The sun.tools.debug.RemoteThread.next method does not always stop
execution of the thread at the next line of the method. This can be
illustrated with jdb and the following
program:

public class BugDemo {
    static int NotAnInstanceVar;
    String StringField;
    static BugDemo BUGDEMO;
    public static void main(String argv[]) {
        int i;
        BugDemo bd, bugdemo;
        bd = new BugDemo("foo");
        BUGDEMO = bd;
        i = argv.length;
        if (i > 0)
            System.out.println("I'm not expecting any trouble...");
        i *= 2;
        try {
            i = 16/i;
        } catch(Exception e) {
            System.out.println("That didn't work at all!");
        }
        System.exit(0);
    }
    public BugDemo(String s) {
        StringField = s;
    }
}

Start jdb on BugDemo and enter the following commands:

load BugDemo
stop in BugDemo.main
run BugDemo
next
next
next
next

The last next should step over line 11. The program runs to
completion.

Another similar situation with this sample:

start jdb and enter the following:

load BugDemo
stop at BugDemo:15
run BugDemo
next

The program takes an exception and completes. It should
stop at the next line - which in this case is in the catch block.

As we are a commercial licensee and have source, we have
looked into this problem a bit further:

The problem is due to the algorithm used in the method
sun.tools.debug.Agent.stepNextThread(). In order to step over a line of
code, a breakpoint is placed at the first instruction of the next source
line. If there is some branching instruction that causes execution to
avoid the next line, such as in an if clause or in the case of an
exception, the program runs until it hits some other breakpoint or
completes. In our sample, a goto instruction in line 11 and an exception
in line 13 caused the flow of control to avoid the breakpoint.