Fall 2014 SPO600 Weekly Schedule

It's Alive!
This SPO600 weekly schedule will be updated as the course proceeds - dates and content are subject to change. The cells in the summary table will be linked to relevant resources and labs as the course progresses.

Summary Table

This is a summary/index table. Please follow the links in each cell for additional detail -- especially for the Deliverables column.

Week	Week of...	Tuesday	Friday	Deliverables (Summary - click for details)
1	Sep 1	(Experience Seneca Orientation - No Class)	Introduction to Software Porting, Portability, Benchmarking, and Optimization / How is code accepted into an open source project?	Set up accounts and a Fedora system
2	Sep 8	Computer Architecture Overview	Compiled C Lab	Blog a commentary on code reviews in two communities (Lab 1) and on the Compiled C Code lab (Lab 2).
3	Sep 15	Linaro Connect 2014 - No classes scheduled - Prepare your presentation on assembly language code		Assembly language presentation ready for presentation next Tuesday (September 23).
4	Sept 22	Assembly language presentations & Assembly Basics	Assembly language presentations, continued	Blog the content of your presentation
5	Sep 29	Assembly lab (Lab 3)	Inline assembler and compiler optimizations / Potential Project Triage – Scan the potential project list from the Linaro Performance site	Blog post about assembly lab (lab 3) and potential projects
6	Oct 6	Guest Speaker: Jon "Maddog" Hall	Codebase analysis / Memory barriers and Atomics	Analyze the platform-specific code in 3 packages.
7	Oct 13	Architecture-specific Code - Why?	Compiler Intrinsics and Picking your Package	Pick your package and blog about it.
Study Week	Oct 20	Study Week FSOSS 2014 on Thursday-Friday
8	Oct 27	Working with the Code	Group hack session - Profiling	Blog about your progress.
9	Nov 3	Profiling review	Group hack session	Post baseline stats for your software.
10	Nov 10	Presentations	AArch64 on x86_64 / Hack session	Blog about your project status - 1st project marks.
11	Nov 17	Optimizing Code	Group hack session - Profiling and optimizing	Post a code review update.
12	Nov 24	Using complier optimizations	Project Work	Post a code review update.
13	Dec 1	Final Presentations	Final Presentations	Code accepted upstream. Blog about it!
Exam Week	Dec 8	Exam Week - No exam in this course!

Evaluation

Category	Percentage	Evaluation Dates
Communication	20%	September 30, October 31, November 21, December 10
Quizzes	10%	May be held during any class, usually at the start of class. A minimum of 5 one-page quizzes will be given. No make-up/retake option is offered if you miss a quiz. Lowest 3 scores will not be counted.
Labs	10%	See deliverables column above.
Project work	60%	November 11 (15%), November 25 (20%), December 10 (25%)

Week 1

Friday (Sep 5)

Introduction to the Problem

• Most software is written in a high-level language which can be compiled into machine code for a specific architecture. However, there is a lot of existing code that contains some architecture-specific code fragments written in Assembly Language (or, in some cases, machine-specific high-level code).
• Reasons for writing code in Assembly Langauge include:
  • Performance
  • Atomic Operations
  • Direct access to hardware features, e.g., CPUID registers
• Most of the historical reasons for including assembler are no longer valid. Modern compilers can out-perform most hand-optimized assembly code, atomic operations can be handled by libraries or compiler intrinsics, and most hardware access should be performed through the operating system or appropriate libraries.
• A new architecture has appeared: Aarch64, which is part of ARMv8. This is the first new computer architecture to appear in several years (at least, the first mainstream computer architecture).
• There are over 1400 software packages/modules present in GNU Linux systems which contain architecture-specific assembly language code or have other portability issues. Most of these packages cannot be built on Aarch64 systems without modification.

Course Projects

In this course, you will:

1. Select two software packages from a list compiled by Steve Macintyre of Linaro. Each of the packages on this list contains assembly language code which is platform-specific.
2. Prepare a fix/patch for the software so that it will run on 64-bit ARM systems (aarch64). This may be done at either of two levels:
   1. Port - Add additional assembly language code for aarch64 (basic solution).
   2. Make Portable - Remove architecture-specific code, replacing it with compiler intrinsics or high-level code so that the software will successfully build on multiple platforms.
3. Benchmark - Prove that your changes do not cause a performance regression on existing platforms, and that (ideally) it improves performance.
4. Upstream your Code - Submitting your code to the upstream (originating) software project so that it can be incorporated into future versions of the software. This will involve going through a code review to ensure that your code is compatible with and acceptable to the upstream community.

General Course Information

• Course resources are linked from the CDOT wiki, starting at http://zenit.senecac.on.ca/wiki/index.php/SPO600 (Quick find: This page will usually be Google's top result for a search on "SPO600").
• Coursework is submitted by blogging.
• Quizzes will be short (1 page) and will be held without announcement at any time. Your lowest three quiz scores will not be counted, so do not worry if you miss one or two.
• Course marks (see Weekly Schedule for dates):
  • 60% - Project Deliverables
  • 20% - Communication (Blog and Wiki writing)
  • 20% - Labs and Quizzes (10% labs - completed/not completed; 10% for quizzes - lowest 3 scores not counted)
• All classes will be held in an Active Learning Classroom -- you are encouraged to bring your own laptop to class. If you do not have a laptop, consider signing one out of the Learning Commons for class, or using a smartphone with an HDMI adapter.
• For more course information, refer to the SPO600 Weekly Schedule (this page), the Course Outline, and SPO600 Course Policies.

Discussion of how open source communities work

(Background for the Code Review Lab (Lab 1) which is homework due Week 2).

Week 1 Deliverables

1. Set up your SPO600 Communication Tools - in particular, set up a blog and add it to Planet CDOT (via the Planet CDOT Feed List).
2. Add yourself to the Fall 2014 SPO600 Participants page (leave the projects columns blank).
3. Generate a pair of keys for SSH and email the public key to your professor.
4. Sign and return the Open Source Professional Option Student Agreement.
5. Set up a Fedora 20 system.

Week 2

Tuesday (Sep 9)

• Computer Architecture overview (see also the Computer Architecture Category)

Friday (Sep 12)

Bring Your Laptop
Classes are held in a Active Learning Classroom. If you have a laptop or other device with a VGA or HDMI output (such as a smartphone!) please bring it. You'll need either a local linux environment or an SSH client -- which is built-in to Linux, Mac, and Chromebook systems, and readily available for Windows, Android, and iOS devices.

• Perform the Compiled C Lab (Lab 2) in groups

Week 2 Deliverables

• Complete and blog your conclusion to the Code Review Lab (Lab 1).
• Blog your conclusion to the Compiled C Lab (Lab 2)

Week 3

This week your professor is at Linaro Connect, an engineering conference run by Linaro - a distributed not-for-profit collaborative technology company focused on Linux on ARM. You can participate remotely and may find some of the sessions interesting.

• Select and prepare to teach the class about a specific small topic related to assembly language / machine language programming.

Week 3 Deliverables

• Be prepared to give your presentation on Tuesday of next week (September 23).

Week 4

Tuesday (Sep 23)

• Give your presentation about an assembly language topic - teach your SPO600 colleagues

Friday (Sep 26)

• Remainder of the presentations about an assembly language topic
• Introduction to assembly language
• Writing x86_64 and aarch64 code in the Assembly language lab (Lab 3)

Week 4 Deliverables

• Blog your presentation

Week 5

Tuesday (Sep 30)

• Assembler Lab

Friday (Oct 3)

• Inline Assembly Language
• Compiler Optimizations

The Linaro Performance Challenge is a project initiated by Jon "Maddog" Hall and Linaro to port or make portable open source software packages which contain platform-specific code and which may not build on the new Aarch64 architecture.

(There are two videos about the challenge, both of which are from late 2013 while the program was being finalized: One of an interview with Maddog and Steve Macintyre, and the other of a Linaro Connect presentation by Maddog).

The list of packages for this project was originally developed by the UK Debian developer Steve Macintyre, who works for Linaro. His focus in developing that list was to find packages that contained assembly language code for x86 (or other platforms) and which did not have assembly language code for ARM, especially Aarch64. As a result, the list includes many false-positives: there are many packages on there that can successfully build on ARM, either due to C work-arounds for the missing assembly code, or conditional compilation of the assembly code, or recent updates to the software, or other reasons.

In this class, we're going to start to triage this list by analyzing which packages exist in the Fedora package set and which have been successfully built for the aarch64 architecture.

The list has been processed with these steps:

1. Grab the list of packages from the Linaro Performance Challenge site.
2. Find out the corresponding names of the packages in the Fedora package set. Most of these will be the same as in the list, but some may be different.
3. Find out which packages have not been built for Fedora on aarch64
4. Divide the resulting list up between the members of the class for further analysis

Please see the Fall 2014 SPO600 Packages by Participant and perform the steps listed there.

Participating in the Linaro Performance Challenge
You are invited to participate in the Linaro Performance Challenge directly, utilizing the work you are doing in the SPO600 course. Doing so may enable you to receive a prize for participation and the chance to win a trip to a Linaro Connect conference. In order to participate in the Challenge, you will need to comply with the technical requirements and rules of the Challenge.

Your participation in this course, and the mark you receive in this course, are independent of your participation in the Linaro Performance Challenge.

In other words: Linaro and Seneca are distinct entities, and although you can participate in both the Challenge and SPO600 with the same project, Seneca assumes no responsibility for your interaction with Linaro, and vice-versa.

Week 5 Deliverables

• Blog about the assembler lab (Lab 3).
• Do the tasks assigned on the Fall 2014 SPO600 Packages by Participant page, and blog about it.

Week 6

Tuesday (Oct 7)

Jon "Maddog" Hall will be joining us for a remote talk and Q&A via Google Hangout.

Friday (Oct 10)

In groups, we'll be analyzing software pacakges from the Packages by Participant list to find the platform-specific code and build instructions.

Week 6 Deliverables

• Pick three additional packages (not the ones done in class) from your section of the Packages by Participant list and find the platform-specific code (or build instructions). Figure out what that code does, and document that on the list page. Blog about your results and your reflections on the task.

Week 7

Tuesday (Oct 14)

Discussion of some of the reasons that platform specific (usually assembly language) code is used in software

Memory Barriers

Memory Barriers ensure that memory accesses are sequenced so that multiple threads, processes, cores, or IO devices see a predictable view of memory.

• Leif Lindholm provides an excellent explanation of memory barriers.
  • Blog series - I recommend this series, especially the introduction, as a very clear explanation of memory barrier issues.
    • Part 1 - Memory Access Ordering - An Introduction
    • Part 2 - Memory Access Ordering Part 2 - Barriers and the Linux Kernel
    • Part 3 - Memory Access Ordering Part 3 - Memory Access Ordering in the ARM Architecture
  • Presentation at Embedded Linux Conference 2010 (Note: Acquire/Release in C++11 and ARMv8 aarch64 appeared after this presentation):
    • Slides
    • Video
• Memory Barriers - A Hardware View for Software Hackers - This is a highly-rated paper that explains memory barrier issues - as the title suggests, it is designed to describe the hardware origin of the problem to software developers. Despite the fact that it is an introduction to the topic, it is still very technical.
• ARM Technical Support Knowlege Article - In what situations might I need to insert memory barrier instructions? - Note that there are some additional mechanisms present in ARMv8 aarch64, including Acquire/Release.
• Kernel Documentation on Memory Barriers - discusses the memory barrier issue generally, and the solutions used within the Linux kernel. This is part of the kernel documentation.
• Acquire-Release mechanisms
  • MSDN Blog Post with a very clear explanation of Acquire-Release.
  • Preshing on Programming post with a good explanation.
  • ARMv8 Instruction Set Architecture Manual (ARM InfoCentre registration required) - See the section on Acquire/Release and Load/Store, especially Load/Store Exclusive (e.g., LDREX)

Atomics

Atomics are operations which must be completed in a single step (or appear to be completed in a single step) without potential interruption.

• Wikipedia has a good basic overview of the need for atomicity in the article on Linerarizability

Friday (Oct 17)

• Compiler Intrinsics
  • The use of compiler intrinsics (e.g., those in gcc, or in another compiler) locks you into the use of that specific compiler (or another that supports the exact same intrinsics), but it provides platform portability. It is better to use language features (e.g., C11 or C++11 atomic and acquire/release syntax) where possible, but since that is often not possible, the use of compiler intrinsics is more maintainable than inline assembly.
  • GCC provides intrinsics (built-in functions) for atomic operations, as documented in the GCC manual:
    • Legacy __sync Built-in Functions for Atomic Memory Access
    • Built-in functions for memory model aware atomic operations
  • The Fedora project has some guidelines/recommendations for the use of these GCC builtins:
    • http://fedoraproject.org/wiki/Architectures/ARM/GCCBuiltInAtomicOperations
• Select your project(s):
  • Criteria:
    1. Interest
    2. Scope
    3. Fit between your skills and project needs
    4. Upstream status (e.g., dead project)

Week 7 Deliverables

1. Select at least two software packages from the Linaro performance web site and/or the Packages page.
2. Record your choice on:
   1. The Participants page - so that your colleagues in class know that you're working on the package(s). Note: Package choices will be approved by your professor, but will usually be accepted on a first-come, first-served basis.
   2. The Linaro performance web site - this will tell other people outside of our group within the Linux-on-ARM community that you are working on the package(s).
3. Investigate and blog about your choice.

Over reading week:

1. Contact the upstream communities for the packages you have selected -- so that they know that you are working on the package and to open a channel of communication for your forthcoming patches (or benchmarks, or other results).
2. Formulate a work plan that will conclude with landing your software changes in the upstream software before the end of the course.
3. Blog about your work plan and what you need to investigate/learn in order to complete your project.

Week 8

Tuesday (Oct 28)

Working with the Code

• Working with GIT
• Working with other version control systems

Looking at How Distributions Package the Code

• Using fedpkg

Friday (Oct 31)

• Benchmarking and Profiling

• Profiling with gprof
  • Build with profiling enabled (use the option -pg with both gcc and ld)
  • Run the profile-enabled executable
  • Analyze the data in the gmon.out file
    • gprof nameOfBinary # Displays text profile including call graph
    • gprof nameOfBinary | gprof2dot | dot | display - # Displays visualization of call graph

Resources

• GProf Manual
• Profiling with GProf

Week 8 Deliverables

• Blog about your progress connecting with the communities associated with your projects and working with the code.

= Tuesday (Nov 4)

• Profiling/baseline benchmarking review
• Hacking session

Friday (Nov 7)

• Hacking session

Week 9 Deliverables

• Blog about your baseline benchmarking/profiling.

Week 10

Tuesday (Nov 11)

• Presentation on your project status

Friday (Nov 14)

• Running AArch64 code on x86
  • Foundation models
  • QEMU Userspace emulation

= Week 10 Deliverables

• Blog about your project status