JVM Structure

JVM Structure

In the JAVA Virtual Machine, There are mainly 4 parts.

• Class Loader
• Execution Engine
• Native Interface
• Runtime Data Area

Class Loader

Class Loader is in charge of loading the class files. The class files are specified by particular file markers at the file header, and ClassLoader is only in charge of the loading process of the class files. As for the execution of the file, it is decided by the Execution Engine.

Native Interface

The function of Native interface is to consilidate other programming language to be used by Java, the original intention of it is to fuse C/C++ programs. The invention period of Java is the time when C/C++ occupies the programming world. To establish its popularity, it has to be able to call C/C++ programs, thus, there develops an area to cope with the code specified by native. The detailed method is to register native method in the Native Method Stack, and load native libraries when executing the Executing Engine.

This method is fading out gradually, unless when coping with the utilization of the haedwares, such as the execution of printing machine by the drive of Java and controll the production machines by Java, which are rare in today’s enterprise application.

Method Area

Method area is sharing by all the Threads, including all the characters all the bytes and some special methods shuch as constructors, besides, Interfaces are also defined here. In a nutshell, all the information of the definition of methods are kept in this area, which is a communal space.

PC Register

Every threads has a program conuter, like a pointer, pointing to the Method Bytecodes(The execution code next to be executed), and then the execution engine points to the next order, which is a tiny space, almost negligible.

Native Method Stack

The execution process is to register the native methods in the Native Method Stack. When the Execution Engine runs, it laods native libraries.S

Stack

Stack is also called Stack Memory which is in charge of the operation of Java program. It is created when the threads are created, and its lifecycle is also followed by the lifecycle of the threads, which means that the Stack memory is released when the threads are end. As for the Stack， there is not the problem of recycling the memory, it is owned privately by the threads. Primative variables and the reference of the objects are all allowcated in the function’s stack memory.

The Stack mainly stores three types of data:

• Local variables: Input parameters, output parameters and the variables in the function.
• Operand Stack: Record the operation of stack input and stack output.
• Frame Data: Includes class files, functions and so on.

The Principle of Stack

Data in the stack exists as a form of Stack Frame. Stack Frame is a memory area, which can be described as a data set relating to the data produced by the execution of Methods and runtime data. For example, when a method A is called, it produces a stack frame F1, and it is pushed into the stack. When method A calls method B, the produced stack frame F2 is also pushed into the Stack. When method B calls method C, the produced Stack Frame F3 is also pushed into the stack

…..

After the execution, F3 is firstly poped out of the stack, then F2, then, F1 …… which follows the principle of FIFO(First in First out).

Heap

Heap structure:

• Young Generation Area

As the picture demonstrated above, Young Generation Area is the birtn, growth, fade area of a class. A class is born here, utilized here, and collected by the garbage recycling mechnism in the end and terminated its life. The Young Generation Area is divided into two parts: Eden space and Survivor Space, all the Classes are new in the Eden Space. The Survivor spaces has two instances: 0 area(Survivor 0 Space) and 1 area(Survivor 1 Space). When the space in the Eden Space runs out, and the program has to create new objects, the garbage recycling mechanism in JVM0 will make the garbage recycling in Eden Space(Minor GC), and terminate the objects that is on longer referenced by other objects in the Eden Space. Then, it will transfer the remaining objects to Survivor 0 Space. If unfortunately, the Survivor 0 Space is also full, it will make garbage recycling in this area, then move the survivors to the Survivor 1 area. What if the Survivor 1 Space is also ful? it will transfer the surviving objects to the Tenured. Then, if the Tenured is also full, the Major GC(Full GC) will be generated, and make cleaning in the Tenured spaces. If after the Fiull GC execution in the Tenured space, it can not save the newly-created objects neither, there will produce a OOM exception OutOfMemoryError.

• Tenured Generation Area

Tenured Generation Area is used to keep the JAVA objects filtered in the Young Generation Area.

• Permanent Area

Permanent Area is a space used to store the residnet data, it is ultilized to deposit metadata of class and interface in the JDK. In other words, it stores the required class information when operating the JDK environment, the data loaded into this area can not be recyled by the garbage collector. Only through shutting down the JVM can it release the memory it takes.

• JDK1.6 & before: There is permanent area
• JDK1.7: There is permenent area, however, the permanent area has stepped out. The constant pool is in heap
• JDK1.6: There does not exist permanent area. The constant pool is in metaspace.

In terms of actual, Method area is like Heap, is the memory that threads share, and it is used to store the data JVM loads: metadata + normal consants + static constants + the codes after compilation etc. Although JVM describe the Method Area as a logical part of the heap, it has another name called Non-Heap, in order to distinguish with the Heap.

Constant Pool is a part of method area, except the methods, interfaces, fileds, versions that Class files contains, another information is constant pool. This part of information will be stored in the run-time constant pool after the loading process of class.

For example, an demo Java program describing the memory distribution is demonstrated below:

• As soon as we run the program, it loads all the Runtime classes into the Heap space. When main() method is found at line 1, Java Runtime creates stack memory to be used by main() method thead.
• We are creating primitive local variables at line 2, so it is created and stored in the stack memory of main() method.
• Since we are creating an Object in line 3, it’s created in Heap memory and Stack Memory contains the reference for it. Similar process occurs when we create Memory object in line 4.
• Now, when we call foo() method in line 5, a block in the top of the stack is crteated to be used by foo() method. Since Java is passby value, a new reference to Object is created in the foo() stack block in line 6.
• A String is created in line7, it goes in the String Pool in the heap space and a reference is created in the foo() stack space for it.
• foo() method is terminated in line8, at this time memory block allocated for foo() in stack becomes free.
• in line 9, main() method terminates and the stack memory created for main() method is destroyed. Also, the program ends at this line. Hence, Java Runtime frees all the memory and end the execution of the program.

Java Garbage Collection

Automatic Garbage Collection

The automatic garbage collection mechanism is a process of looking at heap memory, distinguishing between objects in use and unused objects, and deleting unused objects. As for using objects or referencing objects, it means that your program holds a reference to that object. For unused or unreferenced objects, no part of your program holds a reference.Therefore, the memory used by unreferenced objects can be recycled.

In C - like programming languages, memory allocation and collection are manual.In Java, memory collection is handled automatically by the garbage collector.The basic steps can be described as follows:

• Step 1: Marking

The first step is marking, which distinguishes which memory is in use and which is not.

Referenced objects are identified in blue and unreferenced objects in gold. In the marking phase, it will scan all objects and determine. If all objects in the system are to be scanned, this step can be time-consuming.

• Step 2: Deleting normally

Remove the unreferenced objects and leave the referenced objects and the free pointers.

The memory allocator holds references to free memory, which is linked to a List and can be allocated to new objects when needed.

• Step 3: Deleting With Compacting

To further improve performance, in addition to removing unreferenced objects, users can also compress live referenced objects. Moving reference objects together makes it faster and easier to allocate new memory.

Steps of GC (Garbage Collection)

When JVM make GC process, it usually does not apply this mechanism in Tenured Area and Permanent Area. Almost all the GC process plays its role in Eden Area. Thus, GC are divided into two kinds, one is Minor GC and another one is Major GC or (Full GC).

• Minor GC: Only executes in the Eden Area.

• Major GC: Aiming at Tenured Area, constantly Eden Area GC and Permanent GC.

Minor GC

For Minor GC, it will transfer all the living objects to the Survivor Space. If the Survivor Space cannot hold the objects, the remaining objects will be trandfered into the Old Generation. Once the collection has finished, Eden Area becomes empty.

• When the objects are created in the Eden Area(Including a Survivor space, here refer to it as From space), after a Minor GC, if the objects are living, and can be contained in another Survivor Space(here refer to it as To Space, and To Space had enough memory to hold the living objects from Edne Area and From Space),JVM can use copying algorithm to copy these living objects to another Survivor Space(To Space) and clean the used Eden Area and Survivor Space(From Space). Then, the age of living instances will be set to 1. From now on, once the objects make through a Minor GC its age will be increased by 1 and when its age reaches a threshold(Defult 15, can be configured through -XX:MaxTenuringThreshold), these objects can be refered to as Old Generation.

• HotSpot JVM divides the Young Generation into three parts: one Eden Area and two Survivor Spaces(called From Space and To Space). The defult ratio of these parts are 8:1:1. Under normal conditon, the newly-created objects can be allowcated into Eden Area. After one Minor GC operation, if these instances are living, they will be moved to Survivor Space. The objects will increase its age by 1 if it make through a Minor GC, and when the age reaches a certain degree, it will be transfered into Old Generation. Due to the fact that almost all the Objects fade away after its creation(Over 80%), the GC algorithm in Young Generation is Copying. The fundamental thoughts of Copying Algorithm is to divide the memory into two parts, every time only use one of them, when the memory runs out, copy the living objects over another piece of memory. Copying Algorithm does not produce fragment.

• When the Minor GC begins, objects can only exist in the Survivor space named From, and the space in the To are empty. Then, after one GC, the living objects in the Eden Area are all copied into To, while in the From, the living objects decide their next pace according to their age. If their age reach a certain threshold, they will be transfered in to the Old Generation Area, and the ones whose age do not reach the threshold will be copied into To. After this GC operation, Eden Area and From are all cleared, meanwile, From and To will change their character, in other words, the new To (The From before the last GC) will become From(The To before the last GC). No matter under what condition, the space named To will always be kept empty. Minor GC will repete these process untill To is full. After the fulfillment of To, all the objects will be moved into Old Generation.

The efficency of this algorithm is depend on the living ratio of the object. If the ratio is high, the efficiency will be low. The reason is that the utilizing spaces are comsumed by the copying spaces.

Major GC

Major GC is usually the combinataion of Mark-sweep and mark-compact

For Mark-Sweep:

1. Mark Begin with the root collection, mark the living objects.

2. Sweep Scan all the memory space, recycle the unmarked objects, and keep record of this area by using free-list.

For the shortcomings, this method has a low efficiency(Including recursion and traversal). Besides, when going through GC, the program has to stop, which lead to bad experience of users.

In addition, the main disdvantage is that the spaces this method frees is discrete, whichis not confusion. And the fading objects are seperated in every corner of the spaces, after the cleaning, the natural layout of the space will become a mass.

For Mark-Compact:

The only shortcoming of this algorithm is that the efficiency of this method is not high. It not only has to mark all the living instances, but also needs to neaten the reference location of the living objects.

Comparisions:

Memory efficiency:

Copying > Mark-Sweep > Mark-Compact

Memory uniformity:

Copying = Mark-Compact > Mark-Sweep

Memory utilizatino:

Mark-Compact = Mark-Sweep > Copying