fork() in a for loop

[daniel fitz]

I'm writing a program to stress test the scheduler on a unix distribution. I'm creating a user specified number of processes using fork() and each process should work out prime numbers up to 1000. I'm using a for loop to create the specified number of processes but my program creates too many. I know I need to make sure only the parent forks but i'm just not sure how. Any guidance you could give me would be a massive help. Thanks

Code:

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>
#include <sys/wait.h>

int prime_calc(int prime_range) {
int prime = 1;
int i;
int p;
for(i = 0; i <= prime_range; i++) {
for(p = 2; p <= prime_range; p++) {
if( i != p && i % p == 0) {
prime = 0;
break;
}
}
if(prime == 1) {
printf("Prime: %d\n", i);
}
prime = 1;
}
return 0;
}

int main() {
int prime_range = 1000;
int i;
int pid;
int pid1;
for(i = 0; i < 3; i++) {
pid1 = fork();
}
if(pid1==-1) {
perror("bad fork");
exit(0);
}
if(pid1==0) {
pid = getpid();
printf("The process id is: %d\n", pid);
prime_calc(prim e_range);
}
else {
wait(0);
}
return 0;
}

[donbock]

You want a parent process that forks off some number of child processes. Each child process works out the prime numbers up to 1000. You don't say, but I propose that the parent process has nothing else to do after it gets all of the children going.

This is going to sound like heresy, but I think this is a job for the ignoble goto.

for (i=0; i<nchildren; i++) {
   pid = fork();
   if (pid == 0) goto child_only;
   ...
   }
...
return;

child_only:
...

On another topic, are you certain this will produce the expected load on the operating system? What if the earliest [oldest] children finish their prime number task before the parent finishes forking off their younger siblings?

[johny10151981]

I didn't do much programming using linux and fork,

But I guess you should not use fork in loop its point less,

You can call fork once and then call your prim number counting functions in thread