Thanks for the test. It implies that this is no regression at least ;) Meanwhile I found the missing piece. Actually it's not one but more pieces. First off, other distros seem to have inputattach program that supports Wacom w8001 mode. This is included in linuxconsole (on openSUSE, it's linuxconsoletools package). But our package removes inputattach program because the same named binary is provided by gpm -- and it's a much older one that doesn't contain Wacom support. Then, the inputattach program should have been called by a systemd wacom-inputattach@.service. This is already found in the existing 65-wacom.rules udev rules. However, our package misses this unit file, hence nothing happens. (I'm not sure whether xf86-input-wacom or libwacom should provide it, though.) With those, the wacom_w8001 gets bound via serio on the given tty, and it provides a proper input device... I suppose. Some information found at: http://who-t.blogspot.com/2013/12/wacom-serial-devices-and-inputattach-in.html And my previous udev rules must be superfluous once when this works.