1. Lecture 42: Review for Final Exam

2. About Final Exam Comprehensive
Two double-sided US letter sheet are allowed
12-2pm, Thursday, May 3
Please arrive at 11:55am or earlier, so that we can start right at 12.

3. Overview Processes and Threads Mutual Exclusion Deadlocks
Process Scheduling
Memory Management
File System
OS Security

4. Process Process context Process creation
Address space
Registers
Stack pointer
Program counter
Files opened
Others
Process creation
fork(), exec()
Multiprogramming/Context switching

5. Threads Per-thread context Threads implementation POSIX standards
Stack
Registers
Stack pointer
Program counter
Threads implementation
User-level
Kernel-level
POSIX standards

6. Mutual Exclusion Core issue of multiprogramming Solutions
Disabling interrupts
Strict alternation
Peterson’s solution
Hardware support
Semaphore

7. Semaphore and condition variable
Mutual exclusion is not enough
E.g., producer-consumer problem
We need to block until a condition is satisfied
Semaphore
Condition variable

8. Classical OS problems Dining philosophers problem
Reader-writer problem

9. Deadlocks
Formal definition : A set of processes is deadlocked if each process in the set is waiting for an event that only another process in the set can cause

10. Four Conditions for Deadlock
Mutual exclusion condition
each resource assigned to 1 process or is available
Hold and wait condition
process holding resources can request additional
No preemption condition
previously granted resources cannot forcibly taken away
Circular wait condition
must be a circular chain of 2 or more processes
each is waiting for resource held by next member of the chain

11. Strategies for dealing with deadlocks
just ignore the problem altogether
detection and recovery
Directed circle in resource allocation graph
Algorithm for multiple-copy resources
dynamic avoidance
Safe and unsafe states
Banker’s algorithm
prevention
Attacking the four necessary conditions

12. Process scheduling What are the goals of process scheduling?
Scheduling Algorithms for Batch Systems
First come first serve
Shortest job first
Shortest remaining time next
Scheduling Algorithms for Interactive Systems
Round-Robin scheduling
Priority scheduling
Multiple queues
Lottery scheduling

13. Memory management Memory allocation Memory de-allocation
A new process comes into memory
Memory de-allocation
A process leaves memory
Address translation
Virtual address to physical address
Process isolation
One process cannot access the memory of other processes

14. Memory management approaches
Processes are known in advance
Swapping
Base and limit registers in CPU
Virtual memory

15. Virtual memory Paging Address translation via page table
MMU
Page fault
Speed up address translation
TLB
Large page tables
Multi-level page table
Inverted page table

16. Page Replacement Algorithms
FIFO (First-In-First-Out)
NRU (Not-Recently-Used)
Second Chance
LRU (Least-Recently-Used)
Clock Algorithm(s)
Determining which pages to replace is a decision that could have significant ramifications to system performance. If the wrong pages are replaced, i.e. pages which are referenced again fairly soon, then not only must these pages be fetched again, but a new set of pages must be tagged for removal.
The FIFO scheme has the benefit of being easy to implement, but often removes the wrong pages. Now, we can’t assume that we have perfect knowledge of a process’s reference pattern to memory, but we can assume that most processes are reasonably “well behaved.” In particular, it is usually the case that pages referenced recently will be referenced again soon, and that pages that haven’t been referenced for a while won’t be referenced for a while. This is the basis for a scheme known as LRU—least recently used. We order pages in memory by the time of their last access; the next page to be removed is the page whose last access was the farthest in the past.
A variant of this approach is LFU—least frequently used. Pages are ordered by their frequency of use, and we replace the page which has been used the least frequently of those still in primary storage.
Both these latter two policies have the advantage that they behave better for average programs than FIFO, but the disadvantage that they are costly to implement. It is straightforward to implement LRU for the Unix buffer cache, since the time to order the buffers is insignificant compared to the time required to access them. But we cannot afford to maintain pages in order of last reference.

17. Manage free memory/disk
Bitmap
Linked list

18. Files
Naming
Structure
Types
Access
Attributes
Operations

19. Directories
Directory systems
Path names
Operations

20. File system implmentation
Contiguous allocation
Linked list allocation
Linked list allocation with a table in memory
I-nodes

21. UNIX V7 Directory Implementation
The steps in looking up /usr/ast/mbox

22. Speeding up file operations
Cache
Block read ahead
Reduce disk-head motion

23. Hard links vs. symbolic links
Inode 134
links=2 . 12 .. 14 f1
134 f2
Hard link
inode 134 . 12 .. 14 f1
134 f2
208
special file
data
Symbolic link
/home/gong/f1

24. OS Security Access control Exploiting software Cryptography
Authentication

25. Access control Access control list Capability
File permissions in UNIX systems
-rwxrw-r--

26. Exploiting software Attacks Defenses Buffer overflow, stack smashingNX
StackGuard
ASLR
CFI

27. Cryptography Symmetric-key cryptography Public-key cryptography
Confidentiality
Block cipher
Stream cipher
Integrity & authenticity
HMAC
Public-key cryptography
Public-key encryption
Digital signature

28. Cryptographic hash functions
Preimage resistance
Given h, intractable to find y such that H(y)=h
Second preimage resistance
Given x, intractable to find y≠x such that H(y)=H(x)
Collision resistance
Intractable to find x, y such that y≠x and H(y)=H(x)

29. Authentication Authentication is to prove the identity of a user
Four categories
Something you know
Something you have
Someone you are
Someone you know