On Sun, Dec 13, 2020 at 12:39:26AM +0100, Daniel Lezcano wrote: > On 12/12/2020 21:08, Matthew Garrett wrote: > > Anything that provides a trip point that has no active notifications and > > doesn't provide any information that tells the kernel to poll it. > > I'm not able to create a setup as you describe working correctly with > the forced passive trip point. > > The forced passive trip can not be detected as there is no comparison > with the defined temperature in the thermal_zone_device_update() function.
The logic seems to be in the step_wise thermal governor. I'm not sure why it would be used in thermal_zone_device_update() - the entire point is that we don't get updates from the device? > If my analysis is correct, this 'feature' is broken since years, more > than 8 years to be exact and nobody complained. I've no problem with it being removed if there are no users, but in that case the justification should be rewritten - ACPI table updates aren't a complete replacement for the functionality offered (and can't be used if the lockdown LSM is being used in any case).