> It sounds suspiciously similar to PR10347 though. Thanks guys. I've confirmed that it is the same problem by comparing the backtraces. I have update that bug with the analysis from Paul. Marc