GotchaDifficulty: Intermediate1 min

static synchronized Locks the Class, Not the Instance

Every method is synchronized, one is static, and the shared state is corrupted anyway.

A static synchronized method locks the class object, Registry.class, while an ordinary synchronized method locks the instance it was called on. They are two different locks, so a static and a non-static synchronized method do not keep threads apart from each other.

A class where every method is synchronized and one of them is static: two threads run through it at the same time, and the shared state is corrupted anyway.

java

class Registry {
    private static int total;
    private int localTotal;

    static synchronized void addToTotal(int n) { total += n; }

    synchronized void addToLocal(int n) { localTotal += n; }
}

Line 7 locks the instance it was called on. Line 5 has no instance, because it is static, so it locks the Registry.class object instead.

What each form of synchronized locks
Written	Locks
`synchronized void m()`	`this`, the instance it was called on
`static synchronized void m()`	The class object, shared by every instance
`synchronized (obj) { }`	Whatever `obj` is

Guard state with the lock that matches its scope: a static field with the class lock, an instance field with the instance lock. Better still, give each piece of shared state a private lock object of its own, so the pairing is written down rather than implied.

java

private static final Object TOTAL_LOCK = new Object();
private final Object localLock = new Object();

← → move through sections and cards