Comment 2 for bug 1632197

An update on the symptoms...

Two new manifestations of note:

1) After trying briefly to get the display to freeze by dragging windows on the desktop, and believing that I had failed, I tried switching to a virtual terminal, only to find them all dead (blank screen and unresponsive). When I tried to return to the desktop display, I found that it was frozen except for mouse pointer movement.

2) After working on the desktop for several hours with just one incidence of freezing/reboot, I again attempted to switch to a virtual terminal, to find the screen as I had left it earlier (logged in and grepping a log file), but unresponsive to the keyboard. I tried a number of obvious keyboard sequences and random pecking to no avail, then switched back to the desktop display to find that much of that keyboard input somehow made it to an xterm on the desktop. In other words, on switching to the vt, the keyboard input remained with the desktop. This is all looking very much like a problem with the xserver, but then...

Thinking that I would try dragging some windows between workspaces, I hit <alt>+s to bring up the workspace switcher, only to find to my amazement that the workspace grid had undergone a vertical rotation. Is this a bug or a feature? With all the random keyboard pecking that had been erroneously fed to the desktop, had I stumbled upon an undocumented "feature" which we shall call workspace grid rotation? If the answer is "no", then unity has a serious problem. If the answer is "yes", then I think there is probably enough evidence to blame the xserver for the other reported problems.

This is reaching the point that I am needing to back out of the xserver/compiz/unity upgrade to get back to a stable system. I'll back out one package (or meta-package) at a time, test for stability, and keep going until stability is achieved, and will let you know how it goes.