Name : .

htt
Roll No. : .....
Invigilators Signature : ..
CS/M. Tech (CSE-OLD)/SEM-1/CSEM-107/2011-12
p:/
2011
ADVANCED OPERATING SYSTEM
Time Allotted : 3 Hours Full Marks : 70
/q
The figures in the margin indicate full marks.
Candidates are required to give their answers in their own words
pap
as far as practicable.
Answer question no. 1 and any three questions from the rest from
Group-A and any one from the Group-B.

GROUP A
1. a) Consider nodes A, B, C that all use Lamport's logical
er.

clock concept. The notation A4 C 2 will mean that A

sent a message to C and that when the message was
sent A's logical clock reads 4 and when the message was
wb

received C's logical clock reads 2. Which of the following

are legal sets of events ? Explain your answers.
i) A5 C 7, C 8 A10
ut .

ii) C12 A5, A6 C11 3

b) "In Ricart- Agrawala's mutual exclusion algorithm a

process will get all N-1 REPLY messages only if it is
a c.

eligible to enter in the critical section, so the extra

RELEASE message is not required in this algorithm."
Justify this statement by discussing the Ricart-
Agrawala's mutual exclusion algorithm in detail. 5
in

40516 (O) [ Turn over

CS/M. Tech (CSE-OLD)/SEM-1/CSEM-107/2011-12

c) If the reader processes are in the sequence R1, R2, R3,

htt
R4 in the reader waiting queue and writer processes are
in the sequence W1, W 2 in the writer waiting queue
then discuss the shared file execution sequences of
p:/
these processes by discussing the monitor for readers'
priority problem. If R5 is arrived in the system when R4
is accessing the shared file, then show which one will
/q
gain the control of the monitor and how ? 6
2. a) Fig. below shows events of three processes P1 , P2 and
pap
P3 . Let e ij denotes the jth event of process Pi . Arrows

indicate transmission of message.

er.

Assume the processes use Lamport's logical clocks

wb

where C i denotes the local clock at process Pi . The

initial value of C1 = 0 for every process Pi . Assume that
the increment value is d = 1 for all processes.
i) To each event shown in the figure, assign the
ut .

correct clock value. 3

ii) In Lamport's logical clock if the increment value d
is different at each process, explain your answer.
a c.

3
b) Explain the limitations of the Lamport's logical clock
with the help of a diagram. What is busy waiting
semaphore ? How can it be solved ? 3+5
in

40516 (O) 2
 CS/M. Tech (CSE-OLD)/SEM-1/CSEM-107/2011-12
htt
3. a) What are the different approaches for maintaining

Cache Consistency of the data cached at clients ? If the

Cache Consistency scheme does not address the

p:/
sequential-write sharing issue, then what sort of

problems can arise ? 3+3

/q
b) "A distributed operating system makes collection of
pap
networked machines to act like a virtual uniprocessor."

What are the main advantages of this virtual machine

architecture for a user ? What design issues are

er.
required in distributed operating system to achieve this

goal ? 3+5
wb

4. a) How can basic primitives be used to solve the problem

of race condition ? 3

b) There is, one distributed mutual exclusion algorithm,

ut .

which gives the impression that the message complexity

of a distributed mutual exclusion can be O ( N )

a c.

instead of O(N). Why this algorithm needs 3 N

messages per CS execution ? Derive how each request

in

set's size becomes is equal to N . 6

40516 (O) 3 [ Turn over

CS/M. Tech (CSE-OLD)/SEM-1/CSEM-107/2011-12

c) "In a simple solution to distributed mutual exclusion, a

htt
control site is assigned the task of granting permissions
for the CS execution. To request the CS, a site sends a
REQUEST message to the control site. The control site
p:/
queues up the requests for the CS and grants them
permission, one by one. This method to achieve mutual
exclusion in distributed systems requires only
/q
3 messages per CS execution."
Discuss what prompted Lamport to develop mutual
pap
exclusion also, even it requires more messages [3(N-1)]
per CS invocation. 5
5. a) Write the edge chasing distributed deadlock detection
algorithm for the AND model. Show the probe message
er.
propagation along each edge for the following example
and then determine any deadlock exists there or not.

3+3
wb
ut .
a c.
in

40516 (O) 4
 CS/M. Tech (CSE-OLD)/SEM-1/CSEM-107/2011-12

b) Show the iterations (up to 3rd) for deadlock detection

htt
for Obermarck's algorithm for the following figure : 5
p:/
/q pap
er.
wb

c) How the file system and the virtual memory modules

negotiate for the physical memory usage in the Sprite
file system ? 3
6. a) "In distributed deadlock detection, in some algorithms
ut .

for accessing a remote resource a process site sends the

request message to a designated control site. But in
a c.

some algorithms the process site sends the request

message to the respective site where the resource is
located." Between these two approaches which one is
better ? Explain you answer. 2
in

40516 (O) 5 [ Turn over

CS/M. Tech (CSE-OLD)/SEM-1/CSEM-107/2011-12
htt b) Explain the migration algorithm and the read-

replication algorithm for implementing DSM systems. 5

p:/
c) Define what it means to say that a global state is

inconsistent. Give an example of such a state of a

/q
system consisting of two processes. Explain the

Chandy-Lamport's Global state recording algorithm.

pap
1+3+3

GROUP B
er.

7. a) Mention the conditions for setting the suitable range of

frame size f. Consider a system that contains four

wb

independent periodic tasks : in ( pi , e i , D i ) format :

T1 = (15, 1, 14 ), T2 = ( 20, 2, 26 ), T3 = ( 22, 3, 22 ) . Then

ut .

find the allowable frame sizes.

b) Is the above technique is all-time useful for cyclic

a c.

scheduling, if not what is the solution ? Explain it with

an example. 4+6+4
in

40516 (O) 6
 CS/M. Tech (CSE-OLD)/SEM-1/CSEM-107/2011-12

8. a) Write down the algorithm for Cyclic executive. 4

htt
b) Prove that when preemption is allowed and jobs do not

contend for resources, the EDF algorithm can produce a

p:/
feasible schedule of a set J of jobs with arbitrary release

times and deadlines on a processor if and only if J has

feasible schedules. 4
/q
c) Differentiate between dynamic and static systems. 2
pap
d) Whether the LST algorithm remains optimal for

scheduling jobs when more than one processor is

used ? Explain with example. 4

er.
wb
ut .
a c.
in

40516 (O) 7 [ Turn over