Duplication of Piped output - c

Question

Can someone please explain to me why my output has duplicates in it from the ls command. The normal operation of ls -l | sort does not give me a duplicated output so what could be the issue?
Essentially i'm trying to pipe the output from one command and enter it into another command. The program works so far, but the output is displaying duplicate data. Plus and explanation of why I would need to do a close after dup2 would be really helpful :)
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/wait.h>
// function declarations
void executeLs(int data_pipe[]);
void executeSort(int data_pipe[]);
int main(){
int data_pipe[2]; // array storing the file descriptors
int childls_pid; // ls child process
int childSort_pid; // sort child process
int rc; // return vaue of the pipe
int child_status1;
int child_status2;
rc = pipe(data_pipe);
if(rc == -1){
perror("pipe");
exit(1);
}
childls_pid = fork();
childSort_pid = fork();
// Ls Child process
switch(childls_pid) {
case -1:
perror("fork childLs Error");
exit(1);
case 0:
// inside of child process
executeLs(data_pipe);
exit(0);
default:
break;
}
// Sort child process
switch(childSort_pid) {
case -1:
perror("fork childSort Error");
exit(1);
case 0:
executeSort(data_pipe);
exit(0);
default:
wait(&child_status2);
}
return 0;
}
void executeLs(int data_pipe[]){
// Closes the read file descriptor
close(data_pipe[0]);
dup2(data_pipe[1], STDOUT_FILENO);
// confused as to why this is necessary
close(data_pipe[1]);
execlp("ls", "ls", "-1", NULL);
}
void executeSort(int data_pipe[]){
// close the write file descriptor
close(data_pipe[1]);
dup2(data_pipe[0], STDIN_FILENO);
close(data_pipe[0]);
execlp("sort","sort", NULL);
}

Answer 1

The reason is that you're forking more processes than you intended. When you do:
childls_pid = fork();
childSort_pid = fork();
you're doing the second fork() in both the original parent process and the process created by the first fork(). So you now have the following process tree:
parent
childls
childSort
childSort
In both childls and childls->childSort, childls_pid is 0, so they both execute the case 0: clause that runs executeLs().
You need to move the second fork into code that only runs in the original parent. You can simply move it to after the first switch statement.
The reason you need to close the pipe FDs is because a pipe isn't really closed until all processes that have it open close it. If the child process has the write end of its pipe open, that will keep it from reading EOF on the pipe.
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/wait.h>
// function declarations
void executeLs(int data_pipe[]);
void executeSort(int data_pipe[]);
int main(){
int data_pipe[2]; // array storing the file descriptors
int childls_pid; // ls child process
int childSort_pid; // sort child process
int rc; // return vaue of the pipe
int child_status1;
int child_status2;
rc = pipe(data_pipe);
if(rc == -1){
perror("pipe");
exit(1);
}
childls_pid = fork();
// Ls Child process
switch(childls_pid) {
case -1:
perror("fork childLs Error");
exit(1);
case 0:
// inside of child process
executeLs(data_pipe);
exit(0);
default:
break;
}
childSort_pid = fork();
// Sort child process
switch(childSort_pid) {
case -1:
perror("fork childSort Error");
exit(1);
case 0:
executeSort(data_pipe);
exit(0);
default:
wait(&child_status2);
}
return 0;
}
void executeLs(int data_pipe[]){
// Closes the read file descriptor
close(data_pipe[0]);
dup2(data_pipe[1], STDOUT_FILENO);
// confused as to why this is necessary
close(data_pipe[1]);
execlp("ls", "ls", "-1", NULL);
}
void executeSort(int data_pipe[]){
// close the write file descriptor
close(data_pipe[1]);
dup2(data_pipe[0], STDIN_FILENO);
close(data_pipe[0]);
execlp("sort","sort", NULL);
}