[gpm]Bug with IMPS/2 w/ Microsoft Intellimouse 1.3A on Linux

[Pozsar Balazs]

On Mon, 17 Jun 2002, Peter Berg Larsen wrote:

> > But... one question: Isn't the mouse supposed to 'shut up', after
> > receiving a 0xFF, until it is initialized? (ie. a 0xF4 is sent to it).
>
> No. You are only ensured that you (the host) will not reacive any other
> data before the client is ready. And it signals this with AA 00, if the
> client is a mouse. There is ofcause some mice that don't stick to the
> specifications and sends the AA 00 at once. Thats another reason why I
> suggested a disable stream first.
>
> While you at imps2 couldn't you rewrite it all to automaticly find the
> best imps2 mode the mouse knows? 5 button wheel is missing (which I have
> just bought:) Here is a link about imps2 and how to detect the features of
> present imps2 mice:
>
> http://www.microsoft.com/hwdev/tech/input/mcompat.asp

Checking the X sources it seems that 5-button "imps2" protocol is what
they call "ExplorerPS/2", and the 3-button "imps2" is their "IMPS/2".

All we have to do here is to create a seperate M_exps2, as the M_imps2 is
currently used for the "ExplorerPS/2" protocol, and that's why gpm doesn't
interpret button 4 & 5.

This would consistent with X, but it won't provide auto-detection.

I think though that this is best solution, and detection should be more
global.

I'll send a patch for M_exps2 in a few days :)
And while we are at it, R_exps2 should be implemented too, so you can use
gpm in repeater mode for X, enjoying all five buttons :)

-- 
pozsy