Re: [PATCH] Change behaviour of psmouse-base.c under error conditions.

[Dmitry Torokhov <dtor_core@ameritech.net>]

On Monday 05 September 2005 16:55, Vojtech Pavlik wrote:
> On Mon, Sep 05, 2005 at 04:12:13PM -0500, Dmitry Torokhov wrote:
> > On Monday 05 September 2005 15:58, Bryan O'Donoghue wrote:
> > > 
> > > However, the KVM in question invariably ends up sending a packet like this
> > > 
> > > packet= 0x0 0xff 0x2 0x8
> > > 
> > > Which is completely invalid for PS/2 and IMPS/2, the specification of
> > > PS/2 defines bit 4 in byte 0 as always being 1.
> 
> Just for the record: Bit 4 was supposed to indicate an external (=1)
> mouse, while internal mice would have a '0' there. This then would allow
> to differentiate between two devices on the same bus.
> 
> I've even seen that in action on one very old laptop.
> 
> Of course, extended protocols break that completely.

IBM Trackpoint controllers can force that bit to 1 for internal device
and to 0 for external. If we ever implement pass-through port for
trackpoints we might use it do demultiplex data streams.

-- 
Dmitry