Re: [PATCH v7 2/7] dt-bindings: net: add amlogic,t7-dwmac
From: Chuan Liu
Date: Mon Nov 16 2026 - 09:19:38 EST
Hi Martin, Cc Lucas, Conor,
On 11/16/2026 1:56 AM, Martin Blumenstingl wrote:
On Fri, Nov 13, 2026 at 1:18 PM Lucas Tanure <tanure@xxxxxxxxx> wrote:
Looks good to me in general
The T7 has the same ethernet glue as the G12A but needs two clocks of
its own. The controller reaches memory through a pipeline stage with
its own gate; nothing else claims it, so it is switched off as unused
and transfers never finish. The glue registers share a gate with the
MDIO multiplexer, which probes after the controller has programmed
them: the writes are lost and the port keeps the bootloader's delays.
With a newer core, a clock list of its own and delays that follow
ethernet-controller.yaml, the T7 gets a binding of its own: the PHY
adds the 2ns, and only the receive delay line offers fine tuning, so
only rx-internal-delay-ps is valid here.
Signed-off-by: Lucas Tanure <tanure@xxxxxxxxx>
[...]
+ - description:Please give Chuan Liu a few days to clarify this clock input: [0]
+ APB clock of the glue registers PRG_ETH0 and PRG_ETH1.
+ Without it, they read as zero and ignore writes.
Thanks for the reminder, I have confirmed that Lucas' understanding is right. It is indeed necessary to rely on this clock to access PRG_ETH0 and PRG_ETH1 normally. I have replied in the patch of v5 [0].
[0] https://lore.kernel.org/linux-amlogic/bd4c8a15-ab13-459f-8fb5-fe3f3b04af43@xxxxxxxxxxx/
It would be great if he could provide more details on it. If there's
no reply then I suggest going with Lucas' testing finding though
(which means that clock input is correct)
[0] https://lore.kernel.org/linux-amlogic/CAJX_Q+0seFwDvd6zksLnJjWYJJY1QcLc1gAjr=FN1p+aTdumpw@xxxxxxxxxxxxxx/
--
Best regards,
Chuan