[linux-dvb] Re: dib3000mb questions

[Andreas Oberritter <obi@linuxtv.org>]

(thanks for adding a subject line ;))

On Fri, 2004-10-08 at 09:42, Patrick Boettcher wrote:
> On Thu, 7 Oct 2004, Andreas Oberritter wrote:
> > 1.) Inversion
> > Is this fall through for INVERSION_AUTO intended?
> 
> No. When having INVERSION_AUTO, it is good to skip the
> DIB3000MB_REG_DDS_INV at all. I.e. you don't need to set any value to the
> reg.

OK.

> > 2.) Hierarchy and Alpha value
> >        switch (ofdm->hierarchy_information) {
> >                 case HIERARCHY_NONE:
> >                         deb_setf("none ");
> >                 case HIERARCHY_1:
> >                         deb_setf("alpha=1\n");
> >                         wr(DIB3000MB_REG_VIT_ALPHA,
> > DIB3000MB_VIT_ALPHA_1);
> >                         break;
> >
> > Is this fall through for HIERARCHY_NONE intended?
> 
> As you can see, the reg is called 'DIB3000MB_REG_VIT_ALPHA', This 'switch'
> only determines and sets the Alpha-Value. The alpha value is 1, when if
> the HIERARCHY is NONE.

OK.

> > If yes, why isn't it set to HIERARCHY_AUTO instead of HIERARCHY_1?
> 
> Setting it to auto would force an auto-search for TPS evertime to user
> uses HIERARCHY_NONE, which takes some time.
> 
> The actual hierarchy setting are processed below this 'switch'.

OK.

> > 3.) dib3000mb_read_status seems to behave quite strangely. Shouldn't it
> > rather look like this?
> 
> I corrected this in the HEAD Branch this week, along with some other
> changes.

I see, will merge it by hand.

> What will happen to the FE_REFACTORING Branch? Will it become the new HEAD
> Branch? Or will the changes be merged back?

Once there has been enough positive (or no more negative) feedback for a
while, FE_REFACTORING will surely find its way back into HEAD.

> > 4.) auto_search failed
> > Returning -EINVAL here seems strange because it doesn't look like a
> > user's fault. What is this supposed to mean?
> 
> tuning failed, if autosearch was not successful.

I will replace it by "return 0;", because this is not an error.

I'd like to remove "msleep(70);" from the bottom of
dib3000mb_set_frontend(). What is it good for? The ioctl should IMHO
return immediately instead of sleeping.

Regards,
Andreas