Reputation: 278
I'm busy looking at the filter lock algorithm for n-thread mutual exclusion and I can't seem to understand line 17 of the code. I understand that it is spinning on a condition but not entirely sure what those conditions are. More specifically what (∃k != me) entails.
1 class Filter implements Lock {
2 int[] level;
3 int[] victim;
4 public Filter(int n) {
5 level = new int[n];
6 victim = new int[n]; // use 1..n-1
7 for (int i = 0; i < n; i++) {
8 level[i] = 0;
9 }
10 }
11 public void lock() {
12 int me = ThreadID.get();
13 for (int i = 1; i < n; i++) { //attempt level 1
14 level[me] = i;
15 victim[i] = me;
16 // spin while conflicts exist
17 while ((∃k != me) (level[k] >= i && victim[i] == me)) {};
18 }
19 }
20 public void unlock() {
21 int me = ThreadID.get();
22 level[me] = 0;
23 }
24 }
Upvotes: 9
Views: 7446
Reputation: 427
I can show you my implementation. The Filter class:
package org.example.ch2.example4;
import org.example.ch2.Lock;
import org.example.ch2.ThreadID;
public class Filter implements Lock {
private volatile int[] level;
private volatile int[] victim;
Filter(int n) {
level = new int[n];
victim = new int[n];
for (int i = 0; i < n; i++) {
level[i] = 0;
}
}
@Override
public void lock() {
int me = ThreadID.get();
int n = level.length;
for(int i = 1; i < n; i++) {
level[me] = i; // I'm interested
victim[i] = me; // you go first
for(int k = 0; k < n; k++) { // check other thread ids
if(k == me)
continue; // skip myself
// spin while conflict exists
while(level[k] >= i && victim[i] == me) {} // wait
}
}
}
@Override
public void unlock() {
int me = ThreadID.get();
level[me] = 0;
}
}
Lock interface:
package org.example.ch2;
public interface Lock {
public void lock();
public void unlock();
}
ThreadID class:
package org.example.ch2;
import java.util.HashMap;
public class ThreadID {
private static final HashMap<Long, Integer> map = new HashMap<>();
public static synchronized int get() {
long threadId = Thread.currentThread().threadId();
if(!map.containsKey(threadId)) {
int key = map.size();
map.put(threadId, key);
}
return map.get(threadId);
}
}
Running the example:
package org.example.ch2.example4;
import org.example.ch2.Lock;
public class Example {
private static int count = 0;
public static void main(String[] args) {
int n = 9;
Lock lock = new Filter(n);
Thread[] threads = new Thread[n];
for (int i = 0; i < n; i++) {
threads[i] = new Thread(() -> {
lock.lock();
System.out.println(++count);
lock.unlock();
});
}
for (Thread thread : threads) {
thread.start();
}
// pause before you press Enter
new java.util.Scanner(System.in).nextLine();
}
}
output:
1
2
3
4
5
6
7
8
9
Upvotes: 0
Reputation: 75207
It can be written as:
for (int k = 0; k < n; k++) {
while ((k != me) && (level[k] >= i && victim[i] == me)) {
//spin wait
}
}
Upvotes: 8
Reputation: 500673
My reading of
(∃k != me) (level[k] >= i && victim[i] == me)
is "there exists some k
other than me
such that level[k] >= i && victim[i] == me
".
The loop spins until there is no k
for which the condition holds.
Here is another way to state the same thing:
boolean conflicts_exist = true;
while (conflicts_exist) {
conflicts_exist = false;
for (int k = 1; k < n; k++) {
if (k != me && level[k] >= i && victim[i] == me) {
conflicts_exist = true;
break;
}
}
}
Upvotes: 10