Volker, > BTW, the current discussion for the Linux kernel about a > general async syscall interface, does that also cover fcntl > locks? Just curious. it should do. The key is what mechanism they end up with for event completion. Cheers, Tridge