Prime Sieve with Processes and Pipes #
Goal: Build a pipeline of processes where each filters out multiples of a prime.
v0: Pipe demo - send numbers through a pipe.
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/wait.h>
int
main(int _ac, char* av[])
{
int n = atoi(av[1]);
int pp[2];
pipe(pp);
if (fork()) {
close(pp[0]);
for (int i = 2; i <= n; ++i) {
write(pp[1], &i, sizeof(i));
}
close(pp[1]);
wait(0);
}
else {
close(pp[1]);
int num;
while (read(pp[0], &num, sizeof(num)) > 0) {
printf("%d\n", num);
}
}
return 0;
}
v1: First filter - filter out even numbers.
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/wait.h>
void
filter(int prime, int in_fd)
{
int num;
while (read(in_fd, &num, sizeof(num)) > 0) {
if (num % prime != 0) {
printf("%d\n", num);
}
}
}
int
main(int _ac, char* av[])
{
int n = atoi(av[1]);
int pp[2];
pipe(pp);
if (fork()) {
close(pp[0]);
printf("2\n");
for (int i = 3; i <= n; ++i) {
write(pp[1], &i, sizeof(i));
}
close(pp[1]);
wait(0);
}
else {
close(pp[1]);
filter(2, pp[0]);
}
return 0;
}
v2: Dynamic pipeline - each filter spawns the next on first survivor.
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/wait.h>
void
filter(int prime, int in_fd)
{
int num;
int out_fd = -1;
int pp[2];
while (read(in_fd, &num, sizeof(num)) > 0) {
if (num % prime == 0) continue;
if (out_fd == -1) {
printf("%d\n", num);
pipe(pp);
if (fork()) {
close(pp[0]);
out_fd = pp[1];
}
else {
close(pp[1]);
filter(num, pp[0]);
return;
}
}
else {
write(out_fd, &num, sizeof(num));
}
}
close(in_fd);
if (out_fd != -1) {
close(out_fd);
wait(0);
}
}
int
main(int _ac, char* av[])
{
int n = atoi(av[1]);
int pp[2];
pipe(pp);
if (fork()) {
close(pp[0]);
printf("2\n");
for (int i = 3; i <= n; ++i) {
write(pp[1], &i, sizeof(i));
}
close(pp[1]);
wait(0);
}
else {
close(pp[1]);
filter(2, pp[0]);
}
return 0;
}
Notes:
- Draw process diagram as pipeline forms
- Trace through with n=10
- Why close unused pipe ends?
- Reader needs to get EOF when all writers done
- Prevents file descriptor leaks
- Why wait in each filter?
- Prevents zombie processes
- Pipeline terminates from end back to start
- Compare to sequential sieve - same algorithm, parallel execution
