Guest Additions may be giving invalid information about pointer device -> duplicate of #13968

[Aren Cambre]

It appears that XIQueryDevice() or something else influenced by Guest Additions may be returning invalid information on pointers. Per ​https://code.google.com/p/chromium/issues/detail?id=456222, Chromium is having a problem when running as a Debian or derivative guest OS, possibly because it is being told that the mouse is a touch device.

This is making Chromium difficult to use without a workaround.

[Michael Thayer]

Solving this might be easiest if the relevant VirtualBox, Chromium and X.Org developers all put their heads together. I could be a candidate for VirtualBox. The basic problem is that input devices (in this case our mouse integration device and our USB tablet emulation) report certain capabilities, and Linux and X.Org decide, based on those capabilities and some interesting heuristics, what sort of device it is.

It might (I'm not sure) be helpful if Linux and X.Org could be persuaded to recognise our devices as graphics tablets and not as touchscreens. Our first concern though is that the devices work as expected by the user in as many situations as possible. We also have the problem, especially for the USB tablet, that older releases of X.Org which can no longer be changed by the developers also need to recognise and handle the devices correctly.

In any case, please ask a relevant Chromium developer to comment here so that we can start a discussion. I will see if I can find someone from X.Org.

[Aren Cambre]

Given how much diagnosis is already on the code.google.com ticket, would it make sense to focus this discussion over there?

[Michael Thayer]

Actually this is a duplicate of ticket #13968. The Chromium people seem to be treating this as their bug, which probably makes sense if it is common to several virtualisation platforms.

[Aren Cambre]

This is not duplicate. See comment 142 on ​https://code.google.com/p/chromium/issues/detail?id=456222. There is a bug in how VirtualBox Guest Additions is reporting pointer type. While the Chromium folks found a workaround, the bug in Guest Additions is still there.

[Michael Thayer]

I checked with the X.Org input stack maintainer, who was not able to see a problem with the way the Guest Additions are reporting themselves: the output of "xinput list --long" did not contain an XITouchClass entry. His guess was that it was actually Chromium deciding based on its own heuristics that this is a touch device.

[Aren Cambre]

This problem seems to have gone away.

[Frank Mehnert]

Thanks for the information!