Motivation
I noticed this while looking at something else.
AbstractEpollStreamChannel::spliceQueue is an MPSC queue but only
accessed from the event loop. So it could be just changed to e.g. an
ArrayDeque. This PR instead reverts to using is as an MPSC queue to
avoid dispatching a task to the EL, as appears was the original
intention.
Modification
Change AbstractEpollStreamChannel::spliceQueue to be volatile and lazily
initialized via double-checked locking. Add tasks directly to the queue
from the public methods rather than possibly waking the EL just to
enqueue.
An alternative is just to change PlatformDependent.newMpscQueue() to new
ArrayDeque() and be done with it :)
Result
Less disruptive channel/fd-splicing.
Motivation:
OOME is occurred by increasing suppressedExceptions because other libraries call Throwable#addSuppressed. As we have no control over what other libraries do we need to ensure this can not lead to OOME.
Modifications:
Only use static instances of the Exceptions if we can either dissable addSuppressed or we run on java6.
Result:
Not possible to OOME because of addSuppressed. Fixes https://github.com/netty/netty/issues/9151.
Motivation:
`DefaultFileRegion.transferTo` will return 0 all the time when we request more data then the actual file size. This may result in a busy spin while processing the fileregion during writes.
Modifications:
- If we wrote 0 bytes check if the underlying file size is smaller then the requested count and if so throw an IOException
- Add DefaultFileRegionTest
- Add a test to the testsuite
Result:
Fixes https://github.com/netty/netty/issues/8868.
Motivation:
We have a utility method to check for > 0 and >0 arguments. We should use it.
Modification:
use checkPositive/checkPositiveOrZero instead of if statement.
Result:
Re-use utility method.
* Allow to use native transports when sun.misc.Unsafe is not present on the system
Motivation:
We should be able to use the native transports (epoll / kqueue) even when sun.misc.Unsafe is not present on the system. This is especially important as Java11 will be released soon and does not allow access to it by default.
Modifications:
- Correctly disable usage of sun.misc.Unsafe when -PnoUnsafe is used while running the build
- Correctly increment metric when UnpooledDirectByteBuf is allocated. This was uncovered once -PnoUnsafe usage was fixed.
- Implement fallbacks in all our native transport code for when sun.misc.Unsafe is not present.
Result:
Fixes https://github.com/netty/netty/issues/8229.
Motivation:
We can store the NativeDatagramPacketArray directly in the EpollEventLoop. This removes the need of using FastThreadLocal.
Modifications:
- Store NativeDatagramPacketArray directly in the EpollEventLoop (just as we do with IovArray as well).
Result:
Less FastThreadLocal usage and more consistent code.
Motivation:
Epoll and Kqueue channels have internal state which forces
a single read operation after channel construction. This
violates the Channel#read() interface which indicates that
data shouldn't be delivered until this method is called.
The behavior is also inconsistent with the NIO transport.
Modifications:
- Epoll and Kqueue shouldn't unconditionally read upon
initialization, and instead should rely upon Channel#read()
or auto_read.
Result:
Epoll and Kqueue are more consistent with NIO.
Motivation:
The flush task is currently using flush() which will have the affect of have the flush traverse the whole ChannelPipeline and also flush messages that were written since we gave up flushing. This is not really correct as we should only continue to flush messages that were flushed at the point in time when the flush task was submitted for execution if the user not explicit call flush() by him/herself.
Modification:
Call *Unsafe.flush0() via the flush task which will only continue flushing messages that were marked as flushed before.
Result:
More correct behaviour when the flush task is used.
Motivation:
b215794de3 recently introduced a change in behavior where writeSpinCount provided a limit for how many write operations were attempted per flush operation. However when the write quantum was meet the selector write flag was not cleared, and the channel unsafe flush0 method has an optimization which prematurely exits if the write flag is set. This may lead to no write progress being made under the following scenario:
- flush is called, but the socket can't accept all data, we set the write flag
- the selector wakes us up because the socket is writable, we write data and use the writeSpinCount quantum
- we then schedule a flush() on the EventLoop to execute later, however it the flush0 optimization prematurely exits because the write flag is still set
In this scenario the socket is still writable so the EventLoop may never notify us that the socket is writable, and therefore we may never attempt to flush data to the OS.
Modifications:
- When the writeSpinCount quantum is exceeded we should clear the selector write flag
Result:
Fixes https://github.com/netty/netty/issues/7729
Motivation:
The writeSpinCount currently loops over the same buffer, gathering
write, file write, or other write operation multiple times but will
continue writing until there is nothing left or the OS doesn't accept
any data for that specific write. However if the OS keeps accepting
writes there is no way to limit how much time we spend on a specific
socket. This can lead to unfair consumption of resources dedicated to a
single socket.
We currently don't limit the amount of bytes we attempt to write per
gathering write. If there are many more bytes pending relative to the
SO_SNDBUF size we will end up building iov arrays with more elements
than can be written, which results in extra iteration, conditionals,
and book keeping.
Modifications:
- writeSpinCount should limit the number of system calls we make to
write data, instead of applying to individual write operations
- IovArray should support a maximum number of bytes
- IovArray should support composite buffers of greater than size 1024
- We should auto-scale the amount of data that we attempt to write per
gathering write operation relative to SO_SNDBUF and how much data is
successfully written
- The non-unsafe path should also support a maximum number of bytes,
and respect the IOV_MAX limit
Result:
Write resource consumption can be bounded and gathering writes have
a limit relative to the amount of data which can actually be accepted
by the socket.
Motivation:
We need to set readPending to false when we detect a EOF while issue a read as otherwise we may not unregister from the Selector / Epoll / KQueue and so keep on receving wakeups.
The important bit is that we may even get a wakeup for a read event but will still will only be able to read 0 bytes from the socket, so we need to be very careful when we clear the readPending. This can happen because we generally using edge-triggered mode for our native transports and because of the nature of edge-triggered we may schedule an read event just to find out there is nothing left to read atm (because we completely drained the socket on the previous read).
Modifications:
Set readPending to false when EOF is detected.
Result:
Fixes [#7255].
This reverts commit 413c7c2cd8 as it introduced an regression when edge-triggered mode is used which is true for our native transports by default. With 413c7c2cd8 included it was possible that we set readPending to false by mistake even if we would be interested in read more.
Motivation:
readPending is currently only set to false if data is delivered to the application, however this may result in duplicate events being received from the selector in the event that the socket was closed.
Modifications:
- We should set readPending to false before each read attempt for all
transports besides NIO.
- Based upon the Javadocs it is possible that NIO may have spurious
wakeups [1]. In this case we should be more cautious and only set
readPending to false if data was actually read.
[1] https://docs.oracle.com/javase/7/docs/api/java/nio/channels/SelectionKey.html
That a selection key's ready set indicates that its channel is ready for some operation category is a hint, but not a guarantee, that an operation in such a category may be performed by a thread without causing the thread to block.
Result:
Notification from the selector (or simulated events from kqueue/epoll ET) in the event of socket closure.
Fixes https://github.com/netty/netty/issues/7255
Motivation:
When SO_LINGER is used we run doClose() on the GlobalEventExecutor by default so we need to ensure we schedule all code that needs to be run on the EventLoop on the EventLoop in doClose. Beside this there are also threading issues when calling shutdownOutput(...)
Modifications:
- Schedule removal from EventLoop to the EventLoop
- Correctly handle shutdownOutput and shutdown in respect with threading-model
- Add unit tests
Result:
Fixes [#7159].
Motivation:
If AutoClose is false and there is a IoException then AbstractChannel will not close the channel but instead just fail flushed element in the ChannelOutboundBuffer. AbstractChannel also notifies of writability changes, which may lead to an infinite loop if the peer has closed its read side of the socket because we will keep accepting more data but continuously fail because the peer isn't accepting writes.
Modifications:
- If the transport throws on a write we should acknowledge that the output side of the channel has been shutdown and cleanup. If the channel can't accept more data because it is full, and still healthy it is not expected to throw. However if the channel is not healthy it will throw and is not expected to accept any more writes. In this case we should shutdown the output for Channels that support this feature and otherwise just close.
- Connection-less protocols like UDP can remain the same because the channel may disconnected temporarily.
- Make sure AbstractUnsafe#shutdownOutput is called because the shutdown on the socket may throw an exception.
Result:
More correct handling of write failure when AutoClose is false.
Motivation:
Implementations of DuplexChannel delegate the shutdownOutput to the underlying transport, but do not take any action on the ChannelOutboundBuffer. In the event of a write failure due to the underlying transport failing and application may attempt to shutdown the output and allow the read side the transport to finish and detect the close. However this may result in an issue where writes are failed, this generates a writability change, we continue to write more data, and this may lead to another writability change, and this loop may continue. Shutting down the output should fail all pending writes and not allow any future writes to avoid this scenario.
Modifications:
- Implementations of DuplexChannel should null out the ChannelOutboundBuffer and fail all pending writes
Result:
More controlled sequencing for shutting down the output side of a channel.
Motivation:
We did not correctly handle connect() and disconnect() in EpollDatagramChannel / KQueueDatagramChannel and so the behavior was different compared to NioDatagramChannel.
Modifications:
- Correct implement connect and disconnect methods
- Share connect and related code
- Add tests
Result:
EpollDatagramChannel / KQueueDatagramChannel also supports correctly connect() and disconnect() methods.
Motivation:
1. special handling of ByteBuf with multi nioBuffer rather than type of CompositeByteBuf (eg. DuplicatedByteBuf with CompositeByteBuf)
2. EpollDatagramUnicastTest and KQueueDatagramUnicastTest passed because CompositeByteBuf is converted to DuplicatedByteBuf before write to channel
3. uninitalized struct msghdr will raise error
Modifications:
1. isBufferCopyNeededForWrite(like isSingleDirectBuffer in NioDatgramChannel) checks wether a new direct buffer is needed
2. special handling of ByteBuf with multi nioBuffer in EpollDatagramChannel, AbstractEpollStreamChannel, KQueueDatagramChannel, AbstractKQueueStreamChannel and IovArray
3. initalize struct msghdr
Result:
handle of ByteBuf with multi nioBuffer in EpollDatagramChannel and KQueueDatagramChannel are ok
Motivation:
We currently don't have a native transport which supports kqueue https://www.freebsd.org/cgi/man.cgi?query=kqueue&sektion=2. This can be useful for BSD systems such as MacOS to take advantage of native features, and provide feature parity with the Linux native transport.
Modifications:
- Make a new transport-native-unix-common module with all the java classes and JNI code for generic unix items. This module will build a static library for each unix platform, and included in the dynamic libraries used for JNI (e.g. transport-native-epoll, and eventually kqueue).
- Make a new transport-native-unix-common-tests module where the tests for the transport-native-unix-common module will live. This is so each unix platform can inherit from these test and ensure they pass.
- Add a new transport-native-kqueue module which uses JNI to directly interact with kqueue
Result:
JNI support for kqueue.
Fixes https://github.com/netty/netty/issues/2448
Fixes https://github.com/netty/netty/issues/4231
Make the FileRegion comments about which transports are supported more accurate.
Also, eleminate any outstanding references to FileRegion.transfered as the method was renamed for spelling.
Modifications:
Class-level comment on FileRegion, can call renamed method.
Result:
More accurate documentation and less calls to deprecated methods.
Motivation:
EpollRecvByteAllocatorHandle intends to override the meaning of "maybe more data to read" which is a concept also used in all existing implementations of RecvByteBufAllocator$Handle but the interface doesn't support overriding. Because the interfaces lack the ability to propagate this computation EpollRecvByteAllocatorHandle attempts to implement a heuristic on top of the delegate which may lead to reading when we shouldn't or not reading data.
Modifications:
- Create a new interface ExtendedRecvByteBufAllocator and ExtendedHandle which allows the "maybe more data to read" between interfaces
- Deprecate RecvByteBufAllocator and change all existing implementations to extend ExtendedRecvByteBufAllocator
- transport-native-epoll should require ExtendedRecvByteBufAllocator so the "maybe more data to read" can be propagated to the ExtendedHandle
Result:
Fixes https://github.com/netty/netty/issues/6303.
Motivation:
At the moment only DefaultFileRegion is supported when using the native epoll transport.
Modification:
- Add support for any FileRegion implementation
- Add test case
Result:
Also custom FileRegion implementation are supported when using the epoll transport.
Motivation:
The NIO transport used an IllegalStateException if a user tried to issue another connect(...) while the connect was still in process. For this case the JDK specified a ConnectPendingException which we should use. The same issues exists in the EPOLL transport. Beside this the EPOLL transport also does not throw the right exceptions for ENETUNREACH and EISCONN errno codes.
Modifications:
- Replace IllegalStateException with ConnectPendingException in NIO and EPOLL transport
- throw correct exceptions for ENETUNREACH and EISCONN in EPOLL transport
- Add test case
Result:
More correct error handling for connect attempts when using NIO and EPOLL transport
Motivation:
We need to ensure we also call fireChannelActive() if the Channel is directly closed in a ChannelFutureListener that is belongs to the promise for the connect. Otherwise we will see missing active events.
Modifications:
Ensure we always call fireChannelActive() if the Channel was active.
Result:
No missing events.
Motivation:
In 4.0 AbstractNioByteChannel has a default of 16 max messages per read. However in 4.1 that constraint was applied at the NioSocketChannel which is not equivalent. In 4.1 AbstractEpollStreamChannel also did not have the default of 16 max messages per read applied.
Modifications:
- Make Nio consistent with 4.0
- Make Epoll consistent with Nio
Result:
Nio and Epoll both have consistent ChannelMetadata and are consistent with 4.0.
Motivation:
We use pre-instantiated exceptions in various places for performance reasons. These exceptions don't include a stacktrace which makes it hard to know where the exception was thrown. This is especially true as we use the same exception type (for example ChannelClosedException) in different places. Setting some StackTraceElements will provide more context as to where these exceptions original and make debugging easier.
Modifications:
Set a generated StackTraceElement on these pre-instantiated exceptions which at least contains the origin class and method name. The filename and linenumber are specified as unkown (as stated in the javadocs of StackTraceElement).
Result:
Easier to find the origin of a pre-instantiated exception.
Motivation:
JCTools supports both non-unsafe, unsafe versions of queues and JDK6 which allows us to shade the library in netty-common allowing it to stay "zero dependency".
Modifications:
- Remove copy paste JCTools code and shade the library (dependencies that are shaded should be removed from the <dependencies> section of the generated POM).
- Remove usage of OneTimeTask and remove it all together.
Result:
Less code to maintain and easier to update JCTools and less GC pressure as the queue implementation nt creates so much garbage
Motivation:
The DuplexChannel is currently incomplete and only supports shutting down the output side of a channel. This interface should also support shutting down the input side of the channel.
Modifications:
- Add shutdownInput and shutdown methods to the DuplexChannel interface
- Remove state in NIO and OIO for tracking input being shutdown independent of the underlying transport's socket type. Tracking the state independently may lead to inconsistent state.
Result:
DuplexChannel supports shutting down the input side of the channel
Fixes https://github.com/netty/netty/issues/5175
Motivation:
441aa4c575 introduced a bug in transport-native-epoll where readPending is set to false before a read is attempted, but this should happen before fireChannelRead is called. The NIO transport also only sets the readPending variable to false on the first read in the event loop. This means that if the user only calls read() on the first channelRead(..) the select loop will still listen for read events even if the user does not call read() on subsequent channelRead() or channelReadComplete() in the same event loop run. If the user only needs 2 channelRead() calls then by default they will may get 14 more channelRead() calls in the current event loop, and then 16 more when the event loop is woken up for a read event. This will also read data off the TCP stack and allow the peer to queue more data in the local RECV buffers.
Modifications:
- readPending should be set to false before each call to channelRead()
- make NIO readPending set to false consistent with EPOLL
Result:
NIO and EPOLL transport set readPending to false at correct times which don't read more data than intended by the user.
Fixes https://github.com/netty/netty/issues/5082
Motivation:
There is a spelling error in FileRegion.transfered() as it should be transferred().
Modifications:
Deprecate old method and add a new one.
Result:
Fix typo and can remove the old method later.
Motivation:
bfbef036a8 made EPOLL respect autoRead while in ET mode. However it is possible that we may miss data pending on the RECV queue if autoRead is off. This is because maybeMoreDataToRead is updated after fireChannelRead and if a user calls read() from here maybeMoreDataToRead will be false because it is updated after the fireChannelRead call. The way maybeMoreDataToRead was updated also causes a single channel to continuously read on the event loop and not relinquish and give other channels to try reading.
Modifications:
- Ensure maybeMoreDataToRead is always set after all user events, and is evaluated with readPending to execute a epollInReady on the EventLoop
- Combine the checkResetEpollIn and maybeMoreDataToRead logic to invoke a epollInReady later into the epollInFinally method due to similar responsibilities
- Update unit tests to reflect the user calling read() on the event loop from channelRead()
Result:
EPOLL ET with autoRead set to false will not leave data on the RECV queue.
Motivation:
We need to break out of the read loop for two reasons:
- If the input was shutdown in between (which may be the case when the user did it in the
fireChannelRead(...) method we should not try to read again to not produce any
miss-leading exceptions.
- If the user closes the channel we need to ensure we not try to read from it again as
the filedescriptor may be re-used already by the OS if the system is handling a lot of
concurrent connections and so needs a lot of filedescriptors. If not do this we risk
reading data from a filedescriptor that belongs to another socket then the socket that
was "wrapped" by this Channel implementation.
Modification:
Break the reading loop if the input was shutdown from within the channelRead(...) method.
Result:
No more meaningless exceptions and no risk to read data from wrong socket after the original was closed.
Motivation:
cf171ff525 introduced a change in behavior when dealing with closing channel in the read loop. This changed behavior may use stale state to determine if a channel should be shutdown and may be incorrect.
Modifications:
- Revert the usage of potentially stale state
Result:
Closing a channel in the read loop is based upon current state instead of potentially stale state.
Motivation:
The code of transport-native-epoll missed some things in terms of static keywords, @deprecated annotations and other minor things.
Modifications:
- Add missing @deprecated annotation
- Not using FQCN in javadocs
- Add static keyword where possible
- Use final fields when possible
- Remove throws IOException from method where it is not needed.
Result:
Cleaner code.
Motivation:
In commit acbca192bd we changed to have our native operations which either gall getsockopt or setsockopt throw IOExceptions (to be more specific we throw a ClosedChannelException in some cases). Unfortunally I missed to also do the same for getSoError() and missed to add throws IOException to the native methods.
Modifications:
- Correctly throw IOException from getSoError()
- Add throws IOException to native methods where it was missed.
Result:
Correct declaration of getSoError() and other native methods.
Motivation:
EPOLL does not support autoread when in ET mode.
Modifications:
- EpollRecvByteAllocatorHandle should not unconditionally force reading just because ET is enabled
- AbstractEpollChannel and all derived classes which implement epollInReady must support a variable which indicates
there may be more data to read. The variable will be used when read is called to simulate a EPOLL wakeup and call epollInReady if necessary. This will ensure that if we don't read until EAGAIN that we will try to read again and not rely on EPOLL to notify us.
Result:
EPOLL ET supports auto read.
Motivation:
We should retain the original hostname when connect to a remote peer so the user can still query the origin hostname if getHostString() is used.
Modifications:
Compute a InetSocketAddress from the original remote address and the one returned by the Os.
Result:
Same behavior when using epoll transport and nio transport.
Motivation:
AbstractEpollStreamChannel has a queue which collects splice events. Splice is assumed not to be the most common use case of this class and thus the splice queue could be initialized in a lazy fashion to save memory. This becomes more significant when the number of connections grows.
Modifications:
- AbstractEpollStreamChannel.spliceQueue will be initialized in a lazy fashion
Result:
Less memory consumption for most use cases
Motivation:
We should use OneTimeTask where possible to reduce object creation.
Modifications:
Replace Runnable with OneTimeTask
Result:
Less object creation
Motivation:
If we have a lot of writes going on we currently need to lookup the IovArray for each Channel that does writes. This can have quite some perf overhead. We should not need to do this and just store a reference of the IovArray on the EpollEventLoop itself.
Modifications:
- Remove IoArrayThreadLocal
- Store the IoArray in the EventLoop itself
Result:
Less FastThreadLocal lookups
Motivation:
The EPOLL module was not completly respecting the half closed state. It may have missed events, or procssed events when it should not have due to checking isOpen instead of the appropriate shutdown state.
Modifications:
- use FileDescriptor's isShutdown* methods instead of isOpen to check for processing events.
Result:
Half closed code in EPOLL module is more correct.
Motivation:
transport-native-epoll is designed to be specific to Linux. However there is native code that can be extracted out and made to work on more Unix like distributions. There are a few steps to be completely decoupled but the first step is to extract out code that can run in a more general Unix environment from the Linux specific code base.
Modifications:
- Move all non-Linux specific stuff from Native.java into the io.netty.channel.unix package.
- io.netty.channel.unix.FileDescriptor will inherit all the native methods that are specific to file descriptors.
- io_netty_channel_epoll_Native.[c|h] will only have code that is specific to Linux.
Result:
Code is decoupled and design is streamlined in FileDescriptor.
Motivation:
We should call shutdown(...) on the socket before closing the filedescriptor to ensure it is closed gracefully.
Modifications:
Call shutdown(...) before close.
Result:
Sockets are gracefully shutdown when using native transport.
Motivation:
writeBytes(...) missed to set EPOLLOUT flag when not all bytes were written. This could lead to have the EpollEventLoop not try to flush the remaining bytes once the socket becomes writable again.
Modifications:
- Move setting EPOLLOUT flag logic to one point so we are sure we always do it.
- Move OP_WRITE flag logic to one point as well.
Result:
Correctly try to write pending data if socket becomes writable again.
Motivation:
Commit cf171ff525 changed the way read operations were done. This change introduced a feedback loop between fireException and epollInReady.
Modifications:
- All EPOLL*Channel* classes should not call fireException and also continue to read. Instead a read operation should be executed on the eventloop (if the channel's input is not closed, and other conditions are satisfied)
Result:
Exception processing and channelRead will not be in a feedback loop.
Fixes https://github.com/netty/netty/issues/4091
Motivation:
If is enabled and a channel is half closed it is possible for the EPOLL event loop to get into an infinite loop by continuously being woken up on the EPOLLRDHUP event.
Modifications:
- Ensure that the EPOLLRDHUP event is unregistered for to prevent infinite loop.
Result:
1 less infinite loop.
Motiviation:
The current read loops don't fascilitate reading a maximum amount of bytes. This capability is useful to have more fine grain control over how much data is injested.
Modifications:
- Add a setMaxBytesPerRead(int) and getMaxBytesPerRead() to ChannelConfig
- Add a setMaxBytesPerIndividualRead(int) and getMaxBytesPerIndividualRead to ChannelConfig
- Add methods to RecvByteBufAllocator so that a pluggable scheme can be used to control the behavior of the read loop.
- Modify read loop for all transport types to respect the new RecvByteBufAllocator API
Result:
The ability to control how many bytes are read for each read operation/loop, and a more extensible read loop.
Motivation:
We missed to register for EPOLLRDHUP events when construct the EpollSocketChannel from an existing FileDescriptor. This could cause to miss connection-resets.
Modifications:
Add Native.EPOLLRDHUP to the events we are interested in.
Result:
Connection-resets are detected correctly.
