[PATCH] arm64: dts: nuvoton: Add missing "device_type" property on memory node

[Rob Herring]

On Thu, Jan 8, 2026 at 1:09 AM Andrew Jeffery
<andrew at codeconstruct.com.au> wrote:
>
> On Mon, 2026-01-05 at 13:32 -0600, Rob Herring (Arm) wrote:
> > "device_type" is required for memory nodes, but is missing on Nuvoton
> > npcm845-evb.
> >
> > Signed-off-by: Rob Herring (Arm) <robh at kernel.org>
> > ---
> >  arch/arm64/boot/dts/nuvoton/nuvoton-npcm845-evb.dts | 1 +
> >  1 file changed, 1 insertion(+)
> >
> > diff --git a/arch/arm64/boot/dts/nuvoton/nuvoton-npcm845-evb.dts b/arch/arm64/boot/dts/nuvoton/nuvoton-npcm845-evb.dts
> > index 2638ee1c3846..5edf5d13342d 100644
> > --- a/arch/arm64/boot/dts/nuvoton/nuvoton-npcm845-evb.dts
> > +++ b/arch/arm64/boot/dts/nuvoton/nuvoton-npcm845-evb.dts
> > @@ -17,6 +17,7 @@ chosen {
> >       };
> >
> >       memory at 0 {
> > +             device_type = "memory";
> >               reg = <0x0 0x0 0x0 0x40000000>;
> >       };
> >
>
> Hmm, we have quite a few other offenders under arch/arm/boot/dts/aspeed
> too. I expect this should be caught by CHECK_DTBS=y?
>
> Digging back through my build archives, warnings have cropped up, but
> only for when I'm building the arm64 or arm multi_v{5,7} defconfigsm
> and seemingly not the aspeed_g5 defconfig.
>
> executions/2024Y/2024Y08m/2024Y08m13d10H09M/20240812145816.3301570-1-tmaimon77 at gmail.com.md:      /home/andrew/src/kernel.org/linux/openbmc/build.npcm/arch/arm64/boot/dts/nuvoton/nuvoton-npcm845-evb.dtb: /: memory: 'device_type' is a required property
> executions/2025Y/2025Y04m/2025Y04m03d21H04M/baseline-build:/home/andrew/src/kernel.org/linux/origin/build.multi_v5/arch/arm/boot/dts/nuvoton/nuvoton-npcm750-evb.dtb: /: memory: False schema does not allow {'device_type': ['memory'], 'reg': [[0, 536870912]]}
> executions/2025Y/2025Y04m/2025Y04m04d00H57M/baseline-build:/home/andrew/src/kernel.org/linux/origin/build.multi_v5/arch/arm/boot/dts/nuvoton/nuvoton-npcm750-evb.dtb: /: memory: False schema does not allow {'device_type': ['memory'], 'reg': [[0, 536870912]]}
> executions/2025Y/2025Y04m/2025Y04m04d01H10M/baseline-build:/home/andrew/src/kernel.org/linux/origin/build.multi_v5/arch/arm/boot/dts/nuvoton/nuvoton-npcm750-evb.dtb: /: memory: False schema does not allow {'device_type': ['memory'], 'reg': [[0, 536870912]]}
> executions/2025Y/2025Y04m/2025Y04m08d10H18M/baseline-build:/home/andrew/src/kernel.org/linux/origin/build.multi_v7/arch/arm/boot/dts/nuvoton/nuvoton-npcm750-evb.dtb: /: memory: False schema does not allow {'device_type': ['memory'], 'reg': [[0, 536870912]]}

> executions/2025Y/2025Y07m/2025Y07m21d13H58M/baseline-build:/home/andrew/src/kernel.org/linux/origin/build.arm.multi_v7/arch/arm/boot/dts/nuvoton/nuvoton-npcm750-evb.dtb: / (nuvoton,npcm750-evb): memory: False schema does not allow {'device_type': ['memory'], 'reg': [[0, 536870912]]}

There's a number of problems on arm32 /memory nodes. The main one is
the lack of unit-address, but that can't necessarily be fixed because
bootloaders may depend on finding a /memory node. That unfortunately
includes our own ATAGS to FDT zImage code. If you know these platforms
don't have those issues, then yes, please update them.

Rob