Commit Graph

320 Commits

Author SHA1 Message Date
Trustin Lee
6d1cd0d0cd ReferenceCountException -> IllegalReferenceCountException 2013-06-13 14:00:15 +09:00
Trustin Lee
7eb0f6105d Fix memory leaks 2013-06-13 13:32:47 +09:00
Trustin Lee
175526b6bd Move ReferenceCounted and AbstractReferenceCounted to io.netty.util
- Fixes #1441
- Also move and rename IllegalBufferAccessException to ReferenceCountException
- Prettier reference count exception messages
2013-06-13 13:14:21 +09:00
Trustin Lee
283feda119 Reduce even more garbage by exposing ByteBuf.internalNioBuffer() 2013-06-13 12:40:26 +09:00
Norman Maurer
78c6925921 Clearify when you need to call retain() 2013-06-12 23:13:36 +02:00
Norman Maurer
68c737f0c0 Optimize the way messages are added from one MessageList to another one 2013-06-10 14:07:25 +02:00
Norman Maurer
3be25694d0 Add ChannelHandlerContext.isRemoved() to easily detect the removal of a ChannelHandler while in a method. 2013-06-10 11:16:05 +02:00
Trustin Lee
14158070bf Revamp the core API to reduce memory footprint and consumption
The API changes made so far turned out to increase the memory footprint
and consumption while our intention was actually decreasing them.

Memory consumption issue:

When there are many connections which does not exchange data frequently,
the old Netty 4 API spent a lot more memory than 3 because it always
allocates per-handler buffer for each connection unless otherwise
explicitly stated by a user.  In a usual real world load, a client
doesn't always send requests without pausing, so the idea of having a
buffer whose life cycle if bound to the life cycle of a connection
didn't work as expected.

Memory footprint issue:

The old Netty 4 API decreased overall memory footprint by a great deal
in many cases.  It was mainly because the old Netty 4 API did not
allocate a new buffer and event object for each read.  Instead, it
created a new buffer for each handler in a pipeline.  This works pretty
well as long as the number of handlers in a pipeline is only a few.
However, for a highly modular application with many handlers which
handles connections which lasts for relatively short period, it actually
makes the memory footprint issue much worse.

Changes:

All in all, this is about retaining all the good changes we made in 4 so
far such as better thread model and going back to the way how we dealt
with message events in 3.

To fix the memory consumption/footprint issue mentioned above, we made a
hard decision to break the backward compatibility again with the
following changes:

- Remove MessageBuf
- Merge Buf into ByteBuf
- Merge ChannelInboundByte/MessageHandler and ChannelStateHandler into ChannelInboundHandler
  - Similar changes were made to the adapter classes
- Merge ChannelOutboundByte/MessageHandler and ChannelOperationHandler into ChannelOutboundHandler
  - Similar changes were made to the adapter classes
- Introduce MessageList which is similar to `MessageEvent` in Netty 3
- Replace inboundBufferUpdated(ctx) with messageReceived(ctx, MessageList)
- Replace flush(ctx, promise) with write(ctx, MessageList, promise)
- Remove ByteToByteEncoder/Decoder/Codec
  - Replaced by MessageToByteEncoder<ByteBuf>, ByteToMessageDecoder<ByteBuf>, and ByteMessageCodec<ByteBuf>
- Merge EmbeddedByteChannel and EmbeddedMessageChannel into EmbeddedChannel
- Add SimpleChannelInboundHandler which is sometimes more useful than
  ChannelInboundHandlerAdapter
- Bring back Channel.isWritable() from Netty 3
- Add ChannelInboundHandler.channelWritabilityChanges() event
- Add RecvByteBufAllocator configuration property
  - Similar to ReceiveBufferSizePredictor in Netty 3
  - Some existing configuration properties such as
    DatagramChannelConfig.receivePacketSize is gone now.
- Remove suspend/resumeIntermediaryDeallocation() in ByteBuf

This change would have been impossible without @normanmaurer's help. He
fixed, ported, and improved many parts of the changes.
2013-06-10 16:10:39 +09:00
Luke Wood
c1cbe8bec3 Clarify SnappyFramedDecoder javadoc to make it as clear as possible regarding the default settings 2013-05-20 17:31:04 +01:00
Norman Maurer
81e3c1719a [maven-release-plugin] prepare for next development iteration 2013-05-18 09:59:13 +02:00
Norman Maurer
99caefdf39 [maven-release-plugin] prepare release netty-4.0.0.CR3 2013-05-18 09:57:11 +02:00
Trustin Lee
e58af1bc94 Readability / Fix inception year 2013-05-17 17:59:13 +09:00
kxbmap
3a5aa12000 Add LengthFieldPrepender.lengthAdjustment parameter 2013-05-17 17:58:53 +09:00
Luke Wood
f841056752 Fix SnappyFramedDecoder checksum validation and add a pair of suitable unit tests 2013-05-16 14:10:54 +01:00
Trustin Lee
80f4c0b334 Merge SnappyChecksumUtil into Snappy and make calculateChecksum() public / Overall clean up 2013-05-09 15:23:56 +09:00
Norman Maurer
c43950a03f [maven-release-plugin] prepare for next development iteration 2013-05-08 18:19:51 +02:00
Norman Maurer
ae76502040 [maven-release-plugin] prepare release netty-4.0.0.CR2 2013-05-08 18:19:38 +02:00
Norman Maurer
108c7d9b44 No need to pass the next BufType to the constructor 2013-05-06 07:46:26 +02:00
Norman Maurer
73c35aef4e [#1309] Make sure ReplayDecoder respect isSingleDecode()
* This could cause for example corrupt WebSocketFrame's if they was written from the server
  to the client directly after it send the handshake response.
2013-04-25 08:32:07 +02:00
Trustin Lee
7884574c7b Remove freeInboundBuffer() and freeOutboundBuffer() which has no value
- Fixes #1308

freeInboundBuffer() and freeOutboundBuffer() were introduced in the early days of the new API when we did not have reference counting mechanism in the buffer. A user did not want Netty to free the handler buffers had to override these methods.

However, now that we have reference counting mechanism built into the buffer, a user who wants to retain the buffers beyond handler's life cycle can simply return the buffer whose reference count is greater than 1 in newInbound/OutboundBuffer().
2013-04-25 09:15:55 +09:00
Trustin Lee
1b3d7f5325 Make sure handlerAdded() is called before forwarding the buffer content of the removed handler
- Added a test case that reproduces the problem in ReplayingDecoderTest
- Call newHandler.handlerAdded() *before* oldHandler.handlerRemoved() to ensure newHandlerAdded() is called before forwarding the buffer content of the old handler in replace0().
2013-04-24 19:25:43 +09:00
Trustin Lee
5dd35448a2 Add a failing test that will pass once #1306 is fixed 2013-04-24 18:47:36 +09:00
Trustin Lee
b5989e2449 Reduce exception instantiation overhead in SslHandler / Reduce unnecessary empty array creation
- Added EmptyArrays as an internal utility class
2013-04-24 09:32:53 +09:00
Trustin Lee
8e2e22c270 Change the thread model slightly for new/freeInbound/OutboundBuffer() for future improvement
- Related: #1283
- Make ReplayingDecoder work with the modified thread model
2013-04-23 13:06:27 +09:00
Norman Maurer
9c4bfa44d9 [#1294] Make sure ByteBuf is released once written to channel 2013-04-22 11:04:56 +02:00
Norman Maurer
ca5554dfe7 [#1236] Fix problem where adding a new ChannelHandler could block the eventloop
This change also introduce a few other changes which was needed:
 * ChannelHandler.beforeAdd(...) and ChannelHandler.beforeRemove(...) were removed
 * ChannelHandler.afterAdd(...) -> handlerAdded(...)
 * ChannelHandler.afterRemoved(...) -> handlerRemoved(...)
 * SslHandler.handshake() -> SslHandler.hanshakeFuture() as the handshake is triggered automatically after
   the Channel becomes active
2013-04-19 07:00:50 +02:00
Trustin Lee
4a5dc32224 Fix a bug where LineBasedFrameDecoder does not handle too long lines correctly
- Related: #1287
2013-04-19 13:04:20 +09:00
Andrei Pozolotin
a3e760a003 fix #1234 - duplicate package-info.java errors in eclipse requires release of netty-build v 19 and netty-parent update. 2013-04-05 05:38:05 +09:00
Trustin Lee
6dfa455f9e Fix compiler warnings 2013-04-04 16:31:41 +09:00
Trustin Lee
c03179c81c Clean up try/catch blocks / Prefer 'throw' to 'fireExceptionCaught' 2013-04-04 15:00:07 +09:00
Trustin Lee
3b9994455a Saner field order 2013-04-04 14:53:32 +09:00
Trustin Lee
03955428bc Ensure BufUtil.release() is called even if writeBytes() fails in writeAndRelease() 2013-04-04 14:50:25 +09:00
Trustin Lee
015e60b00f Clean up catch blocks
- No need to have nested try blocks because the catch block catches everything and rethrows it
- No need to do instanceof-checks
2013-04-04 14:48:30 +09:00
Trustin Lee
c25fd78ca0 No need to have nested try blocks
.. because the catch block catches everything and rethrows it
2013-04-04 14:44:52 +09:00
Trustin Lee
788b88b7af Make sure to release the buffer when it's written to the next byte buffer
.. otherwise it will leak.
2013-04-04 14:37:54 +09:00
Trustin Lee
46540578fc Optimize OutputMessageBuf.drainToNextInbound/Outbound()
- Return early when the buffer is empty
- Keep only the number of byte buffers
- Remove unnecessary null check in the loop (because we know buffer is not empty at certain point)
2013-04-04 14:32:48 +09:00
Trustin Lee
8a672c2800 Proper assertion instead of blindly clearing the OutputMessageBuf 2013-04-04 14:11:24 +09:00
Norman Maurer
7ee2adb587 Move drain logic to OutputMessageBuf and optimize it as far as possible 2013-04-03 18:52:57 +02:00
Norman Maurer
1675e61f5b Share same ThreadLocal for all decoder/encoders to minimize memory usage 2013-04-03 18:19:35 +02:00
Norman Maurer
94ef7dc1b9 Optimize to minimize volatile reads to access next buffer in codec framework 2013-04-03 18:03:55 +02:00
Norman Maurer
9828267165 Fix performance regression which was caused by calling the fireInboundBufferUpdated() a way to often
* Also use a ThreadLocal for the decoder buffer to safe space as it is cheap
2013-04-03 17:07:52 +02:00
Norman Maurer
af4b71a00e Remove special handling of Object[] in codec framework (a.k.a unfolding)
- Fixes #1229
- Primarily written by @normanmaurer and revised by @trustin

This commit removes the notion of unfolding from the codec framework
completely.  Unfolding was introduced in Netty 3.x to work around the
shortcoming of the codec framework where encode() and decode() did not
allow generating multiple messages.

Such a shortcoming can be fixed by changing the signature of encode()
and decode() instead of introducing an obscure workaround like
unfolding.  Therefore, we changed the signature of them in 4.0.

The change is simple, but backward-incompatible.  encode() and decode()
do not return anything.  Instead, the codec framework will pass a
MessageBuf<Object> so encode() and decode() can add the generated
messages into the MessageBuf.
2013-04-03 21:44:54 +09:00
Norman Maurer
2e3b8f13ea Revert "Let ProtobufEncoder extend MessageToByteEncoder which makes more sense. Related to [#1222]"
This reverts commit 6a9d055f57.
2013-04-02 07:56:49 +02:00
Norman Maurer
6a9d055f57 Let ProtobufEncoder extend MessageToByteEncoder which makes more sense. Related to [#1222] 2013-04-02 07:17:45 +02:00
Norman Maurer
a0fca93128 [#1222] Correctly handle the case when a MessageToMessageEncoder encodes to ByteBuf and the next outbound buffer is of type ByteBuf 2013-04-02 07:17:10 +02:00
Prajwal Tuladhar
05850da863 enable checkstyle for test source directory and fix checkstyle errors 2013-03-30 13:18:57 +01:00
Norman Maurer
59012390f6 Fix version numbering 2013-03-25 08:01:11 +01:00
Norman Maurer
7d7b676eeb [maven-release-plugin] prepare for next development iteration 2013-03-22 15:20:35 +01:00
Norman Maurer
60fc7dac4d [maven-release-plugin] prepare release netty-4.0.0.CR1 2013-03-22 15:20:11 +01:00
Trustin Lee
8b722d29a7 Add constructor parameters that do not perform type parameter auto-detection for the languages without type parameters
- Fixes #1177
- Add TypeParameterMatcher.get(parameterType)
- Add alternative constructors
2013-03-21 16:11:47 +09:00