/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.lucene.util;
import java.io.Closeable;
import java.lang.ref.WeakReference;
import java.util.Iterator;
import java.util.Map;
import java.util.WeakHashMap;
import java.util.concurrent.atomic.AtomicInteger;
Java's builtin ThreadLocal has a serious flaw: it can take an arbitrarily long amount of time to dereference the things you had stored in it, even once the ThreadLocal instance itself is no longer referenced. This is because there is single, master map stored for each thread, which all ThreadLocals share, and that master map only periodically purges "stale" entries. While not technically a memory leak, because eventually the memory will be reclaimed, it can take a long time and you can easily hit OutOfMemoryError because from the GC's standpoint the stale entries are not reclaimable. This class works around that, by only enrolling WeakReference values into the ThreadLocal, and separately holding a hard reference to each stored value. When you call close
, these hard references are cleared and then GC is freely able to reclaim space by objects stored in it. We can not rely on ThreadLocal.remove()
as it only removes the value for the caller thread, whereas close
takes care of all threads. You should not call close
until all threads are done using the instance. @lucene.internal
/** Java's builtin ThreadLocal has a serious flaw:
* it can take an arbitrarily long amount of time to
* dereference the things you had stored in it, even once the
* ThreadLocal instance itself is no longer referenced.
* This is because there is single, master map stored for
* each thread, which all ThreadLocals share, and that
* master map only periodically purges "stale" entries.
*
* While not technically a memory leak, because eventually
* the memory will be reclaimed, it can take a long time
* and you can easily hit OutOfMemoryError because from the
* GC's standpoint the stale entries are not reclaimable.
*
* This class works around that, by only enrolling
* WeakReference values into the ThreadLocal, and
* separately holding a hard reference to each stored
* value. When you call {@link #close}, these hard
* references are cleared and then GC is freely able to
* reclaim space by objects stored in it.
*
* We can not rely on {@link ThreadLocal#remove()} as it
* only removes the value for the caller thread, whereas
* {@link #close} takes care of all
* threads. You should not call {@link #close} until all
* threads are done using the instance.
*
* @lucene.internal
*/
public class CloseableThreadLocal<T> implements Closeable {
private ThreadLocal<WeakReference<T>> t = new ThreadLocal<>();
// Use a WeakHashMap so that if a Thread exits and is
// GC'able, its entry may be removed:
private Map<Thread,T> hardRefs = new WeakHashMap<>();
// Increase this to decrease frequency of purging in get:
private static int PURGE_MULTIPLIER = 20;
// On each get or set we decrement this; when it hits 0 we
// purge. After purge, we set this to
// PURGE_MULTIPLIER * stillAliveCount. This keeps
// amortized cost of purging linear.
private final AtomicInteger countUntilPurge = new AtomicInteger(PURGE_MULTIPLIER);
protected T initialValue() {
return null;
}
public T get() {
WeakReference<T> weakRef = t.get();
if (weakRef == null) {
T iv = initialValue();
if (iv != null) {
set(iv);
return iv;
} else {
return null;
}
} else {
maybePurge();
return weakRef.get();
}
}
public void set(T object) {
t.set(new WeakReference<>(object));
synchronized(hardRefs) {
hardRefs.put(Thread.currentThread(), object);
maybePurge();
}
}
private void maybePurge() {
if (countUntilPurge.getAndDecrement() == 0) {
purge();
}
}
// Purge dead threads
private void purge() {
synchronized(hardRefs) {
int stillAliveCount = 0;
for (Iterator<Thread> it = hardRefs.keySet().iterator(); it.hasNext();) {
final Thread t = it.next();
if (!t.isAlive()) {
it.remove();
} else {
stillAliveCount++;
}
}
int nextCount = (1+stillAliveCount) * PURGE_MULTIPLIER;
if (nextCount <= 0) {
// defensive: int overflow!
nextCount = 1000000;
}
countUntilPurge.set(nextCount);
}
}
@Override
public void close() {
// Clear the hard refs; then, the only remaining refs to
// all values we were storing are weak (unless somewhere
// else is still using them) and so GC may reclaim them:
hardRefs = null;
// Take care of the current thread right now; others will be
// taken care of via the WeakReferences.
if (t != null) {
t.remove();
}
t = null;
}
}