Skip to contentSkip to conversation

dt-bindings: arm: qcom: add ipq8064 board variants

PATCH In review Christian Marangi 3 participants 6 messages Nov 20, 2025 – Nov 21, 2025

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 dt-bindings: arm: qcom: add ipq8064 board variants

Document QCOM ipq8064 board variants ipq8062, ipq8065, ipq8066, ipq8068, ipq8069 now matched by the QCOM cpufreq nvmem driver if socinfo can't derive the variant from SMEM.

---
 .../devicetree/bindings/arm/qcom.yaml         | 20 +++++++++++++++++++
 1 file changed, 20 insertions(+)
Documentation/devicetree/bindings/arm/qcom.yaml +20 -0
index 18b5ed044f9f..0eb1619fede8 100644
@@ -299,12 +299,32 @@ properties:
               - qcom,ipq5424-rdp466
           - const: qcom,ipq5424
 
+      - items:
+          - const: qcom,ipq8062
+          - const: qcom,ipq8064
+
       - items:
           - enum:
               - mikrotik,rb3011
               - qcom,ipq8064-ap148
           - const: qcom,ipq8064
 
+      - items:
+          - const: qcom,ipq8065
+          - const: qcom,ipq8064
+
+      - items:
+          - const: qcom,ipq8066
+          - const: qcom,ipq8064
+
+      - items:
+          - const: qcom,ipq8068
+          - const: qcom,ipq8064
+
+      - items:
+          - const: qcom,ipq8069
+          - const: qcom,ipq8064
+
       - items:
           - enum:
               - qcom,ipq8074-hk01
Signature · 1 line
2.51.0
#1 lore raw
Quoted 15 lines · Christian Marangi

Document QCOM ipq8064 board variants ipq8062, ipq8065, ipq8066, ipq8068, ipq8069 now matched by the QCOM cpufreq nvmem driver if socinfo can't derive the variant from SMEM.

---
 .../devicetree/bindings/arm/qcom.yaml         | 20 +++++++++++++++++++
 1 file changed, 20 insertions(+)
Documentation/devicetree/bindings/arm/qcom.yaml +0 -0
index 18b5ed044f9f..0eb1619fede8 100644

Documentation/devicetree/bindings/arm/qcom.yaml properties: · line 304

Quoted 6 lines
Documentation/devicetree/bindings/arm/qcom.yaml +3 -0
@@ -299,12 +299,32 @@ properties:
               - qcom,ipq5424-rdp466
           - const: qcom,ipq5424
 
+      - items:
+          - const: qcom,ipq8062
+          - const: qcom,ipq8064

Since 'items' requires that all items are present (and in this order), we would normally have a board name go first.. but I suppose this is some sort of a fix to the issue that sparked this (posting the link for others to have more context)

But since these SoCs do exist, I wouldn't say this is necessarily wrong..

https://lore.kernel.org/linux-arm-msm/[email protected]/

Konrad

#2 lore raw
Quoted 28 lines · Konrad Dybcio
Quoted 15 lines · Christian Marangi

Document QCOM ipq8064 board variants ipq8062, ipq8065, ipq8066, ipq8068, ipq8069 now matched by the QCOM cpufreq nvmem driver if socinfo can't derive the variant from SMEM.

---
 .../devicetree/bindings/arm/qcom.yaml         | 20 +++++++++++++++++++
 1 file changed, 20 insertions(+)
Documentation/devicetree/bindings/arm/qcom.yaml +0 -0
index 18b5ed044f9f..0eb1619fede8 100644

Documentation/devicetree/bindings/arm/qcom.yaml properties: · line 304

Quoted 6 lines
Documentation/devicetree/bindings/arm/qcom.yaml +3 -0
@@ -299,12 +299,32 @@ properties:
               - qcom,ipq5424-rdp466
           - const: qcom,ipq5424
 
+      - items:
+          - const: qcom,ipq8062
+          - const: qcom,ipq8064

Since 'items' requires that all items are present (and in this order), we would normally have a board name go first.. but I suppose this is some sort of a fix to the issue that sparked this (posting the link for others to have more context)

Quoted 3 lines

But since these SoCs do exist, I wouldn't say this is necessarily wrong..

Well we can see this as a ""template"" for device that might be added using the ipq8062 or ipq8065 compatible.

When device with that variant will be added we would have to just add an enum with the real device name on top of it (as first element).

Honestly I should have added these compatible long time ago as on OpenWrt we have tons of device that are ipq8062 or ipq8065 with the compatible structure

"device,name", "qcom,ipq8065", "qcom,ipq8064".

I feel it might be wrong to simply add to the list "qcom,ipq8065" without the "qcom,ipq8064" part.

Quoted 2 lines

https://lore.kernel.org/linux-arm-msm/[email protected]/

Maybe I should add a Ref: tag referencing that?

Signature · 1 line
	Ansuel
maintainer
#3 lore raw
Quoted 36 lines · Christian Marangi
Quoted 28 lines · Konrad Dybcio
Quoted 15 lines · Christian Marangi

Document QCOM ipq8064 board variants ipq8062, ipq8065, ipq8066, ipq8068, ipq8069 now matched by the QCOM cpufreq nvmem driver if socinfo can't derive the variant from SMEM.

---
 .../devicetree/bindings/arm/qcom.yaml         | 20 +++++++++++++++++++
 1 file changed, 20 insertions(+)
Documentation/devicetree/bindings/arm/qcom.yaml +0 -0
index 18b5ed044f9f..0eb1619fede8 100644

Documentation/devicetree/bindings/arm/qcom.yaml properties: · line 304

Quoted 6 lines
Documentation/devicetree/bindings/arm/qcom.yaml +3 -0
@@ -299,12 +299,32 @@ properties:
               - qcom,ipq5424-rdp466
           - const: qcom,ipq5424
 
+      - items:
+          - const: qcom,ipq8062
+          - const: qcom,ipq8064

Since 'items' requires that all items are present (and in this order), we would normally have a board name go first.. but I suppose this is some sort of a fix to the issue that sparked this (posting the link for others to have more context)

Quoted 3 lines

But since these SoCs do exist, I wouldn't say this is necessarily wrong..

Well we can see this as a ""template"" for device that might be added using the ipq8062 or ipq8065 compatible.

Quoted 8 lines

When device with that variant will be added we would have to just add an enum with the real device name on top of it (as first element).

Honestly I should have added these compatible long time ago as on OpenWrt we have tons of device that are ipq8062 or ipq8065 with the compatible structure

"device,name", "qcom,ipq8065", "qcom,ipq8064".

If you don't you have any boards yet, you can do:

items:
- description: ...
- const: qcom,ipq8065
- const: qcom,ipq8064

Just to prevent skipping a board compatible.

But you said you have tons of devices, so...

Rob

#4 lore raw
Quoted 48 lines · Rob Herring
Quoted 36 lines · Christian Marangi
Quoted 28 lines · Konrad Dybcio
Quoted 15 lines · Christian Marangi

Document QCOM ipq8064 board variants ipq8062, ipq8065, ipq8066, ipq8068, ipq8069 now matched by the QCOM cpufreq nvmem driver if socinfo can't derive the variant from SMEM.

---
 .../devicetree/bindings/arm/qcom.yaml         | 20 +++++++++++++++++++
 1 file changed, 20 insertions(+)
Documentation/devicetree/bindings/arm/qcom.yaml +0 -0
index 18b5ed044f9f..0eb1619fede8 100644

Documentation/devicetree/bindings/arm/qcom.yaml properties: · line 304

Quoted 6 lines
Documentation/devicetree/bindings/arm/qcom.yaml +3 -0
@@ -299,12 +299,32 @@ properties:
               - qcom,ipq5424-rdp466
           - const: qcom,ipq5424
 
+      - items:
+          - const: qcom,ipq8062
+          - const: qcom,ipq8064

Since 'items' requires that all items are present (and in this order), we would normally have a board name go first.. but I suppose this is some sort of a fix to the issue that sparked this (posting the link for others to have more context)

Quoted 3 lines

But since these SoCs do exist, I wouldn't say this is necessarily wrong..

Well we can see this as a ""template"" for device that might be added using the ipq8062 or ipq8065 compatible.

Quoted 8 lines

When device with that variant will be added we would have to just add an enum with the real device name on top of it (as first element).

Honestly I should have added these compatible long time ago as on OpenWrt we have tons of device that are ipq8062 or ipq8065 with the compatible structure

"device,name", "qcom,ipq8065", "qcom,ipq8064".

If you don't you have any boards yet, you can do:

Quoted 9 lines

items:
- description: ...
- const: qcom,ipq8065
- const: qcom,ipq8064

Just to prevent skipping a board compatible.

But you said you have tons of devices, so...

Yes the problem is that it might take a while for the DTS to be picked so maybe it's better to push for the description solution?

Signature · 1 line
	Ansuel
#5 lore raw
Quoted 54 lines · Christian Marangi
Quoted 53 lines · Rob Herring
Quoted 36 lines · Christian Marangi
Quoted 28 lines · Konrad Dybcio
Quoted 15 lines · Christian Marangi

Document QCOM ipq8064 board variants ipq8062, ipq8065, ipq8066, ipq8068, ipq8069 now matched by the QCOM cpufreq nvmem driver if socinfo can't derive the variant from SMEM.

---
 .../devicetree/bindings/arm/qcom.yaml         | 20 +++++++++++++++++++
 1 file changed, 20 insertions(+)
Documentation/devicetree/bindings/arm/qcom.yaml +0 -0
index 18b5ed044f9f..0eb1619fede8 100644

Documentation/devicetree/bindings/arm/qcom.yaml properties: · line 304

Quoted 6 lines
Documentation/devicetree/bindings/arm/qcom.yaml +3 -0
@@ -299,12 +299,32 @@ properties:
               - qcom,ipq5424-rdp466
           - const: qcom,ipq5424
 
+      - items:
+          - const: qcom,ipq8062
+          - const: qcom,ipq8064

Since 'items' requires that all items are present (and in this order), we would normally have a board name go first.. but I suppose this is some sort of a fix to the issue that sparked this (posting the link for others to have more context)

Quoted 3 lines

But since these SoCs do exist, I wouldn't say this is necessarily wrong..

Well we can see this as a ""template"" for device that might be added using the ipq8062 or ipq8065 compatible.

Quoted 8 lines

When device with that variant will be added we would have to just add an enum with the real device name on top of it (as first element).

Honestly I should have added these compatible long time ago as on OpenWrt we have tons of device that are ipq8062 or ipq8065 with the compatible structure

"device,name", "qcom,ipq8065", "qcom,ipq8064".

If you don't you have any boards yet, you can do:

items:
- description: ...
- const: qcom,ipq8065
- const: qcom,ipq8064

Quoted 7 lines
Quoted 4 lines

Just to prevent skipping a board compatible.

But you said you have tons of devices, so...

Yes the problem is that it might take a while for the DTS to be picked so maybe it's better to push for the description solution?

DT depends on bindings, it's not so stringent the other way around

Konrad

Quoted 1 line