Re: [PATCH v3 2/3] dt-bindings: i2c-ast2600: Add bindings for AST2600 i2C driver

[Andrew Jeffery]

On Fri, 29 Jul 2022, at 12:33, Ryan Chen wrote:
> Hello Andrew,
>
>> -----Original Message-----
>> From: Andrew Jeffery <andrew@xxxxxxxx>
>> Sent: Friday, July 29, 2022 10:29 AM
>> To: Ryan Chen <ryan_chen@xxxxxxxxxxxxxx>; Joel Stanley <joel@xxxxxxxxx>;
>> Philipp Zabel <p.zabel@xxxxxxxxxxxxxx>; linux-arm-kernel@xxxxxxxxxxxxxxxxxxx;
>> linux-aspeed@xxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx;
>> openbmc@xxxxxxxxxxxxxxxx
>> Cc: BMC-SW <BMC-SW@xxxxxxxxxxxxxx>
>> Subject: Re: [PATCH v3 2/3] dt-bindings: i2c-ast2600: Add bindings for AST2600
>> i2C driver
>> 
>> Hi Ryan,
>> 
>> On Mon, 16 May 2022, at 16:18, ryan_chen wrote:
>> > AST2600 support new register set for I2C controller, add bindings
>> > document to support driver of i2c new register mode controller
>> >
>> > Signed-off-by: ryan_chen <ryan_chen@xxxxxxxxxxxxxx>
>> > ---
>> >  .../bindings/i2c/aspeed,i2c-ast2600.ymal      | 78
>> +++++++++++++++++++
>> >  1 file changed, 78 insertions(+)
>> >  create mode 100644
>> > Documentation/devicetree/bindings/i2c/aspeed,i2c-ast2600.ymal
>> >
>> > diff --git
>> > a/Documentation/devicetree/bindings/i2c/aspeed,i2c-ast2600.ymal
>> > b/Documentation/devicetree/bindings/i2c/aspeed,i2c-ast2600.ymal
>> > new file mode 100644
>> > index 000000000000..7c75f5bac24f
>> > --- /dev/null
>> > +++ b/Documentation/devicetree/bindings/i2c/aspeed,i2c-ast2600.ymal
>> > @@ -0,0 +1,78 @@
>> > +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) %YAML 1.2
>> > +---
>> > +$id: http://devicetree.org/schemas/i2c/aspeed,i2c-ast2600.yaml#
>> > +$schema: http://devicetree.org/meta-schemas/core.yaml#
>> > +
>> > +title: AST2600 I2C Controller on the AST26XX SoCs Device Tree
>> > +Bindings
>> > +
>> > +maintainers:
>> > +  - Ryan Chen <ryan_chen@xxxxxxxxxxxxxx>
>> > +
>> > +allOf:
>> > +  - $ref: /schemas/i2c/i2c-controller.yaml#
>> > +
>> > +properties:
>> > +  compatible:
>> > +    enum:
>> > +      - aspeed,ast2600-i2c
>> 
>> The original driver uses e.g. aspeed,ast2500-i2c-bus for the subordinate
>> controllers. While the register layout changes, I'd prefer we try to use the
>> existing compatibles rather than introducing a new set and causing some
>> confusion.
>> 
>> Further, what you're proposing here is effectively being used to select the
>> driver implementation, which isn't the purpose of the devicetree.
>> 
>> My preference would be to reuse the existing compatibles and instead select
>> the driver implementation via Kconfig. Or, if we can figure out some way to do
>> so, support both register interfaces in the one driver implementation and fall
>> back to the old register interface where the new one isn't available (I don't
>> think this is feasible though).
>> 
> Yes, that the reason go for another driver ast2600 to implement.
> Like others SOC driver implement different generation have diff driver 
> in Kconfig 
> and Makefile.
> Example :
> https://github.com/torvalds/linux/blob/master/drivers/i2c/busses/Makefile#L82-L84
>
>
>> > +
>> > +  reg:
>> > +    minItems: 1
>> > +    items:
>> > +      - description: address offset and range of bus
>> > +      - description: address offset and range of bus buffer
>> > +
>> > +  interrupts:
>> > +    maxItems: 1
>> > +
>> > +  clocks:
>> > +    maxItems: 1
>> > +    description:
>> > +      root clock of bus, should reference the APB
>> > +      clock in the second cell
>> > +
>> > +  resets:
>> > +    maxItems: 1
>> > +
>> > +  bus-frequency:
>> > +    minimum: 500
>> > +    maximum: 2000000
>> > +    default: 100000
>> > +    description: frequency of the bus clock in Hz defaults to 100 kHz
>> > when not
>> > +      specified
>> > +
>> > +  multi-master:
>> > +    type: boolean
>> > +    description:
>> > +      states that there is another master active on this bus
>> > +
>> > +required:
>> > +  - reg
>> > +  - compatible
>> > +  - clocks
>> > +  - resets
>> > +
>> > +unevaluatedProperties: false
>> > +
>> > +examples:
>> > +  - |
>> > +    #include <dt-bindings/interrupt-controller/arm-gic.h>
>> > +    #include <dt-bindings/clock/ast2600-clock.h>
>> > +
>> > +    i2c_gr: i2c-global-regs@0 {
>> > +      compatible = "aspeed,ast2600-i2c-global", "syscon";
>> > +      reg = <0x0 0x20>;
>> > +      resets = <&syscon ASPEED_RESET_I2C>;
>> > +    };
>> > +
>> > +    i2c0: i2c-bus@80 {
>> > +      #address-cells = <1>;
>> > +      #size-cells = <0>;
>> > +      #interrupt-cells = <1>;
>> > +      compatible = "aspeed,ast2600-i2c-bus";
>> 
>> This isn't quite right with respect to your binding description above :)
> Yes, the compatible need to be " aspeed,ast2600-i2c" is that your point ?

Yes, but only if we agree that we should have different compatibles for 
the different drivers. I'm not convinced about that yet.

I think it's enough to have different Kconfig symbols, and select the 
old driver in aspeed_g4_defconfig, and the new driver in 
aspeed_g5_defconfig. Won't that gives us the right outcome without
requiring a new set of compatibles?

Andrew