week 5 os theory Flashcards

Question

what is resource allocation

Answer 1

Resource allocation: Processes may compete for resources such as the CPU, memory, and I/O devices.

Answer 2

The owners of information stored in a multi-user or networked computer system may want to control use of that information, and concurrent processes should not interfere with each other

Answer 3

Programming interface to the services provided by the OS (e.g. open file, read file, etc.) Typically written in C (or C++) Mostly accessed by programs via a high-level Application Program Interface (API) rather than direct system call. Two common APIs: Win32 API for Windows, POSIX API for UNIX-based systems (including virtually all versions of UNIX, Linux, and Mac OS X). The APIs allow application code to access the kernel without requiring root privileges: it transfers control from user mode to kernel mode

Answer 4

open read write close

Answer 5

Register the file with the operating system. Must be called before any operation on the file. Returns an integer called the file descriptor - an index into the list of open files maintained by the OS.

Answer 6

Read data from the file. Returns number of bytes read or 0 for end of file.

Answer 7

Write data to a file. Returns number of bytes written.

Answer 8

De-registers the file with the operating system. No further operations on the file are possible

Answer 9

return negative number

Answer 10

Trapping to the Kernel The user process calls the system call wrapper function from the standard C library The wrapper function issues a low-level trap instruction (in assembly) to switch from user mode to kernel mode There is some ‘magic’ happening to move from user mode to kernel mode: To get around the problem that no call can directly be made from user space to a specific function in kernel space: Before issuing the trap instruction, an index is stored in a well known location (e.g. CPU register, the stack, etc.). Then, once switched into kernel space, the index is used to look up the desired kernel service function, which is then called. Some function calls may take arguments, which may be passed as pointers to structures via registers.

Answer 11

UNIX - one big kernel Consists of everything below the system-call interface and above the physical hardware Provides the file system, CPU scheduling, memory management, and other operating-system functions; a large number of functions for one level Limited to hardware support compiled into the kernel

Answer 12

Most modern operating systems implement kernel modules Uses object-oriented–like approach Each core component is separate Each talks to the others over known interfaces Each is loadable as needed within the kernel, so you could download a new device driver for your OS and load it at run-time, or perhaps when a device is plugged in Overall, similar to layered architecture but with more flexibility, since all require drivers or kernel functionality need not be compiled into the kernel binary. Note that the separation of the modules is still only logical, since all kernel code (including dynamically loaded modules) runs in the same privileged address space (a design now referred to as monolithic), so I could write a module that wipes out the operating system no problem. This leads to the benefits of micro-kernel architecture, which we will look at soon

Answer 13

Moves as much as possible from the kernel into less privileged “user” space (e.g. file system, device drivers, etc.) Communication takes place between user modules using message passing The device driver for, say, a hard disk device can run all logic in user space (e.g. decided when to switch on and off the motor, queuing which sectors to read next, etc.) But when it needs to talk directly to hardware using privileged I/O port instructions, it must pass a message requesting such to the kernel

Answer 14

Easier to develop microkernel extensions Easier to port the operating system to new architectures More reliable (less code is running in kernel mode) - if a device driver fails, it can be re-loaded More secure, since kernel is less-complex and therefore less likely to have security holes. The system can recover from a failed device driver, which would usually cause “a blue screen of death” in Windows or a “kernel panic” in linux

Answer 15

Performance overhead of user space to kernel space communication

Answer 16

The Minix OS and L3/L4 are examples of microkernel architecture

Answer 17

A virtual machine allows to run one operating system (the guest) on another operating system (the host) A virtual machine provides an interface identical to the underlying bare hardware The operating system host creates the illusion that a process has its own processor and (virtual) memory Each guest is provided with a (virtual) copy of underlying computer, so it is possible to install, say, Windows 10 as a guest operating system on Linux.

Answer 18

Presents guest with system similar but not identical to hardware (e.g. Xen Hypervisor) Guest OS must be modified to run on paravirtualized ’hardware’ For example, the kernel is recompiled with all code that uses privileged instructions replaced by hooks into the virtualisation layer After an OS has been successfully modified, para-virtualisation is very efficient, and is often used for providing low-cost rented Internet servers (e.g. Amazon EC2, Rackspace

Answer 19

VMWare implements full virtualisation, such that guest operating systems do not require modification to run upon the virtualised machine. The virtual machine and guest operating system run as a user-mode process on the host operating system As such, the virtual machine must get around some tricky problems to convince the guest operating system that it is running in privileged CPU mode when in fact it is not. Consider a scenario where a process of the guest operating system raises a divide-by-zero error. Without special intervention, this would cause the host operating system immediately to halt the virtual machine process rather than the just offending process of the guest OS. So VMWare must look out for troublesome instructions and replace them at run-time with alternatives that achieve the same effect within user space, albeit with less efficiency But since usually these instructions occur only occasionally, many instructions of the guest operating system can run unmodified on the host CPU