1.
Allocate memory for the new process control block ( task _ struct
).
If memory is not enough, it returns an error
2.
All values from the ancestral data structure are copied to the
descendant data structure. After that, the
fields whose values should
be different from the original ones will be changed. If the user
exceeds the specified number of processes for him or if the number
of processes in the system exceeds the maximum possible value (
which depends on the amount of available memory), the process
creation stops and an error returns
3.
An identifier ( pid ) is generated for the process using a special
algorithm
to guarantee uniqueness
4.
For the descendant, the necessary additional ancestral data
structures (table of file descriptors, information
about the current
directory, table of signal handlers, etc.) are copied
5.
Form the address space of the process
6.
The process data structure is placed in the list and hash table of
system
processes
7.
The process is put into readiness for execution
Using Fork () In Applications
Consider an example of using a fork () system call to create a process. The
description of fork () according to POS IX looks like this:
# Include < unistd . h >
pid _ t fork (); // type pid _
t is an integer
Since executing fork () causes the control of both the ancestor and the
descendant to switch to an operator, which after calling fork () (both begin
to execute one instruction), practically the only difference between the
ancestor and the descendant in the programmer's view is in the code
return fork () .
For descendant fork () returns zero and for ancestor the id ( pid ) of the
created descendant process. When for some reason the descendant was not
created, fork () will return –1. Therefore, the normal code for and fork ()
looks like this:
pid _ t pid ;
if (( pid = fork ()) == - 1)
{ / * error, crash * /) }
if (pid == 0)
{
// is a descendant
}
else
{
// this is the ancestor
printf (" Descendant started with code % d \ n ", pid );
}
Once the process is created, it can access the identifying information by a
system call to getpid () which returns the current process ID, and getppid (),
which returns the process ID of the ancestor .
#include
pid_t mypid, parent_pid;
mypid = getpid ();
parent_pid = getppid ();
