A running program is a collection of “threads” – essentially independent execution streams...
-
Upload
berenice-greene -
Category
Documents
-
view
213 -
download
0
Transcript of A running program is a collection of “threads” – essentially independent execution streams...
Note 5Threads
SE2811 Software Component DesignDr. Rob Hasker (based on slides by Dr. Mark Hornick)
What SE1011 students are told…
• When the main() method is called, the instructions within the method begin to execute in sequence• The program terminates when the main() method finishes executing• But is this really true?
The ugly truth…
• A running program is a collection of “threads” – essentially independent execution streams•When the main() method is called, the instructions within the method begin to execute in sequence on a primary thread• The program terminates when the primary thread, and any additional threads, finish executing
What’s a Thread?
Defining Process: A process is most easily understood as a program or application running on your PC
A process generally has a complete, private set of basic run-time resources, in particular:• Its own memory space• Execution priority• A list of threads that execute within it• A set of credentials with which to execute (usually
yours)• These provide authorization to access various resources
such as files
By default, a Process creates and executes a single, primary Thread
BUT:A process can create and execute more than one thread
The JVM works with the OS to create processes and threads• The underlying OS provides the
essential multiprocessing support
Modern systems: multiple processes run simultaneously
• (On single-CPU PC’s) each process runs individually for a discrete time period
while one process runs, other processes sleep
• The process currently executing changes very rapidly - every few milliseconds
Operating systems use a scheduler to distribute CPU time among processes
• The net effect is that you (the user) observe all processes running simultaneously and continuously
Java application: the JVM creates a Process and a Primary Thread
• The primary thread begins executing the main() method in the main class
Note: other Java programs, like applets, begin execution with an init() method
• If no other threads are created, the process terminates when the primary thread terminates
That is, when there are no more instructions to execute on that thread
Threads wind their way through the code until they run out of instructions to execute
public class App {
public static void main(String[] args) {
App me = new App();
me.method_A();
}
private void method_A() {
// more code here
method_B();
return;
}
private void method_B() {
return;
}
private void method_C() {
// more code here
}
}
Single-threaded application
Where do other threads come from?
1. Additional threads are created by a Swing-based application
Java applications that create and display windows cause Swing to create additional threads
2. Additional threads are created by various Java utility classes
Eg: the Timer class
3. Threads can be created, controlled explicitly
Creating a multi-threaded application using Swing
1. Create a JFrame window containing JButtons, JTextField, etc.
2. Connect the JButtons, etc. to an ActionListener
3. Make the window visible• Once the window is visible, a second thread is
created• All calls to actionPerformed() occur on the
second thread• The Event-Dispatching Thread
Using a Timer is fairly straighforward:
import javax.swing.Timer;
Timer timer = new Timer(timeoutPeriod, eventHandler);
timer.start();
The eventHandler argument to the constructor is a reference to a class that implements Timer ActionListener• That is, eventHandler contains an actionPerformed()
method. This is similar to how Swing events are handled
• Whenever the Timer generates a timeout event, the JVM invokes actionPerformed() on another thread• JVM uses the Event Dispatch thread when available; otherwise a
“worker” thread is created
Explicitly creating additional threads
Thread t = new Thread(r);
t.start();
The r argument to the Thread constructor is a reference to a class that implements the Runnable interface• Runnable declares a single method: public
void run()
When the Thread’s start() method is called, the instructions in the run() method begin executing on the new thread.start() returns essentially immediately; it does not wait for the
started thread to finish execution
The main class may implement the Runnable interface itself:
public class App implements Runnable {
public static void main(String[] args) {
App me = new App();
me.method_A();
}
private void method_A() {
// more code here
method_B();
return;
}
private void method_B() {
Thread t = new Thread(this); // App is runnable!
t.start(); // start executing the run() method
return;
}
public void run() {
// more code here
}
}
public class App implements Runnable{
public static void main(String[] args) {
App me = new App()
me.method_A();
}
private void method_A() {
// more code here
method_B();
return
}
private void method_B() {
Thread t = new Thread(this);
t.start();// execute run() on new thread
return;
}
public void run() {
// more code here
}
}
Both threads execute simultaneously and independently after the secondary thread is started
The secondary thread may execute a method defined in another class that implements Runnable
public class App {
public static void main(String[] args) {
App me = new App();
me.method_A();
}
private void method_A() {
ThreadRunner tr = new ThreadRunner();
Thread t = new Thread(tr);
t.start();
return;
}
private class ThreadRunner implements Runnable { // inner class
public void run() {
// more code here
}
}
}
The secondary thread may execute a lambda expression
public class App {
public static void main(String[] args) {
App me = new App();
me.method_A();
}
private void method_A() {
ThreadRunner tr = new ThreadRunner();
// assume declared x, y, z here...
Thread t = new Thread(()-> method_to_run(x, y, z));
t.start();
return;
}
private void method_to_run(int a, int b, String q) {
System.out.print(a + b);
System.out.print(q);
}
}
This method gets executed in place of the run (when
t.start() is executed)
An application may be designed to execute the same instructions on more than one thread at the same time
public class App implements Runnable {public static void main(String[] args) {
App me = new App();me.method_A();
} private void method_A() {
ThreadRunner tr = new ThreadRunner();Thread t = new Thread(tr);t.start(); // execute run() on new secondary threadmethod_C(); // execute method_C on the primary
thread}private void method_C() {
// more code here } public void run() {
// some other instruction is here method_C(); // execute method_C on the secondary thread}
}
Question:
Is it a good idea to let two (or more) threads execute the same code at the same time?
See https://faculty-web.msoe.edu/hasker/se2811/samples/
Fortunately, Java supports several mechanisms for synchronizing the execution of multiple threads
S
Keeping code thread-safe
The Thread class’s join() method is one way of synchronizing two threads:
Thread t = new Thread(cref);
t.start();
t.join(); // wait for 2nd thread to finish…
method_C(); // …before executing next inst.
• The second Thread starts executing the instructions in the run() method.
• The current thread (the launcher of the second thread) waits until the second thread finishes
If a method is declared to be synchronized, it will only be run on a single thread at a time
public class App implements Runnable {public static void main(String[] args) {
App me = new App()me.method_A();
} private void method_A() {
ThreadRunner tr = new ThreadRunner();Thread t = new Thread(tr);t.start();method_C(); // run method_C on the primary thread
}private synchronized void method_C() {
// More code here } public void run() {
// some other instructions here method_C(); // run method_C on the secondary thread}
}
Once a thread enters a synchronized method, no other threads can enter until the first thread completes execution of the method, and exits the method.
private synchronized void method_C() {
<statement 1>
<statement 2>
<statement 3>
}
Thread x Thread y Thread z
When Thread x leaves the method, the Scheduler arbitrarily allows one of the other threads to enter the method. When the second thread exits, the Scheduler allows another waiting thread to enter.
If all threads get to the method at the same time, the thread that gets to enter the method first is determined arbitrarily by the Scheduler.
Once a thread enters a synchronized method, no other threads can enter until the first thread completes execution of the method, and exits the method.
private synchronized void method_C() {
<statement 1>
<statement 2>
<statement 3>
}
Thread x Thread y Thread z
When Thread x leaves the method, the Scheduler arbitrarily allows one of the other threads to enter the method. When the second thread exits, the Scheduler allows another waiting thread to enter.
If all threads get to the method at the same time, the thread that gets to enter the method first is determined arbitrarily by the Scheduler.
Add syncronized to count_up()
Is synchronized the solution to everything?
Can you think of any disadvantage to making a method synchronized?
If only a few statements within a method need to be guarded against simultaneous execution, use a synchronized block instead of making the entire method synchronized.
private void method_C() {
<safe statement 1>
<safe statement 2>
synchronized( <sync_object> ) {
<unsafe statement 3>
<unsafe statement 4>
<unsafe statement 5>}
<safe statement 6>
}
Thread x Thread y Thread z
When Thread x leaves the block, the Scheduler arbitrarily allows one of the other threads to enter.
The synchronizing object can be any object
• Java’s Object class incorporates the concept of something called a monitor
•Monitors are used to guard the gates of synchronized blocks
•Monitors only become active within a synchronized block
Since every class derives from Object, the class containing a synchronized block can act as the Monitor for the block:
private void method_C() {
<safe statement 1>
<safe statement 2>
synchronized( this ) { // gate down
<unsafe statement 3>
<unsafe statement 4>
<unsafe statement 5>} // gate up
<safe statement 6>
}
Thread x Thread y Thread z
Or any generic Object can act as a Monitor
private Object guard = new Object();private void method_C() {
<safe statement 1>
<safe statement 2>
synchronized( guard ) { // gate down
<unsafe statement 3>
<unsafe statement 4>
<unsafe statement 5>} // gate up
<safe statement 6>
}
Thread x Thread y Thread z
Consider the following code. Suppose all threads reach the for() loop simultaneously.
How do the threads compete to run the for() loop?
private Object guard = new Object();private void method_C() {
<safe statement 1>
<safe statement 2>
for( int i=0; i<100; i++ ) {
synchronized( guard ) { // gate down
<unsafe statement 3>
<unsafe statement 4>
<unsafe statement 5> } // gate up
} // end for
<safe statement 6>
}
Thread x Thread y Thread z
After each thread executes the synchronized section, it can notify the Monitor that another thread can be allowed to enter the synchronized block as soon as it relinquishes ownership of the synchronized section by entering a wait (or exiting the synchronized section)
private Object guard = new Object();private void method_C() {
<safe statement 1>
<safe statement 2>
for( int i=0; i<100; i++ ) {
synchronized( guard ) { // gate down
<unsafe statement 3>
<unsafe statement 4>
<unsafe statement 5>
guard.notify(); // signal waiting threads
guard.wait(); // wait for other threads }
} // end for
<safe statement 6>
}
Thread x Thread y Thread z
Review• Process: running application• memory, priority, permissions, threads
• Thread: execution stream• Essentially: a stream of instructions executed by CPU in a
process context
• Creating threads• Java Swing or JavaFx, Timer, new Thread()
• A thread of a problem• What happens if two threads access the same memory at
the same time• Using synchronized to avoid the problem