Re: [PATCH net-next 5/5] net: mdio: bcm-unimac: implement timestamped MDIO writes

[James Clark]

On Sun, Oct 11, 2026 at 8:22 PM Nicolai Buchwitz <nb@xxxxxxxxxxx> wrote:

> > Let me just confirm how I would need to adapt my patch with that fix:
> > in the pdata branch of the probe, I would call devm_clk_get_optional
> > to find the clock on the mdio node, falling back to pdata->clk. I
> > would still add the chip-specific compatible string because the number
> > of MDC periods has only been measured on that specific hardware.
>
> In the unimac patch you would only need clk_get_rate() and can gate
> write_sts() on it. Not sure about pdata->clk though: bcmgenet passes
> its "enet" MAC clock there, which isn't necessarily the MDC reference.
> If write_sts() only trusts a clock on the mdio node itself, you don't
> need the compatible either.

Sashiko wanted write_sts limited to BCM2711 because the 63.5 to 64.5
window has only been measured on BCM2711 and we don't have evidence
that it applies elsewhere. If it doesn't, then we could potentially
violate what PTP_SYS_OFFSET_EXTENDED promises. I think one could argue
that this is too conservative: clause 22 specifies 64 bits and the
63.5 to 64.5 range falls out of that given a free-running MDC and some
other reasonable assumptions. Andrew made a similar point on 4/5
(https://lore.kernel.org/netdev/c136b0ad-6d7e-4695-b988-55eb37550d01@xxxxxxx/).
Florian said he wanted a compatible string but I am not sure whether
because of this or because of the clock. In any case, I am happy to do
it whichever way you and Florian decide.

In the probe, I believe I still need to add a call to
devm_clk_get_optional() in the pdata branch, which is used when GENET
creates the UniMAC device. At the moment this branch does priv->clk =
pdata->clk, so it would not pick up the new GISB clock from the mdio
node. I think this new clock would live alongside the existing
pdata->clk: the write_sts rate would come from the mdio node clock,
but pdata->clk would be used for gating. I was not planning for this
series to make clk_set use the new GISB clock.

James