[PATCH 1/5] include: linux: Destage VCHIQ interface headers

[Florian Fainelli]

On 8/27/25 23:19, Stefan Wahren wrote:
> Hi Florian,
> 
> Am 28.08.25 um 04:17 schrieb Florian Fainelli:
>>
>>
>> On 8/27/2025 12:05 PM, Stefan Wahren wrote:
>>> Hi,
>>>
>>> Am 27.08.25 um 16:33 schrieb Umang Jain:
>>>> On Wed, Aug 27, 2025 at 02:40:16PM +0200, Laurent Pinchart wrote:
>>>>> Hi Jai,
>>>>>
>>>>> Thank you for the patch.
>>>>>
>>>>> On Wed, Aug 27, 2025 at 11:54:08AM +0530, Jai Luthra wrote:
>>>>>> From: Umang Jain <umang.jain at ideasonboard.com>
>>>>>>
>>>>>> Move the VCHIQ headers from drivers/staging/vc04_services/include to
>>>>>> include/linux/vchiq
>>>>>>
>>>>>> This is done so that they can be shared between the VCHIQ interface
>>>>>> (which is going to be de-staged in a subsequent commit from staging)
>>>>>> and the VCHIQ drivers left in the staging/vc04_services (namely
>>>>>> bcm2835-audio, bcm2835-camera).
>>>>>>
>>>>>> The include/linux/vchiq/ provides a central location to serve both
>>>>>> of these areas.
>>>>> Lots of SoC-specific headers are stored in include/linux/soc/$vendor/.
>>>>> This would be include/linux/soc/bcm/vchiq/ in this case. I'm also fine
>>>>> with include/linux/vchiq/ but other people may have a preference.
>>>> I agree with this point and I might have missed to notice the
>>>> include/linux/soc earlier. That's seems a better location to me since
>>>> it's actually broadcom-specific.
>>> I would expect that headers and source would be more related.
>>>
>>> For example:
>>>
>>> include/linux/soc/bcm
>>>
>>> drivers/soc/bcm/
>>
>> This is not Broadcom code, it is Raspberry Pi AFAICT, therefore, just 
>> like drivers/firmware/raspberrypi.c, we would need some namespacing 
>> here that reflects that, ideally.
> The VCHIQ code originally comes from Broadcom, but the current 
> implementation has been adapted and tested for Raspberry Pi. I'm not 
> against a Raspberry Pi specific namespace.

OK, that's a fair point, but it's not applicable to any other Broadcom 
SoC, more on that below.

> 
> Should this also apply to the DT bindings?

Ideally yes.
-- 
Florian