[v2,1/2] dt-bindings: dma: Add Apple ADMAC

Apple's Audio DMA Controller (ADMAC) is used to fetch and store audio
samples on SoCs from the "Apple Silicon" family.

Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
---

After the v1 discussion, I dropped the apple,internal-irq-destination
property and instead the index of the usable interrupt is now signified
by prepending -1 entries to the interrupts= list. This works when I do
it like this:

  interrupt-parent = <&aic>;
  interrupts = <AIC_IRQ 0xffffffff 0>,
               <AIC_IRQ 626 IRQ_TYPE_LEVEL_HIGH>;

I would find it neat to do it like this:

  interrupts-extended = <0xffffffff>,
                        <&aic AIC_IRQ 626 IRQ_TYPE_LEVEL_HIGH>;

but unfortunately the kernel doesn't pick up on it:

[    0.767964] apple-admac 238200000.dma-controller: error -6: IRQ index 0 not found
[    0.773943] apple-admac 238200000.dma-controller: error -6: IRQ index 1 not found
[    0.780154] apple-admac 238200000.dma-controller: error -6: IRQ index 2 not found
[    0.786367] apple-admac 238200000.dma-controller: error -6: IRQ index 3 not found
[    0.788592] apple-admac 238200000.dma-controller: error -6: no usable interrupt

 .../devicetree/bindings/dma/apple,admac.yaml  | 68 +++++++++++++++++++
 1 file changed, 68 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/dma/apple,admac.yaml

On Tue, Apr 12, 2022 at 12:22:03AM +0200, Martin Povišer wrote:
> Apple's Audio DMA Controller (ADMAC) is used to fetch and store audio
> samples on SoCs from the "Apple Silicon" family.
> 
> Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
> ---
> 
> After the v1 discussion, I dropped the apple,internal-irq-destination
> property and instead the index of the usable interrupt is now signified
> by prepending -1 entries to the interrupts= list. This works when I do
> it like this:
> 
>   interrupt-parent = <&aic>;
>   interrupts = <AIC_IRQ 0xffffffff 0>,
>                <AIC_IRQ 626 IRQ_TYPE_LEVEL_HIGH>;


BTW, just use '-1'. dtc takes negative values (and other expressions).

> 
> I would find it neat to do it like this:
> 
>   interrupts-extended = <0xffffffff>,
>                         <&aic AIC_IRQ 626 IRQ_TYPE_LEVEL_HIGH>;
> 
> but unfortunately the kernel doesn't pick up on it:
> 
> [    0.767964] apple-admac 238200000.dma-controller: error -6: IRQ index 0 not found
> [    0.773943] apple-admac 238200000.dma-controller: error -6: IRQ index 1 not found
> [    0.780154] apple-admac 238200000.dma-controller: error -6: IRQ index 2 not found
> [    0.786367] apple-admac 238200000.dma-controller: error -6: IRQ index 3 not found
> [    0.788592] apple-admac 238200000.dma-controller: error -6: no usable interrupt

We should make this case work. It is less fragile IMO as it doesn't 
depend on the provider's translation of cells.

> 
>  .../devicetree/bindings/dma/apple,admac.yaml  | 68 +++++++++++++++++++
>  1 file changed, 68 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/dma/apple,admac.yaml
> 
> diff --git a/Documentation/devicetree/bindings/dma/apple,admac.yaml b/Documentation/devicetree/bindings/dma/apple,admac.yaml
> new file mode 100644
> index 000000000000..bbd5eaf5f709
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/dma/apple,admac.yaml
> @@ -0,0 +1,68 @@
> +# SPDX-License-Identifier: GPL-2.0-only OR BSD-2-Clause
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/dma/apple,admac.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: Apple Audio DMA Controller (ADMAC)
> +
> +description: |
> +  Apple's Audio DMA Controller (ADMAC) is used to fetch and store audio samples
> +  on SoCs from the "Apple Silicon" family.
> +
> +  The controller has been seen with up to 24 channels. Even-numbered channels
> +  are TX-only, odd-numbered are RX-only. Individual channels are coupled to
> +  fixed device endpoints.
> +
> +maintainers:
> +  - Martin Povišer <povik+lin@cutebit.org>
> +
> +allOf:
> +  - $ref: "dma-controller.yaml#"
> +
> +properties:
> +  compatible:
> +    items:
> +      - enum:
> +          - apple,t6000-admac
> +          - apple,t8103-admac
> +      - const: apple,admac
> +
> +  reg:
> +    maxItems: 1
> +
> +  '#dma-cells':
> +    const: 1
> +    description:
> +      Clients specify single cell with channel number.
> +
> +  dma-channels:
> +    maximum: 24
> +
> +  interrupts:
> +    minItems: 1
> +    maxItems: 4

I'm now confused why this is variable. Put -1 entries on the end if 
that's why it is variable.

This needs some description about there being 1 of 4 outputs being 
connected.

> +
> +required:
> +  - compatible
> +  - reg
> +  - '#dma-cells'
> +  - dma-channels
> +  - interrupts
> +
> +additionalProperties: false
> +
> +examples:
> +  - |
> +    #include <dt-bindings/interrupt-controller/apple-aic.h>
> +    #include <dt-bindings/interrupt-controller/irq.h>
> +
> +    admac: dma-controller@238200000 {
> +      compatible = "apple,t8103-admac", "apple,admac";
> +      reg = <0x38200000 0x34000>;
> +      dma-channels = <24>;
> +      interrupt-parent = <&aic>;
> +      interrupts = <AIC_IRQ 0xffffffff 0>,
> +                   <AIC_IRQ 626 IRQ_TYPE_LEVEL_HIGH>;
> +      #dma-cells = <1>;
> +    };
> -- 
> 2.33.0
> 
>

> On 14. 4. 2022, at 17:43, Rob Herring <robh@kernel.org> wrote:
> 
> On Tue, Apr 12, 2022 at 12:22:03AM +0200, Martin Povišer wrote:
>> Apple's Audio DMA Controller (ADMAC) is used to fetch and store audio
>> samples on SoCs from the "Apple Silicon" family.
>> 
>> Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
>> ---
>> 
>> After the v1 discussion, I dropped the apple,internal-irq-destination
>> property and instead the index of the usable interrupt is now signified
>> by prepending -1 entries to the interrupts= list. This works when I do
>> it like this:
>> 
>>  interrupt-parent = <&aic>;
>>  interrupts = <AIC_IRQ 0xffffffff 0>,
>>               <AIC_IRQ 626 IRQ_TYPE_LEVEL_HIGH>;
> 
> 
> BTW, just use '-1'. dtc takes negative values (and other expressions).

Ha! <-1> didn’t work for me but <(-1)> does.

> 
>> 
>> I would find it neat to do it like this:
>> 
>>  interrupts-extended = <0xffffffff>,
>>                        <&aic AIC_IRQ 626 IRQ_TYPE_LEVEL_HIGH>;
>> 
>> but unfortunately the kernel doesn't pick up on it:
>> 
>> [    0.767964] apple-admac 238200000.dma-controller: error -6: IRQ index 0 not found
>> [    0.773943] apple-admac 238200000.dma-controller: error -6: IRQ index 1 not found
>> [    0.780154] apple-admac 238200000.dma-controller: error -6: IRQ index 2 not found
>> [    0.786367] apple-admac 238200000.dma-controller: error -6: IRQ index 3 not found
>> [    0.788592] apple-admac 238200000.dma-controller: error -6: no usable interrupt
> 
> We should make this case work. It is less fragile IMO as it doesn't 
> depend on the provider's translation of cells.

Then I may send some patch to that end.

>> 
>> .../devicetree/bindings/dma/apple,admac.yaml  | 68 +++++++++++++++++++
>> 1 file changed, 68 insertions(+)
>> create mode 100644 Documentation/devicetree/bindings/dma/apple,admac.yaml
>> 
>> diff --git a/Documentation/devicetree/bindings/dma/apple,admac.yaml b/Documentation/devicetree/bindings/dma/apple,admac.yaml
>> new file mode 100644
>> index 000000000000..bbd5eaf5f709
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/dma/apple,admac.yaml
>> @@ -0,0 +1,68 @@
>> +# SPDX-License-Identifier: GPL-2.0-only OR BSD-2-Clause
>> +%YAML 1.2
>> +---
>> +$id: http://devicetree.org/schemas/dma/apple,admac.yaml#
>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>> +
>> +title: Apple Audio DMA Controller (ADMAC)
>> +
>> +description: |
>> +  Apple's Audio DMA Controller (ADMAC) is used to fetch and store audio samples
>> +  on SoCs from the "Apple Silicon" family.
>> +
>> +  The controller has been seen with up to 24 channels. Even-numbered channels
>> +  are TX-only, odd-numbered are RX-only. Individual channels are coupled to
>> +  fixed device endpoints.
>> +
>> +maintainers:
>> +  - Martin Povišer <povik+lin@cutebit.org>
>> +
>> +allOf:
>> +  - $ref: "dma-controller.yaml#"
>> +
>> +properties:
>> +  compatible:
>> +    items:
>> +      - enum:
>> +          - apple,t6000-admac
>> +          - apple,t8103-admac
>> +      - const: apple,admac
>> +
>> +  reg:
>> +    maxItems: 1
>> +
>> +  '#dma-cells':
>> +    const: 1
>> +    description:
>> +      Clients specify single cell with channel number.
>> +
>> +  dma-channels:
>> +    maximum: 24
>> +
>> +  interrupts:
>> +    minItems: 1
>> +    maxItems: 4
> 
> I'm now confused why this is variable. Put -1 entries on the end if 
> that's why it is variable.

That’s why. Fixed length it is then.

> 
> This needs some description about there being 1 of 4 outputs being 
> connected.

OK. Description there will be.

>> +
>> +required:
>> +  - compatible
>> +  - reg
>> +  - '#dma-cells'
>> +  - dma-channels
>> +  - interrupts
>> +
>> +additionalProperties: false
>> +
>> +examples:
>> +  - |
>> +    #include <dt-bindings/interrupt-controller/apple-aic.h>
>> +    #include <dt-bindings/interrupt-controller/irq.h>
>> +
>> +    admac: dma-controller@238200000 {
>> +      compatible = "apple,t8103-admac", "apple,admac";
>> +      reg = <0x38200000 0x34000>;
>> +      dma-channels = <24>;
>> +      interrupt-parent = <&aic>;
>> +      interrupts = <AIC_IRQ 0xffffffff 0>,
>> +                   <AIC_IRQ 626 IRQ_TYPE_LEVEL_HIGH>;
>> +      #dma-cells = <1>;
>> +    };
>> -- 
>> 2.33.0
>> 
>> 

Martin

> On 14. 4. 2022, at 19:23, Martin Povišer <povik@cutebit.org> wrote:
> 
>> 
>> On 14. 4. 2022, at 17:43, Rob Herring <robh@kernel.org> wrote:
>> 
>> On Tue, Apr 12, 2022 at 12:22:03AM +0200, Martin Povišer wrote:
>>> Apple's Audio DMA Controller (ADMAC) is used to fetch and store audio
>>> samples on SoCs from the "Apple Silicon" family.
>>> 
>>> Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
>>> ---
>>> 
>>> After the v1 discussion, I dropped the apple,internal-irq-destination
>>> property and instead the index of the usable interrupt is now signified
>>> by prepending -1 entries to the interrupts= list. This works when I do
>>> it like this:
>>> 
>>> interrupt-parent = <&aic>;
>>> interrupts = <AIC_IRQ 0xffffffff 0>,
>>> <AIC_IRQ 626 IRQ_TYPE_LEVEL_HIGH>;
>> 
>> 
>> BTW, just use '-1'. dtc takes negative values (and other expressions).
> 
> Ha! <-1> didn’t work for me but <(-1)> does.
> 
>> 
>>> 
>>> I would find it neat to do it like this:
>>> 
>>> interrupts-extended = <0xffffffff>,
>>> <&aic AIC_IRQ 626 IRQ_TYPE_LEVEL_HIGH>;
>>> 
>>> but unfortunately the kernel doesn't pick up on it:
>>> 
>>> [ 0.767964] apple-admac 238200000.dma-controller: error -6: IRQ index 0 not found
>>> [ 0.773943] apple-admac 238200000.dma-controller: error -6: IRQ index 1 not found
>>> [ 0.780154] apple-admac 238200000.dma-controller: error -6: IRQ index 2 not found
>>> [ 0.786367] apple-admac 238200000.dma-controller: error -6: IRQ index 3 not found
>>> [ 0.788592] apple-admac 238200000.dma-controller: error -6: no usable interrupt
>> 
>> We should make this case work. It is less fragile IMO as it doesn't 
>> depend on the provider's translation of cells.
> 
> Then I may send some patch to that end.

Turns out there’s no need. Passing in <0> in place of a phandle reference
looks like what we want:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/of/base.c?h=v5.18-rc2#n1334

(Confirmed by testing to do the right thing with the IRQs.)

Martin