CON00-J. Declare shared variables as volatile to ensure visibility and prevent reordering of statements

Memory that can be shared between threads is called shared memory or heap memory. The term variable is used in the context of this guideline, to refer to both fields and array elements [[JLS 05]]. Variables that are shared between threads are referred to as shared variables. All instance fields, static fields, and array elements are shared variables and are stored in heap memory. Local variables, formal method parameters, or exception handler parameters are never shared between threads and are not affected by the [memory model].

The Java Language Specification defines the Java Memory Model (JMM) which describes possible behaviors of a multi-threaded Java program. If the correctness of the constituting code is based on its execution characteristics in a single threaded environment, the code may be unsafe for use in a multi-threaded environment. Two kinds of hazards, visibiity and reordering, occur when special concurrency primitives are not used in such code.

Visibility refers to the requirement that every thread sees the value of the most recent update to a shared variable. This is essential in multi-threaded programs because the value of a shared variable may be cached and not written to main memory immediately. Consequently, another thread may retrieve a stale value of the variable when attempting to read it from the main memory. Using the volatile keyword mitigates this risk and so does correct synchronization.

The restricted or no reordering requirement refers to ensuring that the execution sequence of a set of statements does not vary when observed by multiple threads. Concurrent executions of code are typically interleaved and statements may be reordered by the compiler or runtime system to facilitate various optimizations. This results in execution orders that are not immediately obvious from an examination of the source-code. Failure to account for this hazard is a common source of data races. Restricting the set of possible reorderings makes it easier to reason about the safety of the code. This risk is mitigated by correctly synchronizing the code and in some cases, by using the volatile keyword.

Even if statements execute in the expected order (program order), caching can prevent the latest values from being reflected in the main memory (visibility hazard). Program order is the execution order that is expected when a single thread is running the statements sequentially, as written in a method.

There are two requirements for writing provably correct multi-threaded code:

1. Happens-before consistency: If two accesses of a shared variable follow the happens-before relationship, data races arising from statement reorderings cannot occur. However, this is necessary but not sufficient for acceptable program behavior.

Consider the following example in which a and b are (shared) global variables or instance fields but r1 and r2 are local variables not accessible by other threads.

Initially, let a = 0 and b = 0.

Because, in Thread 1, the two assignments a = 10; and r1 = b; are not related, the compiler or runtime system is free to reorder them. Similarly in Thread 2, the statements may be freely reordered. Although it may seem counter-intuitive, the Java memory model allows a read to see a write that occurs later in the execution order.

A possible execution order showing actual assignments is:

In this ordering, r1 and r2 read the original values of the variables a and b even though they are expected to see the updated values, 10 and 20. Another possible execution order showing actual assignments is:

In this ordering, r1 and r2 read the values of a and b written from step 3 and 4, even before the statements corresponding to these steps have executed.

"The fact that we allow a read to see a write that comes later in the execution order can sometimes thus result in unacceptable behaviors." [[JLS 05]]. Both, the use of synchronization and the volatile keyword prevent a thread from observing inconsistent values of shared variables.

2. Sequential consistency: This property provides a very strong guarantee that the compiler will not optimize away or reorder any statements. It guarantees that the program is free from data races. It also ensures that each access of a variable is atomic and immediately visible to other threads.

For example, consider a set of statements that are being executed by multiple threads:

If statements 1, 2 and 3 are always executed sequentially by all threads as given in this program order, they are sequentially consistent with respect to each other. The sequential consistency property also requires that a read operation in some thread does not see the value of a future write operation taking place in the same or another thread. Similarly, a read operation is guaranteed to see the value of the last write to the variable from any thread.

The use of sequential consistency as the sole memory model mechanism makes it easy for a programmer to reason with the program's logic in a multithreading scenario, however, introduces a performance penalty because the compiler is prohibited from reordering code for performing complex optimizations. Using volatile variables reduces this performance penalty at the cost of strong sequential consistency guarantees.

Consider two threads that are executing some statements:

Thread 1 and Thread 2 have a happens-before relationship such that Thread 2 does not start before Thread 1 finishes. This is established by the semantics of volatile accesses. Sequential consistency of volatile accesses provides certain visibility and reordering guarantees:

Visibility

A write to a volatile field happens-before every subsequent read of that field. Statements that occur before the write to the volatile field also happen-before the read of the volatile field.

In the previous example, Statement 3 writes to a volatile variable, and statement 4 in the second thread, reads the same volatile variable. The read sees the most recent write (to the same variable v) from statement 3. This may not be true in the happens-before order because a future read can always see the default or previous value of v instead of the one set in the most recent write. This guarantee is provided by the sequential consistency property of volatile accesses.

Reordering

Volatile read and write operations cannot be reordered with respect to each other and in addition, as required by the JMM, volatile read and write operations are also not reordered with respect to operations on nonvolatile variables. When reading the volatile variable, the other thread will also see statements occurring before the write to the volatile variable to have already executed, with prior occurrences of volatile and nonvolatile fields assuming the assigned values.

In the previous example, statement 4 also sees the statements 1 and 2 to have executed and all their operands with the most-up to date values. However, this does not mean that statements 1 and 2 are sequentially consistent with respect to each other. They may be freely reordered by the compiler. In fact, if statement 1 constituted a read of some variable x, it could see the value of a future write to x in statement 2.

Because the guarantees of code present before the volatile write are weaker than sequentially consistent code, volatile as a synchronization primitive, performs better. "Because no locking is involved, declaring fields as volatile is likely to be cheaper than using synchronization, or at least no more expensive. However, if volatile fields are accessed frequently inside methods, their use is likely to lead to slower performance than would locking the entire methods." [[Lea 00]].

"Finally, note that the actual execution order of instructions and memory accesses can be in any order as long as the actions of the thread appear to that thread as if program order were followed, and provided all values read are allowed for by the memory model. This allows the programmer to fully understand the semantics of the programs they write, and it allows compiler writers and virtual machine implementors to perform complex optimizations that a simpler memory model would not permit." [[JPL 06]].

The possible reorderings between volatile and nonvolatile variables are summarized in the matrix shown below. The load and store operations correspond to read and write operations that use the variable. [[Lea 08]]

Noncompliant Code Example (status flag)

This noncompliant code example uses a shutdown() method to set a non-volatile done flag that is checked in the run() method. If one thread invokes the shutdown() method to set the flag, it is possible that another thread might not observe this change. Consequently, the second thread may still observe that done is false and incorrectly invoke the sleep() method.

final class ControlledStop implements Runnable {
  private boolean done = false;
 
  public void run() {
    while (!done) {
      try {
        // ...
        Thread.currentThread().sleep(1000); // Do something
      } catch(InterruptedException ie) { 
        // handle exception
      } 
    } 	 
  }

  protected void shutdown(){
    done = true;
  }
}

Compliant Solution (volatile status flag)

This compliant solution qualifies the done flag as volatile so that updates by one thread are immediately visible to another thread.

final class ControlledStop implements Runnable {
  private volatile boolean done = false;
  // ...
}

Noncompliant Code Example (nonvolatile guard)

This noncompliant code example declares a non-volatile int variable that is initialized in the constructor depending on a security check. In a multi-threading scenario, it is possible that the statements will be reordered so that the boolean flag initialized is set to true before the initialization has concluded. If it is possible to obtain a partially initialized instance of the class in a subclass using a finalizer attack (OBJ04-J. Do not allow partially initialized objects to be accessed), a race condition can be exploited by invoking the getBalance() method to obtain the balance even though initialization is still underway.

class BankOperation {
  private int balance = 0;
  private boolean initialized = false;
 
  public BankOperation() {
    if (!performAccountVerification()) {
      throw new SecurityException("Invalid Account"); 
    }
    balance = 1000;   
    initialized = true; 
  }
  
  private int getBalance() {
    if (initialized == true) {
      return balance;
    }
    else {
      return -1;
    }
  }
}

Compliant Solution (volatile guard)

This compliant solution declares the initialized flag as volatile to ensure that the initialization statements are not reordered.

class BankOperation {
  private int balance = 0;
  private volatile boolean initialized = false; // Declared volatile
  // ...
}

The use of the volatile keyword is inappropriate for composite operations on shared variables (CON01-J. Design APIs that ensure atomicity of composite operations and visibility of results).

Noncompliant Code Example (visibility)

This noncompliant code example consists of two classes, an immutable ImmutablePoint class and a mutable Holder class. Holder is mutable because a new ImmutablePoint instance can be assigned to it using the setPoint() method. If one thread updates the value of the ipoint field, another thread may still see the reference of the old value.

class Holder {
  ImmutablePoint ipoint;
  
  Holder(ImmutablePoint ip) {
   ipoint = ip;
  }
  
  ImmutablePoint getPoint() {
    return ipoint;
  }

  void setPoint(ImmutablePoint ip) {
    this.ipoint = ip;
  }
}

public class ImmutablePoint {
  final int x;
  final int y;

  public ImmutablePoint(int x, int y) {
    this.x = x;
    this.y = y;
  }
}

Compliant Solution (visibility)

This compliant solution declares the ipoint field as volatile so that updates are immediately visible to other threads.

class Holder {
  volatile ImmutablePoint ipoint;
  
  Holder(ImmutablePoint ip) {
    ipoint = ip;
  }
  
  ImmutablePoint getPoint() {
    return ipoint;
  }

  void setPoint(ImmutablePoint ip) {
    this.ipoint = ip;
  }
}

Note that no synchronization is necessary for the setPoint() method because it operates atomically on immutable data, that is, on an instance of ImmutablePoint.

Declaring immutable fields as volatile enables their safe publication, in that, once published, it is impossible to change the state of the sub-object.

Noncompliant Code Example (partial initialization)

Thread-safe objects (which may not be strictly immutable) must declare their nonfinal fields as volatile to ensure that no thread sees any field references before the sub-objects' initialization has concluded. This noncompliant code example does not declare the map field as volatile.

public class Container<K,V> {
  Map<K,V> map;

  public Container() {
    map = new HashMap<K,V>();	
    // Put values in HashMap
  }

  public V get(Object k) {
    return map.get(k);
  }
}

Compliant Solution (proper initialization)

This compliant solution declares the map field as volatile to ensure other threads see an up-to-date HashMap reference and object state.

public class Container<K,V> {
  volatile Map<K,V> map;
  // ...
}

Risk Assessment

Failing to use volatile to guarantee visibility of shared values across multiple thread and prevent reordering of statements can result in unpredictable control flow.

Automated Detection

TODO

Related Vulnerabilities

Search for vulnerabilities resulting from the violation of this rule on the CERT website.

References

[[JLS 05]] Chapter 17, Threads and Locks, section 17.4.5 Happens-before Order, section 17.4.3 Programs and Program Order, section 17.4.8 Executions and Causality Requirements
[[Tutorials 08]] Java Concurrency Tutorial
[[Lea 00]] Sections, 2.2.7 The Java Memory Model, 2.2.5 Deadlock, 2.1.1.1 Objects and locks
[[Bloch 08]] Item 66: Synchronize access to shared mutable data
[[Goetz 06]] 3.4.2. "Example: Using Volatile to Publish Immutable Objects"
[[JPL 06]] 14.10.3. "The Happens-Before Relationship"
[[MITRE 09]] CWE ID 667 "Insufficient Locking", CWE ID 413 "Insufficient Resource Locking", CWE ID 366 "Race Condition within a Thread", CWE ID 567 "Unsynchronized Access to Shared Data"

11. Concurrency (CON)      11. Concurrency (CON)      CON02-J. Always synchronize on the appropriate object