Re: [PATCH v2 1/4] dt-bindings: rtc: add schema for NXP S32G2/S32G3 SoCs

[Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>]

On 10/16/2024 7:08 PM, Alexandre Belloni wrote:

Hello Rob and Alexandre,

Thank you for your careful review!

> On 15/10/2024 16:27:17-0500, Rob Herring wrote:
>> On Tue, Oct 15, 2024 at 04:15:40PM -0500, Rob Herring wrote:
>>> On Tue, Oct 15, 2024 at 01:51:30PM +0300, Ciprian Costea wrote:
>>>> From: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>>>>
>>>> This patch adds the dt-bindings for NXP S32G2/S32G3 SoCs RTC driver.
>>>>
>>>> Co-developed-by: Bogdan-Gabriel Roman <bogdan-gabriel.roman@nxp.com>
>>>> Signed-off-by: Bogdan-Gabriel Roman <bogdan-gabriel.roman@nxp.com>
>>>> Co-developed-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
>>>> Signed-off-by: Ghennadi Procopciuc <ghennadi.procopciuc@nxp.com>
>>>> Signed-off-by: Ciprian Marian Costea <ciprianmarian.costea@oss.nxp.com>
>>>> ---
>>>>   .../devicetree/bindings/rtc/nxp,s32g-rtc.yaml | 102 ++++++++++++++++++
>>>>   1 file changed, 102 insertions(+)
>>>>   create mode 100644 Documentation/devicetree/bindings/rtc/nxp,s32g-rtc.yaml
>>>>
>>>> diff --git a/Documentation/devicetree/bindings/rtc/nxp,s32g-rtc.yaml b/Documentation/devicetree/bindings/rtc/nxp,s32g-rtc.yaml
>>>> new file mode 100644
>>>> index 000000000000..3a77d4dd8f3d
>>>> --- /dev/null
>>>> +++ b/Documentation/devicetree/bindings/rtc/nxp,s32g-rtc.yaml
>>>> @@ -0,0 +1,102 @@
>>>> +# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
>>>> +%YAML 1.2
>>>> +---
>>>> +$id: http://devicetree.org/schemas/rtc/nxp,s32g-rtc.yaml#
>>>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
>>>> +
>>>> +title: NXP S32G2/S32G3 Real Time Clock (RTC)
>>>> +
>>>> +maintainers:
>>>> +  - Bogdan Hamciuc <bogdan.hamciuc@nxp.com>
>>>> +  - Ciprian Marian Costea <ciprianmarian.costea@nxp.com>
>>>> +
>>>> +properties:
>>>> +  compatible:
>>>> +    oneOf:
>>>> +      - enum:
>>>> +          - nxp,s32g2-rtc
>>>> +      - items:
>>>> +          - const: nxp,s32g3-rtc
>>>> +          - const: nxp,s32g2-rtc
>>>> +
>>>> +  reg:
>>>> +    maxItems: 1
>>>> +
>>>> +  interrupts:
>>>> +    maxItems: 1
>>>> +
>>>> +  "#clock-cells":
>>>> +    const: 1
>>>> +
>>>> +  clocks:
>>>> +    items:
>>>> +      - description: ipg clock drives the access to the
>>>> +          RTC iomapped registers
>>>> +
>>>> +  clock-names:
>>>> +    items:
>>>> +      - const: ipg
>>>> +
>>>> +  assigned-clocks:
>>>> +    minItems: 1
>>>> +    items:
>>>> +      - description: Runtime clock source. It must be a clock
>>>> +            source for the RTC module. It will be disabled by hardware
>>>> +            during Standby/Suspend.
>>>> +      - description: Standby/Suspend clock source. It is optional
>>>> +            and can be used in case the RTC will continue ticking during
>>>> +            platform/system suspend. RTC hardware module contains a
>>>> +            hardware mux for clock source selection.
>>>
>>> If the RTC h/w contains a mux, then your mux inputs should be listed in
>>> 'clocks', not here.
>>>
>>>> +
>>>> +  assigned-clock-parents:
>>>> +    description: List of phandles to each parent clock.
>>>> +
>>>> +  assigned-clock-rates:
>>>> +    description: List of frequencies for RTC clock sources.
>>>> +            RTC module contains 2 hardware divisors which can be
>>>> +            enabled or not. Hence, available frequencies are the following
>>>> +            parent_freq, parent_freq / 512, parent_freq / 32 or
>>>> +            parent_freq / (512 * 32)
>>>
>>> In general, assigned-clocks* do not need to be documented and should
>>> never be required.
>>>
>>>> +
>>>> +required:
>>>> +  - compatible
>>>> +  - reg
>>>> +  - interrupts
>>>> +  - "#clock-cells"
>>>> +  - clocks
>>>> +  - clock-names
>>>> +  - assigned-clocks
>>>> +  - assigned-clock-parents
>>>> +  - assigned-clock-rates
>>>> +
>>>> +additionalProperties: false
>>>> +
>>>> +examples:
>>>> +  - |
>>>> +    #include <dt-bindings/interrupt-controller/arm-gic.h>
>>>> +    #include <dt-bindings/interrupt-controller/irq.h>
>>>> +
>>>> +    rtc0: rtc@40060000 {
>>>> +        compatible = "nxp,s32g3-rtc",
>>>> +                   "nxp,s32g2-rtc";
>>>> +        reg = <0x40060000 0x1000>;
>>>> +        interrupts = <GIC_SPI 121 IRQ_TYPE_LEVEL_HIGH>;
>>>> +        #clock-cells = <1>;
>>>> +        clocks = <&clks 54>;
>>>> +        clock-names = "ipg";
>>>> +        /*
>>>> +         * Configuration of default parent clocks.
>>>> +         * 'assigned-clocks' 0-3 IDs are Runtime clock sources
>>>> +         * 4-7 IDs are Suspend/Standby clock sources.
>>>> +         */
>>>> +        assigned-clocks = <&rtc0 2>, <&rtc0 4>;
>>>
>>> That's weird...
>>>
>>>> +        assigned-clock-parents = <&clks 56>, <&clks 55>;
>>>
>>> I'd expect these should be in 'clocks'. I don't think this node should
>>> be a clock provider unless it provides a clock to something outside the
>>> RTC.
>>>
>>> Looks like you are just using assigned-clocks to configure the clock mux
>>> in the RTC. That's way over complicated. Just define a vendor specific
>>> property with the mux settings.
>>
>> I just read v1 and got told use the clock framework...
>>
>> I disagree completely. Tons of h/w blocks have the ability to select
>> (internal to the block) from multiple clock sources. Making the block a
>> clock provider to itself is completely pointless and an overkill, and
>> we *never* do that. Any display controller or audio interface has
>> mutiple clock sources as just 2 examples.
> 
> And in 6 months, we are going to learn that the rtc is used to clock the
> wifi chip or whatever and we are going to need to add everything in the
> CCF and we will have an unused property that we are going to have to
> support forever to avoid breaking the ABI. This already happened...
> 
>>
>> However, I don't see why you need the divider config in DT. Can't you
>> figure out what divider you need based on input frequency? The output
>> frequency should be fixed, right?
>>
>> Rob
> 

Would the following approach be acceptable ?

1. Keep 'assigned-clocks' based implementation as optional (not 
required), in order to take leverage of its scalability for allowing 
different clock source and divisor settings.
2. Implement a default clock muxing setting in the driver for both 
Runtime and Suspend.


Best Regards,
Ciprian