On 3.08.2023 07:06, Pavan Kondeti wrote:
>
> On Wed, Aug 02, 2023 at 03:14:19PM +0200, Konrad Dybcio wrote:
>> On 2.08.2023 15:13, Rohit Agarwal wrote:
>>>
>>> On 8/2/2023 6:12 PM, Konrad Dybcio wrote:
>>>> On 2.08.2023 11:59, Rohit Agarwal wrote:
>>>>> Add dtsi for PMIC pm7550ba found in Qualcomm platforms.
>>>>>
>>>>> Signed-off-by: Rohit Agarwal <quic_rohiagar@xxxxxxxxxxx>
>>>>> ---
>>>> Subject: "PMIC pm7550ba" -> "pm7550ba PMIC"
>>>>
>>>> [...]
>>>>
>>>>> +
>>>>> + pm7550ba_eusb2_repeater: phy@fd00 {
>>>>> + compatible = "qcom,pm8550b-eusb2-repeater";
>>>> A new compatible should be introduced, so that it goes like this:
>>>>
>>>> compatible = "qcom,pm7550ba-eusb2-repeater", "qcom,pm8550b-eusb2-repeater";
>>> Just a doubt, Since the compatible can be same why we need to introduce a new compatible.
>>> Should every soc have a compatible string?
>> If it turns out that we need to add a quirk for PM7550BA 3 years down
>> the line, this approach lets us fix it for users that never updated
>> their device trees.
>>
>
> Trying to make my understanding clear.
>
> eUSB repeater is a peripheral in the PMIC. Do we need a separate
> compatible even if the peripheral is same in two different PMIC chips?
> I believe eUSB peripheral has some identification registers to apply any
> quirks in future.
Perhaps, but keeping the compatible tied to the specific hardware is
the way to go with the device tree. Most components don't have such
information avaiable, and since at introduction time there wasn't
any better name for it, "pm8550b-eusb2-repeater" was chosen.
Konrad
