Partition nodes without a compatible have no restrictions on additional
properties. Fix this by ensuring 'unevaluatedProperties' is set when
there's no compatible property. If there is a compatible property, then
there should be a specific schema for it which will enforce
'unevaluatedProperties'.
Signed-off-by: Rob Herring (Arm) <robh@kernel.org>
---
Documentation/devicetree/bindings/mtd/mtd.yaml | 8 ++------
.../devicetree/bindings/mtd/partitions/fixed-partitions.yaml | 2 +-
.../devicetree/bindings/mtd/partitions/partition.yaml | 12 ++++++++++++
.../mtd/partitions/tplink,safeloader-partitions.yaml | 2 +-
Documentation/devicetree/bindings/mtd/partitions/u-boot.yaml | 2 +-
Documentation/devicetree/bindings/mtd/ti,gpmc-onenand.yaml | 2 +-
6 files changed, 18 insertions(+), 10 deletions(-)
diff --git a/Documentation/devicetree/bindings/mtd/mtd.yaml b/Documentation/devicetree/bindings/mtd/mtd.yaml
index e56dba83f00a..5a2d06c96c0d 100644
--- a/Documentation/devicetree/bindings/mtd/mtd.yaml
+++ b/Documentation/devicetree/bindings/mtd/mtd.yaml
@@ -36,12 +36,8 @@ properties:
- compatible
patternProperties:
- "@[0-9a-f]+$":
- $ref: partitions/partition.yaml
- deprecated: true
-
- "^partition@[0-9a-f]+":
- $ref: partitions/partition.yaml
+ "(^partition)?@[0-9a-f]+$":
+ $ref: /schemas/mtd/partitions/partition.yaml#/$defs/partition-node
deprecated: true
"^otp(-[0-9]+)?$":
diff --git a/Documentation/devicetree/bindings/mtd/partitions/fixed-partitions.yaml b/Documentation/devicetree/bindings/mtd/partitions/fixed-partitions.yaml
index 26bef777353d..4d79f786fcd5 100644
--- a/Documentation/devicetree/bindings/mtd/partitions/fixed-partitions.yaml
+++ b/Documentation/devicetree/bindings/mtd/partitions/fixed-partitions.yaml
@@ -33,7 +33,7 @@ properties:
patternProperties:
"@[0-9a-f]+$":
- $ref: partition.yaml#
+ $ref: partition.yaml#/$defs/partition-node
required:
- "#address-cells"
diff --git a/Documentation/devicetree/bindings/mtd/partitions/partition.yaml b/Documentation/devicetree/bindings/mtd/partitions/partition.yaml
index fe2f7d622e8b..51fd48af55ec 100644
--- a/Documentation/devicetree/bindings/mtd/partitions/partition.yaml
+++ b/Documentation/devicetree/bindings/mtd/partitions/partition.yaml
@@ -149,6 +149,18 @@ then:
# This is a generic file other binding inherit from and extend
additionalProperties: true
+$defs:
+ partition-node:
+ type: object
+ if:
+ not:
+ required: [ compatible ]
+ then:
+ $ref: '#'
+ unevaluatedProperties: false
+ else:
+ $ref: '#'
+
examples:
- |
partitions {
diff --git a/Documentation/devicetree/bindings/mtd/partitions/tplink,safeloader-partitions.yaml b/Documentation/devicetree/bindings/mtd/partitions/tplink,safeloader-partitions.yaml
index a24bbaac3a90..40e6eaab03ce 100644
--- a/Documentation/devicetree/bindings/mtd/partitions/tplink,safeloader-partitions.yaml
+++ b/Documentation/devicetree/bindings/mtd/partitions/tplink,safeloader-partitions.yaml
@@ -38,7 +38,7 @@ properties:
patternProperties:
"^partition-.*$":
- $ref: partition.yaml#
+ $ref: partition.yaml#/$defs/partition-node
required:
- partitions-table-offset
diff --git a/Documentation/devicetree/bindings/mtd/partitions/u-boot.yaml b/Documentation/devicetree/bindings/mtd/partitions/u-boot.yaml
index 327fa872c001..d51bdcb7e585 100644
--- a/Documentation/devicetree/bindings/mtd/partitions/u-boot.yaml
+++ b/Documentation/devicetree/bindings/mtd/partitions/u-boot.yaml
@@ -29,7 +29,7 @@ properties:
patternProperties:
"^partition-.*$":
- $ref: partition.yaml#
+ $ref: partition.yaml#/$defs/partition-node
unevaluatedProperties: false
diff --git a/Documentation/devicetree/bindings/mtd/ti,gpmc-onenand.yaml b/Documentation/devicetree/bindings/mtd/ti,gpmc-onenand.yaml
index 7d3ace4f5505..8db991dee7eb 100644
--- a/Documentation/devicetree/bindings/mtd/ti,gpmc-onenand.yaml
+++ b/Documentation/devicetree/bindings/mtd/ti,gpmc-onenand.yaml
@@ -36,7 +36,7 @@ properties:
patternProperties:
"@[0-9a-f]+$":
- $ref: /schemas/mtd/partitions/partition.yaml
+ $ref: /schemas/mtd/partitions/partition.yaml#/$defs/partition-node
allOf:
- $ref: /schemas/memory-controllers/ti,gpmc-child.yaml
--
2.51.0On Thu, Jan 08, 2026 at 11:53:16AM -0600, Rob Herring (Arm) wrote: > Partition nodes without a compatible have no restrictions on additional > properties. Fix this by ensuring 'unevaluatedProperties' is set when > there's no compatible property. If there is a compatible property, then > there should be a specific schema for it which will enforce > 'unevaluatedProperties'. > > Signed-off-by: Rob Herring (Arm) <robh@kernel.org> > --- > Documentation/devicetree/bindings/mtd/mtd.yaml | 8 ++------ > .../devicetree/bindings/mtd/partitions/fixed-partitions.yaml | 2 +- > .../devicetree/bindings/mtd/partitions/partition.yaml | 12 ++++++++++++ > .../mtd/partitions/tplink,safeloader-partitions.yaml | 2 +- > Documentation/devicetree/bindings/mtd/partitions/u-boot.yaml | 2 +- > Documentation/devicetree/bindings/mtd/ti,gpmc-onenand.yaml | 2 +- > 6 files changed, 18 insertions(+), 10 deletions(-) > > diff --git a/Documentation/devicetree/bindings/mtd/mtd.yaml b/Documentation/devicetree/bindings/mtd/mtd.yaml > index e56dba83f00a..5a2d06c96c0d 100644 > --- a/Documentation/devicetree/bindings/mtd/mtd.yaml > +++ b/Documentation/devicetree/bindings/mtd/mtd.yaml > @@ -36,12 +36,8 @@ properties: > - compatible > > patternProperties: > - "@[0-9a-f]+$": > - $ref: partitions/partition.yaml > - deprecated: true > - > - "^partition@[0-9a-f]+": > - $ref: partitions/partition.yaml > + "(^partition)?@[0-9a-f]+$": > + $ref: /schemas/mtd/partitions/partition.yaml#/$defs/partition-node > deprecated: true This makes partition@ deprecated. Looks unintended or needds commit msg explanation. > > "^otp(-[0-9]+)?$": > diff --git a/Documentation/devicetree/bindings/mtd/partitions/fixed-partitions.yaml b/Documentation/devicetree/bindings/mtd/partitions/fixed-partitions.yaml > index 26bef777353d..4d79f786fcd5 100644 > --- a/Documentation/devicetree/bindings/mtd/partitions/fixed-partitions.yaml > +++ b/Documentation/devicetree/bindings/mtd/partitions/fixed-partitions.yaml > @@ -33,7 +33,7 @@ properties: Best regards, Krzysztof
On Fri, Jan 9, 2026 at 3:36 AM Krzysztof Kozlowski <krzk@kernel.org> wrote: > > On Thu, Jan 08, 2026 at 11:53:16AM -0600, Rob Herring (Arm) wrote: > > Partition nodes without a compatible have no restrictions on additional > > properties. Fix this by ensuring 'unevaluatedProperties' is set when > > there's no compatible property. If there is a compatible property, then > > there should be a specific schema for it which will enforce > > 'unevaluatedProperties'. > > > > Signed-off-by: Rob Herring (Arm) <robh@kernel.org> > > --- > > Documentation/devicetree/bindings/mtd/mtd.yaml | 8 ++------ > > .../devicetree/bindings/mtd/partitions/fixed-partitions.yaml | 2 +- > > .../devicetree/bindings/mtd/partitions/partition.yaml | 12 ++++++++++++ > > .../mtd/partitions/tplink,safeloader-partitions.yaml | 2 +- > > Documentation/devicetree/bindings/mtd/partitions/u-boot.yaml | 2 +- > > Documentation/devicetree/bindings/mtd/ti,gpmc-onenand.yaml | 2 +- > > 6 files changed, 18 insertions(+), 10 deletions(-) > > > > diff --git a/Documentation/devicetree/bindings/mtd/mtd.yaml b/Documentation/devicetree/bindings/mtd/mtd.yaml > > index e56dba83f00a..5a2d06c96c0d 100644 > > --- a/Documentation/devicetree/bindings/mtd/mtd.yaml > > +++ b/Documentation/devicetree/bindings/mtd/mtd.yaml > > @@ -36,12 +36,8 @@ properties: > > - compatible > > > > patternProperties: > > - "@[0-9a-f]+$": > > - $ref: partitions/partition.yaml > > - deprecated: true > > - > > - "^partition@[0-9a-f]+": > > - $ref: partitions/partition.yaml > > + "(^partition)?@[0-9a-f]+$": > > + $ref: /schemas/mtd/partitions/partition.yaml#/$defs/partition-node > > deprecated: true > > This makes partition@ deprecated. Looks unintended or needds commit msg > explanation. No, the only change here is I combined the entries. "partition@" is deprecated when not under "partitions" node. Rob
On Fri, Jan 09, 2026 at 06:25:32PM -0600, Rob Herring wrote: > > > - $ref: partitions/partition.yaml > > > - deprecated: true > > > - > > > - "^partition@[0-9a-f]+": > > > - $ref: partitions/partition.yaml > > > + "(^partition)?@[0-9a-f]+$": > > > + $ref: /schemas/mtd/partitions/partition.yaml#/$defs/partition-node > > > deprecated: true > > > > This makes partition@ deprecated. Looks unintended or needds commit msg > > explanation. > > No, the only change here is I combined the entries. "partition@" is > deprecated when not under "partitions" node. Indeed, I misread the diff. Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com> Best regards, Krzysztof
Hello Rob,
On 08/01/2026 at 11:53:16 -06, "Rob Herring (Arm)" <robh@kernel.org> wrote:
> Partition nodes without a compatible have no restrictions on additional
> properties. Fix this by ensuring 'unevaluatedProperties' is set when
> there's no compatible property. If there is a compatible property, then
> there should be a specific schema for it which will enforce
> 'unevaluatedProperties'.
>
> Signed-off-by: Rob Herring (Arm) <robh@kernel.org>
> ---
Overall thank you very much for this series, I've been sruggling while
defining these and your approach seems very relevant. I am fine with all
the patches in this series, but will wait a bit for others to make a bit
of feedback.
[...]
> +$defs:
> + partition-node:
> + type: object
> + if:
> + not:
> + required: [ compatible ]
> + then:
> + $ref: '#'
> + unevaluatedProperties: false
> + else:
> + $ref: '#'
This, however, is total blackmagic to me. Would you mind explaining what
$ref: '#'
indicates? Is this a placeholder indicating "a reference must be given?
Also I do not understand the final else case, what is it covering?
Thanks for the little explanation (or you can give me a pointer to some
relevant doc I could read).
Thanks,
Miquèl
On Fri, Jan 9, 2026 at 3:02 AM Miquel Raynal <miquel.raynal@bootlin.com> wrote: > > Hello Rob, > > On 08/01/2026 at 11:53:16 -06, "Rob Herring (Arm)" <robh@kernel.org> wrote: > > > Partition nodes without a compatible have no restrictions on additional > > properties. Fix this by ensuring 'unevaluatedProperties' is set when > > there's no compatible property. If there is a compatible property, then > > there should be a specific schema for it which will enforce > > 'unevaluatedProperties'. > > > > Signed-off-by: Rob Herring (Arm) <robh@kernel.org> > > --- > > Overall thank you very much for this series, I've been sruggling while > defining these and your approach seems very relevant. I am fine with all > the patches in this series, but will wait a bit for others to make a bit > of feedback. > > [...] > > > +$defs: > > + partition-node: > > + type: object > > + if: > > + not: > > + required: [ compatible ] > > + then: > > + $ref: '#' > > + unevaluatedProperties: false > > + else: > > + $ref: '#' > > This, however, is total blackmagic to me. Would you mind explaining what > > $ref: '#' > > indicates? Is this a placeholder indicating "a reference must be given? It's what's known as JSON pointers. The '#' is a reference to the top level of this schema. > Also I do not understand the final else case, what is it covering? It's really just there so a $ref to partition.yaml#/$defs/partition-node applies the schema (all of partition.yaml) whether there's a compatible property or not. This all just works around that a schema like this doesn't work: $ref: foo.yaml if: ... then: unevaluatedProperties: false The evaluation of unevaluatedProperties doesn't "see" the $ref being in the parent. So we can't factor out the $ref. Hope that makes sense. Rob
Hi Rob, >> > +$defs: >> > + partition-node: >> > + type: object >> > + if: >> > + not: >> > + required: [ compatible ] >> > + then: >> > + $ref: '#' >> > + unevaluatedProperties: false >> > + else: >> > + $ref: '#' >> >> This, however, is total blackmagic to me. Would you mind explaining what >> >> $ref: '#' >> >> indicates? Is this a placeholder indicating "a reference must be given? > > It's what's known as JSON pointers. The '#' is a reference to the top > level of this schema. > >> Also I do not understand the final else case, what is it covering? > > It's really just there so a $ref to > partition.yaml#/$defs/partition-node applies the schema (all of > partition.yaml) whether there's a compatible property or not. > > This all just works around that a schema like this doesn't work: > > $ref: foo.yaml > if: > ... > then: > unevaluatedProperties: false > > The evaluation of unevaluatedProperties doesn't "see" the $ref being > in the parent. So we can't factor out the $ref. Oooh, ok, fully understood. I think I already faced that problem before. First time I hear about JSON pointers, thanks a lot for the heads up. Regarding this series, it feels like in the end, if I understood the discussion with Krzysztof correctly, there is no modification to bring? Let me know if you plan on sending a v2 of if I shall take v1 otherwise. Thanks, Miquèl
On Wed, Jan 14, 2026 at 7:58 AM Miquel Raynal <miquel.raynal@bootlin.com> wrote: > > Hi Rob, > > >> > +$defs: > >> > + partition-node: > >> > + type: object > >> > + if: > >> > + not: > >> > + required: [ compatible ] > >> > + then: > >> > + $ref: '#' > >> > + unevaluatedProperties: false > >> > + else: > >> > + $ref: '#' > >> > >> This, however, is total blackmagic to me. Would you mind explaining what > >> > >> $ref: '#' > >> > >> indicates? Is this a placeholder indicating "a reference must be given? > > > > It's what's known as JSON pointers. The '#' is a reference to the top > > level of this schema. > > > >> Also I do not understand the final else case, what is it covering? > > > > It's really just there so a $ref to > > partition.yaml#/$defs/partition-node applies the schema (all of > > partition.yaml) whether there's a compatible property or not. > > > > This all just works around that a schema like this doesn't work: > > > > $ref: foo.yaml > > if: > > ... > > then: > > unevaluatedProperties: false > > > > The evaluation of unevaluatedProperties doesn't "see" the $ref being > > in the parent. So we can't factor out the $ref. > > Oooh, ok, fully understood. I think I already faced that problem > before. First time I hear about JSON pointers, thanks a lot for the > heads up. > > Regarding this series, it feels like in the end, if I understood the > discussion with Krzysztof correctly, there is no modification to bring? > Let me know if you plan on sending a v2 of if I shall take v1 otherwise. Right, I think v1 is good to go. Rob
© 2016 - 2026 Red Hat, Inc.