CON29-J. Do not execute interdependent tasks in a bounded thread pool

Tasks that depend on other tasks to be complete should not be executed in the same bounded thread pool. A bounded thread pool is one that has a fixed size.

A form of deadlock called thread starvation deadlock arises when all the threads executing in the pool are blocked on tasks that are waiting on the queue. A blocking operation within a sub-task can also lead to unbounded queue growth. [[Goetz 06]]

Thread starvation deadlock can arise if a task spawns several other tasks in its own thread pool and waits for them to complete. Because of the limited thread pool size, only a fixed number of tasks can execute to completion at a particular time.

This issue is deceptive because the program may appear to function correctly when fewer threads are needed. In some cases, the issue can be mitigated by choosing a larger pool size. However, there is often no easy way to determine a suitable size. Threads in a thread pool may not be recycled if two executing tasks require each other to complete before they can terminate.

Noncompliant Code Example

This noncompliant code example is vulnerable to a thread starvation deadlock. It consists of class ValidationService which performs various input validation tasks such as checking whether a user-supplied field exists in a back-end database.

The fieldAggregator() method accepts a variable number of String arguments and creates a task corresponding to each argument to gain some speedup. The task performs input validation using class ValidateInput.

The class ValidateInput in turn, attempts to sanitize the input by creating a sub-task for each request using class SanitizeInput. All tasks are executed in the same thread pool. The method fieldAggregator() blocks until all the tasks have finished executing. When all results are available, it aggregates the processed inputs as a StringBuilder and returns it to the caller.

public final class ValidationService {
  private final ExecutorService pool;

  public ValidationService(int poolSize) {
    pool = Executors.newFixedThreadPool(poolSize);
  }
  
  public void shutdown() {
    pool.shutdown();
  }
     
  public StringBuilder fieldAggregator(String... inputs) 
    throws InterruptedException, ExecutionException   {
   
    StringBuilder sb = new StringBuilder();
    Future<String>[] results = new Future[inputs.length]; // Stores the results
		
    for(int i = 0; i < inputs.length; i++) { // Submits the tasks to thread pool
      results[i] = pool.submit(new ValidateInput<String>(inputs[i], pool));     
    } 

    for(int i = 0; i < inputs.length; i++) { // Aggregates the results	    	
      sb.append(results[i].get());	    	
    }
    return sb;
  }
}


public final class ValidateInput<V> implements Callable<V> {
  private final V input;
  private final ExecutorService pool;

  ValidateInput(V input, ExecutorService pool) {
    this.input = input;
    this.pool = pool;
  }

  @Override public V call() throws Exception {
    // If validation fails, throw an exception here
    Future<V> future = pool.submit(new SanitizeInput<V>(input)); // Sub-task
    return (V)future.get();
  }
}


public final class SanitizeInput<V> implements Callable<V> {
  private final V input;
	
  SanitizeInput(V input) {
    this.input = input;
  }

  @Override public V call() throws Exception {
    // Sanitize input and return
    return (V)input;	
  }
}

Assume that the caller sets the thread pool size as 6. When ValidationService.fieldAggregator() is invoked with six arguments that are required to be validated, six tasks are submitted to the thread pool. Six more sub-tasks corresponding to SanitizeInput must also execute before these threads can return their results. However, this is not possible because the queue is full with all threads blocked. Furthermore, invoking the shutdown() method does not shutdown the thread pool when it contains active tasks.

This situation can also occur when using single threaded Executors, for example, when the caller creates several sub-tasks and waits for the results.

Compliant Solution (no interdependent threads)

This compliant solution refactors the ValidateInput<V> class so that the tasks corresponding to SanitizeInput are not executed as distinct threads in the thread pool but in the same threads as the ValidateInput tasks. Consequently, the ValidateInput and SanitizeInput tasks are independent of each other, and need not wait for each other to complete. Also, SanitizeInput can be refactored to not implement Callable.

class ValidationService {
 // ...
 public StringBuilder fieldAggregator(String... inputs) 
   throws InterruptedException, ExecutionException {
   // ...
   for (int i = 0; i < inputs.length; i++) {
     results[i] = pool.submit(new ValidateInput<String>(inputs[i])); // Don't pass-in thread pool    
   } 
   // ...
 } 
}

public final class ValidateInput<V> implements Callable<V> { // Does not use same thread pool
  private final String input;
	
  ValidateInput(String input) {
    this.input = input;
  }

  @Override public V call() throws Exception {
    // If validation fails, throw an exception here
    return (V)SanitizeInput.sanitizeString(input);
  }
}

public final class SanitizeInput {  // No longer a Callable task	
  private SanitizeInput() { }

  public static String sanitizeString(String input) {
    // Sanitize input and return
    return input;	
  }
}

Always submit independent tasks to the Executor. Thread starvation issues can be mitigated by choosing a large pool size, however, the limitations of the application when using this approach should be clearly documented.

Note that operations that have further constraints, such as the total number of database connections or total ResultSets open at a particular time, impose an upper bound on the thread pool size as each thread continues to block until the resource becomes available. The other rules of fair concurrency, such as not running time consuming tasks, also apply. When this is not possible, expecting to obtain real time result guarantees from the execution of tasks is conceivably, an unreasonable target.

Sometimes, a private static ThreadLocal variable is used per thread to maintain local state. When using thread pools, ThreadLocal variables should be used only if their lifetime is shorter than that of the corresponding task [[Goetz 06]]. Moreover, such variables should not be used as a communication mechanism between tasks.

Compliant Solution (Unbounded thread pool)

This compliant solution uses a cached thread pool, which dynamically creates new threads as needed and prevents deadlock. However, this implementation may result in resource exhaustion and should not be used in front-end or critical production systems.

public final class ValidationService {
  private final ExecutorService pool;

  public ValidationService(int poolSize) {
    pool = Executors.newCachedThreadPool(poolSize);
  }
  // ...
}

The Executors.newCachedThreadPool() method does the following: [[API 06]]

Creates a thread pool that creates new threads as needed, but will reuse previously constructed threads when they are available. These pools will typically improve the performance of programs that execute many short-lived asynchronous tasks. Calls to execute will reuse previously constructed threads if available. If no existing thread is available, a new thread will be created and added to the pool. Threads that have not been used for sixty seconds are terminated and removed from the cache. Thus, a pool that remains idle for long enough will not consume any resources.

Noncompliant Code Example

This noncompliant code example (based on [[Gafter 06]]) shows a BrowserManager class that has several methods that use a fork-join mechanism, that is, they start threads and wait for them to finish. The methods are called in the sequence perUser(), perProfile and perTab(). The method methodInvoker() spawns several instances of the specified runnable depending on the value of the variable numberOfTimes. A fixed sized thread pool is used to execute the enumerations of tasks created at different levels.

public final class BrowserManager {
  private final ExecutorService pool = Executors.newFixedThreadPool(10);
  private final int numberOfTimes;
  private static AtomicInteger count = new AtomicInteger(0);

  public BrowserManager(int n) {
    numberOfTimes = n;
  }

  public void perUser() {  
    methodInvoker(numberOfTimes, "perProfile"); 
    pool.shutdown();
  }

  public void perProfile() {
    methodInvoker(numberOfTimes, "perTab");   
  }

  public void perTab() {  
    methodInvoker(numberOfTimes, "doSomething");
  }

  public void doSomething() {
    System.out.println(count.getAndIncrement());
  }

  public void methodInvoker(int n, final String method) {
    final BrowserManager fm = this;
    Callable<Object> callable = new Callable<Object>() {
      public Object call() throws Exception {
        Method meth = fm.getClass().getMethod(method);
        return meth.invoke(fm);      
      }
    };  

    Collection< Callable< Object>> collection = Collections.nCopies( n, callable); 
    try {
      Collection< Future< Object>> futures = pool.invokeAll(collection);
    } catch (InterruptedException e) {     
      // Forward to handler  
      Thread.currentThread().interrupt(); // Reset interrupted status
    }
    // ... 
  }

  public static void main(String[] args) {
    BrowserManager manager = new BrowserManager(5);
    manager.perUser();
  }
}

Contrary to what is expected, this program does not print the total count, that is, the number of times doSomething() is invoked. This is because it is susceptible to a thread starvation deadlock because the size of the thread pool (10) does not allow either thread from perTab() to invoke the doSomething() method. The output of the program varies for different values of numberOfTimes and the thread pool size. Note that different threads are allowed to invoke doSomething() in different orders; we are concerned only with the maximum value of count to determine how many times the method executed.

Compliant Solution (CallerRunsPolicy)

To prevent thread starvation, every level (worker) must have a double ended queue where all sub-tasks are queued [[Goetz 06]]. Each level removes the most recently generated sub-task from the queue so that it can process it. When there are no more threads left to process, the current level runs the least-recently created sub-task of another level by picking and removing it from that level's queue (work stealing).

This compliant solution selects and schedules tasks for execution, and consequently, avoids the thread starvation deadlock. It sets the CallerRunsPolicy on a ThreadPoolExecutor, and uses a SynchronousQueue [[Gafter 06]]. The policy dictates that if the thread pool runs out of available threads, any subsequent tasks will run in the thread that invoked execute().

public final class BrowserManager {
  private final static ThreadPoolExecutor pool =
    new ThreadPoolExecutor(0, 10, 60L, TimeUnit.SECONDS,
                           new SynchronousQueue<Runnable>());
  private final int numberOfTimes;
  private static AtomicInteger count = new AtomicInteger(0);

  static {
    pool.setRejectedExecutionHandler(
    new ThreadPoolExecutor.CallerRunsPolicy());
  }

  // ... 
}	

According to Goetz et al. [[Goetz 06]]:

A SynchronousQueue is not really a queue at all, but a mechanism for managing handoffs between threads. In order to put an element on the SynchronousQueue, another thread must already be waiting to accept the handoff. It no thread is waiting but the current pool size is less than the maximum, ThreadPoolExecutor creates a new thread; otherwise the task is rejected according to the saturation policy.

According to the Java API class java.util.concurrent.ThreadPoolExecutor.CallerRunsPolicy documentation [[API 06]],:

A handler for rejected tasks that runs the rejected task directly in the calling thread of the execute method, unless the executor has been shut down, in which case the task is discarded.

In this compliant solution, tasks that have other tasks waiting to accept the handoff are added to the SynchronousQueue when the thread pool is full. For example, tasks corresponding to perTab() are added to the SynchronousQueue because the tasks corresponding to perProfile() are waiting to receive the handoff. Once the pool is full, additional tasks are rejected according to the saturation policy in effect. If the CallerRunsPolicy is used to handle these rejected tasks, all the rejected tasks are executed in the main thread that invoked execute(). When all the threads corresponding to perTab() have finished executing, the next set of tasks corresponding to perProfile() are added to the SynchronousQueue because the handoff will be further used by tasks corresponding to perUser(). Consequently, all tasks are executed in bottom-up fashion.

The caller runs policy allows graceful degradation of service when faced with many requests by distributing the workload across the work queue, the underlying abstraction (such as the queue of a TCP/IP connection) and the caller.

This compliant solution is subject to the vagaries of the thread scheduler which may not optimally schedule the tasks, however, it avoids the thread starvation deadlock.

Risk Assessment

Executing interdependent tasks in a thread pool can lead to denial of service.

Automated Detection

TODO

Related Vulnerabilities

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

References

[[API 06]]
[[Gafter 06]] A Thread Pool Puzzler
[[Goetz 06]] 5.3.3 Deques and work stealing

[!The CERT Sun Microsystems Secure Coding Standard for Java^button_arrow_left.png!]      [!The CERT Sun Microsystems Secure Coding Standard for Java^button_arrow_up.png!]      [!The CERT Sun Microsystems Secure Coding Standard for Java^button_arrow_right.png!]