[Gpm] Synaptics touchpad

[Chris Collins]

Hello Doug,

> The synps2 module 
> ... attached via a PS/2 style auxiliary port
> ... attached to the serial port.  

I am a bit thick.  Which of these two describes the touchpad
on a compaq presario 1685 laptop?

> One of the things we did to help reduce the 'jumpy mouse problem' was
> to add the following smoothing logic inside the mouse handler (gpm & kernel 
> driver):
> 
>     Once a packet has been received from the touch pad
[snip]
>       - Assemble PS/2 style packet in kernel driver
>                   or
>         Assemble GPM_Event structure in GPM.

The path of info is then: an (x) byte packet comes from the
touchpad directly to the GPM module.  The modules interprets 
the information, and returns it to GPM(again) or the kernel
where it is used for screen actions. Yes?

The GPM part is confusing.  What recieves the packet in GPM
and where does it send it to in GPM.  Sending it to the
kernel makes sense but back to GPM confuses me.


>  From the kernel driver stand point.  The pc_keyb.c driver is not a kludge.
> It's just directed towards a generic PS/2 style mouse running in relative
> mode.  For the project referenced by Dr. Shub, I had to patch pc_keyb.c
> to allow an alternative mouse driver to be used instead of the default
> driver in pc_keyb.c (haven't submitted the diffs to Alan Cox yet, however).
> Assuming you wish to write a kernel driver, you'll want my pc_keyb.c mods
> and most definitely the documentation from Synaptic's site.


So let me try to get this straight. A complete solution would be to have
touchpad movement be picked up by a running gpm module (gpmd?).
It sends this massaged info to a (yet to be revised) pc_keyb.c which
correctly interperates the info and makes mousey  movements on the
screen.

If I have this flow of info correct, how can I help test it?
Setting up the latest GPM is involved :)  also testing the
patched kernel vs. non-patched?

Has all this been succesfully test already?


Is there a nice overview of the way simple mouse movements
are read and used through the kernel which could help me
at least understand the problem better to assist in its
repair.

--Chris