<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd"> <html lang="en"> <head> <meta http-equiv="content-type" content="text/html; charset=utf-8"> <title>Compilation and Installation using Autoconf</title> <link rel="stylesheet" type="text/css" href="mesa.css"> </head> <body> <div class="header"> <h1>The Mesa 3D Graphics Library</h1> </div> <iframe src="contents.html"></iframe> <div class="content"> <h1>Compilation and Installation using Autoconf</h1> <ol> <li><p><a href="#basic">Basic Usage</a></li> <li><p><a href="#driver">Driver Options</a> <ul> <li><a href="#xlib">Xlib Driver Options</a></li> <li><a href="#dri">DRI Driver Options</a></li> <li><a href="#osmesa">OSMesa Driver Options</a></li> </ul> </ol> <h2 id="basic">1. Basic Usage</h2> <p> The autoconf generated configure script can be used to guess your platform and change various options for building Mesa. To use the configure script, type: </p> <pre> ./configure </pre> <p> To see a short description of all the options, type <code>./configure --help</code>. If you are using a development snapshot and the configure script does not exist, type <code>./autogen.sh</code> to generate it first. If you know the options you want to pass to <code>configure</code>, you can pass them to <code>autogen.sh</code>. It will run <code>configure</code> with these options after it is generated. Once you have run <code>configure</code> and set the options to your preference, type: </p> <pre> make </pre> <p> This will produce libGL.so and several other libraries depending on the options you have chosen. Later, if you want to rebuild for a different configuration run <code>make realclean</code> before rebuilding. </p> <p> Some of the generic autoconf options are used with Mesa: </p> <dl> <dt><code>--prefix=PREFIX</code></dt> <dd><p>This is the root directory where files will be installed by <code>make install</code>. The default is <code>/usr/local</code>.</p> </dd> <dt><code>--exec-prefix=EPREFIX</code></dt> <dd><p>This is the root directory where architecture-dependent files will be installed. In Mesa, this is only used to derive the directory for the libraries. The default is <code>${prefix}</code>.</p> </dd> <dt><code>--libdir=LIBDIR</code></dt> <dd><p>This option specifies the directory where the GL libraries will be installed. The default is <code>${exec_prefix}/lib</code>. It also serves as the name of the library staging area in the source tree. For instance, if the option <code>--libdir=/usr/local/lib64</code> is used, the libraries will be created in a <code>lib64</code> directory at the top of the Mesa source tree.</p> </dd> <dt><code>--enable-static, --disable-shared</code></dt> <dd><p>By default, Mesa will build shared libraries. Either of these options will force static libraries to be built. It is not currently possible to build static and shared libraries in a single pass.</p> </dd> <dt><code>CC, CFLAGS, CXX, CXXFLAGS</code></dt> <dd><p>These environment variables control the C and C++ compilers used during the build. By default, <code>gcc</code> and <code>g++</code> are used and the debug/optimisation level is left unchanged.</p> </dd> <dt><code>LDFLAGS</code></dt> <dd><p>An environment variable specifying flags to pass when linking programs. These should be empty and <code>PKG_CONFIG_PATH</code> is recommended to be used instead. If needed it can be used to direct the linker to use libraries in nonstandard directories. For example, <code>LDFLAGS="-L/usr/X11R6/lib"</code>.</p> </dd> <dt><code>PKG_CONFIG_PATH</code></dt> <dd><p>The <code>pkg-config</code> utility is a hard requirement for cofiguring and building mesa. It is used to search for external libraries on the system. This environment variable is used to control the search path for <code>pkg-config</code>. For instance, setting <code>PKG_CONFIG_PATH=/usr/X11R6/lib/pkgconfig</code> will search for package metadata in <code>/usr/X11R6</code> before the standard directories.</p> </dd> </dl> <p> There are also a few general options for altering the Mesa build: </p> <dl> <dt><code>--enable-debug</code></dt> <dd><p>This option will enable compiler options and macros to aid in debugging the Mesa libraries.</p> </dd> <dt><code>--disable-asm</code></dt> <dd><p>There are assembly routines available for a few architectures. These will be used by default if one of these architectures is detected. This option ensures that assembly will not be used.</p> </dd> <dt><code>--build=</code></dt> <dt><code>--host=</code></dt> <dd><p>By default, the build will compile code for the architecture that it's running on. In order to build cross-compile Mesa on a x86-64 machine that is to run on a i686, one would need to set the options to:</p> <p><code>--build=x86_64-pc-linux-gnu --host=i686-pc-linux-gnu</code></p> Note that these can vary from distribution to distribution. For more information check with the <a href="https://www.gnu.org/savannah-checkouts/gnu/autoconf/manual/autoconf-2.69/html_node/Specifying-Target-Triplets.html"> autoconf manual</a>. Note that you will need to correctly set <code>PKG_CONFIG_PATH</code> as well. <p>In some cases a single compiler is capable of handling both architectures (multilib) in that case one would need to set the <code>CC,CXX</code> variables appending the correct machine options. Seek your compiler documentation for further information - <a href="https://gcc.gnu.org/onlinedocs/gcc/Submodel-Options.html"> gcc machine dependent options</a></p> <p>In addition to specifying correct <code>PKG_CONFIG_PATH</code> for the target architecture, the following should be sufficient to configure multilib Mesa</p> <code>./configure CC="gcc -m32" CXX="g++ -m32" --build=x86_64-pc-linux-gnu --host=i686-pc-linux-gnu ...</code> </dd> </dl> <h2 id="driver">2. Driver Options</h2> <p> There are several different driver modes that Mesa can use. These are described in more detail in the <a href="install.html">basic installation instructions</a>. The Mesa driver is controlled through the configure options <code>--enable-xlib-glx</code>, <code>--enable-osmesa</code>, and <code>--enable-dri</code>. </p> <h3 id="xlib">Xlib</h3><p> It uses Xlib as a software renderer to do all rendering. It corresponds to the option <code>--enable-xlib-glx</code>. The libX11 and libXext libraries, as well as the X11 development headers, will be need to support the Xlib driver. <h3 id="dri">DRI</h3><p>This mode uses the DRI hardware drivers for accelerated OpenGL rendering. Enable the DRI drivers with the option <code>--enable-dri</code>. See the <a href="install.html">basic installation instructions</a> for details on prerequisites for the DRI drivers. <!-- DRI specific options --> <dl> <dt><code>--with-dri-driverdir=DIR</code> <dd><p> This option specifies the location the DRI drivers will be installed to and the location libGL will search for DRI drivers. The default is <code>${libdir}/dri</code>. <dt><code>--with-dri-drivers=DRIVER,DRIVER,...</code> <dd><p> This option allows a specific set of DRI drivers to be built. For example, <code>--with-dri-drivers="swrast,i965,radeon,nouveau"</code>. By default, the drivers will be chosen depending on the target platform. See the directory <code>src/mesa/drivers/dri</code> in the source tree for available drivers. Beware that the swrast DRI driver is used by both libGL and the X.Org xserver GLX module to do software rendering, so you may run into problems if it is not available. <!-- This explanation might be totally bogus. Kristian? --> <dt><code>--disable-driglx-direct</code> <dd><p> Disable direct rendering in GLX. Normally, direct hardware rendering through the DRI drivers and indirect software rendering are enabled in GLX. This option disables direct rendering entirely. It can be useful on architectures where kernel DRM modules are not available. <dt><code>--enable-glx-tls</code> <dd><p> Enable Thread Local Storage (TLS) in GLX. <dt><code>--with-expat=DIR</code> <dd><p><strong>DEPRECATED</strong>, use <code>PKG_CONFIG_PATH</code> instead.</p> <p>The DRI-enabled libGL uses expat to parse the DRI configuration files in <code>/etc/drirc</code> and <code>~/.drirc</code>. This option allows a specific expat installation to be used. For example, <code>--with-expat=/usr/local</code> will search for expat headers and libraries in <code>/usr/local/include</code> and <code>/usr/local/lib</code>, respectively. </dl> <h3 id="osmesa">OSMesa </h3><p> No libGL is built in this mode. Instead, the driver code is built into the Off-Screen Mesa (OSMesa) library. See the <a href="osmesa.html">Off-Screen Rendering</a> page for more details. It corresponds to the option <code>--enable-osmesa</code>. <!-- OSMesa specific options --> <dl> <dt><code>--with-osmesa-bits=BITS</code> <dd><p> This option allows the size of the color channel in bits to be specified. By default, an 8-bit channel will be used, and the driver will be named libOSMesa. Other options are 16- and 32-bit color channels, which will add the bit size to the library name. For example, <code>--with-osmesa-bits=16</code> will create the libOSMesa16 library with a 16-bit color channel. </dl> <h2 id="library">3. Library Options</h2> <p> The configure script provides more fine grained control over the GL libraries that will be built. More details on the specific GL libraries can be found in the <a href="install.html">basic installation instructions</a>. </div> </body> </html>