The problem is that {IPC_ENGINE}.close_ecdbgd sleeps for only 1 millisecond
(1_000_000 nanoseconds) for up to `ise_ending_timeout' times before exiting
the loop and checking whether the daemon is dead.  Perhaps the intention
was for it to sleep for 1 second each time around the loop, instead of only
1 millisecond (though 1 second seems like a long time).  Currently, it will 
only wait a total of 100 milliseconds.  Or maybe `ise_ending_timeout' defaults
to a value that is too small.