CS 537: Introduction to Operating Systems Fall 2026: Midterm Exam #1 Solution Guide
This exam is closed book, closed notes.
All cell phones must be turned off and put away.
No calculators may be used.
You have two hours to complete this exam.
Write all of your answers on the accu-scan form with a #2 pencil.
These exam questions must be returned at the end of the exam, but we will not
grade anything in this booklet.
You may separate the pages of this exam if it helps you.
Unless stated (or implied) otherwise, you should make the following assumptions:
o The OS manages a single uniprocessor (single core)
o All memory is byte addressable
o The terminology lg means log2
o 210 bytes = 1KB
o 220 bytes = 1MB
o Page table entries require 4 bytes
o Data is allocated with optimal alignment, starting at the beginning of a page
o Assume leading zeros can be removed from numbers (e.g., 0x06 == 0x6).
o Hex numbers are represented with a preceding “0x”
This exam has 92 questions. Each question is worth the same number of points.
Good luck!
,Part 1: Virtualizing the CPU
Designate if the statement is True (a) or False (b).
1) One of the roles of the OS is to directly expose all details of a hardware resource to user-level applications.
False; the OS needs to provide an abstraction of the hardware to applications and may hide details to make
different hardware appear identical to other hardware
2) A program can be composed of multiple processes.
True; Can call fork() multiple times within a program
3) Each process has its own virtualized stack pointer.
True; each process (and each thread even) sees its own value for the stack pointer
4) Different processes accessing the same virtual memory address may see different contents.
True; each process has its own virtual address space that is distinct from others
5) Whenever a process is descheduled by the scheduler, the process is moved to the BLOCKED state.
False; when a process is descheduled, it is moved to the READY state
6) System calls are handled by the OS in kernel mode.
True;
7) A context switch is the same as a time slice.
False; a context switch changes the process that is running; the time slice is the amount of time a process has to
run in RR or a MLFQ
8) A running job may relinquish the CPU and become BLOCKED when it is waiting for the user to type at the
terminal.
True; when a process is waiting for something to happen and there is no reason to schedule it, it is in the blocked
state
9) With cooperative multi-tasking it is possible for a malicious user-level process to hold on to the CPU until
the machine is rebooted.
True; with cooperative multi-tasking there is no preemption, so the only time the OS gets control is when a
process blocks, makes a system call, or exits; if the process is spinning forever in a while() loop, the scheduler
may never get control
10) On a timer interrupt, the OS is responsible for transitioning from user to kernel mode.
False; on an interrupt, the hardware performs the transition from user to kernel mode
11) Multiple processes may be in the READY state at the same time.
True; can have many processes that want to use the CPU
12) A reasonable performance goal for a scheduler is to minimize job throughput (i.e., the number of jobs
completed per second).
, False; would want to MAXIMIZE job throughput
13) If all jobs have the same length (i.e., the same cpu burst length), an FCFS and a SJF scheduler give the same
average response time for the workload.
True; if all jobs have the same length, then SJF (which looks at job length) would default to FCFS to break ties.
14) SJF is provably optimal for preventing starvation.
False; SJF is provably optimal (for a non-preemptive scheduler) for minimizing turnaround time, but does suffer
from starvation (long jobs may never be scheduled if short jobs keep arriving)
15) A non-preemptive scheduler will only schedule a new job when the RUNNING job exits or becomes
BLOCKED.
True; non-preemptive schedulers never deschedule a running process
16) Increasing the time-slice of an RR scheduler is likely to decrease the overhead imposed by scheduling.
True; there will be fewer context switches for the same workload
17) Increasing the time-slice of an RR scheduler makes the scheduler behave more like FCFS.
True; once the time-slice is as long as the job length, it is equivalent to FCFS.
18) Increasing the time-slice of an RR scheduler makes the scheduler behave more like STCF.
False, see above.
19) An MLFQ scheduler assumes the presence of an oracle that knows the length of a job’s CPU burst in
advance.
False; MLFQ changes the priority of a process after it has begun to run as a way of dynamically adapting to the
the length of the CPU bursts that it observes
20) An MLFQ scheduler can preempt running jobs.
True; for example, RR is performed at the same priority.
This exam is closed book, closed notes.
All cell phones must be turned off and put away.
No calculators may be used.
You have two hours to complete this exam.
Write all of your answers on the accu-scan form with a #2 pencil.
These exam questions must be returned at the end of the exam, but we will not
grade anything in this booklet.
You may separate the pages of this exam if it helps you.
Unless stated (or implied) otherwise, you should make the following assumptions:
o The OS manages a single uniprocessor (single core)
o All memory is byte addressable
o The terminology lg means log2
o 210 bytes = 1KB
o 220 bytes = 1MB
o Page table entries require 4 bytes
o Data is allocated with optimal alignment, starting at the beginning of a page
o Assume leading zeros can be removed from numbers (e.g., 0x06 == 0x6).
o Hex numbers are represented with a preceding “0x”
This exam has 92 questions. Each question is worth the same number of points.
Good luck!
,Part 1: Virtualizing the CPU
Designate if the statement is True (a) or False (b).
1) One of the roles of the OS is to directly expose all details of a hardware resource to user-level applications.
False; the OS needs to provide an abstraction of the hardware to applications and may hide details to make
different hardware appear identical to other hardware
2) A program can be composed of multiple processes.
True; Can call fork() multiple times within a program
3) Each process has its own virtualized stack pointer.
True; each process (and each thread even) sees its own value for the stack pointer
4) Different processes accessing the same virtual memory address may see different contents.
True; each process has its own virtual address space that is distinct from others
5) Whenever a process is descheduled by the scheduler, the process is moved to the BLOCKED state.
False; when a process is descheduled, it is moved to the READY state
6) System calls are handled by the OS in kernel mode.
True;
7) A context switch is the same as a time slice.
False; a context switch changes the process that is running; the time slice is the amount of time a process has to
run in RR or a MLFQ
8) A running job may relinquish the CPU and become BLOCKED when it is waiting for the user to type at the
terminal.
True; when a process is waiting for something to happen and there is no reason to schedule it, it is in the blocked
state
9) With cooperative multi-tasking it is possible for a malicious user-level process to hold on to the CPU until
the machine is rebooted.
True; with cooperative multi-tasking there is no preemption, so the only time the OS gets control is when a
process blocks, makes a system call, or exits; if the process is spinning forever in a while() loop, the scheduler
may never get control
10) On a timer interrupt, the OS is responsible for transitioning from user to kernel mode.
False; on an interrupt, the hardware performs the transition from user to kernel mode
11) Multiple processes may be in the READY state at the same time.
True; can have many processes that want to use the CPU
12) A reasonable performance goal for a scheduler is to minimize job throughput (i.e., the number of jobs
completed per second).
, False; would want to MAXIMIZE job throughput
13) If all jobs have the same length (i.e., the same cpu burst length), an FCFS and a SJF scheduler give the same
average response time for the workload.
True; if all jobs have the same length, then SJF (which looks at job length) would default to FCFS to break ties.
14) SJF is provably optimal for preventing starvation.
False; SJF is provably optimal (for a non-preemptive scheduler) for minimizing turnaround time, but does suffer
from starvation (long jobs may never be scheduled if short jobs keep arriving)
15) A non-preemptive scheduler will only schedule a new job when the RUNNING job exits or becomes
BLOCKED.
True; non-preemptive schedulers never deschedule a running process
16) Increasing the time-slice of an RR scheduler is likely to decrease the overhead imposed by scheduling.
True; there will be fewer context switches for the same workload
17) Increasing the time-slice of an RR scheduler makes the scheduler behave more like FCFS.
True; once the time-slice is as long as the job length, it is equivalent to FCFS.
18) Increasing the time-slice of an RR scheduler makes the scheduler behave more like STCF.
False, see above.
19) An MLFQ scheduler assumes the presence of an oracle that knows the length of a job’s CPU burst in
advance.
False; MLFQ changes the priority of a process after it has begun to run as a way of dynamically adapting to the
the length of the CPU bursts that it observes
20) An MLFQ scheduler can preempt running jobs.
True; for example, RR is performed at the same priority.
