Under which circumstances a child process creates another child process using fork?

mitsulas · 12-07-2009, 02:18 PM

Hello,

Consider the following code:

Code:

int main()
{
	int i=0;
	pid_t pid;
	for(i=0;i<2;i++)
	{
		pid=fork();
		if(pid==0){
			printf("This is child %i i=%i\n",getpid(),i);
		}
		else{
			printf("Parent: chid_pid=%i i=%i parent's pid=%i\n",pid,i,getpid());

		}

	}
	
	exit(0);
}

I get the following output:

Parent: chid_pid=4356 i=0 parent's pid=4355
This is child 4356 i=0
This is child 4357 i=1
This is child 4358 i=1
Parent: chid_pid=4357 i=1 parent's pid=4355
Parent: chid_pid=4358 i=1 parent's pid=4356

As I can observe instead of two children(as I expect) processes there are three. This is because child process 4356 creates its own child. Under which circumstances this happens? Why all the messages of the type
"This is child X i=Y" are concentrated one under another? How exactly fork works? Is affected by the fact that I have a dual-core processor?

Thanks in advance

johnsfine · 12-07-2009, 02:41 PM

Quote:

Originally Posted by mitsulas

As I can observe instead of two children(as I expect) processes there are three.

Why would you expect two?

Quote:

This is because child process 4356 creates its own child.

You even understand why there are three.

Quote:

Under which circumstances this happens?

Everything except the return value of fork() is duplicated between the parent and child, so after the fork() the child will do exactly the same things the parent does except where those things depend on the return value of fork().

You did not make continuing through the loop until i is 2 depend on that return value. (see alternate code below)

Quote:

Why all the messages of the type
"This is child X i=Y" are concentrated one under another?

The processes happened to execute in that sequence.

Quote:

How exactly fork works?

Too general a question to answer.

Quote:

Is affected by the fact that I have a dual-core processor?

The sequence in which the processes execute is probably affected by the fact that you have a dual-core processor.

If you wanted just the two child processes you say you expected, notice where I added break to your code in the version below. That means the child does not stay in the loop so it does not create its own child.

Code:

int main()
{
	int i=0;
	pid_t pid;
	for(i=0;i<2;i++)
	{
		pid=fork();
		if(pid==0){
			printf("This is child %i i=%i\n",getpid(),i);
                        break;
		}
		else{
			printf("Parent: chid_pid=%i i=%i parent's pid=%i\n",pid,i,getpid());

		}

	}
	
	exit(0);
}

mitsulas · 12-08-2009, 02:37 AM

Thank you for you answer.

Quote:

Originally Posted by johnsfine

Everything except the return value of fork() is duplicated between the parent and child, so after the fork() the child will do exactly the same things the parent does except where those things depend on the return value of fork().

Why the other child process doesn't make its own child?

johnsfine · 12-08-2009, 08:16 AM

Quote:

Originally Posted by mitsulas

Why the other child process doesn't make its own child?

Everything except the return value of fork() is duplicated

Are you ignoring the fact that the variable i in your program is part of that "everything"? Or are you ignoring the way the value of i affects the subsequent behavior?

As you coded the program, when you create a child in which i is 0, each of the child and the parent will create another child (and in both those children, i will be 1). When you create a child in which i is 1, neither the parent nor the child will create another child.

As you coded it, whether a process goes on to create a child does not depend on whether that process is itself the original or a child. It depends only on the value of i.