Acquiring Source Code for PAs

Getting Source Code

Go back to the home directory by

cd ~

Usually, all works unrelated to system should be performed under the home directory. Other directories under the root of file system (/) are related to system. Therefore, do NOT finish your PAs and Labs under these directories by sudo.

Now acquire source code for PA by the following command:

git clone -b 2017 https://github.com/NJU-ProjectN/ics-pa.git ics2017

A directory called ics2017 will be created. This is the project directory for PAs. Details will be explained in PA1.

Issue the following commands to perform git configuration:

git config --global user.name "161220000-Zhang San" # your student ID and name
git config --global user.email "[email protected]"   # your email
git config --global core.editor vim                 # your favorite editor
git config --global color.ui true

You should configure git with your student ID, name, and email. Before continuing, please read this git tutorial to learn some basics of git.

Enter the project directory ics2017, then run

git branch -m master
bash init.sh

to initialize all the subprojects. This script will pull 4 subprojects from github. We will explain them later. Besides, the script will also add some environment variables into the bash configuration file ~/.bashrc. These variables are defined by absolute path to support the compilation of the subprojects. Therefore, DO NOT move your project to another directory once the initialization finishes, else these variables will become invalid. Particularly, if you use shell other than bash, please set these variables in the corresponding configuration file manually.

Git usage

We will use the branch feature of git to manage the process of development. A branch is an ordered list of commits, where a commit refers to some modifications in the project.

You can list all branches by

git branch

You will see there is only one branch called "master" now.

* master

To create a new branch, use git checkout command:

git checkout -b pa0

This command will create a branch called pa0, and check out to it. Now list all branches again, and you will see we are now at branch pa0:

  master
* pa0

From now on, all modifications of files in the project will be recorded in the branch pa0.

Now have a try! Modify the STU_ID variable in nemu/Makefile.git:

STU_ID=161220000            # your student ID

Run

git status

to see those files modified from the last commit:

On branch pa0
Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git checkout -- <file>..." to discard changes in working directory)

    modified:   nemu/Makefile.git

no changes added to commit (use "git add" and/or "git commit -a")

Run

git diff

to list modifications from the last commit:

diff --git a/nemu/Makefile.git b/nemu/Makefile.git
index c9b1708..b7b2e02 100644
--- a/nemu/Makefile.git
+++ b/nemu/Makefile.git
@@ -1,4 +1,4 @@
-STU_ID = 161220000
+STU_ID = 161221234

  # DO NOT modify the following code!!!

You should see the STU_ID is modified. Now add the changes to commit by git add, and issue git commit:

git add .
git commit

The git commit command will call the text editor. Type modified my STU_ID in the first line, and keep the remaining contents unchanged. Save and exit the editor, and this finishes a commit. Now you should see a log labeled with your student ID and name by

git log

Now switch back to the master branch by

git checkout master

Open nemu/Makefile.git, and you will find that STU_ID is still unchanged! By issuing git log, you will find that the commit log you just created has disappeared!

Don't worry! This is a feature of branches in git. Modifications in different branches are isolated, which means modifying files in one branch will not affect other branches. Switch back to pa0 branch by

git checkout pa0

You will find that everything comes back! At the beginning of PA1, you will merge all changes in branch pa0 into master.

The workflow above shows how you will use branch in PAs:

• before starting a new PA, new a branch pa? and check out to it
• coding in the branch pa? (this will introduce lot of modifications)
• after finish the PA, merge the branch pa? into master, and check out back to master

Compiling and Running NEMU

Now enter nemu/ directory, and compile the project by make:

make

If nothing goes wrong, NEMU will be compiled successfully.

To perform a fresh compilation, type

make clean

to remove the old compilation result, then make again.

To run NEMU, type

make run

However, you will see an error message:

nemu: nemu/src/cpu/reg.c:21: reg_test: Assertion `(cpu.gpr[check_reg_index(i)]._16) == (sample[i] & 0xffff)' failed.

This message tells you that the program has triggered an assertion fail at line 21 of the file nemu/src/cpu/reg.c. If you do not know what is assertion, blame the 程序设计基础 course. If you go to see the line 21 of nemu/src/cpu/reg.c, you will discover the failure is in a test function. This failure is expected, because you have not implemented the register structure correctly. Just ignore it now, and you will fix it in PA1.

To debug NEMU with gdb, type

make gdb

Development Tracing

Once the compilation succeeds, the change of source code will be traced by git. Type

git log

If you see something like

commit 4072d39e5b6c6b6837077f2d673cb0b5014e6ef9
Author: tracer-ics2017 <[email protected]>
Date:   Sun Jul 26 14:30:31 2017 +0800

    > compile NEMU
    161220000
    user
    Linux debian 3.16.0-4-686-pae #1 SMP Debian 3.16.7-3 i686 GNU/Linux
     14:30:31 up  3:44,  2 users,  load average: 0.28, 0.09, 0.07
    3860572d5cc66412bf85332837c381c5c8c1009f

this means the change is traced successfully.

If you see the following message while executing make, this means the tracing fails.

fatal: Unable to create '/home/user/ics2017/.git/index.lock': File exists.

If no other git process is currently running, this probably means a
git process crashed in this repository earlier. Make sure no other git
process is running and remove the file manually to continue.

Try to clean the compilation result and compile again:

make clean
make

If the error message above always appears, please contact us as soon as possible.

Local Commit

Although the development tracing system will trace the change of your code after every successful compilation, the trace record is not suitable for your development. This is because the code is still buggy at most of the time. Also, it is not easy for you to identify those bug-free traces. Therefore, you should trace your bug-free code manually.

When you want to commit the change, type

git add .
git commit --allow-empty

The --allow-empty option is necessary, because usually the change is already committed by development tracing system. Without this option, git will reject no-change commits. If the commit succeeds, you can see a log labeled with your student ID and name by

git log

To filter out the commit logs corresponding to your manual commit, use --author option with git log. For details of how to use this option, RTFM.

Submission

Finally, you should submit your project to the submission website. To submit PA0, put your report file (ONLY .pdf file is accepted) under the project directory.

ics2017
├── 161220000.pdf   # put your report file here
├── init.sh
├── Makefile
├── nanos-lite
├── navy-apps
├── nemu
└── nexus-am

Then go back to the project directory, issue

make submit

This command does 3 things:

1. Cleanup unnecessary files for submission
2. Cleanup unnecessary files in git
3. Create an archive containing the source code and your report. The archive is located in the father directory of the project directory, and it is named by your student ID set in Makefile.

If nothing goes wrong, transfer the archive to your host. Open the archive to double check whether everything is fine. And you can manually submit this archive to the submission website.

RTFSC and Enjoy

If you are new to GNU/Linux and finish this tutorial by yourself, congratulations! You have learn a lot! The most important, you have learn searching the Internet and RTFM for using new tools and trouble-shooting. With these skills, you can solve lots of troubles by yourself during PAs, as well as in the future.

In PA1, the first thing you will do is to RTFSC. If you have troubles during reading the source code, go to RTFM:

• If you can not find the definition of a function, it is probably a library function. Read man for more information about that function.
• If you can not understand the code related to hardware details, refer to the i386 manual.

By the way, you will use C language for programming in all PAs. Here is an excellent tutorial about C language. It contains not only C language (such as how to use printf() and scanf()), but also other elements in a computer system (data structure, computer architecture, assembly language, linking, operating system, network...). It covers most parts of this course. You are strongly recommended to read this tutorial.

Finally, enjoy the journey of PAs, and you will find hardware is not mysterious, so does the computer system! But remember:

• The machine is always right.
• Every line of untested code is always wrong.
• RTFM.