Re: Keyboard handling on non-x86

On Thu, May 19, 2005 at 08:11:25PM -0400, Michael wrote:
> And lastly - why does the wscons protocol option require a device name?
> Defaulting it to /dev/wskbd should do the trick.

I am not sure here. In wscons, there are two fundamentally different
ways to access keyboards in the "we have more than one" case:

 /dev/wskbd   is a multiplex device, which mixes all keyboards assigned to
              the console; so you can plug in a USB keyboard and use that
              as console keyboard instead of the previosly attached
	      PS/2 keyboard.

 /dev/wskbd0 ...
 /dev/wskbdN  are "single" keyboards, not multiplexed together.

For the X server, you should be able to configure multiple heads and
multiple keyboards, each of which would talk "protocol wscons" but use
a different /dev/wskbdX.

I talked to Michael about this before, and he thinks the kernel does not
get this right for raw scancode mode, but I would call that a bug.
(And neither me nor him have actually tested such a setup, AFAIK)

I think the XFree server would not get this right, since there only is
a single xf86Info instance - but I might be missing something.

> Would it be better to write an input driver module for wskbd that uses
> wscons' hardware-independent scancodes instead? Seeing the mess that the
> current code is I'd almost think so.

I'd second this - and the #ifdef linux|freebsd stuff in xf86PosteKbdEvent
should be transformed too - though I'm not sure anyone ever tested it.

Martin