Bug 1281881

Please attach your Xorg.log and the output of libwacom-list-local-devices. Thanks

Created attachment 1094719 [details]
Xorg.log

$ libwacom-list-local-devices

[Device]
Name=Wacom Serial Tablet WACf004
DeviceMatch=serial:0000:0000;
Class=ISDV4
Width=0
Height=0
IntegratedIn=Display;System;
Styli=0xfffff;0xffffe;

[Features]
Reversible=false
Stylus=true
Ring=false
Ring2=false
Touch=false
TouchSwitch=false
StatusLEDs=
NumStrips=0
Buttons=0
---------------------------------------------------------------
[Device]
Name=Wacom Serial Tablet WACf004
DeviceMatch=serial:0000:0000;
Class=ISDV4
Width=0
Height=0
IntegratedIn=Display;System;
Styli=0xfffff;0xffffe;

[Features]
Reversible=false
Stylus=true
Ring=false
Ring2=false
Touch=false
TouchSwitch=false
StatusLEDs=
NumStrips=0
Buttons=0
---------------------------------------------------------------

What's the output of "systemctl status wacom-inputattach" on your machine?

Also, please run evemu-record against the "Wacom Serial Penabled 2FG Touchscreen" device (/dev/input/event14 in the previous log) and touch the screen once. Then attach the output here.

managed to reproduce this here. The problem is that the device is both a touchscreen and a tablet device at the same time. libinput does not yet support tablet devices and xorg-x11-drv-libinput usually skips tablets for that reason. In this case libinput does get assigned because it has both the ID_INPUT_TOUCHSCREEN and the ID_INPUT_TABLET udev property.

That's why the touchscreen portion of it works but the stylus part of it doesn't.

I don't have a good solution to this yet, we're lacking the ability to negate xorg.conf matches, so it's extremely hard to say "Don't match on this". Until I can fix this, you'll have to use a local workaround:

$ cat /etc/X11/xorg.conf.d/99-wacom-serial-override.conf
Section "InputClass"
  Identifier "override serial wacom driver"
  MatchProduct "Wacom Serial Penabled 2FG Touchscreen"
  Driver "wacom
EndSection

Yeah, the wacom-inputattach was loaded and active without errors.

FWIW, I've seen this on Wayland too, but I suppose that's precisely because libinput doesn't support it... or would there be Wayland-specific differences here?

Meanwhile, great to hear you managed to reproduce it. So AFAIU you don't need further info from my end?

no, I've got all I need thanks. Wayland has the same issue because we use the same input stack, i.e. libinput. There are no specific differences that apply here.

I was the poster with the problem on the Fedora help page, mentioned in the Description.

For one session, the tablet started working again, but now it again refuses to work. I don't know if that's significant, but I thought I'd throw it in. I don't know how to reproduce it working again, since if I did, I would do that every time I started the computer. Also, I'm pretty new to Linux.

I just checked the local workaround mentioned in Comment 5, and it works.

For the archives: long-term this actually needs a kernel fix. Having the two different devices off the same event node is not ideal and not how other similar devices work. So expect this to get fixed in the kernel, libinput will then just work.

The fix is in Dmitry's input tree now.

http://git.kernel.org/cgit/linux/kernel/git/dtor/input.git/commit/?id=e0361b70175f0cd6199dd9ed6679632de73973d4

Closing as NEXTRELEASE. If kernel 4.5 ends up in f23 then it'll work there too.

Workaround is outlined in Comment 5.