Opera;ng Systems
CMPSCI 377
Tim Richards
University of MassachuseIs Amherst
1
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
Today’s Class
§ Organiza)onal mee)ng
– Course organiza)on & outline
– Policies
– Prerequisites & course sign-up
§ Intro to Opera)ng systems
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
2
Work
§ Read Chapter 1
§ W01 Quiz on Moodle
§ Homework 1
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
3
Organiza;onal Informa;on
§ Course web page
– Visit hJp://umass-cs-377.github.io
– Moodle 377 page: to turn in assignments, discussion
forums etc
– Piazza
§ Contact info
– richards@cs.umass.edu
§ Primary TA: Teddy Sudol tsudol@cs.umass.edu
§ Secondary TA: Tianbo Gu tbgu@cs.umass.edu
§ Discussion sec)on (Wednesdays 11:15-12:05)
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
3
Prerequisites and Syllabus
§ CMPSCI 230: computer Systems Principles
§ Textbook: Opera)ng System Concepts
(Silberschatz,Galvin,Gagne) 9th ed
§ Course requirements
– Several homeworks plus several in-class assignments (10%)
– Weekly quizzes (10%)
– Several lab/programming assignments (40%)
– 3 exams (40%) – two midterms and a final
§ Late policies and chea)ng
– Automa)c 48 hour late extension
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
Course Organiza;on: Misc
§ Accounts in the Ed-lab: 30+ Linux-based
PCs
§ Discussion sec)on to help you with Lab
assignments and course concepts
§ Office hours:
– Instructor:
W 10:00 - 11:00a, CS 258 or by app
– TA Office hrs and loca)on:
Teddy - M 11:00-1p, CS 207 (star)ng next week)
Tianbo - F 11:00-1p, CS 207
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
Course Requirements
§ Note: Percentages are subject to revision.
§ Programming projects: 40%
– Strict late policy!
§ In-class exams: 40%
§ Homework & Classwork: 10%
§ Quizzes: 10%
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
6
Exams
§ 3 Exams: 2 Midterms and 1 final exam
§ Midterm exam 1
– Feb 18, 2016 (in class)
§ Midterm exam 2
– Mar 24, 2016 (in class)
§ Final exam
– Apr 29, 10:30a-12:30p, ILC S211
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
8
Labs
§ 3-4 Projects
– Focus on topics covered in the class
§ Projects will use C/C++ (and some Java)
§ May work in groups of 2
§ EdLab Accounts or use your own machine
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
7
Plagiarism
§ Chea)ng includes:
– “Borrowing” code from someone
• This includes reading previous solu)ons
–
–
–
–
–
Giving code to someone (even next year)
Copying code from anyone (including the net)
Hiring someone to write your code
Submiong someone else’s code as your own
Looking at anyone else’s code
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
8
Cell Phone and Laptop Policy
class use policy: Don’t!
Cell phones should be off or on silent alert
Tex)ng is strictly prohibited in class
Laptops and tablets may be used in class:
No email, browsing, facebook, twiJer
during class lectures
§ Laptops allowed in the discussion sec)on
§
§
§
§
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
9
Course Outline & Topics
§
§
§
§
§
Introduc)on
Processes and Threads
Memory Management
Storage and File Systems
Distributed Systems
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
10
What’s An Opera;ng System?
§ Defini)on has changed over years
– Originally, very bare bones
– Now, includes more and more
§ Opera)ng System (OS)
– Interface between the user and the
architecture
– Implements a virtual machine that is
– (hopefully) easier to program than raw
hardware.
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
11
OS: Tradi;onal View
§ Interface between user and architecture
– Hides architectural details
§ Implements virtual machine:
– Easier to program than raw hardware
§ Illusionist
– Bigger, faster, reliable
§ Government
– Divides resources
– “Taxes” = overhead
User-level Applications
virtual machine interface
Operating System
physical machine interface
Hardware
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
13
New Developments in OS
§ Opera)ng systems: ac)ve field of research
– Demands on OS’s growing
– New applica)on spaces (Web, Grid, Cloud)
– Rapidly evolving hardware
§ Advent of open-source opera)ng systems
– Linux etc.
– You can contribute to and develop OS’s!
– Excellent research platorm
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
14
OS Most Notable Features
• Services: The OS provides standard services (the
interface) which the hardware implements.
– Examples: the file system, virtual memory, networking, CPU
scheduling, and time-sharing
• Coordination: The OS coordinates multiple
applications and users to achieve fairness and
efficiency (throughput).
– Examples: concurrency, memory protection, networking, and
security.
• Goal: Design an OS so that the machine is convenient
to use (a software engineering problem) and efficient
(a system and engineering problem).
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
Why Study Opera;ng Systems?
• Abstraction: How to get the OS to give users an
illusion of infinite memory, CPUs, resources, world
wide computing, etc.
• System Design: How to make tradeoffs between
– performance and the convenience of OS abstractions,
– performance and the simplicity of OS design, and
– putting functionality in hardware or software.
• Basic Understanding: The OS provides the services
that allow application programs to work at all.
• System Intersection Point: The OS is the point where
hardware and software meet.
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
Why Study Opera;ng Systems?
Not many opera)ng systems are under development, so you are
unlikely to get a job building an OS. However, understanding
opera)ng systems will enable you to use your computer more
effec)vely. They also serve as an excellent example of system
design issues whose results and ideas you will apply elsewhere.
§ Background: To understand this course you must have a
solid basic understanding of hardware (CPU instruc)on sets,
memory hierarchies, I/O systems, etc.) and solid programming
skills (complex data structures, classes as an encapsula)on
mechanism, etc.)
– Obviously, you cannot understand the implica)ons of how
components intersect without understanding the components.
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
Build Large Computer Systems
§ OS as an example of large system design
§ Goals: Fast, reliable, large scale
§ To build these systems, you need to know
– Each computer:
•
•
•
•
•
Architectural details that maJer
C and C++ (niJy griJy & more)
Memory management & locality
Concurrency & scheduling
Disks, network, file systems
– Across cluster:
• Server architectures
• Distributed compu)ng, file systems
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
18
History of Opera;ng Systems
§ And now, for some historical context
– From mainframes to web-based systems
in nine slides
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
19
1. Single-User Computers
§ Hardware: expensive; humans: cheap
§ One user at a )me on console
– Interac)ng with as program runs
§ Computer executes one func)on at a )me
– No overlap: computa)on & I/O
§ User must be at console to debug
§ Mul)ple users = inefficient use of machine
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
20
2. Batch Processing
§ Execute mul)ple “jobs” in batch:
–
–
–
–
Load program
Run
Print results, dump machine state
Repeat
§ Users submit jobs (on cards or tape)
§ Human schedules jobs
§ Opera)ng system loads & runs jobs
§ More efficient use of machine
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
21
3. Overlap I/O and Computa;on
§ Before: machine waits for I/O to complete
§ New approach:
– Allow CPU to execute while wai)ng
– Add buffering
• Data fills “buffer” and then output
– and interrupt handling
• I/O events trigger a signal (“interrupt”)
§ More efficient use of machine
– s)ll one job at a )me
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
22
4. Mul;programming
§ Several programs to run simultaneously
– Run one job un)l I/O
– Run another job, etc.
§ OS manages interac)ons
– Which jobs to run (schedule)
– Protects program’s memory from others
– Decides which to resume when CPU available
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
23
OS Complexity
§ Increased func)onality & complexity
§ First OS failures
– Mul)cs (GE & MIT):
announced 1963, released 1969
– OS/360 released with 1000 known bugs
§ Need to treat OS design scien)fically
§ Managing complexity becomes key to…
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
24
The Renaissance (1970’s)
§ Hardware: cheap; humans: expensive
§ Users share system via terminals
§ The UNIX era
– Mul)cs:
• army of programmers, six years
– UNIX:
• three guys, two years
• “Shell”: composable commands
• No dis)nc)on between programs & data
§ But: response )me & thrashing
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
25
Industrial Revolu;on (1980’s)
§ Hardware very cheap;
§ humans expensive
§ Widespread use of PCs
– IBM PC: 1981, Macintosh: 1984
§ Simple OS (DOS, MacOS)
– No mul)programming,
concurrency, memory protec)on, virtual memory, …
– Later: networking, file-sharing, remote prin)ng…
– GUI added to OS (“WIMP”)
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
26
The Modern Era (1990’s-now)
§ Hardware cheap; processing demands
increasing
§ “Real” opera)ng systems on PC’s
– NT (1991); Mac OS X; Linux
§ Different modali)es:
– Real-)me: Strict or loose deadlines
– Sensor/Embedded: Many small computers
– Parallel: Mul)ple processors, one machine
– Distributed: Mul)ple networked processors
• Think P2P, the Web, Google, cloud
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
27
Architectural Trends
§ Big Changes
– In 50 years, almost every computer
component now
9 orders of magnitude faster, larger, cheaper
1983
1999
0.5
500
$100,000
$500
memory
1 MB
1 GB
network
10 Mbit/s
1 Gb/s
1 GB
1 Tbyte
examples
Millions of
Instructions
Per Second
MIPS
cost/MIPS
disk
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
28
History Lesson
This degree of change has no counterpart in any other area of
business.
Examples:
§ Transporta)on -- over the last 200 years, we have gone from
horseback (10 miles/hour) to the Concorde (1000 miles/hour) 2 orders of magnitude.
§ Communica)on -- at the inven)on of the telephone (voice), TV
(video) and fax (text \& pictures), communica)on went from the
speed of transporta)on to nearly the speed of light - 7 orders of
magnitude.
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
Orders of Magnitude
§ 10^0
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
30
Orders of Magnitude
§ 10^1
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
31
Orders of Magnitude
§ 10^2
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
32
Orders of Magnitude
§ 10^3
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
33
Orders of Magnitude
§ 10^4
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
34
Orders of Magnitude
§ 10^5
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
35
Orders of Magnitude
§ 10^6
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
36
Orders of Magnitude
§ 10^7
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
37
Orders of Magnitude
§ 10^8
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
38
Orders of Magnitude
§ 10^9
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
39
Coming Soon
§ Moore’s Law – running out of steam
§ New “features” coming
– Mul)ple cores
– Unreliable memory
– Serious power/heat constraints
§ Other tradeoffs possible
– Compu)ng power for reliability…
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
40
What’s an OS? Bill Gates says…
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
12
What is important to you?
§ From your perspec)ve, think about the
following:
– What is an opera)ng system?
– What is the most important aspect of an
opera)ng system (according to your
defini)on) and why?
§ Take a moment to think about this…
§ Then, write a short paragraph answering
these ques)ons. 8-digit spire ID and name
at the top please. Hand in at end of class.
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
43
Perspec;ves
§ We look at opera)ng systems from…
§ The User View
– Ease of use (mostly)
– Performance (some)
§ The System View
– Resource Alloca)on
– Program Execu)on
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
44
OS Defini;on
§ An OS covers many roles and func)ons.
§ They are ubiquitous in our every day lives
– sure: laptops, desktops, phones, etc.
– but: cars, toasters, refrigerator, homes, medical
equipment, etc.
§ Defini;on?
– In general, we have no adequate defini)on of an
opera)ng system.
– They exist because they offer some reasonable
way to solve the problem of crea)ng a usable
“compu)ng system”.
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
45
So, what is an OS?
§ Huge Range
– Big (Windows, Mac OSX, Your Car)
– Small (Wearable Computers)
§ We will follow this common def. of an OS:
– A program that constantly executes: kernel
– Libraries that help the kernel: middleware
– Programs that help the OS: system programs
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
46
Computer System Organiza;on
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
47
Computer System Organiza;on
Kernel
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
48
Bootstrap Program
§ How does an OS come to life?
§ The Bootstrap Program
– A small program typically stored in computer
hardware (ROM)
– Ini)alizes computer hardware
– Knows where to “find” the instruc)ons for
the opera)ng system kernel, loads part of it
into memory, and hands over control.
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
49
System Processes
§ OS Startup
– Kernel is loaded into memory
– Begins execu)on
– Starts system processes (daemons)
these run in the “background”
§ Then, we wait…
– The OS waits for an event to occur
– When an event occurs, the OS will step in to
handle the event.
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
50
Events!
§ How do we know when an event happens?
– An interrupt is triggered!
– This originates from either hardware or
so~ware
§ Hardware
– May trigger an interrupt at any
)me by sending a single to the
CPU (via system bus)
§ Sohware
– May trigger an interrupt with a system call
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
51
Interrupt Timeline
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
52
Interrupt Timeline
Process makes I/O
Request
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
53
Interrupt Timeline
Sends trap to indicate
completion.
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
54
Interrupt Timeline
Process suspends.
Transfers I/O to memory
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
55
Interrupt Timeline
Process resumes
execution
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
56
Interrupt Timeline
Another I/O request
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
57
Interrupt Timeline
Transfer done, signals
completion
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
58
Interrupt Timeline
Process suspends,
Handles I/O
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
59
Interrupt Timeline
Execution resumes
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
60
Mul;programming
§ Modern Opera)ng Systems
– Allow mul)ple programs/processes to
execute concurrently.
– Interrupts are key to )me sharing between
processes.
– Events are core to an OS opera)on.
§ The Web
– Also event driven!
– Is the web an opera)ng system?
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
61
Next Time
§ OS and Computer Architecture
UNIVERSITY OF MASSACHUSETTS AMHERST • College of Informa;on and Computer Science
62