Re: [PATCH] dt-bindings: arm: omap-l3-noc: Convert to DT schema
From: Bhargav Joshi
Date: Fri Nov 06 2026 - 08:24:14 EST
Hi,
On Tue, Oct 27, 2026 at 3:52 AM Rob Herring <robh@xxxxxxxxxx> wrote:
>
> On Sat, Oct 03, 2026 at 12:28:32AM +0530, Bhargav Joshi wrote:
> > Convert TI L3 NoC binding from text to DT schema format. Older OMAP3
> > uses the SMX variant which also acts as a bus parent with simple-bus
> > compatible so child nodes are allowed using patternproperties. Mark
> > ti,hwmods deprecated as it is no longer needed and only used by older
> > omap3-l3-smx.
> >
> > Signed-off-by: Bhargav Joshi <j.bhargav.u@xxxxxxxxx>
> > ---
> > .../devicetree/bindings/arm/omap/l3-noc.txt | 23 ----
> > .../bindings/arm/omap/ti,omap-l3-noc.yaml | 149 +++++++++++++++++++++
>
> This should move to bindings/soc/ti/ or bindings/bus/ perhaps.
>
> > 2 files changed, 149 insertions(+), 23 deletions(-)
> >
> > diff --git a/Documentation/devicetree/bindings/arm/omap/l3-noc.txt b/Documentation/devicetree/bindings/arm/omap/l3-noc.txt
> > deleted file mode 100644
> > index 161448da959d2..0000000000000
> > --- a/Documentation/devicetree/bindings/arm/omap/l3-noc.txt
> > +++ /dev/null
> > @@ -1,23 +0,0 @@
> > -* TI - L3 Network On Chip (NoC)
> > -
> > -This version is an implementation of the generic NoC IP
> > -provided by Arteris.
> > -
> > -Required properties:
> > -- compatible : Should be "ti,omap3-l3-smx" for OMAP3 family
> > - Should be "ti,omap4-l3-noc" for OMAP4 family
> > - Should be "ti,omap5-l3-noc" for OMAP5 family
> > - Should be "ti,dra7-l3-noc" for DRA7 family
> > - Should be "ti,am4372-l3-noc" for AM43 family
> > -- reg: Contains L3 register address range for each noc domain.
> > -- ti,hwmods: "l3_main_1", ... One hwmod for each noc domain.
> > -
> > -Examples:
> > -
> > -ocp {
> > - compatible = "ti,omap4-l3-noc", "simple-bus";
> > - #address-cells = <1>;
> > - #size-cells = <1>;
> > - ranges;
> > - ti,hwmods = "l3_main_1", "l3_main_2", "l3_main_3";
> > -};
> > diff --git a/Documentation/devicetree/bindings/arm/omap/ti,omap-l3-noc.yaml b/Documentation/devicetree/bindings/arm/omap/ti,omap-l3-noc.yaml
> > new file mode 100644
> > index 0000000000000..a72a0f4e64332
> > --- /dev/null
> > +++ b/Documentation/devicetree/bindings/arm/omap/ti,omap-l3-noc.yaml
> > @@ -0,0 +1,149 @@
> > +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
> > +%YAML 1.2
> > +---
> > +$id: http://devicetree.org/schemas/arm/omap/ti,omap-l3-noc.yaml#
> > +$schema: http://devicetree.org/meta-schemas/core.yaml#
> > +
> > +title: TI OMAP L3 Network On Chip (NoC) Interconnect
> > +
> > +maintainers:
> > + - Santosh Shilimkar <santosh.shilimkar@xxxxxx>
> > + - Sricharan R <r.sricharan@xxxxxx>
> > +
> > +description:
> > + The L3 interconnect is an implementation of the generic NoC IP provided by
> > + Arteris. OMAP3 uses the SMX variant while OMAP4/5, DRA7, and AM4372 use the
> > + NoC variant.
> > +
> > +properties:
> > + compatible:
> > + oneOf:
> > + - items:
> > + - const: ti,omap3-l3-smx
> > + - const: simple-bus
> > + - enum:
> > + - ti,omap4-l3-noc
> > + - ti,omap5-l3-noc
> > + - ti,dra7-l3-noc
> > + - ti,am4372-l3-noc
> > +
> > + reg:
> > + minItems: 1
> > + maxItems: 3
>
> Need to define what each region is.
>
> > +
> > + interrupts:
> > + minItems: 2
> > + maxItems: 2
>
> Need to define what each interrupt is.
>
sure.
> > +
> > + ti,hwmods:
> > + description: hwmod names for each NoC domain.
> > + $ref: /schemas/types.yaml#/definitions/string-array
> > + deprecated: true
> > +
> > + "#address-cells":
> > + const: 1
> > +
> > + "#size-cells":
> > + const: 1
> > +
> > + ranges: true
> > +
> > +patternProperties:
> > + "^(.*@[0-9a-fA-F]+|regulator-abb-mpu)$":
>
> I suspect that regulator-abb-mpu should be moved out of this node.
Sorry for late reply.
On OMAP3 the ABB registers are both inside the PRM, so prm would be accurate
parent. However, on OMAP3 the L3 SMX node is deliberately the flat bus for the
whole SoC. The OMAP3 prm node is also not a bus so the ABB node can't be nested
there.
So for v2 I'll rename the ABB node to standard regulator@483072f0 instead.
>
> A-F is generally not used for unit-addresses. Do we need it?
will remove it in V2.
>
> Rob