Go Mutex Tutorial

The use of Go when programming highly concurrent applications doesn’t prevent you from writing a system that can feature race conditions. These race conditions can cause unexpected issues with your systems that are both hard to debug and at times, even harder to fix.

Thus, we need to be able to write Go programs that can execute concurrently in a safe manner without impacting performance. This is where the mutex comes into play.

In this tutorial, I’m going to show you some fundamental approaches you can follow within your own Go applications that will help you guard your code from these nasty race conditions.

Video Tutorial

This tutorial is available in video format:

The Theory

Before we dive into the code, let’s have a quick look at the theory and why we need mutexes at all.

So, a mutex, or a mutual exclusion is a mechanism that allows us to prevent concurrent processes from entering a critical section of data whilst it’s already being executed by a given process.

Let’s think about an example where we have a bank balance and a system that both deposits and withdraws sums of money from that bank balance. Within a single threaded, synchronous program, this would be incredibly easy. We could effectively guarantee that it would work as intended every time with a small battery of unit tests.

However, if we were to start introducing multiple threads, or multiple goroutines in Go’s case, we may start to see issues within our code.

1. Imagine we had a customer with a balance of £1,000.
2. The customer deposits £500 to his account
3. One goroutine would see this transaction, read the value at £1,000 and proceed to add the £500 to the existing balance.
4. However, at the same moment, a charge of £700 is applied to his account to pay for his mortgage.
5. This second process reads the account balance of £1,000 before the first is able to add the additional deposit of £500 and proceeds to subtract £700 from his account.
6. The customer checks his bank balance the next day and notices he is down to £300 as the second process was unaware of the first deposit and overwrote the value upon completion.

Obviously, if you were the customer, you would be pretty pissed that the bank had somehow “lost” your deposit of £500 and you would switch banks.

This right here, is an example of a race condition and how, if we aren’t careful, our concurrent programs can start to see issues when we don’t carefully guard the critical sections in our code.

A Simple Example

So, now that we know what the problem is, let’s see how we can fix it using a mutex within our Go system. In order to use mutexes within our code, we need to import the sync package.

package main

import (
    "fmt"
    "sync"
)

var (
    mutex   sync.Mutex
    balance int
)

func init() {
    balance = 1000
}

func deposit(value int, wg *sync.WaitGroup) {
    mutex.Lock()
    fmt.Printf("Depositing %d to account with balance: %d\n", value, balance)
    balance += value
    mutex.Unlock()
    wg.Done()
}

func withdraw(value int, wg *sync.WaitGroup) {
    mutex.Lock()
    fmt.Printf("Withdrawing %d from account with balance: %d\n", value, balance)
    balance -= value
    mutex.Unlock()
    wg.Done()
}

func main() {
    fmt.Println("Go Mutex Example")

	var wg sync.WaitGroup
	wg.Add(2)
    go withdraw(700, &wg)
    go deposit(500, &wg)
    wg.Wait()

    fmt.Printf("New Balance %d\n", balance)
}

So, let’s break down what we have done here. Within both our deposit() and our withdraw() functions, we have specified the first step should be to acquire the mutex using the mutex.Lock() method.

Each of our functions will block until it successfully acquires the Lock. Once successful, it will then proceed to enter it’s critical section in which it reads and subsequently updates the account’s balance. Once each function has performed it’s task, it then proceeds to release the lock by calling the mutex.Unlock() method.

When you execute this code, you should see the following output:

Go Mutex Example
Depositing 500 to account with balance: 1000
Withdrawing 700 from account with balance: 1500
New Balance 800

package main

import (
	"fmt"
	"sync"
)


func main() {
	var b sync.Mutex
	
	b.Lock()
	b.Lock()
	fmt.Println("This never executes as we are in deadlock") 
}

fatal error: all goroutines are asleep - deadlock!

goroutine 1 [semacquire]:
sync.runtime_SemacquireMutex(0x40e024, 0x1174ef00, 0x1, 0x40a0d0)
	/usr/local/go/src/runtime/sema.go:71 +0x40
sync.(*Mutex).lockSlow(0x40e020, 0x40c130)
	/usr/local/go/src/sync/mutex.go:138 +0x120
sync.(*Mutex).Lock(...)
	/usr/local/go/src/sync/mutex.go:81
main.main()
	/tmp/sandbox563268272/prog.go:13 +0xe0