org.apache.cassandra.net.ManyToOneConcurrentLinkedQueue Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of cassandra-all Show documentation
Show all versions of cassandra-all Show documentation
The Apache Cassandra Project develops a highly scalable second-generation distributed database, bringing together Dynamo's fully distributed design and Bigtable's ColumnFamily-based data model.
/*
* 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.cassandra.net;
import java.util.Collection;
import java.util.Iterator;
import java.util.NoSuchElementException;
import java.util.Queue;
import java.util.concurrent.atomic.AtomicReferenceFieldUpdater;
import java.util.function.Consumer;
/**
* A concurrent many-producers-to-single-consumer linked queue.
*
* Based roughly on {@link java.util.concurrent.ConcurrentLinkedQueue}, except with simpler/cheaper consumer-side
* method implementations ({@link #poll()}, {@link #remove()}, {@link #drain(Consumer)}), and padding added
* to prevent false sharing.
*
* {@link #offer(Object)} provides volatile visibility semantics. {@link #offer(Object)} is lock-free, {@link #poll()}
* and all related consumer methods are wait-free.
*
* In addition to that, provides a {@link #relaxedPeekLastAndOffer(Object)} method that we use to avoid a CAS when
* putting message handlers onto the wait queue.
*/
class ManyToOneConcurrentLinkedQueue extends ManyToOneConcurrentLinkedQueueHead implements Queue
{
@SuppressWarnings("unused") // pad two cache lines after the head to prevent false sharing
protected long p31, p32, p33, p34, p35, p36, p37, p38, p39, p40, p41, p42, p43, p44, p45;
ManyToOneConcurrentLinkedQueue()
{
head = tail = new Node<>(null);
}
/**
* See {@link #relaxedIsEmpty()}.
*/
@Override
public boolean isEmpty()
{
return relaxedIsEmpty();
}
/**
* When invoked by the consumer thread, the answer will always be accurate.
* When invoked by a non-consumer thread, it won't always be the case:
* - {@code true} result indicates that the queue IS empty, no matter what;
* - {@code false} result indicates that the queue MIGHT BE non-empty - the value of {@code head} might
* not yet have been made externally visible by the consumer thread.
*/
boolean relaxedIsEmpty()
{
return null == head.next;
}
@Override
public int size()
{
int size = 0;
Node next = head;
while (null != (next = next.next))
size++;
return size;
}
@Override
public E peek()
{
Node next = head.next;
if (null == next)
return null;
return next.item;
}
@Override
public E element()
{
E item = peek();
if (null == item)
throw new NoSuchElementException("Queue is empty");
return item;
}
@Override
public E poll()
{
Node head = this.head;
Node next = head.next;
if (null == next)
return null;
this.lazySetHead(next); // update head reference to next before making previous head node unreachable,
head.lazySetNext(head); // to maintain the guarantee of tail being always reachable from head
E item = next.item;
next.item = null;
return item;
}
@Override
public E remove()
{
E item = poll();
if (null == item)
throw new NoSuchElementException("Queue is empty");
return item;
}
@Override
public boolean remove(Object o)
{
if (null == o)
throw new NullPointerException();
Node prev = this.head;
Node next = prev.next;
while (null != next)
{
if (o.equals(next.item))
{
prev.lazySetNext(next.next); // update prev reference to next before making removed node unreachable,
next.lazySetNext(next); // to maintain the guarantee of tail being always reachable from head
next.item = null;
return true;
}
prev = next;
next = next.next;
}
return false;
}
/**
* Consume the queue in its entirety and feed every item to the provided {@link Consumer}.
*
* Exists primarily for convenience, and essentially just wraps {@link #poll()} in a loop.
* Yields no performance benefit over invoking {@link #poll()} manually - there just isn't
* anything to meaningfully amortise on the consumer side of this queue.
*/
void drain(Consumer consumer)
{
E item;
while ((item = poll()) != null)
consumer.accept(item);
}
@Override
public boolean add(E e)
{
return offer(e);
}
@Override
public boolean offer(E e)
{
internalOffer(e); return true;
}
/**
* Adds the element to the queue and returns the item of the previous tail node.
* It's possible for the returned item to already have been consumed.
*
* @return previously last tail item in the queue, potentially stale
*/
E relaxedPeekLastAndOffer(E e)
{
return internalOffer(e);
}
/**
* internalOffer() is based on {@link java.util.concurrent.ConcurrentLinkedQueue#offer(Object)},
* written by Doug Lea and Martin Buchholz with assistance from members of JCP JSR-166 Expert Group
* and released to the public domain, as explained at http://creativecommons.org/publicdomain/zero/1.0/
*/
private E internalOffer(E e)
{
if (null == e)
throw new NullPointerException();
final Node node = new Node<>(e);
for (Node t = tail, p = t;;)
{
Node q = p.next;
if (q == null)
{
// p is last node
if (p.casNext(null, node))
{
// successful CAS is the linearization point for e to become an element of this queue and for node to become "live".
if (p != t) // hop two nodes at a time
casTail(t, node); // failure is ok
return p.item;
}
// lost CAS race to another thread; re-read next
}
else if (p == q)
{
/*
* We have fallen off list. If tail is unchanged, it will also be off-list, in which case we need to
* jump to head, from which all live nodes are always reachable. Else the new tail is a better bet.
*/
p = (t != (t = tail)) ? t : head;
}
else
{
// check for tail updates after two hops
p = (p != t && t != (t = tail)) ? t : q;
}
}
}
@Override
public boolean contains(Object o)
{
throw new UnsupportedOperationException();
}
@Override
public Iterator iterator()
{
throw new UnsupportedOperationException();
}
@Override
public Object[] toArray()
{
throw new UnsupportedOperationException();
}
@Override
public T[] toArray(T[] a)
{
throw new UnsupportedOperationException();
}
@Override
public boolean containsAll(Collection c)
{
throw new UnsupportedOperationException();
}
@Override
public boolean addAll(Collection c)
{
throw new UnsupportedOperationException();
}
@Override
public boolean removeAll(Collection c)
{
throw new UnsupportedOperationException();
}
@Override
public boolean retainAll(Collection c)
{
throw new UnsupportedOperationException();
}
@Override
public void clear()
{
throw new UnsupportedOperationException();
}
}
class ManyToOneConcurrentLinkedQueueHead extends ManyToOneConcurrentLinkedQueuePadding2
{
protected volatile ManyToOneConcurrentLinkedQueue.Node head;
private static final AtomicReferenceFieldUpdater headUpdater =
AtomicReferenceFieldUpdater.newUpdater(ManyToOneConcurrentLinkedQueueHead.class, Node.class, "head");
@SuppressWarnings("WeakerAccess")
protected void lazySetHead(Node val)
{
headUpdater.lazySet(this, val);
}
}
class ManyToOneConcurrentLinkedQueuePadding2 extends ManyToOneConcurrentLinkedQueueTail
{
@SuppressWarnings("unused") // pad two cache lines between tail and head to prevent false sharing
protected long p16, p17, p18, p19, p20, p21, p22, p23, p24, p25, p26, p27, p28, p29, p30;
}
class ManyToOneConcurrentLinkedQueueTail extends ManyToOneConcurrentLinkedQueuePadding1
{
protected volatile ManyToOneConcurrentLinkedQueue.Node tail;
private static final AtomicReferenceFieldUpdater tailUpdater =
AtomicReferenceFieldUpdater.newUpdater(ManyToOneConcurrentLinkedQueueTail.class, Node.class, "tail");
@SuppressWarnings({ "WeakerAccess", "UnusedReturnValue" })
protected boolean casTail(Node expect, Node update)
{
return tailUpdater.compareAndSet(this, expect, update);
}
}
class ManyToOneConcurrentLinkedQueuePadding1
{
@SuppressWarnings("unused") // pad two cache lines before the tail to prevent false sharing
protected long p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13, p14, p15;
static final class Node
{
E item;
volatile Node next;
private static final AtomicReferenceFieldUpdater nextUpdater =
AtomicReferenceFieldUpdater.newUpdater(Node.class, Node.class, "next");
Node(E item)
{
this.item = item;
}
@SuppressWarnings("SameParameterValue")
boolean casNext(Node expect, Node update)
{
return nextUpdater.compareAndSet(this, expect, update);
}
void lazySetNext(Node val)
{
nextUpdater.lazySet(this, val);
}
}
}
© 2015 - 2024 Weber Informatics LLC | Privacy Policy