-
Notifications
You must be signed in to change notification settings - Fork 34
Building
GitHub will offer you the download options "Source code (zip)" and "Source code (tar.gz)". These are copies of the source, as-is, in the git repository and not suited for distribution. Instead it is highly recommend to use the provided source distribution package, which contains all the necessary dependencies.
The libfvde source code can be build with different compilers:
- Using GNU Compiler Collection (GCC)
- Using Minimalist GNU for Windows (MinGW)
- Using Microsoft Visual Studio
Or directly packaged with different package managers:
- Using Debian package tools (DEB)
- Using RedHat package tools (RPM)
- Using macOS pkgbuild
- Using setup.py
To retrieve the source distribution package go to the downloads page and download the file named:
libfvde-experimental-<version>.tar.gz
To extract the source distribution package run:
tar xfv libfvde-experimental-<version>.tar.gz
This will create the source directory:
libfvde-<version>
Note that the git repository holds the development version of the source code.
If you are not developing it is recommended that you download a source distribution package instead.
To retrieve the source from the git repository make sure to install:
- git
- aclocal
- autoconf
- automake
- autopoint or gettextize
- libtoolize
- pkg-config
- flex
- byacc
To download and prepare the source for building run:
git clone https://github.com/libyal/libfvde.git
cd libfvde/
./synclibs.sh
./autogen.sh
The source from the git repository will not work without synchronizing the library dependencies "./synclibs.sh" and having the autotools generate the necessary files "./autogen.sh".
On Debian based Linux distributions such as Ubuntu install the following packages:
sudo apt install git autoconf automake autopoint libtool pkg-config flex bison
On Fedora based Linux distributions install the following packages:
sudo dnf install git autoconf automake gettext-devel libtool pkg-config flex byacc
On macOS the XCode command line tools no longer provide the autotools you can obtain them e.g. via MacPorts. Install the following packages:
sudo port install git autoconf automake gettext libtool pkgconfig flex byacc
Since Microsoft Visual Studio does not come with a Unix shell or equivalent you'll need to use the Windows PowerShell scripts instead:
git clone https://github.com/libyal/libfvde.git
cd libfvde\
.\synclibs.ps1
.\syncwinflexbison.ps1
.\synczlib.ps1
.\autogen.ps1
Before you build the libfvde source code using the GNU Compiler Collection (GCC) you'll need to have compilation and build tools installed.
- On a Linux system make sure you have build-essential (Debian-based) or the Development Tools (RedHat-based) packages installed.
- On a macOS system make sure you have XCode (with command line tools) or MacPorts (or equivalent) installed.
Also make sure to have the following dependencies including source headers installed:
- flex
- byacc
- zlib (for DEFLATE compression support) (optional but recommended, can be disabled by --with-zlib=no)
- libcrypto (part of OpenSSL) (optional but recommended, can be disabled by --with-openssl=no)
If you want to be able to use fvdemount, make sure that:
- on a Linux system you have libfuse-dev (Debian-based) or fuse-devel (RedHat-based) installed.
- on a macOS system, you have OSXFuse (http://osxfuse.github.com/) installed.
To build the libfvde source code change into the source directory and run the following commands:
./configure
make
You can install the binaries that were build by running:
sudo make install
By default this will install the binaries in /usr/local. If you want to change this to e.g. /usr, add the configuration option --prefix=/usr, e.g.
./configure --prefix=/usr
On Linux make sure libfvde.so is in the library cache. Normally it suffices to run:
sudo ldconfig
To troubleshoot issues or for low-level format analysis libfvde supports verbose and debug output.
To enable verbose and debug output support add --enable-verbose-output and --enable-debug-output to configure, e.g.
./configure --enable-verbose-output --enable-debug-output
This will generate vast amounts of debug information on stderr when the tools are run with -v.
To make a static library add --enable-shared=no to configure, .e.g:
./configure --enable-shared=no
Some distributions provide separate packages for static versions of libraries. Make sure you have a static versions of:
- glibc
- zlib (for DEFLATE compression support) (optional but recommended, can be disabled by --with-zlib=no)
- libcrypto (part of OpenSSL) (optional but recommended, can be disabled by --with-openssl=no)
- fuse (optional, can be disabled by --with-libfuse=no)
To make static executables add --enable-static-executables=yes to configure, .e.g:
./configure --enable-static-executables=yes
Check if the executables can be run.
Disable multi-thread support with --enable-multi-threading-support=no
if you get an error similar to:
libcthreads_read_write_lock_grab_for_write: unable to lock read/write lock for write with error: Deadlock condition detected.
To compile the Python-bindings you'll need the Python development files.
Make sure that on:
- a Linux machine, you have python3-dev (Debian-based) or python3-devel (RedHat-based) installed.
- a macOS machine, XCode if you're using a different Python source tree see macOS notes below.
Adding --enable-python when running ./configure will enable the Python-bindings to be build.
./configure --enable-python
If you want to use Cygwin to build libfvde make sure to have the following packages installed:
- autoconf
- automake
- binutils
- gcc-core
- gcc-g++
- gettext-devel
- libiconv
- libtool
- make
- pkg-config
- flex
- byacc
- zlib-devel (for DEFLATE compression support) (optional but recommended, can be disabled by --with-zlib=no)
- libssl-devel (optional but recommended, can be disabled by --with-openssl=no)
After following the GNU Compiler Collection (GCC) build instructions you should end up with the following DLL:
libfvde/.libs/cygfvde-0.dll
And the following executables:
fvdetools/.libs/fvdeinfo.exe
fvdetools/.libs/fvdemount.exe
fvdetools/.libs/fvdewipekey.exe
Make sure you use define LIBFVDE_DLL_IMPORT before including <libfvde.h>.
To be able to use the DLL on other systems than the build system you'll also need to provide:
- cygwin1.dll
If you find that libtoolize is missing use glibtoolize instead.
macOS comes with multiple platform specific build Python versions. If you install a different version of Python on macOS make sure that when building and running the Python-bindings you're using the correct version otherwise it will likely result in one of the following errors:
Fatal Python error: Interpreter not initialized (version mismatch?)
Fatal Python error: PyThreadState_Get: no current thread
To build against a specific version you can tell configure where to look:
CFLAGS=-I/Library/Frameworks/Python.framework/Versions/2.7/include/ \
LDFLAGS=-L/Library/Frameworks/Python.framework/Versions/2.7/lib/ \
./configure --enable-python --with-pyprefix
Adding --with-pyprefix tells configure to use Python to determine where to install the Python-bindings.
To build libfvde on Sun Solaris make sure that /usr/ccs/bin and /usr/sfw/bin are defined in the PATH environment variable.
To compile libfvde using MinGW you'll need:
- MinGW
- flex
- byacc
- MinGW build of zlib library and source headers (for DEFLATE compression support) (optional but recommended, can be disabled by --with-zlib=no)
To build use:
mingw32-configure --prefix=/opt/local/i386-mingw32 --enable-winapi=yes
mingw32-make
It is recommended that you use WINAPI support but it is possible to compile libfvde without it (--enable-winapi=no). The default behavior is that configure will try to auto-detect MinGW and enable WINAPI support.
If mingw32-configure and mingw32-make are not available you can build it with:
./configure --host=i386-mingw32 --prefix=/opt/local/i386-mingw32 --enable-winapi=yes
make
If this does not work try a script similar to the following:
#!/bin/sh
CC=/opt/local/bin/i386-mingw32-gcc
CXX=/opt/local/bin/i386-mingw32-g++
AR=/opt/local/bin/i386-mingw32-ar
OBJDUMP=/opt/local/bin/i386-mingw32-objdump
RANLIB=/opt/local/bin/i386-mingw32-ranlib
STRIP=/opt/local/bin/i386-mingw32-strip
MINGWFLAGS="-mwin32 -mconsole -march=i586 "
CFLAGS="$MINGWFLAGS"
CXXFLAGS="$MINGWFLAGS"
CC=$CC CXX=$CXX AR=$AR OBJDUMP=$OBJDUMP RANLIB=$RANLIB STRIP=$STRIP ./configure --host=i586-mingw32msvc --prefix=/opt/local/i386-mingw32 --enable-winapi=yes
CC=$CC CXX=$CXX AR=$AR OBJDUMP=$OBJDUMP RANLIB=$RANLIB STRIP=$STRIP CFLAGS="$CFLAGS" CXXFLAGS="$CXXFLAGS" make
If you get compiler errors like:
#error WINAPI file open function for Windows 2000 or earlier NOT implemented yet
That means WINVER is not set or set to a version predating Windows XP (0x0501) and you'll have to set WINVER manually like:
CFLAGS=-DWINVER=0x0501 ./configure --host=i386-mingw32 --enable-winapi=yes
You should end up with the following DLL:
libfvde/.libs/libfvde-1.dll
And the following executables:
fvdetools/.libs/fvdeinfo.exe
fvdetools/.libs/fvdemount.exe
fvdetools/.libs/fvdewipekey.exe
To install libfvde and tools in the MinGW build tree use:
sudo make install
Make sure you use define LIBFVDE_DLL_IMPORT before including <libfvde.h>.
To be able to use the DLL on other systems than the build system you'll also need to provide:
- libgcc_s_dw2-1.dll (or equivalent)
While running make I get an error similar to the following:
libclocale_locale.c: In function 'libclocal_local_get_decimal_point':
libclocale_locale.c:357:2: warning implicit declaration of function 'GetLocaleInfoEx' [-Wimplicit-function-declaration]
libclocale_locale.c:358:7: error: 'LOCALE_NAME_USER_DEFAULT' undeclared (first use in this function)
libclocale_locale.c:358:7: note: each undeclared identifier is reported only once for every function it appears in
The version of MinGW does not support a WINAPI version of Vista or later (0x0600) try setting WINVER to 0x0501.
Before you build libfvde using Microsoft Visual Studio you'll need to have it installed. The libfvde packages comes with Microsoft Visual Studio files for version 2008. Version 2010 is able to convert these files into its newer versions.
The Microsoft Visual Studio express version is sufficient. Note that if you want to build 64-bit version with the express version you'll need at least 2010. Also see the section: Microsoft Visual Studio 2010 express and 64-bit compilation.
Note that if you want to build libfvde from source checked out of git with Visual Studio make sure the autotools are able to make a distribution package of libfvde before trying to build it. You can create distribution package by running: "make dist".
To compile libfvde using Microsoft Visual Studio you'll need:
- zlib (for DEFLATE compression support)
If you want to be able to use fvdemount you'll need Dokan library see the corresponding section below. Otherwise ignore or remove the dokan_dll and fvdemount Visual Studio project files.
To troubleshoot issues or for low-level format analysis libfvde supports verbose and debug output.
To enable verbose and debug output support edit:
common\config_winapi.h
Add the following definitions:
#define HAVE_VERBOSE_OUTPUT 1
#define HAVE_DEBUG_OUTPUT 1
Download the latest zlib source distribution package including from the zlib project site: https://zlib.net/.
Extract the contents of the zlib source distribution package to the directory containing the libfvde-<version>
directory.
Normally the extracted zlib source directory will have a version number e.g. zlib-1.2.8, to remain version independent rename the directory to zlib.
Get the Dokan from: http://dokan-dev.github.io and install the library and driver.
To compile the pyfvde Python-bindings you'll need the Python development files which are part of the Python.org Windows installer package.
The Microsoft Visual Studio files are configured to look in:
C:\Python2.7\
If you want to have the Python development files in a different location you'll have to change the pyfvde Microsoft Visual Studio settings.
Note that the VSDebug build of the Python-bindings will fail since the Python development files do not come with python2.7_d.lib.
If you do not intend to build the Python-bindings you can safely ignore build errors regarding: pyfvde.
Open the file:
msvscpp\libfvde.sln
Note that the project files contain a Release and VSDebug configuration. The VSDebug builds the binaries with debug information. Note that this is not the same as debug output.
Make sure to check if your build environment defines the correct WINVER for your build. The code uses WINAPI version specific functions based on WINVER. You can define a custom WINVER in the Microsoft Visual Studio C++ project files or in common\config_winapi.h
And build the solution. The build files will be places in:
msvscpp\Release\
Another way to build libfvde with Visual Studio is to use MSBuild via the command line. MSBuild can be installed as part of the Microsoft.NET Framework.
First set-up the Visual Studio variables:
C:\Program Files\Microsoft Visual Studio 9.0\VC\bin\vcvars32.bat
Next run MSBuild:
msbuild msvscpp\libfvde.sln /p:Configuration=Release;Platform=Win32
Make sure you use define LIBFVDE_DLL_IMPORT before including <libfvde.h>.
To be able to use the DLL on other systems than the build system you'll also need to provide:
- the Microsoft Visual C++ Run-Time DLL (part of the Visual C++ Redistributable package) libfvde.dll is dependent on:
- zlib.dll
These DLLs can be found in the same directory as libfvde.dll.
To build a 64-bit version of libfvde with Microsoft Visual Studio express you'll need at least the 2010 version.
First make sure to enabling 64-bit compilation support on Microsoft Visual Studio. For Visual Studio 2010 express this can be a tedious process, some relevant links:
- http://msdn.microsoft.com/en-us/library/vstudio/9yb4317s(v=vs.100).aspx
- http://support.microsoft.com/kb/2519277
If you have set up Visual Studio correctly the following should work:
Go to:
Configuration manager -> Active solution platform
Select "<New>
"
- Type or select the new platform: "x64"
- Copy settings from: "Win32"
- Create new project platforms: enabled
Additionally for every project change:
Configuration Properties -> General -> Platform Toolset
Into "Windows7.1SDK" or equivalent.
If you've Python installed on your Visual Studio build machine you can try
running msvscpp-convert
to convert the Visual Studio 2008 solution files to 2010 format and add the x64
settings for you with the option --extend-with-x64
.
msvscpp-convert.py --extend-with-x64 --output-format 2010 msvscpp\libfvde.sln
This will create a subdirectory named vs2010 which contains the Visual Studio 2010 solution files.
To build libfvde using the Debian package tools make sure you have the following packages installed:
sudo apt install autotools-dev build-essential debhelper dh-autoreconf dh-python fakeroot pkg-config flex bison zlib1g-dev libssl-dev libfuse-dev python3-dev python3-setuptools
If you downloaded the source using git make sure to run ./configure at least once to generate the dpkg packaging files.
To build the Debian packages change into the source directory and run:
cp -rf dpkg debian
dpkg-buildpackage -rfakeroot
This will create the following files in the parent directory:
libfvde_<version>-1_<arch>.deb
libfvde-dev_<version>-1_<arch>.deb
libfvde-python3_<version>-1_<arch>.deb
libfvde-tools_<version>-1_<arch>.deb
To install, e.g. the library:
sudo dpkg -i libfvde_<version>-1_<arch>.deb
dpkg-source: error: can't build with source format '3.0 (quilt)': no upstream tarball' found at ../libfvde-<version>.orig.tar.{bz2,gz,lzma,xz}
This means dpkg-buildpackage cannot find the source distribution package. Make sure it is provided as:
../libfvde-<version>.orig.tar.gz
To build libfvde using the RedHat package tools make sure you have the following packages installed:
dnf install rpm-build flex byacc zlib-devel openssl-devel fuse-devel python3-devel
To build:
mv libfvde-experimental-<version>.tar.gz libfvde-<version>.tar.gz
rpmbuild -ta libfvde-<version>.tar.gz
This will create the following files in the rpmbuild directory:
~/rpmbuild/RPMS/<arch>/libfvde-<version>-1.<arch>.rpm
~/rpmbuild/RPMS/<arch>/libfvde-devel-<version>-1.<arch>.rpm
~/rpmbuild/RPMS/<arch>/libfvde-python3-<version>-1.<arch>.rpm
~/rpmbuild/RPMS/<arch>/libfvde-tools-<version>-1.<arch>.rpm
~/rpmbuild/SRPMS/libfvde-<version>-1.src.rpm
To install, e.g. the library:
sudo rpm -ivh libfvde-<version>-1.<arch>.rpm
pkgbuild can be used to create a macOS package.
The following instructions show how to build libfvde.pkg and libfvde.dmg from the command line.
First build libfvde:
./configure --prefix=/usr/local --enable-python --with-pyprefix
make
For System Integrity Protection "/usr/local" is used instead of "/usr".
Next install the build files using DESTDIR
make install DESTDIR=$PWD/tmp
This will make sure that library paths in the dylib file is set correctly for distribution. This is not the case when you use:
./configure --prefix=$PWD/tmp --enable-python --with-pyprefix
You can check the library paths in the dylib by running:
otool -LT tmp/usr/lib/libfvde.1.dylib
After running "make install" the binaries are installed in:
$PWD/tmp/
If you are planning to distribute libfvde.pkg make sure it contains a copy of LGPL license:
mkdir -p $PWD/tmp/usr/share/doc/libfvde
cp AUTHORS COPYING COPYING.LESSER NEWS README $PWD/tmp/usr/share/doc/libfvde
To create the package (directory):
pkgbuild --root $PWD/tmp --identifier com.github.libyal.libfvde --version <version> --ownership recommended ../libfvde-<version>.pkg
To create a distributable disk image:
hdiutil create ../libfvde-<version>.dmg -srcfolder ../libfvde-<version>.pkg -fs HFS+
setup.py can be used to create a Python-binding only build.
python setup.py build
sudo python setup.py install
For more details see:
python setup.py --help