edu(b)log

Resource Handling

Question: What is the difference between a program and a process?

Answer: Processes are created from a program in execution. In order to create a process from a program you need the metadata about the program and its location.

There are tools to examine executables, including PE Explorer from Heaven Tools (Windows® only).

We won't focus on the compiler or the linker, only the topics most closely related to the Operating System.

CPU Resources

The CPU is the most important resource within the computer system. Others include main memory, cache, and hard disk (and other I/Os).

The amount of time a scheduler gives to a certain process is measured in milliseconds. If the job cannot be finished it will be kicked out of the CPU and given another chance sometime later.

The loader is in charge of loading a program into main memory as a process. The number of processes within the system is limited so that resources can be devoted to the most important processes. The dispatcher feeds the next process to the CPU.

Preemptive & Nonpreemptive Scheduling

Preemptive processes add additional overhead because they remain in memory. However, it is usually better because it can lead to improved response times, can be removed from their current processor, and are important in interactive environments.

Nonpreemptive processes run until completion or until they yield control of a processor. They are not great because unimportant processes can block important ones indefinitely.

Scheduling Objectives

• Maximize throughput
• Maximize the number of interactive processes receiving acceptable response times
• Minimize resource utilization
• Avoid indefinite postponement
• Enforce priorities
• Minimize overhead
• Ensure predictability

	CPU	FIFO Waiting	RR Wait	RR TA	Response Time
P1	10 sec	3 sec	17 sec	27 sec	17 sec
P2	20 sec	0 sec	0 sec	40 sec	0 sec
P3	15 sec	20 sec	40 sec	55 sec	40 sec
P4	30 sec	15 sec	15 sec	45 sec	15 sec

You should be able to construct a process timeline from the above table. Also notice that the RR Wait and Response Time columns are equal.

OS Evolution and Concepts

Programs must be submitted through the Olympus server (olympus.acs.ucf.edu) with your PID as the username.

Bath operating systems used batches of punch cards for each program, and had no interaction with the user (non-interactive computing).

Remember (from Computer Architecture) how a simple modern computer executes instructions:

From where are operations fetched? The instruction pointer (IP) a.k.a. program counter (PC) points to the next instruction.

The CPU is usually the fastest part of a computer, followed by cache memory, main memory (10x slower), and hard disk (100,000x slower). So I/O operations take a tremendous amount of time.

OS Evolution and Concepts

Solution: Multiprogramming. Have more than one active (running) program in memory at any one time. This is opposed to sequential execution, overlapping CPU and I/O operations. Different processes compete for a common resource that need to be read or modified concurrently. Dynamic memory management is also introduced, which handles swapping and paging.

Interactive OSs: Time-sharing (quantum) → Reduce the response time

Caching

There are different types of memory caching systems. To prevent a user from accessing the Operating System area of a cache and privileged instructions, you utilize a fence register.

Why don't we allow applications to access I/O devices directly? If you allow this, the same buffer used to receive packets from the network opens up malicious users to all I/O devices.

What if the environment is really safe (e.g. scientific computing with limited access)? You can remove a lot of overhead for large amounts of data and scientific applications if the user is allowed to access I/O devices directly.

TSR Programming

TSR stands for Terminate and Stay Resident. Early C programmers wrote TSR programs (1980s). TSR was a "trick" to override some parts of the Operating System so that more than one program could be used at a time. Some interrupt handlers are rarely used so they were "replaced" with the user's own code. This was useful to run anti-virus software while other programs were running.

In the 1980s the typical memory size was 640kb. The operating system was less than 320kb, but people were still struggling with having enough memory to run their programs.

The 1990s

Microsoft Corporation became dominant with their Windows® operating systems. Applications began to be developed mainly in object-oriented operating systems. Most commercial software is sold as object code (binaries).

Free and open source software became increasingly common (distributed with source code).

Richard Stallman launched the GNU project to recreate and extend tools for AT&T's UNIX operating system. *GNU is Not Unix*

Polling — actively sampling the status of an external device by a client program as a synchronous activity.

What processes benefit from Interrupt Handling? I/O operations and time sharing (e.g. Multics) utilize interrupt handling.

Operating System Environments

Real-time Systems

Require that tasks be performed within a particular (often short) time frame. The autopilot feature of an aircraft must constantly adjust speed, altitude, and direction. Air traffic control, medical patient monitoring, power plant control are much more demanding than an OS like Unix or Windows®.

Real-time operating systems support operations that cannot wait indefinitely, and sometimes cannot wait at all.

Virtual Machines

Virtual Machines (VMs) are used to virtualize a CPU. A VM is software that creates an environment between the computer platform and the end user in which the end user can operate software.

Monolithic Operating System

Every component is contained in the kernel. It is not modularized or well-separated, but it tended to be highly efficient.

The Role of Operating Systems

Fewer than 20% of commerce websites report no unauthorized use, and the number of hacking attempts reported over the past 20 years has increased exponentially. This indicates that operating systems are not reliable (hacker-proof).

The Blaster worm filled the UCF Computer Science network with garbage packets and nearly took the network down.

Zero-Day Attack: The estimated time required for a near-total infection of the world's computers is approximately 30 seconds.

Worms are more active than viruses; they self-duplicate and don't depend on any other hosts. Worms are independent and duplicate on their own.

There is approximately 1 serious error per 1,000 lines of code in unaudited code (20 pages) and 1 error per 5,000 lines of code in examined code (100 pages).

OpenBSD 2.6 contains 1,874,000 lines implying 375 faults; Windows 2000 has 30,000,000 lines of code implying 6,000 serious faults.

On-going problems (via Intel): 50-60 incidents per day on the Internet, 10-12 incidents per day on DSL, 5-6 incidents per day on dial-up.

Typical Users have less than one year computing experience, no computer background, 1GHz machine using 3 applications w/ MS Windows, doesn't make backups, doesn't apply new patches, and is online constantly.

Course Overview

• Introduction — What is an operating system and its structure?
• Hardware/Software Concept — Hardware components, software overview
• Process and Threads — Process/Thread concepts and management
• Asynchronous Concurrent Execution — Mutual exclusion and semaphores
• Concurrent Programming — Condition variables and monitors
• Deadlock — Deadlock problems and solutions

All modern operating systems handle multiple programs concurrently; the OS creates the illusion that you can run multiple programs on one CPU while you'll never actually run two processes at the same time on one CPU.

» Midterm Exam Question: What is an operating system?

The OS is an extended machine (OS provides high level operations using lower level operations), a virtual machine (OS provides virtual CPU memory, devices), and a resource manager (efficient use of shared resources). The main job of the OS is hiding the complexities of the hardware, making the application programs more portable. The OS creates the illusion of multiple copies of the CPU as if all memory and all I/O devices are devoted to the current application.

The Role of OSs

The Operating System fills the gap between applications and the hardware level. In other words, it is is the system software responsible for the direct control and management of hardware and basic system operations. Additionally, it provides a foundation upon which to run application software.

The compiler bridges the gap between machine instructions and high level languages, and between linear memory and data structures.

The OS bridges the gap between limited CPU and memory vs. more needed, secondary memory devices vs. files, and I/O devices vs. high level I/O commands.

The bus connects main memory, CPU, Secondary Storage, and Communication Devices. The delay caused by a bus is minimal.

The ObjectWeb consortium gives the following definition of middleware: "In a distributed computing system, middleware is defined as the software layer that lies between the operating system and the applications on each site of the system."

Intro to Operating Systems (History)

The syllabus is online at http://www.cs.ucf.edu/~jlee/cop4600/syllabus.htm.

The course website is: http://www.cs.ucf.edu/~jlee/cop4600/index.html

For the lecture materials & programming assignments:

username: student
password: iloveos

They will be using the UNIX operating system for grading, so you should make sure that your code compiles and runs natively in UNIX.

The contents of this class have changed for this semester, emphasizing security issues within operating systems (there will be 3-4 lectures on this subject).

• 20% Midterm Exam 1
• 20% Midterm Exam 2
• 20% Final Exam (not comprehensive)
• 25% Programming Projects (5 assignments, 5% each)
• 15% Homework Assignments

No late submissions on homework.

Computer & Operating System History

About 40-50 years ago, you had hardware without operating systems, so in order to run a program you had to have a complete understanding of the hardware and write the application via switches.

Over time, many vendors have started using different types of machines with different I/O devices, so we need an interface between the different hardware devices and the applications we wish to run on it: operating systems.

Check the lecture notes for dates and additional important events.

What is virtual memory? The Operating System uses virtual memory to run programs whose size exceeds that available in the hardware. This is system memory that is simulated by the hard drive. When all the RAM is being used (for example if there are many programs open at the same time) the computer will swap data to the hard drive and back to give the impression that there is slightly more memory.

What are System Calls? The Operating System is an interface between applications and hardware. System calls are the access point through which applications interact with the operating system. Spyware, viruses, and worms are exploiting the fact that the OS interacts with the Internet through system calls by replacing important system calls with their own pointers.

In 1964, PL/1 was chosen as a programming language. From 1965-1969 Bell Lab, GE, and MIT developed the Multics operating system, a project which was funded and cancelled by the US army.

Click here for a history/timeline of UNIX history.

He went a lot into the histry of operating systems and OS names, so if you want to study that you should check out his PowerPoint slides — no reason to copy it all here.

At the end of the semester we want to know what to do if we want to develop a new operating system.