[PATCH v4 1/2] dt-bindings: hwmon: emc2305: Add fan-shutdown-percent property

Guenter Roeck linux at roeck-us.net
Wed Mar 25 15:33:11 PDT 2026


On 3/25/26 14:56, Rob Herring wrote:
> On Fri, Mar 20, 2026 at 05:29:58PM +0200, florin.leotescu at oss.nxp.com wrote:
>> From: Florin Leotescu <florin.leotescu at nxp.com>
>>
>> The EMC2305 fan controller supports multiple independent PWM fan
>> outputs. Some systems require fans to enter a defined safe state
>> during system shutdown or reboot handoff, until firmware or the next
>> boot stage reconfigures the controller.
>>
>> Add an optional "fan-shutdown-percent" property to fan child nodes
>> allowing the shutdown fan speed to be configured per fan output.
> 
> Why not a common fan property in fan-common.yaml?
> 
> We generally specify fan speeds in RPM (and then map RPMs to duty cycle
> for PWM).
> 

The problem here is mentioning "fan speed" instead of referring to pwm
duty cyle in the first place. It is not just misleading, it is wrong.

It is impossible to associate fan speed with a specific duty cycle because
that differs for each fan. It isn't even consistent for the same fan model -
one never knows what fan speed one gets for a given duty cycle. It depends
on the air flow in the chassis and on fan-to-fan deviations. It will change
over the lifetime of a fan. It may even change with the ambient temperature.
Fan controllers use a feedback loop for that purpose, but that doesn't
work here since, after all, this is a shutdown parameter.

I don't think anything but specifying a shutdown duty cycle is
feasible here.

> Also, we have (or can have) cooling levels defined. Perhaps we should
> define the cooling level for shutdown? Maybe other things with cooling
> levels need a shutdown level too?
> 

Unless I am missing something, the thermal subsystem does not currently have
a concept of a "shutdown cooling state". The driver supports registering
with the thermal subsystem, but it is not mandatory. Even if the thermal
subsystem would support shutdown cooling states or a similar concept,
I do not think it would be a good idea to mandate its existence or in a fan
controller driver.

Thanks,
Guenter




More information about the linux-arm-kernel mailing list