Avoiding intermittent core dump at VISProtocolManager::wait_for_shutdown()

Problem:

• Product Name: VisiBroker for C
• Product Version: All
• Product Component: Core
• Platform/OS Version: HP-UX 11.11

Invoking ORB::run() again during the clean up process where the orb is in the midst of executing ORB::shutdown()/destroy() or exit() function call will cause an intermittent core dump at VISProtocolManager::wait_for_shutdown().

ORB::run() is a blocking call and will return to caller"s stack when the ORB::shutdown()/destroy() or exit() is called to do a clean up. The clean up may take a while. The following is a sample stack trace:

#0 0xc7d24c40 in VISProtocolManager::wait_for_shutdown () from /opt/vbroker7.0runtime/lib/liborb_r.sl 
(gdb) bt 
#0 0xc7d24c40 in VISProtocolManager::wait_for_shutdown () from /opt/vbroker7.0runtime/lib/liborb_r.sl 
#1 0xc7c8c0ec in VISORB::run () from /opt/vbroker7.0runtime/lib/liborb_r.sl
 :

The sample code is listed below,

while (true)
 {
   try
   {
      orb->run();
   }
   catch (const CORBA::Exception&)
   {
    // Some exception handling
    break;
   }
 }

Resolution:

ORB::run() function must not be called again when the orb is executing the ORB::shutdown()/destroy() function.