Transcript 1.01
Review: Chapters 1 – 7
Chapter 1:
OS is a layer between user and hardware to make life easier for user and
use hardware efficiently
Control program or resource allocator
Computer organization
CPU(s), memory, and I/O devices connect to a common bus
Devices request CPU attention through interrupts
Storage hierarchy: speed, cost, volatility
Caching: copy frequently used data to faster storage
Multiprogramming: multiple jobs in memory efficiency
Timesharing: frequently switch between jobs interactive, short response
time users get the impression that each has his/her own computer
Dual mode operation: user and kernel modes
Protect OS and users from each other
Privileged instructions executed only in kernel mode
Timer to prevent processes from holding resources forever
1.2
Operating-System Operations
OS is interrupt driven: it sits idle till something happens
Interrupts are generated by devices (hardware)
Traps (or exceptions) are software-generated interrupts due to
software errors, e.g., divide by zero
Request for operating system services (system calls)
Dual-mode operation allows OS to protect itself and other system
components
User mode and kernel mode
Mode bit provided by hardware
Provides ability to distinguish when system is running user code or
kernel code
Some instructions designated as privileged, only executable in
kernel mode
System call changes mode to kernel, return from call resets it to
user
1.3
Transition from User to Kernel Mode
1.4
Chapter 2: OS Services and Structures
OS provides two sets of services for
user convenience and
efficient use of resources
System calls: programming interface to OS services
Typically used through APIs for portability and ease
OS structures
monolithic
layered
microkernel
modular
1.5
Chapter 3: Processes
Process is a program in execution
OS maintains process info in PCB
Process State diagram
Creating and terminating processes (fork)
Process scheduling
Long-, short-, and medium-term schedulers
Scheduling queues
1.6
Scheduling: The Big Picture
Midterm sched.
Jobs
Disk
Job sched.
CPU sched.
In most small and interactive systems (UNIX, WinXP, …),
only the CPU scheduler exists
1.7
Process Lifetime
1.8
CPU Switch From Process to Process
(Context Switch)
When switching occurs,
kernel
Saves state of P0 in
PCB0 (in memory)
Loads state of P1
from PCB1 into
registers
State = values of the
CPU registers, including
the program counter,
stack pointer
1.9
Interprocess Communications Models
Message Passing
Shared Memory
1.10
Chapter 4: Threads
A thread is a basic unit of CPU utilization, a process is composed of
one or more threads
Each thread has: Program counter, stack, registers
Threads share: code, data, OS resources (e.g., open files and signals)
1.11
Single and Multithreaded Processes
Shared among threads
1.12
User level threads vs. kernel threads
1.13
Chapter 5: CPU Scheduling
Process execution: cycle of CPU bursts and I/O bursts
CPU bursts lengths: many short bursts, and few long ones
Scheduler selects one process from ready queue
Dispatcher performs the switching
Scheduling criteria (usually conflicting)
CPU utilization, waiting time, response time, throughput, …
Scheduling Algorithms
FCFS, SJF, Priority, RR, Multilevel Queues, …
1.14
First-Come, First-Served (FCFS) Scheduling
Process
Burst Time
P1
24
P2
3
P3
3
Suppose that the processes arrive in the order: P1 , P2 , P3
The Gantt Chart for the schedule is:
P1
P2
0
24
Waiting time for P1 = 0; P2 = 24; P3 = 27
Average waiting time: (0 + 24 + 27)/3 = 17
1.15
P3
27
30
Multilevel Feedback Queues
1.16
CPU Scheduling
Multiprocessor Scheduling
Processor affinity vs. load balancing
Evaluation of Algorithms
Modeling, simulation, implementation
1.17
Chapter 6: Synchronization
Processor Synchronization
Race condition
Techniques to coordinate access to shared data
Multiple processes manipulating shared data and result depends on
execution order
Critical section problem
Three requirements: mutual exclusion, progress, bounded waiting
Software solution: Peterson’s Algorithm
Hardware support: TestAndSet(), Swap()
Busy waiting (or spinlocks)
Semaphores:
Not busy waiting
wait(), signal() must be atomic moves the CS problem to kernel
Monitors
High-level (programming language) constructs
1.18
Synchronization
Some classical synchronization problems
Consumer-producer
Dining philosopher
Readers-writers
1.19
Chapter 7: Deadlock
A set of blocked processes each holding a resource and waiting to
acquire a resource held by another process in the set
Four necessary (but not sufficient) conditions
Mutual exclusion: only one process at a time can use a resource
Hold and wait: a process holding at least one resource and is waiting
to acquire additional resources held by other processes
No preemption: a resource can be released only voluntarily by the
process holding it, after that process has completed its task
Circular wait: there exists a set {P0, P1, …, P0} of waiting processes
such that P0 is waiting for a resource that is held by P1, P1 is waiting for
a resource that is held by
P2, …, Pn–1 is waiting for a resource that is held by
Pn, and P0 is waiting for a resource that is held by P0
1.20
Deadlock Handling
Prevention: ensure that at least one of the necessary conditions
does not hold
Avoidance: decide for each request whether or not the issuing
process should wait to avoid leaving the system in unsafe state
Resource-allocation graph: single instance of a resource type
Banker’s algorithm: multiple instances of a resource type
Detection and Recovery
Detection algorithm
Recovery: process termination or resource preemption
1.21
Good Luck on the Exam!
1.22