Comment 18 for bug 737792

From the analysis of the reports (this bug and duplicates), it seems that the problem occurs on high-end GPUs and creates a race condition triggering an exception that itself crashes Compiz.

The problem doesn't happen on particular GPUs, ie we cannot just blacklist some of them.

The error is not constant: the system eventually starts. One developer was able to reproduce the anomaly while debugging other fixes.

The fact that it can happen on the livecd is of critical importance. However, since:
a/ it does not happen systematically, and
b/ the system eventually starts (via gnome-session respawning it)

it feels to me, that the bug is not a release blocker. So I am assigning it an importance of 'high'. Besides, we're keeping that problem on the 0day-SRU list, but we don't have high hopes of finding a real solution for it.

We are looking at more reports and comments to help design a workaround that could limit the visual impact of the problem.