Scribd
Upload
19 views3 pages

Questions Exam

Uploaded by

wegiveyouart
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
19 views3 pages

Questions Exam

Uploaded by

wegiveyouart
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

INFO-F404 Oral Exam

Joël Goossens
December 2024

1 Introduction

The exam can be taken either in English or French.


The exam is individual and oral, and you may use the curriculum guide/notes/slides. You will be given two
questions, each related to a different chapter.
You will have a few minutes to consult (optionally) the curriculum guide/notes/slides (e.g., to find a nu-
merical example, recall the main idea of a proof, etc.) before answering the questions orally. During your
response, you can use a blank sheet of paper behind me to write down any necessary explanations.
Afterwards, I will review your answers and initiate a discussion that may include sub-questions on the same
topic, further details, and sometimes examples.

My objective is to assess the depth of your understanding, ranging from superficial (fail) to very deep (suc-
cess). It is not critical to replicate my examples, arguments, or notations, but I will evaluate the correctness
and understanding of your responses.

2 List of questions

2.1 Uniprocessor scheduling


1. Give a periodic implicit deadline systems not schedulable (using RM) with U < 1.
2. Prove the property: Let 𝜏 be a periodic task system, and let S1 and S2 be two schedules of work-
conserving schedulers for 𝜏. The schedule S1 is idle at instant t iff the schedule S2 is idle at instant t,
∀t ≥ 0. What is the purpose of this property?
3. Prove the optimality of RM in the FTP class.
4. For FTP class, synchronous systems, arbitrary deadline: show the non-optimality of RM & DM sched-
ulers. Exhibit the important phenomenon which exists for arbitrary schedule impossible for implicit
or constrained deadline.
5. For RM, synchronous systems and constrained deadline: provide the equation and the fixed-point
iteration process for the computation of the response time of the first job of 𝜏i . Discuss the convergence
of the process and suggest alternative values for initialization.
6. For RM, synchronous systems and constrained deadline find a system with 3 tasks (𝜏1 ≻ 𝜏2 ≻ 𝜏3 ) where
the equation
2
r1
r31 = C3 + ∑ ⌈ 3 ⌉ Cj
Tj
j=1
has two (positive) solutions.

1
7. For FTP class, synchronous systems, arbitrary deadline: provide and prove correct a feasibility interval.
8. For deterministic schedulers and discrete time space, prove that feasible schedules of periodic tasks
are periodic.

9. For FTP class, asynchronous systems, constrained deadline: prove correct the feasibility interval [0, Sn +
P). Explain the meaning of the Si instants.
10. For FTP class (𝜏1 ≻ 𝜏2 ≻ 𝜏3 ), asynchronous systems, constrained deadline: compute the Si values for
the following system (O, C, T) : 𝜏1 : (10, 2, 10), 𝜏2 : (0, 1, 5), 𝜏3 : (1, 1, 3) and then check the schedulability.
11. For FTP class, asynchronous systems, constrained deadline: provide and prove correct the main prop-
erty foundation of the optimal Audsley’s priority assignment.
12. Consider the following task set (O1 = 3, C1 = 2, D1 = 2, T1 = 4; O2 = 0, C2 = 3, D2 = 4, T2 = 8). In the
framework of FTP class (and asynchronous systems, constrained deadline) determine if 𝜏1 is lowest-
priority viable?
13. Prove the optimality of EDF.

14. For EDF, implicit deadline systems: prove the fact that U ≤ 1 is necessary and sufficient.
15. For EDF, synchronous systems, constrained deadline: provide the feasibility interval and the procedure
to determine its upper-bound.
16. LLF: define the notion of laxity, the priority management and the trashing situation using an example.

17. Build an LLF schedule for the following system (C, T, D): (2, 6, 4)(4, 5, 6).
18. For what type(s) of tasks and which scheduling algorithm(s) is the response time of the first job of each
task sufficient to conclude the schedulability of the task?
19. Provide at least three different feasibility intervals, specifying the type of tasks and the scheduler
involved for each.

2.2 Multiprocessor scheduling


20. For partitioning scheduling define and compare FF and BF heuristics.
21. For global scheduling prove that no online optimal scheduler exists.

22. Define the concept of scheduling anomalies, illustrate the phenomenon for global FTP scheduling and
the period parameter.
23. Define the PFair scheduling rule using the notion of lag.
24. Provide the basic sufficient condition for EDF schedulability of implicit deadline tasks, present the
EDF(k) improvement.
25. Build a global DM schedule for the following system upon two processors (C, T): (1, 2)(3, 6)(6, 12).

26. Build an EDF(k) schedule for the following system upon two processors (C, T): (9, 10)(1, 3)(4, 6).

27. Build a PFair schedule for the following system upon two processors (C, T): (2, 3)(2, 3)(4, 6).
28. Build a global EDF schedule for the following system upon two processors (C, T): (1, 2)(3, 6)(6, 12).
29. Provide a necessary and sufficient condition for the feasibility of implicit deadline systems upon mul-
tiprocessors.

30. Half-proof of the fact that partitioning and global scheduling are incomparable: system such that a
partition exists while all global-FTP schedulers fail.

2
2.3 Parallel algorithms
31. Provide the Amdahl’s law and illustrate its consequences.

32. Bitonic Mergesort: principles, algorithms, complexity.


33. Please apply the bitonic pre-processing process in order to convert this arbitrary sequence (length
n = 23 ):

<8, 7, 9, 5, 10, 3, 12, 1>

into a bitonic sequence of length 8.

2.4 Concurrency
34. Solution to the producer-consumer problem, bounded buffers.

35. Describe the Bakery algorithm.


36. Two questions related to the Bakery algorithm:
(a) Construct a scenario showing that the ticket numbers can be unbounded or provide an argument
showing this is not possible;
(b) Construct a scenario showing that the ticket numbers can be identical or provide an argument
showing this is not possible.
37. Describe the algorithm of Ricart-Agrawala (distributed critical sections).
38. Starvation definition. Illustrate the phenomenon using one of the Dekker’s attempts.

39. Live-lock definition. Illustrate the phenomenon using one of the Dekker’s attempts.
40. Describe the Byzantine Generals algorithm.

You might also like