Reputation: 1283
I am not familiar with threads and concurrent programming. I was looking for a simple snippet which would result in a deadlock, here it is :
public class TestLock {
private static class fun {
int a,b;
void read() {System.out.println(a+b);}
void write(int a,int b) {this.a=a;this.b=b;}
}
public static void main (String[] args) throws java.lang.Exception {
final fun d1=new fun();
final fun d2=new fun();
Thread t1=new Thread() {
public void run() {
for(int i=0;i<5;i++) {
synchronized(d2) {
d2.read();
try {
Thread.sleep(50);
} catch (Exception ex) {
ex.printStackTrace();
}
synchronized(d1) {
d1.write(i, i);
}
}
}
};
Thread t2=new Thread() {
public void run() {
for(int i=0;i<5;i++) {
synchronized(d1) {
d1.read();
try {
Thread.sleep(50);
} catch (Exception ex) {
ex.printStackTrace();
}
synchronized(d2) {
d2.write(i, i);
}
}
}
}
};
t1.start();
t2.start();
}
}
Now I was wondering how I could transform this example, using ReentrantLock instead of synchronized, but I don't see how : does fun need to have a ReentrantLock attribute in order to have something like
Thread t1=new Thread() {
public void run() {
for(int i=0;i<5;i++) {
if(d2.lock.tryLock()) {
try {d1.read();Thread.sleep(50);} catch(Exception e) {e.printStackTrace();} finally {d1.lock.unlock();}
if(d2.lock.tryLock()) {
try {d2.write(i, i);} catch(Exception e) {e.printStackTrace();} finally {d2.lock.unlock();}
}
}
}
}
};
or am I missing something entirely ?
Upvotes: 1
Views: 205
Reputation: 691715
Transforming the example using ReentrantLocks would indeed mean using two locks: one associated with d1
and the other one associated with d2
.
And you would replace every entrance in a synchronized block on dX
by a call to lockX.lock()
, and any exit from a synchronized block on dX
by a call to lockX.unlock()`.
Using tryLock()
defeats the purpose, since it returns instead of waiting if the lock can't be acquired.
Upvotes: 1