Skip to content
/ nidhugg Public
forked from nidhugg/nidhugg

Nidhugg is a bug-finding tool which targets bugs caused by concurrency and relaxed memory consistency in concurrent programs. It is particularly useful for programs written in C/pthreads. Currently Nidhugg supports the SC, TSO, PSO, POWER and ARM (partial) memory models.

License

Notifications You must be signed in to change notification settings

kostis/nidhugg

 
 

Repository files navigation

Nidhugg is a bug-finding tool which targets bugs caused by concurrency
and relaxed memory consistency in concurrent programs. It works on the
level of LLVM internal representation, which means that it can be used
for programs written in languages such as C or C++.

Currently Nidhugg supports the SC, TSO, PSO, POWER and ARM memory
models. Target programs should use pthreads for concurrency, and each
thread should be deterministic when run in isolation.

Documentation
*************

The main documentation can be found in the git repository at
https://github.com/nidhugg/nidhugg/blob/master/doc/manual.pdf , or in
the source at doc/manual.pdf, or if installed at
PREFIX/share/doc/nidhugg/manual.pdf .

Installation
************

Nidhugg can be compiled and installed on UNIX-like systems.

Requirements
============

   1. A C++ compiler supporting C++14. For example g++ version 6 or
      higher, or clang++ version 3.8 or higher. (For compiling the
      tool Nidhugg.)

   2. The C compiler clang. (For use by Nidhugg as a frontend for C
      compilation.)

   3. Standard C/C++ libraries.

   4. GNU Autotools (autoconf & automake).

   5. Python version 3.0 or higher.

   6. The LLVM library and headers, version 3.8 or higher. If you are
      compiling the LLVM library from sources, see Compiling LLVM
      below.

   7. Foreign Function Interface (libffi) library and headers.

   8. Boost Unit Test Framework, version 1.64 or higher.

   (9). pkg-config, for generating a fully-featured configure script,
        capable of autodetecting the flags needed for libffi.
        Optional, but recommended.

   (10). Optional: Valgrind

   (11). Optional: For documentation: pdflatex
         (with packages article, inputenc, amssymb)

For Users of Debian-like Systems
--------------------------------

   On a Debian-like Linux distribution (such as e.g. Ubuntu), satisfy
   the requirements by installing the following packages:

   Mandatory:
   clang, libc6, libc6-dev, libstdc++6, autoconf, automake, python3,
   llvm, llvm-dev, libffi-dev, libboost-test-dev, libboost-system-dev,
   pkg-config

   Optional:
   valgrind, texlive-latex-base, texlive-base

Getting Nidhugg
===============

   Git users can clone Nidhugg as follows:

   $ git clone https://github.com/nidhugg/nidhugg.git

   Non git users may find it easier to download the latest revision as
   a zip archive:

   https://github.com/nidhugg/nidhugg/archive/master.zip

Basic Installation
==================

   The following assumes that you are in a shell in the main directory
   of Nidhugg.

   1. Use GNU Autotools to generate the configure script:

      $ autoreconf --install

   2. Build as usual. See Installation Options below for options about e.g.
      installation location.

      $ ./configure
      $ make

   3. Install:

      $ make install

   4. Optionally run test suite (Boost Unit Test Framework required):

      $ make test

      or (Valgrind required)

      $ make valtest

Installation Options
====================

   The configure script is built with GNU autotools, and should accept
   the usual options and environment variables. This section outlines
   some of the typical use cases.

Changing Installation Directory
-------------------------------

   The command 'make install' will install Nidhugg and its
   documentation in the directories which are standard on your
   system. To override this behavior add the switch --prefix to the
   './configure' command:

   $ ./configure --prefix=/your/desired/install/path

Specifying Compiler
-------------------

   When the configure script is invoked, it will by GNU autotools
   magic determine which C++ compiler will be used during
   compilation. In case e.g. your default compiler does not support
   C++14, but you have the compiler g++-6 installed at a
   non-standard location you may want to override this. In order to do
   so, specify the path to g++-6 in CXX when invoking the
   './configure' command:

   $ ./configure CXX='/path/to/g++-6'

Specifying LLVM Installation
----------------------------

   If the configure script does not find your LLVM installation, or
   does not find the installation which you want to use (see output
   from configure), you may specify the installation location of LLVM
   with the --with-llvm switch. Below, /path/to/llvm should be the
   path to the root of the LLVM installation, i.e. the directory which
   contains bin/llvm-config and lib/libllvm.a .

   $ ./configure --with-llvm='/path/to/llvm'

(Re)building PDF Documentation
------------------------------

  By default, make will not rebuild the documentation from LaTeX
  sources. This allows the typical user to simply use the latest PDF
  pushed to the git repository. In order to force (re)building of the
  documentation, configure with the switch
  --enable-build-documentation . This requires LaTeX to be installed.

  $ ./configure --enable-build-documentation

Compiling LLVM
==============

   If you are a user of Nidhugg, we recommend you use a release build of
   LLVM. You can also speed up the build by disabling all the backends.

   $ cd llvm-*.src
   $ mkdir build && cd build
   $ cmake -DCMAKE_INSTALL_PREFIX=/your/llvm/install/path \
           -DCMAKE_BUILD_TYPE=Release \
           -DLLVM_TARGETS_TO_BUILD="" -DLLVM_LINK_LLVM_DYLIB=ON ..
   $ make
   $ make install

Compiling LLVM for Debugging or Developing Nidhugg
**************************************************

   If you are developing Nidhugg or are troubleshooting a bug in
   Nidhugg, we recommend you use a debug build of LLVM, as this will
   allow you to build an assertions-enabled build of Nidhugg.

   $ cd llvm-*.src
   $ mkdir build && cd build
   $ cmake -DCMAKE_INSTALL_PREFIX=/your/llvm/install/path \
           -DCMAKE_BUILD_TYPE=RelWithDebInfo \
           -DLLVM_ENABLE_ASSERTIONS=ON \
           -DLLVM_TARGETS_TO_BUILD="" -DLLVM_LINK_LLVM_DYLIB=ON \
           -DLLVM_OPTIMIZED_TABLEGEN=ON ..
   $ make
   $ make install

License
*******

   Nidhugg is licensed under GPLv3. See COPYING.

   Nidhugg is free software: you can redistribute it and/or modify it
   under the terms of the GNU General Public License as published by
   the Free Software Foundation, either version 3 of the License, or
   (at your option) any later version.

   Nidhugg is distributed in the hope that it will be useful, but
   WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
   General Public License for more details.

LLVM License
============

   Part of the code in the files listed below originate in the LLVM
   Compiler Framework (http://llvm.org, version 3.4.1). Those parts
   are licensed under the University of Illinois/NCSA Open Source
   License as well as under GPLv3. See LLVMLICENSE.TXT.

   src/Interpreter.h
   src/Interpreter.cpp
   src/Execution.cpp
   src/ExternalFunctions.cpp

Contact / Bug Report
********************

   Feedback, questions or bug reports should be reported as issues on
   Github.

About

Nidhugg is a bug-finding tool which targets bugs caused by concurrency and relaxed memory consistency in concurrent programs. It is particularly useful for programs written in C/pthreads. Currently Nidhugg supports the SC, TSO, PSO, POWER and ARM (partial) memory models.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C 87.5%
  • C++ 11.5%
  • M4 0.6%
  • Python 0.2%
  • Makefile 0.1%
  • Shell 0.1%