Chapter 11

JVM Architecture
➢ In this chapter, You will learn
o What is VM, JVM, and Types of JVMs, HotSpot technology
o What will happen when we run java command?
o 5 phases in a program development, compilation & execution
o Why Java compiled code is named bytecode?
o .class file structure
o JVM: specification, implementation, and instance
o JVM Architecture Block diagram, JVM Components
o Explanation on Class Loader subsystem (types & algorithm)
o Class Life Cycle: Loading, Linking, Initialization
o Explanation on custom ClassLoader creation
o Explanation on Five Runtime Data Areas
o Explanation on Thread and Stack frame architecture
o Explanation on Method Calls execution flow with PC Register
o Explanation on Java Native Interface (JNI, native method)
o Explanation on Execution Engine (interpreter & JIT Compiler)
o Draw JVM Architecture with programs execution flow
➢ Memory distribution for all 4 types of variables
➢ Methods execution control flow with Stack frames
➢ Program execution with multiple classes with IS-A &
HAS-A relations.
o Changing JVM max & min memory sizes
o Finding JVM Memory Statistics-
▪ Total memory, Free memory, Used memory
o Stand-alone application lifecycle phases & execution flow
o JVM Architecture Final Diagram with complete internal details
➢ By the end of this chapter- you will learn all Components of JVM,
Runtime Data Areas, and their functionalities with complete
execution flow of static and non-static members of a class.
Learn Java with Compiler and JVM Architectures JVM Architecture

JVM Architecture
In this chapter we will learn the internals of JVM, different components of JVM, ClassLoader
subsystem, classes loading mechanism, runtime areas of JVM, execution engine, JNI, different
activities performed inside JVM when class is loaded and object created, memory distribution
for all 4 types of variables, methods execution procedure, etc…

You, being a programmer, no need to learn JVM internals. Instead if you can gain some basic
knowledge on what is happening inside JVM during our class execution then this knowledge
will help you understanding Java program execution easily. This will also help you finding
correct output of an unknown program in interview written test in addition to an
enhancement of skill to develop high performance applications.

If you compare, Knowing about JVM internals is like knowing about Human anatomy in deep
like hands, mouth, throat, stomach, blood and other internal parts which are helping the
digestive system to perform properly to generate energy from the food we eat. Of course, here
we, no need to know about complete functionality of these parts as we are not doctors and
also not going to do any surgery. But by just knowing outer functionality, we can take complete
care in getting energy generated by eating proper food, at proper time in a proper way to live
more years healthy and happy. By knowing body part functionality we can perform first aid to
save life.

So if you compare JVM with your body parts, Food is nothing but the Program you are given to
JVM, then JVM’s internal components are like hand, mouth, throat, stomach will execute
(digest) the given program and gives the output(energy).

As I said in first paragraph, in this chapter we only learn the essential parts of JVM and their
functionality. If you are interested to learn complete details of JVM, you can refer ”JVM
specification given by Oracle (previously Sun)” or refer a text book “Inside the JVM, by Bill
Venners”.

Let us learn JVM Architecture starts with definition

Definition of Java Virtual Machine:

➔ JVM stands for Java Virtual Machine. It is a Process Based Virtual Machine. JVM is a Java
platform it is responsible to run Java bytecode by translating them into current OS machine
language.JVM provides runtime environment with an equally partitioned identical memory
areas for executing Java bytecode in any OS.

The main reason behind the invention of JVM is to achieve platform independency, means JVM
is invented for running Java compiled code, bytecode, in all available OS irrespective of the OS
in which it is compiled. For this purpose, Oracle Corporation, previously SUN Microsystems,
developed separate JVM for every OS for running Java compiled code by translating it into
current OS specific machine language.

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 1

Learn Java with Compiler and JVM Architectures JVM Architecture

Virtual and Virtual Machine

The concept called Virtual Machine is not newly invented for Java. The VM concept is already
existed in OS and other languages. So Let us first get some idea on the word Virtual and Virtual
Machine, and then we will learn more about Java VM.

Well, what is Virtual?

✓ This real-world is physical. You and your friends in this real-world are physical.
✓ Your dream world is virtual. You and your friends in dream world are virtual.
The meaning of Virtual is not physical. It is a logical representation of one physical thing.

In computer world the meaning of Virtual is:

➔ A logical representation of a physical thing created in a computer through software.
For example:
1) A photo album in a computer is a Virtual family.
2) In a Bike Racing game the Bike is a virtual Bike.
3) In a Temple Run Game the person running and jumping is a virtual human being
In above examples- the family, the bike and the person are created virtually using software in
computer. They are not physically existed inside computer.

Definition of Virtual Machine

1) A Virtual Machine is a software implementation of a computer machine. It creates an
isolated duplication of real computer and allows us to perform operations as we
perform operations in a real computer.

Q2) Well, What is the need of Virtual Machine software, when we can perform operations
directly in computer?
If we perform operations directly in a computer we will have below problems as shown above
1. In one computer we can install only one OS
2. We can run only current OS specific compiled code in a platform dependent way.
3. We cannot use hardware resources effectively

If we perform operations through Virtual machine software will have below benefits
1. In a single computer we can install one OS in another OS
2. We can run any OS compiled program in current OS in a platform independent way
3. We can use hardware resources effectively

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 2

Learn Java with Compiler and JVM Architectures JVM Architecture

Summary: Purposes of Virtual Machine

Basically Virtual machines are used for below two purposes
1) For installing one OS in another OS in the same computer, installing Linux in Windows.
2) For proving runtime environment for executing a language application in a platform
independent way.
3) We can use hardware resources effectively

Types of Virtual Machines

Virtual machines are separated into two major classes, based on their use and degree of
correspondence to a real machine:
1. A System Virtual Machine ( also called as Hardware Virtual Machine)
2. A Process Virtual Machine (also called as Language Virtual Machine or an Application
Virtual Machine, or Managed Runtime Environment)

System Virtual Machine

1) System virtual machine is a software emulation of a computer system.
2) It means System Virtual Machine mimics the entire computer.
3) It provides a platform/runtime enviroment for the execution of a complete operating
system.
4) It will create number of different isolated identical execution environments in the single
computer by partitioning computer memory to install and execute different OS at time.
5) Then allows us to install applications in each OS, run application in this OS as if we work
in real computer. For example we can install Windows XP/Linux Ubuntu in Windows 7/8
OS with the help of Virtual Machine.

Examples of System Virtual Machine software

1. VMware 4. QEMU
2. VirtualBox 5. Citrix Xen
3. Parallels 6. Windows Virtual PC

Process virtual machine

1) Process virtual machine is a software simulation of a computer system.
2) It means Process Virtual Machine does not mimics the entire computer.
3) Process VMs provide runtime environment to execute a single program, and supports a
single process.
4) The purpose of a Process VM is to provide a platform-independent programming
environment that abstracts the details of the underlying hardware or operating system,
and allows a program to execute in the same way on any platform.
5) Process VMs are implemented using an interpreter; for improving performance these
VMs will use just-in-time compilers internally.
Examples of Process VMs
1. PVM - Parrot Virtual Machine for Perl Language
2. JVM - Java Virtual Machine for Java language
3. CLR - Common Language Runtime for .NET Framework
4. PVM - Python Virtual Machine for Python language

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 3

Learn Java with Compiler and JVM Architectures JVM Architecture

Types of JVMs:
The Java 2 SDK, Standard Edition, contains two implementations of the Java virtual machine
• Java HotSpot Client VM
• Java HotSpot Server VM
Java HotSpot Client VM:
The Java HotSpot Client VM is the default virtual machine. By default java program is executed
with client VM. As its name implies, it is tuned for best performance when running applications
in a client environment by reducing application start-up time. The Client version is tuned for
quick loading. It makes use of interpretation.
Java HotSpot Server VM:
The Java HotSpot Server VM is designed for maximum program execution speed for
applications running in a server environment. The Server version loads more slowly, putting
more effort into producing highly optimized JIT compilations, that yield higher performance.

The JVM launcher, java, by default starts client VM, if we want to start the Java HotSpot Server
VM we must use the java command-line option -server when launching an application, as
shown below:
java Test -> class Test is executed by client VM
java -server Test -> class Test is executed by server VM

History of HotSpot
HotSpot is a technology it improves performance of application execution. This technology
provides techniques such as just-in-time compilation and adaptive optimization designed to
improve performance.

The Java HotSpot Performance Engine, first released in April 27, 1999, built on technologies
from the Strongtalk implementation of the Smalltalk programming language originally
developed by Longview Technologies, which traded as "Animorphic". Animorphic's virtual-
machine technology had earlier been successfully used in a Sun research project, the Self
programming language. In 1997 Sun Microsystems purchased Animorphic.

Shortly after acquiring Animorphic, Sun intended to write a new just-in-time (JIT) compiler for
the newly developed virtual machine. This new compiler would give rise to the name
"HotSpot", which derives from the fact that, as the software runs Java bytecode, it continually
analyzes the program's performance for "hot spots" which are frequently or repeatedly
executed. These are then targeted for optimization, leading to high-performance execution
with a minimum of overhead for less performance-critical code. In one report, the JVM beat
some C++ or C code in some benchmarks.

Initially available as an add-on for Java 1.2, HotSpot became the default Sun JVM in Java 1.3.

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 4

Learn Java with Compiler and JVM Architectures JVM Architecture

Features of HotSpot technology

Some of the features of Java HotSpot technology, common to both VM implementations, are
the following.
Adaptive compiler
• Applications are launched using a standard interpreter, but the code is then analyzed as
it runs to detect performance bottlenecks, or "hot spots".
• The Java HotSpot VMs compile those performance-critical portions of the code for a
boost in performance, while avoiding unnecessary compilation of seldom-used code
(most of the program).
• The Java HotSpot VMs also uses the adaptive compiler to decide, on the fly, how best
to optimize compiled code with techniques such as in-lining.
• The runtime analysis performed by the compiler allows it to eliminate guesswork in
determining which optimizations will yield the largest performance benefit.

Rapid memory allocation and garbage collection:

• Java HotSpot technology provides for rapid memory allocation for objects, and it has a
fast, efficient, state-of-the-art garbage collector.

Thread synchronization:
• The Java programming language allows for use of multiple, concurrent paths of
program execution (called "threads").
• Java HotSpot technology provides a thread-handling capability that is designed to scale
readily for use in large, shared-memory multiprocessor servers.

Q) How can we start JVM execution?

A) By using "java" tool we will start JVM execution.

With the above command, JVM process will be launched by occupying some memory from
RAM, Test class bytecode will be loaded into JVM memory and executed, output is displayed.

How can we start multiple JVMs at a time?

A command prompt window can hold only one JVM at a time. Then, if we want to start
multiple instances of JVM at a time, we must run java command from different command
prompts at a time as shown below.

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 5

Learn Java with Compiler and JVM Architectures JVM Architecture

What will happen when we run java command?

When we run java command, JVM is created as a process in OS by occupying some memory
from RAM. For executing the given class, JVM performs internally below three phases as
shown in below diagram:

With respect to above diagram:

1. Class Loading phase:
In class loading phase ClassLoader will load given class bytecode in JVM.

Once JVM process is created, JVM will search the given class bytecode in its memory. If the
given class is not loaded into JVM, JVM will request ClassLoader to load the given class
bytecode. Then ClassLoader will load the given class bytecode for the hard disk folder
which is configured in classpath environment variable.

If the given class file is not found, ClassLoader will throw exception
java.lang.ClassNotFoundException/java.lang.NoClassDefFoundError (both are same).

From Java 8 onwards exception is displaying as

Error: Could not find or load main class
2. Bytecode verification phase:
In Bytecode verification phase loaded bytecode will be verified.

After class loading phase, BytecodeVerifier verifies the loaded bytecode’s internal format.
If the bytecode is in JVM understandable format, it allows interpreter to execute loaded
bytecode. If loaded bytecode is not in the JVM understandable format, it will terminate
execution process by throwing exception "java.lang.ClassFormatError".

3. Execution / interpretation phase:

In the interpretation phase loaded bytecode will be executed and displays output.

After byte code verification completed, interpreter will executed loaded bytecode by
converting them into current OS machine language. While executing bytecode if any logical
mistakes found, JVM will throw associated exception.

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 6

Learn Java with Compiler and JVM Architectures JVM Architecture

Five phases diagram

Java program development, compilation and execution contains five phases, they are
1. Editing phase
2. Compilation phase
3. ClassLoading
4. BytecodeVerification
5. Execution

Q) What is bytecode?
Java bytecode is the instruction set of the Java virtual machine. It is an intermediate code of all
operating systems. The basic need of bytecode is to achieve platform independency through
JVM. Compiler software will convert Java Source code into bytecode, and then JVM will
convert bytecode into machine language code of current OS in which it is running.

Q) Why Java compiled code is named as bytecode?

The word bytecode is derived from instruction sets which have one-byte opcodes followed by
optional parameters. It means, the Java compiled code is named as bytecode based on what a
.class file contains. Actually the .class file is made up of opcode and operands. Each opcode is
of size "1 byte", and in the most of the cases the instructions in the .class file are only the
opcodes. So the java compiled code is named as byteopcode, later they are called as bytecode.

The Java bytecode instruction set is made up of 256 opcodes. As of 2015, 198 are in use, 54 are
reserved for future use, and 3 instructions are set aside as permanently unimplemented.

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 7

Learn Java with Compiler and JVM Architectures JVM Architecture

JVM: a specification, implementation, and instance

JVM is
1. a Specification
2. a Concrete implementation
3. a Runtime instance
JVM is a specification means it is an abstract documentation it provides set of rules and
guidelines for implementing JVM functionality.
JVM is a concrete implementation means it is a software implementation of JVM specification
given by different vendors in the world such as SUN, Oracle, IBM, etc... JVM software
implementation is available for all verities of OS like windows, Linux, Solaris, Mac OS for
running java byte code in all OS in a platform independent way.
JVM is a runtime instance means when we run java command one separate instance of JVM is
created for running given Java application.

JVM Architecture block diagram

When we run java command for executing an application, one instance of JVM will start as
shown below by occupying some memory from RAM. Below diagram will give you pictorial
representation of JVM internal architecture with all its internal components:

JVM Components:
As shown in above diagram JVM is a combination of 4 components
1. ClassLoader subsystem
2. Runtime Data Areas
3. Execution Engine
4. Java Native Interface(JNI)

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 8

Learn Java with Compiler and JVM Architectures JVM Architecture

Runtime Data Areas are responsible to provide memory to store byte code, objects,
parameters, local variables, return values and intermediate results of computations.
ClassLoader subsystem is responsible to load and store given class byte code in to JVM from
both the program and the Java API and also it is responsible to link the loaded class to runtime
application of JVM and initialize the loaded class memory.

Execution engine is responsible to execute the loaded byte code with the help of interpreter
and JIT compiler.
Java Native Interface (JNI): It is responsible to link Java method to native method, executes
native method from a Java program and returns its result to Java program.
Below diagram will show you the interaction between ClassLoader, Execution engine, and JNI.

Explanation on ClassLoader subsystem

ClassLoader is a class it is responsible to perform below three activities
➔ Load, Link and Initialize the given class byte code
Below diagram will given you outlook of a class loader subsystem

ClassLoader subsystem

Loading Phase Linking Phase Initialization Phase

BootStrapClassLoader Verify

ExtensionClassLoader Prepare Initialize

ApplicationClassLoader Resolve

In Class Loading phase ClassLoader will load & store given class bytecode into JVM’s Method
Area. ClassLoader subsystem contains three types of class loaders for loading core Java API
classes and userdefined classes. They are
1. BootstrapClassLoader/PrimordialClassLoader
2. ExtensionClassLoader
3. ApplicationClassLoader/SystemClassLoader

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 9

Learn Java with Compiler and JVM Architectures JVM Architecture

BootStrapClassLoader
1) It is responsible to load Java library classes from BootStrapClasspath
i.e.; from jdk\jre\lib\rt.jar file.
2) It is part of the Java Virtual Machine implementation.
3) It is implemented in native API, so it does not have class name.
4) If we try to display its name, null will be displayed.

ExtensionClassLoader
1) It is responsible to load user defined classes or Java library classes from
ExtensionClasspath, i.e.; from jdk\jre\lib\ext folder.
2) It is part of Java API implementation.
3) This class loader is implemented in Java with class name ExtClassLoader.
4) This class is defined as inner class in sun.misc.Launcher class.
5) So, when we try to display its name it is displayed as sun.misc.Launcher$ExtClassLoader.

ApplicationClassLoader
1) It is responsible to load user defined classes from ApplicationClasspath.
2) It will load classes from current working directory or from the folders configured in
classpath environment variable.
3) This class loader is also implemented in Java with class name AppClassLoader.
4) This class is also defined as inner class in sun.misc.Launcher class.
5) So, when we try to display its name it is displayed as sun.misc.Launcher$AppClassLoader.

ClassLoader Working procedure:

ClassLoader will work on Parent-delegation algorithm. When ClassLoader gets request to load
a class, it will first delegates this class loading request to its parent. The parent class loader, in
turn, delegates this request to its parent. This chain of delegation continues through to the
bootstrap class loader. In this chain of hierarchy the bootstrap class loader is at the top.

1) When JVM come across a type, it checks for that class bytecodes in method area. If it is
already loaded, it makes use of the loaded class bytecode type. If the class is not yet
loaded, JVM requests class loader subsystem to load this class.
2) Then ClassLoader subsystem, first handovers this request to ApplicationClassLoader.
3) As per the Parent-delegation principle, the application loader, without searching for
this class in the folders configured in Classpath environment variable, delegates the
request to its parent ClassLoader ExtensionClassLoader.
4) Then ExtensionClassLoader, in turn delegates request to its parent ClassLoader
BootStrapClassLoader.
5) BootStrapClassLoader is the final parent ClassLoader, then it will search the given class
in BootStrapClasspath(jdk\jre\lib\rt.jar and other jar files available in lib folder)
6) If class is found, it will load the class into JVM. If class is not found,
BootStrapClassLoader will send request back to ExtensionClassLoader.
7) Then ExtensionClassLoader will search this class in ExtensionClasspath(jdk\jre\lib\ext
folder jar files).

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 10

Learn Java with Compiler and JVM Architectures JVM Architecture

8) If class is found in ext folder jar files, it will load the class into JVM. If class is not found,
ExtensionClassLoader sends request back to ApplicationClassLoader.
9) Then ApplicationClassLoader will search this class in ApplicationClasspath(classpath
folders).
10) If class is found in application classpath it will load the class into JVM. If here also class
is not found, then we get an exception "java.lang.NoClassDefFoundError" or
"java.lang.ClassNotFoundException".

11) If class is found in any one of the classpaths, the respective ClassLoader will load this
class into JVM's method area by creating java.lang.Class instance.

12) Then JVM will use the loaded bytecodes to complete the execution of this class.
Below diagram will show you the class loader subsystem architecture
5
Searches in BootStrapClasspath
Bootstrap (jdk\jre\lib\rt.jar file)
ClassLoader
6
4
7
Searches in ExtensionClasspath
Extension (jdk\jre\lib\ext folder)
ClassLoader
8
3

Application Searches in ApplicationClasspath

ClassLoader (classpath folders)
ClassLoader
Sub System 2

or 10
java.lang.ClassNotFoundException
1 10

j.l.C instance
Given
class
Bytecode 11
Stored

Note: AppClassLoader will load class only if class is not found by Bootstap and ExtClassLoader

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 11

Learn Java with Compiler and JVM Architectures JVM Architecture

Q) If the given class is found in Extension directory and alos in local directory, from which
directory this class is loaded and by which class loader?
A) class will be loaded from extension directory by ExtClassLoader
As per parent-delegation algorithm, the extension classpath will be given first priority. Then if
class is given class is found in both extension folder and in current working directory, this class
will loaded from extension folder by ExtClassLoader.
Create and Run below classes to find the class loader through which the given class is loaded
//Example.java //Sample.java
class Example{ package p1;
public class Example{
}
}

//ClassLoaderNamePrinter.java
class ClassLoaderNamePrinter{
public static void main(String[] args) {

System.out.println( java.lang.String.class.getClassLoader());

System.out.println( p1.Sample.class.getClassLoader());

System.out.println( Example.class.getClassLoader());
}
}
Save Example.class in only local folder and Sample.class in both ext folder and local folder,
then Compile and Execute above main method class as shown in below diagram

1. Compile all above three

classes as shown here
2. Create jar file with
p1.Sample class
3. Copy test.jar file into
jdk\jre\lib\ext folder
4. Then execute, observe
output, the class Sample
String class:
Sample class is loaded from ext folder
Example class
by ExtClassLoader, but
not from local directory

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 12

Learn Java with Compiler and JVM Architectures JVM Architecture

The Class Life Time: Loading, Linking, and Initialization

In this section we will discuss the lifetime of a type (class or interface) from its initial entrance
into the virtual machine to its ultimate exit. Every class execution will contain below life cycle
phases:
1. Loading 4. Execution 7. Finalization
2. Linking 5. Instantiation 8. Unloading
3. Initialization 6. Garbage Collection

➔ Loading, linking, and initialization will occur at the beginning of a class’s lifetime
➔ Execution, instantiation (object creation), garbage collection, and finalization will occur in
the middle of a class’s lifetime and
➔ The finalization and unloading of types will occur at the end of a class’s lifetime.
Loading, Linking and initialization
The Class loader subsystem will make types available to the JVM running program through a
process of loading, linking, and initialization. It means, the class loader subsystem is not only
responsible for loading the binary data for classes in to JVM. It must also verify the correctness
of imported classes, allocate and initialize memory for class variables, and assist in the
resolution of symbolic references. All these activities are performed in the order strictly as
shown in the below diagram:

1) Loading is the process of finding the given class, copy its bytecode from .class file and
storing its binary form in JVM’s method area.
2) Linking is the process of incorporating the binary type data into the runtime state of
the virtual machine. Linking is divided into three sub-steps:
▪ Verification, Preparation, and Resolution.
1) In Verification phase ClassLoader will the type is properly formed and fit for use
by the JVM.
2) In Preparation phase static variables memory is allocated with default values
based on their data type.
3) In Resolution phase symbolic references in the constant pool are transferred
into their direct references.
3) In initialization phase static block is executed and the static variables are initialized
with their initial values given in their declaration.
4) In execution phase class execution is stated with main method
5) Once main method execution is completed the class is unloaded from the JVM

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 13

Learn Java with Compiler and JVM Architectures JVM Architecture

Explanation on Runtime Data Areas

Runtime data areas provide memory to store bytecode, objects, method parameters, local
variables, return values, and intermediate results of computations. The Java Virtual Machine
organizes the memory it needs to execute a program into several runtime data areas.

Whenever we run java command for executing a class, the Java launcher, java, starts the Java
runtime environment as a process in OS by occupying some memory from RAM, and further
the entire Java Virtual Machine setup is divided into various run-time data areas that are used
during execution of a program. Some of these data areas are created on Java Virtual Machine
start-up and are destroyed only when the Java Virtual Machine exits. Other data areas are
created per thread. Per-thread data areas are created when a thread is created and destroyed
when the thread exits.
JVM totally contains five runtime areas
1. Method Area
2. Heap Area
3. Java Stacks Area
4. Program Counter Registers Area
5. Native Methods Stacks area
About Method Area and Heap Area
1) In Method area every class bytecode is loaded and stored. It means the given class
static and instance variables declaration statements, blocks, methods, and constructors
logic will be stored in method area.
2) Further for all static variables memory is allocated in this runtime area.
3) Heap is the main memory of JVM.
4) All classes and arrays objects are created in heap area. Means given class instance
variables memory is allocated in heap area.
5) For storing given class bytecode inside JVM, one separate instance is created from
java.lang.Class. As many classes are loading into JVM those many instances are created
for java.lang.Class as shown in the below diagram
6) Below diagram will show you classes bytecode storing and object creation in JVM

7) Both Method area and Heap area are sharable memory areas to all threads.
8) Method area and heap area are created on virtual machine start-up.

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 14

Learn Java with Compiler and JVM Architectures JVM Architecture

9) If there is no sufficient memory in method area for loading new class bytecode or there
is not sufficient memory in heap area for creating new object, then JVM will throw
exception java.lang.OutOfMemoryError.
10) The heap memory size can be increased at the time of setting up of runtime
environment using non standard option as show in the below command
java -xms <size> classname
11) The method area and heap area may be of a fixed size or may be expanded as required
by the computation and may be contracted if a larger heap becomes unnecessary.
Learning JVM architecture theory points is not enough. Let try to understand all above points
practically with a program execution with all four types of variable and their memory
distribution done inside runtime data areas:

With respect to above diagram:

1. When we run above class Sample, the java launcher, will create JVM as process.
2. Once JVM set up is completed, JVM will search for the Sample class bytecode in
Method Area. Since Sample class bytecode is not found, JVM will request class loader
subsystem to load the Sample class byte code into JVM.
3. AppClassLoader will load and store Sample class byte code Inside Method Area by
creating java.lang.Class object.
4. After bytecode verification is completed, in preparation phase memory will be
allocated for static variables a & b in their class context in Method area with default
values zero.
5. Then in initialization phase, static block is executed by creating a stack frame inside
main thread in Java Stack Area, static variables are initialized with assigned values 10
and 20. Once after static block execution is completed stack frame is destroyed.
6. After initialization, in Execution phase, main method logic is executed by creating
separate stack frame in main thread in Java Stack Area. Inside main method stack frame
its parameter args, local variables p, q, s1 and s2 memory is allocated.
7. In Heap Area one empty String[] array object, and two objects of Sample class are
created. Non-static variables x, y memory is allocated in heap area in their objects.
Since we have created two objects from Sample class, two copies of non-static
variables memory is created.

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 15

Learn Java with Compiler and JVM Architectures JVM Architecture

8. Once main method execution completed, its stack frame is destroyed, main thread is
destroyed and Sample class is unloaded, JVM destroyed.
9. In PC registers area one pc register is created for tracking main thread execution.
10. This complete execution is tracked by execution engine one of the JVM components.

Q) How a class byte code is exactly stored in JVM’s method area?

We knew ClassLoader stores the given class bytecodes in JVM by using java.lang.Class object.
The java.lang.Class object internally uses several other classes’ objects for storing given class
bytecode. The classes which are used for storing given class bytecode in JVM are collectively
called Reflection API.

Below classes are called Reflection API classes

1. Class
These 2 classes are available
2. Package in java.lang package
3. Field
4. Constructor These 4 classes are available
5. Method in java.lang.reflect package
6. Modifier

Below diagram will show you java.lang.Class object structure for storing given class bytecode
using above 6 classes.

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 16

Learn Java with Compiler and JVM Architectures JVM Architecture

About Java Stacks area

1) In this runtime area all methods (non-native), blocks and constructors logic is executed.
2) In this runtime area we can create several threads for executing methods at a time
concurrently. JVM by default creates minimum two threads, they are
1) main thread
2) garbage collector thread
3) Main thread is responsible to execute Java methods starts with main method, also
responsible to create objects in heap area if it finds "new" keyword in any method logic
4) Garbage collector thread is responsible to destroy all unused objects from heap area.
Note: Like in C++, in Java we do not have destructors to destroy objects.
5) As each new thread comes into existence, it gets its own pc register and Java stack as
shown in the below diagram. The Java stack is composed of stack frames (or frames). A
stack frame contains the state of one Java method invocation.
6) When we invoke a method or constructor a new stack frame is created. Once this
method execution completed this stack frame is destroyed. Stack Frame holds logic of
this method, its parameters, local variables, its return value, and intermediate
calculations. Logic of this method or constructor is loaded in this from method area

7) Above diagram shows a snapshot of a virtual machine instance in which three threads
are executing. Threads one and two are executing Java methods. Thread three is
executing a native method. As shown in above diagram, in this book we will draw stack
diagram, the stacks are growing downwards. The "top" of each stack is shown at the
bottom of the figure.
Note: Inside a thread if there is no sufficient memory for creating new stack frame for
execution a method, then JVM will throw exception java.lang.StackOverflowError.

Thread Architecture
1) Thread is an independent sequential flow of execution created in JSA.
2) Every thread is internally divided into multiple sub blocks.
3) This sub block technically called as stack frame.
4) As many methods as we invoke those many stack frames are created in this thread
sequentially.

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 17

Learn Java with Compiler and JVM Architectures JVM Architecture

5) Inside a thread only one stack frame is active at any point. When we invoke a method
from the currently executing method, this current frame is paused, new frame is
created for this invoked method, control is sent to this new frame, this invoked method
logic is loaded into this frame from method area, logic is executed. After this method
execution is completed, its stack frame is destroyed, control return to previous method
stack frame from which this method is invoked.
6) If this method has any parameter and local variables, they are all created inside this
method's stack frame, and are destroyed automatically when Stack Frame is destroyed.
7) Note that a frame created by a thread is local to that thread and cannot be referenced
by any other thread, its means it is not sharable like method area and heap area.
Below diagram will show you main thread with several methods stack frames

With respect to above program

1) Example class byte code is loaded and stored in method area
2) Then immediately JVM starts main method execution by creating a separate stack
frame in Java stacks area in main thread.
3) As part of its logic we have called m1() method, then a new stack frame is created for
m1() method execution, main method execution is paused, m1() logic is loaded into
new stack frame, logic is executed, output is displayed. After m1() execution
completed, stack frame is destroyed, control return to main method.
4) In next line of main method, we called m2() method. Then again new stack frame is
created, m2() method logic is loaded, logic executed, output displayed, after its
execution is completed, stack frame is destroyed, control is return to main method.
5) In main method we do not have more statements to execute, main method execution
is completed, and main method stack frame is destroyed, main thread destroyed, JVM
destroyed.
6) After all stack frames are destroyed, thread is destroyed.
7) After all threads destroyed, JVM is destroyed.

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 18

Learn Java with Compiler and JVM Architectures JVM Architecture

Stack Frame architecture

1) Stack Frame is sub block of a thread. When we invoke a method JVM will creates new
stack frame, loads the method logic, and executes this logic.
2) Stack Frame is internally divided into three blocks for executing method logic
They are:
1. Local Variables storage area
2. Operand stack (or method stack)
3. Frame data (or Reference to the runtime constant pool of this method)

▪ All parameters and local variables memory is created in Variables storage area
▪ Method logic is executed in operand stack. To execute method logic, operand stack is
divided into number of blocks. Each block is called slot. This slot size is 32bits (4 bytes - int/
float data type depending of the variable type).
▪ Due to the slot size 4 bytes the result value will be return as 4 bytes value. So due to this
reason in an expression byte, short, char data types are automatically promoted to int data
type. So, the result will coming out from an expression is int type.
▪ The frame data portion will contain data to support constant pool resolution, normal
method return, and exception dispatch information.
Below diagram shows Stack Frame architecture for add method

Some more details of stack frame with “this” and “super” keywords
We will learn in static & Non-static members chapter.

About Program Counter Registers Area

1) In this runtime area, a separate program counter register is created for every thread. It
is used for tracking execution in this thread by storing its instruction address.
2) PC Register is created when new thread is created; after thread execution is completed
it is destroyed automatically.
About Native Methods Stacks Area
1) In Native Methods stack area all native methods are executed.
2) In Java, there are two kinds of methods: Java methods and native methods.
1. A Java method is written in the Java language, compiled to bytecodes, and stored in
class files.
2. A native method is written in some other language, such as C, C++, or assembly,
and compiled to the native machine code of a particular processor.

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 19

Learn Java with Compiler and JVM Architectures JVM Architecture

3) To create native method, we must place native keyword in its prototype and it should
not have body. A Java program interacts with the host OS by invoking native methods.
For example
class Example{
public static native int add(int x, int y); It is a native method.
public static void main(String[] args) {
add(10, 20);
}
}
The above program add(int, int) method is a native method. It is compiled fine, but in
execution JVM will throw exception: java.lang.UnsatisfiedLinkError, because we just defined
native method, but we did not defined it required C program and not linked.
Q) If we define a program in C or C++, how can we link this native logic to native method?
A) By using JNI - Java Native Interface. It is a mediator between Java native method and
original native method for linking its method calls.
About Java Native Interface (JNI):
It is one of the JVM components. It is responsible to link Java method to native method,
executes native method from a Java program and returns its result to Java program.
About Execution Engine:
1) It is one of the JVM Components. It is responsible to execute java bytecode.
2) All executions happening in JVM are controlled by Execution Engine.
3) It uses interpreter, JIT compiler, Profiler, SecurityManager for executing a program
4) Execution Engine will also contains Garbage Collector for destroying unreferenced
objects from Heap Area. Below is Execution Engine Block diagram
Execution Engine

JIT Hotspot Security Garbage

Interpreter
Compiler Profiler Manager Collector

5) Execution engine will comes in two implementations

1) The simplest kind of execution engine is just with interpreter, it just interprets the
bytecodes one at a time.
2) Another kind of execution engine is a just-in-time compiler, it is faster but requires
more memory.
6) JIT Compiler will increase the performance of program execution.
7) JIT Compiler will find the code which is executed repeatedly, generates machine
language for this code only once, buffer it and reuses the same machine language code
for repeated execution. This is how JIT Compiler will give high performance in execution

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 20

Learn Java with Compiler and JVM Architectures JVM Architecture

JVM Architecture final diagram with conclusions

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 21

Learn Java with Compiler and JVM Architectures JVM Architecture

If you really want to learn more details about execution flow of your byte code do below
three things
1. After every class you compile, read its byte code using below command
javap –verbose classname

2. Study a Book called Inside the Java Virtual Machine - by Bill Venners

Naresh i Technologies, Ameerpet, Hyderabad, Ph: 040-23746666, 9000994007 Page 22