[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CABgGipWHztqv=8qNNWSc1srcy0tyKg9tUcD-1qKBO-FzGFfR8w@mail.gmail.com>
Date: Sun, 23 Oct 2022 23:04:14 +0800
From: Andy Chiu <andy.chiu@...ive.com>
To: Rob Herring <robh@...nel.org>
Cc: Michal Simek <michal.simek@....com>, davem@...emloft.net,
kuba@...nel.org, michal.simek@...inx.com,
radhey.shyam.pandey@...inx.com, netdev@...r.kernel.org,
devicetree@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
krzysztof.kozlowski+dt@...aro.org, pabeni@...hat.com,
edumazet@...gle.com, greentime.hu@...ive.com
Subject: Re: [PATCH net-next 2/2] dt-bindings: add mdio frequency description
> > > +- xlnx,mdio-freq: Define the clock frequency of the MDIO bus. If the property
> > > + does not pressent on the DT, then the mdio driver would use
> > > + the default 2.5 MHz clock, as mentioned on 802.3 spc.
> >
> > Isn't it better to specify it based on ccf description. It means &clk and
> > used clock framework to find value?
>
Maybe I missed something, but I'd prefer using a number to define the
frequency of the bus rather than basing on the ccf description for our
use case. First, it is more straightforward because ccf requires us to
define a separate DT node and point to it. And currently we don't need
to do extra management that would benefit from using ccf. All we need
to do with the clock is to get its frequency.
> Or use 'bus-frequency' which IIRC is defined for MDIO.
>
I found that "bus-frequency" first appeared in fsl driver but was
replaced later by the standard one, which is "clock-frequency" as
defined in mdio.yaml. We will submit a v2 patch to configure
non-convention MDIO bus frequency with that DT entry.
Thanks and regards,
Andy
Powered by blists - more mailing lists