cs357 Lecture Notes
Spring 2000
Week 2, Tuesday

The reading for today was Chapter 1, and you should have
prepared answers to questions
1.1, 1.5, 1.8, 1.11, 1.12

Reading for Thursday, 2.1 through 2.4
Prepare an answer for Question 2.4

Warning: Chapter 2 is not a great read -- the level of
	 detail is all over the map.  Good chance to
	 practice hierarchical reading.

Review
------

Last time, I gave an overview of the academic and
administrative structure of the class, and I answered
the question...

What is an operating system?

An operating system is a software layer between application
programmers and hardware.

It is largely defined by:

1) the API, or virtual machine interface

2) the hardware interface

3) the performance consequences of its implementation


Outline
-------

1) questions for today
2) the story hidden in Sections 1.2 - 1.5
3) intro to Chapter 2


Questions
---------

1.1  What are the three main purposes of an OS?

1.5  What is the main advantage of multiprogramming?

1.8  The books stresses the need for an OS to make efficient
     use of the computing hardware.  When is it appropriate for
     the OS to forsake this principle and to "waste" resources.
     Why is such a system not really wasteful?

1.11 Why are distributed systems desireable?

1.12 What the main difficulty that a programmer must overcome
     in writing an OS for a real-time environment?


History of OS
-------------

1.2  Batch systems

OS shares physical memory with one process.

Pro:  simple

Con: lots of idle time because loading is much slower than CPU


1.3  Multiprogramming

OS shares physical memory with multiple processes.

Pro: overlap I/O with CPU on coarse scale -- better CPU utilization

Con: complications like interprocess protection


1.4  Time-sharing systems

As computers get cheaper and people get more expensive,
human utilization becomes more important than CPU utilization.

Support interaction by multitasking -- switching between
processes on a smaller time scale.

Use virtual memory so the total demand can exceed the limits
of physical memory.

Pro: nifty

Con: complicated


1.5  Single user systems

Now computers are very cheap, humans very expensive.

If each computer has a single user,

We don't need multitasking, right?
We don't need inter-process protection, right?
We don't need virtual memory, right?

(Also, these machines were usually not connected to networks,
so their OS often lacked network capability)

WRONG!

Need multitasking for responsiveness (not CPU utilization).

Need protection for reliability and security.

     (for n components, interactions grow as n^2.
     How many inits does a typical Mac boot?)

Need virtual memory to implement protection (not for performance).


Problem:  Hard to add these features as an afterthought.

They need to be central to the design of the OS, and they
need a little help from the hardware.

The history of MacOS and Windows shows how hard it is to
back into these features.

But if you know how to do it, it's not that hard.
The kernel of Linux was written by one person in six months.


Chapter 2
---------

2.1 System model

2.2 I/O

2.3 Storage (memory and disk)

2.4 Memory hierarchy

2.5 Implementation of protection
    dual-mode instructions
    I/O interface
    memory
    cpu


System model:  for now, just a CPU connected to memory
and a disk drive by a bus.

In real systems, there is usually one bus for memory and
one bus for everything else, bit lets keep things simple
for now.


Section 2.2
-----------

How does data get from the CPU to somewhere else.

1) CPU puts signal onto bus to identify the target device,
   then puts a block of data on bus.

   Block size varies depending on the device.

   Data winds up in the controller for the device, which is
   hardware that includes at least a little memory.

   Controller does its thing.

2) CPU might wait for the controller to finish before
   proceeding (synchronous I/O) or go off and do something
   else (asynchronous).

   What kind of something else can it do?

3) If the operation produces a result, the CPU might have to
   poll (ask repeatedly) to see if it is done, or the

   device might generate an interrupt when it is done.

4) If the destination of the result is memory, we have another
   choice.  The CPU can explicitly move the data from the
   device into memory, or the device might move it into
   memory without interrupting the CPU.  The latter is called
   DMA -- direct memory access.