[pvrusb2] Terratec Grabster AV400 and 2.6.37

[Sven Barth]

On 05.02.2011 17:34, Mike Isely wrote:
> On Sat, 5 Feb 2011, Sven Barth wrote:
>
>> Hi again!
>>
>> I could pinpoint the source of the problem in the cx25840 module (it's volume
>> related... something the cx25837 chip on the AV400 does not support), so I'll
>> report this to the linux-media list.
>>
>> But now that I've (again) patched the cx25840 module I can use the AV400
>> without audio problems on my PC. :D
>>
>
>
>> As I hope that this regression will be fixed soon, what needs to be tested so
>> that the AV400 will become a part of the in-kernel version?
>> I have used the composite part of the device for months now without any
>> problems. I'll need to test S-video, but so far I'm totally satisfied with the
>> device (it runs better than on Windows).
>
> Basically I'd like to verify that all the inputs (audio and video
> routing configuration) work correctly; if I remember correctly that was
> really the part of the port that we weren't sure about.  I can't do that
> myself because I don't have access to the hardware.  You do - I can help
> you understand the routing mechanism, and if you can generate a patch
> and verify that it all works, then I will pull it into the driver.
>

I can tell you the following: the default configuration (composite) 
works without problems (including audio). S-video does not work 
correctly though: I see the video input, but it's vertically shifted 
(~halfway with the lower half being visible at the upper half), the 
picture moves/flickers horizontally and I see only outlines (audio works 
correctly).

> Changes that might be needed to the cx25840 driver is outside the scope
> of the pvrusb2 driver, but we can get those submitted to the v4l-dvb
> tree provided the changes don't regress other aspects of that driver.
> Regression testing cx25840 is a serious pain - I can only verify that it
> continues to work correctly with pvrusb2-driven devices but this part is
> used by a lot of other drivers so it's kind of a crap-shoot to ensure
> real stability there across everything when there's a change that looks
> like it might be a problem in other contexts.

Well... I have already reported my findings to linux-media, so let's see 
whether they like my workaround or need to find a better solution. :D

Regards,
Sven