Re: [PATCH] dt-bindings: spi: convert spi_atmel to json-schema

From: Sergiu.Moga
Date: Thu Jun 30 2022 - 02:00:01 EST


Hello Krzysztof,

On 29.06.2022 16:23, Krzysztof Kozlowski wrote:
> On 29/06/2022 14:58, Sergiu Moga wrote:
>> Convert SPI binding for Atmel/Microchip SoCs to Device Tree Schema
>> format.
>>
>> Signed-off-by: Sergiu Moga <sergiu.moga@xxxxxxxxxxxxx>
>> ---
>> .../devicetree/bindings/spi/atmel,spi.yaml | 82 +++++++++++++++++++
>> .../devicetree/bindings/spi/spi_atmel.txt | 36 --------
>> 2 files changed, 82 insertions(+), 36 deletions(-)
>> create mode 100644 Documentation/devicetree/bindings/spi/atmel,spi.yaml
>> delete mode 100644 Documentation/devicetree/bindings/spi/spi_atmel.txt
>>
>> diff --git a/Documentation/devicetree/bindings/spi/atmel,spi.yaml b/Documentation/devicetree/bindings/spi/atmel,spi.yaml
>> new file mode 100644
>> index 000000000000..751618a47235
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/spi/atmel,spi.yaml
>> @@ -0,0 +1,82 @@
>> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
>> +# Copyright (C) 2022 Microchip Technology, Inc. and its subsidiaries
>> +%YAML 1.2
>> +---
>> +$id: http://devicetree.org/schemas/spi/atmel,spi.yaml#
>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>> +
>> +title: Atmel SPI device
>> +
>> +maintainers:
>> + - Mark Brown <broonie@xxxxxxxxxx>
> This should be rather someone from Microchip.
>
>> +
>> +allOf:
>> + - $ref: "spi-controller.yaml#"
> No need for quotes.
>
>> +
>> +properties:
>> + compatible:
>> + oneOf:
>> + - items:
>> + - const: atmel,at91rm9200-spi
>> + - items:
> These are not items, just single entry. Should be combined with above
> and made an 'enum'.
>
>> + - const: microchip,sam9x60-spi
>> + - items:
>> + - const: microchip,sam9x60-spi
>> + - const: atmel,at91rm9200-spi
> This is wrong. Either this is a fallback or it is not. It's not the
> Schroedinger's cat... Maybe your DTS are wrong.

Indeed, this looks unnecessary. This here is only to avoid the error
caused by arm/boot/dts/at91-sam9x60ek.dts, since its spi0 node contains
both compatibles.

>> +
>> + reg:
>> + maxItems: 1
>> +
>> + interrupts:
>> + maxItems: 1
>> +
>> + cs-gpios: true
> No need, coming from spi-controller.yaml.
>
>> +
>> + clock-names:
>> + description:
>> + Tuple listing input clock names, "spi_clk" is a required element.
> Skip description.
>
>> + contains:
>> + const: spi_clk
>> + additionalItems: true
> no additionalItems.
>
>> +
>> + clocks:
>> + maxItems: 1
>> +
>> + atmel,fifo-size:
>> + $ref: /schemas/types.yaml#/definitions/uint32
>> + description: |
>> + Maximum number of data the RX and TX FIFOs can store for FIFO
>> + capable SPI controllers.
> minimum and maximum for values, if it is known.
>
>> +
>> +required:
>> + - compatible
>> + - reg
>> + - interrupts
>> + - clock-names
>> + - clocks
>> +
>> +unevaluatedProperties: false
>> +
>> +examples:
>> + - |
>> + #include <dt-bindings/gpio/gpio.h>
>> + #include <dt-bindings/interrupt-controller/irq.h>
>> +
>> + spi1: spi@fffcc000 {
>> + compatible = "atmel,at91rm9200-spi";
>> + reg = <0xfffcc000 0x4000>;
>> + interrupts = <13 IRQ_TYPE_LEVEL_HIGH 5>;
>> + #address-cells = <1>;
>> + #size-cells = <0>;
>> + clocks = <&spi1_clk>;
>> + clock-names = "spi_clk", "str2";
> This does not make really sense. You have one clock.
>
I am sorry about this. Yes, you are right, the original binding did not
really have it either. I just put it here while testing it to ensure
that the property accepts more than just "spi_clk". It seems that I
forgot to remove it before sending it. Thank you for pointing it out!

>> + cs-gpios = <&pioB 3 GPIO_ACTIVE_HIGH>;
>> + atmel,fifo-size = <32>;
>> +
>> + mmc@0 {
>> + compatible = "mmc-spi-slot";
>> + reg = <0>;
>> + gpios = <&pioC 4 GPIO_ACTIVE_HIGH>; /* CD */
>> + spi-max-frequency = <25000000>;
>> + };
>
> Best regards,
> Krzysztof


Regards,

    Sergiu