Term
| What are the differences between the three kernel types? |
|
Definition
Monolithic kernel - all operating system services are run in kernel mode (eg. Linux)
Microkernel - core OS services are run in kernel mode; everything else run in user mode
Hybrid-kernel - somewhere in between (eg. Windows); some services run in user mode, some run in kernel mode |
|
|
Term
| What do processes facilitate? |
|
Definition
Processes facilitate:
* program structure
* interleaving program execution
* resource management |
|
|
Term
| Process creation (POSIX and Windows) |
|
Definition
POSIX - fork()
* copies the current running process in its entirety
* continues execution of the child process from where fork is called
Windows
* since Windows supports multiple APIs, it checks which API is used during process creation and takes appropriate action |
|
|
Term
| Explain the differences between Stalling's 2 state model, 5 state model, and 7 state model. |
|
Definition
* 2 state model - basic model
* 5 state model - adds blocking I/O support
* 7 state model - adds suspended state support |
|
|
Term
| When do processes terminate? |
|
Definition
Processes terminate when:
* there is an error
* when the process reaches the end of execution via similar instruction
* parent process is terminated |
|
|
Term
| How does the OS do cleanup at process termination? |
|
Definition
OS cleanups at process termination by
* releasing all non-shared memory allocated to process
* releases any resources locked to the process |
|
|
Term
|
Definition
| Threads share address space and resources. |
|
|
Term
| What are the differences between User level threads and Kernel level threads? |
|
Definition
User level threads - thread scheduling managed by program
Kernel level threads - thread scheduling managed by the OS |
|
|
Term
|
Definition
| speedup = (time taken to execute on a single processor) / (time taken to execute on N parallel processors) |
|
|
Term
| The differences between fine and coarse thread granularities. |
|
Definition
Fine grained - threads do pieces of the *same* problem; often threads perform portions of a loop; used to maximize usage of processors
Coarse grained - individual modules assigned a thread; each thread does something *different*; some threads may meet at a join |
|
|
Term
|
Definition
Deadlock
* the state where a process waits for resource it can never get because the resource is locked indefinitely by another process
* this happens when each process in the set is blocked awaiting an event that can only be triggered by another blocked process in the set |
|
|
Term
|
Definition
Starvation
* occurs when process never gets access to a resource due to competition from other resources
* possibly because other processes have higher priority and gets preferential scheduling |
|
|
Term
|
Definition
| Atomic operations are operations that cannot be interrupted |
|
|
Term
| Requirements for Mutual Exclusion (6) |
|
Definition
Requirements for Mutual Exclusion
1) It must be enforced: only one process at a time is allowed into its critical section
2) A process must halt in its critical section for the same resource or shared object.
3) It must not be possible for a process requiring access to a critical section to be delayed indefinitely: no deadlock or starvation
4) When no process is in a critical section, any process that requests entry to its critical section must be permitted to enter w/o delay
6) A process remains inside its critical section for a finite time only |
|
|
Term
|
Definition
Semaphores = a programming mechanism provided by the OS
* used by processes to coordinate access to a resource
* modeled as a variable with an int value
-semaphore may be initialized with a nonnegative int value
-semWait decrements the semaphore value; if the value becomes negative, the process executing the semWait is blocked
-semSignal increments the semaphore value; if the resulting value is <= 0, then a process blocked by a semWait operation (if any) is blocked |
|
|
Term
|
Definition
3 ways to use sempahores
* create critical sections
* use to count/limit resource usage
* used to signal other threads |
|
|
Term
| What 3 conditions make it possible to have deadlock? |
|
Definition
3 conditions that make it possible to have deadlock:
* hold and wait
* no preemption
* mutual exclusion |
|
|
Term
| What is the 4th condition that ensures deadlock? |
|
Definition
|
|
Term
|
Definition
Circular Wait:
A closed chain of processes exists such that each process holds at least one resource needed by the next process in the chain |
|
|
Term
|
Definition
Deadlock avoidance
* by not allowing processes to starve or resource request to be granted
* use Banker's algorithm
* know resource that all processes need beforehand |
|
|
Term
| Stalling vs Tanenbaum definitions of Thread |
|
Definition
Thread
* Stalling definition - a thread is the execution portion of a process which follows an execution path(trace) through one or more programs; this may interleave with that of other processes; thus, a process has an execution state (running, ready, etc) and a dispatching priority and is the entity that is scheduled dispatched by the OS
* Tanenbaum's definition - threads are processes that share the same address space |
|
|
Term
| The relationship between processes and threads in a modern OS like Windows? |
|
Definition
The relationship between processes and threads in a modern OS like Windows:
A process can have multiple threads. |
|
|
Term
| Under which condition can it take longer for a program to execute on an infinite number processors than it takes to execute on one processor? |
|
Definition
Under which condition can it take longer for a program to execute on an infinite number processors than it takes to execute on one processor:
If there is communication overhead that increases as the number of processor increases. |
|
|
Term
| What is a major disadvantage of User Level Threads? |
|
Definition
A major disadvantage of User Level Threads:
They cannot take advantage of multiple processors. |
|
|
Term
| Who does the scheduling for Kernel Level Threads? |
|
Definition
| The operating system does the scheduling for Kernel Level Threads. |
|
|
Term
| Where should memory shared by threads be allocated? |
|
Definition
| Memory shared by threads be allocated on the heap. |
|
|
Term
| What are reasons to use threads in a single-user multi-processing system? |
|
Definition
Reasons to use threads in a single-user multi-processing system:
* foreground and background work
* asynchronous processing
* modular program structure |
|
|
Term
| What is the term for when 2 threads try to modify the same area of memory in an undesirable manner? |
|
Definition
| Collision is when 2 threads try to modify the same area of memory in an undesirable manner. |
|
|
Term
| Define process control block |
|
Definition
Process control block:
A data structure that contains information to allow the OS to preempt and resume a process |
|
|
Term
True or False?
In Stalling's 5 state process model, when a process is added to the ready queue, the process is in the ready state and is ready to be run on the CPU. |
|
Definition
True.
In Stalling's 5 state process model, when a process is added to the ready queue, the process is in the ready state and is ready to be run on the CPU. |
|
|
Term
| Why does the suspended state exist? |
|
Definition
| Suspended state allows memory swapping to occur. |
|
|
Term
| What is one of the commands to create a process on a POSIX system? |
|
Definition
| fork() is one of the commands to create a process on a POSIX system |
|
|
Term
| What executable does Windows run when a user types "help.bat" in a terminal? |
|
Definition
| When a user types "help.bat" in a terminal, Windows runs cmd.exe. |
|
|
Term
| Why does the collection of processes that the OS manages have some processes in kernel mode and some processes in user mode? |
|
Definition
The collection of processes that the OS manages has some processes in kernel mode and some processes in user mode because:
switching between the 2 modes occur during normal operation of a program |
|
|
Term
| The purpose of a system call? |
|
Definition
| The purpose of a system call is to access operating system services. |
|
|
Term
| What data structure stores the system calls in the kernel? |
|
Definition
| The TRAP table stores the system calls in the kernel. |
|
|
Term
| In a microkernel design, OS services is requested via what mechanism? |
|
Definition
| In a microkernel design, OS services is requested via MESSAGES. |
|
|
Term
| What advantages does the seL4 microkernel have over monolithic kernel designs? |
|
Definition
| The advantage that the seL4 microkernel has over monolithic kernel designs is that it has been proven to be secure using mathematics. |
|
|
Term
| Before making a system call, what needs to be setup? |
|
Definition
| Before making a system call, registers need to be setup so that dispatch knows which system call to execute. |
|
|
Term
| A system call is executed from what space? |
|
Definition
| A system call is executed from user space. It starts from user space and then enters kernel space. |
|
|
Term
| The device driver serves what purpose? |
|
Definition
| The device driver handles communication to and from hardware. |
|
|
Term
| What does the function write do? |
|
Definition
| The function write sends data to an output device. |
|
|
Term
| What is a function of the dispatch handler? |
|
Definition
| The dispatch handler looks up the interrupt routine from the Trap table. |
|
|
Term
| Why is Link a system call? |
|
Definition
| Link is a system call because it needs to alter a directory file and it requires low level access to the disk drive. |
|
|
Term
True or False?
A program contains mutual exclusion, hold and wait, and no preemption conditions. The program has the potential to deadlock though deadlock may not occur. |
|
Definition
True.
A program contains mutual exclusion, hold and wait, and no preemption conditions. The program has the potential to deadlock though deadlock may not occur. |
|
|
Term
| When does the Banker algorithm deny a process resource access request? |
|
Definition
| The Banker algorithm denies a process resource access request when the resource request from the process will result in a state where deadlock may occur. |
|
|
Term
| Why can't mutual exclusion be excluded from the OS? |
|
Definition
| Mutual exclusion cannot be excluded from the OS because some resources require mutual exclusion, therefore it must be supported by the OS. |
|
|
Term
| In the dining philosopher problem, if we placed all 5 chopsticks in the middle and the philosopher can use any chopstick, what will happen? |
|
Definition
| Deadlock and starvation can occur. |
|
|
Term
| From Stalling's perspective of read and write with respect to concurrency, what kind of operations are file reads and file writes. |
|
Definition
| File read and file write are both write operations. |
|
|
Term
| In the dining philosopher problem, will there still be deadlock if we had 6 chopsticks and 5 philosophers? |
|
Definition
| No, our concurrency problem is solved. |
|
|
Term
| When does the Banker algorithm deny a process creation request? |
|
Definition
| The Banker algorithm denies a process creation request if creating the process has the potential to prevent one or all of the processes that is currently executing from running to completion. |
|
|
Term
| Why didn't the Babbage Difference Engine have an OS? |
|
Definition
| The Babbage Difference Engine didn't have an OS because it was designed for one task so it didn't need an OS. |
|
|
Term
| A time sharing system IBM 701 allowed what? |
|
Definition
| A time sharing system IBM 701 allowed more than one user to use a mainframe at the same time. |
|
|
Term
|
Definition
Batch System:
The term given to a system where the next program to be run is loaded automatically when the current program finishes. |
|
|
Term
| What does the batch system require to load the next program? |
|
Definition
| The batch program requires a monitor program to load the next program. |
|
|
Term
|
Definition
OS
* degree of abstraction from hardware
* provides services to programmers/users to make their life easier
* more services and higher levels of abstraction for more powerful programs |
|
|
Term
|
Definition
Time sharing
* multiple users connected at once; sharing of computer
* required multiprogramming features; can preempt currently running programs to allow others to run
* security and isolation of users important; know which connected device to send output
* eg. IBM 701(1958) |
|
|
Term
|
Definition
Virtual Memory
* invented in 1956
* first used in ATLAS computer 1962
* can switch virtual memory block in and out of main memory |
|
|
Term
| What are some services provided by a modern OS? |
|
Definition
Services provided by a modern OS
* GUI
* input device support (mice, keyboard)
* networking services
* file mgmt
* multiprocessor support
* support for multitasking
* power mgmt support
* graphics card
* hardware support
* extensible to new hardware
* security |
|
|
Term
|
Definition
CPU
* computers have one or more
* basic computational cycle: fetch-decode-execute
* each processor has a set of instructions to execute
* provides instructions to support OS |
|
|
Term
|
Definition
Registers
* store commands, operands, other info
* special purpose registers
-PC program counter: contains address of next instruction to be fetched
-IR instruction register: contains current instruction being executed
-AR address register: RAM address being read/written
-PSW program status word: contains bits that are set and cleared by OS to indicate CPU state
-SP stack pointer: points to top of stack of memory; stack holds local vars/params for function being executed
* OS manages registers |
|
|
Term
|
Definition
Interrupts
* preempt currently running code
* hardware or software based
* sequence of events:
1) program running
2) interrupt
3) chip saves subset of current registers
4) runs interrupt handler
5) goes back to where it left off |
|
|
Term
|
Definition
I/O Devices
* CPU & Memory are the only devices that are not I/O
* OS coordinates access to all devices
* manufacturer provides interface
* each device has a controller that decodes and performs commands sent from CPU
* device drivers are software that sends commands and receives responses from the device |
|
|
Term
| What is an OS? (Stalling vs Tanebaum) |
|
Definition
* Stalling's definition of OS:
An OS makes a computer more convenient and efficient, and allows for development, testing, and introduction of new system functions.
* Tanebaum's definition of OS
An OS provides an abstract layer on top of hardware to make programming easier; it is a resource manager. |
|
|
Term
|
Definition
GUI - Graphical User Interface
* provides common controls
* manages output of multiple UIs
* easy to use interface
* graphical representation of other systems |
|
|
Term
|
Definition
Multiprocessing - the capability for an OS to have multiple running processes
* during normal operations some processes may be blocked, so it may be more efficient to run a different process till the original process is unblocked
* OS may switch between different process
* more efficient use of resources |
|
|
Term
| What does the allocation of the CPU to the process? |
|
Definition
|
|
Term
|
Definition
* An abstraction of the memory system.
* Represented by integer values |
|
|
Term
| Describe the relationship between processes and address space. |
|
Definition
* Each process has its own address space
* Since address spaces are not shared between processes, sharing pointers between processes is meaningless
* Usually memory allocated to one process cannot be accessed from another process, unless special shared memory is allocated |
|
|
Term
| File System - what is their function? |
|
Definition
* Manage persistent data
* Provides device independent interface to devices (eg. disks, USB keys)
* Manage I/O devices (each device is a file) |
|
|
Term
| What is Interprocess Communications (IPC)? |
|
Definition
* A mechanism allowing processes to work together
- allows programs to be modular
- the OS gives each process a different address, but it refers back to the same block of physical memory |
|
|
Term
|
Definition
| A serial input/output mechanism. |
|
|
Term
| What are sockets used for? |
|
Definition
| Communication over a network. |
|
|
Term
| What are mutexes and semaphores for? |
|
Definition
| They are specialized communication mechanism for process coordination. |
|
|
Term
| Do kernel mode processes all share the same address space? |
|
Definition
|
|
Term
|
Definition
* Internet access
- support for network protocols and devices
* Resource sharing (eg. files, devices) |
|
|
Term
| What is auditing used for? |
|
Definition
* Monitoring of resource usage (availability, amount)
* Debugging
* Keep track of errors |
|
|
Term
| Describe the concept of the OS being separated into layers design. |
|
Definition
* The kernel layer interacts with the hardware
* OS and application software interact through APIs
* APIs can map into system calls |
|
|
Term
|
Definition
* Kernel Mode - privileged mode of operation; programs running in this mode have access to all system resources
- all processes share the same address space
- communication between kernel modules is very efficient
- All active components are in memory, typically
* User Mode - less privileged mode |
|
|
Term
| Elaborate on the microkernel. |
|
Definition
* makes kernel development more modular
* first microkernel provided memory mgmt, IPC functionaility and thread mgmt
* everything else implemented as user space modules
* more fault tolerant
* Mach was a microkernel Unix-like system |
|
|
