I am having some trouble with a ReetrantLock(true) in an application that checks the difference between two images and shows similarity based on a percentage.
For some reason my input thread for reading in the minimum percent for being "similar" calls lock.lock() and that particular thread just deadlocks indefinitely but I can't seem to figure out why only that thread is deadlocking.
In the SSCCE below in the main method main(String[]) the inline Thread is used to get input from the console and if I type any number in it correctly stores it but as soon as it calls lock.lock() it deadlocks indefinitely and I am unsure why because the ReetrantLock has been told to be fair to caller threads and try to order them for when they called.
private static volatile boolean running = false;
public static void main(String[] args)
{
webcam2 webcam = new webcam2();
webcam.start();
(new Thread(new Runnable()
{
#Override
public void run()
{
Scanner scanner = new Scanner(System.in);
while (running)
{
System.out.print("Enter Double: ");
double val = scanner.nextDouble();
lock.lock(); // locks indefinatly here
if (val < 0.0D) reset = true;
dif = val;
System.out.println("Searching for value: " + dif);
lock.unlock();
}
scanner.close();
}
})).start();
}
private static double dif = 0.0D;
private static boolean reset = false;
private static ReentrantLock lock = new ReentrantLock(true);
#Override
public void run()
{
try
{
while (running)
{
// show image on window
lock.lock();
if (reset == true)
{
reset = false;
lock.unlock();
doComplexStuffToImage();
}
lock.lock();
doComplexStuffToImage();
lock.unlock();
}
}
catch (Exception e)
{
e.printStackTrace();
}
}
private static void doComplexStuffToImage()
{
try
{
Thread.sleep(1000);
}
catch(InterruptedException ie)
{
//swallow exception
}
}
public void start()
{
new Thread(this).start();
running = true;
}
public void stop()
{
running = false;
}
One thread is locking the ReentrantLock and not releasing it. What "reentrant" means is you can call lock many times, but you must call unlock() the same number of times. You lock twice, and unlock once so you are not actually unlocking the lock, thus no other process gets a chance.
Related
I'm trying to implement a simple read/write lock for a resource accessed concurrently by multiple threads. The workers randomly try reading or writing to a shared object. When a read lock is set, workers should not be able to write until the lock is released. When a write lock is set, read and write are not permitted.
Although my implementation seems to work, I believe it is conceptually wrong.
A read operation taking place should allow for more read operations happening at the same time, resulting in the overall number of reads being larger than the number of writes. My program yields numbers that follow the probability of these operations being performed by a worker.
I feel like my implementation is actually not concurrent at all, but I'm having a hard time identifying the mistake. I would really appreciate being pointed in the right direction.
Main class that dispatches and terminates workers:
class Main {
private static final int THREAD_NUMBER = 4;
public static void main(String[] args) {
// creating workers
Thread[] workers = new Thread[THREAD_NUMBER];
for (int i = 0; i < THREAD_NUMBER; i++) {
workers[i] = new Thread(new Worker(i + 1));
}
System.out.println("Spawned workers: " + THREAD_NUMBER);
// starting workers
for (Thread t : workers) {
t.start();
}
try {
Thread.sleep((long) 10000);
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
}
// stopping workers
System.out.println("Stopping workers...");
for (Thread t : workers) {
t.interrupt();
}
}
}
The Resource class:
class Resource {
enum ResourceLock {
ON,
OFF
}
private static Resource instance = null;
private ResourceLock writeLock = ResourceLock.OFF;
private ResourceLock readLock = ResourceLock.OFF;
private Resource() {}
public static synchronized Resource getInstance() {
if (instance == null) {
instance = new Resource();
}
return instance;
}
public ResourceLock getWriteLock() {
return writeLock;
}
public ResourceLock getReadLock() {
return readLock;
}
public void setWriteLock() {
writeLock = ResourceLock.ON;
}
public void setReadLock() {
readLock = ResourceLock.ON;
}
public void releaseWriteLock() {
writeLock = ResourceLock.OFF;
}
public void releaseReadLock() {
readLock = ResourceLock.OFF;
}
}
And finally the Worker class:
import java.util.Random;
class Worker implements Runnable {
private static final double WRITE_PROB = 0.5;
private static Random rand = new Random();
private Resource res;
private int id;
public Worker(int id) {
res = Resource.getInstance();
this.id = id;
}
public void run() {
message("Started.");
while (!Thread.currentThread().isInterrupted()) {
performAction();
}
}
private void message(String msg) {
System.out.println("Worker " + id + ": " + msg);
}
private void read() {
synchronized(res) {
while (res.getWriteLock() == Resource.ResourceLock.ON) {
try {
wait();
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
}
}
res.setReadLock();
// perform read
try {
Thread.sleep((long) 500);
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
}
res.releaseReadLock();
res.notifyAll();
}
message("Finished reading.");
}
private void write() {
synchronized(res) {
while (res.getWriteLock() == Resource.ResourceLock.ON || res.getReadLock() == Resource.ResourceLock.ON) {
try {
wait();
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
}
}
res.setWriteLock();
// perform write
try {
Thread.sleep((long) 500);
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
}
res.releaseWriteLock();
res.notifyAll();
}
message("Finished writing.");
}
private void performAction() {
double r = rand.nextDouble();
if (r <= WRITE_PROB) {
write();
} else {
read();
}
}
}
The reasoning behind having two separate locks for read and write is that I want to have the ability to atomise both operations and their queries for the lock.
Here is an example of the output I'm getting with a 0.5 write probability:
Spawned workers: 4
Worker 2: Started.
Worker 3: Started.
Worker 1: Started.
Worker 4: Started.
Worker 2: Finished writing.
Worker 4: Finished reading.
Worker 1: Finished writing.
Worker 3: Finished writing.
Worker 1: Finished reading.
Worker 4: Finished writing.
Worker 2: Finished reading.
Worker 4: Finished reading.
Worker 1: Finished reading.
Worker 3: Finished writing.
Worker 1: Finished writing.
Worker 4: Finished writing.
Worker 2: Finished writing.
Worker 4: Finished writing.
Worker 1: Finished reading.
Worker 3: Finished writing.
Worker 1: Finished writing.
Worker 4: Finished reading.
Worker 2: Finished writing.
Stopping workers...
Worker 4: Finished writing.
Worker 1: Finished writing.
Worker 3: Finished reading.
Worker 2: Finished reading.
Help much appreciated.
You are performing the entire operation within a synchronized block, so there is no concurrency. Further, there is no precedence towards any lock kind, as at most one thread can own a lock. Not performing the entire operation in a synchronized block won’t work with your current code, as every reader does a readLock = ResourceLock.OFF at the end, regardless of how many readers are there. Without a counter, you can’t support multiple readers correctly.
Besides that, it’s a strange code structure, to provide a Resource class maintaining the state but leaving it entirely up to the callers to do the right thing with it. That’s not the way to deal with responsibility and encapsulation.
An implementation may look like
class ReadWriteLock {
static final int WRITE_LOCKED = -1, FREE = 0;
private int numberOfReaders = FREE;
private Thread currentWriteLockOwner;
public synchronized void acquireReadLock() throws InterruptedException {
while(numberOfReaders == WRITE_LOCKED) wait();
numberOfReaders++;
}
public synchronized void releaseReadLock() {
if(numberOfReaders <= 0) throw new IllegalMonitorStateException();
numberOfReaders--;
if(numberOfReaders == FREE) notifyAll();
}
public synchronized void acquireWriteLock() throws InterruptedException {
while(numberOfReaders != FREE) wait();
numberOfReaders = WRITE_LOCKED;
currentWriteLockOwner = Thread.currentThread();
}
public synchronized void releaseWriteLock() {
if(numberOfReaders!=WRITE_LOCKED || currentWriteLockOwner!=Thread.currentThread())
throw new IllegalMonitorStateException();
numberOfReaders = FREE;
currentWriteLockOwner = null;
notifyAll();
}
}
It simply uses a counter of acquired read locks, setting the counter to -1 when there is a write lock (so write locks can not be nested). Acquiring a read lock may succeed whenever there is no write lock, so there is no need to implement precedence for them, the possibility to succeed when another thread already has a real lock, is sufficient. In fact, when having a significantly larger number of readers than writers, you may encounter the “starving writer” problem.
The worker simplifies to
class Worker implements Runnable {
private static final double WRITE_PROB = 0.5;
private static final Random rand = new Random();
private final ReadWriteLock theLock;
private final int id;
public Worker(int id, ReadWriteLock lock) {
theLock = lock;
this.id = id;
}
public void run() {
message("Started.");
while(!Thread.currentThread().isInterrupted()) {
performAction();
}
}
private void message(String msg) {
System.out.println("Worker " + id + ": " + msg);
}
private void read() {
try {
theLock.acquireReadLock();
} catch(InterruptedException e) {
Thread.currentThread().interrupt();
return;
}
// perform read
try {
Thread.sleep(500);
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
}
finally { theLock.releaseReadLock(); }
message("Finished reading.");
}
private void write() {
try {
theLock.acquireWriteLock();
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
return;
}
// perform write
try {
Thread.sleep(500);
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
}
finally { theLock.releaseWriteLock(); }
message("Finished writing.");
}
private void performAction() {
double r = rand.nextDouble();
if (r <= WRITE_PROB) {
write();
} else {
read();
}
}
}
Note that I avoided global variables here. The lock should get passed to the constructor. It’s also important that the methods return when being interrupted during the lock acquisition. Self interrupting and retrying the acquisition like in your original code will lead to an infinite loop, as the next wait would again throw an InterruptedException after you restored the current thread’s interrupted state. Of course, proceeding without having the lock would be wrong too, so the only valid options are not restoring the interrupted state or returning immediately.
The only change to your main program is to construct a pass the lock instance:
ReadWriteLock sharedLock = new ReadWriteLock();
// creating workers
Thread[] workers = new Thread[THREAD_NUMBER];
for (int i = 0; i < THREAD_NUMBER; i++) {
workers[i] = new Thread(new Worker(i + 1, sharedLock));
}
System.out.println("Spawned workers: " + THREAD_NUMBER);
// starting workers
for (Thread t : workers) {
t.start();
}
try {
Thread.sleep(10000);
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
}
// stopping workers
System.out.println("Stopping workers...");
for (Thread t : workers) {
t.interrupt();
}
This is the simple implementation for ReadWriteLock with more priority given to write operation:
public class ReadWriteLock{
private int readers = 0;
private int writers = 0;
private int writeRequests = 0;
public synchronized void lockRead() throws InterruptedException{
while(writers > 0 || writeRequests > 0){
wait();
}
readers++;
}
public synchronized void unlockRead(){
readers--;
notifyAll();
}
public synchronized void lockWrite() throws InterruptedException{
writeRequests++;
while(readers > 0 || writers > 0){
wait();
}
writeRequests--;
writers++;
}
public synchronized void unlockWrite() throws InterruptedException{
writers--;
notifyAll();
}
}
Source: http://tutorials.jenkov.com/java-concurrency/read-write-locks.html
First of all, this is not a homework.
I have written a piece of code so that:
Thread-1 prints 1,4,7,... (diff is 3)
Thread-2 prints 2,5,8,...
Thread-3 prints 3,6,9,...
And the final output should be:
1,2,3,4,5,6,7,8,9,...
Here's the code that works wonderfully well:
package threadAlgo;
public class ControlOrder {
volatile Monitor monitor = new Monitor();
public static void main(String[] args) {
ControlOrder order = new ControlOrder();
Thread one = new Thread(new Task(order.monitor, 1));
one.setName("Thread-1");
Thread two = new Thread(new Task(order.monitor, 2));
two.setName("Thread-2");
Thread three = new Thread(new Task(order.monitor, 3));
three.setName("Thread-3");
one.start();
two.start();
three.start();
}
}
class Monitor {
int threadNumber = 1;
}
class Task implements Runnable {
private Monitor monitor;
private int myThreadNumber;
private int currentCount;
Task(Monitor monitor, int myThreadNumber) {
this.monitor = monitor;
this.myThreadNumber = myThreadNumber;
this.currentCount = myThreadNumber;
}
#Override
public void run() {
while (true) {
while (monitor.threadNumber != myThreadNumber) {
synchronized (monitor) {
try {
monitor.wait(100); //DOESN'T WORK WITHOUT THE TIMEOUT!!!
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
synchronized (monitor) {
if (monitor.threadNumber == myThreadNumber) {
System.out.println(Thread.currentThread().getName() + ": " + currentCount);
currentCount = currentCount + 3;
}
try {
Thread.sleep(500);
} catch (InterruptedException e) {
e.printStackTrace();
}
if (myThreadNumber == 3) {
monitor.threadNumber = 1;
} else {
monitor.threadNumber = myThreadNumber + 1;
}
monitor.notifyAll();
}
}
}
}
The only problem is that if I use wait() instead of wait(timeout), then the thread halts.
UPDATE:
Wait condition (while loop) should be inside synchronized block. A lesson for beginners, including me.
You should always
perform notifyAll/notify in conjunction with a change in state.
check the state change before using wait() in a loop.
If you call notify() and no wait() is waiting, then the signal is lost, so unless you check a state change, (or timeout) you can block forever waiting for a signal which doesn't change.
Say, I've got a javafx.concurrent.Task nested into Thread, ie.:
Task task = new Task();
Thread thread = new Thread(task);
thread.start();
How can I in this situation pause and/or stop executing task nad resume its work?
There is no easy way, except using the deprecated suspend() and resume() methods on the Thread class.
In case you are sure your task doesn't enter synchronized code that would work
Otherwise you would have to have halt points in your task were you check if the Task has been halted and if so call wait() on some object to block the thread. A call to notify on the object would then wake the Thread up and resume. Below is a chunk of pseudo code for that approach.
Note that for this to work as expected you need check the halt variable frequently in the task code.
class MyTask{
volatile boolean halt = false;
Object o = new Object();
public void run(){
while(notDone) {
if (halt) halt();
}
}
private halt(){
synchronized (o){o.wait()}
}
public resume(){
halt = false;
synchronized (o){o.notify()}
}
public suspend(){
halt=true;
}
}
import java.io.IOException;
public class TestThread {
static class PausableRunnable implements Runnable{
volatile boolean shouldHalt = false;
private final Object lock=new Object();
public void run(){
while(true){
if(shouldHalt)halt();
try {
Thread.sleep(500);
} catch (InterruptedException e) {
e.printStackTrace();
}
System.out.print(".");
}
}
private void halt(){
synchronized (lock){
try {
lock.wait();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
void pause(){
shouldHalt = true;
}
void resume(){
synchronized (lock){
shouldHalt=false;
lock.notify();
}
}
}
public static void maipn(String[] args) throws IOException {
PausableRunnable pr = new PausableRunnable();
Thread t = new Thread(pr);
t.start();
while(true) {
char c = (char) System.in.read();
if (c == 'p') {
System.out.println("Pausing");
pr.pause();
}
if (c == 'r') {
System.out.println("Resuming");
pr.resume();
}
}
}
}
The following code was summed up the application, the application randomly was locked in
while (flag)
This code, running on my machine gets caught, in another machine he finished normally
The output generated here is:
INIT
END
before while
before flag
after flag
Code:
package threads;
public class Run implements Runnable {
private Thread thread;
private boolean flag = true;
public void init() {
thread = new Thread(this);
thread.setName("MyThread");
thread.start();
}
#Override
public void run() {
try {
int i = 0;
while (i < 1000) {
i++;
}
System.out.println("before flag");
flag = false;
System.out.println("after flag");
} catch (Exception e) {
e.printStackTrace();
} finally {
flag = false;
}
}
public void end() {
thread.interrupt();
thread = null;
System.out.println("before while");
while (flag) {
// try { Thread.sleep(100);} catch (InterruptedException e) {}
}
;
System.out.println("after while");
}
public static void main(String[] args) {
Run r = new Run();
System.out.println("INIT");
r.init();
System.out.println("END");
r.end();
}
}
Why when I change the value of flag the main thread does not pass through loop?
Change
private boolean flag = true;
to
private volatile boolean flag = true;
Without volatile, there is no guarantee the waiting thread needs to see the value get updated. HotSpot might even inline while(flag) to while(true) if the loop spins enough times.
See Memory Consistency Errors.
Also, what you're doing is called a spinlock. Normally you should use thread.join() instead. A spinlock is wasteful of resources because the waiting thread is actually working (checking a variable) the entire time it is supposed to be waiting.
I am currently working on a project where I am to have essentially 10 threads that are "sleeping". At random one of these 10 threads is to "wake up" and start doing some work. I just want to see if I am headed in the right direction. So should I just create each instance of the thread for instance.
Thread thread0 = new Thread(new doWork());
...
Thread thread9 = new Thread(new doWork());
and just not start them and then when they are to "wake" just call the start() method on the particular thread..
or should I start each thread but have them wait() until I call the notify() method?
or should I start the thread and use sleep() and then call the interrupt() method?
Which approach seems to be better and why?
Any insight is greatly appreciated.
edit Will this be acceptable??
import java.util.Random;
public class Client {
private static Thread [] clients = new Thread[10];
public static void main(String[] args){
createClients();
randomWake();
}// end main()
static void createClients(){
Thread client0 = new Thread(new ClientThread(0));
clients[0] = client0;
Thread client1 = new Thread(new ClientThread(1));
clients[1] = client1;
Thread client2 = new Thread(new ClientThread(2));
clients[2] = client2;
Thread client3 = new Thread(new ClientThread(3));
clients[3] = client3;
Thread client4 = new Thread(new ClientThread(4));
clients[4] = client4;
Thread client5 = new Thread(new ClientThread(5));
clients[5] = client5;
Thread client6 = new Thread(new ClientThread(6));
clients[6] = client6;
Thread client7 = new Thread(new ClientThread(7));
clients[7] = client7;
Thread client8 = new Thread(new ClientThread(8));
clients[8] = client8;
Thread client9 = new Thread(new ClientThread(9));
clients[9] = client9;
for(int i = 0; i < clients.length; i++)
clients[i].start();
}// end createClients()
static void randomWake(){
Random rand = new Random();
int randomNumber = rand.nextInt(10);
clients[randomNumber].interrupt();
}// end randomWake()
static class ClientThread implements Runnable{
private int clientNumber;
public ClientThread(int clientNumber){
this.clientNumber = clientNumber;
}// end ClientThread(int clientNumber)
public void run(){
while(!Thread.interrupted()){}
System.out.println("Client " + clientNumber + " is awake!");
}// end run()
}// end class ClientThread
}// end class Client
In case there is a maximum amount of sleep time
You probably will need to implement the following Thread class:
public class DoWork extends Thread {
public void run () {
while(true) {
Thread.Sleep((int) Math.floor(Math.random()*10000));
//do some work
}
}
}
Where 10000 is the maximum time in milliseconds a thread should sleep.
In case there is no maximum amount of sleep time
You probably will need to implement the following Thread class:
public class DoWork extends Thread {
public void run () {
while(true) {
Thread.Sleep(1);
if(Math.random() < 0.005d) {
//do some work
}
}
}
}
where 0.005 is the probability of running the method a certain millisecond.
notify and wait are used to implement Semaphores: this are objects that prevent two threads to manipulate the same object at the same time (since some objects could end up in an illegal state).
How about using semaphores?
class DoWork extends Runnable {
private final Semaphore semaphore;
DoWork(Semaphore semaphore) {
this.semaphore = semaphore;
}
#Override
public void run() {
while (true) {
semaphore.acquire();
//do some work
}
}
}
The main program can create an array of Semaphores, and an equal number of Threads running DoWork instances, so that each DoWork instance has its own semaphore. Each time the main program calls sema[i].release(), The run() method of the corresponding DoWork instance will "do some work" and then go back to waiting.
It doesn't make much sense your answer, so not sure what you really want to achieve. But for what you describe you should put all threads waiting on the same lock and just notify the lock (it will awake only one randomly)
But as that doesn't make much sense, I guess you want to achieve something different.
Check this question regarding sleep vs wait: Difference between wait() and sleep()
Check this one. This is how I would solve it if I were not to use ThreadPooling (which is very correct as the others have said) and so that I can see how wait(),notify() and Thread.sleep() work. Checking google you will see (e.g. Thread.sleep and object.wait) that the mainly wait() and notify() are used for communication between threads and Thread.sleep is used so that you can pause your program.
-Part of this answer is based on this: http://tutorials.jenkov.com/java-concurrency/thread-signaling.html#missedsignals. You can check in the code to see the steps that you need to take (comment out some parts of the code) in order to make your program hang, so that you realize how to work with missed signals. The iterations needed for your program to hang are not fixed.
-The programm will run forever. You will need to work on it a bit in order to fix that.
Main
public class Main
{
public static void main(String[] args)
{
Manager mgr = new Manager("manager");
mgr.start();
}
}
Manager
public class Manager extends Thread
{
private final Object lock = new Object();
private boolean wasSignalled = false;
private DoWork[] workThreads = new DoWork[5];
public Manager(String name){
super(name);
workThreads[0] = new DoWork(this,"work 0");
workThreads[1] = new DoWork(this,"work 1");
workThreads[2] = new DoWork(this,"work 2");
workThreads[3] = new DoWork(this,"work 3");
workThreads[4] = new DoWork(this,"work 4");
}
public void wakeUP()
{
synchronized (this.lock) {
wasSignalled = true;
this.lock.notify();
}
}
public void pauseAndWait()
{
synchronized (this.lock) {
if(!wasSignalled)
{
try {
this.lock.wait();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
//clear signal and continue running.
wasSignalled = false;
}
}
public void run ()
{
int i=0;
while(true)
{
i++;
System.out.println(" manager ...: "+i+" ");
int choose = 0 + (int)(Math.random() * ((4 - 0) + 1));
//choose=0; for debugginng
if(!workThreads[choose].isAlive()){
workThreads[choose].start();
}
else{
workThreads[choose].wakeUP();
}
//wait to be notified by DoWork thread when its job
//is done
pauseAndWait();
}
}
}
DoWork
public class DoWork extends Thread
{
private final Object lock = new Object();
private boolean wasSignalled = false;
private Manager managerThread;
public DoWork(Manager managerThread,String name){
super(name);
this.managerThread=managerThread;
}
public void wakeUP()
{
synchronized (this.lock) {
//check what happens without wasSignalled flag
//step #1: comment out wasSignalled = true;
wasSignalled = true;
this.lock.notify();
}
}
public void pauseAndWait()
{
synchronized (this.lock) {
//check what happens without wasSignalled flag
//step #2: comment out the if block
if(!wasSignalled)
{
try {
this.lock.wait();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
//check what happens without wasSignalled flag
//step #3: comment out wasSignalled = false;
//clear signal and continue running.
wasSignalled = false;
}
}
public void run ()
{
int i=0;
while(true)
{
i++;
try {
System.out.print(this.getName()+" going to sleep ...: "+i+" ");
//check what happens without wasSignalled flag
//step #4: put sleep time to Thread.sleep(0);
//simulate worker thread job
Thread.sleep(1000);
System.out.println(" woke up ... ");
} catch (InterruptedException e) {
System.out.println(" worker thread: job simulation error:"+e);
}
//if worker thread job simulation is done (sleep for 4 times)
//then suspend thread and wait to be awaken again
if(i>4)
{
System.out.println(this.getName()+" notifying main ...: "+i+" \n");
i=0;
managerThread.wakeUP();
// thread does not get destroyed, it stays in memory and when the manager
// thread calls it again it will wake up do its job again
pauseAndWait();
}
}
}
}