> That still leaves the segmentation faults in test t.lock2 on S390. Have > you any thoughts/theories/ideas on what's causing that? Unfortunately, I can't reproduce this on my own. I'll see if the autobuilder fails again.