You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mina.apache.org by "Emmanuel Lecharny (JIRA)" <ji...@apache.org> on 2019/04/03 14:24:00 UTC
[jira] [Comment Edited] (DIRMINA-1101) InvalidMarkException on
session.write when using CompressionFilter.
[ https://issues.apache.org/jira/browse/DIRMINA-1101?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16808774#comment-16808774 ]
Emmanuel Lecharny edited comment on DIRMINA-1101 at 4/3/19 2:23 PM:
--------------------------------------------------------------------
As Jonathan said, the risk is that it breaks something in existing use code base, something I really can't guarantee with a modified 2.0.X.
OTOH, 2.1 is really close to 2.0, and there should be no impact on your code if you switch to this version. All in all, we just added an event, which is handled by default in the {{IoHandlerAdapter}}, so your {{IoHandler}} implementation should not be hit unless you were using the SSL notification (check that: [http://mina.apache.org/mina-project/2.1-vs-2.0.html).]
Is that the case ?
was (Author: elecharny):
As Jonathan said, the risk is that it breaks something in existing use code base, something I really can't guarantee with a modified 2.0.X.
OTOH, 2.1 is really close to 2.0, and there should be no impact on your code if you switch to this version. All in all, we just added an event, which is handled by default in the \{IoHandlerAdapter}, so your \{IoHandler} implementation should not be hit unless you were using the SSL notification (check that: [http://mina.apache.org/mina-project/2.1-vs-2.0.html).]
Is that the case ?
> InvalidMarkException on session.write when using CompressionFilter.
> -------------------------------------------------------------------
>
> Key: DIRMINA-1101
> URL: https://issues.apache.org/jira/browse/DIRMINA-1101
> Project: MINA
> Issue Type: Bug
> Affects Versions: 2.0.20
> Reporter: Jörg Michelberger
> Assignee: Emmanuel Lecharny
> Priority: Major
>
> I'm updated from a MINA 2.0.7 to the 2.0.20 and am a user of CompressionFilter. Writing of Messages fails with a InvalidMarkException.
> Reproducible Test is:
> * Copy MINA Core Test org.apache.mina.core.service.AbstractIoServiceTest to MINA Compression Filter org.apache.mina.filter.compression Test Packages package.
> * Adapt package statement to org.apache.mina.filter.compression.
> * Add Compression to acceptor and connector
> * acceptor.getFilterChain().addLast("compression", new CompressionFilter());
> acceptor.getFilterChain().addLast("logger", new LoggingFilter());
> acceptor.getFilterChain().addLast("codec",
> new ProtocolCodecFilter(new TextLineCodecFactory(StandardCharsets.UTF_8)));
> * connector.getFilterChain().addLast("compression", new CompressionFilter());
> connector.getFilterChain().addLast("logger", new LoggingFilter());
> connector.getFilterChain().addLast("codec",
> new ProtocolCodecFilter(new TextLineCodecFactory(StandardCharsets.UTF_8)));
> * Set a Breakpoint to java.nio.Buffer.reset() method, where the InvalidMarkException is thrown.
> * Run Debug Testfile on org.apache.mina.filter.compression.AbstractIoServiceTest
> After the Exception the session is immediatelly scheduled for disconnect.
> It seems that there is a discrepancy between the mark() and reset() calls on the underlaying Buffer. In case of compression, a Buffer with the compressed content is created and is wrapped with the original Buffer in a FilteredWriteRequest, because CompressionFilter is a WriteRequestFilter. This is in WriteRequestFilter.filterWrite()
> In DefaultIoFilterChain$HeadFilter.filterWrite() is then the mark() call, which is done on the compressed Buffer.
> In AbstractPollingIoProcessor.writeBuffer() is the reset() call, which is in this case done on the original Buffer, leading to the Exception.
> It seems that the change at date 16.02.2016
> SHA-1: 44b58469f84ce991074cdc187b1c1f23b94cf445
> * Don't try to reset a message when it's not a IoBuffer
> which reassignes the buf before reset() is called broke it. The buf before reassign looks much better as the right to reset() in this case.
>
> {{java.nio.InvalidMarkException}}
> {{ at java.nio.Buffer.reset(Buffer.java:306)}}
> {{ at org.apache.mina.core.buffer.AbstractIoBuffer.reset(AbstractIoBuffer.java:425)}}
> {{ at org.apache.mina.core.polling.AbstractPollingIoProcessor$Processor.writeBuffer(AbstractPollingIoProcessor.java:1131)}}
> {{ at org.apache.mina.core.polling.AbstractPollingIoProcessor$Processor.flushNow(AbstractPollingIoProcessor.java:994)}}
> {{ at org.apache.mina.core.polling.AbstractPollingIoProcessor$Processor.flush(AbstractPollingIoProcessor.java:921)}}
> {{ at org.apache.mina.core.polling.AbstractPollingIoProcessor$Processor.run(AbstractPollingIoProcessor.java:688)}}
> {{ at org.apache.mina.util.NamePreservingRunnable.run(NamePreservingRunnable.java:64)}}
> {{ at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)}}
> {{ at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)}}
> {{ at java.lang.Thread.run(Thread.java:748)}}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)