Uppsala University Operating Systems

1
Uppsala UniversityOperating Systems
UNIX Brahim Hnichhttp//www.csd.uu.se/brahim/os
1.html
2
Overview

• History and Evolution
• Overview of the UNIX system
• Fundamental concepts
• Key system calls
• Implementation of UNIX

3
Portable UNIX

• UNIX was written in C
• moving it to a new machine (porting it) is easy
• C compiler for the new machine
• writing device drivers for the new machine I/O
• small amount of machine-dependent code
• interrupt handlers
• memory management routines (assembly language)
• Portable C compiler could be retargeted to
  produce code for any reasonable machine with very
  few effort

4
Berkley UNIX

• University of California at Berkley acquired UNIX
• modified the system substantially
• introduced virtual memory, paging,
• implementation of file system changed much
  faster
• signal handling was made more reliable
• Networking introduced, TCP/IP standard
• added substantial number of utilities
• new editor (vi)
• new shell (csh)
• Pascal and Lisp compilers
• Berkley UNIX became well established in the
  academic, research, and defense worlds

5
Standard UNIX

• By late 80s two different, and quite incompatible
  versions of UNIX
• The first serious attempt to reconcile the two
  flaws, was by IEEE standard board (1003.1)
• attempt to standard failed
• Every group is developing UNIX in his own way
• gt UNIX became huge and complicated, rather than
  small and simple!!

6
UNIX Goals I

• UNIX an interactive timesharing system
• designed by programmers, for programmers
• sophisticated!
• Engaged in complex system development
• UNIX allows people to work together and share
  data in controlled ways
• different from the personal computer model of a
  single beginner working alone with a word
  processor

7
UNIX Goals II

• What a good programmer requires from the system?
• simple
• elegant
• consistent
• examples
• ls A
• rm A (principle of least surprise)
• grep ard f

8
Interfaces to UNIX
USERS
User interface
User Mode
Standard utility programs (shell, editors,
compilers, etc)
Library interface
Standard library (open, close, read, write, fork,
etc)
System call interface
UNIX operating system (process management, memory
management, the file system, I/O, etc)
Kernel Mode
Hardware (CPU, memory, disks, terminals,etc)
9
Logging Into UNIX

• To use UNIX, you first login
• login
• password
• Necessary to provide security
• resources
• file system
• Doesnt keep all user names and passwords in
  secret files
• UNIX scheme
• the password file contains one entry for each
  user (login, ID, encrypted password, home
  directory,)
• when user logs in, password is encrypted
• compares encrypted password with password file
• () eliminates the danger that the password file
  will leak somehow

10
UNIX Shell

• After a successful login, the shell initializes
  itself and types a prompt character
• When the user types a command line
• the shell extracts the first word from it
• assumes it is the name of a program to be run
• searches for this program, if found executes it
• A program, like the shell, does not need to open
  the terminal in order to read from or write to
• standard input
• standard output
• standard error
• Redirection of input/output
• e.g. sort ltin gtout
• Shell scripts files containing shell commands

11
Files and Directories in UNIX

• A UNIX file sequence of bytes containing
  arbitrary information
• No distinction between ASCII, binary, or any
  other kinds of files
• The meaning of the bits in a file is entirely up
  to the files owner
• Files can be protected by 9 bits right bits
• Files can be grouped in directories
• Directories are stored as files, and hence can be
  treated as files
• Directories can contain subdirectories

r
r
r
w
x
w
w
x
x
Everyone else
group
owner
12
UNIX Utility Programs

• Files and directory manipulation commands
• Filters
• Compilers and program development tools
• Text processing
• System administration
• Miscellaneous

13
Processes in UNIX I

• The only active entities in a UNIX system are
  processes
• Each process runs a single program
• UNIX is a multiprogramming system multiple
  processes
• Even when a logged user user is absent, multiple
  background processes are running daemons
• e.g. the cron daemon wakes up once a minute, doe
  ssome work, the goes to sleep
• Fork command creates processes
• e.g. pid fork()

14
Processes in UNIX II

• The ability to form a tree of processes gt
  timesharing
• Processes in UNIX communicate with each other
  using a form of message passing
• It is also possible to create a channel for
  communication between 2 processes pipes
• e.g. sort ltf head
• Processes can communicate by software interrupts
• process sends a signal to another process
• processes can tell the system what they want to
  happen when a signal arrives
• Each user is identified by an integer uid
  (password file), every process automatically
  acquires uid
• Superuser uid 0 (root) has the power to read
  or write all files in the system

15
UNIX Memory Model

• Each UNIX process has an address space consisting
  of 3 segments
• text segment machine instruction
• data segment storage for the programs
  variables,
• initialized part (Data)
• uninitialized part (BSS)
• stack segment starts at top of the virtual
  address space and grows down wards
• UNIX supports shared text segments

16
UNIX File System

• accessing files
• open file
• OS checks access rights
• return a file descriptor, if access is allowed
• absolute path
• relative path to the working directory
• linking is allowed in UNIX
• locking to implement mutual exclusion without
  semaphores
• specify the file to be locked
• the starting bytes
• the number of bytes
• two kinds of locks
• shared
• exclusive

17
Input/Output in UNIX

• UNIX solution to access devices, is to integrate
  them into the file system special files
• Each I/O device is assigned a path name (/dev)
• These special files can be accessed the same way
  as any other files
• e.g. cp file /dev/lp
• Usual file protection applies to I/O devices
• Special files
• block special files
• character special files

18
UNIX System Calls
Process Management
Files and Directories Management
pid fork()
Create a child process
fdcreate(name,mode)
Create a new file
swaitpid(pid, status, opts)
Wait for a child to terminate
fdopen(name,how)
Open a file for reading or writing
sexecve(name, argv, envp)
Replace a process core image
Close an open file
sclose(fd)
exit(status)
Terminate execution
nread(fd,buffer,nbytes)
Read data from file into a buffer
ssigaction(sig, act,oact)
Specify action to take for a signal
nwrite(fd,buffer,nbytes)
Write data from buffer to file
skill(pid,sig)
Send a signal to process
poslssek(fd,offset,whence)
Move the file pointer somewhere
residualalarm(seconds)
Schedule a SIGALRM signal later
sstart(name,buf)
Read and return info. about file
pause()
Suspend the caller unit until next signal
smkdir(name,mode)
Create a new directory
srmdir(name)
Delete an empty directory
slink(name1,name2)
Create a new directory entry for an old file
Memory Management
sunlink(name)
Remove a directory entry
sizebrk(addr)
Set the size of data segment
schdir(dirname)
Change the working directory
schmod(name,mode)
Change a files protection bits
Input/Output Management
scfsetospeed(termios, speed)
Set the output speed
scfsetispeed(termios, speed)
Set the input speed
scfgetospeed(termios, speed)
Get the output speed
scfgetispeed(termios, speed)
Get the input speed
stcsetattr(fd,opt,termios)
Set terminal attributes
stcgetattr(fd,termios)
Get terminal attributes
19
UNIX Implementation

• Machine-dependent kernel
• directly drives the H/W
• has to be rewritten from scratch whenever UNIX is
  ported to a new machine
• interrupt handlers
• low-level I/O system device drivers
• part of the memory management S/W
• Machine-independent kernel same for all machines
• system call handling
• process management
• scheduling
• pipes
• paging and swapping
• file system
• high-level part of I/O system

20
Implementation of Processes in UNIX

• Every process has two parts
• user part
• kernel part
• becomes active only when a system is invoked
• it has its own stack
• it has its own program counter
• Kernel maintains two key data structures
• process table
• resident all time in memory
• contains information needed for all processes
• user structure
• paged out or swapped, when its associated process
  is not in memory
• process table information
• scheduling parameters
• memory image
• signals
• user structure contains info. that is not needed
  when process is not in memory
• machine registers
• system call state

21
Process Creation

• Fork is executed gt trap to the kernel
• kernel looks for a free slot in the process table
  for use by the child
• copies all the information from parents process
  table entry to childs entry
• allocates memory for the childs data and stack
  segment (exact copy of the parents segment)
• the user structure is kept adjacent to the stack
  segment, and copied along with it
• child is ready to run

22
Scheduling Algorithm

• Two-level scheduling algorithm
• low-level algorithm picks the process to run
  next from the set of processes in main memory
• high-level algorithm moves processes between
  memory and disk

23
Implementation of Memory Man. in UNIX Swapping

• Swapping
• swapper it is the high-level scheduler, which
  handles movement between memory and disk
• events that lead to swapping
• a fork system call needed memory for a child
  process
• a BRK system call needed to expand a data segment
• a stack became larger and ran out of the space
  allocated to it
• free storage in memory and on the swap device is
  handled by linked lists of holes
• uses first-fit

24
Implementation of Memory Man. in UNIX Paging

• Processes need not be entirely in memory to run
• The only required information for a process to
  run
• page tables
• user structure
• The pages of the text, data, and stack segment
  are brought in dynamically, once at a time, as
  they are referenced
• If the user structure and page table are not in
  memory, the process cannot run until the swapper
  brings them in
• Paging is implemented partly by the main kernel,
  and partly by a new process called page daemon
• Page daemon is started up periodically so it can
  look around to see if there is any work for it to
  do
• If page daemon discovers that the number of free
  pages in memory is too low, it initiates action
  to free up more pages
• when page fault occurs, OS checks if there exist
  free page frame in the free list, otherwise the
  process is suspended until the page daemon has
  freed a page frame
• Page replacement algorithm global alg.
• executed by the paging daemon
• modified version of the clock alg.

25
Implementation of the UNIX File System

• I-nodes (UNIX)
• associated with each file is a little table
  called I-node (index-node)

Addresses of data blocks
Addresses of data blocks
I-node
Single indirect block
Attributes
Addresses of data blocks
Disk addresses
Double indirect block
Triple indirect block
26
Implementation Input/Output in UNIX

• I/O in UNIX is implemented by a collection of
  device drivers
• By providing standard interfaces between the
  drivers and the rest of the OS, most of the I/O
  system can be put into the machine independent
  part of the kernel
• I/O system is split into two patrs
• handling of the block special files
• handling of the character special files

Reading/writing files
cooked interface raw interface
User Space
File system
Line disciplines
Buffer cache
Kernel
C-list
Disk driver
Terminal driver
Keyboard
Disk
Terminal
27
Summary

• History and Evolution
• Overview of the UNIX system
• Fundamental concepts
• Key system calls
• Implementation of UNIX
• Whats next?
• I go to sleep, you prepare for your exam! GOOD
  LUCK and hope not to see you in this course again
  o)