Re: [PATCH 3/3] hwmon: cros-ec-hwmon: Add Chromium-EC HWMON driver

From: Guenter Roeck
Date: Thu Apr 13 2017 - 17:34:43 EST


On Fri, Apr 07, 2017 at 03:00:10PM -0700, Moritz Fischer wrote:
> From: Moritz Fischer <mdf@xxxxxxxxxx>
>
> This adds a hwmon driver for the Chromium EC's fans
> and temperature sensors.
>
> Signed-off-by: Moritz Fischer <mdf@xxxxxxxxxx>
> ---
>
> This one still needs some work, but I figured some early feedback might not hurt.
> Specifically I was wondering if using the devm_hwmon_register_with_info() is
> preferable to the devm_hwmon_register_with_groups().
>

Please use devm_hwmon_register_with_info().

> The EC has a bunch of additional features such as setting thermal limits etc,
> which I'd still like to add but I figured I'll get some feedback on what I got so far.
>

Those would probably be more appropriate for a thermal driver.

> Thanks,
>
> Moritz
>
> ---
> drivers/hwmon/Kconfig | 8 ++
> drivers/hwmon/Makefile | 1 +
> drivers/hwmon/cros-ec-hwmon.c | 244 ++++++++++++++++++++++++++++++++++++++++++
> 3 files changed, 253 insertions(+)
> create mode 100644 drivers/hwmon/cros-ec-hwmon.c
>
> diff --git a/drivers/hwmon/Kconfig b/drivers/hwmon/Kconfig
> index 0649d53f3..3b9155f 100644
> --- a/drivers/hwmon/Kconfig
> +++ b/drivers/hwmon/Kconfig
> @@ -1254,6 +1254,14 @@ config SENSORS_PCF8591
> These devices are hard to detect and rarely found on mainstream
> hardware. If unsure, say N.
>
> +config SENSORS_CROS_EC
> + tristate "ChromeOS EC hwmon"
> + depends on MFD_CROS_EC
> + help
> + If you say yes here you get hwmon support that will expose the
> + ChromeOS internal sensors for fanspeed and temperature to the
> + Linux hwmon subsystem.
> +
> source drivers/hwmon/pmbus/Kconfig
>
> config SENSORS_PWM_FAN
> diff --git a/drivers/hwmon/Makefile b/drivers/hwmon/Makefile
> index 5509edf..e59b5da 100644
> --- a/drivers/hwmon/Makefile
> +++ b/drivers/hwmon/Makefile
> @@ -134,6 +134,7 @@ obj-$(CONFIG_SENSORS_PC87360) += pc87360.o
> obj-$(CONFIG_SENSORS_PC87427) += pc87427.o
> obj-$(CONFIG_SENSORS_PCF8591) += pcf8591.o
> obj-$(CONFIG_SENSORS_POWR1220) += powr1220.o
> +obj-$(CONFIG_SENSORS_CROS_EC) += cros-ec-hwmon.o
> obj-$(CONFIG_SENSORS_PWM_FAN) += pwm-fan.o
> obj-$(CONFIG_SENSORS_S3C) += s3c-hwmon.o
> obj-$(CONFIG_SENSORS_SCH56XX_COMMON)+= sch56xx-common.o
> diff --git a/drivers/hwmon/cros-ec-hwmon.c b/drivers/hwmon/cros-ec-hwmon.c
> new file mode 100644
> index 0000000..29d8b06
> --- /dev/null
> +++ b/drivers/hwmon/cros-ec-hwmon.c
> @@ -0,0 +1,244 @@
> +/*
> + * Copyright (c) 2017, National Instruments Corp.
> + *
> + * Chromium EC Fan speed and temperature sensor driver
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; version 2 of the License.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + */
> +
> +#include <linux/device.h>
> +#include <linux/kernel.h>
> +#include <linux/slab.h>
> +#include <linux/sysfs.h>
> +#include <linux/spi/spi.h>
> +#include <linux/module.h>
> +#include <linux/err.h>
> +#include <linux/of_platform.h>
> +#include <linux/hwmon.h>
> +#include <linux/hwmon-sysfs.h>
> +#include <linux/bitops.h>
> +#include <linux/mfd/cros_ec.h>
> +
> +struct cros_ec_hwmon_priv {
> + struct cros_ec_device *ec;
> + struct device *hwmon_dev;
> +
> + struct attribute **attrs;
> +
> + struct attribute_group attr_group;
> + const struct attribute_group *groups[2];
> +};
> +
> +#define KELVIN_TO_MILLICELSIUS(x) (((x) - 273) * 1000)
> +
> +static int __cros_ec_hwmon_probe_fans(struct cros_ec_hwmon_priv *priv)
> +{
> + int err, idx;
> + uint16_t data;
> +
> + for (idx = 0; idx < EC_FAN_SPEED_ENTRIES; idx++) {
> + err = cros_ec_read_mapped_mem16(priv->ec,
> + EC_MEMMAP_FAN + 2 * idx,
> + &data);

Kind of interesting. ectool and ec code all assume that there is an
endianness match. Guess we can do the same. Just add a respective comment.

> + if (err)
> + return err;
> +
> + if (data == EC_FAN_SPEED_NOT_PRESENT)
> + break;
> + }
> +
> + return idx;
> +}
> +
> +static int __cros_ec_hwmon_probe_temps(struct cros_ec_hwmon_priv *priv)
> +{
> + uint8_t data;
> + int err, idx;
> +
> + err = cros_ec_read_mapped_mem8(priv->ec, EC_MEMMAP_THERMAL_VERSION,
> + &data);
> +
> + /* if we have a read error, or EC_MEMMAP_THERMAL_VERSION is not set,
> + * most likely we don't have temperature sensors ...
> + */
> + if (err || !data)
> + return 0;
> +
> + for (idx = 0; idx < EC_TEMP_SENSOR_ENTRIES; idx++) {

This doesn't cover EC_MEMMAP_THERMAL_VERSION >= 2 which supports
an additional set of thermal sensors.

> + err = cros_ec_read_mapped_mem8(priv->ec,
> + EC_MEMMAP_TEMP_SENSOR + idx,
> + &data);

You can read all data in one go using the provided callback function.

> + if (err)
> + return idx;
> +
> + /* this assumes that they're all good up to idx */
> + switch (data) {
> + case EC_TEMP_SENSOR_NOT_PRESENT:
> + case EC_TEMP_SENSOR_ERROR:
> + case EC_TEMP_SENSOR_NOT_POWERED:
> + case EC_TEMP_SENSOR_NOT_CALIBRATED:

Not sure if we can assume sensors to be in sequential order.
ectool doesn't make that assumption (it only does it for fans).
It might be safer to generate attributes for all possible sensors
and then use the is_visible function to determine which ones are
enabled.

> + return idx;
> + default:
> + continue;
> + };
> + }
> +
> + return idx;
> +}
> +
> +static ssize_t cros_ec_hwmon_read_fan_rpm(struct device *dev,
> + struct device_attribute *attr,
> + char *buf)
> +{
> + uint16_t data;
> + int err;
> + struct sensor_device_attribute *sattr = to_sensor_dev_attr(attr);
> + struct cros_ec_hwmon_priv *priv = dev_get_drvdata(dev);
> +
> + err = cros_ec_read_mapped_mem16(priv->ec,
> + EC_MEMMAP_FAN + 2 * sattr->index,
> + &data);
> + if (err)
> + return err;
> +

Watch out for EC_FAN_SPEED_STALLED.

> + return sprintf(buf, "%d\n", data);
> +}
> +
> +static ssize_t cros_ec_hwmon_read_temp(struct device *dev,
> + struct device_attribute *attr,
> + char *buf)
> +{
> + uint8_t data;
> + int err, tmp;
> +
> + struct sensor_device_attribute *sattr = to_sensor_dev_attr(attr);
> + struct cros_ec_hwmon_priv *priv = dev_get_drvdata(dev);
> +
> + err = cros_ec_read_mapped_mem8(priv->ec,
> + EC_MEMMAP_TEMP_SENSOR + 1 * sattr->index,
> + &data);
> + if (err)
> + return err;
> +
> + switch (data) {
> + case EC_TEMP_SENSOR_NOT_PRESENT:
> + case EC_TEMP_SENSOR_ERROR:
> + case EC_TEMP_SENSOR_NOT_POWERED:
> + case EC_TEMP_SENSOR_NOT_CALIBRATED:
> + dev_info(priv->ec->dev, "Failure: result=%d\n", data);

Please no logging noise.

> + return -EIO;
> + }
> +
> + /* make sure we don't overflow when adding offset*/
> + tmp = data + EC_TEMP_SENSOR_OFFSET;
> +
> + return sprintf(buf, "%d\n", KELVIN_TO_MILLICELSIUS(tmp));

Overall, a REG_TO_MILLICELSIUS() macro might be easier.

> +}
> +
> +static int cros_ec_hwmon_probe(struct platform_device *pdev)
> +{
> + struct cros_ec_device *ec = dev_get_drvdata(pdev->dev.parent);
> + struct cros_ec_hwmon_priv *ec_hwmon;
> + struct sensor_device_attribute *attr;
> + int num_fans, num_temps, i;
> +
> + ec_hwmon = devm_kzalloc(&pdev->dev, sizeof(*ec_hwmon), GFP_KERNEL);
> + if (!ec_hwmon)
> + return -ENOMEM;
> + ec_hwmon->ec = ec;
> +
> + num_fans = __cros_ec_hwmon_probe_fans(ec_hwmon);
> + if (num_fans < 0)
> + return num_fans;
> +
> + num_temps = __cros_ec_hwmon_probe_temps(ec_hwmon);
> + if (num_fans < 0)
> + return num_temps;
> +
> + ec_hwmon->attrs = devm_kzalloc(&pdev->dev,
> + sizeof(*ec_hwmon->attrs) *
> + (num_fans + num_temps + 1),
> + GFP_KERNEL);
> + if (!ec_hwmon->attrs)
> + return -ENOMEM;
> +
> + for (i = 0; i < num_fans; i++) {
> + attr = devm_kzalloc(&pdev->dev, sizeof(*attr), GFP_KERNEL);
> + if (!attr)
> + return -ENOMEM;
> + sysfs_attr_init(&attr->dev_attr.attr);
> + attr->dev_attr.attr.name = devm_kasprintf(&pdev->dev,
> + GFP_KERNEL,
> + "fan%d_input",
> + i);
> + if (!attr->dev_attr.attr.name)
> + return -ENOMEM;
> +
> + attr->dev_attr.show = cros_ec_hwmon_read_fan_rpm;
> + attr->dev_attr.attr.mode = S_IRUGO;
> + attr->index = i;
> + ec_hwmon->attrs[i] = &attr->dev_attr.attr;
> +
> + }
> +
> + for (i = 0; i < num_temps; i++) {
> + attr = devm_kzalloc(&pdev->dev, sizeof(*attr), GFP_KERNEL);
> + if (!attr)
> + return -ENOMEM;
> + sysfs_attr_init(&attr->dev_attr.attr);
> + attr->dev_attr.attr.name = devm_kasprintf(&pdev->dev,
> + GFP_KERNEL,
> + "temp%d_input",
> + i);
> + if (!attr->dev_attr.attr.name)
> + return -ENOMEM;
> +
> + attr->dev_attr.show = cros_ec_hwmon_read_temp;
> + attr->dev_attr.attr.mode = S_IRUGO;
> + attr->index = i;
> + ec_hwmon->attrs[i + num_fans] = &attr->dev_attr.attr;
> +
> + }
> +
> + ec_hwmon->attr_group.attrs = ec_hwmon->attrs;
> + ec_hwmon->groups[0] = &ec_hwmon->attr_group;
> +
> + ec_hwmon->hwmon_dev = devm_hwmon_device_register_with_groups(&pdev->dev,
> + "ec_hwmon", ec_hwmon, ec_hwmon->groups);
> +
hwmon_dev is only used in this function. There is no need to store it in
ec_hwmon.

> + if (IS_ERR(ec_hwmon->hwmon_dev))
> + return PTR_ERR(ec_hwmon->hwmon_dev);
> +
return PTR_ERR_OR_ZERO(hwmon_dev);

> + platform_set_drvdata(pdev, ec_hwmon);
> +
I don't see where this is used.

> + return 0;
> +}
> +
> +#ifdef CONFIG_OF
> +static const struct of_device_id cros_ec_hwmon_of_match[] = {
> + { .compatible = "google,cros-ec-hwmon" },
> + {},
> +};
> +MODULE_DEVICE_TABLE(of, cros_ec_hwmon_of_match);
> +#endif
> +
> +static struct platform_driver cros_ec_hwmon_driver = {
> + .probe = cros_ec_hwmon_probe,
> + .driver = {
> + .name = "cros-ec-hwmon",
> + .of_match_table = of_match_ptr(cros_ec_hwmon_of_match),
> + },
> +};
> +module_platform_driver(cros_ec_hwmon_driver);
> +
> +MODULE_LICENSE("GPL v2");
> +MODULE_DESCRIPTION("ChromeOS EC Hardware Monitor driver");
> +MODULE_ALIAS("platform:cros-ec-hwmon");
> +MODULE_AUTHOR("Moritz Fischer <mdf@xxxxxxxxxx>");
> --
> 2.7.4
>