Re: [PATCH] Revert "Input: trackpoint - add new trackpoint firmware ID"

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

Hi Aaron,

On Tue, Jan 02, 2018 at 09:57:55PM +0800, Aaron Ma wrote:
> No, it is not a regression of this commit.
> 
> ThinkPad X1 Yoga 2nd:
>    trackpoint (ID: 01)
> 
> ThinkPad X1 Yoga 3rd:
>    trackpoint (ID: 03)
> 
> Both laptop's trackpoints have the same behavior.
> Writing "speed" of sysfs is failed.
> 
> Override the ID and force loading drivers/input/mouse/elantech.c,
> it causes too many failure and trackpoint stops work.

Right, because it does not support Elantech *touchpad* protocol, that is
not a surprise.

> 
> The ID of "2.4.18 READ SECONDARY ID (x"E1")" in TrackPoint specification
> does not indicate any other vendors but only trackpoint.

Exactly. If ID does not match, it is not an IBM trackpoint device.

> Elantech uses 0x03e9.
> ALPS uses 0x00e6/0x00e7/0x00ec.
> 
> Maybe the windows tool's is wrong like Linux driver before.

I am not sure what you mean by that.

Anyway, I played with my Carbons a bit, and it seems that the patch
should indeed be reverted. I believe that neither the Elantech nor ALPS
trackpoints support the IBM trackpoint protocol; none of the extended
features (sensitivity, inertia, etc) work when we register them as
TTPS/2 devices. They should continue to be registered as "Generic PS/2"
as that's that they support.

I understand that you want scroll mode working with trackpoints, but
forcing them to pretend that they are TTPS/2 devices is not the proper
way of doing that. Write udev rules that would set
ID_INPUT_POINTINGSTICK property on all input devices connected to a
pass-through serio ports on LENOVO devices, and you should be set (just
make sure you cover both PS/2 pass-through and RMI pass-through
options).

Thanks.

-- 
Dmitry