Comment 222 for bug 41301

I have this bug on FreeBSD 8.2, running the latest X server from ports (xorg-server-1.7.7_1,1 and KDE 4.6.5).

Specifically, what I am experiencing closely matches comment #61 which Farzad Battiwalla wrote on 2010-08-02 in https://bugs.launchpad.net/ubuntu/+source/xorg-server/+bug/362359 . In fact, I learned a lot from his description, and I believe this should enable a knowledgeable person to isolate the error.

A minor addition to his description:

- Once you left-click in a window which has no context menu on a right-click, you are basically stuck. This is for example true for the pin/unpin button in the window border - if you left-click this once, all further mouse events are directed to that button. Even if the mouse can be moved into another window, the balloon help for the pin/unpin button will pop up there once the mouse stops moving.

- I can still switch between windows on the current desktop using ALT-TAB. That is to say, the keyboard input will be focused to that window, but not the mouse buttons if the mouse is stuck as described above.

- I could get out of the stuck pin/unpin button by using KDE's CTL-ALT-ESC window kill function: Press CTL-ALT-ESC and then RETURN in the window whose pin/unpin button receives the mouse events, and one can then do a right-click hopefully in another window (e.g., the background) with a context menu, which then enables to move the mouse focus to another window by once more right-clicking there (for example the desktop panner of the KDE panel).

- In summary, it is possible to redirect the mouse input to a new window if one can manage to bring up a context menu by right-clicking in the window which currently is stuck receiving mouse events, and then right-clicking in the window which one desires to become stuck next. The latter then better have a right-click context menu as well...

As I said, Farzad's description is the most accurate one and should enable correction of the problem.

I have the problem now for several hours because I am doing a long-running compilation which I do not want to interrupt, and with these instructions I (barely) manage to control my desktop (and write this report).

One last comment: I seem to remember (from earlier experiences) that it does not suffice to just restart the X server. Instead, I have to reboot to get rid of the problem for a longer time. Which to me seems to indicate that some hardware register got mangled, or that the X server is accessing some fixed address it shouldn't (maybe usage of already freed stack space?), or whatever. But take this with a grain of salt as I am not sure of this observation.