I2C driver directory

[Luc Verhaegen]

On Wed, Sep 29, 2004 at 09:24:24PM -0400, Vladimir Dergachev wrote:
> 
> Well, actually I had a much smaller ambition - GATOS code includes drivers
> for tuners, bt828 chips, sound processors and demodulators.
> 
> These have always been built as separate modules, but I kept them in the 
> same directory as main ATI driver - simpler for everyone when we were 
> working in separate tree.
> 
> This makes no sense when we are merging code. I believe all these 
> I2C drivers can be easily used by other drivers if they need to, at worst 
> with minimal changes. I would certainly be glad if other developers found 
> them useful.
> 
> As for tv/tmds/lvds encoders - my first priority is to get GATOS code in 
> so people can use their cards with newer versions of X.Org.
> 
> Yes, separate modules for each chip.
> 
> No idea - really !
> 
> Since we are talking about this - does via/unichrome support VIP bus ? Do 
> you know of any cards that have other chips hanging off it ?
> 
>                         best
> 
>                           Vladimir Dergachev
I'm not really acquainted with VIP, VESA is still gaurding this 
apparently largely abandoned interface. Wouldn't it just be an 8bit 
databus with clocking and sync and probably i2c?

About other graphics cards:

I scavenged a voodoo 3000, it has a Bt869 tv encoder.

The EPIA V class boards have a VT1621 tv encoder attached to a trident.

There's VT1622, VT1622A, VT1623, CH7005, CH7019B tv/lvds encoders 
attached to unichromes (support was in the driver for FS454, SAA711x tv 
encoders). Alan Cox cleaned up VIA code for SAA711x tv tuners on the 
unichrome.

The nvidia cores on X-box have had several tv-encoders attached, 
conexant, FS454, ...

There are probably many more examples like this.

It will take a bit before i'm able to have trident and unichrome use the 
same VT162x code though (i'd need to get me an epia V class board for 
one :p) so it might be a good idea to go ahead with the gatos devices 
and gradually move to something more "ambitious" and extend it to other 
drivers. 

About the maven, there will probably be other i2c addressable devices 
which are highly driver specific.

In any case, if hardware designers are able to work modularly with these 
devices, driver developers should be able to do so too :)

Luc Verhaegen.