README.md

Shared State Concurrency

In the shared memory concurrency model multiple threads can access the same memory location at the same time.

Suppose we have a vector and we want to access/share that vector in several threads. In particular, assume that each thread takes an index i, the vector, and prints to stdout element located at index i.

A simple solution is to use Arc<T> like the following

use std::sync::Arc;
use std::thread;

pub fn main() {
    let ptr1 = Arc::new(vec![1, 2, 3]);
    let mut handles = vec![];

    for i in 0..ptr1.len() {
        let ptr2 = Arc::clone(&ptr1);

        let handle = thread::spawn(move || {
            println!("{}", ptr2[i]);
        });
        handles.push(handle);
    }

    for handle in handles {
        handle.join().unwrap();
    }
}

Next, suppose we want to mutate the vector inside the threads. We could implement it using Mutex<T>.

use std::sync::{Arc, Mutex};
use std::thread;

pub fn main() {
    let ptr1 = Arc::new(Mutex::new(vec![0, 1, 2]));
    let mut handles = vec![];

    println!("Before increasing: {:?}", *ptr1);

    for _ in 0..6 {
        let ptr2 = Arc::clone(&ptr1);

        let handle = thread::spawn(move || {
            let mut vptr = ptr2.lock().unwrap();
            // mutate data
            vptr[0] += 1;
            vptr[1] += 1;
            vptr[2] += 1;
        });
        handles.push(handle);
    }

    for handle in handles {
        handle.join().unwrap();
    }

    println!("After increasing: {:?}", *ptr1);
}

However, Mutex<T> comes with the risk of creating deadlocks. In the following piece of code we have two vectors and we try to mutually access these vectors inside two different threads in different order which results in a deadlock.

use std::sync::{Arc, Mutex};
use std::{thread, time};

pub fn main() {
    let ptr1 = Arc::new(Mutex::new(vec![0, 1, 2]));
    let ptr2 = Arc::new(Mutex::new(vec![3, 4, 5]));
    let mut handles = vec![];

    for i in 0..2 {
        let ptr3 = Arc::clone(&ptr1);
        let ptr4 = Arc::clone(&ptr2);

        let handle = thread::spawn(move || {
            let one_sec = time::Duration::from_secs(1);
            if i & 1 == 0 {
                let _lock1 = ptr3.lock().unwrap();
                thread::sleep(one_sec);
                let _lock2 = ptr4.lock().unwrap();
            } else {
                let _lock1 = ptr4.lock().unwrap();
                thread::sleep(one_sec);
                let _lock2 = ptr3.lock().unwrap();
            }
        });
        handles.push(handle);
    }

    for handle in handles {
        handle.join().unwrap();
    }

    // This line is unreachable
    println!("Done.");
}