Re: [PATCH] arm64: dts: qcom: ipq9574: Use 'usb-phy' for node names

[mr . nuke . me]

On 12/10/25 12:03 AM, Krzysztof Kozlowski wrote:
> On 09/12/2025 22:59, mr.nuke.me@xxxxxxxxx wrote:
> > On 12/9/25 3:48 PM, Krzysztof Kozlowski wrote:
> > > On 09/12/2025 17:26, mr.nuke.me@xxxxxxxxx wrote:
> > > > On 12/9/25 10:17 AM, Dmitry Baryshkov wrote:
> > > > > On Tue, Dec 09, 2025 at 10:07:54AM -0600, Alexandru Gagniuc wrote:
> > > > > > The devicetree spec allows node names of "usb-phy". So be more
> > > > > > specific for the USB PHYs, and name the nodes "usb-phy" instead of
> > > > > > just "phy".
> > > > >
> > > > > Why? "phy" is more generic.
> > > >
> > > > Hi Dmitry,
> > > >
> > > > The goal is to be more specific. I find usb-phy, ethernet-phy and others
> > >
> > > We do not have such goal. Where did you find that goal documented?
> >
> > If the goal isn't to be specific, clear, and readable, what is it? Why not be generic, and call subnodes node@, or dev@ ?
>
>
> Did you read the spec you referred to? What sort of class of devices
> represents "node"?

My statement was intended to be a reduction ad absurdum to the generic
naming argument, rather than my public exam on the dt 0.4 spec.

I find it useful to have node names that identify the function as
clearly as possible, or to see ethernet-phy and usb-phy under
/proc/device-tree/soc@0/ and /sys/bus/platform/devices/. That was
_my_ goal. Obviously, you and Dmitry disagree with that goal.

Alex