[
https://issues.apache.org/jira/browse/DIRMINA-1101?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jörg Michelberger updated DIRMINA-1101:
---------------------------------------
Description:
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)}}
was:
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$HeadFilter.filterWrite() is then the mark() call, which
is done on the compressed Buffer.
In AbstractPollingIoProcessor.writeBuffer() is the reset() call, which is in
this case done on the original Buffer, leading to the Exception.
{{java.nio.InvalidMarkException}}
{{ at java.nio.Buffer.reset(Buffer.java:306)}}
{{ at
org.apache.mina.core.buffer.AbstractIoBuffer.reset(AbstractIoBuffer.java:425)}}
{{ at
org.apache.mina.core.polling.AbstractPollingIoProcessor$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)}}
> 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
> 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)
