[PATCH net-next 06/10] dt-bindings: soc: mediatek: move ilm in a dedicated dts node

[Lorenzo Bianconi]

> On Mon, Mar 20, 2023 at 05:58:00PM +0100, Lorenzo Bianconi wrote:
> > Since the cpuboot memory region is not part of the RAM SoC, move ilm in
> > a deidicated syscon node.
> > This patch helps to keep backward-compatibility with older version of
> > uboot codebase where we have a limit of 8 reserved-memory dts child
> > nodes.
> 
> Maybe, but breaks the ABI. It also looks like a step backwards. Fix your 
> u-boot.

Can you please give some more details about ABI breakage? mtk_wed driver is
supposed to be backward compatible with respect to the older dts description
(please take a look to patch 5/10 and 8/10).

> 
> > 
> > Signed-off-by: Lorenzo Bianconi <lorenzo at kernel.org>
> > ---
> >  .../arm/mediatek/mediatek,mt7622-wed.yaml     | 14 +++---
> >  .../soc/mediatek/mediatek,mt7986-wo-ilm.yaml  | 45 +++++++++++++++++++
> >  2 files changed, 53 insertions(+), 6 deletions(-)
> >  create mode 100644 Documentation/devicetree/bindings/soc/mediatek/mediatek,mt7986-wo-ilm.yaml
> > 
> > diff --git a/Documentation/devicetree/bindings/arm/mediatek/mediatek,mt7622-wed.yaml b/Documentation/devicetree/bindings/arm/mediatek/mediatek,mt7622-wed.yaml
> > index 7f6638d43854..5d2397ec5891 100644
> > --- a/Documentation/devicetree/bindings/arm/mediatek/mediatek,mt7622-wed.yaml
> > +++ b/Documentation/devicetree/bindings/arm/mediatek/mediatek,mt7622-wed.yaml
> > @@ -32,14 +32,12 @@ properties:
> >    memory-region:
> >      items:
> >        - description: firmware EMI region
> > -      - description: firmware ILM region
> >        - description: firmware DLM region
> >        - description: firmware CPU DATA region
> >  
> >    memory-region-names:
> >      items:
> >        - const: wo-emi
> > -      - const: wo-ilm
> >        - const: wo-dlm
> >        - const: wo-data
> >  
> > @@ -51,6 +49,10 @@ properties:
> >      $ref: /schemas/types.yaml#/definitions/phandle
> >      description: mediatek wed-wo cpuboot controller interface.
> >  
> > +  mediatek,wo-ilm:
> > +    $ref: /schemas/types.yaml#/definitions/phandle
> > +    description: mediatek wed-wo ilm interface.
> > +
> >  allOf:
> >    - if:
> >        properties:
> > @@ -63,6 +65,7 @@ allOf:
> >          memory-region: false
> >          mediatek,wo-ccif: false
> >          mediatek,wo-cpuboot: false
> > +        mediatek,wo-ilm: false
> >  
> >  required:
> >    - compatible
> > @@ -97,11 +100,10 @@ examples:
> >          reg = <0 0x15010000 0 0x1000>;
> >          interrupts = <GIC_SPI 205 IRQ_TYPE_LEVEL_HIGH>;
> >  
> > -        memory-region = <&wo_emi>, <&wo_ilm>, <&wo_dlm>,
> > -                        <&wo_data>;
> > -        memory-region-names = "wo-emi", "wo-ilm", "wo-dlm",
> > -                              "wo-data";
> > +        memory-region = <&wo_emi>, <&wo_dlm>, &wo_data>;
> > +        memory-region-names = "wo-emi", "wo-dlm", "wo-data";
> >          mediatek,wo-ccif = <&wo_ccif0>;
> >          mediatek,wo-cpuboot = <&wo_cpuboot>;
> > +        mediatek,wo-ilm = <&wo_ilm>;
> >        };
> >      };
> > diff --git a/Documentation/devicetree/bindings/soc/mediatek/mediatek,mt7986-wo-ilm.yaml b/Documentation/devicetree/bindings/soc/mediatek/mediatek,mt7986-wo-ilm.yaml
> > new file mode 100644
> > index 000000000000..2a3775cd941e
> > --- /dev/null
> > +++ b/Documentation/devicetree/bindings/soc/mediatek/mediatek,mt7986-wo-ilm.yaml
> > @@ -0,0 +1,45 @@
> > +# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
> > +%YAML 1.2
> > +---
> > +$id: http://devicetree.org/schemas/soc/mediatek/mediatek,mt7986-wo-ilm.yaml#
> > +$schema: http://devicetree.org/meta-schemas/core.yaml#
> > +
> > +title: MediaTek Wireless Ethernet Dispatch (WED) WO ILM firmware interface for MT7986
> 
> Either this region is some memory or it's a device. Sounds like the 
> former and this is not a 'syscon'.

ILM and DLM are mmio regions used to store Wireless Ethernet Dispatch firmware
sections.
AFAIU reserved-memory nodes are supposed to be used just for RAM areas, correct?
What do you think would be the best dts node type for ILM and DLM?

Another possibility I was discussing with Felix is to move ILM and DLM definitions
in WED node reg section, something like:

wed0: wed at 15010000 {
	compatible = "mediatek,mt7986-wed";
	reg = <0 0x15010000 0 0x1000>,
	      <0 0x151e0000 0 0x8000>,
	      <0 0x151e8000 0 0x2000>;
	...
};

Doing so we need to get rid of syscon in WED compatible string, right? Is it acceptable?

Regards,
Lorenzo

> 
> > +
> > +maintainers:
> > +  - Lorenzo Bianconi <lorenzo at kernel.org>
> > +  - Felix Fietkau <nbd at nbd.name>
> > +
> > +description:
> > +  The MediaTek wo-ilm (Information Lifecycle Management) provides a configuration
> > +  interface for WiFi critical data used by WED WO firmware. WED WO controller is
> > +  used to perform offload rx packet processing (e.g. 802.11 aggregation packet
> > +  reordering or rx header translation) on MT7986 soc.
> > +
> > +properties:
> > +  compatible:
> > +    items:
> > +      - enum:
> > +          - mediatek,mt7986-wo-ilm
> > +      - const: syscon
> > +
> > +  reg:
> > +    maxItems: 1
> > +
> > +required:
> > +  - compatible
> > +  - reg
> > +
> > +additionalProperties: false
> > +
> > +examples:
> > +  - |
> > +    soc {
> > +      #address-cells = <2>;
> > +      #size-cells = <2>;
> > +
> > +      syscon at 151e0000 {
> > +        compatible = "mediatek,mt7986-wo-ilm", "syscon";
> > +        reg = <0 0x151e0000 0 0x8000>;
> > +      };
> > +    };
> > -- 
> > 2.39.2
> > 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 228 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-mediatek/attachments/20230322/d0cc61b5/attachment-0001.sig>