How do you read stderr when using execvp, running a command that outputs and error in the terminal using this script will result in an empty buffer (no text)

int cmd(char * const *cmd, char * output)
{
    int pipefds[2], r, status, x;  
    pid_t pid;

    if (pipe(pipefds) == -1){
        return -1;
    }
    if ( (pid = fork()) == -1){
        return -1;
    }

    if (pid == 0)
    {
        dup2(pipefds[1], STDOUT_FILENO);
        close(pipefds[0]);
        close(pipefds[1]);
        execvp(cmd[0] , cmd);
    }
    else
    {
        close(pipefds[1]);
        x = read(pipefds[0], output, 16384);
        wait(NULL);
    }
    return 0;
}

all the resources that I found online do not seem to mention where/how to pipe the error into the buffer, what are the three file descriptors in pipefds one for stdin another for stdout what about the last one stderr?

>Solution :

Before calling execvp, set it up so that file descriptor STDERR_FILENO (i.e. 2) goes to the pipe.

You can simply dup it an extra time, e.g.:

dup2(pipefds[1], STDOUT_FILENO);
dup2(pipefds[1], STDERR_FILENO);

now both file descriptors 1 (stdout) and 2 (stderr) point to the same pipe.