Building a crawler in Rust: Synchronization (Atomic Types and Barriers)

Last week we saw how to design a crawler. Today we are going to see implementation details: how to use Rust's synchrnoization primitives to make our crawler as efficient as possible.

This post is an excerpt from my book Black Hat Rust

Building a crawler in Rust:

Atomic types

Atomic types, like mutexes, are shared-memory types: they can be safely shared between multiple threads.

They allow not to have to use a mutex, and thus and all the ritual around lock() which may introduce bugs such as deadlocks.

You should use an atomic if you want to share a boolean or an integer (such as a counter) across threads instead of a Mutex<bool> or Mutex<i64>.

Operations on atomic types require an ordering argument. The reason is out of the topic of this book, but you can read more about it on this excellent post: Explaining Atomics in Rust.

To keep things simple, use Ordering::SeqCst which provides the strongest guarantees.


use std::sync::atomic::{AtomicUsize, Ordering};
use std::sync::Arc;
use std::thread;

fn main() {
    // creating a new atomic
    let my_atomic = AtomicUsize::new(42);

    // adding 1
    my_atomic.fetch_add(1, Ordering::SeqCst);

    // geting the value
    assert!(my_atomic.load(Ordering::SeqCst) == 43);

    // substracting 1
    my_atomic.fetch_sub(1, Ordering::SeqCst);

    // replacing the value, Ordering::SeqCst);
    assert!(my_atomic.load(Ordering::SeqCst) == 10);

    // other avalable operations
    // fetch_xor, fetch_or, fetch_nand, fetch_and...

    // creating a new atomic that can be shared between threads
    let my_arc_atomic = Arc::new(AtomicUsize::new(4));

    let second_ref_atomic = my_arc_atomic.clone();
    thread::spawn(move|| {, Ordering::SeqCst);

The available types are:

  • AtomicBool
  • AtomicI8
  • AtomicI16
  • AtomicI32
  • AtomicI64
  • AtomicIsize
  • AtomicPtr
  • AtomicU8
  • AtomicU16
  • AtomicU32
  • AtomicU64
  • AtomicUsize

You can learn more about atomic type in the Rust doc.


A barrier is like a sync.WaitGroup in Go: it allows multiples concurrent operations to synchronize.

use tokio::sync::Barrier;
use std::sync::Arc;

async fn main() {
    // number of concurrent operations
    let barrier = Arc::new(Barrier::new(3));

    let b2 = barrier.clone()
    tokio::spawn(async move {
        // do things

     let b3 = barrier.clone()
    tokio::spawn(async move {
        // do things


    println!("This will print only when all the three concurrent operations have terminated");

Want to learn more? Get my book Black Hat Rust where we build a crawler in Rust to scrape vulnerabilities and gather data about our targets.

1 email / week to learn how to (ab)use technology for fun & profit: Programming, Hacking & Entrepreneurship.
I hate spam even more than you do. I'll never share your email, and you can unsubscribe at any time.

Tags: hacking, programming, rust, tutorial

Want to learn Rust, Cryptography and Security? Get my book Black Hat Rust!