[linux-dvb] [PATCH] [RFC] remove static dependencies on dvb-pll

Michael Krufky mkrufky at linuxtv.org
Sat Jun 2 20:15:34 CEST 2007


Uwe Bugla wrote:
> Am Samstag, 2. Juni 2007 18:56 schrieben Sie:
>   
> However, the output applying this against 2.6.22-rc3 causes other trouble:
I told you that it wouldn't apply against 2.6.22-rc3.  The patch was
generated against the mercurial repository.

> OKIDOK! If you can promise me and others that this is your "final" version I 
> am inclined to do the following now:
>   
This is _not_ the final version of the patch.  There is still one caller
of dvb_pll_configure (dvb-usb-i2c), and that must be removed before I
can remove the dvb_pll_configure export, and store the pointer to the
dvb_pll_desc inside the priv structure, instead of storing the id.

Trent also asked me to convert the #define's into an enum... I think
that's a good idea, but I have reason not to do that yet.  (some other
patches in my sandbox here depend on those being integer -- I havent yet
decided whether or not to share those patches yet)

> 1. Adjust your work for being applied against 2.6.22-rc3
>   
^^^ Please don't.  This patch is _not_ intended for 2.6.22-rc3.  I have
some other cleanups to do, and I only shared this patch with you because
of your complaining on the mailing list.  I have shown you this patch,
because I nacked your other patch, which made you upset.  You wanted me
to provide an alternate solution.  Showing you this patch was my
"proof-of-concept", and this was my way to show you that I have a better
solution, and it will be applied to the repository, when the time comes,
provided that the other dvb developers are OK with it.

> 2. Test your work
>   

By all means, please feel free to test it against the mercurial repository.
> 3. Send it in again in a corrected version (without offsets, rejects and fuzz 
> factors plus the pre-condition of having applied Trent's work for dst 
> customization before applying your stuff).
>   
^^^ Please don't.  This is my work, and I will see it through, myself.
> You've offered lots of help now for which I am very thankful.
> So please have some patience and let me do the rest, OK?
>   
Apply it to your own tree and do whatever testing you like.  This patch
is _not_ for mainline... at least not yet.  As I have explained above,
when I am ready to include this patch, I will ask the relevant
maintainers for their acks, and I will have it pushed into the master
repository, where it can receive sufficient testing before the 2.6.23
merge window opens.

I feel strongly against rushing in a patch without being sure that all
approve of it.  Since this patch touches functionality that many
developers have had their hands in, it is only fair to give them all
sufficient time to look it over.
> There is no discussion that your work (plus Trent's inputs of course) is far 
> more better than mine or Markus's.
>
>   
>> The patch was generated against v4l-dvb.hg ... I can't promise that it
>> will apply cleanly against 2.6.22-rc3, although I see no reason why
>> not...  You may run into a hunk failure when it comes to dvb_pll_desc
>> *dvb_pll_opera1.
>>
>> Cheers,
>>
>> Mike
>>     
>
> Best Regards and Thanks
>
> Uwe
>
> P. S.: I still do not understand the essence of this pll library as far as
> - performance
> - stability
> - other functionalities
>   
To make a long story short, the devices supported by the dvb-pll library
are similar enough that, through some abstraction, a single module
library is able to support a wide variety of devices.  The performance
if the driver leaves nothing to be desired.  It is stable, and functions
as well as it needs to.

The only downside to this module is that it forces all supported devices
to fit into a certain criteria, and some (many) devices are too complex
to be supported by dvb-pll.  When such a device shows up, we either
extend the functionality of dvb-pll to be able to support the additional
feature, or we create a new, separate module to handle the new device.
> are concerned. Would be really keen on understanding the theoretical 
> background / use of this Gerd-Knorr-relic.
>   
Please drop this now.  I do not care to get into this discussion.  There
is no reason to point more fingers at any more people.

-Mike



More information about the linux-dvb mailing list