On a lengthy find, if no output is found in the first five seconds, libxffmtubo is considering the find terminated but it is still running in the background. Apparently TuboWait(gpointer fork_object) is doing the wait on the wrong child.
The bug is in waitpid() which erroneously returns a program termination status. Seems like an operating system race condition, which occurs when the process is being stopped. Putting in a kill(pid,SIGCONT) apparently works to ensure the process is not in stopped state when the waitpid() is performed. This affects FreeBSD 5.1 and might not affect other systems, but the workaround will keep it safe all around.