CS264: Program Analysis

Ras Bodik, WF 11-12:30, 310 Soda

Course Description

Advanced techniques for statically analyzing programs are discussed. These techniques allow one to answer computationally hard questions about programs in an efficient albeit conservative way. They are also referred to as abstract interpretation since the algorithms interpret the program on a simplified abstract domain. The techniques are useful in compilers in order to generate more efficient code and in other programming language environments such as debuggers and code quality checkers. Time permitting, we will also study program representations and and techniques for dynamic analysis.

Textbook

• Principles of Program Analysis by Nielson, Nielson, Hankin
• research papers

Outline of Syllabus

• Techniques:
  • Dataflow analysis
  • Abstract interpretation
  • Constraint analysis
• Applications:
  • Pointer analysis
  • Bug finding
• Program Representations
  • Single Static Assignment form
  • Program Dependence Graph

Paper Summaries

You will write short summaries of the research papers covered in class, including the papers covered by guest lecturers. the summaries should be emailed to cs264 imail eecs berkeley edu. The summaries will be due before the paper is discussed in class. They will graded Pass/Fail.

Format: the summaries will have five short paragraphs:

1. Summary: what’s the problem? what’s the solution?
   • paraphrase the paper
   • offer a different spin
   • cut through the hype (if applicable)
2. Key new ideas: what’s good about the paper?
   • what new insights or techniques made this paper possible?
   • or simply, what did you like?
3. Limitations: what’s weak in the paper?
   • limitations not mentioned in the paper
   • impact of limitations, whether mentioned or not
4. What next?
• applications of the technique
• future directions
• cs264 project ideas
5. Missing background
   • which parts of the paper you didn't understand
   • background you didn't have to understand the paper
   • unfamiliar formalisms: e.g., type inference rules, semantics
   • unfamiliar techniques: e.g., model checking
   • unfamiliar lingo: e.g., happens-before relation

Schedule

Acknowledgement: Slides are adopted from Mooly Sagiv. Notes come from students in Mooly's class at Tel Aviv.

News

Page last updated: Tue, 4/12/05 5:04 PM

Friday 3/4:

• Summaries: as I mentioned in class, the default is that summaries are needed for each we read. The summary for Monday is due 1;30pm, just before the talk. Other summaries are due the day before the lecture.

Tuesday 3/1:

• HW1 has been assigned. Due in a week.

Friday 2/25:

• The project: The first bit of information on the project has been posted. Summary: find a project topic, recruit a project partner (if you need one), write a 1-2-page-long project proposal and email it to me by Sat, Mar 12.
• The homework: I was busy with the faculty candidate, so the homework is not ready yet. I'll send out an email when the homework is posted.

Tuesday 2/15:

• Bad news on the textbook: the publisher is unable to deliver the textbook before Mar 31 or so. It looks like we'll be relying on papers instead.
• If you missed the talk by Shaz Qadeer (on Jan 24), I have a CD with a video of his talk (thanks to Jimmy).

Friday 2/11:

• Link to the paper to be discussed on Wednesday has been added. Please read Sections 1-4.1 and send me your summary by Tuesday noon.

Sunday:

• Sumit Gulwani (UCB) will be the guest speaker on Friday, Feb 4. Links to his paper are on the course schedule. Yes, you will have to submit a written summary, but the rules are slightly different this time because I want you to give Sumit some feedback on his talk, which he'll take with him on his interview tour this spring. This feedback should take the form of the sixth (and seventh, eighth, ninth, if necessary) paragraph in your summary; the rest of the format stays the same. I recommend that you read the paper after his talk the way his audience will. The summaries will be forwarded to Sumit, of course, so use civil language. :-) The summaries are due Tuesday Feb 7 at noon.

Thursday:

• The schedule now contains links to past lectures and notes.
• As I suspected, the textbook is not in the bookstore yet. The promised day is Feb 3rd.

Friday:

• Copies of Chapter 1 are now available. I placed a few of them in the mailbox by my door (773 Soda).
• The bookstore will apparently have the books on Tuesday, but don't make a trip there until I can't verify the books are in stock.