Motivation:
The current PROXY protocol implementation does not have support for optional Type-Length-Value fields. This pull requests adds the TLV values as specified in the PROXY protocol specification (http://www.haproxy.org/download/1.5/doc/proxy-protocol.txt) and adds support for arbitrary TLVs.
Modifications:
The existing HAProxyMessage implements an additional TLV reading operation. A small bug in the AF_UNIX reader which didn’t set the reader index correctly was also fixed.
Result:
The PROXY protocol supports TLVs
Motivation:
A `SeekAheadNoBackArrayException` used as check for `ByteBuf#hasArray`. The catch of exceptions carries a large overhead on stack trace filling, and this should be avoided.
Modifications:
- Remove the class `SeekAheadNoBackArrayException` and replace its usage with `if` statements.
- Use methods from `ObjectUtils` for better readability.
- Make private methods static where it make sense.
- Remove unused private methods.
Result:
Less of exception handling logic, better performance.
Motivation:
JdkZlibDecoder will allocate a new buffer when the previous buffer is filled with inflated data, but JZlibDecoder will attempt to use the same buffer by resizing. This leads to inconsistent results when these two decoders that are intended to be functionality equivalent.
Modifications:
- JdkZlibDecoder should attempt to resize and reuse the existing buffer instead of creating multiple buffers
Result:
Fixes https://github.com/netty/netty/issues/6804
Motivation:
If a full HttpResponse with a Content-Length header is encoded by the HttpContentEncoder subtypes the Content-Length header is removed and the message is set to Transfer-Encoder: chunked. This is an unnecessary loss of information about the message content.
Modifications:
- If a full HttpResponse has a Content-Length header, the header is adjusted after encoding.
Result:
Complete messages continue to have the Content-Length header after encoding.
Motivation:
If the destination buffer is completely filled during a call to OpenSslEngine#wrap(..) we may return NEED_UNWRAP because there is no data pending in the SSL buffers. However during a handshake if the SSL buffers were just drained, and filled up the destination buffer it is possible OpenSSL may produce more data on the next call to SSL_write. This means we should keep trying to call SSL_write as long as the destination buffer is filled and only return NEED_UNWRAP when the destination buffer is not full and there is no data pending in OpenSSL's buffers.
Modifications:
- If the handshake produces data in OpenSslEngine#wrap(..) we should return NEED_WRAP if the destination buffer is completely filled
Result:
OpenSslEngine returns the correct handshake status from wrap().
Fixes https://github.com/netty/netty/issues/6796.
Motivation:
QueryStringDecoder has several problems:
- doesn't decode correctly path part with `+` (plus) sign in it,
- doesn't cut a `fragment` (after `#`) from query string (see RFC 3986),
- doesn't work correctly with encoding,
- treat `%%` as a percent character escaping (it's don't described in RFC).
Modifications:
- leave `+` chars in a `path` part of uri string,
- ignore `fragment` part (after `#`),
- correctly work with encoding.
- don't treat `%%` as escaping for the `%`.
Result:
Fixed issues from #6745.
Motivation:
ChannelPipeline will happily add a handler to a closed Channel's pipeline and will call handlerAdded(...) but will not call handlerRemoved(...).
Modifications:
Check if pipeline was destroyed and if so not add the handler at all but propergate an exception.
Result:
Fixes [#6768]
Motivation:
Allow subclasses of HttpObjectEncoder other than HttpServerCodec to override the isContentAlwaysEmpty method
Modification:
Change the method visibility from package private to protected
Result:
Fixes#6761
Motivation:
We need to exclude the netty-tcnative classes from the all jar when building it as otherwise there may be classloader issues.
Modifications:
Add io/netty/internal/tcnative to the exclude list.
Result:
Fixes [#6791]
Motivation:
ByteBufUtil provides a hexDump method. For debugging purposes it is often useful to decode that hex dump to get the original content, but no such method exists.
Modifications:
- Add ByteBufUtil#decodeHexDump
Result:
ByteBufUtil#decodeHexDump is available to make debugging easier.
Motivation:
RedisDecoder can get into an infinite loop while decoding bulk strings if the final \r and \n to indicate the end of content are split on ByteBuf boundaries.
Modifications:
- We should break out of the decode loop if remainingBulkLength is 0 and we don't have enough data to read EOL
Result:
No more infinite loop in RedisDecoder#decodeBulkStringContent.
Motivation:
We used strstr to find the path to the library, which fails if the library is contained in a directory that also matches the library name.
Modifications:
- Introduce netty_unix_util_strstr_last which will return a pointer which points to the last accourance and so not fails if the direct also matches the library name.
Result:
Be able to load the library in all cases.
Motivation:
Fix the regression recently introduced that causes already encoded responses to be encoded again as gzip
Modification:
instead of just looking for IDENTITY, anything set for Content-Encoding should be respected and left as-is
added unit tests to capture this use case
Result:
Fixes#6784
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
RFC 1945 (see section 3.1) says that request lines may not have a version in which case the request is assumed to be HTTP/0.9. We don't necessarily want to support that but the existing Exception should indicate the possibility of the request being HTTP/0.9 and give the user a chance to track it down.
Modifications
Indicate in the Exception's message that the request is possibly HTTP/0.9.
Result
Fixes#6739
Motivation:
The first parameter of SctpMessage is protocolIdentifier, and the second is streamIdentifier. So we need to swap the parameters in encode method
Modification:
Fix order
Result:
SctpOutboundByteStreamHandler works correctly.
Motivation:
The status 308 is defined by RFC7538.
This RFC has currently the state Proposed Standard since 2 years, but the status code is already handle by all browsers (Chrome, Firefox, Edge, Safari, …).
To let developer handles easily this status code, it is added into this list.
Modifications:
Added this status code in the list of all status codes and changed the valudOf() method
Result:
Status code 308 included
Motivation:
We only can call eventLoop() if we are registered on an EventLoop yet. As we just did this without checking we spammed the log with an error that was harmless.
Modifications:
Check if registered on eventLoop before try to deregister on close.
Result:
Fixes [#6770]
Motivation:
If the remote server returns an invalid response in the form "000 \r\n"
(i.e. a three digit code, then space, but no details), null is added
as a singletonList to the response being constructed.
This seems unexpected and it would be easier to handle an empty
details list in client code.
Modifications:
If detail is null (because frame.isReadable() returned false after
reading the separator), initialise DefaultSmtpResponse with an empty
list instead of a list containing a single null value.
Result:
When encountering this malformed server response, a DefaultSmtpResponse
with a code but no details will be created.
Motivation
RFC 6066 (https://tools.ietf.org/html/rfc6066#page-6) says that the hostname in the SNI extension is ASCII encoded but Netty decodes it using UTF-8.
Modifications
Use ASCII instead of UTF-8
Result
Fixes#6717
Motivation:
If the remote server returns a 4xx/5xx error in response to
a DATA command (or earlier command if using pipelining),
SmtpRequestEncoder can become stuck in an invalid state,
not allowing any requests to be sent.
This makes the channel unusable and the connection has to be closed,
or the encoder handler has to be replaced.
Modifications:
If a RSET command is written to the channel, the contentExpected
flag is set to false, and the RSET is written to the channel.
Result:
Sending a RSET command after a server 4xx/5xx error will make it
possible to use the current connection for new mail transactions.
Motivation:
The native epoll transport allows to wrap an existing filedescriptor, we should support the same in the native kqueue transport.
Modifications:
Add constructors that allow to wrap and existing filedescriptor.
Result:
Featureset of native transports more on par.
Motivation:
For our native libraries in netty we support shading, to have this work on runtime the user needs to set a system property. This code should shared.
Modifications:
Move logic to NativeLbiraryLoader and so share for all native libs.
Result:
Less code duplication and also will work for netty-tcnative out of the box once it support shading
Motivation
SslHandler should release any type of SSLEngine if it implements the ReferenceCounted interface
Modifications
Change condition to check for ReferenceCounted interface
Result
Better use of interfaces
Motivation:
`FormattingTuple.getArgArray()` is never used.
In the `MessageFormatter` it is possible to make
some improvements, e.g. replace `StringBuffer`
with `StringBuilder`, avoid redundant allocations, etc.
Modifications:
- Remove `argArray` field from the `FormattingTuple`.
- In `MessageFormatter`:
- replace `StringBuffer` with `StringBuilder`,
- replace `HashMap` with `HashSet` and make it lazy initialized.
- avoid redundant allocations (`substring()`, etc.)
- use appropriate StringBuilder's methods for the some `Number` values.
- Porting unit tests from `slf4j`.
Result:
Less GC load on logging with internal `MessageFormatter`.
Motivation:
IPv4/6 validation methods use allocations, which can be avoided.
IPv4 parse method use StringTokenizer.
Modifications:
Rewriting IPv4/6 validation methods to avoid allocations.
Rewriting IPv4 parse method without use StringTokenizer.
Result:
IPv4/6 validation and IPv4 parsing faster up to 2-10x.
Motivation:
UnixResolverDnsServerAddressStreamProvider allows the default name server address stream to be null, but there should always be a default stream to fall back to ([1] Search Strategy).
UnixResolverDnsServerAddressStreamProvider currently shuffles the names servers are multiple are present, but the defined behavior is to try them sequentially [2].
[1] Search Strategy Section - https://developer.apple.com/legacy/library/documentation/Darwin/Reference/ManPages/man5/resolver.5.html
[2] DESCRIPTION/nameserver Section - https://developer.apple.com/legacy/library/documentation/Darwin/Reference/ManPages/man5/resolver.5.html
Modifications:
- UnixResolverDnsServerAddressStreamProvider should always use the first file provided to derive the default domain server address stream. Currently if there are multiple domain names in the file identified by the first argument of the constructor then one will be selected at random.
- UnixResolverDnsServerAddressStreamProvider should return name servers sequentially.
- Reduce access level on some methods which don't have known use-cases externally.
Result:
Fixes https://github.com/netty/netty/issues/6736
Motivation:
MacOS will throw an error when attempting to set the IP_TOS socket option if IPv6 is available, and also when getting the value for IP_TOS.
Modifications:
- Socket#setTrafficClass and Socket#getTrafficClass should try to use IPv6 first, and check if the error code indicates the protocol is not supported before trying IPv4
Result:
Fixes https://github.com/netty/netty/issues/6741.
Motivation:
We not correctly handle LE buffers when try to read the packet length out of the buffer and just assume it always is a BE buffer.
Modifications:
Correctly account for the endianess of the buffer when reading the packet lenght.
Result:
Fixes [#6709].
Motivation:
We need to include the native modules on the bom to ensure there is no version missmatch.
Modifications:
Add native modules.
Result:
Fixes [#6738]
Motivation:
SslHandler#wrapNonAppData may be able to return early if it is called from a unwrap method and the status is NEED_UNWRAP. This has been observed to occur while using the OpenSslEngine and can avoid allocation of an extra ByteBuf of size 2048.
Modifications:
- Return early from SslHandler#wrapNonAppData if NEED_UNWRAP and we are called from an unwrap method
Result:
Less buffer allocations and early return from SslHandler#wrapNonAppData.
Motivation:
We should only try to load jdk.internal.misc.Unsafe if we run on Java9+ to eliminate noise in the log.
Modifications:
- Move javaVersion() and related methods to PlatformDependent0 to be able to use these in the static initializer without creating a cycle.
- Only try to load jdk.internal.misc.Unsafe when running in Java9+
Result:
Less noise in the log when running pre java9.
Motivation:
As we now include native code for multiple platforms we need to generate an uber all jar before release it from the staging repository. For this the uber-staging profile can be used. To create a snapshot uber jar the uber-snapshot profile can be used.
Modifications:
- Add uber-staging and uber-snapshot profile
- Correct comment in pom.xml file to show usage.
Result:
Easier to create snapshot and release uber jars.
Motivation:
A previous change allocated a new thread local string builder if it
was getting too large. This is a good change, these string builders
can accidentally get too large and then never shrunk and that is sort
of a memory leak. However, the change allocates an entirely new string
builder which is more allocations than necessary. Instead, we can trim
the string builder if its too large, this only allocates an extra
backing array instead of a whole new object.
Modifications:
If the string builder is above a threshold, we trim the string builder
and then ensure its capacity is reasonable to we do not allocate too
much as we start using the string builder.
Result:
The thread local string builder do not serve as a memory yet we do not
allocate too many new objects.
Motivation:
`NetUtil`'s methods `isValidIpV6Address` and `getIPv6ByName` incorrectly validate some IPv6 addresses.
Modifications:
- `getIPv6ByName`: add checks for single colon at the start or end.
- `isValidIpV6Address`: fix checks for the count of colons and use `endOffset` instead of `ipAddress.length()` for the cases with the brackets or '%'.
Result:
More correct implementation of `NetUtil#isValidIpV6Address` and `NetUtil#getIPv6ByName`.
Motivation:
To ensure the release plugin works correctly we need to ensure all modules are included during build.
Modification:
- Include all modules
- Skip compilation and tests for native code when not supported but still include the module and build the jar
Result:
Build and release works again
__Motivation__
`HttpClientCodec` skips HTTP decoding on the connection after a successful HTTP CONNECT response is received.
This behavior follows the spec for a client but pragmatically, if one creates a client to use a proxy transparently, the codec becomes useless after HTTP CONNECT.
Ideally, one should be able to configure whether HTTP CONNECT should result in pass-through or not. This will enable client writers to continue using HTTP decoding even after HTTP CONNECT.
__Modification__
Added overloaded constructors to accept `parseHttpPostConnect`. If this parameter is `true` then the codec continues decoding even after a successful HTTP CONNECT.
Also fixed a bug in the codec that was incrementing request count post HTTP CONNECT but not decrementing it on response. Now, the request count is only incremented if the codec is not `done`.
__Result__
Easier usage by HTTP client writers who wants to connect to a proxy but still decode HTTP for their users for subsequent requests.
Motivation:
ByteToMessageDecoder#handlerRemoved will immediately release the cumulation buffer, but it is possible that a child class may still be using this buffer, and therefore use a dereferenced buffer.
Modifications:
- ByteToMessageDecoder#handlerRemoved and ByteToMessageDecoder#decode should coordinate to avoid the case where a child class is using the cumulation buffer but ByteToMessageDecoder releases that buffer.
Result:
Child classes of ByteToMessageDecoder are less likely to reference a released buffer.
Motivation:
We only used the openssl version to detect if Ocsp is supported or not which is not good enough as even the version is correct it may be compiled without support for OCSP (like for example on ubuntu).
Modifications:
Try to enable OCSP while static init OpenSsl and based on if this works return true or false when calling OpenSsl.isOcspSupported().
Result:
Correctly detect if OSCP is supported.
Motivation:
Some JUnit assert calls can be replaced by simpler.
Modifications:
Replacement with a more suitable methods.
Result:
More informative JUnit reports.
Motivation:
In OpenSsl init code we create a SelfSignedCertificate which we not explicitly delete. This can lead to have the deletion delayed.
Modifications:
Delete the SelfSignedCertificate once done with it.
Result:
Fixes [#6716]
Motivation:
The javadocs for ByteBuf#ensureWritable(int, boolean) indicate that it should not throw, and instead the return code should indicate the result of the operation. Due to a bug in AbstractByteBuf it is possible for a resize to be attempted on a buffer that may exceed maxCapacity() and therefore throw.
Modifications:
- If there is not enough space in the buffer, and force is false, then a resize should not be attempted
Result:
AbstractByteBuf#ensureWritable(int, boolean) enforces the javadoc constraints and does not throw.
Motivation:
SSL_write requires a fixed amount of bytes for overhead related to the encryption process for each call. OpenSslEngine#wrap(..) will attempt to encrypt multiple input buffers until MAX_PLAINTEXT_LENGTH are consumed, but the size estimation provided by calculateOutNetBufSize may not leave enough room for each call to SSL_write. If SSL_write is not able to completely write results to the destination buffer it will keep state and attempt to write it later. Netty doesn't account for SSL_write keeping state and assumes all writes will complete synchronously (by attempting to allocate enough space to account for the overhead) and feeds the same data to SSL_write again later which results in corrupted data being generated.
Modifications:
- OpenSslEngine#wrap should only produce a single TLS packet according to the SSLEngine API specificaiton [1].
[1] https://docs.oracle.com/javase/8/docs/api/javax/net/ssl/SSLEngine.html#wrap-java.nio.ByteBuffer:A-int-int-java.nio.ByteBuffer-
- OpenSslEngine#wrap should only consider a single buffer when determining if there is enough space to write, because only a single buffer will ever be consumed.
Result:
OpenSslEngine#wrap will no longer produce corrupted data due to incorrect accounting of space required in the destination buffers.
Motivation:
AtomicIntegerFieldUpdater#get is unnecessary, I think use simple volatile read is cleaner
Modifications:
Replace code STATE_UPDATER.get(this) to state in SingleThreadEventExecutor
Result:
Cleaner code
