[linux-dvb] idea on how to break the static dependencies on demodulator modules

Andrew de Quincey adq_dvb at lidskialf.net
Wed Apr 26 23:49:40 CEST 2006


On Wednesday 26 April 2006 22:38, Johannes Stezenbach wrote:
> Hi Andrew,
>
> On Tue, Apr 25, 2006, Andrew de Quincey wrote:
> > Hi, a tree with these changes is now available at:
> > http://linuxtv.org/hg/~quincy/v4l-dvb-attach
> >
> > This has had the various feedbacks from various people applied - let me
> > know what you think.
>
> -  you have five changesets, however it is really just one
>    lump of code as the changes are not logically seperate,
>    and additionally in the wrong order.
>
>    If Mauro would forward these as-is to
>    Linus it would break bisecting for some people :-(
>    The kernel must compile (and should work) after each
>    changeset applied.

Oh right! I didn't know that: I was trying to break it up for easier review. I 
did not realise people would want to compile the kernel for each changeset. I 
can reorder the changes so they will compile - or would you prefer just a 
single huge diff?

> -  adding symbol_put() to the _release() functions means
>    the drivers cannot be used without dvb_attach() anymore
>    (not a problem for me, but I just wanted to mention it)

Ah yeah - I've been trying to think of a way to fix that. One of the reasons 
why this is still under review.

> - fe->ops->write() should be called fe->ops->pll_write(), no?

Hmm, no actually. The write pointer is meant to be a general purpose way for 
writing "stuff" to the demod; you're not meant to use it directly. The inline 
functions wrap it up into easy to use calls. Not all the calls write to the 
pll - some of them (e.g. tda1004x) use it to write to the demod).

> - please never ever write
>      if (foo) bar;
>   always use
>      if (foo)
> 	     bar;

Sure.



More information about the linux-dvb mailing list