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.

不要使用root账户做实验!!!

从现在开始, 所有与系统相关的配置工作已经全部完成, 你已经没有使用root账户的必要. 继续使用root账户进行实验, 会改变实验相关文件的权限属性, 可能会导致开发跟踪系统无法正常工作; 更严重的, 你的误操作可能会无意中损坏系统文件, 导致虚拟机无法启动! 往届有若干学长因此而影响了实验进度, 甚至由于损坏了实验相关的文件而影响了分数, 请大家引以为鉴, 不要贪图方便, 否则后果自负!

不过, 今年的实验配置中已经进行了相关的限制: 在container中不能直接切换到root账户.

如果你仍然不理解为什么要这样做, 你可以阅读这个页面: Why is it bad to login as root? 正确的做法是: 永远使用你的普通账号做那些安分守己的事情(例如写代码), 当你需要进行一些需要root权限才能进行的操作时, 使用sudo.

Now acquire source code for PA by the following command:

git clone https://github.com/NJU-ProjectN/ics2016

A directory called ics2016 will be created. This is the project directory for PAs. Details will be explained in PA1.

Git configuration and usage

Issue the following commands:

git config --global user.name "151220000-Zhang San"    # your student ID and name
git config --global user.email "zhangsan@foo.com"    # 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.

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

Enter the project directory ics2016. 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 config/Makefile.git:

STU_ID=151220000            # your student ID

Run

git status

to see those files modified from the last commit:

On branch ics2016
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:   config/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/config/Makefile.git b/config/Makefile.git
index c9b1708..b7b2e02 100644
--- a/config/Makefile.git
+++ b/config/Makefile.git
@@ -1,4 +1,4 @@
-STU_ID = 151220000
+STU_ID = 151221234

  # 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. 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 config/Makefile.git, and you will find that STU_ID is still unchanged! By issuing git log, you will find the commit log you just created 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 above workflow 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

Before compiling the project, you should install the readline library:

apt-get install libreadline-dev

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

make

If nothing goes wrong, NEMU will be compiled successfully.

What happened?

You should know how a program is generated in the 程序设计基础 course. But do you have any idea about what happened when a bunch of information is output to the screen during make is executed?

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-ics2016 <tracer@njuics.org>
Date:   Sun Jul 26 14:30:31 2016 +0800

    > compile NEMU
    151220000
    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/ics2016/.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.

开发跟踪

我们使用git对你的实验过程进行跟踪, 不合理的跟踪记录会影响你的成绩. 往届有学长"完成"了某部分实验内容, 但我们找不到相应的git log, 最终该部分内容被视为没有完成. git log是独立完成实验的最有力证据, 完成了实验内容却缺少合理的git log, 不仅会损失大量分数, 还会给抄袭判定提供最有力的证据. 因此, 请你注意以下事项:

  • 请你不定期查看自己的git log, 检查是否与自己的开发过程相符.
  • 提交往届代码将被视为没有提交.
  • 不要把你的代码上传到公开的地方.
  • 总是在工程目录下进行开发, 不要在其它地方进行开发, 然后一次性将代码复制到工程目录下, 这样git将不能正确记录你的开发过程.
  • 不要修改Makefile中与开发跟踪相关的内容.
  • 不要删除我们要求创建的分支, 否则会影响我们的脚本运行, 从而影响你的成绩
  • 不要清除git log

偶然的跟踪失败不会影响你的成绩. 如果上文中的错误信息总是出现, 请尽快联系我们.

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. Then issue

make submit

This command does 2 things:

  1. Clean all unnecessary files for submission
  2. 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.

Reminder

This ends PA0. And there is no 必答题 in PA0.

results matching ""

    No results matching ""