Re: [PATCH 00/18] media: rcar: Streams support

[Tomi Valkeinen <tomi.valkeinen+renesas@ideasonboard.com>]

Hi,

On 19/02/2025 16:56, Niklas Söderlund wrote:
> On 2025-02-19 16:41:24 +0200, Tomi Valkeinen wrote:
>> Hi,
>>
>> On 19/02/2025 16:22, Niklas Söderlund wrote:
>>> Hi Tomi,
>>>
>>> I'm happy to see this, nice work.
>>>
>>> Unfortunately it does not apply to the media-tree. As this series is a
>>
>> It should, if you apply the dependencies first. But one of the dependencies
>> from you is a bit old and that doesn't apply without some conflicts
>> resolved. So... Yes, doesn't apply =).
> 
> :-)
> 
>>
>>> mix of fixes and new features as well as covering multiple drivers. Do
>>> you think it would make sens to break out the fixes per driver which we
>>> could review and apply ASAP and the new features which we can review but
>>> need to wait until the dependencies are meet?
>>
>> Yes. I think that's the RAW8 and RAW10 fixes. However, I can't test them
>> without streams and GMSL2, so...
> 
> Sure but the fixes for the format setup registers don't depend on
> streams support, right? The fixes are properly tested but depend on new
> features to enable hardware to produce the test environment. It's miles
> better then the BSP testing done when these things where first
> written...

If you say so. I'm reluctant to apply fixes that I can only test with a 
huge pile of other stuff =).

>>
>>> If the new features depends on stuff that is hard to upport, maybe just
>>> focus on the fixes to get them out of the way?
>>>
>>> On 2025-02-19 15:48:54 +0200, Tomi Valkeinen wrote:
>>>> Add streams support to Renesas rcar platform driver.
>>>>
>>>> The series attempts to keep compatibility with the current upstream.
>>>> However, in upstream there's some kind of custom multi-stream support
>>>> implemented to the rcar driver, which breaks at patch "media: rcar-csi2:
>>>> Simplify rcsi2_calc_mbps()".
>>>
>>> I would not worry about breaking the make shift multi-stream, it was
>>> based on our first attempt to allow streams for GMSL many years ago and
>>> it was bonkers ;-) As long as it don't break a single streams, even from
>>> subdevs that do not themself support streams I'm happy.
>>
>> Alright. Then you're happy with this series. I hope =). I don't have
>> anything to test this with, and only the v4h board.
> 
> You can test single stream on V4H using the staging GMSL2 driver and the
> TPG. I will ofc also test it so no need if it's troublesome for you to
> rewire the V4H you have to this configuration.

Yes, that I have tested. So I have used my V4H board with two GMSL2 
deserializers + imx219 cameras, with unpublished GMSL drivers, and some 
other stuff. And I have tested (just the patches in this series, and the 
dependencies, no "other stuff") with the upstream staging driver with 
deser TPG.

I meant that I can't test this with anything "real" except with the full 
multi-stream gmsl setup. It would be great to be able to just plug in a 
single imx219 or such, but, nope.

  Tomi