Question 5

Privileged mode is the same as kernel mode in terms of processor state.

F
Not true is a ring based security system

Privileged instructions cannot be attempted from user mode.

F
Can be attempted, but will be trapped by the kernel

A microkernel is a kernel containing many components that are

F
optimized to reduce resident memory size

Operating systems always provide both command line as well as

F
graphical user interfaces.

A blocking send() and blocking receive() is known as a rendezvous. T

Only the parent and child processes can use named pipes for
F
communication.

A significant problem with priority scheduling algorithms is determining

the length of the next CPU burst. T
I’d argue that it is false based on P271. The textbook states a major problem is indefinite F
blocking/starvation. Priorities determine the CPU burst length
The rate-monotonic scheduling algorithm is a static priority algorithm. T

Priority Inversion occurs when a higher-priority process needs to access

a data structure that is currently being accessed by a lower-priority T
process

A critical section refers to where a process is accessing/updating shared

T
data.

Compile-time binding is the method of binding instructions and data to

T
memory performed by most modern general-purpose operating systems.

An address generated by a CPU is referred to as a logical address. T

Host-attached storage is a special purpose storage system that is

F
accessed remotely over a data network

Seek time and rotational latency are the two components of positioning
T
time for a disk head.

An acyclic-graph directory structure is less flexible than a simple treestructured

F
directory structure.
 An absolute path name must always begin at the root. T

The execution of the application is suspended when a blocking system

T
call is issued.

A dedicated device cannot be used concurrently by several processes or

T
threads.

An access matrix (for allowing domains to access resources) is

F
generally dense.

The "key" scheme for implementing revocation (in access control

T
systems) allows selective revocation.

Question 6

a. Give two advantages of Virtual Memory compared to a simple memory system with load-time binding and
base and limit registers.

Process isolation

Dynamic memory allocation

b. Give two disadvantages of Virtual Memory compared to a simple memory system with load-time binding and
base and limit registers

i. More complicated

ii. Can’t page in and out?

iii. Thrashing?

c. Give one advantage of a static priority algorithm for real-time scheduling, and give one example of such an
algorithm.

i. Rate Monotonic for Real-Time Systems. Scheduling is more guaranteed? There doesn’t seem to be a
good answer if it is static priority

d. Give one advantage of a dynamic priority algorithm for real-time scheduling, and give one example of such an
algorithm.

i. Theoretically optimal, would use 100% CPU if context switching was free

ii. Earliest Deadline First

e. Give two advantages of using interrupts rather than polling for interaction between a host computer and a
device controller.

i. Takes a long time to interrupt happen

f. Give one disadvantage of using interrupts rather than polling for interaction between a host computer and a
device controller.

i. Context switching

Question 6.
 A. Advantages Virtual memory

a. Programs can be larger than physical memory

b. abstracts main memory into an extremely large uniform array of storage, separating logical memory as
viewed by the user from physical memory.

B. Disadvantages virtual memory

a. Substantially decreased performance

b. Very difficult to implement

C. Static priority algorithm

a. Advantage: Assigns higher priorities to tasks that require the CPU more often.

b. Example: Rate-monotonic scheduling

D. dynamic priority algorithm,

a. Advantage: Compared to static, allows priorities to change

b. Example: Earliest-deadline-first

E. Advantages of interrupts

a. Can handle immediately

b. Interrupts can have priorities

F. Disadvantages of interrupts

a. Might require debouncing to prevent numerous triggers within a small window.

Question 7

a. a

i. Hold and Wait

1. Prevention Method: A process must be holding at least one resource and waiting to acquire
additional resources that are currently being held by other processes.

2. Disadvantages:

a. Resource utilisation is low since recourses may be allocated but unused for long
periods.

b. Starvation is possible. A process that needs several popular resources may have to wait
indefinitely because at least one of the resources that it needs is always allocated to
some other process.

ii. No Preemption

1. Prevention Method: Recourses cannot be preempted. A resource can be released only

voluntarily by the process holding it, after that process has completed its task.

2. Disadvantage:

a. Cannot be applied to all data types. e.g. mutexes and semaphores.

iii. Circular Wait

 1. Prevention method: A set of waiting processes must exist such that the first process is waiting
for a resource from the second. The second waits for the third, and so on. The last process is
waiting for a resource from the first.

2. Disadvantage:

a. Can get stuck if all end up waiting and none release.

b. Access List

i. Attached to each object who can access and how. Access revocation for user is easy. Just remove the
person from the list

ii. Advantage: If a small number of files of files have a lot of users with different requirements.

c. Capability List

i. Capability list is per user and says what that person can do overall. Can be transferred because one
process can run on behalf of another

ii. Advantage: If a large number of users have a small number os files with different requirements

Question 6. (20 marks)

(a) Give two advantages of the EDF scheduling algorithm over the rate-monotonic scheduling
algorithm? (4 marks)

(1). EDF uses full computational power of processor whereas rate-monotonic only uses up to
µ = N(21/n-1) where n is the number of processes to be scheduled.

(2). EDF assigns priorities dynamically, so priority levels of tasks can change as time goes on. Rate-monotonic assigns
fixed priorities to tasks, therefore it is unreliable in meeting deadlines.

Some more:
1. Rate monotonic requires that processes are periodic, EDF does not.
2. EDF can schedule processes without missing deadlines that rate monotonic can’t due to static priorities.

(b) What are the two types of latency that affect the performance of real-time systems and how can these be
reduced?
 (4 marks)

(1). Interrupt Latency

The time period between the interrupt when it reaches the CPU, to the start of the routine servicing the interrupt.
Interrupt latency can be reduced by minimising the amount of time that the Kernel disables interrupts for, ie. code the
kernel well to be efficient.

(2). Dispatch Latency
The time it takes for the scheduling dispatcher to stop one process, and start another is known as dispatch latency.
This can be reduced by implementing a preemptive kernel.
Question 6 (continued)
(c) What is the difference between deadlock prevention and deadlock avoidance? (4 marks)

Deadlock avoidance: means, whenever a request is made for a particular resource by a particular process, the system
looks at the available resources, if the future resource needs for process's already in use resources, determine the
possibility of a deadlock in case the resource is granted. If possible, don't grant the resource, if not possible then grant
the resource.

Deadlock prevention: make sure that at least one of the condition for deadlock to occur is not fulfilled at anytime. This
can be achieved in the way resources are requested and granted in the system.

Similar to a police officer controlling traffic (Avoidance) vs. Traffic lights (Prevention)

Source: http://stackoverflow.com/questions/2485608/what-the-difference-between-deadlock-avoidance-and-deadlock-
prevention

Deadlock prevention prevents deadlocks from occuring by restricting how resource requests can be made, ensuring that
the four necessary conditions can not possibly occur.

Deadlock avoidance requires the system to know information in advance of resource requirements. The resource-
allocation state is dynamically examined to ensure no circular-wait condition (one of the necessary conditions) can occur
when a request is made.

tl;dr deadlock prevent = make sure deadlock is impossible by ensuring a necessary condition can never be satisfied.,
deadlock avoidance = at runtime refuse resource allocation if it would deadlock

(d) How is a limit register used for protecting main memory? (4 marks)

They are used to ensure user process only accesses memory that ‘belongs’ to that particular process. It is privileged
activity that only the kernel has the privilege to change the contents of limit registers (and base registers). If the user
program tries to access outside of the base - limit registers, it will result in a trap to the operating system, which in turn
treats the attempt as a fatal error.

Question 6 (continued)
(e) Give two scenarios where using polling may be better than using interrupts for interaction between a host
computer and a device controller. (4 marks)

Polling might be a better in cases where the polls frequently ‘hit’ what the host was requesting from the device controller.
IE - To reduce the number of context switches if a device needs to be polled very frequently.

If the context switch time is very large on a machine, then polling may be the better option to avoid context switches
where possible.

In bare metal code when there is no thread support, the CPU is often just looping until it a task is received. Therefore
instead of not doing anything in that spare time, a better alternative could be polling the device controller.

Question 7 (20 marks)

(a) Explain one advantage and one disadvantage of the SSTF disk scheduling algorithm? (4 marks)
Advantage - Reduces total seek time in comparison to FCFS.

Disadvantage - It is possible that starvation can occur where the disk head only stays in one area of the disk if it is busy.
Requests can occur any time, a burst of short-seek requests could starve a long-seek request since the short-seek requests
will be continually serviced on arrival. (requests can arrive WHILST the head is moving around the platter, remember)

(b) Name five common file attributes that an operating system keeps track of and associates with a file? (5
marks)

Name - Name of the file

File Identifier - Unique id used by the File System
File Type - type of file indicating the characteristics of the file [Not on unix tho? So maybe just the other 6]
Size - Size of the file (usually in bytes)
Location - Pointer to a device & the location on the device.
Protection - File system permission level / access control eg. read, write, execute
Time, Date & User Creation - Date of file creation, and sometimes last modified.

(c) What is the one main disadvantage to using a linear list to implement a directory structure? What is
one step that can be taken to compensate for this problem? (3 marks)

Linear lists are slow to search, at worst case O(n) run time. This is not ideal as the directory structure is used frequently
by users. Implementing a cache could compensate this problem by storing regularly used directory information and
improving search time.
P 552, the textbook suggests fixing O(n) by storing the list in the adt such as a balanced tree.

You can use a hashed table + linear list, with the file name used as a hash-key to do fast lookups of an entry in the linear
list.

(d) Give one advantage and one disadvantage of having access rights managed by the kernel rather
than by the user. (4 marks)

Advantage -

It is considered more secure as users are restricted from exploiting system calls, along with giving other potentially
malicious users full access privileges.

Disadvantage -

Kernel has dictatorship over all access rights, has full control and can perform malicious attacks. Also is inconvenient by
having only one source to manage all access rights.

Question 7 (continued)
(e) Explain two key differences between hardware virtualization and hardware emulation. (4 marks)

Virtualization - simulates parts of the computer’s hardware. Most operations happen on the actual hardware of the
computer for performance reasons. This is why virtualization is faster than emulation.

Emulation - In emulation the virtual machine simulates the entire computer in software. This provides interoperability as
it allows one computer’s OS to be able to run on another computer (of different architecture) such that the emulator is
written for. This means that even if the computer architecture is different, you can still run the OS on a different machine.

Question 8. (20 marks)

You have been asked to design an operating system for a new domestic robot that uses a multiprocessor
computer to respond to voice commands and uses video cameras to navigate around the house.

For each of the indicated aspects of the operating system:

(i) Describe your suggested design alternative,

(ii) Give one other less favourable choice,

(iii) Explain two advantages of the suggested alternative compared to the less favourable choice

(a) Scheduling algorithm (5 marks)

(i) Suggested solution (ii) Less favoured alternative

(iii) Two advantages of suggested solution

(i) Desired = Multi-Level queue scheduling

(ii) less desired Multi-Level Feedback Queue
(iii) I chose the multilevel queue scheduling so that processes with the highest priority (“The don’t crash into anything
process”) get as much processing time as they need.

I don’t like this answer much, but I’m not in a position to improve it. It’s a multiprocessor computer and these scheduling
algos are shit. https://en.wikipedia.org/wiki/Multiprocessor_scheduling#Algorithms
Maybe just pick a good and a bad algo and compare those instead.
Given that its multiprocessor, perhaps you could specify using asymmetric or symmetric multiprocessing alongside a
scheduling algorithm for the OS as a whole? Ie. SMP + EDF, suggested EDF since robots generally have real time
requirements

(b) File System Implementation (5 marks)

(The file system stores things like maps for the house, dictionaries for commands, and individual
programs for tsks like cleaning.)

(i) Suggested solution (ii) Less favoured alternative

(iii) Two advantages of suggested solution

(i).
For this I wasn’t sure whether it meant directory structure or file system structure.
For File System, I chose (i) Unix, then (ii) Linked Allocation with File Allocation Table. With (iii) as: A map of a
particular room may change with moving furniture. This will mean files will need to be changed in the middle.

For Directory structure I would have: (i) Graph Structure, (ii) Acyclic graph structure.with (iii) as: A room may is
connected to many other rooms and may be connected in a loop. (Cyclic Graph).

(c) Processor Affinity (i.e., whether particular tasks prefer particular processors). (5 marks)

(i) Suggested solution (ii) Less favoured alternative

(iii) Two advantages of suggested solution

I think we should be talking about process affinity (Soft Affinity / Hard Affinity / No Affinity)
For this, I would probably choose:
(i) Soft Affinity
(ii) Hard Affinity
(iii) Try to keep process one processor to avoid costly context switches; however, with soft affinity allow for processes to
switch processes to help reduce CPU overloading

There are lots of ways you can spin this, but I’d go hard affinity as my main choice - it guarantees that the process will
run on a specific processor(s) unlike soft. This would avoid slow access to other process’s memory as well. Note, this
could potentially reduce overall computational power available to the process (video processing could take more than one
cpu worth of power).
I would agree with hard affinity, since you could make the assumption the OS might have real time requirements you
could probably hand wave that its easier to provide with hard affinity - a CPU dedicated to control can’t be overloaded by
people trying to stream cat videos or some shit.

(d) Protection and Access Control

(5 marks)

(i) Suggested solution (ii) Less favoured alternative

(iii) Two advantages of suggested solution

This one is a funny one as I didn’t really think it was necessary.
(i) Have the kernel control access & protection.
(ii) Have the user control access and protection??

As per above, but I’d add:

(i) Lock key access protection (faster than other methods because its a simple does the key match the lock check, thus
saving on processing time - maybe relevant for a mobile robot). Also probably not a multi user system beyond root and
the default robot.
(i) Worst protection would be global table due to long lookup time.

Question 6

a) Describe what a Context Switch does and list 3 values are involved.

It is when a CPU switches to another process and saves the state into the PCB and reads the state from the
other processes PCB.

STATE, IDENTIFIER, ADDRESS SPACE, PARENT ID, CHILD ID.

b) For memory swapping, describe the difference between Backing Store and Roll Out Roll In.

Backing store is a large and fast disk able to store all copies of all memory images while Roll Out Roll in is priority
based scheduling for the memory.

c) What is the difference between internal and external Fragmentation?

Internal: allocated memory may be slightly larger than requested memory.
External: total memory space exists to satisfy request, bit is not contiguous.

d) What are atomic instructions and give an example of how they are used for synchronisation?

Idk an atomic increment, can be used with counting semaphores for synchronisation. Atomic instructions are
hardware level CPU instructions that cannot be interrupted and are executed in sequential order even across
 multiple CPUs (ie. in a multicore system or muliprocessor). They can be used to implement mutexes or locks
such as by the following pseudocode:

Do {

While (test_and_set(lock)) ; // do nothing

// crit section

Lock = false;

} while (true);

Test_and_set will return the current value of the lock and then set it to one atomically.

Question 5
EXAMPLE: Linux is an open source operating system T/F

A process is a passive entity, while a program is an active entity FALSE

The program counter of a single threaded process specifies the next memory location of the local variables FALSE
to be fetched

A multi threaded process has one program counter per thread TRUE

Privileged instructions can only be executed in kernel mode TRUE

When a process is in running state, it is waiting for event to complete FALSE

The process control block must contain the values of all CPU registers TRUE

A context switch only occurs when the process changes state TRUE

For a many to one threading model, multiple kernel threads are mapped to one user thread FALSE

For bounded waiting, the number of times that other processes are allowed to enter the critical sections, is TRUE
limited after a process has made a request to enter its critical section and before that request is granted

Roll out, roll in is a swapping variant used for time sliced based scheduling algorithms FALSE

Major part of swap time is execution time; total execution time is directly proportional to the amount of FALSE?
memory swapped

The STBR points to the segment table’s location in memory TRUE

If a page has the valid/invalid bit set to “invalid” the page is not in the physical address space of the TRUE
process
 A lazy swapper only swaps a page into memory, when needed TRUE

A page fault means that an error has occurred with current page being used by the process FALSE?

A RAID 4 system keeps duplicates of each disk used FALSE

C-LOOK treats the sectors as a circular list that wraps around from the last sectors to the first one TRUE

To recover from crashes, a log file system records each metadata update to the file system as a transaction TRUE

Read, write and seek commands are used with character based IO (streamed) devices FALSE

A domain switch allows a process to change domains TRUE

Question 8

a) Scheduling Algorithm
i) Solution: First come first served
ii) Less favoured: Round robin
iii) Two advantages of solution: First come first served reduces context switching
which is a waste of processing, especially on a battery powered device. Round
robin involves unnecessary context switches which is a waste of battery life.
b) File System Implementation
i) Solution: Extent-based system
ii) Less favoured: Contiguous
iii) Two advantages of solution: Allows allocation of larges blocks of disk space for
faster writing. Also allows for easier reading via large blocks.