Re: reserved-memory: node naming - generic name vs binding recommendation

On Thu, Mar 19, 2026 at 10:27=E2=80=AFAM Vitor Soares <[email protected]> wr=
ote:
>
> Hi,
>
> I have a question about node naming conventions for static reserved-memor=
y
> allocations.
>
> The reserved-memory.yaml binding recommends that node names reflect the p=
urpose
> of the node (e.g. "framebuffer" or "dma-pool"). The base Devicetree
> Specification (section 2.2.1) on the other hand recommends using generic =
names
> from its suggested list, which includes "memory", implying a "memory@<add=
r>"
> form.
>
> For nodes with a static allocation, is the binding's guidance to use purp=
ose-
> reflecting names the recommendation, taking precedence over the base spec=
's
> generic "memory@" convention?

"memory" is for memory nodes (i.e. ones with 'device_type =3D
"memory"'), so not really for reserved-memory nodes.

Yes, we would prefer some generic name, but there isn't one defined.
Until one is defined, I have little opinion about what name is used
(beyond global conventions like don't use '_'). Names reflecting the
purpose for /reserved-memory is so widely used already, I don't really
think it is something we can go back and fix.

Rob