[mpeg2] Re: Porting kfir for kernel 2.6

Anssi Hannula anssi.hannula at gmail.com
Sun Feb 13 20:20:14 CET 2005


Klaus Schmidinger wrote:
> Anssi Hannula wrote:
> 
>> Klaus Schmidinger wrote:
>>
>>> Anssi Hannula wrote:
>>>
>>>> Klaus Schmidinger wrote:
>>>>
>>>>> Anssi Hannula wrote:
>>>>>
>>>>>>
>>>>>> Here is a new patch to test (debug=2). I slightly changed the 
>>>>>> kfir_read function and also put a BUG() in kfir_close so that the 
>>>>>> kernel dumps a backtrace in a log when the device closes, so that 
>>>>>> we can see what closes it.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> Looks like we're finally getting somewhere here!
>>>>>
>>>>> After loading the driver successfully I did
>>>>>
>>>>>   cat /dev/video > x.x
>>>>>
>>>>> and it actually wrote TS data into that file.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> Now that's progress :)
>>>>
>>>>> After a while I did Ctrl+C to cancel the 'cat', but that
>>>>> completely froze the entire system :-(
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> Try this one.
>>>
>>>
>>>
>>>
>>> This one actually appears to work - no more freeze when closing
>>> the device :-)
>>>
>>> I've done the 'cat /dev/video > x.x' twice, and everything worked
>>> as expected (log attached).
>>>
>>> Now my next step will be to test it together with my 'sky' plugin
>>> and an actual VDR instance. I hope to be able to do so tomorrow.
> 
> 
> I did that test now.
> First of all, I get video and audio just fine!
> Even the annoying green line that was always present at the bottom of the
> screen is gone now :-) Don't know whether this has anything to do with your
> modifications or the different videodev on kernel 2.6 - I tend to think
> the latter is true.
> 
> There is still a small problem with buffers running full, but I guess
> this is a thread problem that has nothing to do with the kfir device, since
> that device does deliver lots of data, while the rest of VDR is apparently
> not starting to process it in time. But once it has started, it runs fine.
> 
> The last problem I'm still having may be harder to solve, and I had it
> all the time I have been using the kfir card (and it still persists on
> kernel 2.6). It has to do with block artefacts after the first recording.
> I'll start a new thread for this, so that we can finally finish this one.
> 
>>> Thanks for all your support!
>>
>>
>>
>> That's great :)
>>
>> My next step is to do some final cleanings and documentation updates, 
>> and release a patch.
>>
>> But what after that?
>> Is there someone who has write access to the CVS and is willing to 
>> create a 2.6 branch for kfir? Or should I just release the patch on my 
>> homepage? etc.etc.
> 
> 
> I believe I do have write access to the mpeg2 CVS.
> You could just send me the changed files (preferrably the complete 
> files, not
> a patch - to rule out any mistakes I might make ;-) and I can check them in
> for you.
> 
> Thanks again for a job well done!

Well, here's the whole kfir-directory. I modified all the sourcefiles, 
because they had erratic formatting. I also made few changes to README 
and made a NEWS entry.
Could the best option be to branch the current CVS as linux_2_4, and 
then modify the MAIN for 2.6?

BTW, I deleted some obsolete files from the driver which have newer 
versions already in the 2.6 kernel tree.

And, please test this before committing, just to be sure.

-- 
Anssi Hannula
-------------- next part --------------
A non-text attachment was scrubbed...
Name: kfir-2.6-driver.tar.bz2
Type: application/octet-stream
Size: 39089 bytes
Desc: not available
Url : http://www.linuxtv.org/pipermail/mpeg2/attachments/20050213/8191f042/kfir-2.6-driver.tar-0001.obj


More information about the mpeg2 mailing list