[linux-dvb] Re: ALPS tdmb7 conflict with cx22702

[Holger Waechtler <holger@qanu.de>]

wolfgang@leila.ping.de wrote:

> Hi,
> IIRC the TDMD7 uses the TDA10046 as a demodulator.

We were talking about the TDMB7.

> Wouldn´t it be better to use a proper check for the chip which 
> responds to the I2C access instead of binding a frontend to some 
> specific card?

Not really, since only 0x7f ( = 127) i2c adresses are available and many 
chips can get configured to several i2c addresses this approach only 
causes problems. Most current frontend drivers contain only about 20-30% 
shared code, the remaining are the different setup tables and 
complicated probing functions for all flavors of possible card/frontend 
combinations.

All this autoprobing stuff has never been a good idea and was the root 
of quite a lot error-prone code in the past. I would suggest for all new 
drivers to make the frontend code a "library" which has nothing to do 
with the kernel i2c code and neither relies on the dvb_i2c code.

Then the card driver (which is the only instance that knows about the 
hardware anyway) would call the library function by simply passing it's 
i2c_xfer function pointer and maybe pointers to the card-specific 
register initialisation tables. The frontend driver would then only 
contain the code that can really get shared, all drivers could get as 
clean and lean as they were once when they worked for a single card only.

what do you think about this approach?

Holger