Name: bsT130419 Date: 10/02/2001


java version "1.2.2"
Solaris VM (build Solaris_JDK_1.2.2_05, native threads, sunwjit)

If a subprocess is started and a SIGSTOP is sent to it, then a subsequent
Process.destroy() call will fail to kill the process (although a ``kill -TERM''
will).

  Interesting note: if the process is sent a SIGSTOP and then a SIGCONT, the
Process.destroy() call will also fail. (That is, if the subprocess
*ever* gets a SIGSTOP, it seems to stop the Process.destroy() from
succeeding). Unless one sends their own SIGTERM after the SIGCONT.

Anyway, here is the code (copied from the bug I mentioned initially), except
for some comment changes.

import java.io.*;

public class Foo {

    public static void main(String args[])
    {
        try
        {
            Process foo = Runtime.getRuntime().exec("java Bar");
            System.out.println("Started Bar");

            // Wait 10 seconds
            // at this point, suspend this process
            // and send some of the signals mentioned
            // in the bugreport to the ``Bar'' subprocess

            Thread.currentThread().sleep(10000);

            // try to kill the process
            System.out.println("Killing runjava");
            foo.destroy(); // doesn't work

            // get exit value -- will throw exception if process not killed!
            foo.exitValue();
        }
        catch (InterruptedException e) {System.out.println(e.getMessage());}
        catch (IOException e) {System.out.println(e.getMessage());}
        catch (IllegalThreadStateException e) {System.out.println(e.getMessage());}
    }
}


public class Bar {

    public static void main(String args[])
    {
        // Just sit here.
        Object foo = new Object();
        synchronized (foo) {
            try {
                foo.wait();
            } catch (InterruptedException e) {}
        }
    }
}

(Review ID: 132325)
======================================================================