[PATCH v2 2/2] arm64: dts: qcom: Add msm8916 CoreSight components

[Mathieu Poirier]

On 8 May 2015 at 11:57, Ivan T. Ivanov <ivan.ivanov at linaro.org> wrote:
>
>> On May 8, 2015, at 7:17 PM, Mathieu Poirier <mathieu.poirier at linaro.org> wrote:
>>
>> On 8 May 2015 at 08:17, Ivan T. Ivanov <ivan.ivanov at linaro.org> wrote:
>>>
>>> On Fri, 2015-05-08 at 08:13 -0600, Mathieu Poirier wrote:
>>>> On 8 May 2015 at 07:47, Ivan T. Ivanov ivanov at linaro.org> wrote:
>>>>> On Fri, 2015-05-08 at 07:38 -0600, Mathieu Poirier wrote:
>>>>>> On 7 May 2015 at 09:36, Ivan T. Ivanov ivanov at linaro.org> wrote:
>>>>>>> Add initial set of CoreSight components found on Qualcomm's 8x16 chipset.
>>>>>>>
>>>>>>>
>>>>>>> +       replicator at 824000 {
>>>>>>> +               compatible = "qcom,coresight-replicator", "arm,primecell";
>>>>>>
>>>>>> Shouldn't it be "qcom,coresight-replicator1x" ?
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> True, I still wonder, why we have to have this compatible string?
>>>>> Drivers are probed by amba_id and "arm,primecell", after all.
>>>>>
>>>>
>>>> Drivers have their own compatible strings for historical reasons,
>>>> something I've been meaning to fix for a long time now...
>>>>
>>>
>>> Yep, I see that they have been platform drivers in the past, but now
>>> they are not, except coresight-replicator driver. IMHO, having
>>> additional compatible string could lead just to confusion.
>>
>> I did a little more research on this and based on what I found in the
>> kernel it may not need "fixing" after all.  The majority of drivers
>> that do specify "arm,primecell" also specify a device-specific
>> compatible string.  And in the case of CoreSight devices were
>> implementers can do pretty much whatever they  want with the ID
>> strings, it is only a matter of time before we need to call something
>> like of_device_is_compatible() to fix a quirk.
>>
>> Unless someone heavy asks to remove the device-specific compatible
>> strings I'd prefer keeping the current trend set forth by other
>> drivers and as such, will ask you to add the "1x" in this bindings.
>
>
> Well, I don’t strongly object against this “1x”, I will add it.
> My point is that if we can dynamically detect device version,
> which we can do in this case, it will be more robust to do it
> in this way.

I agree with you.  The device specific bindings will come handy when
there is a problem with the device version, something that is bound to
happen.

>
> If there are not issues with patch 1/2, I will like to fix and
> resend only this patch.

I'm good with 1/2, just this patch will be fine.

>
> Regards,
> Ivan
>
>