Commit Graph

6133 Commits

Author SHA1 Message Date
Norman Maurer
cd2eaebd92 [#1895] Fix IllegalStateException which was produced during failing ChunkedWrite after the channel was closed 2013-10-08 06:36:59 +02:00
Trustin Lee
8864792499 Fixes the problem where the promise of the outbound operation that causes a channel closure is notified after channelInactive()
- Fixes #1897
2013-10-08 12:24:32 +09:00
Trustin Lee
3c7d458c6e Fix the problem where HashedWheelTimer puts a timeout into an incorrect place
- the stopIndex of a timeout is calculated based on the start time of the worker thread and the current tick count for greater accuracy
2013-10-07 17:09:00 +09:00
Norman Maurer
f4edb2f689 [#1893] Fix bug in JdkZlibDecoder which did not let it handle large data 2013-10-07 08:07:53 +02:00
Norman Maurer
44293291a7 Use direct ByteBuf for the test to make sure it is not copied 2013-10-07 08:07:38 +02:00
Norman Maurer
7ad3c1e10c [#1890] Correctly expand ByteBuffer array in all cases
The problem was that we did not handle the case correctly when doubling the array was not enough. We need to keep doubling until everything fits in.
2013-10-06 15:30:51 +02:00
Norman Maurer
2366c2846d [#1881] Use correct field name in NullPointerException 2013-10-04 09:54:21 +02:00
Bill Gallagher
805a9dff8e disable debugging output during test 2013-10-04 06:58:52 +02:00
Norman Maurer
eb24b572dc [#1885] Correctly close leak detected also on non started HashedWheelTimer 2013-10-02 06:44:55 +02:00
Norman Maurer
a4a207fc4f Fix checkstyle 2013-10-02 06:44:10 +02:00
Bill Gallagher
a5220198ba use enum for state in LocalChannel 2013-10-02 06:32:03 +02:00
Bill Gallagher
f80109f423 use enum for state in EmbeddedChannel 2013-10-01 21:08:45 +02:00
Norman Maurer
4bca1c3fe3 [#1876] Make use of proper state machine in WebSocket08FrameDecoder for performance reasons 2013-10-01 10:20:01 +02:00
Norman Maurer
1ecd1e01a5 [#1880] Use ByteBufAllocator when read bytes into new chunks 2013-10-01 09:27:02 +02:00
Norman Maurer
5a59cb9fb7 [#1879] Correctly validate close frame 2013-10-01 07:32:07 +02:00
Norman Maurer
aaafdf909d [#1878] Fix leak of ByteBuf when masked payload is used 2013-10-01 07:18:16 +02:00
Norman Maurer
328f67fdfe [#1874] WebSocket08FrameDecoder may leak memory if channel is closed before the full frame was received 2013-09-30 20:43:30 +02:00
Bill Gallagher
20d5361403 [#1832] - Channel writability change notifications sometimes fail to fire 2013-09-30 20:09:28 +02:00
Norman Maurer
512908f993 [#1875] Correctly check the readerIndex when try to read a byte from AbstractByteBuf 2013-09-30 14:49:03 +02:00
Jeff Pinner
321990a115 SPDY: add SPDY/3.1 support
- with Michael Schore <mschore@twitter.com>
2013-09-28 12:58:44 +02:00
Norman Maurer
c2101d3c56 Cache underlying ByteBuffers and count in ChannelOutboundBuffer.Entry to reduce object creation and so GC pressure
Beside this it also helps to reduce CPU usage as nioBufferCount() is quite expensive when used on CompositeByteBuf which are
nested and contains a lot of components
2013-09-26 21:01:40 +02:00
Norman Maurer
fc805a0787 CompositeByteBuf.isDirect() should return true if its only backed by direct buffers 2013-09-26 20:50:59 +02:00
Norman Maurer
213d195909 [#1865] Only use internalNioBuffer when one of the read* or write* methods are used. This is neccessary to prevent races as those can happen when a slice or duplicate is shared between different Channels
that are not assigned to the same EventLoop. In general get* operations should always be safe to be used from different Threads.

This aslo include unit tests that show the issue
2013-09-25 17:18:47 +02:00
Norman Maurer
52660dfcfa Propagate channelWritabilityChanged() through the pipeline after flushing. Related to [#1861] 2013-09-24 14:19:53 +02:00
Norman Maurer
f35ba4f80f Correctly handle automatically suspend/resume in ChunkedWriteHandler. Related to [#1861]
The old implementation was broken and could lead to pending message never be picked up again until the user either explicit called flush or
resumeTransfer().
2013-09-24 14:19:38 +02:00
Norman Maurer
cd5f9a2212 Introduce a new ChannelOption called DATAGRAM_CHANNEL_ACTIVE_ON_REGISTRATION. Related to [#1830]
This ChannelOption allows to tell the DatagramChannel implementation to be active as soon as they are registrated to their EventLoop. This can be used to make it possible to write to a not bound DatagramChannel.
The ChannelOption is marked as @deprecated as I'm looking for a better solution in master which breaks default behaviour with 4.0 branch.
2013-09-24 11:51:54 +02:00
Norman Maurer
5aa2b7e9f7 [#1855] Try to calculate the correct amount of written bytes to update the ChannelProgressiveFuture 2013-09-24 07:50:02 +02:00
Arron Norwell
5a0d7fe19a HttpRequestEncoder should append '/' to absolute path requests only when needed 2013-09-22 13:57:10 +02:00
Norman Maurer
1c7e107bef [#1851] EmptyByteBuf.isWritable(..) and isReadable(...) should not throw IndexOutOfBoundsException 2013-09-21 20:40:08 +02:00
Sasha Zverev
71c062167d Slip in DefaultOioSocketChannelConfig (setAllowHalfClosure used to ignore argument) 2013-09-21 20:14:12 +02:00
Norman Maurer
a855f19ad6 [#1853] Optimize gathering writes for CompositeByteBuf that are only backed by one ByteBuffer 2013-09-19 07:29:21 +02:00
Norman Maurer
0e1dcb91ff [#1852] Fix bug in UnpooledDirectByteBuf.nioBuffer(...) implementation 2013-09-18 20:48:40 +02:00
Norman Maurer
3cfcf09af8 More efficient handling of incomplete writes.
The problem with the old way was that we always set the OP_WRITE when the buffer could not be written
until the write-spin-count was reached. This means that in some cases the channel was still be writable
but we just was not able to write out the data quick enough. For this cases we should better break out the
write loop and schedule a write to be picked up later in the EventLoop, when other tasks was executed.
The OP_WRITE will only be set if a write actual returned 0 which means there is no more room for writing data
and this we need to wait for the os to notify us.
2013-09-17 07:01:05 +02:00
Norman Maurer
8dc57f6933 [#1836] Add comment to explain why read is triggered 2013-09-17 06:58:23 +02:00
Greg Soltis
335f3807a5 Fix nioBuffer implementation for CompositeByteBuf 2013-09-16 06:57:32 +02:00
Norman Maurer
c0b9ef9386 [#1833] Add testcase for fix 2013-09-14 11:53:30 +02:00
Phillip Schichtel
d885394b8a [#1833] Replace ; with & only in the QueryString and not the whole URI 2013-09-14 11:52:39 +02:00
Trustin Lee
4aab9e0773 No need to use an unreleasable buffer - just wrap an array 2013-09-13 14:09:51 -07:00
Norman Maurer
87a6a68059 [#1830] Add testcase for write to not bound DatagramChannel impls and revert change in OIO as it breaks things as the udnerlying socket lazy binds 2013-09-12 09:29:40 +02:00
Norman Maurer
50106f3d41 Make DatagramChannel impls .isActive() return true when the underlying Channel is open. This also fixes [#1830] 2013-09-11 20:38:04 +02:00
Norman Maurer
98633e0aee Allow to only register a Channel via AbstractBootstrap and bind/connect it later. Related to [#1829] 2013-09-11 20:37:53 +02:00
Norman Maurer
343ecaa111 Add test to show that writing a CompositeByteBuf when SslHandler is in use works. Related to [#1825] 2013-09-11 07:29:02 +02:00
Norman Maurer
16cf807fc5 Set memory for compiler as otherwise it sometimes fails here 2013-09-10 18:39:02 +02:00
Norman Maurer
3b44c602db [#1821] Fix IndexOutOfBoundsException which was thrown if the last component was removed but other components was left 2013-09-10 18:38:52 +02:00
Norman Maurer
12f325767c [#1819] Remove HTTP compressor and decompressor from the pipeline after websockets handshake completes 2013-09-09 11:55:25 +02:00
Norman Maurer
c89f1b0a03 [#1818] Pass through message as they are when no compression is needed 2013-09-09 11:43:45 +02:00
Norman Maurer
198dde33b3 Optimize the http-codec to be much faster then before.
This includes the following changes:
  * Rewrite HttpObjectDecoder to use ByteToMessageDecoder
  * Optimize parsing of the header
  * Allow to disable the header validation for performance reason
  * Not need to validate kwown header names and values
  * Minimize access of ThreadLocals
  * Optimize parsing of initial line
2013-09-05 20:21:04 +02:00
Norman Maurer
6716dca17a Use a Thread-local based direct buffer pool if non pooled allocator is used 2013-09-05 11:58:03 +02:00
Trustin Lee
95576d6559 Ensure operationProgressed is invoked even on completion
- Fixes #1809
2013-09-05 18:36:48 +09:00
Norman Maurer
e46ba9e8de Move encode of chunked content to an extra method, this allows for inline. Related to #1812
encode of chunked content is not the most common pattern so moving it to an extra method makes it possible to inline the rest as it is a smaller method now.
2013-09-05 10:15:51 +02:00