Re: saa7146: please test the vb2 conversion!

[Hans Verkuil <hverkuil-cisco@xs4all.nl>]

On 07/04/2023 00:43, Stefan Herdler wrote:
> Hi Hans,
> 
> I'm sorry for my late replay, but it was quiet troublesome to build
> a kernel with the patch, also running on the rest of my hardware.
> 
> 
> On 24/03/23 22:21, Stefan Herdler wrote:
>> Hi Hans,
>>
>> great to read, that it is finally done, thank you for your work!
>>
>>
>> On 24/03/23 11:40 Hans Verkuil wrote:
>>> On 24/03/2023 11:37, Hans Verkuil wrote:
>>>> Hi all,
>>>>
>>>> I finished the vb2 conversion and tested what I could test. I am missing
>>>> 'full featured' hardware, so I could not test the analog video capture part
>> There is some miss understanding.
>> At VDR 'full featured' refers to all DVB-cards with decoder and OSD.
>>
>> I wasn't aware, that this definition doesn't seem to be common, sorry.
>>>> of that. It's not clear to me if VBI capture is also supported on those
>>>> cards, if so, then that needs to be tested as well.
> 
> I have talked to the "cable-guys" and they didn't know about VBI capture.
> 
> The whole analog capture of this cards was never very popular with VDR.
> The signal had to be encoded on the fly, this was never reliable as the
> PVR-cards with encoder.
> 
> I think it is o.k. to leave it untested.
>>>>
>>>> Note that there is one userspace-facing change: the VBI output settings
>>>> are kept, even if the vbi device is closed by the application.
>>>>
>>>> Before you had to open the vbi device, format the slice VBI output, and
>>>> write sliced VBI data to it. Closing the device would reset how VBI output
>>>> behaves. That is not in spec with the V4L2 API. The format is kept after
>>>> the device is closed.
>>>>
>>>> Any application that uses VBI output and that wants to keep the same
>>>> behavior would have to call VIDIOC_S_FMT with a struct v4l2_sliced_vbi_format
>>>> with a service_set field set to 0 to indicate that you don't want to
>>>> output any VBI anymore.
> 
> VBI output is used to switch the aspect-ratio via WSS.
> this should be supported by any av7110 card.
> 
> The software is run a daemon or plugin, so the userspace-facing change
> shouldn't matter.
> 
> I'll test this as soon as possible.
> 
> 
> 
> 
> I've done only basic testing so far, but unfortunately it already failed.
> 
> The test:
> Switch to a channel[*] and view the decoded video with tvtime.
> 
> The resulting picture is corrupted.
> Almost green with some pink traces at the outlines.
> 
> It reminds me to YCbCr component-yideo on a RGB-input.
> Maybe the input-format of saa7146 not set correctly?
> 
> The OSD is equally affected, but the card seems to run stable.

That's weird. When you are in this state, can you run
'v4l2-ctl -V -d /dev/videoX' for the video device that tvtime
is using? I'll try to test it with tvtime as well next week.
I have done my tests using qvidcap and qv4l2, and that looked fine.

Regards,

	Hans

> 
> 
> 
> * I used VDR for this, but it shouldn't matter.
> 
> Regards
> Stefan
> 
>>>>
>>>> If this is a problem, then I can make a module option that selects the old
>>>> behavior.
>>>>
>>>> BTW, if anyone has a spare full-featured card (i.e. with analog video capture
>>>> as well), then I would love to take it off your hands so that I can test that
>>>> myself!
>> There are only DVB-C boards with analog features.
>>
>> I personally never had cable-TV nor own any DVB-C cards.
>> But I try to find such a card with an analog module on it.
>>>>
>>>> This series has been tested on the two Hexium boards, the mxb board, and two
>>>> av7710 boards (DVB-C and DVB-T).
>>
>> I can test on the DVB-S hardware.
>>
>> But let me finish the API-documentation fist, it is almost done.
>> There are only the complains from chackpatch left to fix, I hope it is done quickly.
>>
>>
>> Regards
>> Stefan
>>
>>
>>>
>>> It does help if I point to the patches :-)
>>>
>>> The patch series is here:
>>>
>>> https://patchwork.linuxtv.org/project/linux-media/list/?series=10140
>>>
>>> It's also in my git tree:
>>>
>>> https://git.linuxtv.org/hverkuil/media_tree.git/log/?h=saa7146-clean
>>>
>>> Regards,
>>>
>>> 	Hans