[RFC PATCH 0/2] another 'interrupts-extended' replacement needed?

[RFC PATCH 0/2] another 'interrupts-extended' replacement needed?

[Wolfram Sang]

Hi Rob,

I want to fix the following 'dtbs_check' issue:

arch/arm/boot/dts/renesas/sh73a0-kzm9g.dtb: accelerometer@1d (adi,adxl345): interrupts-extended: [[18, 2, 4], [18, 3, 4]] is too long
	from schema $id: http://devicetree.org/schemas/iio/accel/adi,adxl345.yaml#

However, this series uncovers another place where 'interrupts-extended'
is not recognized as an alternative to 'interrupts'. Namely, when this
is part of a 'dependencies:' like in this binding:

arch/arm/boot/dts/renesas/sh73a0-kzm9g.dtb: accelerometer@1d (adi,adxl345): 'interrupts' is a dependency of 'interrupt-names'
	from schema $id: http://devicetree.org/schemas/iio/accel/adi,adxl345.yaml#

Now, I wonder if this also needs to be fixed in dtschema? Or do we need
another way to express the dependency?

Thanks for any pointers,

   Wolfram


Wolfram Sang (2):
  dt-bindings: iio: accel: adxl345: document second interrupt
  ARM: dts: renesas: kzm9g: name interrupts for accelerometer

 .../bindings/iio/accel/adi,adxl345.yaml          | 16 +++++++++++-----
 arch/arm/boot/dts/renesas/sh73a0-kzm9g.dts       |  1 +
 2 files changed, 12 insertions(+), 5 deletions(-)

-- 
2.47.2

[RFC PATCH 1/2] dt-bindings: iio: accel: adxl345: document second interrupt

[Wolfram Sang]

The pinout of all the supported chips in this binding have two interrupt
pins. Document the second one, too, even though the Linux driver
currently does not support the second interrupt. Boards may have it
wired nonetheless.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
---

Passes dt_binding_check. But uncovers another issue. See coverletter.

 .../bindings/iio/accel/adi,adxl345.yaml          | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)

diff --git a/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml b/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml
index a23a626bfab6..806b0250ec07 100644
--- a/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml
+++ b/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml
@@ -35,11 +35,16 @@ properties:
   spi-3wire: true
 
   interrupts:
-    maxItems: 1
+    minItems: 1
+    maxItems: 2
 
   interrupt-names:
-    items:
-      - enum: [INT1, INT2]
+    oneOf:
+      - items:
+          - enum: [INT1, INT2]
+      - items:
+          - const: INT1
+          - const: INT2
 
 dependencies:
   interrupts: [ interrupt-names ]
@@ -84,7 +89,8 @@ examples:
             spi-cpol;
             spi-cpha;
             interrupt-parent = <&gpio0>;
-            interrupts = <0 IRQ_TYPE_LEVEL_HIGH>;
-            interrupt-names = "INT2";
+            interrupts = <0 IRQ_TYPE_LEVEL_HIGH>,
+                         <1 IRQ_TYPE_LEVEL_HIGH>;
+            interrupt-names = "INT1", "INT2";
         };
     };
-- 
2.47.2

Re: [RFC PATCH 1/2] dt-bindings: iio: accel: adxl345: document second interrupt

[Geert Uytterhoeven]

On Tue, 7 Oct 2025 at 01:20, Wolfram Sang
<wsa+renesas@sang-engineering.com> wrote:
> The pinout of all the supported chips in this binding have two interrupt
> pins. Document the second one, too, even though the Linux driver
> currently does not support the second interrupt. Boards may have it
> wired nonetheless.
>
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [RFC PATCH 1/2] dt-bindings: iio: accel: adxl345: document second interrupt

[Rob Herring]

On Mon, Oct 06, 2025 at 07:21:17PM +0200, Wolfram Sang wrote:
> The pinout of all the supported chips in this binding have two interrupt
> pins. Document the second one, too, even though the Linux driver
> currently does not support the second interrupt. Boards may have it
> wired nonetheless.
> 
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> ---
> 
> Passes dt_binding_check. But uncovers another issue. See coverletter.
> 
>  .../bindings/iio/accel/adi,adxl345.yaml          | 16 +++++++++++-----
>  1 file changed, 11 insertions(+), 5 deletions(-)
> 
> diff --git a/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml b/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml
> index a23a626bfab6..806b0250ec07 100644
> --- a/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml
> +++ b/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml
> @@ -35,11 +35,16 @@ properties:
>    spi-3wire: true
>  
>    interrupts:
> -    maxItems: 1
> +    minItems: 1
> +    maxItems: 2
>  
>    interrupt-names:
> -    items:
> -      - enum: [INT1, INT2]
> +    oneOf:
> +      - items:
> +          - enum: [INT1, INT2]
> +      - items:
> +          - const: INT1
> +          - const: INT2

This is better written as:

minItems: 1
items:
  - enum: [INT1, INT2]
  - const: INT2

It is mainly better because using 'oneOf' results in poor error 
messages.

Rob

Re: [RFC PATCH 1/2] dt-bindings: iio: accel: adxl345: document second interrupt

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 450 bytes --]


> > +    oneOf:
> > +      - items:
> > +          - enum: [INT1, INT2]
> > +      - items:
> > +          - const: INT1
> > +          - const: INT2
> 
> This is better written as:
> 
> minItems: 1
> items:
>   - enum: [INT1, INT2]
>   - const: INT2
> 
> It is mainly better because using 'oneOf' results in poor error 
> messages.

But wouldn't that allow naming both interrupts INT2? Or is this
magically prevented somehow?


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [RFC PATCH 1/2] dt-bindings: iio: accel: adxl345: document second interrupt

[Geert Uytterhoeven]

Hi Wolfram,

On Sun, 12 Oct 2025 at 02:01, Wolfram Sang
<wsa+renesas@sang-engineering.com> wrote:
> > > +    oneOf:
> > > +      - items:
> > > +          - enum: [INT1, INT2]
> > > +      - items:
> > > +          - const: INT1
> > > +          - const: INT2
> >
> > This is better written as:
> >
> > minItems: 1
> > items:
> >   - enum: [INT1, INT2]
> >   - const: INT2
> >
> > It is mainly better because using 'oneOf' results in poor error
> > messages.
>
> But wouldn't that allow naming both interrupts INT2? Or is this
> magically prevented somehow?

    "That's because we require strings to be unique entries, so [...]
twice will be rejected."

https://lore.kernel.org/all/20251009200054.GA3245555-robh@kernel.org/

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [RFC PATCH 1/2] dt-bindings: iio: accel: adxl345: document second interrupt

[Geert Uytterhoeven]

Hi Rob,

On Thu, 9 Oct 2025 at 23:22, Rob Herring <robh@kernel.org> wrote:
> On Mon, Oct 06, 2025 at 07:21:17PM +0200, Wolfram Sang wrote:
> > The pinout of all the supported chips in this binding have two interrupt
> > pins. Document the second one, too, even though the Linux driver
> > currently does not support the second interrupt. Boards may have it
> > wired nonetheless.
> >
> > Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> > ---
> >
> > Passes dt_binding_check. But uncovers another issue. See coverletter.
> >
> >  .../bindings/iio/accel/adi,adxl345.yaml          | 16 +++++++++++-----
> >  1 file changed, 11 insertions(+), 5 deletions(-)
> >
> > diff --git a/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml b/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml
> > index a23a626bfab6..806b0250ec07 100644
> > --- a/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml
> > +++ b/Documentation/devicetree/bindings/iio/accel/adi,adxl345.yaml
> > @@ -35,11 +35,16 @@ properties:
> >    spi-3wire: true
> >
> >    interrupts:
> > -    maxItems: 1
> > +    minItems: 1
> > +    maxItems: 2
> >
> >    interrupt-names:
> > -    items:
> > -      - enum: [INT1, INT2]
> > +    oneOf:
> > +      - items:
> > +          - enum: [INT1, INT2]
> > +      - items:
> > +          - const: INT1
> > +          - const: INT2
>
> This is better written as:
>
> minItems: 1
> items:
>   - enum: [INT1, INT2]
>   - const: INT2
>
> It is mainly better because using 'oneOf' results in poor error
> messages.

For those that missed the other email, where you explained about
listing strings like "INT2" twice:

    "That's because we require strings to be unique entries".

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

[RFC PATCH 2/2] ARM: dts: renesas: kzm9g: name interrupts for accelerometer

[Wolfram Sang]

Per the updated binding documentation, both interrupts must be named.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
---

This uncovers another issue. See coverletter.

 arch/arm/boot/dts/renesas/sh73a0-kzm9g.dts | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm/boot/dts/renesas/sh73a0-kzm9g.dts b/arch/arm/boot/dts/renesas/sh73a0-kzm9g.dts
index 1ce07d0878dc..0a9cd61bcb5f 100644
--- a/arch/arm/boot/dts/renesas/sh73a0-kzm9g.dts
+++ b/arch/arm/boot/dts/renesas/sh73a0-kzm9g.dts
@@ -209,6 +209,7 @@ accelerometer@1d {
 		reg = <0x1d>;
 		interrupts-extended = <&irqpin3 2 IRQ_TYPE_LEVEL_HIGH>,
 				      <&irqpin3 3 IRQ_TYPE_LEVEL_HIGH>;
+		interrupt-names = "INT1", "INT2";
 	};
 
 	rtc@32 {
-- 
2.47.2

Re: [RFC PATCH 2/2] ARM: dts: renesas: kzm9g: name interrupts for accelerometer

[Geert Uytterhoeven]

On Mon, 6 Oct 2025 at 19:21, Wolfram Sang
<wsa+renesas@sang-engineering.com> wrote:
> Per the updated binding documentation, both interrupts must be named.
>
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [RFC PATCH 0/2] another 'interrupts-extended' replacement needed?

[Rob Herring]

On Mon, Oct 06, 2025 at 07:21:16PM +0200, Wolfram Sang wrote:
> Hi Rob,
> 
> I want to fix the following 'dtbs_check' issue:
> 
> arch/arm/boot/dts/renesas/sh73a0-kzm9g.dtb: accelerometer@1d (adi,adxl345): interrupts-extended: [[18, 2, 4], [18, 3, 4]] is too long
> 	from schema $id: http://devicetree.org/schemas/iio/accel/adi,adxl345.yaml#
> 
> However, this series uncovers another place where 'interrupts-extended'
> is not recognized as an alternative to 'interrupts'. Namely, when this
> is part of a 'dependencies:' like in this binding:
> 
> arch/arm/boot/dts/renesas/sh73a0-kzm9g.dtb: accelerometer@1d (adi,adxl345): 'interrupts' is a dependency of 'interrupt-names'
> 	from schema $id: http://devicetree.org/schemas/iio/accel/adi,adxl345.yaml#
> 
> Now, I wonder if this also needs to be fixed in dtschema? Or do we need
> another way to express the dependency?

In this case, the 'interrupt-names: [ interrupts ]' dependency should be 
dropped because that's *always* the dependency and the core handles it 
correctly:

dependencies:
  interrupt-names:
    oneOf:
      - required:
          - interrupts
      - required:
          - interrupts-extended

Rob

Re: [RFC PATCH 0/2] another 'interrupts-extended' replacement needed?

[Rob Herring]

On Thu, Oct 9, 2025 at 3:35 PM Rob Herring <robh@kernel.org> wrote:
>
> On Mon, Oct 06, 2025 at 07:21:16PM +0200, Wolfram Sang wrote:
> > Hi Rob,
> >
> > I want to fix the following 'dtbs_check' issue:
> >
> > arch/arm/boot/dts/renesas/sh73a0-kzm9g.dtb: accelerometer@1d (adi,adxl345): interrupts-extended: [[18, 2, 4], [18, 3, 4]] is too long
> >       from schema $id: http://devicetree.org/schemas/iio/accel/adi,adxl345.yaml#
> >
> > However, this series uncovers another place where 'interrupts-extended'
> > is not recognized as an alternative to 'interrupts'. Namely, when this
> > is part of a 'dependencies:' like in this binding:
> >
> > arch/arm/boot/dts/renesas/sh73a0-kzm9g.dtb: accelerometer@1d (adi,adxl345): 'interrupts' is a dependency of 'interrupt-names'
> >       from schema $id: http://devicetree.org/schemas/iio/accel/adi,adxl345.yaml#
> >
> > Now, I wonder if this also needs to be fixed in dtschema? Or do we need
> > another way to express the dependency?
>
> In this case, the 'interrupt-names: [ interrupts ]' dependency should be
> dropped because that's *always* the dependency and the core handles it
> correctly:
>
> dependencies:
>   interrupt-names:
>     oneOf:
>       - required:
>           - interrupts
>       - required:
>           - interrupts-extended

However, the "interrupts: [ interrupt-names ]" depend is not handled
correctly. It would allow 'interrupts-extended' without
'interrupt-names'.

I'll make a fix.

Rob

Re: [RFC PATCH 0/2] another 'interrupts-extended' replacement needed?

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 293 bytes --]


> However, the "interrupts: [ interrupt-names ]" depend is not handled
> correctly. It would allow 'interrupts-extended' without
> 'interrupt-names'.
> 
> I'll make a fix.

So, even after your fix, the above dependency needs to stay in this
binding document? Did I get this right?


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [RFC PATCH 0/2] another 'interrupts-extended' replacement needed?

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 211 bytes --]


> So, even after your fix, the above dependency needs to stay in this
> binding document? Did I get this right?

Keeping the dependency and using your improved dt-schema makes all
warnings go away. Thank you!


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]