[gpm]Touchpads

[Alessandro Rubini]

> I would suggest that every driver have an additional functioncall
> that reset that device to normal state (if it has one), which is
> called when gpm stops.

That's a good idea. And since the mouse type is becoming complex, I'd
also switch to tagged-initialization for the data structure instead of
listing fields as it is done now. Since gpm is only compiled with gcc
I don't see anything bad in that change. But again, I'll let the
maintainer speak.

> Secondly how would I tell gpm that there are two different
> mice/protocols on the same device?

It's not supported. To have that, you'd need to implement protocol
autodetection on each data packet. It's not that hard if you know in
advance what the two protocols are, but it's hard to do it in the
non-device-specific way.

A quick workaround would be a mouse type that does such detection for
those two types, but proliferating mouse types to this aim is bad
practice (in theory you might have a new mouse type for each
ps2+something pair).

Or the standard synps2 driver might accept legacy ps2 packets. If you
implement it that way I'm sure it will be accepted in the standard ps2
decoder.

/alessandro