[Bug libgcj/14751] New: java program compiled with gcj never release system handle resources allocated by new threads on windows platform

[Bug libgcj/14751] New: java program compiled with gcj never release system handle resources allocated by new threads on windows platform

[alessandro dot pepi at email dot it]

A program compiled with gcj on Windows platform (i use the thisiscool 
distribution) never release system handle resources allocated by new threads 
until the main thread exit.
When a compiled java application create a new thread, the system allocate a 
new thread handle and some other system handles. When the new thread exits the 
thread handle is released but 3 other system handles (3 per thread) will be 
released only when application terminate its execution. Below a java test 
program that i have comiled whith the latest windows thisiscool distribution 
of gcj:

import java.util.*;

public class TestThread {
	
public void createThread(final int j){	        
    		Runnable run = new Runnable(){
    			public void run(){
			  System.out.println("I am thread number "+j);    
			  tsleep(2000,null);
    			};
    		};
    		try{
    		  new Thread(run).start();
    		}catch(Exception e){
    		  System.out.println(e.toString());
    		};	
};	

public void Test(){
	for (int i = 0;i<100;i++){
		createThread(i);
		tsleep(250,null);
	};
};
    
public static void tsleep(long t,String s){
    if (s!=null)
	System.out.println(s);
    try{
    	Thread.sleep(t);
    }catch(Exception e){
    	System.out.println(e.toString());
    }
};
    

public static void main(String[] args) {
   tsleep(5000,"Begin test : look at handles resource on Task Manager before 
starting ");
    TestThread test = new TestThread();
    test.Test();
    tsleep(1000,"executing garbage collector ...");
    Runtime.getRuntime().runFinalization();
    Runtime.getRuntime().gc();
    tsleep(10000,"End test : now look again at handles resource on Task 
Manager before exiting ");
}

};

-- 
           Summary: java program compiled with gcj never release system
                    handle resources allocated by new threads on windows
                    platform
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libgcj
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: alessandro dot pepi at email dot it
                CC: alessandro dot pepi at email dot it,gcc-bugs at gcc dot
                    gnu dot org,java-prs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14751

[Bug libgcj/14751] [win32] java program compiled with gcj never release system handle resources allocated by new threads on windows platform

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|java program compiled with  |[win32] java program
                   |gcj never release system    |compiled with gcj never
                   |handle resources allocated  |release system handle
                   |by new threads on windows   |resources allocated by new
                   |platform                    |threads on windows platform


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14751

[Bug libgcj/14751] [win32] java program compiled with gcj never release system handle resources allocated by new threads on windows platform

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-04-29 00:52 -------
Confirmed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2004-04-29 00:52:51
               date|                            |
            Version|unknown                     |3.4.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14751

[Bug libgcj/14751] [win32] java program compiled with gcj never release system handle resources allocated by new threads on windows platform

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-04-29 00:52 -------
*** Bug 15195 has been marked as a duplicate of this bug. ***

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ovidr at users dot
                   |                            |sourceforge dot net


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14751

[Bug libgcj/14751] [win32] java program compiled with gcj never release system handle resources allocated by new threads on windows platform

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-07-03 22:59 -------
*** Bug 16342 has been marked as a duplicate of this bug. ***

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |efly at javaresearch dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14751

[Bug libgcj/14751] [win32] thread creation leaks system handle resources

[ovidr at users dot sourceforge dot net]

------- Additional Comments From ovidr at users dot sourceforge dot net  2004-09-10 16:15 -------
I think the problem is more than just thread creation.  I have ripped out the 
syncExec code from SWT (Display.syncExec()), and only use three threads in a 
loop.

It requires swt-win32-3062.dll (SWT 3 final) in the directory for 
PostThreadMessage calls.

Handles seem to increase ad infinitum, and memory seems to grow slowly as well 
(unlike sun's java) but I don't know if it is ever collected.

gcj -o threadswitch.exe -fjni --main=org.eclipse.swt.internal.win32.OS OS.java

I am using gcj 3.4.1 final (with or without Mohan's patch  
http://gcc.gnu.org/ml/java/2004-09/msg00048.html) as well as:
gcc version 3.5.0 20040730 (experimental)

---
package org.eclipse.swt.internal.win32;

class OS implements Runnable {
  public static final native int GetCurrentProcessId();

  public static final native int GetCurrentThreadId();

  public static final native boolean PostThreadMessageW(int idThread, int Msg, 
int wParam, int lParam);

  public static final native boolean PostThreadMessageA(int idThread, int Msg, 
int wParam, int lParam);

  public static final int WM_NULL = 0x0;

  static {
    System.loadLibrary("swt-win32-3062");
  }

  public static void main(String[] args) {
    final OS t1 = new OS();

    Thread t3 = new Thread() {
      public void run() {
        while (true) {

          try {
            t1.runAsyncMessages();
            Thread.sleep(1);
          } catch (Exception e) {
            e.printStackTrace();
          }
        }
      }
    };
    t3.start();
    
    Thread t2 = new Thread() {
      public void run() {
        while (true) {

          try {
            System.out.println("Thread: " + Thread.currentThread());
            t1.syncExec(t1);
            Thread.sleep(1);
          } catch (Exception e) {
            e.printStackTrace();
          }
        }
      }
    };

    t2.start();

  }

  Thread thread;
  Object messageLock = new Object();
  RunnableLock[] messages;
  int messageCount;
  int threadId;
  int processId;
  Thread syncThread;

  public OS() {
    threadId = GetCurrentThreadId();
    processId = GetCurrentProcessId();
    thread = Thread.currentThread();
    System.out.println("Thread: " + Thread.currentThread());
  }

  public static final boolean PostThreadMessage(int idThread, int Msg, int 
wParam, int lParam) {
    // IsUnicode = !IsWin32s && !IsWin95;
    // if (IsUnicode) return PostThreadMessageW (idThread, Msg, wParam, 
lParam);
    return PostThreadMessageA(idThread, Msg, wParam, lParam);
  }

  public void run() {
    System.out.println("TS: " + Thread.currentThread());
  }

  boolean isValidThread() {
    return thread == Thread.currentThread();
  }

  void wakeThread() {
    PostThreadMessage(threadId, WM_NULL, 0, 0);
  }

  void addLast(RunnableLock lock) {
    synchronized (messageLock) {
      if (messages == null)
        messages = new RunnableLock[4];
      if (messageCount == messages.length) {
        RunnableLock[] newMessages = new RunnableLock[messageCount + 4];
        System.arraycopy(messages, 0, newMessages, 0, messageCount);
        messages = newMessages;
      }
      messages[messageCount++] = lock;
      if (messageCount == 1)
        this.wakeThread();

    }
  }

  protected void syncExec(Runnable runnable) {
    if (this.isValidThread()) {
      if (runnable != null)
        runnable.run();
      return;
    }
    if (runnable == null) {
      // display.wake();
      return;
    }
    RunnableLock lock = new RunnableLock(runnable);
    /*
     * Only remember the syncThread for syncExec.
     */
    lock.thread = Thread.currentThread();
    synchronized (lock) {
      addLast(lock);
      boolean interrupted = false;
      while (!lock.done()) {
        try {
          lock.wait();
        } catch (InterruptedException e) {
          interrupted = true;
        }
      }
      if (interrupted) {
        Thread.currentThread().interrupt();
      }
      if (lock.throwable != null) {
        System.err.println("Throwable != null");
      }
    }
  }

  boolean runAsyncMessages() {
    if (messageCount == 0)
      return false;
    RunnableLock lock = removeFirst();
    if (lock == null)
      return true;
    synchronized (lock) {
      syncThread = lock.thread;
      try {
        lock.run();
      } catch (Throwable t) {
        lock.throwable = t;
        System.err.println("FAILED EXEC");
      } finally {
        syncThread = null;
        lock.notifyAll();
      }
    }
    return true;
  }

  RunnableLock removeFirst() {
    synchronized (messageLock) {
      if (messageCount == 0)
        return null;
      RunnableLock lock = messages[0];
      System.arraycopy(messages, 1, messages, 0, --messageCount);
      messages[messageCount] = null;
      if (messageCount == 0)
        messages = null;
      return lock;
    }
  }

  class RunnableLock {
    Runnable runnable;
    Thread thread;
    Throwable throwable;

    RunnableLock(Runnable runnable) {
      this.runnable = runnable;
    }

    boolean done() {
      return runnable == null || throwable != null;
    }

    void run() {
      if (runnable != null)
        runnable.run();
      runnable = null;
    }
  }

}









-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14751

[Bug libgcj/14751] [win32] thread creation leaks system handle resources

[ovidr at users dot sourceforge dot net]

------- Additional Comments From ovidr at users dot sourceforge dot net  2004-09-10 18:12 -------
swt-win32-3062.dll and source are here:

ftp://download.eclipse.org/R-3.0-200406251208/swt-3.0-win32.zip

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14751

[Bug libgcj/14751] [win32] thread creation leaks system handle resources

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-09-11 19:14 -------
Subject: Bug 14751

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	membar@gcc.gnu.org	2004-09-11 19:13:51

Modified files:
	libjava        : ChangeLog win32-threads.cc 
	libjava/include: win32-threads.h 

Log message:
	PR libgcj/14751
	* win32-threads.cc (_Jv_ThreadInitData): Zero out thread
	handle in newly-allocated _Jv_Thread_t.
	(_Jv_ThreadDestroyData): Close thread handle.
	(_Jv_ThreadStart): Remove obsolete comment.
	Store handle of newly-created thread in _Jv_Thread_t.
	* include/win32-threads.h: #define WIN32_LEAN_AND_MEAN
	before including <windows.h>
	#define _Jv_HaveCondDestroy

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libjava/ChangeLog.diff?cvsroot=gcc&r1=1.3050&r2=1.3051
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libjava/win32-threads.cc.diff?cvsroot=gcc&r1=1.11&r2=1.12
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libjava/include/win32-threads.h.diff?cvsroot=gcc&r1=1.9&r2=1.10



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14751

[Bug libgcj/14751] [win32] thread creation leaks system handle resources

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-09-11 19:32 -------
Fixed for 4.0.0.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|---                         |4.0.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14751