* Decouble EventLoop details from the IO handling for each transport to allow easy re-use of code and customization
Motiviation:
As today extending EventLoop implementations to add custom logic / metrics / instrumentations is only possible in a very limited way if at all. This is due the fact that most implementations are final or even package-private. That said even if these would be public there are the ability to do something useful with these is very limited as the IO processing and task processing are very tightly coupled. All of the mentioned things are a big pain point in netty 4.x and need improvement.
Modifications:
This changeset decoubled the IO processing logic from the task processing logic for the main transport (NIO, Epoll, KQueue) by introducing the concept of an IoHandler. The IoHandler itself is responsible to wait for IO readiness and process these IO events. The execution of the IoHandler itself is done by the SingleThreadEventLoop as part of its EventLoop processing. This allows to use the same EventLoopGroup (MultiThreadEventLoupGroup) for all the mentioned transports by just specify a different IoHandlerFactory during construction.
Beside this core API change this changeset also allows to easily extend SingleThreadEventExecutor / SingleThreadEventLoop to add custom logic to it which then can be reused by all the transports. The ideas are very similar to what is provided by ScheduledThreadPoolExecutor (that is part of the JDK). This allows for example things like:
* Adding instrumentation / metrics:
* how many Channels are registered on an SingleThreadEventLoop
* how many Channels were handled during the IO processing in an EventLoop run
* how many task were handled during the last EventLoop / EventExecutor run
* how many outstanding tasks we have
...
...
* Implementing custom strategies for choosing the next EventExecutor / EventLoop to use based on these metrics.
* Use different Promise / Future / ScheduledFuture implementations
* decorate Runnable / Callables when submitted to the EventExecutor / EventLoop
As a lot of functionalities are folded into the MultiThreadEventLoopGroup and SingleThreadEventLoopGroup this changeset also removes:
* AbstractEventLoop
* AbstractEventLoopGroup
* EventExecutorChooser
* EventExecutorChooserFactory
* DefaultEventLoopGroup
* DefaultEventExecutor
* DefaultEventExecutorGroup
Result:
Fixes https://github.com/netty/netty/issues/8514 .
Motivation:
We should leave the responsibility to choose the EventExecutor for a ChannelHandler to the user for more flexibility and to keep things simple.
Modification:
- Change method signatures to take an EventExecutor and not an EventExecutorGroup
- Remove special ChannelOption that allowed to enable / disable EventExecutor pinning
Result:
Simpler and more flexible code.
Motivation:
Custom Netty ThreadLocalRandom and ThreadLocalRandomProvider classes are no longer needed and can be removed.
Modification:
Remove own ThreadLocalRandom
Result:
Less code to maintain
Motivation:
PlatformDependent.newConcurrentHashMap() is no longer needed so it could be easily removed and new ConcurrentHashMap<>() inlined instead of invoking PlatformDependent.newConcurrentHashMap().
Modification:
Use ConcurrentHashMap provided by the JDK directly.
Result:
Less code to maintain.
Motivation:
We can use the diamond operator these days.
Modification:
Use diamond operator whenever possible.
Result:
More modern code and less boiler-plate.
Motivation:
Netty uses own Integer.compare and Long.compare methods. Since Java 7 we can use Java implementation instead.
Modification:
Remove own implementation
Result:
Less code to maintain
Motivation:
Invoking ChannelHandlers is not free and can result in some overhead when the ChannelPipeline becomes very long. This is especially true if most handlers will just forward the call to the next handler in the pipeline. When the user extends Channel*HandlerAdapter we can easily detect if can just skip the handler and invoke the next handler in the pipeline directly. This reduce the overhead of dispatch but also reduce the call-stack in many cases.
Modifications:
Detect if we can skip the handler when walking the pipeline.
Result:
Reduce overhead for long pipelines.
Benchmark (extraHandlers) Mode Cnt Score Error Units
DefaultChannelPipelineBenchmark.propagateEventOld 4 thrpt 10 267313.031 ± 9131.140 ops/s
DefaultChannelPipelineBenchmark.propagateEvent 4 thrpt 10 824825.673 ± 12727.594 ops/s
Motivation:
While we are not yet quite sure if we want to require Java11 as minimum we are at least sure we want to use java8 as minimum.
Modifications:
Change minimum version to java8 and update some tests which failed compilation after this change.
Result:
Use Java8 as minimum and be able to use Java8 features.
Motivation:
testChannelInitializerEventExecutor() did sometimes fail as we sometimes miss to count down the latch. This can happen when we remove the handler from the pipeline before channelUnregistered(...) was called for it.
Modifications:
Countdown the latch in handlerRemoved(...).
Result:
Fix flaky test.
Motivation:
testWriteTaskRejected was racy as we did not ensure we dispatched all events to the executor before shutting it down.
Modifications:
Add a latch to ensure we dispatched everything.
Result:
Fix racy test that failed sometimes before.
Motiviation:
Because of how we implemented the registration / deregistration of an EventLoop it was not possible to wrap an EventLoop implementation and use it with a Channel.
Modification:
- Introduce EventLoop.Unsafe which is responsible for the actual registration.
- Move validation of EventLoop / Channel combo to the EventLoop
- Add unit test that verifies that wrapping works
Result:
Be able to wrap an EventLoop and so add some extra functionality.
Motivation:
We should access the Constructor of the passed in class in the Constructor of ReflectiveChannelFactory only to reduce the overhead but also fail-fast.
Modifications:
Access the Constructor early.
Result:
Fails fast and less performance overhead.
Motivation:
At the moment it’s possible to have a Channel in Netty that is not registered / assigned to an EventLoop until register(...) is called. This is suboptimal as if the Channel is not registered it is also not possible to do anything useful with a ChannelFuture that belongs to the Channel. We should think about if we should have the EventLoop as a constructor argument of a Channel and have the register / deregister method only have the effect of add a Channel to KQueue/Epoll/... It is also currently possible to deregister a Channel from one EventLoop and register it with another EventLoop. This operation defeats the threading model assumptions that are wide spread in Netty, and requires careful user level coordination to pull off without any concurrency issues. It is not a commonly used feature in practice, may be better handled by other means (e.g. client side load balancing), and therefore we propose removing this feature.
Modifications:
- Change all Channel implementations to require an EventLoop for construction ( + an EventLoopGroup for all ServerChannel implementations)
- Remove all register(...) methods from EventLoopGroup
- Add ChannelOutboundInvoker.register(...) which now basically means we want to register on the EventLoop for IO.
- Change ChannelUnsafe.register(...) to not take an EventLoop as parameter (as the EventLoop is supplied on custruction).
- Change ChannelFactory to take an EventLoop to create new Channels and introduce ServerChannelFactory which takes an EventLoop and one EventLoopGroup to create new ServerChannel instances.
- Add ServerChannel.childEventLoopGroup()
- Ensure all operations on the accepted Channel is done in the EventLoop of the Channel in ServerBootstrap
- Change unit tests for new behaviour
Result:
A Channel always has an EventLoop assigned which will never change during its life-time. This ensures we are always be able to call any operation on the Channel once constructed (unit the EventLoop is shutdown). This also simplifies the logic in DefaultChannelPipeline a lot as we can always call handlerAdded / handlerRemoved directly without the need to wait for register() to happen.
Also note that its still possible to deregister a Channel and register it again. It's just not possible anymore to move from one EventLoop to another (which was not really safe anyway).
Fixes https://github.com/netty/netty/issues/8513.
Motivation:
We should remove the ChannelPool and related implementations. It is often the case that having protocol knowledge can result in more effective pooling and ChannelPool currently doesn’t have this knowledge. This responsibility is assumed to be implemented at layers higher in the stack than Netty.
Modifications:
Remove io.netty.channel.pool.*
Result:
Less code to maintain, fixes https://github.com/netty/netty/issues/8549.
Motivation:
Due a race in DefaultChannelPipeline / AbstractChannelHandlerContext it was possible to have only handlerRemoved(...) called during tearing down the pipeline, even when handlerAdded(...) was never called. We need to ensure we either call both of none to guarantee a proper lifecycle of the handler.
Modifications:
- Enforce handlerAdded(...) / handlerRemoved(...) semantics / ordering
- Add unit test.
Result:
Fixes https://github.com/netty/netty/issues/8676 / https://github.com/netty/netty/issues/6536 .
* Handling AUTO_READ should not be the responsibility of DefaultChannelPipeline but the Channel itself.
Motivation:
At the moment we do automatically call read() in the DefaultChannelPipeline when fireChannelReadComplete() / fireChannelActive() is called and the Channel is using auto read. This is nice in terms of sharing code but imho is not the responsibility of the ChannelPipeline implementation but the responsibility of the Channel implementation.
Modifications:
Move handing of auto read from DefaultChannelPipeline to Channel implementations.
Result:
More clear responsibiliy and not depending on implemention details of the ChannelPipeline.
Motivation:
executeAfterEventLoopIteration is an Unstable API and isnt used in Netty. We should remove it to reduce complexity.
Changes:
This reverts commit 77770374fb.
Result:
Simplify implementation / cleanup.
Motivation:
8331248671 did make some changes to fix a race in ChannelInitializer when using with a custom EventExecutor. Unfortunally these where a bit racy and so the testcase failed sometimes.
Modifications:
- More correct fix when using a custom EventExecutor
- Adjust the testcase to be more correct.
Result:
Proper fix for https://github.com/netty/netty/issues/8616.
Motivation:
Most of the maven modules do not explicitly declare their
dependencies and rely on transitivity, which is not always correct.
Modifications:
For all maven modules, add all of their dependencies to pom.xml
Result:
All of the (essentially non-transitive) depepdencies of the modules are explicitly declared in pom.xml
Motivation:
The ChannelInitializer may be invoked multipled times when used with a custom EventExecutor as removal operation may be done asynchronously. We need to guard against this.
Modifications:
- Change Map to Set which is more correct in terms of how we use it.
- Ensure we only modify the internal Set when the handler was removed yet
- Add unit test.
Result:
Fixes https://github.com/netty/netty/issues/8616.
Motivation:
java.nio.channels.spi.AbstractSelectableChannel.register(...) need to obtain multiple locks during execution which may produce a long wait time if we currently select. This lead to multiple CI failures in the past.
Modifications:
Ensure the register call takes place on the EventLoop.
Result:
No more flacky CI test timeouts.
Motivation:
During benchmarks two methods showed up as "hot method too big". We can easily make these smaller by factor out some less common code-path to an extra method and so allow inlining.
Modifications:
Factor out less common code path to an extra method.
Result:
Hot methods can be inlined.
Motivation:
Our HeadContext in DefaultChannelPipeline does handle inbound and outbound but we only marked it as outbound. While this does not have any effect in the current code-base it can lead to problems when we change our internals (this is also how I found the bug).
Modifications:
Construct HeadContext so it is also marked as handling inbound.
Result:
More correct code.
Motivation:
This transport is unique because it uses Java's blocking IO (java.io / java.net) under the hood. However it is not clear if this transport is actually useful so it should be removed.
Modifications:
- Remove OIO transport and RXTX transport which depend on it.
- Remove Oio*Sctp* implementations
- Remove PerThreadEventLoop* which was only used by OIO transport.
Result:
Fixes https://github.com/netty/netty/issues/8510.
Motivation:
We plan to remove the OIO based transports in Netty 5 so we should mark these as deprecated already.
Modifications:
Mark all OIO based transports as deprecated.
Result:
Give the user a heads-up for removal.
Motivation:
When the Selector throws an IOException during our EventLoop processing we should rebuild it and transfer the registered Channels. At the moment we will continue trying to use it which will never work.
Modifications:
- Rebuild Selector when an IOException is thrown during any select*(...) methods.
- Add unit test.
Result:
Fixes https://github.com/netty/netty/issues/8566.
Motivation:
Besides an error caused by closing socket in Windows a bunch of other errors may happen at this place which won't be somehow logged. For instance any VirtualMachineError as OutOfMemoryError will be simply ignored. The library should at least log the problem.
Modification:
Added logging of the throwable object.
Result:
Fixes#8499.
Motivation:
There is a racy UnsupportedOperationException instead because the task removal is delegated to MpscChunkedArrayQueue that does not support removal. This happens with SingleThreadEventExecutor that overrides the newTaskQueue to return an MPSC queue instead of the LinkedBlockingQueue returned by the base class such as NioEventLoop, EpollEventLoop and KQueueEventLoop.
Modifications:
- Catch the UnsupportedOperationException
- Add unit test.
Result:
Fix#8475
Motivation:
There are currently many more places where this could be used which were
possibly not considered when the method was added.
If https://github.com/netty/netty/pull/8388 is included in its current
form, a number of these places could additionally make use of the same
BYTE_ARRAYS threadlocal.
There's also a couple of adjacent places where an optimistically-pooled
heap buffer is used for temp byte storage which could use the
threadlocal too in preference to allocating a temp heap bytebuf wrapper.
For example
https://github.com/netty/netty/blob/4.1/buffer/src/main/java/io/netty/buffer/ByteBufUtil.java#L1417.
Modifications:
Replace new byte[] with PlatformDependent.allocateUninitializedArray()
where appropriate; make use of ByteBufUtil.getBytes() in some places
which currently perform the equivalent logic, including avoiding copy of
backing array if possible (although would be rare).
Result:
Further potential speed-up with java9+ and appropriate compile flags.
Many of these places could be on latency-sensitive code paths.
Motivation:
It has shown that the used test timeout may be too low when the CI is busy.
Modifications:
Increase timeout to 3 seconds.
Result:
Less false-positives.
Motivation:
Currently we may end up in the situation that we incremented the pending bytes before submitting the AbstractWriteTask but never decrement these again if the submitting of the task fails. This may result in incorrect watermark handling.
Modifications:
- Correctly decrement pending bytes if subimitting of task fails and also ensure we recycle it correctly.
- Add unit test.
Result:
Fixes https://github.com/netty/netty/issues/8343.
Motivation:
Unless the 'io.netty.noKeySetOptimization' system property is set,
registering a SelectableChannel instance to a NioEventLoop results
in a ClassCastException:
io.netty.channel.nio.SelectedSelectionKeySetSelector cannot be cast
to java.nio.channels.spi.AbstractSelector
Modifications:
Instead of 'selector', pass 'unwrappedSelector' to SelectableChannel.
Result:
It is possible to register a SelectableChannel instance without
setting the 'io.netty.noKeySetOptimization' system property.
Motivation:
Add an option (through a SelectStrategy return code) to have the Netty event loop thread to do busy-wait on the epoll.
The reason for this change is to avoid the context switch cost that comes when the event loop thread is blocked on the epoll_wait() call.
On average, the context switch has a penalty of ~13usec.
This benefits both:
The latency when reading from a socket
Scheduling tasks to be executed on the event loop thread.
The tradeoff, when enabling this feature, is that the event loop thread will be using 100% cpu, even when inactive.
Modification:
Added SelectStrategy option to return BUSY_WAIT
Epoll loop will do a epoll_wait() with no timeout
Use pause instruction to hint to processor that we're in a busy loop
Result:
When enabled, minimizes impact of context switch in the critical path
Motivation:
In Java8 and earlier we used reflection to replace the used key set if not otherwise told. This does not work on Java9 and later without special flags as its not possible to call setAccessible(true) on the Field anymore.
Modifications:
- Use Unsafe to instrument the Selector with out special set when sun.misc.Unsafe is present and we are using Java9+.
Result:
NIO transport produce less GC on Java9 and later as well.
Motivation:
We need to implement remove() by ourselves to make it work on Java7 as otherwise it will throw an AbstractMethodError. This is a followup of c1a335446d.
Modifications:
Just implemented remove()
Result:
Works on Java7 as well.
Motivation:
c1a335446d reimplemented remove(...) and contains(...) in a way which made it not work anymore when used by the Selector.
Modifications:
Partly revert changes in c1a335446d.
Result:
Works again as expected
Motivation:
Our SelectedSelectionKeySet does not correctly implement various methods which can be done without any performance overhead.
Modifications:
Implement iterator(), contains(...) and remove(...)
Result:
Related to https://github.com/netty/netty/issues/8242.
Motivation:
It seems to sometimes confuse people what to do to replace setMaxMessagePerRead(...).
Modifications:
Add some more details to the javadocs about the correct replacement.
Result:
Related to https://github.com/netty/netty/issues/8214.
