[jira] Commented: (AMQCPP-325) Failover transport exception can cause CPU spin

["Timothy Bish (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/activemq/browse/AMQCPP-325?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=62974#action_62974 ] 

Timothy Bish commented on AMQCPP-325:
-------------------------------------

patches welcome.

> Failover transport exception can cause CPU spin
> -----------------------------------------------
>
>                 Key: AMQCPP-325
>                 URL: https://issues.apache.org/activemq/browse/AMQCPP-325
>             Project: ActiveMQ C++ Client
>          Issue Type: Bug
>          Components: Transports
>    Affects Versions: 3.2.3
>         Environment: Windows 2003,  Windows 7, Windows XP
>            Reporter: FangYu
>            Assignee: Timothy Bish
>   Original Estimate: 2 weeks
>  Remaining Estimate: 2 weeks
>
> We find that when use the failover transport to connect the broker, if failed to connect, then the thread will not release and become a DEADLESS LOOP cause CPU spin.
> See the source code:
> file: ~\activemq-cpp-library-3.2.3\src\main\activemq\transport\failover\failovertransport.cpp
> 500 bool FailoverTransport::iterate() {
> ......................................................................
> 510        if( connectedTransport != NULL || closed || connectionFailure != NULL ) {
> 511            return false;
> ......................................................................
> 702 }
> file: ~\activemq-cpp-library-3.2.3\src\main\activemq\threads\compositetaskrunner.cpp
> 97 void CompositeTaskRunner::run() {
> 98 
> 99     try {
> 100 
> 101         while( true ) {
> 102 
> 103            synchronized( &mutex ) {
> 104                 pending = false;
> 105                 if( shutDown ) {
> 106                     return;
> 107                  }
> 108              }
> 109 
> 110             if( !this->iterate() ) {
> 111 
> 112                 // wait to be notified.
> 113                 synchronized( &mutex ) {
> 114                     if( shutDown ) {
> 115                         return;
> 116                     }
> 117                     while( !pending ) {
> 118                         mutex.wait();
> 119                     }
> 120                 }
> 121             }
> 122         }
> 123     }
> 124     AMQ_CATCH_NOTHROW( Exception )
> 125     AMQ_CATCHALL_NOTHROW()
> 126 
> 127     // Make sure we notify any waiting threads that thread
> 128     // has terminated.
> 129     synchronized( &mutex ) {
> 130         threadTerminated = true;
> 131         mutex.notifyAll();
> 132     }
> 133 }
> ............................................................................
> 160 bool CompositeTaskRunner::iterate() {
> 161
> 162     synchronized( &tasks ) {
> 163 
> 164         auto_ptr< Iterator<CompositeTask*> > iter( tasks.iterator() );
> 165 
> 166         while( iter->hasNext() ) {
> 167 
> 168             CompositeTask* task = iter->next();
> 169 
> 170            if( task->isPending() ) {
> 171                task->iterate() ;
> 173
> 174                // Always return true, so that we check again for
> 175                // any of the other tasks that might now be pending.
> 176                return true;
> 177            }
> 178        }
> 179     }
> 180 
> 181     return false;
> 182 }

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.