Motivation:
In linux it is possible to write more then one buffer withone syscall when sending datagram messages.
Modifications:
Not copy CompositeByteBuf if it only contains direct buffers.
Result:
More performance due less overhead for copy.
Motivation:
Due incorrect usage of CompositeByteBuf a buffer leak was introduced.
Modifications:
Correctly handle tests with CompositeByteBuf.
Result:
No more buffer leaks
Motivation:
On linux with glibc >= 2.14 it is possible to send multiple DatagramPackets with one syscall. This can be a huge performance win and so we should support it in our native transport.
Modification:
- Add support for sendmmsg by reuse IovArray
- Factor out ThreadLocal support of IovArray to IovArrayThreadLocal for better separation as we use IovArray also without ThreadLocal in NativeDatagramPacketArray now
- Introduce NativeDatagramPacketArray which is used for sendmmsg(...)
- Implement sendmmsg(...) via jni
- Expand DatagramUnicastTest to test also sendmmsg(...)
Result:
Netty now automatically use sendmmsg(...) if it is supported and we have more then 1 DatagramPacket in the ChannelOutboundBuffer and flush() is called.
Motivation:
On linux it is possible to use the sendMsg(...) system call to write multiple buffers with one system call when using datagram/udp.
Modifications:
- Implement the needed changes and make use of sendMsg(...) if possible for max performance
- Add tests that test sending datagram packets with all kind of different ByteBuf implementations.
Result:
Performance improvement when using CompoisteByteBuf and EpollDatagramChannel.
Motivation:
InetAddress.getByName(...) uses exceptions for control flow when try to parse IPv4-mapped-on-IPv6 addresses. This is quite expensive.
Modifications:
Detect IPv4-mapped-on-IPv6 addresses in the JNI level and convert to IPv4 addresses before pass to InetAddress.getByName(...) (via InetSocketAddress constructor).
Result:
Eliminate performance problem causes by exception creation when parsing IPv4-mapped-on-IPv6 addresses.
Motivation:
There is a bug in the JettySslEngineTest where the interface receiving a message does not do a latch.countDown().
This causes each test to be subject to the CountDownLatch timeout period instead of being notified right when an event occurs.
Modifications:
- The JettySslEngineTest message handler will call the appropriate latch.countDown after a message is received
Result:
JettySslEngineTest will not be subject to waiting the timeout period even if the message is correctly received
Motivation:
Outbound flow control does not properly remove the head of queue after
it's written. This will cause streams with multiple frames to get stuck
and not send all of the data.
Modifications:
Modified the DefaultHttp2OutboundFlowController to properly remove the
head of the pending write queue once a queued frame has been written.
Added an integration test that sends a large message to verify that all
DATA frames are properly collected at the other end.
Result:
Outbound flow control properly handles several queued messages.
Motivation:
Once the interrupted flag in SingleThreadEventLoopB was set it was never reset.
That way the eventloop thread was always in an interrupted state and all calls
to Thread.sleep/LockSupport.parkNanos would fail.
Modifications:
Reset interrupted flag.
Result:
Number of GC runs when executing the test is down to zero (from 20 - 50).
Looking at the heap dump shows >10x decrease in memory consumption e.g.
the number of ForkJoinTasks is down to 25k from 350k previously. Similar
is true for BlockingQueue.Node objects.
Motivation:
We received a bug-report that the ByteBuf.refCnt() does sometimes not show the correct value when release() and refCnt() is called from different Threads.
Modifications:
Add test-case which shows that all is working like expected
Result:
Test-case added which shows everything is ok.
Motivation:
Because of a bug a NPE was thrown when tried to encode HTTP to SPDY and no X-SPDY-Associated-To-Stream-ID was present.
Modifications:
Use 0 as default value when X-SPDY-Associated-To-Stream-ID not present.
Result:
No NPE anymore.
Motivation:
This fixes bug #2848 which caused Recycler to become unbounded and cache infinite number of objects with maxCapacity that's not a power of two. This can result in general sluggishness of the application and OutOfMemoryError.
Modifications:
The test for maxCapacity has been moved out of test to check if the buffer has filled. The buffer is now also capped at maxCapacity and cannot grow over it as it jumps from one power of two to the other.
Additionally, a unit test was added to verify maxCapacity is honored even when it's not a power of two.
Result:
With these changes the user is able to use a custom maxCapacity number and not have it ignored. The unit test assures this bug will not repeat itself.
Motivation:
We failed to release buffers on protocolErrors which caused buffer leaks when using HTTP/2
Modifications:
Release buffer on protocol errors
Result:
No more buffer leaks
Motivation:
Netty only supports a java NPN implementation provided by npn-api and npn-boot.
There is no java implementation for ALPN.
ALPN is needed to be compliant with the HTTP/2 spec.
Modifications:
-SslContext and JdkSslContext to support ALPN
-JettyNpn* class restructure for NPN and ALPN common aspects
-Pull in alpn-api and alpn-boot optional dependencies for ALPN java implementation
Result:
-Netty provides access to a java implementation of APLN
Motivation:
The SingleThreadEventLoopTest allocated up to half a gigabyte of memory per run,
causing up to 50 GC runs in ~20 seconds.
Modification:
Charlie Hunt identified TLAB allocations to be the root cause of this excessive
memory usage [1]. By reusing the Executor in every test we can reduce the
memory usage by >50%.
Result:
Lower Memory Usage and fewer Garbage Collector runs.
Helps to resolve GitHub issue #2841.
[1] https://twitter.com/charlesjhunt/status/505351389317722112
Motivation:
Sometimes the SingleThreadEventLoopTest did fail on our CI. This was because of GC pressure produced by Thread.sleep(...) when interrupted as it creates a new InterruptedException all the time (and need to fill it).
Modifications:
Replace Thread.sleep(...) with LockSupport.parkNanos(...) to eliminate exception overhead.
Result:
SingleThreadEventLoopTest produce a lot less garbage.
Motivation:
A recent refactoring of the outbound flow controller interface
introduced a bug when writing data. We're no longer properly handling
the completion of the write (i.e. updating stream state/handling error).
Modifications:
Updated AbstractHttp2ConnectionHandler.writeData to properly handle the
completion of the write future.
Result:
DATA writes now perform post-write cleanup.
The 'bin/' directory is currently not in the .gitignore file.
Eclipse creates this directory to support IDE operations.
These directory is generated and not related to source code and therefore does not belong in git.
Modifications:
-Add 'bin/' to .gitignore
Result:
-No more 'bin/' directory confusion for eclipse users.
Motivation:
We currently have a mix of "Observer" and "Listener" for interface
names. We should make them all "Listener" to be consistent.
Modifications:
Renamed Http2DataObserver->Http2DataListener and
Http2FrameObserver->Http2FrameListener.
Result:
Listener interface names are consistent.
Motivation:
The priority information reported by the HTTP/2 to HTTP tranlsation layer is not correct in all situations.
The HTTP translation layer is not using the Http2Connection.Listener interface to track tree restructures.
This incorrect information is being sent up to clients and is misleading.
Modifications:
-Restructure InboundHttp2ToHttpAdapter to allow a default data/header mode
-Extend this interface to provide an optional priority translation layer
Result:
-Priority information being correctly reported in HTTP/2 to HTTP translation layer
-Cleaner code with seperation of concerns (optional priority conversion).
Motivation:
Currently, window maintenance is automatically performed when a flow
control window drops below half its initial size. We should provide a
way for advanced applications to determine whether or not this should be
done on a per-stream basis.
Modifications:
Modifying DefaultHttp2InboundFlowController to allow enabling/disabling
of window maintenance per stream.
Result:
Inbound flow control window maintenance will be dynamically
configurable.
Motivation:
In EpollSocketchannel.doWriteFileRegion(...) we need to make sure we write until sendFile(...) returns either 0 or all is written. Otherwise we may not get notified once the Channel is writable again.
This is the case as we use EPOLL_ET.
Modifications:
Always write until either sendFile returns 0 or all is written.
Result:
No more hangs when writing DefaultFileRegion can happen.
Motivation:
To make it more easy to shutdown an EventExecutorGroup / EventLoopGroup we should let both of them extend AutoCloseable.
Modifications:
Let EventExecutorGroup extend AutoCloseable and impement it.
Result:
Easier shutdown of EventExecutorGroup and EventLoopGroup
Motivation:
Our last-minute change that made AbstractChannelHandlerContext implement PausableChannelEventExecutor broke the socksproxy example.
Modifications:
AbstractChannelHandlerContext does not inherit from PausableChannelEventExecutor anymore. Instead we'll allocate an extra object on demand.
Result:
AbstractChannelHandlerContext.executor().newPromise() returns the correct type.
Related issue: #2821
Motivation:
There's no way for a user to change the default ZlibEncoder
implementation.
It is already possible to change the default ZlibDecoder implementation.
Modification:
Add a new system property 'io.netty.noJdkZlibEncoder'.
Result:
A user can disable JDK ZlibEncoder, just like he or she can disable JDK
ZlibDecoder.
Motivation:
HTTP/2 draft 14 came out a couple of weeks ago and we need to keep up
with the spec.
Modifications:
-Revert back to dispatching FullHttpMessage objects instead of individual HttpObjects
-Corrections to HttpObject comparitors to support test cases
-New test cases to support sending headers immediatley
-Bug fixes cleaned up to ensure the message flow is terminated properly
Result:
Netty HTTP/2 to HTTP/1.x translation layer will support the HTTP/2 draft message flow.
Motivation:
This is just some general cleanup to get rid of the FrameWriter inner
interface withing Http2InboundFlowController. It's not necessary since
the flow controller can just use the Http2FrameWriter to send
WINDOW_UPDATE frames.
Modifications:
Updated DefaultHttp2InboundFlowController to use Http2FrameWriter.
Result:
The inbound flow control code is somewhat less smelly :).
Motivation:
We have some duplicated code that can be reused.
Modifications:
Create package private class called CodecUtil that now contains the shared code / helper method.
Result:
Less code-duplication
Motivation:
ByteToMessageCodec miss to check for @Sharable annotation in one of its constructors.
Modifications:
Ensure we call checkForSharableAnnotation in all constructors.
Result:
After your change, what will change.
Motivation:
We're currently out-of-spec with HTTP/2 in that we don't include padding
in the flow control logic.
Modifications:
Modified both DefaultHttp2InboundFlowController and
DefaultHttp2OutboundFlowController to properly take padding into
account. Outbound is more complicated since padding has to be properly
accounted for when splitting frames.
Result:
HTTP/2 codec properly flow controls padding.
Motivation:
The current interface and implementation for HTTP/2 priority tree events does not notify listeners of all parent change events. As a result operations which depend upon knowing about parent change events may be missing events and result in stale data. This interface also allows new listeners to easily consume priority tree change events.
Modifications:
-Http2Connection.Listener interface will change to support notifications on every node after the priority has changed and tree events have settled
-This will affect the outbound flow controller, DefaultHttp2Connection, and other listeners using the old interface
Result:
A modified (hopefully simplified and correct) Listener interface to get priority tree change event notification
Motivation:
This is addressing a TODO in the outbound flow controller. We currently
have a separate writer interface passed into the outbound flow
controller. This is confusing and limiting as to how the flow controller
can perform its writes (e.g. no control over flushing). Instead it would
be better to just let the flow controller use the Http2FrameWriter
directly.
Modifications:
- Added a new Http2DataWriter interface, which is extended by
Http2FrameWriter and Http2OutboundFlowController.
- Removed automatic flushing from Http2DataWriter in order to facilitate
optimizing the case where there are multiple writes.
- Updated DefaultHttp2OutboundFlowController to properly optimize
flushing of the ChannelHandlerContext when multiple writes occur.
Result:
Code is greatly simplified WRT outbound flow control and flushes are
optimized for flow-controlled DATA frames.
Motivation:
Currently we do more memory copies then needed.
Modification:
- Directly use heap buffers to reduce memory copy
- Correctly release buffers to fix buffer leak
Result:
Less memory copies and no leaks
Motivation:
There were no way to efficient write a CompositeByteBuf as we always did a memory copy to a direct buffer in this case. This is not needed as we can just write a CompositeByteBuf as long as all the components are buffers with a memory address.
Modifications:
- Write CompositeByteBuf which contains only direct buffers without memory copy
- Also handle CompositeByteBuf that have more components then 1024.
Result:
More efficient writing of CompositeByteBuf.
Motivation:
There is not need todo redunant reads of head in peakNode as we can just spin on next() until it becomes visible.
Modifications:
Remove redundant reads of head in peakNode. This is based on @nitsanw's patch for akka.
See https://github.com/akka/akka/pull/15596
Result:
Less volatile access.
Motivation:
The recent changes f8bee2e94c to use a FJP introduced a regression on Android that cause a NPE.
Modifications:
- Use AtomicReferenceFieldUpdater so it works also on Android
- Fix inspection warnings
Result:
Netty works again on Android too.
Motivation:
We used the wrong EventExecutor to notify for bind failures if a late registration was done.
Modifications:
Use the correct EventExecutor to notify and only use the GlobelEventExecutor if the registration fails itself.
Result:
The correct Thread will do the notification.
Motivation:
The unit tests in codec-socks contained redundant casts and empty test classes.
Modifications:
- Remove redundant casts
- Delete empty test classes
Result:
Cleanup
The HTTP/2 codec is missing from the netty-all maven dependency.
This means third party libraries have to pull in additional dependencies
above netty-all to use this codec.
Modifications:
Add HTTP/2 codec to netty-all
Result:
Third party libraries don't need to pull in additional dependencies
to use HTTP/2 codec.
Motivation:
HTTP/2 draft 14 came out a couple of weeks ago and we need to keep up
with the spec.
Modifications:
- Removed use of segment throughout.
- Added new setting for MAX_FRAME_SIZE. Used by the frame reader/writer
rather than a constant.
- Added new setting for MAX_HEADER_LIST_SIZE. This is currently unused.
- Expanded the header size to 9 bytes. The frame length field is now 3
bytes and added logic for checking that it falls within the valid range.
Result:
Netty will support HTTP/2 draft 14 framing. There will still be some
work to do to be compliant with the HTTP adaptation layer.
Motivation:
The example mis handle two elements:
1) Last message is a LastHttpContent and is not taken into account by
the server handler
2) The client makes a sync on last write (chunked) but there is no flush
before, therefore the sync is waiting forever.
Modifications:
1) Take into account the message LastHttpContent in simple Get.
2) Removes sync but add flush for each post and multipost parts
Results:
Example is no more blocked after get test.
Should be done also in 4.0 and Master (similar changes)
Motivation:
Recently we changed the default value of SOMAXCONN that is used when we can not determine it by reading /proc/sys/net/core/somaxconn. While doing this we missed to update the javadocs to reflect the new default value that is used.
Modifications:
List correct default value in the javadocs of SOMAXCONN.
Result:
Correct javadocs.
Motivation:
In GitHub issue #2767 a bug was reported that the IPv4
default route leads to the ipfilter package denying
instead of accepting all addresses.
While the issue was reported for Netty 3.9, this bug
also applies to Netty 4 and higher.
Modifications:
When computing the subnet address from the CIDR prefix,
correctly handle the case where the prefix is set to zero.
Result:
Ipfilter accepts all addresses when passed the
IPv4 default route.
Motivation:
The test procedure is unstable when testing quick time (factor less or equal to 1). Changing to default 10ms in this case will force time to be correct and time to be checked only when factor is >= 2.
Modifications:
When factor is <= 1, minimalWaitBetween is 10ms
Result:
Hoping this version is finally stable.
Motivation:
It seems that in certain conditions, the write back from the server is so quick that the handler has no time to compute traffic shaping. So 10ms of wait before acknowledging is added in server side.
Modifications:
Add 10ms waiting before server ackonwledge the client.
Result:
The timing is now suppsed to be stable.
Motivation:
Test seems not yet stable (20ms instead of 40ms), so this small fix provides a "lightest" check.
If this is not passing again in master, one could deactivate this test by commenting the assertTrue in line 431.
Modification:
Divide by 2 the to be compared elapse time.
Result:
Locally this test is now passing.
