[linux-dvb] tda1004x update

[Hartmut Hackmann]

Michael Krufky wrote:
> Hartmut Hackmann wrote:
> 
>> Michael Krufky wrote:
>>
>>> Hartmut Hackmann wrote:
>>>
>>>> I just committed an update of the tda1004x dvb-t driver to the 
>>>> dvb-kernel
>>>> CVS. It only affects the tda10046 (hopefully):
>>>>
>>>> - fixed oddities at firmware download from eeprom.
>>>> - more tolerant vs crystal frequency offset. It should now work with up
>>>>   to -/+100ppm deviation. This can be the cause why some boards like
>>>>   the Lifeview just didn't work for some users.
>>>> - lower sampling clock to 48Mhz for low IF and 53MHz for 36MHz. 
>>>> Together
>>>>   with some other minor changes, this increases the sensitivity a bit.
>>>> - doesn't touch the GPIOs any longer. This needs to be handled by the
>>>>   board layer.
>>>>
>>>> The latter change might cause incompatibilities. However, it is 
>>>> necessary
>>>> since the pins are used very differently. I could not find problems 
>>>> with
>>>> boards i have access to. 
>>>
>>>
>>> What is the status of the testings with this patch?  Your 
>>> announcement mentions some possible incompatabilities...
>>>
>>> Mauro has already send your v4l patches to -mm... Do they depend on 
>>> this one?
>>
>>
>> I haven't heard anything but few reports that it gave the slight 
>> performance
>> improvement i expected.
>> I tested with 4 different cards: Lifeview, Philips Tiger and Europa 
>> and the
>> Medion. The Philips cards are at least compatible with the Compro and 
>> the ASUS.
>> I can get hands on one other important card in short term.
>> So there is quite some test coverage.
>> The GPIO thing can cause trouble with modern hybrid cards like the 
>> ASUS which
>> need proper handling of the GPIOs.
>>
>> Hermann, can you roll back the tda1004x driver to see whether the old 
>> version
>> works with your card?
>>
>> This went a bit fast. I didn't expect that Mauro would update the patches
>> again...
>>
>> Hartmut
> 
> 
> Hartmut-
> 
> Hmmm.... I do agree that Mauro might have sent the v4l patches too 
> quickly to -mm... However, -mm itself is somewhat of a testing area.
> 
> If I understand it correctly, the patches in v4l are only adding dvb 
> support, and will not break anything unrelated.  Is this correct?
> 
> If that is correct, then it is okay for Mauro to have sent them.  Now, I 
> need to know from you whether the corresponding dvb patch to tda1004x 
> should also be sent to -mm?  IIRC, the patch that Mauro sent to -mm 
> might depend on this dvb patch.  If so, let me know, and I'll send it to 
> akpm.
> 
> Unless, of course, you think it has the potential to break other 
> boards.  In which case, let Mauro know and he can have akpm drop it from 
> -mm until it can get some more testing in cvs.
> 
> -Mike
> 

Hi, Mike

Just to clarify this: i don't blame Mauro, i was just surprised about
the fast 2nd update...

There are 3 recent patches from me:
The first one fixes the SEMAC-L problem with the tda8290. I am sure this
one is correct.

The other 2 patches plus the tda1004x update are dvb-t only. They will
not break compatibility with known cards and compile with the old and
the new version of the tda1004x driver.

Regarding the tda1004x update: Please let me check whether the recently
added boards work with the old version of the tda1004x (need to solve a
very specific issue). If this is the case, i will recommend not to
update this module right now but leave more time for compatibility
checks. I will let you know the next days.

I hope this clarifies things.

Best regards
    Hartmut