Transcript OS2-Warp-Spr-2001-sect-1-group

Operating System/2 Warp
Shawn Ortiz
John McAveney
Andrew Bates
Paul Graf
Kevin Tougher
History of OS/2
• Developed by IBM in collaboration with Microsoft
– OS/2 originally developed to take the place of DOS
– Wanted to build an OS with multitasking
• 1987 OS/2 1.00
– 1st OS with built in multitasking based on hardware
support for a PC
– One program displayed on screen at a time, others
running in the background
– command line interface
History of OS/2 continued
• 1988 OS/2 1.10 Standard Edition
– Graphical User Interface
– Supported FAT hard drives
• 1988 OS/2 1.10 Extended Edition
– Database Manager and Communications Manager
• 1989 OS/2 1.20 SE and EE
– High Performance File System
History of OS/2 continued
• 1990 IBM and Microsoft split
– IBM took control of all OS/2 versions 1.X and OS/2 2.00
– Microsoft continued development of Windows and
OS/2 3.00, which it renamed Windows NT
• 1991 OS/2 1.30 SE and EE
– More device drivers
– More fonts
– Improved swapping algorithm to enhance performance
History of OS/2 continued
• 1992 OS/2 2.00
– 1st true 32 bit operating system for the PC
– Virtual DOS Machines
– Workplace Shell: Object Oriented User Interface
• 1993 OS/2 2.11
– For users with Windows 3.1 already installed
– Used existing version of Windows 3.1 to run Windows
programs
History of OS/2 continued
• 1994 OS/2 Warp 3
– 1st OS for the PC with Internet support built in
• Web Explorer, Ultimail, Graphical and Textual FTP
• 1995 OS/2 Warp Connect
– Combined features of Warp 3 with network connectivity
and tools
• 1996 OS/2 Warp Server
– Combined Warp 3 with IBM’s LAN Server 4.0 product
History of OS/2 continued
• 1996 OS/2 Warp 4
– VoiceType Navigation
• Requires no additional software
– Built in Java
• Requires no additional software
– IBM calls it “Universal Client” because of unparalleled
network connectivity.
• Able to connect to LAN Server, Warp Server, Windows NT
Server, Novell NetWare, Windows 95, Warp Connect, FTP,
SLIP, PPP, NetBIOS and many more
Some Commercial Users of OS/2
•
•
•
•
•
•
•
•
AT&T
First Union Bank
Wal-Mart
Audi
Toyota
Mitsubishi
Sony
Starbucks
•
•
•
•
•
•
•
•
Sprint
Nations Bank
USMC
US Navy
GEICO
Ford
FAA
Delta Airlines
Task Scheduling
• Task management or scheduling is one of the
primary functions of the OS/2 Warp operating
system.
• The OS/2 kernel manages the execution of all
tasks running on the system.
• The scheduler allocates CPU time to each process
based on its priority and whether it is capable of
running or not.
Characteristics of the OS/2 Warp
Operating System:
•
•
•
•
Priority Based Multitasking System
Dynamic
Preemptive
Uses Round-robin Scheduling
Priority Based Multitasking
System
• OS/2 Warp supports user-level prioritization.
• In the OS/2 Warp operating system, each task is
assigned a priority.
• The priority of each task can be determined
directly by the programmer, or, if no priority is
assigned, the operating system assigns a default
priority.
Priority Based Multitasking
System (cont.)
• Programs that have higher priorities, and are ready
to run, are given access to the processor before
programs with lower priorities.
• All other tasks must wait until the higher priority
task becomes blocked before they may have CPU
time.
Priority Levels
• OS/2 Warp supports 128 priority levels.
• These are divided into four classes, each with 32
sublevels.
• The priority values within the sublevels range
from 0 – 31, with higher numbers representing
higher priorities.
• Threads are also grouped into priority classes.
Priority Classes
The four priority classes that determine how the
processor schedules a thread’s execution are, ranked
from highest to lowest priority:
1.
2.
3.
4.
Time-critical (priority 3)
Server (priority 4)
Regular (priority 2)
Idle-time (priority 1)
Time-critical (Priority 3)
•
•
Programs that need to have access to the CPU
very quickly
Time critical tasks take a very small portion of
the CPU's time. Even after all time critical tasks
receive all the CPU time they require, there is
still plenty of time left for all the other tasks
running on the computer.
Examples:
–
–
Communication and Networking tasks
Real-time robotics applications
Server (Priority 4)
•
Server class programs are used by Warp Server.
Example:
–
–
Used to process requests for data by client
workstations and to get the data requested ready to
transfer across the network.
Processing these types of tasks at the Server class
ensures that administrative tasks such as creating new
user IDs do not interfere with them.
Regular (Priority 2)
•
•
This is the default priority class assigned to tasks
when the programmer hasn’t directly specified a
priority.
The Regular class is the priority class in which
most application programs run.
Idle-time (Priority 1)
•
•
When a task is assigned an Idle-time priority, it
only receives CPU time after all other higher
priority tasks have become blocked.
Most of the time the processor is doing very
little, if anything, so there is lots of time for idle
class programs to run.
Scheduling Thread Execution
To schedule a thread’s execution:
1. The system first checks the thread’s priority
class.
2. The system then checks the priority value that
was assigned to that thread.
3. The thread with the higher priority will then
be executed.
Round-robin Scheduling
• Because many tasks will run at the same priority
level, round-robin scheduling is used.
• Round-robin scheduling provides regular time
slices of CPU time to tasks running at the same
priority level. This limits each task to a short burst
of processor time.
• When a task has taken the maximum time slice, or
when a higher priority program is ready to run, the
OS/2 Warp task dispatcher preempts the task and
rotates to the next ready process.
OS/2 Warp is Dynamic
• Each thread receives a priority when it is created
(it inherits the priority of the thread that started it).
• The operating system can change the priority level
of the thread based on the dynamic conditions to
the user’s environment.
OS/2 Warp is Dynamic (cont.)
Example:
– it is possible for some lower priority tasks to suffer
starvation for CPU time.
– When a task becomes starved for CPU time, the OS/2
task scheduler can boost the priority of the starved task
by one priority level to make sure each task receives at
least some CPU time.
– After the task receives its time slice, it is reduced to its
base priority.
Partitioning
• Way of dividing a single physical disk into two or
more “disks”
– lose some parallelism
• OS/2 Warp uses either FAT or HPFS partitioning
schemes
– uses FDISK to manage and create partitions
– Up to four primary partitions, and no logical partitions;
– Up to three primary partitions, and any number of
logical partitions.
FAT Partitioning
• block size is 512 Bytes
• file sizes rounded up to the next integral block size
• only allows for small hard drives, have to use
clusters to extend size of manageable drive
• file sizes rounded up to nearest integral cluster
size
– makes for large files
FAT Partitioning (cont’d)
• directories are arranged in linear fashion
– if the array gets too large, it can be a pain to
search through
• for space allocation, it finds the first
available block
– increases disk fragmentation
The HPFS
• block size still 512 Bytes
• cluster size is always equal to the physical
drive’s size, irregardless of how large it is
– less wasted space
• don’t have to worry about making partitions
too large
• directories are arranged in tree structure
– makes searching faster and easier
The HPFS (cont’d)
• space allocation
• makes sure block is physically close to it’s
directory
– decreases fragmentation
– never reaches level that is noticeable
HPFS Advantages
• longer file names (254 characters in length)
• higher performance
– FAT decreases in performance as drive size goes up
• HPFS allocates sectors only
– much less wasted space
– much better granularity
• less fragmentation
• incompatible with windows 95/98/NT/2000
– no worrying about windows messing with your files
HPFS Disadvantages
• more complex
• requires 300 kB of memory for code, plus
more for the cache
• need third party drivers to access under
DOS and Linux
• Overhead is more than made up for by
increase in performace
– Especially true for large drives
Extended Attributes (EA)
• any extra information about a file
– E.g. icon file
• allows for longer file names (>11 characters)
• HPFS contains native support for EA, FAT does
not
• For FAT drives, OS/2 stores EA in one file in the
root directory
– WP ROOT.SF
– Conatins all EA for all files on the drive that need it
• Can get rather large
Memory Management
• Dividing up user memory to accommodate for
multiple processes - multiprogramming
• Effective memory management essential for
efficient processor utilization (ex. processor
blocking on I/O)
Use of Virtual Memory
• Most modern multiprogramming environments
(including OS/2 Warp) utilize Virtual Memory
Virtual Memory - memory allocated on a separate
disk
• Two techniques using VM - segmentation and
paging
Paging
• Static partitions of main memory into small
chunks (page frames)
• Process chunks (pages) are stored onto page
frames
• OS/2 default page size - 4KB
• Use of page table maps page to specific page
frame - allows for multiple page frames for one
process
Paging with Swapping
• OS/2 carries out paging using swapping
• Swapping - allows the use of some other disk file
as an extension of main memory
• Why? - OS/2 (as do most multiprogramming
environments) needs more memory than is
physically in main memory
• Result - OS/2’s page table now must know
whereabouts of each page (physically in memory
or not)
• Page faults occur when a page not in main
memory is requested - process blocks temporarily
Paging with Swapping (contd..)
• OS/2’s use of process swapping increases
flexibility
• Too much swapping - thrashing
• Thrashing - processor spends majority of time
swapping memory
• Thrashing causes your machine to slow to a crawl
• Memory MISManagement of OS/2 Warp?
Segmentation
• Dynamically divides main memory into segments
• Provides processes with protection - one process
cannot access the segments of another
• OS/2 1.x and beyond (Intel 80286 was the first
Intel chip to implement a segment model)
• OS/2 segmentation provides a segment hierarchy
ex. - which processes can perform “dangerous”
operations, etc.
• Use of segment table - segment number, within
segment offset
Segmentation(contd.)
• Segmentation was underutilized by OS/2 users:
– Intel 8086 chip (predecessor to 80286) emulated
segmentation poorly- users couldn’t distinguish
difference and became disillusioned
– Many 80286 applications used 8086 legacy software did not utilize segmentation capabilities of 80286
– 80286 (OS2 1.x) limited segment size (16KB). 80386
(OS/2 2.x) broadened (4GB) segment size, but most
users were using techniques other than segmentation
Flat Memory Model
• Memory model currently used by OS/2
• Flat Memory - memory is regarded as a single
large linear address space of 4GB
• Cannot disable segmentation on OS/2, only bypass
it - combine entire program (data and code) into
one large segment
• Programmer must only worry about within
segment offsets, no segment numbers
Flat Memory Model (contd.)
• Global address space - entire 4GB linear address
space
• Each process has distinct process address space threads belonging to same process also share
process address space
• Process address spaces not quite 4GB - some
memory used by operating system
Flat Memory Model (contd.)
• Mapping process address space into global address
space in OS/2:
4GB
System Region
512MB
Process Address
Space
0
• OS/2 limits process address space to 512MB
• Space above 512MB is the system region - used
for operating system resident processes
Threading
• Supports preemptive multithreading
• Uses 3 methods to determine priority:
– Privilege Settings
– States
– Priority Classes
Privilege Settings
• On 386 processors and above, there are 4 levels, or
“rings”, of privilege: ring 0, 1, 2, and 3
• Determines level of access, and scheduling method
• OS/2 primarily uses ring 0 and ring 3
– Ring 0 is highest level, for kernel and device drivers
– Ring 3 is lowest level, for all user applications (Word, email, etc)
• Ring 0 is cooperative, while ring 3 is preemptive
States
• 3 different states: running, ready, and blocked
– Running is whatever thread is executing at the time
– Ready is a thread ready to be run, but is not actually running
– Blocked is a sleeping thread, waiting for I/O or another thread
• Threads in a blocked state are not considered for execution when a
thread switch occurs
Priority Classes
• 4 classes: idle (I), regular (R), foreground server or fixed high (S), and
time-critical (TC)
– Idle threads lie in the background until nothing else is running. Example:
passive virus scanner
– Regular threads are ordinary threads. Baseline class for comparison to
other classes
– Foreground server or fixed high threads are like ordinary threads, except
they don’t lose priority when placed in the background
– Time-critical threads are the most important threads, and are executed
before any other class of threads
• Additional value, called delta, added to each class. Value ranges from
0 to 31, and can be dynamically changed