Also tried with big mutex enforcement (via qmutexlocker covering every procedure that touched the data) and found that a) the performance isn't affected that bad as I was remembering; b) the segfault is still here. So, apparently not a concurrence problem. -- Andrea