netty5/transport-native-kqueue/src/main/java/io/netty/channel/kqueue/KQueueEventLoop.java
Norman Maurer c9aaa93d83
Allow to specify a EventLoopTaskQueueFactory for various EventLoopGroup implementations (#9247)
Motivation:

Sometimes it is desirable to be able to use a different Queue implementation for the EventLoop of a Channel. This is currently not possible without resort to reflection.

Modifications:

- Add a new constructor to Nio|Epoll|KQueueEventLoopGroup which allows to specify a factory which is used to create the task queue. This was the user can override the default implementation.
- Add test

Result:

Be able to change Queue that is used for the EventLoop.
2019-06-21 09:05:19 +02:00

394 lines
16 KiB
Java

/*
* Copyright 2016 The Netty Project
*
* The Netty Project 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 io.netty.channel.kqueue;
import io.netty.channel.EventLoop;
import io.netty.channel.EventLoopGroup;
import io.netty.channel.MultithreadEventLoopGroup;
import io.netty.channel.SelectStrategy;
import io.netty.channel.SingleThreadEventLoop;
import io.netty.channel.kqueue.AbstractKQueueChannel.AbstractKQueueUnsafe;
import io.netty.channel.unix.FileDescriptor;
import io.netty.channel.unix.IovArray;
import io.netty.util.IntSupplier;
import io.netty.util.collection.IntObjectHashMap;
import io.netty.util.collection.IntObjectMap;
import io.netty.util.concurrent.RejectedExecutionHandler;
import io.netty.util.internal.ObjectUtil;
import io.netty.util.internal.PlatformDependent;
import io.netty.util.internal.logging.InternalLogger;
import io.netty.util.internal.logging.InternalLoggerFactory;
import java.io.IOException;
import java.util.Queue;
import java.util.concurrent.Executor;
import java.util.concurrent.atomic.AtomicIntegerFieldUpdater;
import static java.lang.Math.min;
/**
* {@link EventLoop} which uses kqueue under the covers. Only works on BSD!
*/
final class KQueueEventLoop extends SingleThreadEventLoop {
private static final InternalLogger logger = InternalLoggerFactory.getInstance(KQueueEventLoop.class);
private static final AtomicIntegerFieldUpdater<KQueueEventLoop> WAKEN_UP_UPDATER =
AtomicIntegerFieldUpdater.newUpdater(KQueueEventLoop.class, "wakenUp");
private static final int KQUEUE_WAKE_UP_IDENT = 0;
static {
// Ensure JNI is initialized by the time this class is loaded by this time!
// We use unix-common methods in this class which are backed by JNI methods.
KQueue.ensureAvailability();
}
private final boolean allowGrowing;
private final FileDescriptor kqueueFd;
private final KQueueEventArray changeList;
private final KQueueEventArray eventList;
private final SelectStrategy selectStrategy;
private final IovArray iovArray = new IovArray();
private final IntSupplier selectNowSupplier = new IntSupplier() {
@Override
public int get() throws Exception {
return kqueueWaitNow();
}
};
private final IntObjectMap<AbstractKQueueChannel> channels = new IntObjectHashMap<AbstractKQueueChannel>(4096);
private volatile int wakenUp;
private volatile int ioRatio = 50;
KQueueEventLoop(EventLoopGroup parent, Executor executor, int maxEvents,
SelectStrategy strategy, RejectedExecutionHandler rejectedExecutionHandler,
MultithreadEventLoopGroup.EventLoopTaskQueueFactory queueFactory) {
super(parent, executor, false, newTaskQueue(queueFactory), newTaskQueue(queueFactory),
rejectedExecutionHandler);
selectStrategy = ObjectUtil.checkNotNull(strategy, "strategy");
this.kqueueFd = Native.newKQueue();
if (maxEvents == 0) {
allowGrowing = true;
maxEvents = 4096;
} else {
allowGrowing = false;
}
changeList = new KQueueEventArray(maxEvents);
eventList = new KQueueEventArray(maxEvents);
int result = Native.keventAddUserEvent(kqueueFd.intValue(), KQUEUE_WAKE_UP_IDENT);
if (result < 0) {
cleanup();
throw new IllegalStateException("kevent failed to add user event with errno: " + (-result));
}
}
private static Queue<Runnable> newTaskQueue(
MultithreadEventLoopGroup.EventLoopTaskQueueFactory queueFactory) {
if (queueFactory == null) {
return newTaskQueue0(DEFAULT_MAX_PENDING_TASKS);
}
return queueFactory.newTaskQueue(DEFAULT_MAX_PENDING_TASKS);
}
void add(AbstractKQueueChannel ch) {
assert inEventLoop();
AbstractKQueueChannel old = channels.put(ch.fd().intValue(), ch);
// We either expect to have no Channel in the map with the same FD or that the FD of the old Channel is already
// closed.
assert old == null || !old.isOpen();
}
void evSet(AbstractKQueueChannel ch, short filter, short flags, int fflags) {
assert inEventLoop();
changeList.evSet(ch, filter, flags, fflags);
}
void remove(AbstractKQueueChannel ch) throws Exception {
assert inEventLoop();
int fd = ch.fd().intValue();
AbstractKQueueChannel old = channels.remove(fd);
if (old != null && old != ch) {
// The Channel mapping was already replaced due FD reuse, put back the stored Channel.
channels.put(fd, old);
// If we found another Channel in the map that is mapped to the same FD the given Channel MUST be closed.
assert !ch.isOpen();
} else if (ch.isOpen()) {
// Remove the filters. This is only needed if it's still open as otherwise it will be automatically
// removed once the file-descriptor is closed.
//
// See also https://www.freebsd.org/cgi/man.cgi?query=kqueue&sektion=2
ch.unregisterFilters();
}
}
/**
* Return a cleared {@link IovArray} that can be used for writes in this {@link EventLoop}.
*/
IovArray cleanArray() {
iovArray.clear();
return iovArray;
}
@Override
protected void wakeup(boolean inEventLoop) {
if (!inEventLoop && WAKEN_UP_UPDATER.compareAndSet(this, 0, 1)) {
wakeup();
}
}
private void wakeup() {
Native.keventTriggerUserEvent(kqueueFd.intValue(), KQUEUE_WAKE_UP_IDENT);
// Note that the result may return an error (e.g. errno = EBADF after the event loop has been shutdown).
// So it is not very practical to assert the return value is always >= 0.
}
private int kqueueWait(boolean oldWakeup) throws IOException {
// If a task was submitted when wakenUp value was 1, the task didn't get a chance to produce wakeup event.
// So we need to check task queue again before calling kqueueWait. If we don't, the task might be pended
// until kqueueWait was timed out. It might be pended until idle timeout if IdleStateHandler existed
// in pipeline.
if (oldWakeup && hasTasks()) {
return kqueueWaitNow();
}
long totalDelay = delayNanos(System.nanoTime());
int delaySeconds = (int) min(totalDelay / 1000000000L, Integer.MAX_VALUE);
return kqueueWait(delaySeconds, (int) min(totalDelay - delaySeconds * 1000000000L, Integer.MAX_VALUE));
}
private int kqueueWaitNow() throws IOException {
return kqueueWait(0, 0);
}
private int kqueueWait(int timeoutSec, int timeoutNs) throws IOException {
int numEvents = Native.keventWait(kqueueFd.intValue(), changeList, eventList, timeoutSec, timeoutNs);
changeList.clear();
return numEvents;
}
private void processReady(int ready) {
for (int i = 0; i < ready; ++i) {
final short filter = eventList.filter(i);
final short flags = eventList.flags(i);
final int fd = eventList.fd(i);
if (filter == Native.EVFILT_USER || (flags & Native.EV_ERROR) != 0) {
// EV_ERROR is returned if the FD is closed synchronously (which removes from kqueue) and then
// we later attempt to delete the filters from kqueue.
assert filter != Native.EVFILT_USER ||
(filter == Native.EVFILT_USER && fd == KQUEUE_WAKE_UP_IDENT);
continue;
}
AbstractKQueueChannel channel = channels.get(fd);
if (channel == null) {
// This may happen if the channel has already been closed, and it will be removed from kqueue anyways.
// We also handle EV_ERROR above to skip this even early if it is a result of a referencing a closed and
// thus removed from kqueue FD.
logger.warn("events[{}]=[{}, {}] had no channel!", i, eventList.fd(i), filter);
continue;
}
AbstractKQueueUnsafe unsafe = (AbstractKQueueUnsafe) channel.unsafe();
// First check for EPOLLOUT as we may need to fail the connect ChannelPromise before try
// to read from the file descriptor.
if (filter == Native.EVFILT_WRITE) {
unsafe.writeReady();
} else if (filter == Native.EVFILT_READ) {
// Check READ before EOF to ensure all data is read before shutting down the input.
unsafe.readReady(eventList.data(i));
} else if (filter == Native.EVFILT_SOCK && (eventList.fflags(i) & Native.NOTE_RDHUP) != 0) {
unsafe.readEOF();
}
// Check if EV_EOF was set, this will notify us for connection-reset in which case
// we may close the channel directly or try to read more data depending on the state of the
// Channel and also depending on the AbstractKQueueChannel subtype.
if ((flags & Native.EV_EOF) != 0) {
unsafe.readEOF();
}
}
}
@Override
protected void run() {
for (;;) {
try {
int strategy = selectStrategy.calculateStrategy(selectNowSupplier, hasTasks());
switch (strategy) {
case SelectStrategy.CONTINUE:
continue;
case SelectStrategy.BUSY_WAIT:
// fall-through to SELECT since the busy-wait is not supported with kqueue
case SelectStrategy.SELECT:
strategy = kqueueWait(WAKEN_UP_UPDATER.getAndSet(this, 0) == 1);
// 'wakenUp.compareAndSet(false, true)' is always evaluated
// before calling 'selector.wakeup()' to reduce the wake-up
// overhead. (Selector.wakeup() is an expensive operation.)
//
// However, there is a race condition in this approach.
// The race condition is triggered when 'wakenUp' is set to
// true too early.
//
// 'wakenUp' is set to true too early if:
// 1) Selector is waken up between 'wakenUp.set(false)' and
// 'selector.select(...)'. (BAD)
// 2) Selector is waken up between 'selector.select(...)' and
// 'if (wakenUp.get()) { ... }'. (OK)
//
// In the first case, 'wakenUp' is set to true and the
// following 'selector.select(...)' will wake up immediately.
// Until 'wakenUp' is set to false again in the next round,
// 'wakenUp.compareAndSet(false, true)' will fail, and therefore
// any attempt to wake up the Selector will fail, too, causing
// the following 'selector.select(...)' call to block
// unnecessarily.
//
// To fix this problem, we wake up the selector again if wakenUp
// is true immediately after selector.select(...).
// It is inefficient in that it wakes up the selector for both
// the first case (BAD - wake-up required) and the second case
// (OK - no wake-up required).
if (wakenUp == 1) {
wakeup();
}
// fallthrough
default:
}
final int ioRatio = this.ioRatio;
if (ioRatio == 100) {
try {
if (strategy > 0) {
processReady(strategy);
}
} finally {
runAllTasks();
}
} else {
final long ioStartTime = System.nanoTime();
try {
if (strategy > 0) {
processReady(strategy);
}
} finally {
final long ioTime = System.nanoTime() - ioStartTime;
runAllTasks(ioTime * (100 - ioRatio) / ioRatio);
}
}
if (allowGrowing && strategy == eventList.capacity()) {
//increase the size of the array as we needed the whole space for the events
eventList.realloc(false);
}
} catch (Throwable t) {
handleLoopException(t);
}
// Always handle shutdown even if the loop processing threw an exception.
try {
if (isShuttingDown()) {
closeAll();
if (confirmShutdown()) {
break;
}
}
} catch (Throwable t) {
handleLoopException(t);
}
}
}
@Override
protected Queue<Runnable> newTaskQueue(int maxPendingTasks) {
return newTaskQueue0(maxPendingTasks);
}
private static Queue<Runnable> newTaskQueue0(int maxPendingTasks) {
// This event loop never calls takeTask()
return maxPendingTasks == Integer.MAX_VALUE ? PlatformDependent.<Runnable>newMpscQueue()
: PlatformDependent.<Runnable>newMpscQueue(maxPendingTasks);
}
/**
* Returns the percentage of the desired amount of time spent for I/O in the event loop.
*/
public int getIoRatio() {
return ioRatio;
}
/**
* Sets the percentage of the desired amount of time spent for I/O in the event loop. The default value is
* {@code 50}, which means the event loop will try to spend the same amount of time for I/O as for non-I/O tasks.
*/
public void setIoRatio(int ioRatio) {
if (ioRatio <= 0 || ioRatio > 100) {
throw new IllegalArgumentException("ioRatio: " + ioRatio + " (expected: 0 < ioRatio <= 100)");
}
this.ioRatio = ioRatio;
}
@Override
public int registeredChannels() {
return channels.size();
}
@Override
protected void cleanup() {
try {
try {
kqueueFd.close();
} catch (IOException e) {
logger.warn("Failed to close the kqueue fd.", e);
}
} finally {
// Cleanup all native memory!
changeList.free();
eventList.free();
}
}
private void closeAll() {
try {
kqueueWaitNow();
} catch (IOException e) {
// ignore on close
}
// Using the intermediate collection to prevent ConcurrentModificationException.
// In the `close()` method, the channel is deleted from `channels` map.
AbstractKQueueChannel[] localChannels = channels.values().toArray(new AbstractKQueueChannel[0]);
for (AbstractKQueueChannel ch: localChannels) {
ch.unsafe().close(ch.unsafe().voidPromise());
}
}
private static void handleLoopException(Throwable t) {
logger.warn("Unexpected exception in the selector loop.", t);
// Prevent possible consecutive immediate failures that lead to
// excessive CPU consumption.
try {
Thread.sleep(1000);
} catch (InterruptedException e) {
// Ignore.
}
}
}