The bug popped up when one UE was doing TCP UL iperf and a second UE
attempted to connect. A crash of the eNB happened because we changed
the wrong NFAPI structure in a TTI where both UEs are scheduled.

Something like that.