There is support in the spec for PS/2 guests believe it or not, it's
just mostly undocumented. Coincidentally, I was actually the one who
wrote the drivers for this in Benjamin's tree. Using that, we can
actually just create a serio port that goes through the rmi4 driver and
communicate with that using the normal TrackPoint driver. We have a
working implementation of this here: 
https://github.com/bentiss/linux/tree/synaptics-rmi4-smbus-v4.2-rc6%2B

Cheers,
	Stephen Chandler Paul

On Wed, 2015-08-19 at 14:27 -0700, Dmitry Torokhov wrote:

On Wed, Aug 19, 2015 at 09:52:00AM -0400, Benjamin Tissoires wrote:

quoted

On Tue, Aug 18, 2015 at 3:06 PM, Benjamin Tissoires
[off-list ref] wrote:

quoted

Hi Nick,

thanks for the report.

On Mon, Aug 17, 2015 at 10:31 PM, Nick Bowler <nbowler@draconx.ca

quoted

wrote:
Hi,

I'm not sure if this is actually a Linux issue but figured I'd
at least
report it here to start...

I have a Lenovo Thinkpad X250, with the newfangled trackpoint
buttons.
I have a problem with missing button press or release events:
sometimes
pressing a button has no effect, and sometimes releasing a
button has no
effect (the latter is especially annoying, as the button
remains depressed
as far as any applications are concerned).

After some testing, the problem apparently depends on the
position of my
hands over the touchpad.  I can reliably reproduce it as
follows: place
two fingers on the touchpad, then press the button repeatedly. 
 Watching
with evtest, several events (could be either press or release)
will
simply be missin, although the kernel never does anything weird
like
send two release events in a row.  There are no problems if
there are
0 or 1 fingers on the touchpad.

My current kernel version is 4.1.6, although the problem occurs
in all
versions that I tried.

Here is the evtest result from placing two fingers on the
touchpad, then
pressing the left trackpoint button 10 times, counting 1 second
between
each press.  As you can see, only 3 presses and 3 release
events total
were sent by the kernel, sometimes with many physical button
presses
between the press and its corresponding release:

OK, so this is definitively weird. My first idea would be a
firmware
problem. I have asked Chandler to reproduce it on his t450 and
see if
we observe it on our laptops too.

Update on this one : Chandler reproduced it on the t450, so it's
likely that all of these sensors are affected. It is still unclear
if
it is a firmware bug or a driver problem, but the chances are huge
that this is a firmware bug. Anyway, that's one more reason to push
towards RMI4 over SMBus for these sensors: the bug is not present
with
this protocol :)

By the way, how are we going to handle Trackpoint (PS/2 device) with
touchpads in RMI4 mode? I do not recall anything in RMI4 spec, but I
looked at it quite some time ago.

Thanks.