How do I work with RwLock for read-write locking in Rust?

Walkthrough

An RwLock (read-write lock) is a synchronization primitive that allows multiple readers OR a single writer to access data concurrently. Unlike a Mutex which provides exclusive access to a single thread, RwLock distinguishes between read and write access.

Key characteristics:

Multiple readers — Many threads can hold read locks simultaneously
Single writer — Only one thread can hold a write lock, with no readers
Read-write mutual exclusion — Writers block readers, readers block writers

RwLock is ideal when:

Data is read frequently but written infrequently
Read operations are time-consuming
You want better concurrency than a Mutex allows

The standard library provides std::sync::RwLock, but for production code, consider parking_lot::RwLock for better performance.

Code Examples

Basic RwLock Usage

use std::sync::RwLock;
 
fn main() {
    let lock = RwLock::new(5);
    
    // Multiple readers can access simultaneously
    {
        let r1 = lock.read().unwrap();
        let r2 = lock.read().unwrap();
        println!("Reader 1: {}", *r1);
        println!("Reader 2: {}", *r2);
    } // Readers released
    
    // Writer needs exclusive access
    {
        let mut w = lock.write().unwrap();
        *w += 1;
        println!("Writer: {}", *w);
    }
}

Shared Cache with RwLock

use std::sync::{Arc, RwLock};
use std::thread;
use std::collections::HashMap;
 
struct Cache {
    data: RwLock<HashMap<String, String>>,
}
 
impl Cache {
    fn new() -> Self {
        Self {
            data: RwLock::new(HashMap::new()),
        }
    }
    
    fn get(&self, key: &str) -> Option<String> {
        let data = self.data.read().unwrap();
        data.get(key).cloned()
    }
    
    fn insert(&self, key: String, value: String) {
        let mut data = self.data.write().unwrap();
        data.insert(key, value);
    }
    
    fn remove(&self, key: &str) -> Option<String> {
        let mut data = self.data.write().unwrap();
        data.remove(key)
    }
}
 
fn main() {
    let cache = Arc::new(Cache::new());
    
    // Initial data
    cache.insert("config".to_string(), "default".to_string());
    
    let mut handles = vec![];
    
    // Spawn reader threads
    for i in 0..5 {
        let cache_clone = Arc::clone(&cache);
        let handle = thread::spawn(move || {
            if let Some(value) = cache_clone.get("config") {
                println!("Reader {}: got config = {}", i, value);
            } else {
                println!("Reader {}: config not found", i);
            }
        });
        handles.push(handle);
    }
    
    // Spawn writer thread
    let cache_clone = Arc::clone(&cache);
    let writer = thread::spawn(move || {
        cache_clone.insert("config".to_string(), "updated".to_string());
        println!("Writer: updated config");
    });
    
    for handle in handles {
        handle.join().unwrap();
    }
    writer.join().unwrap();
}

Read vs Write Lock Semantics

use std::sync::{Arc, RwLock};
use std::thread;
 
fn main() {
    let data = Arc::new(RwLock::new(vec![1, 2, 3]));
    let mut handles = vec![];
    
    // Spawn multiple readers - they run concurrently
    for i in 0..3 {
        let data_clone = Arc::clone(&data);
        let handle = thread::spawn(move || {
            let read_guard = data_clone.read().unwrap();
            println!("Reader {}: {:?}", i, *read_guard);
            // Simulate work while holding read lock
            thread::sleep(std::time::Duration::from_millis(100));
            println!("Reader {} done", i);
        });
        handles.push(handle);
    }
    
    // Give readers a moment to start
    thread::sleep(std::time::Duration::from_millis(10));
    
    // Writer waits for all readers to finish
    let data_clone = Arc::clone(&data);
    let writer = thread::spawn(move || {
        println!("Writer: trying to acquire write lock...");
        let mut write_guard = data_clone.write().unwrap();
        println!("Writer: acquired write lock");
        write_guard.push(4);
        println!("Writer: added element");
    });
    
    for handle in handles {
        handle.join().unwrap();
    }
    writer.join().unwrap();
    
    println!("Final data: {:?}", *data.read().unwrap());
}

Try Read and Try Write

use std::sync::{Arc, RwLock};
use std::thread;
use std::time::Duration;
 
fn main() {
    let data = Arc::new(RwLock::new(0));
    
    let data_clone = Arc::clone(&data);
    let writer = thread::spawn(move || {
        let mut guard = data_clone.write().unwrap();
        println!("Writer acquired lock");
        
        // Hold the lock for a while
        thread::sleep(Duration::from_millis(500));
        
        *guard = 42;
        println!("Writer done");
    });
    
    thread::sleep(Duration::from_millis(50)); // Let writer acquire lock
    
    let data_clone = Arc::clone(&data);
    let reader = thread::spawn(move || {
        // Try to read without blocking
        match data_clone.try_read() {
            Ok(guard) => println!("Got read: {}", *guard),
            Err(e) => println!("Could not read: {}", e),
        }
        
        // Try to write without blocking
        match data_clone.try_write() {
            Ok(mut guard) => {
                *guard = 100;
                println!("Got write, set to 100");
            }
            Err(e) => println!("Could not write: {}", e),
        }
    });
    
    writer.join().unwrap();
    reader.join().unwrap();
    
    println!("Final value: {}", *data.read().unwrap());
}

Upgrading Read Lock to Write Lock

use std::sync::{Arc, RwLock};
 
fn main() {
    // Standard RwLock doesn't support upgrading
    // You must release the read lock before acquiring write lock
    // This can cause race conditions - be careful!
    
    let data = Arc::new(RwLock::new(HashMap::<String, i32>::new()));
    
    // Pattern: check then update
    fn get_or_insert(cache: &RwLock<HashMap<String, i32>>, key: &str, default: i32) -> i32 {
        // First, try to read
        {
            let read_guard = cache.read().unwrap();
            if let Some(&value) = read_guard.get(key) {
                return value;
            }
        } // Release read lock
        
        // Now acquire write lock
        {
            let mut write_guard = cache.write().unwrap();
            // MUST check again - another thread may have inserted
            write_guard.entry(key.to_string()).or_insert(default);
            *write_guard.get(key).unwrap()
        }
    }
    
    let result = get_or_insert(&data, "count", 0);
    println!("Result: {}", result);
}
 
use std::collections::HashMap;

Using parking_lot::RwLock

// Add to Cargo.toml:
// [dependencies]
// parking_lot = "0.12"
 
// parking_lot provides a better RwLock:
// - No poisoning (no unwrap needed)
// - Smaller memory footprint
// - Faster performance
// - Supports read lock upgrading
 
use parking_lot::RwLock;
use std::sync::Arc;
use std::thread;
 
fn main() {
    let data = Arc::new(RwLock::new(vec![1, 2, 3]));
    
    let data_clone = Arc::clone(&data);
    let handle = thread::spawn(move || {
        let read_guard = data_clone.read();
        println!("Reader: {:?}", *read_guard);
    });
    
    handle.join().unwrap();
    
    // No unwrap needed with parking_lot
    let mut write_guard = data.write();
    write_guard.push(4);
    
    println!("Final: {:?}", *data.read());
}

Read-Heavy Workload Example

use std::sync::{Arc, RwLock};
use std::thread;
use std::time::Instant;
 
struct Database {
    users: RwLock<Vec<(u32, String)>>,
}
 
impl Database {
    fn new() -> Self {
        let mut users = Vec::new();
        for i in 0..1000 {
            users.push((i, format!("user_{}", i)));
        }
        Self {
            users: RwLock::new(users),
        }
    }
    
    fn find_user(&self, id: u32) -> Option<String> {
        let users = self.users.read().unwrap();
        users.iter()
            .find(|(uid, _)| *uid == id)
            .map(|(_, name)| name.clone())
    }
    
    fn add_user(&self, id: u32, name: String) {
        let mut users = self.users.write().unwrap();
        users.push((id, name));
    }
    
    fn user_count(&self) -> usize {
        self.users.read().unwrap().len()
    }
}
 
fn main() {
    let db = Arc::new(Database::new());
    let start = Instant::now();
    
    let mut handles = vec![];
    
    // 10 reader threads
    for i in 0..10 {
        let db_clone = Arc::clone(&db);
        handles.push(thread::spawn(move || {
            for j in 0..100 {
                let id = (i * 100 + j) % 1000;
                if let Some(name) = db_clone.find_user(id as u32) {
                    // Process user
                    let _ = name.len();
                }
            }
        }));
    }
    
    // 1 writer thread
    let db_clone = Arc::clone(&db);
    handles.push(thread::spawn(move || {
        for i in 1000..1010 {
            db_clone.add_user(i, format!("new_user_{}", i));
            thread::sleep(std::time::Duration::from_micros(100));
        }
    }));
    
    for handle in handles {
        handle.join().unwrap();
    }
    
    println!("Total users: {}", db.user_count());
    println!("Time: {:?}", start.elapsed());
}

Deadlock Prevention

use std::sync::{Arc, RwLock};
use std::thread;
 
fn main() {
    let data1 = Arc::new(RwLock::new(0));
    let data2 = Arc::new(RwLock::new(0));
    
    let d1 = Arc::clone(&data1);
    let d2 = Arc::clone(&data2);
    
    let handle1 = thread::spawn(move || {
        // Always acquire locks in the same order!
        let mut g1 = d1.write().unwrap();
        let mut g2 = d2.write().unwrap();
        *g1 += 1;
        *g2 += 1;
    });
    
    let d1 = Arc::clone(&data1);
    let d2 = Arc::clone(&data2);
    
    let handle2 = thread::spawn(move || {
        // Same order as handle1 - prevents deadlock
        let mut g1 = d1.write().unwrap();
        let mut g2 = d2.write().unwrap();
        *g1 += 10;
        *g2 += 10;
    });
    
    handle1.join().unwrap();
    handle2.join().unwrap();
    
    println!("data1: {}, data2: {}", *data1.read().unwrap(), *data2.read().unwrap());
}

RwLock vs Mutex Comparison

use std::sync::{Arc, RwLock, Mutex};
use std::thread;
use std::time::Instant;
 
fn main() {
    const NUM_READERS: usize = 10;
    const NUM_READS: usize = 100_000;
    
    // Test with RwLock
    let rwlock_data = Arc::new(RwLock::new(0i32));
    let start = Instant::now();
    
    let mut handles = vec![];
    for _ in 0..NUM_READERS {
        let data = Arc::clone(&rwlock_data);
        handles.push(thread::spawn(move || {
            for _ in 0..NUM_READS {
                let _ = *data.read().unwrap();
            }
        }));
    }
    
    for h in handles {
        h.join().unwrap();
    }
    
    let rwlock_time = start.elapsed();
    
    // Test with Mutex
    let mutex_data = Arc::new(Mutex::new(0i32));
    let start = Instant::now();
    
    let mut handles = vec![];
    for _ in 0..NUM_READERS {
        let data = Arc::clone(&mutex_data);
        handles.push(thread::spawn(move || {
            for _ in 0..NUM_READS {
                let _ = *data.lock().unwrap();
            }
        }));
    }
    
    for h in handles {
        h.join().unwrap();
    }
    
    let mutex_time = start.elapsed();
    
    println!("RwLock: {:?}", rwlock_time);
    println!("Mutex: {:?}", mutex_time);
    println!("RwLock is {}x faster for read-heavy workloads", 
        mutex_time.as_secs_f64() / rwlock_time.as_secs_f64());
}

Summary

Method	Description	Blocks When
`read()`	Acquire read lock	Write lock held
`try_read()`	Non-blocking read attempt	Would block
`write()`	Acquire write lock	Any lock held
`try_write()`	Non-blocking write attempt	Would block

RwLock vs Mutex:

Feature	RwLock	Mutex
Concurrent readers	✅ Multiple	❌ Only one
Write exclusivity	✅ Exclusive	✅ Exclusive
Memory overhead	Higher	Lower
Best for	Read-heavy	Write-heavy or equal R/W

When to Use RwLock:

Data is read much more often than written
Read operations take significant time
Multiple threads need concurrent read access
You want to maximize read throughput

When to Use Mutex Instead:

Read and write frequencies are similar
Operations are very fast (lock overhead dominates)
You need to hold the lock while performing complex operations
Simple cases where performance doesn't matter

Key Points:

Multiple readers can hold read locks simultaneously
Writers need exclusive access (no readers or other writers)
Use try_read() and try_write() to avoid blocking
Be careful with lock ordering to prevent deadlocks
Consider parking_lot::RwLock for better performance
Always check data again after upgrading (read → write) to avoid races