Operations On Processes, Creation and Exit

Objectives

1. Understand basics of process creation and destruction

Notes

• In linux each process has a unique process id.
• Fork will create a new process
  • There are other mechanisms as well, but they are not important here.
  • vfork, clone
• The fork command creates an identical copy of the process
  • copy of the stack, heap, text, data, ... memory segments
  • Copy of the registers (ie both will start executing at the same point)
• But they will not:
  • Have the same process id
  • have the same return value from fork
    • The old (parent) process will get the new (child's) process id
    • The child will get a 0.
  • The child will have parent process id of the parent, not the parent's parent.
  • Some IPC mechanisms will not be inherited.
  • Utilization stats will be set to 0 in the child.
  • See man 2 fork.
• The parent can choose:
  • Continue to execute: a parallel program for example
  • Wait for the child: the shell without job control.
• Linux/unix implements a copy on write for memory
  • Many calls to fork are immediately followed by a exec
  • So memory duplication is not needed.
• Calls to exit will terminate the process
  • This is the same as a call to return.
  • There are other mechanisms as well, but they are not important here.
  • Most resources are freed.
    • In linux a small amount of data is retained until the parent acknowledges the exit. (zombie)
  • In *nux the parent is sent a signal that a child process has exited (SIGCHILD)
    • The parent can receive this signal with wait
    • And get the exit status of the child.
    • If the parent is gone, systemd gets the message.
      • Or some other suitable process
  • A process without a parent is called an orphan.