Mailing List archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[linux-dvb] Re: Proposal for new frontend architecture



Andrew de Quincey wrote:

On Wednesday 22 Sep 2004 16:06, Andrew de Quincey wrote:

On Wednesday 22 Sep 2004 15:41, Gerd Knorr wrote:

This is similar to the proposal that Holger and I suggested

Yep, is somewhat simliar.


- we just took it
a step further and removed the frontend dependencies on kernel i2c
altogether.

No way, I'd like to veto that.

why? please elaborate.

The background of the read/write_demod_reg() function pointers in the card_spec struct is that for almost all demods and tuner ICs the chip adresses are configurable by strapping some config pins high or low.


The advantages of removing kernel i2c from the frontends
altogether weree

* It makes it easier for people to reuse the frontends in places where
kernel i2c is not usable

... and makes it harder for people which use kernel i2c anyway for other
reasons.


- does this actually occur though?

Good question. Before trying to solve a problem you should better
verify there actually is one ;)


* It also removes any trace of autoprobing problems since they're not
kernel i2c_drivers.

.class will do that just fine as well. If the dvb subsystem stops doing
autoprobing itself and .class prevents any non-dvb i2c modules playing
with the i2c bus on dvb cards everything is fine.

OK, this plan sounds good to me.

Oh - I take it this doesn't preclude having the PLL code in the adapters - I did like that. You'd just pass two function calls init_pll() and set_pll() as part of the initialisation structure.

it's useful for other cards, too: usually the PLL/Synthesizer bandswitch and Chargepump settings are completely board-dependent and by no mean portable. Keeping them in the generic frontend code makes not much sense.

Holger





Home | Main Index | Thread Index