* Bumped the rate control rate from 50 events/s to 128.
* Added rate control for all CONTINUATION frames.
* Added rate control for invalid PUSH_PROMISE frames.
* Added rate control for RST_STREAM frames.
* Added rate control for all SETTINGS frames.
* Fixed growth of header block accumulation buffer.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Changed default implementation of Session.Listener.onNewStream() and Stream.Listener.onDataAvailable() to auto-discard DATA frames.
For normal cases, these methods are overridden and the application is in full control.
For test cases, these methods may not be overridden and the default implementation conveniently avoids buffer leaks.
Fixed flaky test RawHTTP2ProxyTest.testRawHTTP2Proxy() due to the bad assumption that the first DATA frame ends the stream (so an aggregator is needed), and a copy/paste error in ServerToProxyToClient while processing DATA frames.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Fixed test case that was racy.
When the DATA frames arrived at the server before the call to consumeAvailable(), they were read and the client flow control window re-opened.
If it happened that the DATA frames arrived at the server after the call to consumeAvailable(), the client flow control window was not re-opened, making the test flaky.
Fixed by avoiding the race in the test.
Added over-release buffer tracking, add leak tracking to H2 tests, fix client leaks in tests.
Also reviewed the places that required re-opening of the flow control window in case the DATA frames are not read.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Signed-off-by: Ludovic Orban <lorban@bitronix.be>
Co-authored-by: Ludovic Orban <lorban@bitronix.be>
* Changes the H2 semantic of `Stream.readData()` so that it is `readData()` that enlarges the flow control window, and not anymore the release of the `Stream.Data`.
This allows applications to buffer in memory by retaining the `Stream.Data` instances more than the H2 flow control window.
* Updated `FlowControlStrategyTest` after `Stream.readData()` semantic changes.
* Updated `DataDemandTest` after `Stream.readData()` semantic changes.
* Updated documentation.
Signed-off-by: Lachlan Roberts <lachlan@webtide.com>
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Signed-off-by: Ludovic Orban <lorban@bitronix.be>
Co-authored-by: gregw <gregw@webtide.com>
Co-authored-by: Simone Bordet <simone.bordet@gmail.com>
Co-authored-by: Ludovic Orban <lorban@bitronix.be>
* Updating various old/moved URL references found across project (`jetty-10.0.x`) (#10098)
+ Now that the migration of `https://eclipse.org/jetty/` to `https://eclipse.dev/jetty/` has occurred, it is time to review the URI use in our project
+ Updated URLs in poms
+ Added more URIs to XmlConfiguration
+ Updated URLs in module files
+ Updated URLs in documentation
+ Updated URLs in HTML
+ Correcting bad double-scheme URLs (eg: `http://https://www.eclipse...`)
+ Updating text in *.mod files
+ Removing `/current/` from path `/jetty/documentation/current/`
+ Fixing mailing list URL
+ Fixing github URL references in jsps
---------
Signed-off-by: Joakim Erdfelt <joakim.erdfelt@gmail.com>
In case of `request.onComplete(Response.CompleteListener l)`, only the complete event should be delivered to `l`, not all the events (in case `l` implements other listener interfaces).
Only `Response.CompleteListener` passed to `request.send(Response.CompleteListener)` should receive all events.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Fixes#10087 - Flaky EventsHandlerTest due to trailers.
HTTP/2 trailers may arrive and be processed before the application reads request chunks.
Avoid the race condition by storing the trailers aside and returning them during reads.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Remove usage of instanceof Content.Chunk.Error
* Updated AsyncContent to accept a transient failures
* Updated AsyncContent to accept a transient failure with inputstream
* Make a test case to resemble a failing tck test
* Add more test cases more like the tck test
* Fixed processing of EOF frames for pushed requests.
Before, the EOF frame was not added to the HTTP2Stream queue.
When the pushed request was completing it tried to consume the available content, but it was not finding EOF and so was emitting a reset frame, causing the TCK failure.
Now the EOF frame is always added to the HTTP2Stream queue, so it is properly consumed when completing the pushed request.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Fix H2 trailers frames
Signed-off-by: Ludovic Orban <lorban@bitronix.be>
---------
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Signed-off-by: Ludovic Orban <lorban@bitronix.be>
Co-authored-by: Simone Bordet <simone.bordet@gmail.com>
Co-authored-by: Ludovic Orban <lorban@bitronix.be>
* IdleTimeout review
+ pass TimeoutException through all APIs
+ HttpConnection now passes on TimeoutException to HttpChannel.onFailure
* More ServerTests for idletimeout
* Recreated a ServerTimeoutsTest for multiple transports
* more robust tests
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* merged work from @sbordet and @gregw
* Various improvements to CyclicTimeouts.
* Improved reset of the earliest timeout before iteration.
* Removed check for getExpireNanoTime() == -1, since it's a valid value.
* When onExpired(Expirable) returns false, the Expirable should arrange to move its timeout in the future.
* fix keystore to please BoringSSL + use correct temp path
Signed-off-by: Ludovic Orban <lorban@bitronix.be>
* Fixed ErrorResponseAndCallback succeeded() and failed() to call super.failed() in all cases to complete the wrapped callback.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Revert "Fixed ErrorResponseAndCallback succeeded() and failed() to call super.failed() in all cases to complete the wrapped callback."
This reverts commit 5ac57c13e0.
* WIP idleTimeout
* WIP idleTimeout
* Added context wrapper for idle timeout listener
* updates from review
---------
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Signed-off-by: Ludovic Orban <lorban@bitronix.be>
Co-authored-by: Simone Bordet <simone.bordet@gmail.com>
Co-authored-by: Ludovic Orban <lorban@bitronix.be>
* Improved reset of the earliest timeout before iteration.
* Removed check for getExpireNanoTime() == -1, since it's a valid value.
* When onExpired(Expirable) returns false, the Expirable should arrange to move its timeout in the future.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Restored dispatch=true for pushed requests.
* Restored tests that use the Servlet APIs to push.
* Ensured that pushed streams have request content EOF.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Fixes#9684
+ Restore LargeHeaderTest Test
+ Fix Bad Content-Length produced if write + error occurs
+ Fix race between callback failure and error handling failure
+ Introduce new ResponseCompleteTest to attempt to capture complete race issue
+ More DEBUG + ignoring failure for completeStream
+ Removed ErrorResponse in favour or an errorMode
---------
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Co-authored-by: gregw <gregw@webtide.com>
Co-authored-by: Simone Bordet <simone.bordet@gmail.com>
* Optimize Content-Length: 0 handling
Create and use a pre-encoded HttpFields.CONTENT_LENGTH_0 constant
Used the constant in more places
renamed the `putLongField` methods of HttpFields.Mutable to just `put`
Fixed wrong docs examples
* Removed unnecessary constructors from MetaData, MetaData.Request and MetaData.Response.
* Removed MetaData.Request.getURIString() (available as getHttpURI().toString()).
* Renamed MetaData.getFields() -> getHttpFields(), as they can be headers or trailers.
* Renamed MetaData.Request.getURI() -> getHttpURI().
* Normalized handling of contentLength, now always -1 (rather than Long.MIN_VALUE) if unknown.
* Permutated MetaData.Response constructor parameters to be consistent with MetaData.Request.
* MetaData.Request's method and httpURI must be non-null.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Deleted the Ut8fStringBuffer and Utf8Appendable classes in favour of just having Utf8StringBuilder
* Simplified the Utf8StringBuilder and CharsetStringBuilder APIs and improved their exception throwing.
* Allow for replacement characters without throwing
---------
Signed-off-by: Lachlan Roberts <lachlan@webtide.com>
Signed-off-by: gregw <gregw@webtide.com>
Co-authored-by: Lachlan Roberts <lachlan@webtide.com>
Co-authored-by: Joakim Erdfelt <joakim.erdfelt@gmail.com>
The problem was that the retain/release idiom in HTTP2Session.onData()
was not respected.
Now after the call to HTTP2Stream.process(), the Data object is released.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Convert class BadMessageException to a HttpMessage.RuntimeException to allows different types of HttpExceptions exceptions.
This follows the pattern of the QuietException interface.
* Introduced ResponseListeners to reduce per-request allocation.
* Removed ResponseNotifier in favor of ResponseListeners.
* Introduced RequestListeners to reduce the per-request allocation.
* Removed RequestNotifier in favor of RequestListeners.
* Removed unnecessary sharing of HttpRequest.requestListeners.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
With issue #9166, ByteBufferPool was removed and replaced by
RetainableByteBufferPool. Since ByteBufferPool was used by
AbstractConnector, this change broke backwards compatibility with
third-party connectors such as junixsocket-jetty.
Since there's no longer any other ByteBufferPool, rename the
RetainableByteBufferPool interface, and thereby not only reinstate
compatibility with existing third-party libraries but also save a few
keystrokes.
https://github.com/eclipse/jetty.project/issues/9284
Signed-off-by: Christian Kohlschütter <christian@kohlschutter.com>
* silence stack traces in IdleTimeoutTest by refactoring test + avoid execution of handlers over closed streams
Signed-off-by: Ludovic Orban <lorban@bitronix.be>
* Fixes#9210 - Jetty 12 - Review Pool and Pool.Entry
* Extracted interface Pool, renamed implementation to ConcurrentPool.
* Extracted Pool.Entry as interface.
* Moved StrategyType to ConcurrentPool.
* Made Pool.Factory.wrap() work in order to wrap Pool instances.
* Removed constructors that were explicitly taking Pool parameters, replaced by a single Pool.Factory parameter.
* Added javadocs.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
`Retainable`s that return false from `canRetain()` now are noops if `retain()` is called, which allows for a simpler calling convention.
`AsyncContent` has also been reworked to allocate less and be clearer in its use of `canRetain()`.
* Fixes#8993 - Retainability of special Chunks
* Restored Jetty 11's AsyncContentProducer and related classes in jetty-ee9-nested module (src and test).
* Introduced Retainable.canRetain().
* Removed Chunk.isTerminal() and replaced it with alternative method calls.
* Clarified AsyncContent.write() in case of Chunk.Error.
* Removed AsyncContent.write(Chunk, Callback) because it was making the API confusing.
For example, AsyncContent.close() clashing with write(EOF, NOOP), or
AsyncContent.fail(x) clashing with write(Chunk.Error, NOOP), etc.
* Improved usage of Chunk.from(..., Retainable).
* Improved usage of Chunk.slice().
* Using from() in MultiPart, rather than duplicating code.
* Fixed MultiPart.Parser.Listener.onPartContent() javadocs.
* Renamed non-retaining Chunk.from() to Chunk.asChunk().
* Removed Chunk.slice() methods, inlining them where necessary.
* Carefully reviewed all usages of read()-like methods that return a Retainable instance to make sure it is released.
* Updated HTTP/2 and HTTP/3 usages of Stream.Data to follow the correct retain/release semantic.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Reorganization of jetty-client classes.
* Moved oej.client.api to oej.client
* Moved oej.client.util to oej.client
* Moved implementation classes to oej.client.internal
* Moved transports to oej.client.transport
* Moved transport implementation classes to oej.client.transport.internal
* Moved TunnelRequest to oej.client.internal.
* Moved FastCGI transport classes to o.e.j.fcgi.transport
* Moved FastCGI transport implementation classes to o.e.j.fcgi.transport.internal
* Updated WebSocket core client to use only exported, non-internal, oej.client classes.
* Expanded oej.client.Destination APIs:
- added: getOrigin(), isSecure(), getProxy(), getConnectionPool(), getHttpClient(), send(..).
- removed: getScheme(), getHost(), getPort() because they don't uniquely identify a Destination anymore (Origin does)
* Moved destination sweeper functionality from HttpDestination to HttpClient.
HttpDestination does not implement close() anymore, now relies on LifeCycle.stop()
* Moved HttpReceiver.storeCookie() logic to HttpClient.putCookie() to avoid exposing CookieManager.
* Moved HttpClient.getAcceptEncodingField() to ContentDecoder.Factories
* Avoid public/protected Logger instances.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Alternative Handler architecture.
All Handlers are Processors, which now return a boolean to indicate the request has been accepted.
The request/response/callback are no longer modal, so there is no race with the boolean return.
Optimized PathMappings.
Avoid iterations if only ServletPathSpec instances
Avoid tests for empty mappings.
Better reset implementation
Improve suffix matching
Improve exact matching
Renamed HttpStream.getNanoTimeStamp() to getNanoTime().
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Co-authored-by: Simone Bordet <simone.bordet@gmail.com>
* Improved locking for HttpReceiver.ContentSource.
Improved response failure code path.
Now either responseFailure() must be called, or exchange.responseComplete() followed by HttpReceiver.abort().
Fixed failAndClose() for HTTP/2 and HTTP/3: the connection must not be closed, stream.reset() is sufficient.
Fixed flaky test HttpClientDemandTest.testTwoListenersWithDifferentDemand().
Fixed DistributionTests.testVirtualThreadPool().
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
In Jetty 12, demand after last frame invokes onDataAvailable(), so the latch needs to be counted down two times.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Joakim Erdfelt
Simone Bordet
Lachlan
Ludovic Orban
Greg Wilkins
Jan Bartel
Olivier Lamy
Olivier Lamy
dependabot[bot]
Christian Kohlschütter