Re: [PATCH net-next 3/3] dpll: zl3073x: add hwmon support for input reference frequencies

On 25/03/2026 11:19, Ivan Vecera wrote:

On 3/24/26 10:36 PM, Jakub Kicinski wrote:

quoted

On Tue, 24 Mar 2026 13:59:19 +0100 Ivan Vecera wrote:

quoted

quoted

quoted

In the hardware monitoring world one would have min/max attributes and
one or more alarm attributes. I never heard about the idea of 
reporting
deviations, and for typical hardware monitoring attributes it does not
really make sense. What would be a "deviation" of a temperature/
voltage/current/power/humidity sensor ? Such attributes typically have
an operational range, and they are allowed and even expected to
fluctuate within that range.

Ivan, my take on all the above is that using the HWmon interface 
here is
stretching it too much. I think it would be better to move debugfs
and/or netlink events.

I'd rather avoid debugfs... My proposal is to expose absolute measured
frequency attribute of dpll-pin and follow phase-offset-monitor
functionality:

So:
* add real-frequency attribute to dpll pin
* add real-frequency-monitor attribute dpll device
* user will be able to enable/disable monitoring by enabling/disabling
    real-frequency-monitor feature (similarly to phase-offset-monitor)

Thoughts?

I don't have a strong opinion. IDK where to draw the line between DPLL
and "random functionality some devices may have". We have 3 DPLL
maintainers, let's see what their majority opinion is..

I think that line has already been crossed by temperature reporting,
which has been there since the beginning. Actual frequency measurement
is, in my opinion, much closer to the DPLL world than temperature 
reporting. :-)

Frequency measurements are DPLL-world properties indeed. Temperature
reporting was added as it has some influence on the stability of DPLL 
devices as well as on any oscillators.

Anyway, I have the relevant patch-set already prepared, so I can submit
it if you want to see it.

Feel free to submit, I'll be happy to review it.