Java Puzzler: busy wait threads stop working - java

This is some sort of a Java Puzzler, that I stumbled across and can't really explain. Maybe somebody can?
The following program hangs after a short time. Sometimes after 2 outputs, sometimes after 80, but almost always before terminating correctly. You might have to run it a few times, if it doesn't happen the first time.
public class Main {
public static void main(String[] args) {
final WorkerThread[] threads = new WorkerThread[]{ new WorkerThread("Ping!"), new WorkerThread("Pong!") };
threads[0].start();
threads[1].start();
Runnable work = new Runnable() {
private int counter = 0;
public void run() {
System.out.println(counter + " : " + Thread.currentThread().getName());
threads[counter++ % 2].setWork(this);
if (counter == 100) {
System.exit(1);
}
}
};
work.run();
}
}
class WorkerThread extends Thread {
private Runnable workToDo;
public WorkerThread(String name) {
super(name);
}
#Override
public void run() {
while (true){
if (workToDo != null) {
workToDo.run();
workToDo = null;
}
}
}
public void setWork(Runnable newWork) {
this.workToDo = newWork;
}
}
Now, it's clear that busy waiting loops are not a great idea in general. But this not about improving, it's about understanding what is happening.
Since everything works as expected when WorkerThread.setWork() is synchronized or when the WorkerThread.workToDo field is set to volatile I suspect a memory issue.
But why exactly is it happening? Debugging doesn't help, once you start stepping through, everything behaves as expected.
An explanation would be appreciated.

The first problem is that you are setting the Runnable workToDo from the main thread and then reading it in the 2 forked threads without synchronization. Any time you modify a field in multiple threads, it should be marked as volatile or someone synchronized.
private volatile Runnable workToDo;
Also, because multiple threads are doing counter++ this also needs to be synchronized. I recommend an AtomicInteger for that.
private AtomicInteger counter = new AtomicInteger(0);
...
threads[counter.incrementAndGet() % 2].setWork(this);
But I think the real problem may be one of race conditions. It is possible for both threads to set the workToDo to be the Runnable and then have them both return and set it back to be null so they will just spin forever. I'm not sure how to fix that.
1. threads[0] has it's `workToDo` set to the runnable. It calls `run()`.
2. at the same time threads[1] also calls `run()`.
3. threads[0] sets the `workToDo` on itself and threads[1] to be the runnable.
4. at the same time threads[1] does the same thing.
5. threads[0] returns from the `run()` method and sets `workToDo` to be `null`.
6. threads[1] returns from the `run()` method and sets `workToDo` to be `null`.
7. They spin forever...
And, as you mention, the spin loop is crazy but I assume this is a demonstration thread program.

The problem occurs right between these lines:
workToDo.run();
workToDo = null;
Suppose the following sequence of events occurs:
- Original Runnable runs. "Ping!".setWork() called
- Ping! thread realizes workToDo != null, calls run(), the stops between those two lines
- "Pong!".setWork() called
- Pong! thread realizes workToDo != null, calls run()
- "Ping!".setWork() called
- Ping! thread resumes, sets workToDo = null, ignorantly discarding the new value
- Both threads now have workToDo = null, and the counter is frozen at 2,...,80
Program hangs

my 2 cents....
import java.util.concurrent.atomic.AtomicReference;
class WorkerThread extends Thread {
private AtomicReference<Runnable> work;
public WorkerThread(String name) {
super(name);
work = new AtomicReference<Runnable>();
}
#Override
public void run() {
while (true){
Runnable workToDo = work.getAndSet(null);
if ( workToDo != null ) {
workToDo.run();
}
}
}
public void setWork(Runnable newWork) {
this.work.set(newWork);
}
}

Related

Why thread still locks after synchronised method is executed?

I'm reading java multithreading tutorial which says thread only gives up key until it completes synchronised method, however when I run the following code (about 20 times):
public class SyncDemo implements Runnable{
#Override
public void run() {
for (int i = 0; i < 10; i++) {
sync();
}
}
private synchronized void sync() {
System.out.println(Thread.currentThread().getName());
}
public static void main(String[] args) {
SyncDemo s = new SyncDemo();
Thread a = new Thread(s, "a");
Thread b = new Thread(s, "b");
a.start();
b.start();
}
}
it only prints a then b, which I expect a mixed sequence of them because current thread will unlock every time after sync() is executed inside the loop? Thus giving the other thread a chance to print its name?
There is nothing in your program that would demand a certain execution order. So the run time will schedule the threads in a way that makes most sense in the current situation. Factors that may influence the order are number of processors, load situation, ...

Unexpected thread behavior. Visibility

I have the following code:
public static boolean turn = true;
public static void main(String[] args) {
Runnable r1 = new Runnable() {
public void run() {
while (true) {
while (turn) {
System.out.print("a");
turn = false;
}
}
}
};
Runnable r2 = new Runnable() {
public void run() {
while (true) {
while (!turn) {
System.out.print("b");
turn = true;
}
}
}
};
Thread t1 = new Thread(r1);
Thread t2 = new Thread(r2);
t1.start();
t2.start();
}
In class we've learned about "Visibility" problems that may occur when using un-synchronized code.
I understand that in order to save time, the compiler will decide the grab turn to the cache in the CPU for the loop, meaning that the thread will not be aware if the turn value was changed in the RAM because he doesn't check it.
From what I understand, I would expected the code to run like this:
T1 will see turn as true -> enter loop and print -> change turn to false -> gets stuck
T2 will think turn hasn't changed -> will get stuck
I would expect that if T1 will start before T2: only 'a' will be printed and both threads will run in an infinite loop without printing anything else
However, when I'm running the code sometimes I get a few "ababa...." before both threads will stuck.
What am I missing ?
EDIT:
The following code does what I expect it: the thread will run in a infinite loop:
public class Test extends Thread {
boolean keepRunning = true;
public void run() {
long count = 0;
while (keepRunning) {
count++;
}
System.out.println("Thread terminated." + count);
}
public static void main(String[] args) throws InterruptedException {
Test t = new Test();
t.start();
Thread.sleep(1000);
t.keepRunning = false;
System.out.println("keepRunning set to false.");
}
}
How are they different from each other ?
When I run the code, sometimes I get a few "ababa...." before both threads will stuck.
I suspect that what is happening is that the behavior is changing when the code is JIT compiled. Before JIT compilation the writes are visible because the interpreter is doing write-throughs. After JIT compilation, either the cache flushes or the reads have been optimized away ... because the memory model allows this.
What am I missing ?
What you are missing is that you are expecting unspecified behavior to be consistent. It doesn't have to be. After all, it is unspecified! (This is true, even if my proposed explanation above is incorrect.)
The fact that turn isn't volatile doesn't mean that your code WILL break, just that it MIGHT break. For all we know, the thread could see false or true at any given moment. Caches could just be randomly flushed for no reason in particular, the thread could retain its cache, etc etc.
It could be because your code is experiencing side effects from System.out.print, which internally writes to a synchronized method:
521 private void write(String s) {
522 try {
523 synchronized (this) {
(Source - DocJar)
The memory effects of synchronized could be flushing the cache and therefore impact your code.
As #Stephen C said, it could also be the JIT, which might hoist the boolean check because it assumes that the value can't change due to another thread.
So out of the three different possibilities mentioned so far, they could all be factors to contribute to how your code behaves. Visibility is a factor, not a determiner.

java threads alternative to synchronization

I am new to concurrent programming and I am facing few issues with the below code using Java threads.
Status Class (this class tracks the position availability):
public class Status {
private static Map<String, Boolean> positions = new HashMap<>();
static {
//Initially all positions are free (so set to true)
positions.put("A", true);
positions.put("B", true);
}
public synchronized void occupyOrClear(String position,
boolean status) {
positions.put(position, status);
}
public boolean isClear(String position) {
return positions.get(position);
}
}
MyThread Class:
public class MyThread implements Runnable {
private String[] positions;
private String customer;
public MyThread(String customer, String[] positions) {
this.positions = positions;
this.customer = customer;
}
private Status status = new Status();
public void run() {
for (int i = 0; i < positions.length;) {
String position = positions[i];
if (status.isClear(position)) {
// position occupied now
status.occupyOrClear(position, false);
System.out.println(position + " occupied by :"+customer);
try {
//my real application logic goes below (instead of sleep)
Thread.sleep(2000);
} catch (InterruptedException inteExe) {
System.out.println(" Thread interrupted ");
}
// Now clear the position
status.occupyOrClear(position, true);
System.out.println(position + " finished & cleared by:"+customer);
i++;
} else {
try {
Thread.sleep(1000);
} catch (InterruptedException inteExe) {
System.out.println(" Thread interrupted ");
}
}
}
}
}
ThreadTest Class:
public class ThreadTest {
public static void main(String[] args) {
String[] positions = { "A", "B"};
Status status = new Status();
Thread customerThread1 = new Thread(new MyThread(status, "customer1", positions));
Thread customerThread2 = new Thread(new MyThread(status, "customer2", positions));
Thread customerThread3 = new Thread(new MyThread(status, "customer3", positions));
customerThread1.start();
customerThread2.start();
customerThread3.start();
}
}
Even though I have used 'synchronized' I could notice that some times Thread3 is picking up prior to Thread2 and could you please help me to resolve this issue and to acheive the following results ?
(1) Always customerThread1 should take the positions first and then
followed by customerThread2 and then customerThread3 (etc...)
(2) As soon as the A's position is freed by customerThread1, the
position should be immediately picked up by customerThread2 (rather
than customerThread2 and customerThread3 waiting till all positions
are done by customerThread1).And as soon as customerThread2 finishes
position 'A', then customerThread3 should pick it up, etc..
(3) As soon as the position (A, B, etc..) is freed/available, the next
customerThread should pick it up immediately.
(4) The solution should avoid all race conditions
There are several fundamental problems.
You have broken code and already noticed that it doesn’t work. But instead of asking how to fix that broken code, you are asking for alternatives with higher performance. You will never manage to write working programs with that attitude.
Apparently, you have no idea, what synchronized does. It acquires a lock on a particular object instance which can be held by one thread only. Therefore, all code fragments synchronizing on the same object are enforced to be executed ordered, with the necessary memory visibility. So your code fails for two reasons:
You are creating multiple instances of Status accessing the same objects referenced by a static variable. Since all threads use different locks, this access is entirely unsafe.
Your occupyOrClear is declared synchronized, but your method isClear is not. So even if all threads were using the same lock instance for occupyOrClear, the result of isClear remained unpredictable due to its unsafe access to the map.
You have code of the form
if(status.isClear(position)) { status.occupyOrClear(position, false); …
which matches the check-then-act anti-pattern. Even if each of these two method calls were thread-safe, this sequence still remained unsafe, because between these two invocations, everything can happen, most notably, the condition, the thread just checked, may change without the thread noticing. So two or more threads could invoke isClear, receiving true and then proceed with occupyOrClear.
You are using Thread.sleep.
You can try with the following pseudocode:
main() {
//some concurrent queues, eg ConcurrentLinkedQueue
Queue t1Tasks = new Queue("A","B","C");
Queue t2Tasks = new Queue();
Queue t3Tasks = new Queue();
Thread t1 = new PThread(t1Tasks,t2Tasks,"customer1");
Thread t2 = new PThread(t2Tasks,t3Tasks,"customer2");
Thread t3 = new PThread(t3Tasks,null,"customer3");
}
PThread {
Queue q1,q2;
PThread(Queue q1, Queue q2,...){}
run() {
while (item = q1.get()) {
//process item
q2.put(item); //to be processed by next thread
}
}
}

why is this thread safe?

Because it always prints out '3'. No synchronization needed? I am testing this simple thing because I am having a trouble in a real multiple thread problem, which isn't good to illustrate the problem, because it's large. This is a simplified version to showcase the situation.
class Test {
public static int count = 0;
class CountThread extends Thread {
public void run()
{
count++;
}
}
public void add(){
CountThread a = new CountThread();
CountThread b = new CountThread();
CountThread c = new CountThread();
a.start();
b.start();
c.start();
try {
a.join();
b.join();
c.join();
} catch (InterruptedException ex) {
ex.printStackTrace();
}
}
public static void main(String[] args) {
Test test = new Test();
System.out.println("START = " + Test.count);
test.add();
System.out.println("END: Account balance = " + Test.count);
}
Because it always prints out '3'. No synchronization needed?
It is not thread safe and you are just getting lucky. If you run this 1000 times, or on different architectures, you will see different output -- i.e. not 3.
I would suggest using AtomicInteger instead of a static field ++ which is not synchronized.
public static AtomicInteger count = new AtomicInteger();
...
public void run() {
count.incrementAndGet();
}
...
Seems to me like count++ is fast enough to finish until you invoke 'run' for the other class. So basically it runs sequential.
But, if this was a real life example, and two different threads were usingCountThread parallelly, then yes, you would have synchronization problem.
To verify that, you can try to print some test output before count++ and after, then you'll see if b.start() is invoking count++ before a.start() finished. Same for c.start().
Consider using AtomicInteger instead, which is way better than synchronizing when possible -
incrementAndGet
public final int incrementAndGet()
Atomically increments by one the current value.
This code is not thread-safe:
public static int count = 0;
class CountThread extends Thread {
public void run()
{
count++;
}
}
You can run this code a million times on one system and it might pass every time. This does not mean is it is thread-safe.
Consider a system where the value in count is copied to multiple processor caches. They all might be updated independently before something forces one of the caches to be copied back to main RAM. Consider that ++ is not an atomic operation. The order of reading and writing of count may cause data to be lost.
The correct way to implement this code (using Java 5 and above):
public static java.util.concurrent.atomic.AtomicInteger count =
new java.util.concurrent.atomic.AtomicInteger();
class CountThread extends Thread {
public void run()
{
count.incrementAndGet();
}
}
It's not thread safe just because the output is right. Creating a thread causes a lot of overhead on the OS side of things, and after that it's just to be expected that that single line of code will be done within a single timeslice. It's not thread safe by any means, just not enough potential conflicts to actually trigger one.
It is not thread safe.
It just happened to be way to short to have measurable chance to show the issue. Consider counting to much higher number (1000000?) in run to increase chance of 2 operations on multiple threads to overlap.
Also make sure your machine is not single core CPU...
To make the class threadsafe either make count volatile to force memory fences between threads, or use AtomicInteger, or rewrite like this (my preference):
class CountThread extends Thread {
private static final Object lock = new Object();
public void run()
{
synchronized(lock) {
count++;
}
}
}

Is a notify signalled on thread finish? Why does this code sample work?

I am looking in some puzzles for threads and I can't figure out why the following consistently prints 999999:
class Job extends Thread {
private Integer number = 0;
public void run() {
for (int i = 1; i < 1000000; i++) {
number++;
}
}
public Integer getNumber() {
return number;
}
}
public class Test {
public static void main(String[] args)
throws InterruptedException {
Job thread = new Job();
thread.start();
synchronized (thread) {
thread.wait();
}
System.out.println(thread.getNumber());
}
}
There is no notify on the same lock (and spurious wakeup seem to be ignored).
If a thread finishes does a notify get signalled or something?
How come main prints the result and not get "stuck" waiting?
In the Javadoc for Java 7 Thread.join(long)
This implementation uses a loop of this.wait calls conditioned on this.isAlive. As a thread terminates the this.notifyAll method is invoked. It is recommended that applications not use wait, notify, or notifyAll on Thread instances.
Using a Thread directly this way is considered bad practical. Note: wait() could end for any number of reasons, possibly spuriously.
Based on a puzzler related to #Voo's comment. The point is you shouldn't play with the internal behaviour of Thread as this is more likely to lead to confusion.
public static String getName() {
return "MyProgram";
}
public static void main(String... args) {
new Thread() {
public void run() {
System.out.println("My program is " + getName());
}
}.start();
}
What does this program print?
For clarification, I have modified your code to this:
Job thread = new Job();
thread.start();
final Object lock = new Object();
synchronized (lock) { lock.wait(); }
System.out.println(thread.getNumber());
Now it blocks. That's a first-hand confirmation of what #Nitram has explained in his answer. If you care to have a look at the Thread implementation code, it will be quite obvious why this is the observed behavior.
NOTE: This answer has been edited extensively.
The reason for this behaviour is, that "someone" is calling notifyAll internally. This "someone" is the JVM itself as you can "see" in the C sources here:
http://hg.openjdk.java.net/jdk7/hotspot/hotspot/file/f95d63e2154a/src/share/vm/runtime/thread.cpp
In line 1531 the method ensure_join calls notifyAll. This is the counterpart to the wait calls in java.lang.Thread#join (as noted by Marko and others).
ensure_join in turn is called in line 1664 in the method JavaThread::exit.
Since this is "internal bookkeeping" nobody should rely on this behaviour.
Simply put a Thread notifies all waiting threads once the execution of the threads ends. Its not the proposed why to do this, but it works. To synchronize on the end of a thread rather use Thread.join.
A thread object is automatically notified when the thread finishes, that's why the main thread doesn't get stuck.
Well....notify serves purpose of premature notifying to the threads waiting on the locked object. if you don't use Notify ,then certainly when it finishes it releases lock.
So that is equivalent to notify
no it's not..Consider the situation below.
class Job extends Thread {
private Integer number = 0;
public void run() {
synchronized(this) {
for (int i = 1; i < 1000000; i++) {
number++;
}
notify(); //releases lock here and your main thread continues
do sumthing...
do sumthing...
}
}
public Integer getNumber() {
return number;
}
}
if you don't use notify() ...lock will be released only after you do all your sumthings..

Categories

Resources