docs/autoconf.html - platform/external/mesa3d - Gitiles

 <!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/or 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>--sysconfdir=DIR</code></dt>
 <dd><p>This option specifies the directory where the configuration
 files will be installed. The default is <code>${prefix}/etc</code>.
 Currently there's only one config file provided when dri drivers are
 enabled - it's <code>drirc</code>.</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 configuring 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 set the compiler debug/optimisation levels (if the user
 hasn't already set them via the CFLAGS/CXXFLAGS) and macros to aid in
 debugging the Mesa libraries.</p>

 <p>Note that enabling this option can lead to noticeable loss of performance.</p>

 <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. GL 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-glx</code> and <code>--enable-osmesa</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-glx=xlib</code> or <code>--enable-glx=gallium-xlib</code>.

 <h3 id="dri">DRI</h3><p>This mode uses the DRI hardware drivers for
 accelerated OpenGL rendering. To enable use <code>--enable-glx=dri
 --enable-dri</code>.

 <!-- 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>${libdir}/dri</code>.
 <dt><code>--with-dri-drivers=DRIVER,DRIVER,...</code>
 <dd><p> This option
 allows a specific set of DRI drivers to be built. For example,
 <code>--with-dri-drivers="swrast,i965,radeon,nouveau"</code>. By
 default, the drivers will be chosen depending on the target platform.
 See the directory <code>src/mesa/drivers/dri</code> in the source tree
 for available drivers. Beware that the swrast DRI driver is used by both
 libGL and the X.Org xserver GLX module to do software rendering, so you
 may run into problems if it is not available.
 <!-- This explanation might be totally bogus. Kristian? -->
 <dt><code>--disable-driglx-direct</code>
 <dd><p> Disable direct rendering in
 GLX. Normally, direct hardware rendering through the DRI drivers and
 indirect software rendering are enabled in GLX. This option disables
 direct rendering entirely. It can be useful on architectures where
 kernel DRM modules are not available.
 <dt><code>--enable-glx-tls</code> <dd><p>
 Enable Thread Local Storage (TLS) in
 GLX.
 <dt><code>--with-expat=DIR</code>
 <dd><p><strong>DEPRECATED</strong>, use <code>PKG_CONFIG_PATH</code> instead.</p>
 <p>The DRI-enabled libGL uses expat to
 parse the DRI configuration files in <code>${sysconfdir}/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 libraries
 that will be built.

 </div>
 </body>
 </html>
	<!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/or 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>--sysconfdir=DIR</code></dt>
	<dd><p>This option specifies the directory where the configuration
	files will be installed. The default is <code>${prefix}/etc</code>.
	Currently there's only one config file provided when dri drivers are
	enabled - it's <code>drirc</code>.</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 configuring 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 set the compiler debug/optimisation levels (if the user
	hasn't already set them via the CFLAGS/CXXFLAGS) and macros to aid in
	debugging the Mesa libraries.</p>

	<p>Note that enabling this option can lead to noticeable loss of performance.</p>

	<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. GL 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-glx</code> and <code>--enable-osmesa</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-glx=xlib</code> or <code>--enable-glx=gallium-xlib</code>.

	<h3 id="dri">DRI</h3><p>This mode uses the DRI hardware drivers for
	accelerated OpenGL rendering. To enable use <code>--enable-glx=dri
	--enable-dri</code>.

	<!-- 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>${libdir}/dri</code>.
	<dt><code>--with-dri-drivers=DRIVER,DRIVER,...</code>
	<dd><p> This option
	allows a specific set of DRI drivers to be built. For example,
	<code>--with-dri-drivers="swrast,i965,radeon,nouveau"</code>. By
	default, the drivers will be chosen depending on the target platform.
	See the directory <code>src/mesa/drivers/dri</code> in the source tree
	for available drivers. Beware that the swrast DRI driver is used by both
	libGL and the X.Org xserver GLX module to do software rendering, so you
	may run into problems if it is not available.
	<!-- This explanation might be totally bogus. Kristian? -->
	<dt><code>--disable-driglx-direct</code>
	<dd><p> Disable direct rendering in
	GLX. Normally, direct hardware rendering through the DRI drivers and
	indirect software rendering are enabled in GLX. This option disables
	direct rendering entirely. It can be useful on architectures where
	kernel DRM modules are not available.
	<dt><code>--enable-glx-tls</code> <dd><p>
	Enable Thread Local Storage (TLS) in
	GLX.
	<dt><code>--with-expat=DIR</code>
	<dd><p><strong>DEPRECATED</strong>, use <code>PKG_CONFIG_PATH</code> instead.</p>
	<p>The DRI-enabled libGL uses expat to
	parse the DRI configuration files in <code>${sysconfdir}/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 libraries
	that will be built.

	</div>
	</body>
	</html>