A Detailed Study with Examples of Classes and Methods

First published in JavaWorld, January 1997

---

Welcome to another installment of Under The Hood. This column aims to give Java developers a glimpse of the mysterious mechanisms clicking and whirring beneath their running Java programs. This month's article continues the discussion of the bytecode instruction set of the Java virtual machine by examining the manner in which the Java virtual machine handles exception throwing and catching, including the relevant bytecodes. This article does not discuss finally clauses -- that's next month's topic. Subsequent articles will discuss other members of the bytecode family.

Exceptions
Exceptions allow you to smoothly handle unexpected conditions that occur as your programs run. To demonstrate the way the Java virtual machine handles exceptions, consider a class named NitPickyMath that provides methods that perform addition, subtraction, multiplication, division, and remainder on integers. NitPickyMath performs these mathematical operations the same as the normal operations offered by Java's "+", "-", "*", "/", and "%" operators, except the methods in NitPickyMath throw checked exceptions on overflow, underflow, and divide-by-zero conditions. The Java virtual machine will throw an ArithmeticException on an integer divide-by-zero, but will not throw any exceptions on overflow and underflow. The exceptions thrown by the methods of NitPickyMath are defined as follows:

class OverflowException extends Exception {
}

class UnderflowException extends Exception {
}

class DivideByZeroException extends Exception {
}


A simple method that catches and throws exceptions is the remainder method of class NitPickyMath:

static int remainder(int dividend, int divisor)
    throws DivideByZeroException {

    try {
        return dividend % divisor;
    }
    catch (ArithmeticException e) {
        throw new DivideByZeroException();
    }
}


The remainder method simply performs the remainder operation upon the two ints passed as arguments. The remainder operation throws an ArithmeticException if the divisor of the remainder operation is a zero. This method catches this ArithmeticException and throws a DivideByZeroException.

The difference between a DivideByZero and an ArithmeticException exception is that the DivideByZeroException is a checked exception and the ArithmeticException is unchecked. Because the ArithmeticException is unchecked, a method need not declare this exception in a throws clause even though it might throw it. Any exceptions that are subclasses of either Error or RuntimeException are unchecked. (ArithmeticException is a subclass of RuntimeException.) By catching ArithmeticException and then throwing DivideByZeroException, the remainder method forces its clients to deal with the possibility of a divide-by-zero exception, either by catching it or declaring DivideByZeroException in their own throws clauses. This is because checked exceptions, such as DivideByZeroException, thrown within a method must be either caught by the method or declared in the method's throws clause. Unchecked exceptions, such as ArithmeticException, need not be caught or declared in the throws clause.

javac generates the following bytecode sequence for the remainder method:

The main bytecode sequence for remainder:

   0 iload_0 // Push local variable 0 (arg passed as divisor)
   1 iload_1 // Push local variable 1 (arg passed as dividend)
   2 irem // Pop divisor, pop dividend, push remainder
   3 ireturn // Return int on top of stack (the remainder)

The bytecode sequence for the catch (ArithmeticException) clause:

   4 pop // Pop the reference to the ArithmeticException
                           // because it isn't used by this catch clause.

   5 new #5 <Class DivideByZeroException>
                           // Create and push reference to new object of class
                           // DivideByZeroException.

DivideByZeroException

   8 dup // Duplicate the reference to the new
                           // object on the top of the stack because it
                           // must be both initialized
                           // and thrown. The initialization will consume
                           // the copy of the reference created by the dup.

   9 invokenonvirtual #9 <Method DivideByZeroException.<init>()V>
                           // Call the constructor for the DivideByZeroException
                           // to initialize it. This instruction
                           // will pop the top reference to the object.

  12 athrow // Pop the reference to a Throwable object, in this
                           // case the DivideByZeroException,
                           // and throw the exception.



The bytecode sequence of the remainder method has two separate parts. The first part is the normal path of execution for the method. This part goes from pc offset zero through three. The second part is the catch clause, which goes from pc offset four through twelve.

The irem instruction in the main bytecode sequence may throw an ArithmeticException. If this occurs, the Java virtual machine knows to jump to the bytecode sequence that implements the catch clause by looking up and finding the exception in a table. Each method that catches exceptions is associated with an exception table that is delivered in the class file along with the bytecode sequence of the method. The exception table has one entry for each exception that is caught by each try block. Each entry has four pieces of information: the start and end points, the pc offset within the bytecode sequence to jump to, and a constant pool index of the exception class that is being caught. The exception table for the remainder method of class NitPickyMath is shown below:

Exception table:
   from to target type
     0 4 4 <Class java.lang.ArithmeticException>


The above exception table indicates that from pc offset zero through three, inclusive, ArithmeticException is caught. The try block's endpoint value, listed in the table under the label "to", is always one more than the last pc offset for which the exception is caught. In this case the endpoint value is listed as four, but the last pc offset for which the exception is caught is three. This range, zero to three inclusive, corresponds to the bytecode sequence that implements the code inside the try block of remainder. The target listed in the table is the pc offset to jump to if an ArithmeticException is thrown between the pc offsets zero and three, inclusive.

If an exception is thrown during the execution of a method, the Java virtual machine searches through the exception table for a matching entry. An exception table entry matches if the current program counter is within the range specified by the entry, and if the exception class thrown is the exception class specified by the entry (or is a subclass of the specified exception class). The Java virtual machine searches through the exception table in the order in which the entries appear in the table. When the first match is found, the Java Virtual Machine sets the program counter to the new pc offset location and continues execution there. If no match is found, the Java virtual machine pops the current stack frame and rethrows the same exception. When the Java virtual machine pops the current stack frame, it effectively aborts execution of the current method and returns to the method that called this method. But instead of continuing execution normally in the previous method, it throws the same exception in that method, which causes the Java virtual machine to go through the same process of searching through the exception table of that method.

A Java programmer can throw an exception with a throw statement such as the one in the catch (ArithmeticException) clause of remainder, where a DivideByZeroException is created and thrown. The bytecode that does the throwing is shown in the following table:

The athrow instruction pops the top word from the stack and expects it to be a reference to an object that is a subclass of Throwable (or Throwable itself). The exception thrown is of the type defined by the popped object reference.

Play Ball!: a Java virtual machine simulation
The applet below demonstrates a Java virtual machine executing a sequence of bytecodes. The bytecode sequence in the simulation was generated by javac for the playBall method of the class shown below:

class Ball extends Exception {
}

class Pitcher {

    private static Ball ball = new Ball();

    static void playBall() {
        int i = 0;
        while (true) {
            try {
                if (i % 4 == 3) {
                    throw ball;
                }
                ++i;
            }
            catch (Ball b) {
                i = 0;
            }
        }
    }
}


The bytecodes generated by javac for the playBall method are shown below:

   0 iconst_0 // Push constant 0
   1 istore_0 // Pop into local var 0: int i = 0;
                          // The try block starts here (see exception table, below).
   2 iload_0 // Push local var 0
   3 iconst_4 // Push constant 4
   4 irem // Calc remainder of top two operands
   5 iconst_3 // Push constant 3
   6 if_icmpne 13 // Jump if remainder not equal to 3: if (i % 4 == 3) {
                          // Push the static field at constant pool location #5,
                          // which is the Ball exception itching to be thrown
   9 getstatic #5 <Field Pitcher.ball LBall;>
  12 athrow // Heave it home: throw ball;
  13 iinc 0 1 // Increment the int at local var 0 by 1: ++i;
                          // The try block ends here (see exception table, below).
  16 goto 2 // jump always back to 2: while (true) {}
                          // The following bytecodes implement the catch clause:
  19 pop // Pop the exception reference because it is unused
  20 iconst_0 // Push constant 0
  21 istore_0 // Pop into local var 0: i = 0;
  22 goto 2 // Jump always back to 2: while (true) {}

Exception table:
   from to target type
     2 16 19 <Class Ball>


The playball method loops forever. Every fourth pass through the loop, playball throws a Ball and catches it, just because it is fun. Because the try block and the catch clause are both within the endless while loop, the fun never stops. The local variable i starts at 0 and increments each pass through the loop. When the if statement is true, which happens every time i is equal to 3, the Ball exception is thrown.

The Java virtual machine checks the exception table and discovers that there is indeed an applicable entry. The entry's valid range is from 2 to 15, inclusive, and the exception is thrown at pc offset 12. The exception caught by the entry is of class Ball, and the exception thrown is of class Ball. Given this perfect match, the Java virtual machine pushes the thrown exception object onto the stack, and continues execution at pc offset 19. The catch clause merely resets int i to 0, and the loop starts over.

To drive the simulation, just press the "Step" button. Each press of the "Step" button will cause the Java virtual machine to execute one bytecode instruction. To start the simulation over, press the "Reset" button. To cause the Java virtual machine to repeatedly execute bytecodes with no further coaxing on your part, press the "Run" button. The Java virtual machine will then execute the bytecodes until the "Stop" button is pressed. The text area at the bottom of the applet describes the next instruction to be executed. Happy clicking.

Resources

• The book The Java Virtual Machine Specification, by Tim Lindholm and Frank Yellin (ISBN 0-201-63452-X), part of The Java Series (http://www.aw.com/cp/javaseries.html), from Addison-Wesley, is be the definitive Java virtual machine reference.
  http://www.aw.com/cp/lindholm-yellin.html

This article was first published under the name How the Java virtual machine handles exceptions in JavaWorld, a division of Web Publishing, Inc., January 1997.

Talk back!

Have an opinion? Be the first to post a comment about this article.

About the author

Bill Venners provides custom software development and consulting services in Silicon Valley under the name Artima Software Company. He has been object oriented for 5 years, primarily working in C++ on MS Windows. Before that he did a lot of C on UNIX and assembly language on various microprocessors. He is currently attempting to write the Great American Java Book.