When wait system call is used

Question

The theory says that, if wait is not called parent wont be getting information about terminated child and child becomes zombie. But when we create a process, zombies are not created even if we are not calling wait. My question is whether the wait is called automatically?

Answer 1

In many languages, calling a sub process will call wait() for you. For example, in ruby or perl, you often shell out like this:

#!/usr/bin/ruby
system("ls /tmp")
`ls /tmp`

This is doing a bunch of magic for you, including calling wait() . In fact, Ruby must wait for the process to exit anyway to collect the output before the program can continue.

You can easily create zombies like this:

#!/usr/bin/ruby
if fork
  sleep 1000 # Parent ignoring the child
else
  exec "ls /tmp" # short-lived child
end

When we manually fork/exec, there is no magic calling wait() for us, and a zombie will be created. But when the parent exits, the zombie child will get re-parented to init , which will always call wait() to clean up zombies.