2019-10-08 Introduction to Semantic Patching of C programs with Coccinelle (HSPC1W19)

Date:Tuesday, October 8, 2019, 10:00 - 17:00
Location:LRZ Building, Garching/Munich, Boltzmannstr. 1, Kursraum 2, H.U.010
Contents:

 

``Coccinelle is a program matching and transformation engine which provides the language SmPL (Semantic Patch Language) for specifying desired matches and transformations in C code. Coccinelle was initially targeted towards performing collateral evolutions in Linux. Such evolutions comprise the changes that are needed in client code in response to evolutions in library APIs, and may include modifications such as renaming a function, adding a function argument whose value is somehow context-dependent, and reorganizing a data structure. Beyond collateral evolutions, Coccinelle is successfully used (by us and others) for finding and fixing bugs in systems code.'' (http://coccinelle.lip6.fr/)

 

The maintenance of a large software project can be very demanding.
External factors like evolving third-party software library APIs, or constantly changing hardware platforms might require significant code adaptions for the code to run efficiently, or to run at all.
Failure in coping with this can lead to obsolescence, loss of performance, incompatibility, vendor lock-in, bugs.

Have you ever wondered how to detect and manipulate specified classes of C code constructs, be it for code analysis, or better, to restructure an arbitrarily large codebase according to a specified, non-trivial `pattern', without writing a C compiler ?

In this training we introduce you to a tool to do exactly this: match and restructure C codebases in a programmatic, formal way.

After this training, you shall be able to write your own code transformations, be it for a refactoring, performance improvement, paving the way to an experimental fork, or for debug/analysis reasons.

The training will also show how to analyze code looking for interesting patterns (e.g. bugs), integrate with your Python scripts to achieve the custom transformations you need, and leverage Coccinelle's limited C++ support.
 
Special mention will go to performance-oriented transformations, of interest of HPC practitioners.

Tentative program:

  • Invoking Coccinelle. Important command line options.
  • Semantic matching. Matching C language constructs.
  • Semantic patching. Modifying C language constructs.
  • Elements of the SmPL language. SmPL patch structure, rules.
  • Simple SmPL rules. Overview of matchable items.
  • Rules with inheritance. Reusing rules: multiple patches for complex transformations.
  • Scripting. Extensions and customization.
  • Review of common transformations and use cases in HPC. Examples of what to do and what NOT to do.
  • Hands-on exercises.

Contents are subject to modifications.

Prerequisites:

Good C/C++ knowledge required. HPC experience recommended.

Language:English
Teachers:Dr. Michele Martone (LRZ) (you can contact me in case of questions)
Registration:Via the LRZ registration form. ; please choose course HSPC1W19.