Code Monkey home page Code Monkey logo

install-klee's Introduction

How to build KLEE

This is a repository for all my notes about the installation of KLEE. This README.md contains a step by step manual for building KLEE with all its dependencies. Additionally, I wrote a small script for ubuntu, that automatically executes all the commands listed here. If you build KLEE for the first time, I strongly recommend executing the commands manually, but if you end up doing it again, just save your time with the script.


Manual build step by step

Introduction

There are a ton of installation instructions for KLEE out there in the web. This is yet another manual, but it tries to be a little different. First of all, it is not Ubuntu specific. It works on Ubuntu, but it also works on other distros like Arch Linux. Furthermore, this setup do not use the sudo command or any kind of installation. You get a pure local build directory full of all the necessary tools and nothing else. Thereby, you can have multiple klee versions and setups on the same machine. In order to uninstall this tools simply remove the generated build dirs.

The resulting directory structure:

build
├── klee
├── klee-uclibc
├── llvm
├── minisat
├── stp
└── z3

storage-usage

The whole files, that are needed during the build process, needs at least 2 GB of storage. This manual uses version control systems (git, svn) to download the source files. Thereby each file is stored twice: in the version control and in the checkout-folder. The version control is not really useful or necessary, so this manual removes these files with commands like rm -rf {.git,.svn}. You can leave this commands out, but remember, that this will likely double the amount of storage to at least 4 GB in total.

Usefull Links:

Step 0: Install required tools for the build

Ubuntu (16.04)

sudo apt-get install bc bison build-essential cmake curl flex git libboost-all-dev libcap-dev libncurses5-dev python-minimal python-pip subversion unzip zlib1g-dev

Arch Linux

sudo pacman -S bc bison boost cmake curl flex gcc git libcap ncurses python2 python2-pip subversion zlib

Step 1: LLVM

Checkout sourcecode of the core and relevant projects

svn co https://llvm.org/svn/llvm-project/llvm/tags/RELEASE_342/final llvm
svn co https://llvm.org/svn/llvm-project/cfe/tags/RELEASE_342/final llvm/tools/clang
svn co https://llvm.org/svn/llvm-project/compiler-rt/tags/RELEASE_342/final llvm/projects/compiler-rt
svn co https://llvm.org/svn/llvm-project/libcxx/tags/RELEASE_342/final llvm/projects/libcxx
svn co https://llvm.org/svn/llvm-project/test-suite/tags/RELEASE_342/final/ llvm/projects/test-suite

rm -rf llvm/.svn
rm -rf llvm/tools/clang/.svn
rm -rf llvm/projects/compiler-rt/.svn
rm -rf llvm/projects/libcxx/.svn
rm -rf llvm/projects/test-suite/.svn

Build the binaries

The llvm-testsuite, that is used later for make check-all needs a python2. Maybe the default on your system is python3. So you have to add the --with-python-option with your path to a python2 executable.

cd llvm
./configure --enable-optimized --disable-assertions --enable-targets=host --with-python="/usr/bin/python2"
make -j `nproc`

make -j `nproc` check-all
cd ..

Step 2: Minisat

git clone --depth 1 https://github.com/stp/minisat.git
# Commit ID: 37dc6c67e2af26379d88ce349eb9c4c6160e8543 (more than 2 years old)
rm -rf minisat/.git

cd minisat
make
cd ..

Step 3: STP

git clone --depth 1 --branch 2.1.2 https://github.com/stp/stp.git
rm -rf stp/.git

cd stp
mkdir build
cd build
cmake \
 -DBUILD_SHARED_LIBS:BOOL=OFF \
 -DENABLE_PYTHON_INTERFACE:BOOL=OFF \
 -DMINISAT_INCLUDE_DIR="../../minisat/" \
 -DMINISAT_LIBRARY="../../minisat/build/release/lib/libminisat.a" \
 -DCMAKE_BUILD_TYPE="Release" \
 -DTUNE_NATIVE:BOOL=ON ..
make -j `nproc`
cd ../..

Step 4: uclibc and the POSIX environment model

git clone --depth 1 --branch klee_uclibc_v1.0.0 https://github.com/klee/klee-uclibc.git
rm -rf klee-uclibc/.git

cd klee-uclibc
./configure \
 --make-llvm-lib \
 --with-llvm-config="../llvm/Release/bin/llvm-config" \
 --with-cc="../llvm/Release/bin/clang"
make -j `nproc`
cd ..

Step 5: Z3

git clone --depth 1 --branch z3-4.4.1 https://github.com/Z3Prover/z3.git
rm -rf z3/.git

cd z3
python scripts/mk_make.py
cd build
make -j `nproc`

# partialy copied from make install target
mkdir -p ./include
mkdir -p ./lib
cp ../src/api/z3.h ./include/z3.h
cp ../src/api/z3_v1.h ./include/z3_v1.h
cp ../src/api/z3_macros.h ./include/z3_macros.h
cp ../src/api/z3_api.h ./include/z3_api.h
cp ../src/api/z3_algebraic.h ./include/z3_algebraic.h
cp ../src/api/z3_polynomial.h ./include/z3_polynomial.h
cp ../src/api/z3_rcf.h ./include/z3_rcf.h
cp ../src/api/z3_interp.h ./include/z3_interp.h
cp ../src/api/z3_fpa.h ./include/z3_fpa.h
cp libz3.so ./lib/libz3.so
cp ../src/api/c++/z3++.h ./include/z3++.h

cd ../..

Step 6: KLEE

This is the only step in the setup, where we need absolute path in the variables. I prefer having my self-compiled binaries in a build-folder inside my home directory, but you are free to place it wherever you want. Simply replace /home/user/build/ with anything, that fits your needs.

git clone --depth 1 --branch v1.2.0 https://github.com/klee/klee.git
rm -rf klee/.git

cd klee
./configure \
 LDFLAGS="-L/home/user/build/minisat/build/release/lib/" \
 --with-llvm=/home/user/build/llvm/ \
 --with-llvmcc=/home/user/build/llvm/Release/bin/clang \
 --with-llvmcxx=/home/user/build/llvm/Release/bin/clang++ \
 --with-stp=/home/user/build/stp/build/ \
 --with-uclibc=/home/user/build/klee-uclibc \
 --with-z3=/home/user/build/z3/build/ \
 --enable-posix-runtime

make -j `nproc` ENABLE_OPTIMIZED=1

# Copy Z3 libraries to a place, where klee can find them
cp ../z3/build/lib/libz3.so ./Release+Asserts/lib/

make -j `nproc` check
cd ..

A small note: I have tried this setup on several systems and this last check has never finished without errors. I am not absolutely sure, if this is normal or not. From my experience around 4 up to 9 failing test cases seems to be normal, if everything seems to work.

Step 7: Link some executables

TODO: Link KLEE to ~/bin/

TODO: error while loading shared libraries: libkleeRuntest.so.1.0: cannot open shared object file: No such file or directory -> ln -s libkleeRuntest.so libkleeRuntest.so.1.0

Solution for common errors

During the ./configure command of Klee

checking for vc_setInterfaceFlags in -lstp... no
Could not link with libstp
checking for vc_setInterfaceFlags in -lstp... no
configure: error: Unable to link with libstp. Check config.log to see what went wrong

and in the corresponding config.log

configure:5121: checking for vc_setInterfaceFlags in -lstp
configure:5146: g++ -o conftest -g -O2   conftest.cpp -lstp -L.../stp/build/lib -lminisat   >&5
.../stp/build/lib/libstp.a(RunTimes.cpp.o): In function `RunTimes::getDifference[abi:cxx11]()':
.../stp/build/../lib/AST/RunTimes.cpp:118: undefined reference to `Minisat::memUsed()'
...

In other words, the compiler cannot find a lot of minisat functions. This problem is caused by the shared library for minisat, that must be added and must be found during the compilation process. Make sure, that you are giving the correct path to minisat in the LDFLAGS. See step 6 for details.

During runs of klee

.../bin/klee: error while loading shared libraries: libz3.so: cannot open shared object file: No such file or directory

Klee cannot find the libz3.so library of Z3. The easiest solution is to copy the library to the lib directory of Klee. See step 6 for details.

install-klee's People

Watchers

James Cloos avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.