3 */
> +			       /* returns volt in bytes 1,2       */
> +#define CTL_GET_FAN_CNCT 0x20  /* returns in bytes 1-6   */
> +			       /* 0 for no connect       */
> +			       /* 1 for 3pin, 2 for 4pin */
> +#define CTL_GET_FAN_RPM	 0x21  /* works like CTL_GET_TMP */
> +#define CTL_SET_FAN_FPWM 0x23  /* byte 1 is fan number              */
> +			       /* byte 2 is percentage from 0 - 100 */
> +

Please use tabs consistently after the defined name, eg

#define OUT_BUFFER_SIZE	63
#define IN_BUFFER_SIZE	16
#define LABEL_LENGTH	10

> +struct ccp_device {
> +	struct hid_device *hdev;
> +	struct device *hwmondev;

Not used anywhere.

> +	struct mutex mutex;
> +	int pwm[6];
> +	int fan_enable[6];
> +	char fan_label[6][LABEL_LENGTH];
> +

Unnecessary empty line.

> +};
> +
> +/* send 63 byte buffer and receive response in same buffer */
> +static int send_usb_cmd(struct ccp_device *ccp, u8 *buffer)
> +{
> +	int ret;
> +	struct usb_device *udev = hid_to_usb_dev(ccp->hdev);
> +	int actual_length;
> +
> +
Unnecessary empty line

> +	mutex_lock(&ccp->mutex);
> +
> +	ret = usb_bulk_msg(udev,
> +			usb_sndintpipe(udev, 2),
> +			buffer,
> +			OUT_BUFFER_SIZE,
> +			&actual_length,
> +			1000);

Too many line breaks. All over the place. Please only split lines where
necessary (line length limit is now 100).

> +	if (ret < 0) {
> +		hid_err(ccp->hdev,
> +			"usb_bulk_msg send failed: %d", ret);
> +		goto exit;
> +	}
> +
> +	ret = usb_bulk_msg(udev,
> +			usb_rcvintpipe(udev, 1),
> +			buffer,
> +			IN_BUFFER_SIZE,
> +			&actual_length,
> +			1000);
> +	if (ret) {
> +		hid_err(ccp->hdev,
> +			"usb_bulk_msg receive failed: %d", ret);
> +		goto exit;
> +	}
> +
> +exit:
> +	mutex_unlock(&ccp->mutex);
> +	return ret;
> +}
> +
> +/* for commands, which return just a number depending on a channel: */
> +/* get_temp, get_volt, get_fan_rpm */
> +static int get_data(struct ccp_device *ccp, int command, int channel, long *val)
> +{
> +	int ret = 0;

Unnecessary initialization (almost everywhere).

> +	u8 *buffer;
> +
> +	buffer = kzalloc(OUT_BUFFER_SIZE, GFP_KERNEL);
> +	if (buffer == 0)
> +		return -ENOMEM;

AFAICS all those commands end up locking anyway. Might as well
move the buffer into struct ccp_device and use it from there
(ie lock while using it).

> +
> +	buffer[0] = command;
> +	buffer[1] = channel;
> +	ret = send_usb_cmd(ccp, buffer);
> +	if (ret)
> +		return -EIO;

Please do not replace error codes - not just here, everywhere.

> +
> +	*val = (buffer[1] << 8) + buffer[2];
> +
> +	kfree(buffer);
> +	return ret;

Just return 0;

> +}
> +
> +static int set_pwm(struct ccp_device *ccp, int channel, long val)
> +{
> +	int ret = 0;
> +	u8 *buffer;
> +
> +	if (val > 255)
> +		return -EINVAL;

How about val < 0 ?

> +
> +	ccp->pwm[channel] = val;
> +
> +	/* The Corsair Commander Pro uses values from 0-100 */
> +	val = val << 8;
> +	val = val / 255;
> +	val = val * 100;
> +	val = val >> 8;

	val = DIV_ROUND_CLOSEST(val * 100, 255);

would be much easier to read.

> +
> +	buffer = kzalloc(OUT_BUFFER_SIZE, GFP_KERNEL);
> +	if (buffer == 0)

	if (!buffer)
> +		return -ENOMEM;
> +
> +	buffer[0] = CTL_SET_FAN_FPWM;
> +	buffer[1] = channel;
> +	buffer[2] = val;
> +	ret = send_usb_cmd(ccp, buffer);
> +
> +	kfree(buffer);
> +	return ret == 0 ? 0 : -EIO;
> +}
> +
> +static int get_fan_mode_label(struct ccp_device *ccp, int channel)
> +{
> +	int ret = 0;
> +	int mode;
> +	u8 *buffer;
> +
> +	buffer = kzalloc(OUT_BUFFER_SIZE, GFP_KERNEL);
> +	if (buffer == 0)
> +		return -ENOMEM;
> +
> +	buffer[0] = CTL_GET_FAN_CNCT;
> +	ret = send_usb_cmd(ccp, buffer);
> +	if (ret)
> +		goto exit;
> +
> +	mode = buffer[channel+1];

Space around operators (use checkpatch --strict to find all locations).

> +
> +	switch (mode) {
> +	case 0:
> +		scnprintf(ccp->fan_label[channel], LABEL_LENGTH,
> +			  "fan%d nc", channel+1);
> +		break;
> +	case 1:
> +		scnprintf(ccp->fan_label[channel], LABEL_LENGTH,
> +			  "fan%d 3pin", channel+1);
> +		break;
> +	case 2:
> +		scnprintf(ccp->fan_label[channel], LABEL_LENGTH,
> +			  "fan%d 4pin", channel+1);
> +		break;
> +	default:
> +		dev_err(&ccp->hdev->dev,
> +			"Mode Description %d not implemented", mode);

If this is an error that needs to be logged you have to return an error.
I would suggest to name the label something like "fanX other" instead
in this situation.

> +		break;
> +	}
> +
> +exit:
> +	kfree(buffer);
> +	return ret == 0 ? 0 : -EIO;

While it is unacceptable, I am curious why you override error codes everywhere.

> +}
> +
> +static int get_voltages(struct ccp_device *ccp, int channel, long *val)
> +{
> +	int ret = 0;
> +
> +	ret = get_data(ccp, CTL_GET_VOLT, channel, val);
> +
> +	return ret == 0 ? 0 : -EIO;
> +}
> +
> +static int get_temp(struct ccp_device *ccp, int channel, long *val)
> +{
> +	int ret = 0;
> +
> +	ret = get_data(ccp, CTL_GET_TMP, channel, val);
> +	*val = *val * 10;
> +
> +	return ret == 0 ? 0 : -EIO;
> +}
> +
> +static int get_rpm(struct ccp_device *ccp, int channel, long *val)
> +{
> +	int ret = 0;
> +
> +	if (ccp->fan_enable[channel] != 1)

Valid values are 0 and 1, so != 1 is unnecessary. Use
	if (!ccp->fan_enable[channel])
instead.

> +		return -ENODATA;
> +
> +	ret = get_data(ccp, CTL_GET_FAN_RPM, channel, val);
> +
> +	return ret == 0 ? 0 : -EIO;
> +}
> +
> +static int ccp_read_string(struct device *dev, enum hwmon_sensor_types type,
> +			   u32 attr, int channel, const char **str)
> +{
> +	int ret = 0;
> +	struct ccp_device *ccp = dev_get_drvdata(dev);
> +
> +	switch (type) {
> +	case hwmon_fan:
> +		switch (attr) {
> +		case hwmon_fan_label:
> +			ret = get_fan_mode_label(ccp, channel);
> +			*str = ccp->fan_label[channel];
> +			break;
> +		default:
> +			ret = -EINVAL;
> +			break;
> +		}
> +		break;
> +	default:
> +		ret = -EINVAL;
> +		break;
> +	}
> +
> +	return ret;
> +}
> +
> +static int ccp_read(struct device *dev, enum hwmon_sensor_types type,
> +		    u32 attr, int channel, long *val)
> +{
> +	int ret = 0;
> +	struct ccp_device *ccp = dev_get_drvdata(dev);
> +
> +	switch (type) {
> +	case hwmon_temp:
> +		switch (attr) {
> +		case hwmon_temp_input:
> +			ret = get_temp(ccp, channel, val);
> +			break;
> +		default:
> +			ret = -EINVAL;
> +			break;
> +		}
> +		break;
> +	case hwmon_fan:
> +		switch (attr) {
> +		case hwmon_fan_input:
> +			ret = get_rpm(ccp, channel, val);
> +			break;
> +		case hwmon_fan_enable:
> +			*val = ccp->fan_enable[channel];
> +			break;
> +		default:
> +			ret = -EINVAL;
> +			break;
> +		}
> +		break;
> +	case hwmon_pwm:
> +		switch (attr) {
> +		case hwmon_pwm_input:
> +			*val = ccp->pwm[channel];

This returns 0 if pwm wasn't set. Is this indeed not readable from the
device ?

> +			break;
> +		default:
> +			ret = -EINVAL;
> +			break;
> +		}
> +		break;
> +	case hwmon_in:
> +		switch (attr) {
> +		case hwmon_in_input:
> +			ret = get_voltages(ccp, channel, val);
> +			break;
> +		default:
> +			ret = -EINVAL;
> +			break;
> +		}
> +		break;
> +	default:
> +		ret = -EINVAL;
> +	}
> +	return ret;
> +};
> +
> +static int ccp_write(struct device *dev, enum hwmon_sensor_types type,
> +		     u32 attr, int channel, long val)
> +{
> +	int ret = 0;
> +	struct ccp_device *ccp = dev_get_drvdata(dev);
> +
> +	switch (type) {
> +	case hwmon_fan:
> +		switch (attr) {
> +		case hwmon_fan_enable:
> +			ccp->fan_enable[channel] = val;

Needs to be validated. Accepting random values does not make sense here.

> +			break;
> +		default:
> +			ret = -EINVAL;

-EOPNOTSUPP for those cases.

> +			break;
> +		}
> +		break;
> +	case hwmon_pwm:
> +		switch (attr) {
> +		case hwmon_pwm_input:
> +			set_pwm(ccp, channel, val);

Do not ignore errors.

> +			break;
> +		default:
> +			ret = -EINVAL;
> +			break;
> +		}
> +		break;
> +	default:
> +		ret = -EINVAL;
> +		break;
> +	}
> +	return ret;
> +};
> +
> +static umode_t ccp_is_visible(const void *data, enum hwmon_sensor_types type,
> +			      u32 attr, int channel)
> +{
> +	switch (type) {
> +	case hwmon_chip:
> +		switch (attr) {
> +		case hwmon_chip_update_interval:
> +			return 0644;

		default:

missing everywhere here.

> +		}
> +		break;
> +	case hwmon_temp:
> +		switch (attr) {
> +		case hwmon_temp_input:
> +			return 0444;
> +		}
> +		break;
> +	case hwmon_fan:
> +		switch (attr) {
> +		case hwmon_fan_input:
> +			return 0444;
> +		case hwmon_fan_label:
> +			return 0444;
> +		case hwmon_fan_enable:
> +			return 0644;
> +		}
> +		break;
> +	case hwmon_pwm:
> +		switch (attr) {
> +		case hwmon_pwm_input:
> +			return 0644;
> +		}
> +		break;
> +	case hwmon_in:
> +		switch (attr) {
> +		case hwmon_in_input:
> +			return 0444;
> +		}
> +		break;
> +	default:
> +		break;
> +	}
> +	return 0;
> +};
> +
> +static const struct hwmon_ops ccp_hwmon_ops = {
> +	.is_visible = ccp_is_visible,
> +	.read = ccp_read,
> +	.write = ccp_write,
> +	.read_string = ccp_read_string,
> +};
> +
> +static const struct hwmon_channel_info *ccp_info[] = {
> +	HWMON_CHANNEL_INFO(chip,
> +			   HWMON_C_REGISTER_TZ | HWMON_C_UPDATE_INTERVAL),
> +	HWMON_CHANNEL_INFO(temp,
> +			   HWMON_T_INPUT | HWMON_T_MAX,
> +			   HWMON_T_INPUT | HWMON_T_MAX,
> +			   HWMON_T_INPUT | HWMON_T_MAX,
> +			   HWMON_T_INPUT | HWMON_T_MAX
> +			   ),
> +	HWMON_CHANNEL_INFO(fan,
> +			   HWMON_F_INPUT | HWMON_F_ENABLE | HWMON_F_LABEL,
> +			   HWMON_F_INPUT | HWMON_F_ENABLE | HWMON_F_LABEL,
> +			   HWMON_F_INPUT | HWMON_F_ENABLE | HWMON_F_LABEL,
> +			   HWMON_F_INPUT | HWMON_F_ENABLE | HWMON_F_LABEL,
> +			   HWMON_F_INPUT | HWMON_F_ENABLE | HWMON_F_LABEL,
> +			   HWMON_F_INPUT | HWMON_F_ENABLE | HWMON_F_LABEL
> +			   ),
> +	HWMON_CHANNEL_INFO(pwm,
> +			   HWMON_PWM_INPUT,
> +			   HWMON_PWM_INPUT,
> +			   HWMON_PWM_INPUT,
> +			   HWMON_PWM_INPUT,
> +			   HWMON_PWM_INPUT,
> +			   HWMON_PWM_INPUT
> +			   ),
> +	HWMON_CHANNEL_INFO(in,
> +			   HWMON_I_INPUT,
> +			   HWMON_I_INPUT,
> +			   HWMON_I_INPUT
> +			   ),
> +	NULL
> +};
> +
> +static const struct hwmon_chip_info ccp_chip_info = {
> +	.ops = &ccp_hwmon_ops,
> +	.info = ccp_info,
> +


Another unnecessary empty line.

> +};
> +
> +static int ccp_probe(struct hid_device *hdev, const struct hid_device_id *id)
> +{
> +	struct ccp_device *ccp;
> +	int ret = 0;
> +
> +	ret = hid_parse(hdev);
> +	if (ret) {
> +		hid_err(hdev, "hid_parse failed\n");
> +		goto exit;

		return ret;

> +	}
> +
> +	ccp = devm_kzalloc(&hdev->dev, sizeof(struct ccp_device), GFP_KERNEL);
> +	if (ccp == NULL)
> +		goto exit;

		return -ENOMEM;

> +
> +	mutex_init(&ccp->mutex);
> +
> +	ccp->fan_enable[0] = 1;
> +	ccp->fan_enable[1] = 1;
> +	ccp->fan_enable[2] = 1;
> +	ccp->fan_enable[3] = 1;
> +	ccp->fan_enable[4] = 1;
> +	ccp->fan_enable[5] = 1;
> +
> +	hid_set_drvdata(hdev, ccp);
> +
> +	ccp->hdev = hdev;
> +	ccp->hwmondev = devm_hwmon_device_register_with_info(&hdev->dev,
> +				"corsaircpro",
> +				ccp,
> +				&ccp_chip_info,
> +				0);

This can return an ERR_PTR.

> +
> +exit:
> +	return ret;
> +}
> +
> +static void ccp_remove(struct hid_device *hdev)
> +{
> +	struct ccp_device *ccp;
> +
> +	ccp = hid_get_drvdata(hdev);
> +	mutex_destroy(&ccp->mutex);

The memory is about to be released, so this doesn't really add value.
Just drop the remove function and forget about mutex_destroy().

> +}
> +
> +static const struct hid_device_id ccp_devices[] = {
> +	{ HID_USB_DEVICE(USB_VENDOR_ID_CORSAIR,
> +			 USB_PRODUCT_ID_CORSAIR_COMMANDERPRO) },
> +	{ HID_USB_DEVICE(USB_VENDOR_ID_CORSAIR,
> +			 USB_PRODUCT_ID_CORSAIR_1000D) },
> +	{ }
> +};
> +
> +static struct hid_driver ccp_driver = {
> +	.name = "corsair-cpro",
> +	.id_table = ccp_devices,
> +	.probe = ccp_probe,
> +	.remove = ccp_remove
> +};
> +
> +MODULE_DEVICE_TABLE(hid, ccp_devices);
> +MODULE_LICENSE("GPL v2");
> +
> +module_hid_driver(ccp_driver);
> -- 
> 2.27.0
> 
>