Am 08.04.24 um 17:48 schrieb Ilpo Järvinen:
On Mon, 8 Apr 2024, Armin Wolf wrote:Currently, the driver does only supports a custom sysfs interface to allow userspace to read the fan speed. Add support for the standard hwmon interface so users can read the fan speed with standard tools like "sensors". Signed-off-by: Armin Wolf <W_Armin@xxxxxx> --- drivers/acpi/Makefile | 1 + drivers/acpi/fan.h | 2 ++ drivers/acpi/fan_core.c | 7 ++++ drivers/acpi/fan_hwmon.c | 78 ++++++++++++++++++++++++++++++++++++++++ 4 files changed, 88 insertions(+) create mode 100644 drivers/acpi/fan_hwmon.c diff --git a/drivers/acpi/Makefile b/drivers/acpi/Makefile index d69d5444acdb..9a2e03acc1be 100644 --- a/drivers/acpi/Makefile +++ b/drivers/acpi/Makefile @@ -83,6 +83,7 @@ obj-$(CONFIG_ACPI_TINY_POWER_BUTTON) += tiny-power-button.o obj-$(CONFIG_ACPI_FAN) += fan.o fan-objs := fan_core.o fan-objs += fan_attr.o +fan-objs += fan_hwmon.o obj-$(CONFIG_ACPI_VIDEO) += video.o obj-$(CONFIG_ACPI_TAD) += acpi_tad.o diff --git a/drivers/acpi/fan.h b/drivers/acpi/fan.h index e7b4b4e4a55e..45c2637566da 100644 --- a/drivers/acpi/fan.h +++ b/drivers/acpi/fan.h @@ -56,4 +56,6 @@ struct acpi_fan { int acpi_fan_get_fst(struct acpi_device *device, struct acpi_fan_fst *fst); int acpi_fan_create_attributes(struct acpi_device *device); void acpi_fan_delete_attributes(struct acpi_device *device); + +int devm_acpi_fan_create_hwmon(struct acpi_device *device); #endif diff --git a/drivers/acpi/fan_core.c b/drivers/acpi/fan_core.c index ff72e4ef8738..6bbdbb914e95 100644 --- a/drivers/acpi/fan_core.c +++ b/drivers/acpi/fan_core.c @@ -7,6 +7,7 @@ * Copyright (C) 2022 Intel Corporation. All rights reserved. */ +#include <linux/kconfig.h> #include <linux/kernel.h> #include <linux/module.h> #include <linux/init.h> @@ -336,6 +337,12 @@ static int acpi_fan_probe(struct platform_device *pdev) if (result) return result; + if (IS_REACHABLE(CONFIG_HWMON)) { + result = devm_acpi_fan_create_hwmon(device); + if (result) + return result; + } + result = acpi_fan_create_attributes(device); if (result) return result; diff --git a/drivers/acpi/fan_hwmon.c b/drivers/acpi/fan_hwmon.c new file mode 100644 index 000000000000..4f2bec8664f4 --- /dev/null +++ b/drivers/acpi/fan_hwmon.c @@ -0,0 +1,78 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * fan_hwmon.c - hwmon interface for the ACPI Fan driver + * + * Copyright (C) 2024 Armin Wolf <W_Armin@xxxxxx> + */ + +#include <linux/acpi.h> +#include <linux/hwmon.h> +#include <linux/limits.h> + +#include "fan.h" + +static umode_t acpi_fan_is_visible(const void *drvdata, enum hwmon_sensor_types type, u32 attr, + int channel) +{ + return 0444; +} + +static int acpi_fan_read(struct device *dev, enum hwmon_sensor_types type, u32 attr, int channel, + long *val) +{ + struct acpi_device *adev = dev_get_drvdata(dev); + struct acpi_fan_fst fst; + int ret; + + switch (type) { + case hwmon_fan: + ret = acpi_fan_get_fst(adev, &fst); + if (ret < 0) + return ret; + + switch (attr) { + case hwmon_fan_input: + if (fst.speed > LONG_MAX) + return -EOVERFLOW; + + *val = fst.speed; + return 0; + case hwmon_fan_fault: + *val = (fst.speed == U32_MAX); + return 0;Is it okay to return 0 in this case?
Hi, i think so, since the value of the attribute (with the meaning of "is the rpm value ok?") is being correctly stored in val. If acpi_fan_get_fst() fails, we already return a negative error code. Thanks, Armin Wolf
+ default: + break; + } + break; + default: + break; + } + + return -EOPNOTSUPP; +} + +static const struct hwmon_ops acpi_fan_ops = { + .is_visible = acpi_fan_is_visible, + .read = acpi_fan_read, +}; + +static const struct hwmon_channel_info * const acpi_fan_info[] = { + HWMON_CHANNEL_INFO(fan, + HWMON_F_INPUT | HWMON_F_FAULT),One line.
