Name: rlT66838 Date: 07/30/97

Software: Solaris 2.5.1
Hardware: Ultra Sparc 1

The following source core makes the Java VM grow in
size indefinitely indicating a memory leak in the VM
or the Solaris motif libraries.

Here's the java source code:

import java.awt.*;

public class MemoryLeak7
{
public static void main( String args[] )
{
for( ;; )
{
for( int i = 0; i < 16; i++ )
{
Frame frame = new Frame();
frame.setSize( 300, 300 );
frame.setTitle( "Leaky Frame" );
frame.setVisible( true );
frame.setVisible( false );
frame.dispose();
frame = null;
}
System.gc();
System.runFinalization();

try
{
System.out.println( "---------- Sleeping ----------" );
Thread.currentThread().sleep( 5000 );
System.out.println( "------ Finished Sleeping ------" );
}
catch( Exception e )
{
System.err.println( "Thread interrupted" );
}
} // end: infinite for loop
} // end: main
}
company - Sterling Diagnostic Imaging, Inc. , email - ###@###.###
======================================================================


###################### so#3383176 - Thu14May98-15:45 BELLEY ###############3
# java -fullversion
java full version "Solaris_JDK_1.1.5_02"

Bug number 4068618 still occurs on a cleanly installed 128MB Ultra 5
running 2.5.1. Per the install instructions, I installed patches 103566-25
and 103640-17.

I was monitoring memory usage by simply doing ls -la /proc/xxx on the pid
of the job. Left running, this program will CRASH the machine (first, it
reports a SYSSEGV error, and the machine is left in a state that requires a
reboot to clear).

The source code I was using is as follows: (slightly modified from the
reported bug:)

import java.awt.*;
import java.util.*;


public class MemoryLeak7
{
  public static void main( String args[] )
    {
      Frame frame;
      Runtime rt = Runtime.getRuntime();
      for( ;; )
{
for( int i = 0; i < 16; i++ )
{
frame = new Frame();
frame.setSize( 300, 300 );
frame.setTitle( "Leaky Frame" );
frame.setVisible( true );
frame.setVisible( false );
frame.dispose();
       
}
rt.gc();
System.gc();
System.runFinalization();
rt.gc();
try
{
System.out.println( "---------- Sleeping ----------" );
Thread.currentThread().sleep( 1000 );
System.out.println( "------ Finished Sleeping ------" );
System.out.println("Total Memory: "+rt.freeMemory());
}
catch( Exception e )
{
System.err.println( "Thread interrupted" );
}
} // end: infinite for loop
    } // end: main
}


For what it is worth, we tried this on a system running 2.6, and the same
problem occurs.
######################