On Wed, Jun 10, 2026 at 09:39:25AM +0200, Krzysztof Kozlowski wrote:
> On Tue, Jun 09, 2026 at 11:33:03AM -0600, Mathieu Poirier wrote:
> > On Tue, 9 Jun 2026 at 11:06, Frank Li <[email protected]> wrote:
> > >
> > > On Tue, Jun 09, 2026 at 10:40:06AM -0600, Mathieu Poirier wrote:
> > > > [You don't often get email from [email protected]. Learn why
> > > > this is important at https://aka.ms/LearnAboutSenderIdentification ]
> > > >
> > > > On Fri, Jun 05, 2026 at 04:36:18AM -0700, Laurentiu Mihalcea wrote:
> > > > > From: Laurentiu Mihalcea <[email protected]>
> > > > >
> > > > > The names of the carveout regions are derived using the names of the
> > > > > reserved memory devicetree nodes, which are referenced using the
> > > > > "memory-region" property. This adds a restriction on the names of said
> > > > > devicetree nodes, often bearing specific names such as: "vdevbuffer",
> > > > > "vdev0vring0", "rsc-table", etc... This goes against the devicetree
> > > > > specification's recommendation, which states that the devicetree node
> > > > > names should be generic.
> > > >
> > > > I don't see what is so restrictive in using the node name of the
> > > > reserved-memory
> > > > regions. Function of_reserved_mem_region_to_resource() is already
> > > > doing all the
> > > > parsing, packaging everything in a neat and easy to use "struct
> > > > resource". What
> > > > will you gain with this new "memory-region-names" that can't be done
> > > > with the
> > > > current solution?
> > >
> > > DT Binding check can't find such wrong if node name is not what expected.
> > > Binding can't restrict memory's node name because there ware not specific
> > > compatible string for it.
> > >
> >
> > But what "wrong" could that be, and what kind of restriction are you
> > hoping to enforce? What specific problem are you hoping to solve?
> >
> > I'll wait to see what the DT people think about this - I personally
> > don't see the value in it.
>
> I see no point in this commit, but maybe because the commit msg is just
> misleading. It mixes node names with names for phandles which are two
> separate things.
For example:
rsc_table: rsc-table@90000000
{ ret = <0x90000000>;
no-map;
}
m4 {
...
memory-region = <&rsc_table>;
}
If you change node name "rsc-table" to "memory", driver will failure
because it parse node name "rsc-table", which phandle point to. but no
binding to restrict node name to "rsc-table". So rsc-table became hidden
ABI.
if use memory-region-names, we can restrict memory-region-name to
"rsc-table" earsily.
Frank
>
> Plus this change actually makes nothing - no names are restricted to any
> meaningful values!
>
> Best regards,
> Krzysztof
>
