According to Goetz and colleagues [Goetz 2006]:
Client-side locking entails guarding client code that uses some object X with the lock X uses to guard its own state. In order to use client-side locking, you must know what lock X uses.
While client-side locking is acceptable when the thread-safe class commits to and clearly documents its locking strategy, Goetz and colleagues caution against its misuse [Goetz 2006]:
If extending a class to add another atomic operation is fragile because it distributes the locking code for a class over multiple classes in an object hierarchy, client-side locking is even more fragile because it entails putting locking code for class C into classes that are totally unrelated to C. Exercise care when using client-side locking on classes that do not commit to their locking strategy.
The documentation of a class that supports client-side locking should explicitly state its applicability. For example, the class java.util.concurrent.ConcurrentHashMap<K,V> should not be used for client-side locking because its documentation [API 2006] states that:
However, even though all operations are thread-safe, retrieval operations do not entail locking, and there is not any support for locking the entire table in a way that prevents all access. This class is fully interoperable with
Hashtablein programs that rely on its thread safety but not on its synchronization details.
Use of client-side locking is permitted only when the documentation of the class recommends it. For example, the documentation of the synchronizedList() wrapper method of java.util.Collections class [API 2006] states
In order to guarantee serial access, it is critical that all access to the backing list is accomplished through the returned list. It is imperative that the user manually synchronize on the returned list when iterating over it. Failure to follow this advice may result in non-deterministic behavior.
...
Goetz and colleagues describe the fragility of class extension for adding functionality to thread-safe classes [Goetz 2006]:
Extension is more fragile than adding code directly to a class, because the implementation of the synchronization policy is now distributed over multiple, separately maintained source files. If the underlying class were to change its synchronization policy by choosing a different lock to guard its state variables, the subclass would subtly and silently break because it no longer used the right lock to control concurrent access to the base class's state.
...
If the IPAddressList class were modified to use block synchronization on a private final lock object, as recommended by rule LCK00-J. Use private final lock objects to synchronize classes that may interact with untrusted code, the PrintableIPAddressList subclass would silently break. Moreover, if a wrapper such as Collections.synchronizedList() were used, it would be difficult for a client to determine the type of the class being wrapped to extend it [Goetz 2006].
Compliant Solution (Composition)
...
In this case, composition allows the PrintableIPAddressList class to use its own intrinsic lock independent of the underlying list class's lock. The underlying collection lacks a requirement for thread-safety because the PrintableIPAddressList wrapper prevents direct access to its methods by publishing its own synchronized equivalents. This approach provides consistent locking even when the underlying class changes its locking policy in the future [Goetz 2006].
Risk Assessment
Using client-side locking when the thread-safe class fails to commit to its locking strategy can cause data inconsistencies and deadlock.
Rule | Severity | Likelihood | Remediation Cost | Priority | Level |
|---|---|---|---|---|---|
LCK11-J | low | probable | medium | P4 | L3 |
Bibliography
[API 2006] | Class |
8.2, Synchronization and Collection Classes | |
4.4.1, Client-side Locking; 4.4.2, Composition; and 5.2.1, | |
[Lee 2009] | Map & Compound Operation |
...