2022-03-08 15:21:35

by Andrew Lunn

[permalink] [raw]
Subject: Re: [PATCH net-next 2/3] dt-bindings: net: micrel: Configure latency values and timestamping check for LAN8814 phy

> > Thanks for the reply, but you did not answer my question:
> >
> > Does this mean the hardware itself cannot tell you it is missing the
> > needed hardware?
> >
> > Don't you have different IDs in register 2 and 3 for those devices with clock
> > register and those without?
> >
>

> The purpose of this option is, if both PHY and MAC supports
> timestamping then always timestamping is done in PHY. If
> timestamping need to be done in MAC we need a way to stop PHY
> timestamping. If this flag is used then timestamping is taken care
> by MAC.

This is not a valid use of DT, since this is configuration, not
describing the hardware. There has been recent extension in the UAPI
to allow user space to do this configuration. Please look at that
work.

> Sorry I answered wrong. Latency values vary depending on the position of PHY in board.
> We have used this PHY in different hardware's, where latency values differs based on PHY positioning.
> So we used latency option in DTS file.
> If you have other ideas or I'm wrong please let me know?

So this is a function of the track length between the MAC and the PHY?
How do you determine these values? There is no point having
configuration values if you don't document how to determine what value
should be used.

Andrew


2022-03-08 15:48:12

by Richard Cochran

[permalink] [raw]
Subject: Re: [PATCH net-next 2/3] dt-bindings: net: micrel: Configure latency values and timestamping check for LAN8814 phy

On Tue, Mar 08, 2022 at 02:54:37PM +0100, Andrew Lunn wrote:
> This is not a valid use of DT, since this is configuration, not
> describing the hardware. There has been recent extension in the UAPI
> to allow user space to do this configuration. Please look at that
> work.

Yes, I had an RFC up that hopefully will merge soon.

In the mean time, just implement the PHC/time stamping in your PHY
driver unconditionally.

Thanks,
Richard