Skip to contentSkip to conversation

arm: bcmbca: Update BCM4908 dts for BCMBCA merge

RFC PATCH In review William Zhang 5 participants 26 messages Jul 12, 2022 – Jul 18, 2022

Subsystems (2)
lore
Share & tools
lore

William Zhang

author maintainer
#0 lore raw

cover arm: bcmbca: Update BCM4908 dts for BCMBCA merge

Prepare the device tree related changes for the ARCH_BCM4908 merging into ARCH_BCMBCA

In order to reduce the number of compatible strings and keep the bcmbca yaml file concise, I propose to use enum to list all the board and chip variants under the same compatible string for the same chip family.

William Zhang (3):
dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA
dt-bindings: arm64: bcm4908: remove binding document
arm64: dts: bcmbca: update bcm4808 board dts file

 .../bindings/arm/bcm/brcm,bcm4908.yaml        | 42 -------------------
 .../bindings/arm/bcm/brcm,bcmbca.yaml         | 17 ++++++++
 .../bcm4908/bcm4906-netgear-r8000p.dts        |  2 +-
 .../bcm4906-tplink-archer-c2300-v1.dts        |  2 +-
 .../bcm4908/bcm4908-asus-gt-ac5300.dts        |  2 +-
 .../bcm4908/bcm4908-netgear-raxe500.dts       |  2 +-
 6 files changed, 21 insertions(+), 46 deletions(-)
 delete mode 100644 Documentation/devicetree/bindings/arm/bcm/brcm,bcm4908.yaml
Signature · 1 line
2.34.1
author maintainer
#1 lore raw

RFC PATCH 1/3 dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

Merge BCM4908 SoC device tree description into BCMBCA and combined all BCM4908 chip variants into the same BCM4908 chip family item.

Each compatible string represent the whole chip family. The board variants and chip varints go into the first and second enum in the compatible string item list.

---

 .../bindings/arm/bcm/brcm,bcmbca.yaml           | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)
Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml +17 -0
index d9dc4f22f4a5..906c3e1de372 100644
@@ -28,6 +28,23 @@ properties:
           - const: brcm,bcm47622
           - const: brcm,bcmbca
 
+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500
+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408
+          - const: brcm,bcmbca
+
       - description: BCM4912 based boards
         items:
           - enum:
Signature · 1 line
2.34.1
#2 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

Quoted 2 lines · William Zhang

Merge BCM4908 SoC device tree description into BCMBCA and combined all BCM4908 chip variants into the same BCM4908 chip family item.

Merge means you combine some entries, so I would expect to see the removal here as well.

Quoted 24 lines

Each compatible string represent the whole chip family. The board variants and chip varints go into the first and second enum in the compatible string item list.

---

 .../bindings/arm/bcm/brcm,bcmbca.yaml           | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)
Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml +5 -0
index d9dc4f22f4a5..906c3e1de372 100644
@@ -28,6 +28,23 @@ properties:
           - const: brcm,bcm47622
           - const: brcm,bcmbca
 
+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908

Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml properties: · line 45

Quoted 10 lines
Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml +10 -0
+              - asus,gt-ac5300
+              - netgear,raxe500
+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408

This is wrong. brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml properties: · line 50

Quoted 5 lines
Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml +2 -0
+          - const: brcm,bcmbca
+
       - description: BCM4912 based boards
         items:
           - enum:

Best regards,
Krzysztof

author maintainer
#3 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

Quoted 7 lines · Krzysztof Kozlowski
Quoted 2 lines · William Zhang

Merge BCM4908 SoC device tree description into BCMBCA and combined all BCM4908 chip variants into the same BCM4908 chip family item.

Merge means you combine some entries, so I would expect to see the removal here as well.

Will combine with the removal patch as you pointed out

Quoted 29 lines
Quoted 29 lines

Each compatible string represent the whole chip family. The board variants and chip varints go into the first and second enum in the compatible string item list.

---

  .../bindings/arm/bcm/brcm,bcmbca.yaml           | 17 +++++++++++++++++
  1 file changed, 17 insertions(+)
Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml +10 -0
index d9dc4f22f4a5..906c3e1de372 100644
@@ -28,6 +28,23 @@ properties:
            - const: brcm,bcm47622
            - const: brcm,bcmbca
 
+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500
+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
Quoted 9 lines

Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml properties: · line 45

Quoted 5 lines
Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml +5 -0
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408

This is wrong. brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

For 4908 board variant, it will need to be followed by 4908 chip. Sorry for the basic question but is there any requirement to enforce this kind of rule? I would assume dts writer know what he/she is doing and select the right combination.

Quoted 9 lines

Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml properties: · line 50

Quoted 5 lines
Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml +2 -0
+          - const: brcm,bcmbca
+
        - description: BCM4912 based boards
          items:
            - enum:

Best regards,
Krzysztof

#4 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

Quoted 16 lines · William Zhang
Quoted 16 lines
Quoted 15 lines
Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml +15 -0
+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500
+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408
Quoted 7 lines
Quoted 3 lines

This is wrong. brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

For 4908 board variant, it will need to be followed by 4908 chip. Sorry for the basic question but is there any requirement to enforce this kind of rule? I would assume dts writer know what he/she is doing and select the right combination.

The entire point of DT schema is to validate DTS. Combination like above prevents that goal.

Best regards,
Krzysztof

author maintainer
#5 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

Quoted 25 lines · Krzysztof Kozlowski
Quoted 16 lines · William Zhang
Quoted 16 lines
Quoted 15 lines
Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml +15 -0
+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500
+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408
Quoted 7 lines
Quoted 3 lines

This is wrong. brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

For 4908 board variant, it will need to be followed by 4908 chip. Sorry for the basic question but is there any requirement to enforce this kind of rule? I would assume dts writer know what he/she is doing and select the right combination.

Quoted 5 lines

The entire point of DT schema is to validate DTS. Combination like above prevents that goal.

Best regards,
Krzysztof

Understand the DT schema purpose. But items property allows multiple enums in the list which gives a lot of flexibility but make it hard to validate. I am not familiar with DT schema, is there any directive to specify one enum value depending on another so dts validation tool can report error if combination is wrong?

This is our preferred format of all bcmbca compatible string especially when we could have more than 10 chip variants for the same chip family and we really want to work on the chip family id. We will make sure they are in the right combination in our own patch and patch from other contributors. Would this work? If not, I will probably have to revert the change of 4908(maybe append brcm,bcmbca as this chip belongs to the same bca group) and use "enum board variant", "const main chip id", "brcm,bca" for all other chips as our secondary choice.

maintainer
#6 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

Quoted 37 lines · William Zhang
Quoted 25 lines · Krzysztof Kozlowski
Quoted 16 lines · William Zhang
Quoted 16 lines
Quoted 15 lines
Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml +15 -0
+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500
+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408
Quoted 7 lines
Quoted 3 lines

This is wrong. brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

For 4908 board variant, it will need to be followed by 4908 chip. Sorry for the basic question but is there any requirement to enforce this kind of rule? I would assume dts writer know what he/she is doing and select the right combination.

Quoted 5 lines

The entire point of DT schema is to validate DTS. Combination like above prevents that goal.

Best regards,
Krzysztof

Understand the DT schema purpose. But items property allows multiple enums in the list which gives a lot of flexibility but make it hard to validate. I am not familiar with DT schema, is there any directive to specify one enum value depending on another so dts validation tool can report error if combination is wrong?

Quoted 8 lines

This is our preferred format of all bcmbca compatible string especially when we could have more than 10 chip variants for the same chip family and we really want to work on the chip family id. We will make sure they are in the right combination in our own patch and patch from other contributors. Would this work? If not, I will probably have to revert the change of 4908(maybe append brcm,bcmbca as this chip belongs to the same bca group) and use "enum board variant", "const main chip id", "brcm,bca" for all other chips as our secondary choice.

I'm not sure why I didn't even receive 1/3 and half of discussion e-mails.

You can't just put all strings into a single bag and allow mixing them in any combos. Please check how it's properly handled in the current existing binding:
Documentation/devicetree/bindings/arm/bcm/brcm,bcm4908.yaml

Above binding enforces that non-matching compatible strings are not used together.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

maintainer
#7 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

Quoted 50 lines · Rafał Miłecki
Quoted 37 lines · William Zhang
Quoted 25 lines · Krzysztof Kozlowski
Quoted 16 lines · William Zhang
Quoted 16 lines
Quoted 15 lines
Documentation/devicetree/bindings/arm/bcm/brcm,bcmbca.yaml +15 -0
+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500
+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408
Quoted 7 lines
Quoted 3 lines

This is wrong. brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

For 4908 board variant, it will need to be followed by 4908 chip. Sorry for the basic question but is there any requirement to enforce this kind of rule? I would assume dts writer know what he/she is doing and select the right combination.

Quoted 5 lines

The entire point of DT schema is to validate DTS. Combination like above prevents that goal.

Best regards,
Krzysztof

Understand the DT schema purpose. But items property allows multiple enums in the list which gives a lot of flexibility but make it hard to validate. I am not familiar with DT schema, is there any directive to specify one enum value depending on another so dts validation tool can report error if combination is wrong?

Quoted 8 lines

This is our preferred format of all bcmbca compatible string especially when we could have more than 10 chip variants for the same chip family and we really want to work on the chip family id. We will make sure they are in the right combination in our own patch and patch from other contributors. Would this work? If not, I will probably have to revert the change of 4908(maybe append brcm,bcmbca as this chip belongs to the same bca group) and use "enum board variant", "const main chip id", "brcm,bca" for all other chips as our secondary choice.

I'm not sure why I didn't even receive 1/3 and half of discussion e-mails.

Quoted 7 lines

You can't just put all strings into a single bag and allow mixing them in any combos. Please check how it's properly handled in the current existing binding:
Documentation/devicetree/bindings/arm/bcm/brcm,bcm4908.yaml

Above binding enforces that non-matching compatible strings are not used together.

I just noticed you're actually removing brcm,bcm4908.yaml in the 2/3 so you must be aware of that file.

So you see a cleanly working binding in the brcm,bcm4908.yaml but instead copying it you decided to wrote your own one from scratch. Incorrectly.

This smells of NIH (not invented here). Please just use that binding I wrote and move if it needed.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

author maintainer
#8 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

Hi Rafal,

Quoted 62 lines · Rafał Miłecki
Quoted 50 lines · Rafał Miłecki
Quoted 37 lines · William Zhang
Quoted 25 lines · Krzysztof Kozlowski
Quoted 16 lines · William Zhang
Quoted 16 lines
Quoted 15 lines

+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500
+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408

Quoted 7 lines
Quoted 3 lines

This is wrong.  brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

For 4908 board variant, it will need to be followed by 4908 chip. Sorry for the basic question but is there any requirement to enforce this kind of rule?  I would assume dts writer know what he/she is doing and select the right combination.

Quoted 5 lines

The entire point of DT schema is to validate DTS. Combination like above prevents that goal.

Best regards,
Krzysztof

Understand the DT schema purpose. But items property allows multiple enums in the list which gives a lot of flexibility but make it hard to validate. I am not familiar with DT schema, is there any directive to specify one enum value depending on another so dts validation tool can report error if combination is wrong?

Quoted 8 lines

This is our preferred format of all bcmbca compatible string especially when we could have more than 10 chip variants for the same chip family and we really want to work on the chip family id.  We will make sure they are in the right combination in our own patch and patch from other contributors. Would this work? If not, I will probably have to revert the change of 4908(maybe append brcm,bcmbca as this chip belongs to the same bca group) and use "enum board variant", "const main chip id", "brcm,bca" for all other chips as our secondary choice.

I'm not sure why I didn't even receive 1/3 and half of discussion e-mails.

Quoted 7 lines

You can't just put all strings into a single bag and allow mixing them in any combos. Please check how it's properly handled in the current existing binding:
Documentation/devicetree/bindings/arm/bcm/brcm,bcm4908.yaml

Above binding enforces that non-matching compatible strings are not used together.

I just noticed you're actually removing brcm,bcm4908.yaml in the 2/3 so you must be aware of that file.

Quoted 6 lines

So you see a cleanly working binding in the brcm,bcm4908.yaml but instead copying it you decided to wrote your own one from scratch. Incorrectly.

This smells of NIH (not invented here). Please just use that binding I wrote and move if it needed.

Not mean to discredit any of your work and I did copy over your binding and combine them into one SoC entry to the new bcmbca.yaml and add you as one of the maintainer to this file. As this change would certainly concern you, that's why I sent RFC first. As I explained in the cover letter, the purpose of the change is to reduce the number of compatible strings and keep one entry for one chip family due to possible large number of chip variants. But since there is no way to validate the combination, I will copy the existing 4908 bindings as they are now but I would propose to append "brcm, bcmbca" as it is part of bcmbca chip. And for the other chips, we would just use enum "board variant", const "main chip id", const "brcm,bca". Does that sound good to you?

maintainer
#9 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

Quoted 72 lines · William Zhang

Hi Rafal,

Quoted 62 lines · Rafał Miłecki
Quoted 50 lines · Rafał Miłecki
Quoted 37 lines · William Zhang
Quoted 25 lines · Krzysztof Kozlowski
Quoted 16 lines · William Zhang
Quoted 16 lines
Quoted 15 lines

+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500
+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408

Quoted 7 lines
Quoted 3 lines

This is wrong.  brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

For 4908 board variant, it will need to be followed by 4908 chip. Sorry for the basic question but is there any requirement to enforce this kind of rule?  I would assume dts writer know what he/she is doing and select the right combination.

Quoted 5 lines

The entire point of DT schema is to validate DTS. Combination like above prevents that goal.

Best regards,
Krzysztof

Understand the DT schema purpose. But items property allows multiple enums in the list which gives a lot of flexibility but make it hard to validate. I am not familiar with DT schema, is there any directive to specify one enum value depending on another so dts validation tool can report error if combination is wrong?

Quoted 8 lines

This is our preferred format of all bcmbca compatible string especially when we could have more than 10 chip variants for the same chip family and we really want to work on the chip family id.  We will make sure they are in the right combination in our own patch and patch from other contributors. Would this work? If not, I will probably have to revert the change of 4908(maybe append brcm,bcmbca as this chip belongs to the same bca group) and use "enum board variant", "const main chip id", "brcm,bca" for all other chips as our secondary choice.

I'm not sure why I didn't even receive 1/3 and half of discussion e-mails.

Quoted 7 lines

You can't just put all strings into a single bag and allow mixing them in any combos. Please check how it's properly handled in the current existing binding:
Documentation/devicetree/bindings/arm/bcm/brcm,bcm4908.yaml

Above binding enforces that non-matching compatible strings are not used together.

I just noticed you're actually removing brcm,bcm4908.yaml in the 2/3 so you must be aware of that file.

Quoted 6 lines

So you see a cleanly working binding in the brcm,bcm4908.yaml but instead copying it you decided to wrote your own one from scratch. Incorrectly.

This smells of NIH (not invented here). Please just use that binding I wrote and move if it needed.

Quoted 9 lines

Not mean to discredit any of your work and I did copy over your binding and combine them into one SoC entry to the new bcmbca.yaml and add you as one of the maintainer to this file. As this change would certainly concern you, that's why I sent RFC first. As I explained in the cover letter, the purpose of the change is to reduce the number of compatible strings and keep one entry for one chip family due to possible large number of chip variants. But since there is no way to validate the combination, I will copy the existing 4908 bindings as they are now

Right. I believe we need that.

Quoted 4 lines

but I would propose to append "brcm, bcmbca" as it is part of bcmbca chip. And for the other chips, we would just use enum "board variant", const "main chip id", const "brcm,bca". Does that sound good to you?

Nitpicking: you meant "brcm,bcmbca" (typo) but sounds absolutely fine!

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

author maintainer
#10 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

Quoted 86 lines · Rafał Miłecki
Quoted 72 lines · William Zhang

Hi Rafal,

Quoted 62 lines · Rafał Miłecki
Quoted 50 lines · Rafał Miłecki
Quoted 37 lines · William Zhang
Quoted 25 lines · Krzysztof Kozlowski
Quoted 16 lines · William Zhang
Quoted 16 lines
Quoted 15 lines

+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500
+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408

Quoted 7 lines
Quoted 3 lines

This is wrong.  brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

For 4908 board variant, it will need to be followed by 4908 chip. Sorry for the basic question but is there any requirement to enforce this kind of rule?  I would assume dts writer know what he/she is doing and select the right combination.

Quoted 5 lines

The entire point of DT schema is to validate DTS. Combination like above prevents that goal.

Best regards,
Krzysztof

Understand the DT schema purpose. But items property allows multiple enums in the list which gives a lot of flexibility but make it hard to validate. I am not familiar with DT schema, is there any directive to specify one enum value depending on another so dts validation tool can report error if combination is wrong?

Quoted 8 lines

This is our preferred format of all bcmbca compatible string especially when we could have more than 10 chip variants for the same chip family and we really want to work on the chip family id.  We will make sure they are in the right combination in our own patch and patch from other contributors. Would this work? If not, I will probably have to revert the change of 4908(maybe append brcm,bcmbca as this chip belongs to the same bca group) and use "enum board variant", "const main chip id", "brcm,bca" for all other chips as our secondary choice.

I'm not sure why I didn't even receive 1/3 and half of discussion e-mails.

Quoted 7 lines

You can't just put all strings into a single bag and allow mixing them in any combos. Please check how it's properly handled in the current existing binding:
Documentation/devicetree/bindings/arm/bcm/brcm,bcm4908.yaml

Above binding enforces that non-matching compatible strings are not used together.

I just noticed you're actually removing brcm,bcm4908.yaml in the 2/3 so you must be aware of that file.

Quoted 6 lines

So you see a cleanly working binding in the brcm,bcm4908.yaml but instead copying it you decided to wrote your own one from scratch. Incorrectly.

This smells of NIH (not invented here). Please just use that binding I wrote and move if it needed.

Quoted 9 lines

Not mean to discredit any of your work and I did copy over your binding and combine them into one SoC entry to the new bcmbca.yaml and add you as one of the maintainer to this file. As this change would certainly concern you, that's why I sent RFC first.  As I explained in the cover letter, the purpose of the change is to reduce the number of compatible strings and keep one entry for one chip family due to possible large number of chip variants.  But since there is no way to validate the combination, I will copy the existing 4908 bindings as they are now

Right. I believe we need that.

Quoted 6 lines
Quoted 4 lines

but I would propose to append "brcm, bcmbca" as it is part of bcmbca chip. And for the other chips, we would just use enum "board variant", const "main chip id", const "brcm,bca".  Does that sound good to you?

Nitpicking: you meant "brcm,bcmbca" (typo) but sounds absolutely fine!

Yup its a typo. Will append "brcm,bcmbca" and send out new patch.

maintainer
#11 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

In reply to William Zhang
Quoted 76 lines · William Zhang

Hi Rafal,

Quoted 66 lines · Rafał Miłecki
Quoted 54 lines · Rafał Miłecki
Quoted 41 lines · William Zhang
Quoted 28 lines · Krzysztof Kozlowski
Quoted 19 lines · William Zhang
Quoted 16 lines
Quoted 7 lines

+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500

Quoted 8 lines

+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408

Quoted 3 lines

This is wrong.  brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

Quoted 7 lines

For 4908 board variant, it will need to be followed by 4908 chip. Sorry
for the basic question but is there any requirement to enforce this kind
of rule?  I would assume dts writer know what he/she is doing and select
the right combination.

Quoted 6 lines

The entire point of DT schema is to validate DTS.
Combination like above
prevents that goal.

Best regards,
Krzysztof

Understand the DT schema purpose. But items property allows multiple enums in the list which gives a lot of flexibility but make it hard to validate. I am not familiar with DT schema, is there any directive to specify one enum value depending on another so dts validation tool can report error if combination is wrong?

Quoted 8 lines

This is our preferred format of all bcmbca compatible string especially when we could have more than 10 chip variants for the same chip family and we really want to work on the chip family id.  We will make sure they are in the right combination in our own patch and patch from other contributors. Would this work? If not, I will probably have to revert the change of 4908(maybe append brcm,bcmbca as this chip belongs to the same bca group) and use "enum board variant", "const main chip id", "brcm,bca" for all other chips as our secondary choice.

I'm not sure why I didn't even receive 1/3 and half of discussion e-mails.

Quoted 7 lines

You can't just put all strings into a single bag and allow mixing them in any combos. Please check how it's properly handled in the current existing binding:
Documentation/devicetree/bindings/arm/bcm/brcm,bcm4908.yaml

Above binding enforces that non-matching compatible strings are not used together.

I just noticed you're actually removing brcm,bcm4908.yaml in the 2/3 so you must be aware of that file.

Quoted 6 lines

So you see a cleanly working binding in the brcm,bcm4908.yaml but instead copying it you decided to wrote your own one from scratch. Incorrectly.

This smells of NIH (not invented here). Please just use that binding I wrote and move if it needed.

Quoted 11 lines

Not mean to discredit any of your work and I did copy over your binding and combine them into one SoC entry to the new bcmbca.yaml and add you as one of the maintainer to this file. As this change would certainly concern you, that's why I sent RFC first. As I explained in the cover letter, the purpose of the change is to reduce the number of compatible strings and keep one entry for one chip family due to possible large number of chip variants. But since there is no way to validate the combination, I will copy the existing 4908 bindings as they are now but I would propose to append "brcm, bcmbca" as it is part of bcmbca chip. And for the other chips, we would just use enum "board variant", const "main chip id", const "brcm,bca". Does that sound good to you?

If you want fewer combinations of compatibles, adding a genericish "brcm,bcmbca" is not going to help. Is there much value to adding it? What can you do with that information (and nothing else) is the question to ask.

Rob

author maintainer
#12 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

Quoted 89 lines · Rob Herring
Quoted 76 lines · William Zhang

Hi Rafal,

Quoted 66 lines · Rafał Miłecki
Quoted 54 lines · Rafał Miłecki
Quoted 41 lines · William Zhang
Quoted 28 lines · Krzysztof Kozlowski
Quoted 19 lines · William Zhang
Quoted 16 lines
Quoted 7 lines

+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500

Quoted 8 lines

+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408

Quoted 3 lines

This is wrong.  brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

Quoted 7 lines

For 4908 board variant, it will need to be followed by 4908 chip. Sorry
for the basic question but is there any requirement to enforce this kind
of rule?  I would assume dts writer know what he/she is doing and select
the right combination.

Quoted 6 lines

The entire point of DT schema is to validate DTS.
Combination like above
prevents that goal.

Best regards,
Krzysztof

Understand the DT schema purpose. But items property allows multiple enums in the list which gives a lot of flexibility but make it hard to validate. I am not familiar with DT schema, is there any directive to specify one enum value depending on another so dts validation tool can report error if combination is wrong?

Quoted 8 lines

This is our preferred format of all bcmbca compatible string especially when we could have more than 10 chip variants for the same chip family and we really want to work on the chip family id.  We will make sure they are in the right combination in our own patch and patch from other contributors. Would this work? If not, I will probably have to revert the change of 4908(maybe append brcm,bcmbca as this chip belongs to the same bca group) and use "enum board variant", "const main chip id", "brcm,bca" for all other chips as our secondary choice.

I'm not sure why I didn't even receive 1/3 and half of discussion e-mails.

Quoted 7 lines

You can't just put all strings into a single bag and allow mixing them in any combos. Please check how it's properly handled in the current existing binding:
Documentation/devicetree/bindings/arm/bcm/brcm,bcm4908.yaml

Above binding enforces that non-matching compatible strings are not used together.

I just noticed you're actually removing brcm,bcm4908.yaml in the 2/3 so you must be aware of that file.

Quoted 6 lines

So you see a cleanly working binding in the brcm,bcm4908.yaml but instead copying it you decided to wrote your own one from scratch. Incorrectly.

This smells of NIH (not invented here). Please just use that binding I wrote and move if it needed.

Quoted 11 lines

Not mean to discredit any of your work and I did copy over your binding and combine them into one SoC entry to the new bcmbca.yaml and add you as one of the maintainer to this file. As this change would certainly concern you, that's why I sent RFC first. As I explained in the cover letter, the purpose of the change is to reduce the number of compatible strings and keep one entry for one chip family due to possible large number of chip variants. But since there is no way to validate the combination, I will copy the existing 4908 bindings as they are now but I would propose to append "brcm, bcmbca" as it is part of bcmbca chip. And for the other chips, we would just use enum "board variant", const "main chip id", const "brcm,bca". Does that sound good to you?

Quoted 5 lines

If you want fewer combinations of compatibles, adding a genericish "brcm,bcmbca" is not going to help. Is there much value to adding it? What can you do with that information (and nothing else) is the question to ask.

Since we moved all the Broadcom Broadband BCA origin SoC to ARCH_BCMBCA arch, I think it makes sense to have that as the last part of the compatible string, not really for reducing the number. And for that matter, we have agreed to not change anything to Rafal dts definitions including the number of compat strings but just appending brcm,bca. So we are all good now.

Quoted 2 lines

Rob

#13 lore raw

Re: [RFC PATCH 1/3] dt-bindings: arm64: bcmbca: Merge BCM4908 into BCMBCA

In reply to Rafał Miłecki
Quoted 47 lines · Rafał Miłecki
Quoted 37 lines · William Zhang
Quoted 25 lines · Krzysztof Kozlowski
Quoted 16 lines · William Zhang
Quoted 16 lines
Quoted 15 lines

+      - description: BCM4908 Family based boards
+        items:
+          - enum:
+              # BCM4908 SoC based boards
+              - brcm,bcm94908
+              - asus,gt-ac5300
+              - netgear,raxe500
+              # BCM4906 SoC based boards
+              - brcm,bcm94906
+              - netgear,r8000p
+              - tplink,archer-c2300-v1
+          - enum:
+              - brcm,bcm4908
+              - brcm,bcm4906
+              - brcm,bcm49408

Quoted 7 lines
Quoted 3 lines

This is wrong.  brcm,bcm94908 followed by brcm,bcm4906 does not look like valid list of compatibles.

For 4908 board variant, it will need to be followed by 4908 chip. Sorry for the basic question but is there any requirement to enforce this kind of rule?  I would assume dts writer know what he/she is doing and select the right combination.

Quoted 5 lines

The entire point of DT schema is to validate DTS. Combination like above prevents that goal.

Best regards,
Krzysztof

Understand the DT schema purpose. But items property allows multiple enums in the list which gives a lot of flexibility but make it hard to validate. I am not familiar with DT schema, is there any directive to specify one enum value depending on another so dts validation tool can report error if combination is wrong?

Quoted 8 lines

This is our preferred format of all bcmbca compatible string especially when we could have more than 10 chip variants for the same chip family and we really want to work on the chip family id.  We will make sure they are in the right combination in our own patch and patch from other contributors. Would this work? If not, I will probably have to revert the change of 4908(maybe append brcm,bcmbca as this chip belongs to the same bca group) and use "enum board variant", "const main chip id", "brcm,bca" for all other chips as our secondary choice.

Quoted 2 lines

I'm not sure why I didn't even receive 1/3 and half of discussion e-mails.

You are copied on all 4 emails (including cover letter).

Signature · 1 line
Florian
author maintainer
#14 lore raw

RFC PATCH 2/3 dt-bindings: arm64: bcm4908: remove binding document

In reply to William Zhang

bcm4908 binding document are now merged into bcmbca.

---

 .../bindings/arm/bcm/brcm,bcm4908.yaml        | 42 -------------------
 1 file changed, 42 deletions(-)
 delete mode 100644 Documentation/devicetree/bindings/arm/bcm/brcm,bcm4908.yaml
Documentation/devicetree/bindings/arm/bcm/brcm,bcm4908.yaml deleted +0 -42
deleted file mode 100644
index 9b745531ff04..000000000000
@@ -1,42 +0,0 @@
-# SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause
-%YAML 1.2
----
-$id: http://devicetree.org/schemas/arm/bcm/brcm,bcm4908.yaml#
-$schema: http://devicetree.org/meta-schemas/core.yaml#
-
-title: Broadcom BCM4908 device tree bindings
-
-description:
-  Broadcom BCM4906 / BCM4908 / BCM49408 Wi-Fi/network SoCs with Brahma CPUs.
-
-maintainers:
-  - Rafał Miłecki <[email protected]>
-
-properties:
-  $nodename:
-    const: '/'
-  compatible:
-    oneOf:
-      - description: BCM4906 based boards
-        items:
-          - enum:
-              - netgear,r8000p
-              - tplink,archer-c2300-v1
-          - const: brcm,bcm4906
-          - const: brcm,bcm4908
-
-      - description: BCM4908 based boards
-        items:
-          - enum:
-              - asus,gt-ac5300
-              - netgear,raxe500
-          - const: brcm,bcm4908
-
-      - description: BCM49408 based boards
-        items:
-          - const: brcm,bcm49408
-          - const: brcm,bcm4908
-
-additionalProperties: true
-
-...
Signature · 1 line
2.34.1
#15 lore raw

Re: [RFC PATCH 2/3] dt-bindings: arm64: bcm4908: remove binding document

Quoted 3 lines · William Zhang

bcm4908 binding document are now merged into bcmbca.

This must be squashed with previous one.

Best regards,
Krzysztof

author maintainer
#16 lore raw

RFC PATCH 3/3 arm64: dts: bcmbca: update bcm4808 board dts file

In reply to William Zhang

Update compatible string based on the new bcmbca binding rule for BCM4908 famliy based boards

---

 arch/arm64/boot/dts/broadcom/bcm4908/bcm4906-netgear-r8000p.dts | 2 +-
 .../dts/broadcom/bcm4908/bcm4906-tplink-archer-c2300-v1.dts     | 2 +-
 arch/arm64/boot/dts/broadcom/bcm4908/bcm4908-asus-gt-ac5300.dts | 2 +-
 .../arm64/boot/dts/broadcom/bcm4908/bcm4908-netgear-raxe500.dts | 2 +-
 4 files changed, 4 insertions(+), 4 deletions(-)
arch/arm64/boot/dts/broadcom/bcm4908/bcm4906-netgear-r8000p.dts +1 -1
index 2dd028438c22..d1e9531f8bfe 100644
@@ -7,7 +7,7 @@
 #include "bcm4906.dtsi"
 
 / {
-	compatible = "netgear,r8000p", "brcm,bcm4906", "brcm,bcm4908";
+	compatible = "netgear,r8000p", "brcm,bcm4906", "brcm,bcmbca";
 	model = "Netgear R8000P";
 
 	memory@0 {
arch/arm64/boot/dts/broadcom/bcm4908/bcm4906-tplink-archer-c2300-v1.dts +1 -1
index 064f7f549665..8d91d0a475e3 100644
@@ -7,7 +7,7 @@
 #include "bcm4906.dtsi"
 
 / {
-	compatible = "tplink,archer-c2300-v1", "brcm,bcm4906", "brcm,bcm4908";
+	compatible = "tplink,archer-c2300-v1", "brcm,bcm4906", "brcm,bcmbca";
 	model = "TP-Link Archer C2300 V1";
 
 	memory@0 {
arch/arm64/boot/dts/broadcom/bcm4908/bcm4908-asus-gt-ac5300.dts +1 -1
index 04f8524b5335..787c7ddf9102 100644
@@ -6,7 +6,7 @@
 #include "bcm4908.dtsi"
 
 / {
-	compatible = "asus,gt-ac5300", "brcm,bcm4908";
+	compatible = "asus,gt-ac5300", "brcm,bcm4908", "brcm,bcmbca";
 	model = "Asus GT-AC5300";
 
 	memory@0 {
arch/arm64/boot/dts/broadcom/bcm4908/bcm4908-netgear-raxe500.dts +1 -1
index 3c2cf2d238b6..23b96c663239 100644
@@ -3,7 +3,7 @@
 #include "bcm4908.dtsi"
 
 / {
-	compatible = "netgear,raxe500", "brcm,bcm4908";
+	compatible = "netgear,raxe500", "brcm,bcm4908", "brcm,bcmbca";
 	model = "Netgear RAXE500";
 
 	memory@0 {
Signature · 1 line
2.34.1
#17 lore raw

Re: [RFC PATCH 3/3] arm64: dts: bcmbca: update bcm4808 board dts file

Quoted 2 lines · William Zhang

Update compatible string based on the new bcmbca binding rule for BCM4908 famliy based boards

Typo - family

Please explain why breaking the ABI (and users of these DTS_ is acceptable.

Quoted 5 lines

---

Best regards,
Krzysztof

#18 lore raw

Re: [RFC PATCH 3/3] arm64: dts: bcmbca: update bcm4808 board dts file

Quoted 7 lines · Krzysztof Kozlowski
Quoted 2 lines · William Zhang

Update compatible string based on the new bcmbca binding rule for BCM4908 famliy based boards

Typo - family

Please explain why breaking the ABI (and users of these DTS_ is acceptable.

This will be largely targeted towards Rafal who supports these kinds of devices with an upstream kernel. My understanding is that this is OK because we will always ship a DTB matching the Linux kernel, and I believe this is true for both the way that William and his group support these devices, as well as how OpenWrt, buildroot or other build systems envision to support these devices.

Rafal, does that sound about right?

Signature · 1 line
Florian
#19 lore raw

Re: [RFC PATCH 3/3] arm64: dts: bcmbca: update bcm4808 board dts file

Quoted 9 lines · Florian Fainelli
Quoted 7 lines · Krzysztof Kozlowski
Quoted 2 lines · William Zhang

Update compatible string based on the new bcmbca binding rule for BCM4908 famliy based boards

Typo - family

Please explain why breaking the ABI (and users of these DTS_ is acceptable.

Quoted 8 lines

This will be largely targeted towards Rafal who supports these kinds of devices with an upstream kernel. My understanding is that this is OK because we will always ship a DTB matching the Linux kernel, and I believe this is true for both the way that William and his group support these devices, as well as how OpenWrt, buildroot or other build systems envision to support these devices.

Rafal, does that sound about right?

I am fine, just maybe mention it in the commit because it literally breaks the DTSes.

I assume you considered all possible uses outside of Linux like U-Boot, BSD etc?

Best regards,
Krzysztof

author maintainer
#20 lore raw

Re: [RFC PATCH 3/3] arm64: dts: bcmbca: update bcm4808 board dts file

Quoted 8 lines · Krzysztof Kozlowski
Quoted 7 lines · Florian Fainelli
Quoted 6 lines · Krzysztof Kozlowski
Quoted 2 lines · William Zhang

Update compatible string based on the new bcmbca binding rule for BCM4908 famliy based boards

Typo - family

Quoted 8 lines
Quoted 8 lines
Quoted 1 line

Please explain why breaking the ABI (and users of these DTS_ is acceptable.

This will be largely targeted towards Rafal who supports these kinds of devices with an upstream kernel. My understanding is that this is OK because we will always ship a DTB matching the Linux kernel, and I believe this is true for both the way that William and his group support these devices, as well as how OpenWrt, buildroot or other build systems envision to support these devices.

Yes that is correct for bca group.

Quoted 11 lines
Quoted 2 lines

Rafal, does that sound about right?

I am fine, just maybe mention it in the commit because it literally breaks the DTSes.

I assume you considered all possible uses outside of Linux like U-Boot, BSD etc?

Best regards,
Krzysztof

The reason for this patch is to keep the bcmbca board dts in the same format and keep everything in the same yaml file. Understand 4908 was already upstream but luckily there is no driver in linux and u-boot that uses these 4908 compatible strings. They are only used in the board dts as far as I can see. So it does not really break anything in the end, unless someone use them in any driver but never upstream their code...

Rafal, please let us know if this is okay with you or any concern/possible break of existing system.

#21 lore raw

Re: [RFC PATCH 3/3] arm64: dts: bcmbca: update bcm4808 board dts file

Quoted 6 lines · William Zhang
Quoted 3 lines

Best regards,
Krzysztof

The reason for this patch is to keep the bcmbca board dts in the same format and keep everything in the same yaml file.

Not a good reason to change compatibles. You can have the same format and keep everything in same YAML file without replacing compatibles.

Quoted 5 lines

Understand 4908 was
already upstream but luckily there is no driver in linux and u-boot that uses these 4908 compatible strings. They are only used in the board dts as far as I can see. So it does not really break anything in the end, unless someone use them in any driver but never upstream their code...

So maybe just briefly mention it in the commit msg?

Best regards,
Krzysztof

author maintainer
#22 lore raw

Re: [RFC PATCH 3/3] arm64: dts: bcmbca: update bcm4808 board dts file

Quoted 11 lines · Krzysztof Kozlowski
Quoted 6 lines · William Zhang
Quoted 3 lines

Best regards,
Krzysztof

The reason for this patch is to keep the bcmbca board dts in the same format and keep everything in the same yaml file.

Not a good reason to change compatibles. You can have the same format and keep everything in same YAML file without replacing compatibles.

Well the existing 4908 compatible string is not the same format as we are proposing here: "board variant", "chip variant", "brcm, bcmbca"

Quoted 8 lines
Quoted 5 lines

Understand 4908 was
already upstream but luckily there is no driver in linux and u-boot that uses these 4908 compatible strings. They are only used in the board dts as far as I can see. So it does not really break anything in the end, unless someone use them in any driver but never upstream their code...

So maybe just briefly mention it in the commit msg?

I can do that for sure.

Quoted 2 lines

Best regards,
Krzysztof

maintainer
#23 lore raw

Re: [RFC PATCH 3/3] arm64: dts: bcmbca: update bcm4808 board dts file

In reply to Florian Fainelli
Quoted 9 lines · Florian Fainelli
Quoted 7 lines · Krzysztof Kozlowski
Quoted 2 lines · William Zhang

Update compatible string based on the new bcmbca binding rule for BCM4908 famliy based boards

Typo - family

Please explain why breaking the ABI (and users of these DTS_ is acceptable.

Quoted 8 lines

This will be largely targeted towards Rafal who supports these kinds of devices with an upstream kernel. My understanding is that this is OK because we will always ship a DTB matching the Linux kernel, and I believe this is true for both the way that William and his group support these devices, as well as how OpenWrt, buildroot or other build systems envision to support these devices.

Rafal, does that sound about right?

Right - in all cases I'm aware of - Linux gets shipped with DTB files. So such change won't actually break anything in real world.

#24 lore raw

Re: [RFC PATCH 3/3] arm64: dts: bcmbca: update bcm4808 board dts file

Quoted 18 lines · Rafał Miłecki
Quoted 17 lines · Florian Fainelli
Quoted 8 lines · Krzysztof Kozlowski
Quoted 2 lines · William Zhang

Update compatible string based on the new bcmbca binding rule for BCM4908 famliy based boards

Typo - family

Please explain why breaking the ABI (and users of these DTS_ is acceptable.

This will be largely targeted towards Rafal who supports these kinds of devices with an upstream kernel. My understanding is that this is OK because we will always ship a DTB matching the Linux kernel, and I believe this is true for both the way that William and his group support these devices, as well as how OpenWrt, buildroot or other build systems envision to support these devices.

Quoted 4 lines
Quoted 1 line

Rafal, does that sound about right?

Right - in all cases I'm aware of - Linux gets shipped with DTB files. So such change won't actually break anything in real world.

We don't really talk here about Linux, but other projects, like bootloaders or *BSD...

Best regards,
Krzysztof

maintainer
#25 lore raw

Re: [RFC PATCH 3/3] arm64: dts: bcmbca: update bcm4808 board dts file

Quoted 19 lines · Krzysztof Kozlowski
Quoted 18 lines · Rafał Miłecki
Quoted 17 lines · Florian Fainelli
Quoted 8 lines · Krzysztof Kozlowski
Quoted 2 lines · William Zhang

Update compatible string based on the new bcmbca binding rule for BCM4908 famliy based boards

Typo - family

Please explain why breaking the ABI (and users of these DTS_ is acceptable.

This will be largely targeted towards Rafal who supports these kinds of devices with an upstream kernel. My understanding is that this is OK because we will always ship a DTB matching the Linux kernel, and I believe this is true for both the way that William and his group support these devices, as well as how OpenWrt, buildroot or other build systems envision to support these devices.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

Rafal, does that sound about right?

Right - in all cases I'm aware of - Linux gets shipped with DTB files. So such change won't actually break anything in real world.

We don't really talk here about Linux, but other projects, like bootloaders or *BSD...

Right, let me more specific.

BCM4908 uses pkgtb firmware images. Those images contain:
1. bootfs (atf, u-boot, kernel, DTB files)
2. rootfs (filesystem)

So when you flash BCM4908 firmware it always contains:
1. U-Boot and DTB for it
2. Kernel and DTB for it
(+ more stuff)

There isn't any on-flash DTB file that doesn't get updated when flashing a new image.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel