Skip to content

ualberta-smr/varclang

 
 

Repository files navigation

Variability-Aware Clang

A variability-aware analysis tool built on top of the clang (release 6.0) source code.

This was developed as part of SMR in collaboration with IBM.

Team: Jacob Reckhard, Ryan Shukla, and Sarah Nadi
Emails: [email protected], [email protected], [email protected]

What it is

This is a tool for performing a variability-aware analysis. What it does is it takes your source code, and it looks through it for errors that could occur due to variability. It's to be used in a similar manner to your regular compiler, except instead of compiling your code to an executable, it will instead search it for possible bugs.

Why do we need this

Take a look at the following code snippet

#ifdef Animals
    Item cat = 10.99;
    Item dog = 13.99;
    Item mouse = 9.99;
#endif
#ifdef Computers
    Item keyboard = 34.99;
    Item monitor = 99.99;
    Item mouse = 4.99;
#endif

Lets say this is your web store. It all looks good, if you want to sell animals you can, if you want to sell computer parts, that works fine too. However, if you try and sell both at once, your code doesn't even compile. In this simple example it's an easy fix, however, it could be a non trivial problem. The worst part is, it's not immediately obvious. This compile error might not come up until much later when you have written lots of other code depending on the erroneous code.

How to Build

In order to build clang, you need to have llvm, so go and clone that. We only need the one branch.
$ git clone --single-branch -b release_60 https://github.com/llvm-mirror/llvm.git llvm
Then clone this project into the proper spot in the llvm source tree
$ git clone https://github.com/jacobrec/varclang.git llvm/tools/clang
Then make and enter a build directory. In source builds are not supported
$ mkdir build && cd build
Then run cmake to generate build files
$ cmake -G "Unix Makefiles" ../llvm
Finally, build the project
$ make
Since this is just a modified version of clang, the official documentation is a great place to go for build options.

The cmake command I used was

CXX=clang++ CC=clang cmake -DCMAKE_BUILD_TYPE=DEBUG -DLLVM_INCLUDE_TESTS=OFF -DLLVM_USE_LINKER=lld -DLLVM_TARGETS_TO_BUILD=X86 -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DLLVM_INCLUDE_EXAMPLES=OFF -DLLVM_BUILD_EXAMPLES=OFF -DLLVM_BUILD_TESTS=OFF -G "Ninja" ../.llvm

How to Use

Now that you have built the project. If you go into build/bin, there will be a whole bunch of binaries. For this, there are two important ones. clang and clang++.

Lets say you have a c++ file you want to analyze. If this file is named file.cpp. You would type

build/bin/clang++ -fsyntax-only file.cpp

This will do the variability aware analysis with every ifdef it reaches. If you include any header files from the standard library, this can be a problem, as you probably don't want to do your analysis on all of the standard library, you just care about your code. So, you can create a config file to specify the macros to perform variability aware analysis on.

This config file is simple. It is just a newline separated list of macro names to analyze.

To specify you want to use your config file, you do this.
build/bin/clang++ -fsyntax-only -Xclang -var-config=my_config_file file.cpp
It is important that the flag -Xclang immediately precedes the -var-config flag.

For C programs, it is the same, except you use build/bin/clang instead of build/bin/clang++

See the wiki for more info

The wiki contains an overview on what was modified from the original clang codebase. It also documents design decisions and known limitations.

Info on clang

Information on Clang: http://clang.llvm.org/ Building and using Clang: http://clang.llvm.org/get_started.html Information on the LLVM project: http://llvm.org/

About

Mirror of official clang git repository located at http://llvm.org/git/clang. Updated every five minutes.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C++ 71.0%
  • C 18.3%
  • Objective-C 6.2%
  • Objective-C++ 1.9%
  • HTML 1.1%
  • Python 0.8%
  • Other 0.7%