[linux-dvb] Thoughts about supporting more complex tuners

Andrew de Quincey adq_dvb at lidskialf.net
Fri Apr 14 16:03:48 CEST 2006

On Friday 14 April 2006 14:47, treza at chez.com wrote:
> Maybe that remark is off topic but...
> The following issues shall be addressed in the new tuner api :
> - There must be a way to try several tuners detections as
> one single USB IDs ( or PCI ? ) can correspond to several
> devices with different tuners.
> - Some tuners needs additional calibration data stored in
> EEPROM, there is a chicken-and-egg problem in the current
> implementation between the detection of the tuner and
> getting additional parameters for the tuner calibration (
> and the EEPROM is not always on the same I2C sub-bus as the
> Tuner chip )

BTW: Just to be 100% clear with everyone by "tuner" I mean the analogue signal 
reception bit of a frontend. "demod" is obvious. And "frontend" is the entire 
signals part of a card (tuner+demodulator+any other bits like diseqc chips).

In the new API, there will be XXX_attach() methods for the tuners like there 
are for the demodulators.

So for your first point. We support this for demodulators already - in 
budget-av.c several demodulator XXX_attach() methods are called in turn for 
some cards since the same PCI IDs sometimes have different demodulators as 
you say. With the new API, the same can now be done for tuners as well.

For the second point, again with the XXX_attach() methods, any necessary 
configuration data can be read by the card driver and passed in as parameters 
to the tuner specific XXX_attach() method.

Do these meet your concerns? Please let me know!

More information about the linux-dvb mailing list