org.jctools_voltpatches.queues.MpscChunkedArrayQueue Maven / Gradle / Ivy
Show all versions of voltdbclient Show documentation
/*
* Licensed 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.jctools_voltpatches.queues;
import static org.jctools_voltpatches.queues.CircularArrayOffsetCalculator.allocate;
import static org.jctools_voltpatches.util.UnsafeAccess.UNSAFE;
import static org.jctools_voltpatches.util.UnsafeRefArrayAccess.REF_ARRAY_BASE;
import static org.jctools_voltpatches.util.UnsafeRefArrayAccess.REF_ELEMENT_SHIFT;
import static org.jctools_voltpatches.util.UnsafeRefArrayAccess.lvElement;
import static org.jctools_voltpatches.util.UnsafeRefArrayAccess.soElement;
import java.lang.reflect.Field;
import java.util.AbstractQueue;
import java.util.Iterator;
import org.jctools_voltpatches.util.Pow2;
abstract class MpscChunkedArrayQueuePad1 extends AbstractQueue {
long p01, p02, p03, p04, p05, p06, p07;
long p10, p11, p12, p13, p14, p15, p16, p17;
}
abstract class MpscChunkedArrayQueueProducerFields extends MpscChunkedArrayQueuePad1 {
protected long producerIndex;
}
abstract class MpscChunkedArrayQueuePad2 extends MpscChunkedArrayQueueProducerFields {
long p01, p02, p03, p04, p05, p06, p07;
long p10, p11, p12, p13, p14, p15, p16, p17;
}
abstract class MpscChunkedArrayQueueColdProducerFields extends MpscChunkedArrayQueuePad2 {
protected long maxQueueCapacity;
protected long producerMask;
protected E[] producerBuffer;
protected volatile long producerLimit;
protected boolean isFixedChunkSize = false;
}
abstract class MpscChunkedArrayQueuePad3 extends MpscChunkedArrayQueueColdProducerFields {
long p0, p1, p2, p3, p4, p5, p6, p7;
long p10, p11, p12, p13, p14, p15, p16, p17;
}
abstract class MpscChunkedArrayQueueConsumerFields extends MpscChunkedArrayQueuePad3 {
protected long consumerMask;
protected E[] consumerBuffer;
protected long consumerIndex;
}
/**
* An MPSC array queue which starts at initialCapacity and grows to maxCapacity in linked chunks
* of the initial size. The queue grows only when the current buffer is full and elements are not copied on
* resize, instead a link to the new buffer is stored in the old buffer for the consumer to follow.
*
*
* @param
*/
public class MpscChunkedArrayQueue extends MpscChunkedArrayQueueConsumerFields
implements MessagePassingQueue, QueueProgressIndicators {
long p0, p1, p2, p3, p4, p5, p6, p7;
long p10, p11, p12, p13, p14, p15, p16, p17;
private final static long P_INDEX_OFFSET;
private final static long C_INDEX_OFFSET;
private final static long P_LIMIT_OFFSET;
static {
try {
Field iField = MpscChunkedArrayQueueProducerFields.class.getDeclaredField("producerIndex");
P_INDEX_OFFSET = UNSAFE.objectFieldOffset(iField);
}
catch (NoSuchFieldException e) {
throw new RuntimeException(e);
}
try {
Field iField = MpscChunkedArrayQueueConsumerFields.class.getDeclaredField("consumerIndex");
C_INDEX_OFFSET = UNSAFE.objectFieldOffset(iField);
}
catch (NoSuchFieldException e) {
throw new RuntimeException(e);
}
try {
Field iField = MpscChunkedArrayQueueColdProducerFields.class.getDeclaredField("producerLimit");
P_LIMIT_OFFSET = UNSAFE.objectFieldOffset(iField);
}
catch (NoSuchFieldException e) {
throw new RuntimeException(e);
}
}
private final static Object JUMP = new Object();
public MpscChunkedArrayQueue(final int maxCapacity) {
this(Math.max(2, Pow2.roundToPowerOfTwo(maxCapacity / 8)), maxCapacity, false);
}
/**
* @param initialCapacity the queue initial capacity. If chunk size is fixed this will be the chunk size.
* Must be 2 or more.
* @param maxCapacity the maximum capacity will be rounded up to the closest power of 2 and will be the
* upper limit of number of elements in this queue. Must be 4 or more and round up to a larger
* power of 2 than initialCapacity.
* @param fixedChunkSize if true the queue will grow in fixed sized chunks the size of initial capacity,
* otherwise chunk size will double on each resize until reaching the maxCapacity
*/
public MpscChunkedArrayQueue(final int initialCapacity, int maxCapacity, boolean fixedChunkSize) {
if (initialCapacity < 2) {
throw new IllegalArgumentException("Initial capacity must be 2 or more");
}
if (maxCapacity < 4) {
throw new IllegalArgumentException("Max capacity must be 4 or more");
}
if (Pow2.roundToPowerOfTwo(initialCapacity) >= Pow2.roundToPowerOfTwo(maxCapacity)) {
throw new IllegalArgumentException(
"Initial capacity cannot exceed maximum capacity(both rounded up to a power of 2)");
}
int p2capacity = Pow2.roundToPowerOfTwo(initialCapacity);
// leave lower bit of mask clear
long mask = (p2capacity - 1) << 1;
// need extra element to point at next array
E[] buffer = allocate(p2capacity + 1);
producerBuffer = buffer;
producerMask = mask;
consumerBuffer = buffer;
consumerMask = mask;
maxQueueCapacity = ((long)Pow2.roundToPowerOfTwo(maxCapacity)) << 1;
isFixedChunkSize = fixedChunkSize;
soProducerLimit(mask); // we know it's all empty to start with
}
@Override
public final Iterator iterator() {
throw new UnsupportedOperationException();
}
@Override
public boolean offer(final E e) {
if (null == e) {
throw new NullPointerException();
}
long mask;
E[] buffer;
long pIndex;
while (true) {
long producerLimit = lvProducerLimit();
pIndex = lvProducerIndex();
// lower bit is indicative of resize, if we see it we spin until it's cleared
if ((pIndex & 1) == 1) {
continue;
}
// pIndex is even (lower bit is 0) -> actual index is (pIndex >> 1)
// mask/buffer may get changed by resizing -> only use for array access after successful CAS.
mask = this.producerMask;
buffer = this.producerBuffer;
// a successful CAS ties the ordering, lv(pIndex)-[mask/buffer]->cas(pIndex)
// assumption behind this optimization is that queue is almost always empty or near empty
if (producerLimit <= pIndex) {
int result = offerSlowPath(mask, buffer, pIndex, producerLimit);
switch (result) {
case 0:
break;
case 1:
continue;
case 2:
return false;
case 3:
resize(mask, buffer, pIndex, consumerIndex, maxQueueCapacity, e);
return true;
}
}
if (casProducerIndex(pIndex, pIndex + 2)) {
break;
}
}
final long offset = modifiedCalcElementOffset(pIndex, mask);
soElement(buffer, offset, e);
return true;
}
private int offerSlowPath(long mask, E[] buffer, long pIndex, long producerLimit) {
int result;
final long consumerIndex = lvConsumerIndex();
final long maxQueueCapacity = this.maxQueueCapacity;
long bufferCapacity = getCurrentBufferCapacity(mask, maxQueueCapacity);
result = 0;// 0 - goto pIndex CAS
if (consumerIndex + bufferCapacity > pIndex) {
if (!casProducerLimit(producerLimit, consumerIndex + bufferCapacity)) {
result = 1;// retry from top
}
}
// full and cannot grow
else if (consumerIndex == (pIndex - maxQueueCapacity)) {
result = 2;// -> return false;
}
// grab index for resize -> set lower bit
else if (casProducerIndex(pIndex, pIndex + 1)) {
result = 3;// -> return true
}
else {
result = 1;// failed resize attempt, retry from top
}
return result;
}
/**
* This method assumes index is actually (index << 1) because lower bit is used for resize. This is
* compensated for by reducing the element shift. The computation is constant folded, so there's no cost.
*/
private static long modifiedCalcElementOffset(long index, long mask) {
return REF_ARRAY_BASE + ((index & mask) << (REF_ELEMENT_SHIFT - 1));
}
/**
* {@inheritDoc}
*
* This implementation is correct for single consumer thread use only.
*/
@SuppressWarnings("unchecked")
@Override
public E poll() {
final E[] buffer = consumerBuffer;
final long index = consumerIndex;
final long mask = consumerMask;
final long offset = modifiedCalcElementOffset(index, mask);
Object e = lvElement(buffer, offset);// LoadLoad
if (e == null) {
if (index != lvProducerIndex()) {
// poll() == null iff queue is empty, null element is not strong enough indicator, so we must
// check the producer index. If the queue is indeed not empty we spin until element is
// visible.
do {
e = lvElement(buffer, offset);
} while (e == null);
}
else {
return null;
}
}
if (e == JUMP) {
final E[] nextBuffer = getNextBuffer(buffer, mask);
return newBufferPoll(nextBuffer, index);
}
soElement(buffer, offset, null);
soConsumerIndex(index + 2);
return (E) e;
}
/**
* {@inheritDoc}
*
* This implementation is correct for single consumer thread use only.
*/
@SuppressWarnings("unchecked")
@Override
public E peek() {
final E[] buffer = consumerBuffer;
final long index = consumerIndex;
final long mask = consumerMask;
final long offset = modifiedCalcElementOffset(index, mask);
Object e = lvElement(buffer, offset);// LoadLoad
if (e == null && index != lvProducerIndex()) {
// peek() == null iff queue is empty, null element is not strong enough indicator, so we must
// check the producer index. If the queue is indeed not empty we spin until element is visible.
while ((e = lvElement(buffer, offset)) == null)
;
}
if (e == JUMP) {
return newBufferPeek(getNextBuffer(buffer, mask), index);
}
return (E) e;
}
@SuppressWarnings("unchecked")
private E[] getNextBuffer(final E[] buffer, final long mask) {
final long nextArrayOffset = nextArrayOffset(mask);
final E[] nextBuffer = (E[]) lvElement(buffer, nextArrayOffset);
soElement(buffer, nextArrayOffset, null);
return nextBuffer;
}
private long nextArrayOffset(final long mask) {
return modifiedCalcElementOffset(mask + 2, Long.MAX_VALUE);
}
private E newBufferPoll(E[] nextBuffer, final long index) {
final long offsetInNew = newBufferAndOffset(nextBuffer, index);
final E n = lvElement(nextBuffer, offsetInNew);// LoadLoad
if (n == null) {
throw new IllegalStateException("new buffer must have at least one element");
}
soElement(nextBuffer, offsetInNew, null);// StoreStore
soConsumerIndex(index + 2);
return n;
}
private E newBufferPeek(E[] nextBuffer, final long index) {
final long offsetInNew = newBufferAndOffset(nextBuffer, index);
final E n = lvElement(nextBuffer, offsetInNew);// LoadLoad
if (null == n) {
throw new IllegalStateException("new buffer must have at least one element");
}
return n;
}
private long newBufferAndOffset(E[] nextBuffer, final long index) {
consumerBuffer = nextBuffer;
consumerMask = (nextBuffer.length - 2) << 1;
final long offsetInNew = modifiedCalcElementOffset(index, consumerMask);
return offsetInNew;
}
@Override
public final int size() {
/*
* It is possible for a thread to be interrupted or reschedule between the read of the producer and
* consumer indices, therefore protection is required to ensure size is within valid range. In the
* event of concurrent polls/offers to this method the size is OVER estimated as we read consumer
* index BEFORE the producer index.
*/
long after = lvConsumerIndex();
while (true) {
final long before = after;
final long currentProducerIndex = lvProducerIndex();
after = lvConsumerIndex();
if (before == after) {
return (int) (currentProducerIndex - after) >> 1;
}
}
}
private long lvProducerIndex() {
return UNSAFE.getLongVolatile(this, P_INDEX_OFFSET);
}
private long lvConsumerIndex() {
return UNSAFE.getLongVolatile(this, C_INDEX_OFFSET);
}
private void soProducerIndex(long v) {
UNSAFE.putOrderedLong(this, P_INDEX_OFFSET, v);
}
private boolean casProducerIndex(long expect, long newValue) {
return UNSAFE.compareAndSwapLong(this, P_INDEX_OFFSET, expect, newValue);
}
private void soConsumerIndex(long v) {
UNSAFE.putOrderedLong(this, C_INDEX_OFFSET, v);
}
private long lvProducerLimit() {
return producerLimit;
}
private boolean casProducerLimit(long expect, long newValue) {
return UNSAFE.compareAndSwapLong(this, P_LIMIT_OFFSET, expect, newValue);
}
private void soProducerLimit(long v) {
UNSAFE.putOrderedLong(this, P_LIMIT_OFFSET, v);
}
@Override
public long currentProducerIndex() {
return lvProducerIndex();
}
@Override
public long currentConsumerIndex() {
return lvConsumerIndex();
}
@Override
public int capacity() {
return (int) (maxQueueCapacity / 2);
}
@Override
public boolean relaxedOffer(E e) {
return offer(e);
}
@SuppressWarnings("unchecked")
@Override
public E relaxedPoll() {
final E[] buffer = consumerBuffer;
final long index = consumerIndex;
final long mask = consumerMask;
final long offset = modifiedCalcElementOffset(index, mask);
Object e = lvElement(buffer, offset);// LoadLoad
if (e == null) {
return null;
}
if (e == JUMP) {
final E[] nextBuffer = getNextBuffer(buffer, mask);
return newBufferPoll(nextBuffer, index);
}
soElement(buffer, offset, null);
soConsumerIndex(index + 2);
return (E) e;
}
@SuppressWarnings("unchecked")
@Override
public E relaxedPeek() {
final E[] buffer = consumerBuffer;
final long index = consumerIndex;
final long mask = consumerMask;
final long offset = modifiedCalcElementOffset(index, mask);
Object e = lvElement(buffer, offset);// LoadLoad
if (e == JUMP) {
return newBufferPeek(getNextBuffer(buffer, mask), index);
}
return (E) e;
}
@Override
public int fill(Supplier s, int batchSize) {
long mask;
E[] buffer;
long pIndex;
int claimedSlots;
while (true) {
long producerLimit = lvProducerLimit();
pIndex = lvProducerIndex();
// lower bit is indicative of resize, if we see it we spin until it's cleared
if ((pIndex & 1) == 1) {
continue;
}
// pIndex is even (lower bit is 0) -> actual index is (pIndex >> 1)
// mask/buffer may get changed by resizing -> only use for array access after successful CAS.
mask = this.producerMask;
buffer = this.producerBuffer;
// a successful CAS ties the ordering, lv(pIndex)->[mask/buffer]->cas(pIndex)
// we want 'limit' slots, but will settle for whatever is visible to 'producerLimit'
long batchIndex = Math.min(producerLimit, pIndex + 2 * batchSize);
if (pIndex == producerLimit || producerLimit < batchIndex) {
int result = offerSlowPath(mask, buffer, pIndex, producerLimit);
switch (result) {
case 1:
continue;
case 2:
return 0;
case 3:
resize(mask, buffer, pIndex, consumerIndex, maxQueueCapacity, s.get());
return 1;
}
}
// claim limit slots at once
if (casProducerIndex(pIndex, batchIndex)) {
claimedSlots = (int) ((batchIndex - pIndex) / 2);
break;
}
}
int i = 0;
for (i = 0; i < claimedSlots; i++) {
final long offset = modifiedCalcElementOffset(pIndex + 2 * i, mask);
soElement(buffer, offset, s.get());
}
return claimedSlots;
}
private void resize(long mask, E[] buffer, long pIndex, final long consumerIndex,
final long maxQueueCapacity, final E e) {
int newBufferLength = getNextBufferCapacity(buffer, maxQueueCapacity);
final E[] newBuffer = allocate(newBufferLength);
producerBuffer = newBuffer;
producerMask = (newBufferLength - 2) << 1;
final long offsetInOld = modifiedCalcElementOffset(pIndex, mask);
final long offsetInNew = modifiedCalcElementOffset(pIndex, producerMask);
soElement(newBuffer, offsetInNew, e);
soElement(buffer, nextArrayOffset(mask), newBuffer);
final long available = maxQueueCapacity - (pIndex - consumerIndex);
if (available <= 0) {
throw new IllegalStateException();
}
// invalidate racing CASs
soProducerLimit(pIndex + Math.min(mask, available));
// make resize visible to consumer
soElement(buffer, offsetInOld, JUMP);
// make resize visible to the other producers
soProducerIndex(pIndex + 2);
}
private int getNextBufferCapacity(E[] buffer, final long maxQueueCapacity) {
int newBufferLength = buffer.length;
if (isFixedChunkSize) {
newBufferLength = buffer.length;
}
else {
if (buffer.length - 1 == maxQueueCapacity) {
throw new IllegalStateException();
}
newBufferLength = 2 * buffer.length - 1;
}
return newBufferLength;
}
protected long getCurrentBufferCapacity(long mask, final long maxQueueCapacity) {
// consider replacing if with subclass
return (!isFixedChunkSize && mask + 2 == maxQueueCapacity) ? maxQueueCapacity
: mask;
}
@Override
public int fill(Supplier s) {
long result = 0;// result is a long because we want to have a safepoint check at regular intervals
final int capacity = capacity();
do {
final int filled = fill(s, MpmcArrayQueue.RECOMENDED_OFFER_BATCH);
if (filled == 0) {
return (int) result;
}
result += filled;
} while (result <= capacity);
return (int) result;
}
@Override
public void fill(Supplier s,
WaitStrategy w,
ExitCondition exit) {
while (exit.keepRunning()) {
while (fill(s, MpmcArrayQueue.RECOMENDED_OFFER_BATCH) != 0) {
continue;
}
int idleCounter = 0;
while (fill(s, MpmcArrayQueue.RECOMENDED_OFFER_BATCH) == 0 && exit.keepRunning()) {
idleCounter = w.idle(idleCounter);
}
}
}
@Override
public void drain(Consumer c, WaitStrategy w, ExitCondition exit) {
int idleCounter = 0;
while (exit.keepRunning()) {
E e = relaxedPoll();
if (e == null) {
idleCounter = w.idle(idleCounter);
continue;
}
idleCounter = 0;
c.accept(e);
}
}
@Override
public int drain(Consumer c) {
return drain(c, capacity());
}
@Override
public int drain(final Consumer c, final int limit) {
/**
* Impl note: there are potentially some small gains to be had by manually inlining relaxedPoll() and hoisting
* reused fields out to reduce redundant reads.
*/
int i = 0;
E m;
for (; i < limit && (m = relaxedPoll()) != null; i++) {
c.accept(m);
}
return i;
}
}