summaryrefslogtreecommitdiffstats
path: root/drivers/acpi/fan_core.c
AgeCommit message (Collapse)AuthorFilesLines
2022-02-25ACPI: fan: Add additional attributes for fine grain controlSrinivas Pandruvada1-1/+1
Add additional attributes, which helps in implementing algorithm in the user space to optimize fan control. These attributes are presented in the same directory as the existing performance state attributes. Additional attributes: 1. Support of fine grain control Publish support of presence of fine grain control so that fan speed can be tuned correctly. This attribute is called "fine_grain_control". 2. fan speed Publish the actual fan rpm in sysfs. Knowing fan rpm is helpful to reduce noise level and use passive control instead. Also fan performance may not be same over time, so the same control value may not be enough to run the fan at a speed. So a feedback value of speed is helpful. This sysfs attribute is called "fan_speed_rpm". Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2022-02-25ACPI: fan: Properly handle fine grain controlSrinivas Pandruvada1-26/+68
When _FIF object specifies support for fine grain control, then fan speed can be set from 0 to 100% with the recommended minimum "step size" via _FSL object. Here the control value doesn't need to match any value from _FPS object. Currently we have a simple solution implemented which just pick maximum control value from _FPS to display the actual state, but this is not optimal when there is a big window between two control values in _FPS. Also there is no way to set to any speed which doesn't match control values in _FPS. The system firmware can start the fan at speed which doesn't match any control value. To support fine grain control (when supported) via thermal sysfs: - cooling device max state is not _FPS state count but it will be 100 / _FIF.step_size Step size can be from 1 to 9. - cooling device current state is _FST.control / _FIF.step_size - cooling device set state will set the control value cdev.curr_state * _FIF.step_size plus any adjustment for 100%. By the spec, when control value do not sum to 100% because of _FIF.step_size, OSPM may select an appropriate ending Level increment to reach 100%. There is no rounding during calculation. For example if step size is 6: thermal sysfs cooling device max_state = 100/6 = 16 So user can set any value from 0-16. If the system boots with a _FST.control which is not multiples of step_size, the thermal sysfs cur_state will be based on the range. For example for step size = 6: _FST.control thermal sysfs cur_state ------------------------------------------------ 0-5 0 6-11 1 .. .. 90-95 15 96-100 16 While setting the _FST.control, the compensation will be at the last step for cur_state = 16, which will set the _FST.control to 100. Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2022-02-25ACPI: fan: Optimize struct acpi_fan_fifSrinivas Pandruvada1-1/+7
We don't need u64 to store the information about _FIF. There are two booleans (fine_grain_ctrl and low_speed_notification) and one field step_size which can take value from 1-9. There are no internal users of revision field. So convert all fields to u8, by not directly extracting the _FIF info the struct. Use an intermediate buffer to extract and assign. This will help to do u32 math using these fields. No functional changes are expected. Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2022-02-25ACPI: fan: Separate file for attributes creationSrinivas Pandruvada1-0/+416
Move the functionality of creation of sysfs attributes under acpi device to a new file fan_attr.c. This cleans up the core fan code, which just use thermal sysfs interface. The original fan.c is renamed to fan_core.c. No functional changes are expected. Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>