Commit Graph

148 Commits

Author SHA1 Message Date
nmittler
2a2059d976 Adding UniformStreamByteDistributor
Motivation:

The current priority algorithm can yield poor per-stream goodput when either the number of streams is high or the connection window is small. When all priorities are the same (i.e. priority is disabled), we should be able to do better.

Modifications:

Added a new UniformStreamByteDistributor that ignores priority entirely and manages a queue of streams.  Each stream is allocated a minimum of 1KiB on each iteration.

Result:

Improved goodput when priority is not used.
2015-11-19 16:49:12 -08:00
nmittler
96f9b0b91b Remote flow controller incorrectly updates stream state
Motivation:

The `DefaultHttp2RemoteFlowController` does not correctly determine `hasFrame` when updating the stream state for the distributor. Adding a check to enforce `hasFrame` when `streamableBytes > 0` causes several test failures.

Modifications:

Modified `DefaultHttp2RemoteFlowController` to simplify the writing logic and to correct the bookkeeping for `hasFrame`.

Result:

The distributors are always called with valid arguments.
2015-11-18 11:32:18 -08:00
nmittler
8accc52b03 Forking Twitter's hpack
Motivation:

The twitter hpack project does not have the support that it used to have.  See discussion here: https://github.com/netty/netty/issues/4403.

Modifications:

Created a new module in Netty and copied the latest from twitter hpack master.

Result:

Netty no longer depends on twitter hpack.
2015-11-14 10:13:32 -08:00
Norman Maurer
2ecce8fa56 [maven-release-plugin] prepare for next development iteration 2015-11-10 22:59:33 +01:00
Norman Maurer
6a93f331d3 [maven-release-plugin] prepare release netty-4.1.0.Beta8 2015-11-10 22:50:57 +01:00
Scott Mitchell
2f81364522 DefaultHttp2HeadersTest updates
Motivation:
Recently a bug was found in DefaultHttp2Headers where the state of the headers could be corrupted due to the extra tracking to make pseudo headers first during iteration. Unit tests did not catch this bug.

Modifications:
- Update unit tests to cover more methods

Result:
Unit tests for DefaultHttp2Headers have better code coverage.
2015-11-07 10:14:00 -08:00
Louis Ryan
7cc320ce47 Fix memory leak in DefaultHttp2Headers
Motivation:

Memory leak makes headers non-reusable.

Modifications:

Correctly reset firstNonPseudo header reference

Result:

No leak
2015-11-06 07:08:57 -08:00
nmittler
6504d52b94 Add HTTP/2 local flow control option for auto refill
Motivation:

For many HTTP/2 applications (such as gRPC) it is necessary to autorefill the connection window in order to prevent application-level deadlocking.

Consider an application with 2 streams, A and B.  A receives a stream of messages and the application pops off one message at a time and makes a request on stream B. However, if receiving of data on A has caused the connection window to collapse, B will not be able to receive any data and the application will deadlock.  The only way (currently) to get around this is 1) use multiple connections, or 2) manually refill the connection window.  Both are undesirable and could needlessly complicate the application code.

Modifications:

Add a configuration option to DefaultHttp2LocalFlowController, allowing it to autorefill the connection window.

Result:

Applications can configure HTTP/2 to avoid inter-stream deadlocking.
2015-11-05 15:47:10 -08:00
Scott Mitchell
91b8ef3d10 HTTP/2 PriorityStreamByteDistributor exceptions and reentry
Motivation:
PriorityStreamByteDistributor saves exception state and attempts to reset state. This could be simplified by just throwing a connection error and closing the connection. PriorityStreamByteDistributor also does not handle or detect re-entry in the distribute method.

Motivation:
- PriorityStreamByteDistributor propagate an INTERNAL_ERROR if an exception occurs during writing
- PriorityStreamByteDistributor to handle re-entry on the write method

Result:
PriorityStreamByteDistributor exception code state simplified, and re-entry is detected.
2015-11-03 13:22:11 -08:00
Trustin Lee
8f334885ef Reject the first SETTINGS ack on HTTP/2 Preface
Motivation:

Http2ConnectionHandler verifies if the first frame after the preface is
a SETTINGS frame.  However, it does not reject the SETTING ack frame
which is not expected actually.

Modifications:

Reject a SETTINGS-ack frame as well

Result:

When the first frame is a SETTINGS-ack frame, connection does not
proceed to further frame handling. (simplicity)
2015-11-03 11:23:54 +09:00
Scott Mitchell
19658e9cd8 HTTP/2 Headers Type Updates
Motivation:
The HTTP/2 RFC (https://tools.ietf.org/html/rfc7540#section-8.1.2) indicates that header names consist of ASCII characters. We currently use ByteString to represent HTTP/2 header names. The HTTP/2 RFC (https://tools.ietf.org/html/rfc7540#section-10.3) also eludes to header values inheriting the same validity characteristics as HTTP/1.x. Using AsciiString for the value type of HTTP/2 headers would allow for re-use of predefined HTTP/1.x values, and make comparisons more intuitive. The Headers<T> interface could also be expanded to allow for easier use of header types which do not have the same Key and Value type.

Motivation:
- Change Headers<T> to Headers<K, V>
- Change Http2Headers<ByteString> to Http2Headers<CharSequence, CharSequence>
- Remove ByteString. Having AsciiString extend ByteString complicates equality comparisons when the hash code algorithm is no longer shared.

Result:
Http2Header types are more representative of the HTTP/2 RFC, and relationship between HTTP/2 header name/values more directly relates to HTTP/1.x header names/values.
2015-10-30 15:29:44 -07:00
Scott Mitchell
d66520db1b Http2ConnectionHandler.BaseBuilder exception cleanup
Motivation:
Http2ConnectionHandler.BaseBuilder is constructing objects which have 'close' methods, but is not calling these methods in the event of an exception.

Modifications:
- Objects which implement 'close' should have this method called if an exception is thrown and the build operation can not complete normally.

Result:
Objects are closed even if the build process encounters an error.
2015-10-29 11:09:51 -07:00
Norman Maurer
077af8c019 Remove encoderMaxConcurrentStreams
Motivation:

Remove encoderMaxConcurrentStreams(...) and use the default settings. Also throw an exception if server mode is used.

Modifications:

- Remove encoderMaxConcurrentStreams(...) method
- Throw exception if server mode is used and trying to enforce conncurrent streams.

Result:

Correctly support settings stuff via builder
2015-10-15 10:11:39 +02:00
Norman Maurer
e419533498 Remove unused parameter from method declaration.
Motivation:

We had an unused paramter on a method, we should just remove it to keep code clean.

Modifications:

- Remove parameter
- Fix typo in javadoc

Result:

Cleanup done.
2015-10-07 09:18:22 +02:00
Norman Maurer
2ff2806ada [maven-release-plugin] prepare for next development iteration 2015-10-02 09:03:29 +02:00
Norman Maurer
5a43de10f7 [maven-release-plugin] prepare release netty-4.1.0.Beta7 2015-10-02 09:02:58 +02:00
Scott Mitchell
06c3ae07a0 DefaultHttp2ConnectionDecoder write ping buffer
Motivation:
DefaultHttp2ConnectionDecoder writes a ACK when receiving a ping frame and sends the same data buffer it received. The data buffer is also passed to the listener, but the indexes are shared between the send and the listener. We should ensure the indexes are independent for these two operations.

Modifications:
- Call slice on the buffer that is being sent

Result:
Listener now has access to a buffer that will not appear to be already consumed.
2015-10-02 11:40:21 -07:00
Scott Mitchell
284e3702d8 Http2ConnectionHandler Builder instead of constructors
Motivation:
Using the builder pattern for Http2ConnectionHandler (and subclasses) would be advantageous for the following reasons:
1. Provides the consistent construction afforded by the builder pattern for 'optional' arguments. Users can specify these options 1 time in the builder and then re-use the builder after this.
2. Enforces that the Http2ConnectionHandler's internals (decoder Http2FrameListener) are initialized after construction.

Modifications:
- Add an extensible builder which can be used to build Http2ConnectionHandler objects
- Update classes which inherit from Http2ConnectionHandler

Result:
It is easier to specify options and construct Http2ConnectionHandler objects.
2015-10-01 13:51:03 -07:00
Scott Mitchell
1485a87e25 Http2ConnectionHandler and Http2FrameListener cyclic dependency
Motivation:
It is often the case that implementations of Http2FrameListener will want to send responses when data is read. The Http2FrameListener needs access to the Http2ConnectionHandler (or the encoder contained within) to be able to send responses. However the Http2ConnectionHandler requires a Http2FrameListener instance to be passed in during construction time. This creates a cyclic dependency which can make it difficult to cleanly accomplish this relationship.

Modifications:
- Add Http2ConnectionDecoder.frameListener(..) method to set the frame listener. This will allow the listener to be set after construction.

Result:
Classes which inherit from Http2ConnectionHandler can more cleanly set the Http2FrameListener.
2015-09-30 15:41:15 -07:00
Scott Mitchell
0e9545e94d Http2RemoteFlowController stream writibility listener
Motivation:
For implementations that want to manage flow control down to the stream level it is useful to be notified when stream writability changes.

Modifications:
- Add writabilityChanged to Http2RemoteFlowController.Listener
- Add isWritable to Http2RemoteFlowController

Result:
The Http2RemoteFlowController provides notification when writability of a stream changes.
2015-09-28 13:47:24 -07:00
nmittler
7ab132f28a Making HTTP/2 stream byte assignment pluggable
Motivation:

The DefaultHttp2RemoteFlowController has become very large and is getting difficult to understand and maintain. It is also desirable for some applications to be able to disable the priority algorithm altogether for performance reasons.

Modifications:

Abstract the stream byte assignment logic (renamed allocation->assignment for clarity) behind an interface `StreamByteAssigner` with a single implementation `PriorityStreamByteAssigner`.

Result:

Goes some way towards supporting #4246
2015-09-25 14:00:12 -07:00
Scott Mitchell
93011dd315 DefaultHttp2RemoteFlowController not allocating all available bytes
Motivation:
DefaultHttp2RemoteFlowController's allocation algorithm may not allocate all bytes that are available in the connection window. If the 'fair share' based upon weight is not fully used by sibling nodes it was not correctly re-distributed to other sibilings which may be able to utilize part / all of that share.

Modifications:
- Add a unit test which demonstrates the issue.
- Modify the allocation algorithm to ensure all available bytes are allocated.

Result:
Fixes https://github.com/netty/netty/issues/4266
2015-09-25 11:12:04 -07:00
Scott Mitchell
c372f69118 http2.HttpConversionUtil :authority conversion error
Motiviation:
The http2 spec https://tools.ietf.org/html/rfc7540#section-8.1.2.3 states that the :authority header should be copied into the HOST header when converting from HTTP/2 to HTTP/1.x. We currently have an extension header to preserve the authority.

Modifications:
- Remove AUTHORITY extension header
- HTTP/2 :authority should map to HOST header when converting to HTTP/1.x.

Result:
More spec compliant.
2015-09-23 17:06:52 -07:00
Scott Mitchell
ed4928f62a StreamBufferingEncoderTest leak
Motivation:
Buffer leak in StreamBufferingEncoderTest

Modifications:
- Make sure buffers are released in StreamBufferingEncoderTest

Result:
Fixes https://github.com/netty/netty/issues/4230
2015-09-23 16:49:39 -07:00
Scott Mitchell
edb91afcd6 Http2LifecycleManager.onException rename
Motivation:
Http2LifecycleManager.onException takes a Throwable as a paramter and not an Exception. There are also onConnectionError and onStreamError methods in the codec. We should rename this method to onError for consistency and clarity.

Modifications:
- Rename Http2LifecycleManager.onException to Http2LifecycleManager.onError

Result:
More consistent and clarified interface.
2015-09-23 16:48:36 -07:00
Scott Mitchell
24c9407080 DefaultHttp2RemoteFlowController may not write all pending bytes
Motivation:
DefaultHttp2RemoteFlowController attempts to write as many bytes as possible to transition the channel to not writable, and then relies on notification of channelWritabilityChange to continue writing. However the amount of bytes written by DefaultHttp2RemoteFlowController may not be the same number of bytes that is actually written to the channel due to other ChannelHandlers (SslHandler, compression, etc...) in the pipeline. This means there is a potential for the DefaultHttp2RemoteFlowController to be waiting for a channel writaiblity change event that will never come, and thus not write all queued data.

Modifications:
- DefaultHttp2RemoteFlowController should write pending bytes until there are no more, or until the channel is not writable.

Result:
DefaultHttp2RemoteFlowController will write all pending data.
Fixes https://github.com/netty/netty/issues/4242
2015-09-23 16:39:24 -07:00
nmittler
ec20902613 Don't set HTTP/2 flow controller ctx to null
Motivation:

We currently set the flow controller ChannelHandlerContexts to null when the channel becomes inactive. This is bad :)

Modifications:

Just remove that code in Http2ConnectionHandler

Result:

Fixes #4240
2015-09-22 07:25:42 -07:00
fratboy
6241bb059c [#4244] Convert urlencoded uri to http2 path correctly
Motivation:

HttpConversionUtil.toHttp2Headers does not convert urlencoded uri to http2 path properly.

Modifications:

Use getRawPath(), getRawQuery(), getRawFragment() in java.net.URI when converts to http2 path

Result:

HttpConversionUtil.toHttp2Headers does not urldecode uri unproperly.
2015-09-21 16:30:59 -07:00
Scott Mitchell
c7e3f6c6fd HTTP/2 defines using String instead of CharSequence
Motivation:
Http2CodecUtils has some static variables which are defined as Strings instead of CharSequence. One of these defines is used as a header name and should be AsciiString.

Modifications:
- Change the String defines in Http2CodecUtils to CharSequence

Result:
Types are more consistently using CharSequence and adding the upgrade header will require less work.
2015-09-16 14:55:33 -07:00
Scott Mitchell
1d4d5fe312 DefaultHttp2Headers should throw exception of type Http2Exception
Motivation:
The DefaultHttp2Headers code is throwing a IllegalArgumentException if an invalid character is detected. This is being ignored by the HTTP/2 codec instead of generating a GOAWAY.

Modifications:
- Throw a Http2Exception of type PROTOCOL_ERROR in accordance with https://tools.ietf.org/html/rfc7540#section-8.1.2.6
- Update examples which were building invalid headers

Result:
More compliant with https://tools.ietf.org/html/rfc7540#section-8.1.2.6
2015-09-16 13:47:05 -07:00
Scott Mitchell
15450af2e7 DefaultHttp2FrameWriter ping payload size check
Motivation:
The HTTP/2 spec states that the ping frame length must be 8 and is otherwise an error https://tools.ietf.org/html/rfc7540#section-6.7. The DefaultHttp2FrameReader enforces this, but the DefaultHttp2FrameWriter allows invalid frames to be written. We should not allow invalid ping frames to be written to the network.

Modifications:
- DefaultHttp2FrameWriter checks the frame size to be 8, or throws an exception

Result:
Fixes https://github.com/netty/netty/issues/3721
2015-09-16 10:25:07 -07:00
Scott Mitchell
59600f1812 HTTP/2 to HTTP/1.x headers conversion more accessible
Motivation:
Currently there is a HttpConversionUtil.addHttp2ToHttpHeaders which requires a FullHttpMessage, but this may not always be available. There is no interface that can be used with just Http2Headers and HttpHeaders.

Modifications:
- add an overload for HttpConversionUtil.addHttp2ToHttpHeaders which does not take FullHttpMessage

Result:
An overload for HttpConversionUtil.addHttp2ToHttpHeaders exists which does not require FullHttpMessage.
2015-09-16 10:02:48 -07:00
Scott Mitchell
ba11879c9f HTTP/2 codec heap buffer usage
Motivation:
The HTTP/2 codec has a few static buffers sent over the network which are allocated on the heap. This results in a copy operation when the buffer is sent out on the network.

Modifications:
- Ensure these static buffers are allocated using direct memory.

Result:
No copy operation necessary when writing static buffers to network.
2015-09-14 13:12:02 -07:00
Scott Mitchell
f89dfb0bd5 Deprecation cleanup for HTTP headers
Motivaion:
The HttpHeaders and DefaultHttpHeaders have methods deprecated due to being removed in future releases, but no replacement method to use in the current release. The deprecation policy should not be so aggressive as to not provide any non-deprecated method to use.

Modifications:
- Remove deprecated annotations and javadocs from methods which are the best we can do in terms of matching the master's api for 4.1

Result:
There should be non-deprecated methods available for HttpHeaders in 4.1.
2015-09-09 14:30:21 -07:00
Scott Mitchell
47726991b2 HTTP/2 Header Name Validation
Motivation:
The HTTP/2 header name validation was removed, and does not currently exist.

Modifications:
- Header name validation for HTTP/2 should be restored and set to the default mode of operation.

Result:
HTTP/2 header names are validated according to https://tools.ietf.org/html/rfc7540
2015-09-09 13:59:08 -07:00
Scott Mitchell
50d1f0a680 Http2Headers.iterator() comment correction
Motivation:
The javadoc comments on Http2Headers.iterator() are incorrect.

Modifications:
- Correct and clarify the javadoc for Http2Headers.iterator()

Result:
Javadoc for Http2Headers.iterator() is more correct.
2015-09-04 12:43:26 -07:00
Norman Maurer
34de2667c7 [maven-release-plugin] prepare for next development iteration 2015-09-02 11:45:20 +02:00
Norman Maurer
2eb444ec1d [maven-release-plugin] prepare release netty-4.1.0.Beta6 2015-09-02 11:36:11 +02:00
Scott Mitchell
41ee9148e5 HTTP/2 InboundHttp2ToHttpAdapterTest serverChannel NPE
Motivation:
InboundHttp2ToHttpAdapterTest.bootstrapEnv does not wait for the serverConnectedChannel to be initialized before returning. Some methods rely only this behavior and throw a NPE because it may not be set.

Modifications:
- Add a CountDownLatch to ensure the serverConnectedChannel is initialized

Result:
No more NPE.
2015-09-01 10:38:04 -07:00
Scott Mitchell
0736a3bc35 HTTP/2 SimplePromiseAggregator tryFailure not consistent with setFailure
Motivation:
The SimplePromiseAggregator.setFailure allows a failure to occur before newPromise is called, but tryFailure doesn't. These methods should be consistent.

Modifications:
- tryFailure should use the same logic as setFailure

Result:
Consistent failure routines.
2015-09-01 10:35:10 -07:00
Scott Mitchell
50cc647804 DefaultPropertyKey private member variable accessed outside scope
Motivation:
DefaultPropertyKey.index is currently private and accessed outside the class's scope.

Modifications:
- Change access level to package private

Result:
No chance of synthetic method generation for accessing this field
2015-08-28 08:54:11 -07:00
Scott Mitchell
0365927951 HTTP/2 InboundHttp2ToHttpAdapterTest race condition
Motivation:
The latches in InboundHttp2ToHttpAdapterTest were volatile and reset during the tests. This resulted in race conditions and sometimes the tests would be waiting on old latches that were not the same latches being counted down when messages were received.

Modifications:
- Remove volatile latches from tests

Result:
More reliable tests with less race conditions.
2015-08-28 08:50:08 -07:00
Scott Mitchell
b6a4f5de9d Refactor of HttpUtil and HttpHeaderUtil
Motivation:
There currently exists http.HttpUtil, http2.HttpUtil, and http.HttpHeaderUtil. Having 2 HttpUtil methods can be confusing and the utilty methods in the http package could be consolidated.

Modifications:
- Rename http2.HttpUtil to http2.HttpConversionUtil
- Move http.HttpHeaderUtil methods into http.HttpUtil

Result:
Consolidated utilities whose names don't overlap.
Fixes https://github.com/netty/netty/issues/4120
2015-08-27 08:49:58 -07:00
Scott Mitchell
14dc571956 Http2ConnectionHandler channelInactive sequencing
Motivation:
ByteToMessageDecoder may call decode after channelInactive is called. This will lead to a NPE.

Modifications:
- Call super.channelInactive() before we process the event in Http2ConnectionHandler

Result:
No more NPE in decode.
2015-08-26 15:31:47 -07:00
Scott Mitchell
99d6a97b4a HTTP to HTTP/2 translation errors (round 2)
Motivation:
Commit 0d8ce23c83 failed to fix the Host header processing. Host is not a URI but is instead defined in https://tools.ietf.org/html/rfc3986#section-3.2.2 as host        = IP-literal / IPv4address / reg-name

Modifications:
- Host should not be treated as a URI.
- We should be more explicit about required fields, and unexpected input by throwing exceptions.

Result:
Translation from HTTP/1.x to HTTP/2 is more correct.
2015-08-24 20:37:58 -07:00
Scott Mitchell
85c79dbbe4 HTTP to HTTP/2 tranlation errors
Motivation:
HttpUtil.toHttp2Headers is currently not translating HTTP request headers to HTTP/2 request headers correctly.  The path, scheme, and authority are tranlation process are not respecting the HTTP/2 RFC https://tools.ietf.org/html/rfc7540#section-8.1.2.3 and HTTP RFC https://tools.ietf.org/html/rfc7230#section-5.3.

Modifications:
- path, scheme, authority must be set according to rules defined in https://tools.ietf.org/html/rfc7540#section-8.1.2.3
- HTTP/1.x URIs must be handled as defined in https://tools.ietf.org/html/rfc7230#section-5.3

Result:
More correct translation from HTTP/1.x requests to HTTP/2 requests.
2015-08-21 11:33:10 -07:00
Scott Mitchell
ac9ae14bd9 HTTP/2 SimpleChannelPromiseAggregator failure condition
Motivation:
If a SimpleChannelPromiseAggregator is failed before any new promises are generated, the failure is not propegated through to the aggregated promise.

Modifications:
- Failures should be allowed to occur even if no new promises have been generated

Result:
Failures are always allowed.
2015-08-21 11:26:31 -07:00
Scott Mitchell
08477eaf03 HTTP/2 Graceful Shutdown Timeout
Motivation:
If any streams are still active the graceful shutdown code will wait until they are all closed before the connection is closed. In some situations this event may never occur, and thus a timeout should be supported so the socket can be closed even if all streams haven't been closed.

Modifications:
- Add a configurable timeout for when the graceful shutdown process is attempted.
- Update unit tests to be faster, and use this graceful timeout

Result:
Local endpoint can protect from local or remote issues which prevent the channel from being closed during the graceful shutdown process.
2015-08-20 13:27:18 -07:00
Scott Mitchell
34dfa7a2d8 DefaultHttp2ConnectionEncoder private constructors on inner classes
Motivation:
DefaultHttp2ConnectionEncoder.FlowControlledHeaders and DefaultHttp2ConnectionEncoder.FlowControlledData have private constructors which may result in static factory methods being generated to construct instances of these classes.

Modifications:
- Make constructors public for these private classes

Result:
Accessor for inner class constructor more correct and no possibiliy of synthetic method generation.
2015-08-19 13:30:24 -07:00
Scott Mitchell
ba6ce5449e Headers Performance Boost and Interface Simplification
Motivation:
A degradation in performance has been observed from the 4.0 branch as documented in https://github.com/netty/netty/issues/3962.

Modifications:
- Simplify Headers class hierarchy.
- Restore the DefaultHeaders to be based upon DefaultHttpHeaders from 4.0.
- Make various other modifications that are causing hot spots.

Result:
Performance is now on par with 4.0.
2015-08-17 08:50:11 -07:00