On 5/4/23 04:36, Naresh Solanki wrote:
Hi Krzysztof,
[ ... ]
No, we cannot, because we asked you to fix things there. Your entire
explanation about compatible and driver is not related to the comment
you received: bindings should be complete. You argue that bindings do
not have to be complete, because of something with driver. This is not
related. Bindings are not for driver.
I understand that complete bindings are important, but as the driver is already merged and functional, my immediate goal is to enable its use on my machine. I will work on a separate patch to include the interrupts in both binding & driver.
As a follow-up, since it came up in a separate context:
Bindings and driver are independent of each other. _Bindings_
are supposed to be complete. However, the existence of a property
in the bindings description does not have to be reflected in
the driver.
FWIW, you _could_ have added the device to the list of trivial
devices. The only really mandatory property is vdd, and every
chip has that. All other properties are really about configuration
and/or fan properties, and I don't even know how to describe fan
properties (such as pulses per revolution, pwm parameters,
fan speed limits, the relationship between pwm outputs
and fan inputs, the relationship between fan speed input
and pwm output, or fan spin-up requirements) in devicetree.
Guenter
