

Operating Systems 
CSCI 402, Spring 2018, All Sections

Click here to see a PREVIEW of important rules
that was posted before the semester started.
This is an undergraduate course on computer operating systems.
(But only graduate students are permitted to be in this class.
USC undergraduate students must take CS 350 in order to get credit for OS.
If you are an undergraduate student, you cannnot be in this class
and you cannot get credit for Operating System if you take this class.
Please check with your adviser to see which Operating System class you need to take!)
In addition to exploring concepts such as synchronization, virtual memory,
processes, file systems and virtualization, students will develop elements
of a fairly complete operating system during the course of the semester.


General Information


Instructor 
Bill Cheng
(click to see office hours)
Email:
<bill.cheng@usc.edu>. (Please do not send HTMLonly emails.
They will not be read.)

 DEN Section (29945D+29946D)
 PM Section (29971D)
 TT Section (30331D)

Time 
MW 9:30am  10:50am 
MW 12:00pm  1:50pm 
TT 9:30am  10:50am 
Location 
OHE 100D 
GFS 116 
GFS 116 
TA 
Ben Yan,
Email:
<wumoyan@usc.edu>
Office Hours: Fri 8:00am  10:00am in SAL open lab (in front of SAL 126)

Long Li,
Email:
<longl@usc.edu>
Office Hours: Mon 3:00pm  4:00pm in SAL open lab (in front of SAL 126)

Yue Shi,
Email:
<yueshi@usc.edu>
Office Hours: M/W 10:30am  11:30am in SAL open lab (in front of SAL 126)

Course Producer 
Chujun Geng, Email: <chujunge@usc.edu>,
Helpdesk Hours: Tue/Thu 3:30pm  5:00pm in SAL open lab (in front of SAL 126)


Graders 

Midterm Exam 
during class time, Wed, 3/21/2018 (firm)
(if your nunki login ID begins with "a", "b", or "c", please go to RTH 109)

during class time, Wed, 3/21/2018 (firm)

during class time, Thu, 3/22/2018 (firm)

Final Exam 
8am10am, Fri, 5/4/2018 (firm), in THH 102,
(THH is located in section 4E of the campus map).

11am1pm, Fri, 5/4/2018 (firm).

8am10am, Tue, 5/8/2018 (firm).



Class Resources


Description 
: 
textbooks, topics covered, grading policies, additional resources, etc.

Lectures 
: 
information about lectures (and lectures slides in PDF format).

Videos 
: 
information about DEN lectures and discussion sections videos.

Discussions 
: 
information about discussion sections.

Projects 
: 
programming assignments (please also see important information about the class projects
below.)

Participation 
: 
how to earn extra credit for class participation.

Newsgroup 
: 
Google Group for discussing
course materials and programming assignments. You are required to be
a member of this group. (This group is by invitation only.)
Please do not send request to join this group until after the first lecture.



News

(in reversed chronological order)
 5/6/2018:
 Below is the grade normalization information for kernel3.
Please note that this only applies to the graderdependent part of your grade.
If you are graded by Aparna Subbaraya <asubbara@usc.edu>,
her kernel3 average was 91.13 with a standard deviation of 17.80.
If you are graded by Erli Wang <erliwang@usc.edu>,
his kernel3 average was 86.99 with a standard deviation of 14.13.
If you are graded by Shreesh Kulkarni <shreeshk@usc.edu>,
his kernel3 average was 80.12 with a standard deviation of 29.99.
The overall class average for kernel3 was 86.30 with a standard deviation of 21.61.
To figure out your normalized score for kernel3, here's what you can do.
If your graderdependent part of your grade is X and your grader's average is A with a standard deviation of D,
then Y=(XA)/D is the number of standard deviations away your score is from your grader's average.
Therefore, your normalized graderdependent part of your grade would be
86.30+Y*21.61 (i.e., same number of
standard deviation away from the overall class average). Your minimum score
is still one point if you have submitted something for grading.
As I have mentioned in Lecture 1, although we assume that we have a bellshaped curve,
when your score is normalized, linear interpolation is used. It's clearly not perfect since
the actual curve will never be bellshaped and linear interpolation is not the same as bellshapedcurve interpolation.
But this is what was announced at the beginning of the semester, and therefore, we will stick to this
particular way of normailzation for all the programming assignments for the rest of the semester, knowing that it's not perfect.
 4/25/2018:
The final exam will be closed book, closed notes, and closed everything (and no "cheat sheet").
Also, no calculators, cell phones, or any electronic gadgets are allowed.
Please bring a photo ID. Your ID will be collected at the beginning
of the exam and will be returned to you when you turn in your exam. There will be assigned seating.
The final exam will cover everything from slide 21 of
Lecture 13 to slide 37 of
Lecture 15
PLUS from slide 29 of Lecture 17 to the last slide of
Lecture 30.
Also included are discussion section slides from Week 9
through Week 14.
Since the 2nd part of the course depends on stuff covered by the midterm,
I cannot say that I will not ask anything covered by the midterm and
you do need to know the material covered by the midterm. Therefore, it would
be more appropriate to say that the final exam will focus on the material
not covered by the midterm.
Regarding what types of questions will be on the exam, please see
the Exams section of the course description web page.
Regarding regrade policy, please see
the Regrade section of the course description web page.
Please note that if you are asked to run the Stride Scheduling algorithm, to get any credit,
you must run the one described in Lecture 29
(and not the one in the textbook).
Here is a quick summary of the final exam topics (not all topics covered may be listed):
 Ch 3  Basic Concepts
 Ch 4  OperatingSystem Design
 devices
 virtual machines, microkernels
 Ch 5  Processor Management
 threads implementations
 interrupts
 scheduling
 Ch 6  File Systems
 the basics of file systems
 performance improvements
 crash resiliency
 directories and naming
 RAID, flash memory, case studies
 Ch 7  Memory Management
 Kernel assignments 2 & 3
 spec
 FAQ
 my posts to class Google Group
 4/21/2018:
 Below is the grade normalization information for kernel2.
Please note that this only applies to the graderdependent part of your grade.
If you are graded by Aparna Subbaraya <asubbara@usc.edu>,
her kernel2 average was 96.57 with a standard deviation of 4.33.
If you are graded by Erli Wang <erliwang@usc.edu>,
his kernel2 average was 93.83 with a standard deviation of 14.29.
If you are graded by Shreesh Kulkarni <shreeshk@usc.edu>,
his kernel2 average was 96.08 with a standard deviation of 6.71.
The overall class average for kernel2 was 95.54 with a standard deviation of 9.35.
To figure out your normalized score for kernel2, here's what you can do.
If your graderdependent part of your grade is X and your grader's average is A with a standard deviation of D,
then Y=(XA)/D is the number of standard deviations away your score is from your grader's average.
Therefore, your normalized graderdependent part of your grade would be
95.54+Y*9.35 (i.e., same number of
standard deviation away from the overall class average). Your minimum score
is still one point if you have submitted something for grading.
As I have mentioned in Lecture 1, although we assume that we have a bellshaped curve,
when your score is normalized, linear interpolation is used. It's clearly not perfect since
the actual curve will never be bellshaped and linear interpolation is not the same as bellshapedcurve interpolation.
But this is what was announced at the beginning of the semester, and therefore, we will stick to this
particular way of normailzation for all the programming assignments for the rest of the semester, knowing that it's not perfect.
 3/31/2018:
 Below is the grade normalization information for kernel1.
Please note that this only applies to the graderdependent part of your grade.
If you are graded by Aparna Subbaraya <asubbara@usc.edu>,
her kernel1 average was 90.00 with a standard deviation of 16.44.
If you are graded by Erli Wang <erliwang@usc.edu>,
his kernel1 average was 87.86 with a standard deviation of 23.02.
If you are graded by Shreesh Kulkarni <shreeshk@usc.edu>,
his kernel1 average was 93.13 with a standard deviation of 10.69.
The overall class average for kernel1 was 90.37 with a standard deviation of 17.52.
To figure out your normalized score for kernel1, here's what you can do.
If your graderdependent part of your grade is X and your grader's average is A with a standard deviation of D,
then Y=(XA)/D is the number of standard deviations away your score is from your grader's average.
Therefore, your normalized graderdependent part of your grade would be
90.37+Y*17.52 (i.e., same number of
standard deviation away from the overall class average). Your minimum score
is still one point if you have submitted something for grading.
As I have mentioned in Lecture 1, although we assume that we have a bellshaped curve,
when your score is normalized, linear interpolation is used. It's clearly not perfect since
the actual curve will never be bellshaped and linear interpolation is not the same as bellshapedcurve interpolation.
But this is what was announced at the beginning of the semester, and therefore, we will stick to this
particular way of normailzation for all the programming assignments for the rest of the semester, knowing that it's not perfect.
 3/13/2018:
 Below is the udpated grade normalization information for warmup2.
Please note that this only applies to the graderdependent part of your grade.
If you are graded by Aparna Subbaraya <asubbara@usc.edu>,
her warmup2 average was 83.06 with a standard deviation of 26.09.
If you are graded by Erli Wang <erliwang@usc.edu>,
his warmup2 average was 80.26 with a standard deviation of 28.59.
If you are graded by Shreesh Kulkarni <shreeshk@usc.edu>,
his warmup2 average was 77.28 with a standard deviation of 31.61.
The overall class average for warmup2 was 80.19 with a standard deviation of 28.96.
To figure out your normalized score for warmup2, here's what you can do.
If your score is X and your grader's average is A with a standard deviation of D,
then Y=(XA)/D is the number of standard deviations away your score is from your grader's average.
Therefore, your normalized score would be 80.19+Y*28.96 (i.e., same number of
standard deviation away from the overall class average). Your minimum score
is still one point if you have submitted something for grading.
As I have mentioned in Lecture 1, although we assume that we have a bellshaped curve,
when your score is normalized, linear interpolation is used. It's clearly not perfect since
the actual curve will never be bellshaped and linear interpolation is not the same as bellshapedcurve interpolation.
But this is what was announced at the beginning of the semester, and therefore, we will stick to this
particular way of normailzation for all the programming assignments for the rest of the semester, knowing that it's not perfect.
 3/13/2018:
 Below is the udpated grade normalization information for warmup1.
Please note that this only applies to the graderdependent part of your grade.
If you are graded by Aparna Subbaraya <asubbara@usc.edu>,
her warmup1 average was 87.21 with a standard deviation of 24.16.
If you are graded by Erli Wang <erliwang@usc.edu>,
his warmup1 average was 91.36 with a standard deviation of 17.41.
If you are graded by Shreesh Kulkarni <shreeshk@usc.edu>,
his warmup1 average was 90.54 with a standard deviation of 19.11.
The overall class average for warmup1 was 89.72 with a standard deviation of 20.49.
To figure out your normalized score for warmup1, here's what you can do.
If your score is X and your grader's average is A with a standard deviation of D,
then Y=(XA)/D is the number of standard deviations away your score is from your grader's average.
Therefore, your normalized score would be 89.72+Y*20.49 (i.e., same number of
standard deviation away from the overall class average). Your minimum score
is still one point if you have submitted something for grading.
As I have mentioned in Lecture 1, although we assume that we have a bellshaped curve,
when your score is normalized, linear interpolation is used. It's clearly not perfect since
the actual curve will never be bellshaped and linear interpolation is not the same as bellshapedcurve interpolation.
But this is what was announced at the beginning of the semester, and therefore, we will stick to this
particular way of normailzation for all the programming assignments for the rest of the semester, knowing that it's not perfect.
 3/7/2018:
The midterm exam will be closed book,
closed notes, and closed everything (and no "cheat sheet").
Also, no calculators, cell phones, or any electronic gadgets are allowed.
Please bring a photo ID. Your ID will be collected at the beginning
of the exam and will be returned to you when you turn in your
exam. There will be assigned seating.
The midterm exam will cover everything from the beginning of the
semester to slide 19 of Lecture 17 on 3/5,6/2018,
MINUS Chapter 5 (i.e., material in Ch 5 is excluded from the midterm).
Regarding what types of questions will be on the midterm, please see
the Exams section of the course description web page
and slides 20 through 28 of Lecture 17 on 3/5,7/2018.
Regarding regrade policy, please see
the Regrade section of the course description web page.
Here is a quick summary of the midterm exam topics (not all topics covered may be listed):
 Ch 1  Introduction
 introduction
 a simple OS
 files
 Ch 2  Multithreaded Programming
 thread creation, termination, synchronization
 thread safety, deviations
 Ch 3  Basic Concepts
 context switching, I/O
 dynamic storage allocation
 static linking and loading
 booting
 Ch 4  OperatingSystem Design
 a simple system
 storage management
 Warmup assignments 1 & 2
 specs
 FAQs
 my posts to class Google Group
 Kernel assignment 1
 spec
 FAQ
 my posts to class Google Group
Please note that kernel 1 is included in the midterm coverage but Chaper 5 is not.
This mean that I can ask weenixspecific questions.
 3/6/2018:
 Below is the grade normalization information for warmup2.
If you are graded by Aparna Subbaraya <asubbara@usc.edu>,
her warmup2 average was 85.81 with a standard deviation of 27.93.
If you are graded by Erli Wang <erliwang@usc.edu>,
his warmup2 average was 83.62 with a standard deviation of 30.90.
If you are graded by Shreesh Kulkarni <shreeshk@usc.edu>,
his warmup2 average was 80.89 with a standard deviation of 33.51.
The overall class average for warmup2 was 83.43 with a standard deviation of 30.95.
To figure out your normalized score for warmup2, here's what you can do.
If your score is X and your grader's average is A with a standard deviation of D,
then Y=(XA)/D is the number of standard deviations away your score is from your grader's average.
Therefore, your normalized score would be 83.43+Y*30.95 (i.e., same number of
standard deviation away from the overall class average). Your minimum score
is still one point if you have submitted something for grading.
As I have mentioned in Lecture 1, although we assume that we have a bellshaped curve,
when your score is normalized, linear interpolation is used. It's clearly not perfect since
the actual curve will never be bellshaped and linear interpolation is not the same as bellshapedcurve interpolation.
But this is what was announced at the beginning of the semester, and therefore, we will stick to this
particular way of normailzation for all the programming assignments for the rest of the semester, knowing that it's not perfect.
 2/18/2018:
 Below is the grade normalization information for warmup1.
If you are graded by Aparna Subbaraya <asubbara@usc.edu>,
her warmup1 average was 89.90 with a standard deviation of 26.98.
If you are graded by Erli Wang <erliwang@usc.edu>,
his warmup1 average was 95.53 with a standard deviation of 19.51.
If you are graded by Shreesh Kulkarni <shreeshk@usc.edu>,
his warmup1 average was 92.90 with a standard deviation of 24.08.
The overall class average for warmup1 was 92.78 with a standard deviation of 23.83.
To figure out your normalized score for warmup1, here's what you can do.
If your score is X and your grader's average is A with a standard deviation of D,
then Y=(XA)/D is the number of standard deviations away your score is from your grader's average.
Therefore, your normalized score would be 92.78+Y*23.83 (i.e., same number of
standard deviation away from the overall class average). Your minimum score
is still one point if you have submitted something for grading.
As I have mentioned in Lecture 1, although we assume that we have a bellshaped curve,
when your score is normalized, linear interpolation is used. It's clearly not perfect since
the actual curve will never be bellshaped and linear interpolation is not the same as bellshapedcurve interpolation.
But this is what was announced at the beginning of the semester, and therefore, we will stick to this
particular way of normailzation for all the programming assignments for the rest of the semester, knowing that it's not perfect.
 2/6/2018:
 I'm still quite sick and I don't think I can handle giving two lectures a day for the next 2 days.
So, I will still give a lecture tomorrow in the DEN section so it can be recorded and I'm canceling PM and TT sections lectures
on tomorrow and Thursday. There will be no roll sheet signing and everyone will get credit.
 Office hour tomorrow (Wed, 2/7/2018) is moved to 1:301:55pm followed by 2:353:10pm.
Sorry about the inconvenience.
 2/5/2018:
 Office hour tomorrow (Tue, 2/6/2018) is canceled since I'm still sick.
Sorry about the inconvenience.
 2/5/2018:
 PM section lecture today and TT section lecture tomorrow are canceled because I'm sick.
I will still lecture today during the DEN section so the lecture can be recorded. Please
watch this lecture video before class on Wednesday/Thursday. No roll sheet signing
today and tomorrow. Everyone will get credit for it.
 Office hour today is canceled for the same reason.
If you have questions about the programming assignments, please see the CP and the TAs.
Sorry about the inconvenience and the short notice.
 1/7/2018:
 Watch this area for important announcements.
 To get user ID and password for accessing protected
area of this web site, please visit the request access page after semester starts and submit the requested information.
(You do not have to be registered for the course to get the password. You just need to have an USC email address.)
 Please do not send request to join the
class Google Group until after the first lecture.


Prerequisites

In the official syllabus, it is listed that the prerequisites are:
(CSCI 201L or CSCI 455x) and (EE 357 or EE 352L)
Please see:
Apparently, they are the prerequisites for undergraduate students only.
The CS department would waive these prerequisites for graduate students.
Since undergraduate students are required to take CS 350 for OS credit,
there should only be graduate students enrolled in CS 402. Therefore,
these prerequisites are really not prerequisites.
They should be considered recommended preparation for graduate students.
The basic idea behind these prerequisites is that you are expected to know
how to program and you are expected to know something about computer architecture
(such as what the CPU does).


Important Information about Programming Assignments

The programming assignments of this class will be very demanding.
You will be required to write C code. Since C is
a proper subset of C++, knowing C++ well would give you enough
background. However, some of the things that available in C++,
such as strings and streams, are not be available in C. So, you need
to know how to do things such as
manipulating nullterminated array of characters
(using functions such as strchr, strrchr, strlen, strcmp, strncpy, etc.)
and performing console and file I/O
(using functions such as printf/snprintf, fread/fwrite, read/write, fgets, etc.)
in C.
No other programming language will be accepted.
We will not teach C in this class.
You are expected to pick up C on your own if you are not familiar with it.
You should also get familiar with the Unix
development environment (vi/pico/emacs, cc/gcc, make, etc.)
You are expected to know how to use Unix. If you are not familiar with Unix,
please read Unix for the Beginning Mage,
a tutorial written by Joe Topjian.
If you forgot how to use Unix, please summary of some commonly used Unix commands.
The kernel programming assignments must run on Ubuntu 12.04 or Ubuntu 14.04.
Therefore, you should install
Ubuntu 14.04 (or Ubuntu 12.04) on your laptop or desktop as soon as possible.
If you do not have a personal laptop or desktop that runs Windows or Mac OS X, please contact the instructor as soon as possible.
Please note that the preferred version of Ubuntu is
Ubuntu 14.04 (unless you have a laptop with only 2GB of memory
or a slow CPU, then you should install Ubuntu 12.04)
These days, I have been using VagrantBox (i.e., Vagrant with Virtualbox)
to install and run Ubuntu 14.04. I think it has a better integration with Windows 10 than
other systems. If you are running Windows, I would recommend installing Vagrant on your laptop/desktop.
If a student registered late for this class or could not be present
at the beginning of the semester,
he/she is still required to turn all projects and homeworks
on time or he/she will receive a score of 0 for these assignments.
No exceptions!


