Threading in C/C++ programming language using pthreads

In this post, we will write a simple C script for performing a task concurrently using the POSIX pthread api in linux/unix like operating systems. For more details on concurrency, see my previous post about parallel computing in Python.

Simple script to make use of the pthread api

#include <pthread.h> //pthread
#include <unistd.h> //sleep
#include <stdio.h> // printf
#include <string.h> //strcpy

struct sData {char text[100];};


void *myfunc(void *voidData) {
    struct sData *data = voidData;
    for (int i=0; i<8; i++) {
        printf ("(i = %d) data text is %s \n", i, data->text);
        sleep(1);
    }
    return NULL;
}

void anotherfunc() {
    for (int i=0; i<3; i++) {
        sleep(1); //sleep for 1s
        printf("(i = %d) this is another func! \n", i);
    }
}

int main() {
    struct sData sd;
    int tret;
    pthread_t tid; // declare a new variable of type pthread_t tid to identify the thread in the system
    strcpy(sd.text, "earthinversion");
    tret = pthread_create(&tid, NULL, myfunc, &sd); // args: pointer to thread id, attributes, function name, arguments to this function
    anotherfunc();
    pthread_join(tid, NULL); 

    return 0;
}

In the main() function above, we define a data sd, and integer tret to store the output of the thread creation. We store text value in the sd.text, which will be argument for the myfunc. Then we create a variable tid of the data type pthread to identify the thread in the system. To create a thread, we need four arguments - pointer to the thread id, attributes (like detached state, scheduling policy, stack address, etc. Here, we specify it to be NULL, hence the default attributes will be used), name of the function and the argument to the function. Here, the function takes only one argument. If multiple arguments need to be given, then we need to use struct.

We then execute function anotherfunc() in the main thread. Since, the function anotherfunc() will finish before myfunc, we need to wait for the myfunc. To do this, we call the pthread_join. The second argument of the pthread_join is the return value of the thread. In this case, we declare it to be NULL.

Compilation

To compile the codes, we need to use the compiler with pthread library.

gcc threading.c -o threading

When you execute the threading, you should get similar return:

(i = 0) data text is earthinversion 
(i = 1) data text is earthinversion 
(i = 0) this is another func! 
(i = 2) data text is earthinversion 
(i = 1) this is another func! 
(i = 3) data text is earthinversion 
(i = 2) this is another func! 
(i = 4) data text is earthinversion 
(i = 5) data text is earthinversion 
(i = 6) data text is earthinversion 
(i = 7) data text is earthinversion

Notice how the two functions’ lines are interleaved — that’s the concurrency: while the worker thread prints its 8 lines, the main thread is running anotherfunc() at the same time. Once anotherfunc() returns, pthread_join blocks main until the worker’s remaining iterations finish.

Recap

Without scrolling up — what’s the threading pattern?

• A thread runs concurrently inside the same process and shares memory with it.
• pthread_create(&tid, NULL, fn, arg) launches a worker running fn(arg) (bundle multiple args in a struct).
• The main thread keeps going (here, anotherfunc()), so output interleaves.
• pthread_join(tid, NULL) waits for the worker to finish before main returns.
• Compile with gcc file.c -o out -pthread.

Further reads

1. Thread functions in C/C++ — GeeksforGeeks.