You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

1967 lines
91 KiB
HTML

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html>
<!-- Copyright (C) 1988-2018 Free Software Foundation, Inc.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.3 or
any later version published by the Free Software Foundation; with no
Invariant Sections, the Front-Cover texts being (a) (see below), and
with the Back-Cover Texts being (b) (see below). A copy of the
license is included in the section entitled "GNU
Free Documentation License".
(a) The FSF's Front-Cover Text is:
A GNU Manual
(b) The FSF's Back-Cover Text is:
You have freedom to copy and modify this GNU Manual, like GNU
software. Copies published by the Free Software Foundation raise
funds for GNU development. -->
<!-- Created by GNU Texinfo 6.4, http://www.gnu.org/software/texinfo/ -->
<head>
<title>Installing GCC</title>
<meta name="description" content="Installing GCC">
<meta name="keywords" content="Installing GCC">
<meta name="resource-type" content="document">
<meta name="distribution" content="global">
<meta name="Generator" content="makeinfo">
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<style type="text/css">
<!--
a.summary-letter {text-decoration: none}
blockquote.indentedblock {margin-right: 0em}
blockquote.smallindentedblock {margin-right: 0em; font-size: smaller}
blockquote.smallquotation {font-size: smaller}
div.display {margin-left: 3.2em}
div.example {margin-left: 3.2em}
div.lisp {margin-left: 3.2em}
div.smalldisplay {margin-left: 3.2em}
div.smallexample {margin-left: 3.2em}
div.smalllisp {margin-left: 3.2em}
kbd {font-style: oblique}
pre.display {font-family: inherit}
pre.format {font-family: inherit}
pre.menu-comment {font-family: serif}
pre.menu-preformatted {font-family: serif}
pre.smalldisplay {font-family: inherit; font-size: smaller}
pre.smallexample {font-size: smaller}
pre.smallformat {font-family: inherit; font-size: smaller}
pre.smalllisp {font-size: smaller}
span.nolinebreak {white-space: nowrap}
span.roman {font-family: initial; font-weight: normal}
span.sansserif {font-family: sans-serif; font-weight: normal}
ul.no-bullet {list-style: none}
-->
</style>
</head>
<body lang="en">
<h1 class="settitle" align="center">Installing GCC</h1>
<a name="index-Configuration"></a>
<a name="index-Installing-GCC_003a-Configuration"></a>
<p>Like most GNU software, GCC must be configured before it can be built.
This document describes the recommended configuration procedure
for both native and cross targets.
</p>
<p>We use <var>srcdir</var> to refer to the toplevel source directory for
GCC; we use <var>objdir</var> to refer to the toplevel build/object directory.
</p>
<p>If you obtained the sources via SVN, <var>srcdir</var> must refer to the top
<samp>gcc</samp> directory, the one where the <samp>MAINTAINERS</samp> file can be
found, and not its <samp>gcc</samp> subdirectory, otherwise the build will fail.
</p>
<p>If either <var>srcdir</var> or <var>objdir</var> is located on an automounted NFS
file system, the shell&rsquo;s built-in <code>pwd</code> command will return
temporary pathnames. Using these can lead to various sorts of build
problems. To avoid this issue, set the <code>PWDCMD</code> environment
variable to an automounter-aware <code>pwd</code> command, e.g.,
<code>pawd</code> or &lsquo;<samp>amq -w</samp>&rsquo;, during the configuration and build
phases.
</p>
<p>First, we <strong>highly</strong> recommend that GCC be built into a
separate directory from the sources which does <strong>not</strong> reside
within the source tree. This is how we generally build GCC; building
where <var>srcdir</var> == <var>objdir</var> should still work, but doesn&rsquo;t
get extensive testing; building where <var>objdir</var> is a subdirectory
of <var>srcdir</var> is unsupported.
</p>
<p>If you have previously built GCC in the same directory for a
different target machine, do &lsquo;<samp>make distclean</samp>&rsquo; to delete all files
that might be invalid. One of the files this deletes is <samp>Makefile</samp>;
if &lsquo;<samp>make distclean</samp>&rsquo; complains that <samp>Makefile</samp> does not exist
or issues a message like &ldquo;don&rsquo;t know how to make distclean&rdquo; it probably
means that the directory is already suitably clean. However, with the
recommended method of building in a separate <var>objdir</var>, you should
simply use a different <var>objdir</var> for each target.
</p>
<p>Second, when configuring a native system, either <code>cc</code> or
<code>gcc</code> must be in your path or you must set <code>CC</code> in
your environment before running configure. Otherwise the configuration
scripts may fail.
</p>
<p>To configure GCC:
</p>
<div class="smallexample">
<pre class="smallexample">% mkdir <var>objdir</var>
% cd <var>objdir</var>
% <var>srcdir</var>/configure [<var>options</var>] [<var>target</var>]
</pre></div>
<a name="Distributor-options"></a>
<h3 class="heading">Distributor options</h3>
<p>If you will be distributing binary versions of GCC, with modifications
to the source code, you should use the options described in this
section to make clear that your version contains modifications.
</p>
<dl compact="compact">
<dt><code>--with-pkgversion=<var>version</var></code></dt>
<dd><p>Specify a string that identifies your package. You may wish
to include a build number or build date. This version string will be
included in the output of <code>gcc --version</code>. This suffix does
not replace the default version string, only the &lsquo;<samp>GCC</samp>&rsquo; part.
</p>
<p>The default value is &lsquo;<samp>GCC</samp>&rsquo;.
</p>
</dd>
<dt><code>--with-bugurl=<var>url</var></code></dt>
<dd><p>Specify the URL that users should visit if they wish to report a bug.
You are of course welcome to forward bugs reported to you to the FSF,
if you determine that they are not bugs in your modifications.
</p>
<p>The default value refers to the FSF&rsquo;s GCC bug tracker.
</p>
</dd>
</dl>
<a name="Target-specification"></a>
<h3 class="heading">Target specification</h3>
<ul>
<li> GCC has code to correctly determine the correct value for <var>target</var>
for nearly all native systems. Therefore, we highly recommend you do
not provide a configure target when configuring a native compiler.
</li><li> <var>target</var> must be specified as <samp>--target=<var>target</var></samp>
when configuring a cross compiler; examples of valid targets would be
m68k-elf, sh-elf, etc.
</li><li> Specifying just <var>target</var> instead of <samp>--target=<var>target</var></samp>
implies that the host defaults to <var>target</var>.
</li></ul>
<a name="Options-specification"></a>
<h3 class="heading">Options specification</h3>
<p>Use <var>options</var> to override several configure time options for
GCC. A list of supported <var>options</var> follows; &lsquo;<samp>configure
--help</samp>&rsquo; may list other options, but those not listed below may not
work and should not normally be used.
</p>
<p>Note that each <samp>--enable</samp> option has a corresponding
<samp>--disable</samp> option and that each <samp>--with</samp> option has a
corresponding <samp>--without</samp> option.
</p>
<dl compact="compact">
<dt><code>--prefix=<var>dirname</var></code></dt>
<dd><p>Specify the toplevel installation
directory. This is the recommended way to install the tools into a directory
other than the default. The toplevel installation directory defaults to
<samp>/usr/local</samp>.
</p>
<p>We <strong>highly</strong> recommend against <var>dirname</var> being the same or a
subdirectory of <var>objdir</var> or vice versa. If specifying a directory
beneath a user&rsquo;s home directory tree, some shells will not expand
<var>dirname</var> correctly if it contains the &lsquo;<samp>~</samp>&rsquo; metacharacter; use
<code>$HOME</code> instead.
</p>
<p>The following standard <code>autoconf</code> options are supported. Normally you
should not need to use these options.
</p><dl compact="compact">
<dt><code>--exec-prefix=<var>dirname</var></code></dt>
<dd><p>Specify the toplevel installation directory for architecture-dependent
files. The default is <samp><var>prefix</var></samp>.
</p>
</dd>
<dt><code>--bindir=<var>dirname</var></code></dt>
<dd><p>Specify the installation directory for the executables called by users
(such as <code>gcc</code> and <code>g++</code>). The default is
<samp><var>exec-prefix</var>/bin</samp>.
</p>
</dd>
<dt><code>--libdir=<var>dirname</var></code></dt>
<dd><p>Specify the installation directory for object code libraries and
internal data files of GCC. The default is <samp><var>exec-prefix</var>/lib</samp>.
</p>
</dd>
<dt><code>--libexecdir=<var>dirname</var></code></dt>
<dd><p>Specify the installation directory for internal executables of GCC.
The default is <samp><var>exec-prefix</var>/libexec</samp>.
</p>
</dd>
<dt><code>--with-slibdir=<var>dirname</var></code></dt>
<dd><p>Specify the installation directory for the shared libgcc library. The
default is <samp><var>libdir</var></samp>.
</p>
</dd>
<dt><code>--datarootdir=<var>dirname</var></code></dt>
<dd><p>Specify the root of the directory tree for read-only architecture-independent
data files referenced by GCC. The default is <samp><var>prefix</var>/share</samp>.
</p>
</dd>
<dt><code>--infodir=<var>dirname</var></code></dt>
<dd><p>Specify the installation directory for documentation in info format.
The default is <samp><var>datarootdir</var>/info</samp>.
</p>
</dd>
<dt><code>--datadir=<var>dirname</var></code></dt>
<dd><p>Specify the installation directory for some architecture-independent
data files referenced by GCC. The default is <samp><var>datarootdir</var></samp>.
</p>
</dd>
<dt><code>--docdir=<var>dirname</var></code></dt>
<dd><p>Specify the installation directory for documentation files (other
than Info) for GCC. The default is <samp><var>datarootdir</var>/doc</samp>.
</p>
</dd>
<dt><code>--htmldir=<var>dirname</var></code></dt>
<dd><p>Specify the installation directory for HTML documentation files.
The default is <samp><var>docdir</var></samp>.
</p>
</dd>
<dt><code>--pdfdir=<var>dirname</var></code></dt>
<dd><p>Specify the installation directory for PDF documentation files.
The default is <samp><var>docdir</var></samp>.
</p>
</dd>
<dt><code>--mandir=<var>dirname</var></code></dt>
<dd><p>Specify the installation directory for manual pages. The default is
<samp><var>datarootdir</var>/man</samp>. (Note that the manual pages are only extracts
from the full GCC manuals, which are provided in Texinfo format. The manpages
are derived by an automatic conversion process from parts of the full
manual.)
</p>
</dd>
<dt><code>--with-gxx-include-dir=<var>dirname</var></code></dt>
<dd><p>Specify
the installation directory for G++ header files. The default depends
on other configuration options, and differs between cross and native
configurations.
</p>
</dd>
<dt><code>--with-specs=<var>specs</var></code></dt>
<dd><p>Specify additional command line driver SPECS.
This can be useful if you need to turn on a non-standard feature by
default without modifying the compiler&rsquo;s source code, for instance
<samp>--with-specs=%{!fcommon:%{!fno-common:-fno-common}}</samp>.
See &ldquo;Spec Files&rdquo; in the main manual
</p>
</dd>
</dl>
</dd>
<dt><code>--program-prefix=<var>prefix</var></code></dt>
<dd><p>GCC supports some transformations of the names of its programs when
installing them. This option prepends <var>prefix</var> to the names of
programs to install in <var>bindir</var> (see above). For example, specifying
<samp>--program-prefix=foo-</samp> would result in &lsquo;<samp>gcc</samp>&rsquo;
being installed as <samp>/usr/local/bin/foo-gcc</samp>.
</p>
</dd>
<dt><code>--program-suffix=<var>suffix</var></code></dt>
<dd><p>Appends <var>suffix</var> to the names of programs to install in <var>bindir</var>
(see above). For example, specifying <samp>--program-suffix=-3.1</samp>
would result in &lsquo;<samp>gcc</samp>&rsquo; being installed as
<samp>/usr/local/bin/gcc-3.1</samp>.
</p>
</dd>
<dt><code>--program-transform-name=<var>pattern</var></code></dt>
<dd><p>Applies the &lsquo;<samp>sed</samp>&rsquo; script <var>pattern</var> to be applied to the names
of programs to install in <var>bindir</var> (see above). <var>pattern</var> has to
consist of one or more basic &lsquo;<samp>sed</samp>&rsquo; editing commands, separated by
semicolons. For example, if you want the &lsquo;<samp>gcc</samp>&rsquo; program name to be
transformed to the installed program <samp>/usr/local/bin/myowngcc</samp> and
the &lsquo;<samp>g++</samp>&rsquo; program name to be transformed to
<samp>/usr/local/bin/gspecial++</samp> without changing other program names,
you could use the pattern
<samp>--program-transform-name='s/^gcc$/myowngcc/; s/^g++$/gspecial++/'</samp>
to achieve this effect.
</p>
<p>All three options can be combined and used together, resulting in more
complex conversion patterns. As a basic rule, <var>prefix</var> (and
<var>suffix</var>) are prepended (appended) before further transformations
can happen with a special transformation script <var>pattern</var>.
</p>
<p>As currently implemented, this option only takes effect for native
builds; cross compiler binaries&rsquo; names are not transformed even when a
transformation is explicitly asked for by one of these options.
</p>
<p>For native builds, some of the installed programs are also installed
with the target alias in front of their name, as in
&lsquo;<samp>i686-pc-linux-gnu-gcc</samp>&rsquo;. All of the above transformations happen
before the target alias is prepended to the name&mdash;so, specifying
<samp>--program-prefix=foo-</samp> and <samp>program-suffix=-3.1</samp>, the
resulting binary would be installed as
<samp>/usr/local/bin/i686-pc-linux-gnu-foo-gcc-3.1</samp>.
</p>
<p>As a last shortcoming, none of the installed Ada programs are
transformed yet, which will be fixed in some time.
</p>
</dd>
<dt><code>--with-local-prefix=<var>dirname</var></code></dt>
<dd><p>Specify the
installation directory for local include files. The default is
<samp>/usr/local</samp>. Specify this option if you want the compiler to
search directory <samp><var>dirname</var>/include</samp> for locally installed
header files <em>instead</em> of <samp>/usr/local/include</samp>.
</p>
<p>You should specify <samp>--with-local-prefix</samp> <strong>only</strong> if your
site has a different convention (not <samp>/usr/local</samp>) for where to put
site-specific files.
</p>
<p>The default value for <samp>--with-local-prefix</samp> is <samp>/usr/local</samp>
regardless of the value of <samp>--prefix</samp>. Specifying
<samp>--prefix</samp> has no effect on which directory GCC searches for
local header files. This may seem counterintuitive, but actually it is
logical.
</p>
<p>The purpose of <samp>--prefix</samp> is to specify where to <em>install
GCC</em>. The local header files in <samp>/usr/local/include</samp>&mdash;if you put
any in that directory&mdash;are not part of GCC. They are part of other
programs&mdash;perhaps many others. (GCC installs its own header files in
another directory which is based on the <samp>--prefix</samp> value.)
</p>
<p>Both the local-prefix include directory and the GCC-prefix include
directory are part of GCC&rsquo;s &ldquo;system include&rdquo; directories. Although these
two directories are not fixed, they need to be searched in the proper
order for the correct processing of the include_next directive. The
local-prefix include directory is searched before the GCC-prefix
include directory. Another characteristic of system include directories
is that pedantic warnings are turned off for headers in these directories.
</p>
<p>Some autoconf macros add <samp>-I <var>directory</var></samp> options to the
compiler command line, to ensure that directories containing installed
packages&rsquo; headers are searched. When <var>directory</var> is one of GCC&rsquo;s
system include directories, GCC will ignore the option so that system
directories continue to be processed in the correct order. This
may result in a search order different from what was specified but the
directory will still be searched.
</p>
<p>GCC automatically searches for ordinary libraries using
<code>GCC_EXEC_PREFIX</code>. Thus, when the same installation prefix is
used for both GCC and packages, GCC will automatically search for
both headers and libraries. This provides a configuration that is
easy to use. GCC behaves in a manner similar to that when it is
installed as a system compiler in <samp>/usr</samp>.
</p>
<p>Sites that need to install multiple versions of GCC may not want to
use the above simple configuration. It is possible to use the
<samp>--program-prefix</samp>, <samp>--program-suffix</samp> and
<samp>--program-transform-name</samp> options to install multiple versions
into a single directory, but it may be simpler to use different prefixes
and the <samp>--with-local-prefix</samp> option to specify the location of the
site-specific files for each version. It will then be necessary for
users to specify explicitly the location of local site libraries
(e.g., with <code>LIBRARY_PATH</code>).
</p>
<p>The same value can be used for both <samp>--with-local-prefix</samp> and
<samp>--prefix</samp> provided it is not <samp>/usr</samp>. This can be used
to avoid the default search of <samp>/usr/local/include</samp>.
</p>
<p><strong>Do not</strong> specify <samp>/usr</samp> as the <samp>--with-local-prefix</samp>!
The directory you use for <samp>--with-local-prefix</samp> <strong>must not</strong>
contain any of the system&rsquo;s standard header files. If it did contain
them, certain programs would be miscompiled (including GNU Emacs, on
certain targets), because this would override and nullify the header
file corrections made by the <code>fixincludes</code> script.
</p>
<p>Indications are that people who use this option use it based on mistaken
ideas of what it is for. People use it as if it specified where to
install part of GCC. Perhaps they make this assumption because
installing GCC creates the directory.
</p>
</dd>
<dt><code>--with-gcc-major-version-only</code></dt>
<dd><p>Specifies that GCC should use only the major number rather than
<var>major</var>.<var>minor</var>.<var>patchlevel</var> in filesystem paths.
</p>
</dd>
<dt><code>--with-native-system-header-dir=<var>dirname</var></code></dt>
<dd><p>Specifies that <var>dirname</var> is the directory that contains native system
header files, rather than <samp>/usr/include</samp>. This option is most useful
if you are creating a compiler that should be isolated from the system
as much as possible. It is most commonly used with the
<samp>--with-sysroot</samp> option and will cause GCC to search
<var>dirname</var> inside the system root specified by that option.
</p>
</dd>
<dt><code>--enable-shared[=<var>package</var>[,&hellip;]]</code></dt>
<dd><p>Build shared versions of libraries, if shared libraries are supported on
the target platform. Unlike GCC 2.95.x and earlier, shared libraries
are enabled by default on all platforms that support shared libraries.
</p>
<p>If a list of packages is given as an argument, build shared libraries
only for the listed packages. For other packages, only static libraries
will be built. Package names currently recognized in the GCC tree are
&lsquo;<samp>libgcc</samp>&rsquo; (also known as &lsquo;<samp>gcc</samp>&rsquo;), &lsquo;<samp>libstdc++</samp>&rsquo; (not
&lsquo;<samp>libstdc++-v3</samp>&rsquo;), &lsquo;<samp>libffi</samp>&rsquo;, &lsquo;<samp>zlib</samp>&rsquo;, &lsquo;<samp>boehm-gc</samp>&rsquo;,
&lsquo;<samp>ada</samp>&rsquo;, &lsquo;<samp>libada</samp>&rsquo;, &lsquo;<samp>libgo</samp>&rsquo;, and &lsquo;<samp>libobjc</samp>&rsquo;.
Note &lsquo;<samp>libiberty</samp>&rsquo; does not support shared libraries at all.
</p>
<p>Use <samp>--disable-shared</samp> to build only static libraries. Note that
<samp>--disable-shared</samp> does not accept a list of package names as
argument, only <samp>--enable-shared</samp> does.
</p>
<p>Contrast with <samp>--enable-host-shared</samp>, which affects <em>host</em>
code.
</p>
</dd>
<dt><code>--enable-host-shared</code></dt>
<dd><p>Specify that the <em>host</em> code should be built into position-independent
machine code (with -fPIC), allowing it to be used within shared libraries,
but yielding a slightly slower compiler.
</p>
<p>This option is required when building the libgccjit.so library.
</p>
<p>Contrast with <samp>--enable-shared</samp>, which affects <em>target</em>
libraries.
</p>
</dd>
<dt><code><a name="with-gnu-as"></a>--with-gnu-as</code></dt>
<dd><p>Specify that the compiler should assume that the
assembler it finds is the GNU assembler. However, this does not modify
the rules to find an assembler and will result in confusion if the
assembler found is not actually the GNU assembler. (Confusion may also
result if the compiler finds the GNU assembler but has not been
configured with <samp>--with-gnu-as</samp>.) If you have more than one
assembler installed on your system, you may want to use this option in
connection with <samp>--with-as=<var>pathname</var></samp> or
<samp>--with-build-time-tools=<var>pathname</var></samp>.
</p>
<p>The following systems are the only ones where it makes a difference
whether you use the GNU assembler. On any other system,
<samp>--with-gnu-as</samp> has no effect.
</p>
<ul>
<li> &lsquo;<samp>hppa1.0-<var>any</var>-<var>any</var></samp>&rsquo;
</li><li> &lsquo;<samp>hppa1.1-<var>any</var>-<var>any</var></samp>&rsquo;
</li><li> &lsquo;<samp>sparc-sun-solaris2.<var>any</var></samp>&rsquo;
</li><li> &lsquo;<samp>sparc64-<var>any</var>-solaris2.<var>any</var></samp>&rsquo;
</li></ul>
</dd>
<dt><code><a name="with-as"></a>--with-as=<var>pathname</var></code></dt>
<dd><p>Specify that the compiler should use the assembler pointed to by
<var>pathname</var>, rather than the one found by the standard rules to find
an assembler, which are:
</p><ul>
<li> Unless GCC is being built with a cross compiler, check the
<samp><var>libexec</var>/gcc/<var>target</var>/<var>version</var></samp> directory.
<var>libexec</var> defaults to <samp><var>exec-prefix</var>/libexec</samp>;
<var>exec-prefix</var> defaults to <var>prefix</var>, which
defaults to <samp>/usr/local</samp> unless overridden by the
<samp>--prefix=<var>pathname</var></samp> switch described above. <var>target</var>
is the target system triple, such as &lsquo;<samp>sparc-sun-solaris2.7</samp>&rsquo;, and
<var>version</var> denotes the GCC version, such as 3.0.
</li><li> If the target system is the same that you are building on, check
operating system specific directories (e.g. <samp>/usr/ccs/bin</samp> on
Sun Solaris 2).
</li><li> Check in the <code>PATH</code> for a tool whose name is prefixed by the
target system triple.
</li><li> Check in the <code>PATH</code> for a tool whose name is not prefixed by the
target system triple, if the host and target system triple are
the same (in other words, we use a host tool if it can be used for
the target as well).
</li></ul>
<p>You may want to use <samp>--with-as</samp> if no assembler
is installed in the directories listed above, or if you have multiple
assemblers installed and want to choose one that is not found by the
above rules.
</p>
</dd>
<dt><code><a name="with-gnu-ld"></a>--with-gnu-ld</code></dt>
<dd><p>Same as <a href="#with-gnu-as"><samp>--with-gnu-as</samp></a>
but for the linker.
</p>
</dd>
<dt><code>--with-ld=<var>pathname</var></code></dt>
<dd><p>Same as <a href="#with-as"><samp>--with-as</samp></a>
but for the linker.
</p>
</dd>
<dt><code>--with-stabs</code></dt>
<dd><p>Specify that stabs debugging
information should be used instead of whatever format the host normally
uses. Normally GCC uses the same debug format as the host system.
</p>
</dd>
<dt><code>--with-tls=<var>dialect</var></code></dt>
<dd><p>Specify the default TLS dialect, for systems were there is a choice.
For ARM targets, possible values for <var>dialect</var> are <code>gnu</code> or
<code>gnu2</code>, which select between the original GNU dialect and the GNU TLS
descriptor-based dialect.
</p>
</dd>
<dt><code>--enable-multiarch</code></dt>
<dd><p>Specify whether to enable or disable multiarch support. The default is
to check for glibc start files in a multiarch location, and enable it
if the files are found. The auto detection is enabled for native builds,
and for cross builds configured with <samp>--with-sysroot</samp>, and without
<samp>--with-native-system-header-dir</samp>.
More documentation about multiarch can be found at
<a href="https://wiki.debian.org/Multiarch">https://wiki.debian.org/Multiarch</a>.
</p>
</dd>
<dt><code>--enable-sjlj-exceptions</code></dt>
<dd><p>Force use of the <code>setjmp</code>/<code>longjmp</code>-based scheme for exceptions.
&lsquo;<samp>configure</samp>&rsquo; ordinarily picks the correct value based on the platform.
Only use this option if you are sure you need a different setting.
</p>
</dd>
<dt><code>--enable-vtable-verify</code></dt>
<dd><p>Specify whether to enable or disable the vtable verification feature.
Enabling this feature causes libstdc++ to be built with its virtual calls
in verifiable mode. This means that, when linked with libvtv, every
virtual call in libstdc++ will verify the vtable pointer through which the
call will be made before actually making the call. If not linked with libvtv,
the verifier will call stub functions (in libstdc++ itself) and do nothing.
If vtable verification is disabled, then libstdc++ is not built with its
virtual calls in verifiable mode at all. However the libvtv library will
still be built (see <samp>--disable-libvtv</samp> to turn off building libvtv).
<samp>--disable-vtable-verify</samp> is the default.
</p>
</dd>
<dt><code>--disable-multilib</code></dt>
<dd><p>Specify that multiple target
libraries to support different target variants, calling
conventions, etc. should not be built. The default is to build a
predefined set of them.
</p>
<p>Some targets provide finer-grained control over which multilibs are built
(e.g., <samp>--disable-softfloat</samp>):
</p><dl compact="compact">
<dt><code>arm-*-*</code></dt>
<dd><p>fpu, 26bit, underscore, interwork, biendian, nofmult.
</p>
</dd>
<dt><code>m68*-*-*</code></dt>
<dd><p>softfloat, m68881, m68000, m68020.
</p>
</dd>
<dt><code>mips*-*-*</code></dt>
<dd><p>single-float, biendian, softfloat.
</p>
</dd>
<dt><code>powerpc*-*-*, rs6000*-*-*</code></dt>
<dd><p>aix64, pthread, softfloat, powercpu, powerpccpu, powerpcos, biendian,
sysv, aix.
</p>
</dd>
</dl>
</dd>
<dt><code>--with-multilib-list=<var>list</var></code></dt>
<dt><code>--without-multilib-list</code></dt>
<dd><p>Specify what multilibs to build. <var>list</var> is a comma separated list of
values, possibly consisting of a single value. Currently only implemented
for aarch64*-*-*, arm*-*-*, riscv*-*-*, sh*-*-* and x86-64-*-linux*. The
accepted values and meaning for each target is given below.
</p>
<dl compact="compact">
<dt><code>aarch64*-*-*</code></dt>
<dd><p><var>list</var> is a comma separated list of <code>ilp32</code>, and <code>lp64</code>
to enable ILP32 and LP64 run-time libraries, respectively. If
<var>list</var> is empty, then there will be no multilibs and only the
default run-time library will be built. If <var>list</var> is
<code>default</code> or &ndash;with-multilib-list= is not specified, then the
default set of libraries is selected based on the value of
<samp>--target</samp>.
</p>
</dd>
<dt><code>arm*-*-*</code></dt>
<dd><p><var>list</var> is a comma separated list of <code>aprofile</code> and <code>rmprofile</code>
to build multilibs for A or R and M architecture profiles respectively. Note
that, due to some limitation of the current multilib framework, using the
combined <code>aprofile,rmprofile</code> multilibs selects in some cases a less
optimal multilib than when using the multilib profile for the architecture
targetted. The special value <code>default</code> is also accepted and is equivalent
to omitting the option, ie. only the default run-time library will be enabled.
</p>
<p>The table below gives the combination of ISAs, architectures, FPUs and
floating-point ABIs for which multilibs are built for each accepted value.
The union of these options is considered when specifying both <code>aprofile</code>
and <code>rmprofile</code>.
</p>
<table>
<tr><td width="15%">Option</td><td width="28%">aprofile</td><td width="30%">rmprofile</td></tr>
<tr><td width="15%">ISAs</td><td width="28%"><code>-marm</code> and <code>-mthumb</code></td><td width="30%"><code>-mthumb</code></td></tr>
<tr><td width="15%">Architectures<br><br><br><br><br><br></td><td width="28%">default architecture<br>
<code>-march=armv7-a</code><br>
<code>-march=armv7ve</code><br>
<code>-march=armv8-a</code><br><br><br></td><td width="30%">default architecture<br>
<code>-march=armv6s-m</code><br>
<code>-march=armv7-m</code><br>
<code>-march=armv7e-m</code><br>
<code>-march=armv8-m.base</code><br>
<code>-march=armv8-m.main</code><br>
<code>-march=armv7</code></td></tr>
<tr><td width="15%">FPUs<br><br><br><br><br></td><td width="28%">none<br>
<code>-mfpu=vfpv3-d16</code><br>
<code>-mfpu=neon</code><br>
<code>-mfpu=vfpv4-d16</code><br>
<code>-mfpu=neon-vfpv4</code><br>
<code>-mfpu=neon-fp-armv8</code></td><td width="30%">none<br>
<code>-mfpu=vfpv3-d16</code><br>
<code>-mfpu=fpv4-sp-d16</code><br>
<code>-mfpu=fpv5-sp-d16</code><br>
<code>-mfpu=fpv5-d16</code><br></td></tr>
<tr><td width="15%">floating-point ABIs<br><br></td><td width="28%"><code>-mfloat-abi=soft</code><br>
<code>-mfloat-abi=softfp</code><br>
<code>-mfloat-abi=hard</code></td><td width="30%"><code>-mfloat-abi=soft</code><br>
<code>-mfloat-abi=softfp</code><br>
<code>-mfloat-abi=hard</code></td></tr>
</table>
</dd>
<dt><code>riscv*-*-*</code></dt>
<dd><p><var>list</var> is a single ABI name. The target architecture must be either
<code>rv32gc</code> or <code>rv64gc</code>. This will build a single multilib for the
specified architecture and ABI pair. If <code>--with-multilib-list</code> is not
given, then a default set of multilibs is selected based on the value of
<samp>--target</samp>. This is usually a large set of multilibs.
</p>
</dd>
<dt><code>sh*-*-*</code></dt>
<dd><p><var>list</var> is a comma separated list of CPU names. These must be of the
form <code>sh*</code> or <code>m*</code> (in which case they match the compiler option
for that processor). The list should not contain any endian options -
these are handled by <samp>--with-endian</samp>.
</p>
<p>If <var>list</var> is empty, then there will be no multilibs for extra
processors. The multilib for the secondary endian remains enabled.
</p>
<p>As a special case, if an entry in the list starts with a <code>!</code>
(exclamation point), then it is added to the list of excluded multilibs.
Entries of this sort should be compatible with &lsquo;<samp>MULTILIB_EXCLUDES</samp>&rsquo;
(once the leading <code>!</code> has been stripped).
</p>
<p>If <samp>--with-multilib-list</samp> is not given, then a default set of
multilibs is selected based on the value of <samp>--target</samp>. This is
usually the complete set of libraries, but some targets imply a more
specialized subset.
</p>
<p>Example 1: to configure a compiler for SH4A only, but supporting both
endians, with little endian being the default:
</p><div class="smallexample">
<pre class="smallexample">--with-cpu=sh4a --with-endian=little,big --with-multilib-list=
</pre></div>
<p>Example 2: to configure a compiler for both SH4A and SH4AL-DSP, but with
only little endian SH4AL:
</p><div class="smallexample">
<pre class="smallexample">--with-cpu=sh4a --with-endian=little,big \
--with-multilib-list=sh4al,!mb/m4al
</pre></div>
</dd>
<dt><code>x86-64-*-linux*</code></dt>
<dd><p><var>list</var> is a comma separated list of <code>m32</code>, <code>m64</code> and
<code>mx32</code> to enable 32-bit, 64-bit and x32 run-time libraries,
respectively. If <var>list</var> is empty, then there will be no multilibs
and only the default run-time library will be enabled.
</p>
<p>If <samp>--with-multilib-list</samp> is not given, then only 32-bit and
64-bit run-time libraries will be enabled.
</p></dd>
</dl>
</dd>
<dt><code>--with-endian=<var>endians</var></code></dt>
<dd><p>Specify what endians to use.
Currently only implemented for sh*-*-*.
</p>
<p><var>endians</var> may be one of the following:
</p><dl compact="compact">
<dt><code>big</code></dt>
<dd><p>Use big endian exclusively.
</p></dd>
<dt><code>little</code></dt>
<dd><p>Use little endian exclusively.
</p></dd>
<dt><code>big,little</code></dt>
<dd><p>Use big endian by default. Provide a multilib for little endian.
</p></dd>
<dt><code>little,big</code></dt>
<dd><p>Use little endian by default. Provide a multilib for big endian.
</p></dd>
</dl>
</dd>
<dt><code>--enable-threads</code></dt>
<dd><p>Specify that the target
supports threads. This affects the Objective-C compiler and runtime
library, and exception handling for other languages like C++.
On some systems, this is the default.
</p>
<p>In general, the best (and, in many cases, the only known) threading
model available will be configured for use. Beware that on some
systems, GCC has not been taught what threading models are generally
available for the system. In this case, <samp>--enable-threads</samp> is an
alias for <samp>--enable-threads=single</samp>.
</p>
</dd>
<dt><code>--disable-threads</code></dt>
<dd><p>Specify that threading support should be disabled for the system.
This is an alias for <samp>--enable-threads=single</samp>.
</p>
</dd>
<dt><code>--enable-threads=<var>lib</var></code></dt>
<dd><p>Specify that
<var>lib</var> is the thread support library. This affects the Objective-C
compiler and runtime library, and exception handling for other languages
like C++. The possibilities for <var>lib</var> are:
</p>
<dl compact="compact">
<dt><code>aix</code></dt>
<dd><p>AIX thread support.
</p></dd>
<dt><code>dce</code></dt>
<dd><p>DCE thread support.
</p></dd>
<dt><code>lynx</code></dt>
<dd><p>LynxOS thread support.
</p></dd>
<dt><code>mipssde</code></dt>
<dd><p>MIPS SDE thread support.
</p></dd>
<dt><code>no</code></dt>
<dd><p>This is an alias for &lsquo;<samp>single</samp>&rsquo;.
</p></dd>
<dt><code>posix</code></dt>
<dd><p>Generic POSIX/Unix98 thread support.
</p></dd>
<dt><code>rtems</code></dt>
<dd><p>RTEMS thread support.
</p></dd>
<dt><code>single</code></dt>
<dd><p>Disable thread support, should work for all platforms.
</p></dd>
<dt><code>tpf</code></dt>
<dd><p>TPF thread support.
</p></dd>
<dt><code>vxworks</code></dt>
<dd><p>VxWorks thread support.
</p></dd>
<dt><code>win32</code></dt>
<dd><p>Microsoft Win32 API thread support.
</p></dd>
</dl>
</dd>
<dt><code>--enable-tls</code></dt>
<dd><p>Specify that the target supports TLS (Thread Local Storage). Usually
configure can correctly determine if TLS is supported. In cases where
it guesses incorrectly, TLS can be explicitly enabled or disabled with
<samp>--enable-tls</samp> or <samp>--disable-tls</samp>. This can happen if
the assembler supports TLS but the C library does not, or if the
assumptions made by the configure test are incorrect.
</p>
</dd>
<dt><code>--disable-tls</code></dt>
<dd><p>Specify that the target does not support TLS.
This is an alias for <samp>--enable-tls=no</samp>.
</p>
</dd>
<dt><code>--disable-tm-clone-registry</code></dt>
<dd><p>Disable TM clone registry in libgcc. It is enabled in libgcc by default.
This option helps to reduce code size for embedded targets which do
not use transactional memory.
</p>
</dd>
<dt><code>--with-cpu=<var>cpu</var></code></dt>
<dt><code>--with-cpu-32=<var>cpu</var></code></dt>
<dt><code>--with-cpu-64=<var>cpu</var></code></dt>
<dd><p>Specify which cpu variant the compiler should generate code for by default.
<var>cpu</var> will be used as the default value of the <samp>-mcpu=</samp> switch.
This option is only supported on some targets, including ARC, ARM, i386, M68k,
PowerPC, and SPARC. It is mandatory for ARC. The <samp>--with-cpu-32</samp> and
<samp>--with-cpu-64</samp> options specify separate default CPUs for
32-bit and 64-bit modes; these options are only supported for i386,
x86-64, PowerPC, and SPARC.
</p>
</dd>
<dt><code>--with-schedule=<var>cpu</var></code></dt>
<dt><code>--with-arch=<var>cpu</var></code></dt>
<dt><code>--with-arch-32=<var>cpu</var></code></dt>
<dt><code>--with-arch-64=<var>cpu</var></code></dt>
<dt><code>--with-tune=<var>cpu</var></code></dt>
<dt><code>--with-tune-32=<var>cpu</var></code></dt>
<dt><code>--with-tune-64=<var>cpu</var></code></dt>
<dt><code>--with-abi=<var>abi</var></code></dt>
<dt><code>--with-fpu=<var>type</var></code></dt>
<dt><code>--with-float=<var>type</var></code></dt>
<dd><p>These configure options provide default values for the <samp>-mschedule=</samp>,
<samp>-march=</samp>, <samp>-mtune=</samp>, <samp>-mabi=</samp>, and <samp>-mfpu=</samp>
options and for <samp>-mhard-float</samp> or <samp>-msoft-float</samp>. As with
<samp>--with-cpu</samp>, which switches will be accepted and acceptable values
of the arguments depend on the target.
</p>
</dd>
<dt><code>--with-mode=<var>mode</var></code></dt>
<dd><p>Specify if the compiler should default to <samp>-marm</samp> or <samp>-mthumb</samp>.
This option is only supported on ARM targets.
</p>
</dd>
<dt><code>--with-stack-offset=<var>num</var></code></dt>
<dd><p>This option sets the default for the -mstack-offset=<var>num</var> option,
and will thus generally also control the setting of this option for
libraries. This option is only supported on Epiphany targets.
</p>
</dd>
<dt><code>--with-fpmath=<var>isa</var></code></dt>
<dd><p>This options sets <samp>-mfpmath=sse</samp> by default and specifies the default
ISA for floating-point arithmetics. You can select either &lsquo;<samp>sse</samp>&rsquo; which
enables <samp>-msse2</samp> or &lsquo;<samp>avx</samp>&rsquo; which enables <samp>-mavx</samp> by default.
This option is only supported on i386 and x86-64 targets.
</p>
</dd>
<dt><code>--with-fp-32=<var>mode</var></code></dt>
<dd><p>On MIPS targets, set the default value for the <samp>-mfp</samp> option when using
the o32 ABI. The possibilities for <var>mode</var> are:
</p><dl compact="compact">
<dt><code>32</code></dt>
<dd><p>Use the o32 FP32 ABI extension, as with the <samp>-mfp32</samp> command-line
option.
</p></dd>
<dt><code>xx</code></dt>
<dd><p>Use the o32 FPXX ABI extension, as with the <samp>-mfpxx</samp> command-line
option.
</p></dd>
<dt><code>64</code></dt>
<dd><p>Use the o32 FP64 ABI extension, as with the <samp>-mfp64</samp> command-line
option.
</p></dd>
</dl>
<p>In the absence of this configuration option the default is to use the o32
FP32 ABI extension.
</p>
</dd>
<dt><code>--with-odd-spreg-32</code></dt>
<dd><p>On MIPS targets, set the <samp>-modd-spreg</samp> option by default when using
the o32 ABI.
</p>
</dd>
<dt><code>--without-odd-spreg-32</code></dt>
<dd><p>On MIPS targets, set the <samp>-mno-odd-spreg</samp> option by default when using
the o32 ABI. This is normally used in conjunction with
<samp>--with-fp-32=64</samp> in order to target the o32 FP64A ABI extension.
</p>
</dd>
<dt><code>--with-nan=<var>encoding</var></code></dt>
<dd><p>On MIPS targets, set the default encoding convention to use for the
special not-a-number (NaN) IEEE 754 floating-point data. The
possibilities for <var>encoding</var> are:
</p><dl compact="compact">
<dt><code>legacy</code></dt>
<dd><p>Use the legacy encoding, as with the <samp>-mnan=legacy</samp> command-line
option.
</p></dd>
<dt><code>2008</code></dt>
<dd><p>Use the 754-2008 encoding, as with the <samp>-mnan=2008</samp> command-line
option.
</p></dd>
</dl>
<p>To use this configuration option you must have an assembler version
installed that supports the <samp>-mnan=</samp> command-line option too.
In the absence of this configuration option the default convention is
the legacy encoding, as when neither of the <samp>-mnan=2008</samp> and
<samp>-mnan=legacy</samp> command-line options has been used.
</p>
</dd>
<dt><code>--with-divide=<var>type</var></code></dt>
<dd><p>Specify how the compiler should generate code for checking for
division by zero. This option is only supported on the MIPS target.
The possibilities for <var>type</var> are:
</p><dl compact="compact">
<dt><code>traps</code></dt>
<dd><p>Division by zero checks use conditional traps (this is the default on
systems that support conditional traps).
</p></dd>
<dt><code>breaks</code></dt>
<dd><p>Division by zero checks use the break instruction.
</p></dd>
</dl>
</dd>
<dt><code>--with-llsc</code></dt>
<dd><p>On MIPS targets, make <samp>-mllsc</samp> the default when no
<samp>-mno-llsc</samp> option is passed. This is the default for
Linux-based targets, as the kernel will emulate them if the ISA does
not provide them.
</p>
</dd>
<dt><code>--without-llsc</code></dt>
<dd><p>On MIPS targets, make <samp>-mno-llsc</samp> the default when no
<samp>-mllsc</samp> option is passed.
</p>
</dd>
<dt><code>--with-synci</code></dt>
<dd><p>On MIPS targets, make <samp>-msynci</samp> the default when no
<samp>-mno-synci</samp> option is passed.
</p>
</dd>
<dt><code>--without-synci</code></dt>
<dd><p>On MIPS targets, make <samp>-mno-synci</samp> the default when no
<samp>-msynci</samp> option is passed. This is the default.
</p>
</dd>
<dt><code>--with-lxc1-sxc1</code></dt>
<dd><p>On MIPS targets, make <samp>-mlxc1-sxc1</samp> the default when no
<samp>-mno-lxc1-sxc1</samp> option is passed. This is the default.
</p>
</dd>
<dt><code>--without-lxc1-sxc1</code></dt>
<dd><p>On MIPS targets, make <samp>-mno-lxc1-sxc1</samp> the default when no
<samp>-mlxc1-sxc1</samp> option is passed. The indexed load/store
instructions are not directly a problem but can lead to unexpected
behaviour when deployed in an application intended for a 32-bit address
space but run on a 64-bit processor. The issue is seen because all
known MIPS 64-bit Linux kernels execute o32 and n32 applications
with 64-bit addressing enabled which affects the overflow behaviour
of the indexed addressing mode. GCC will assume that ordinary
32-bit arithmetic overflow behaviour is the same whether performed
as an <code>addu</code> instruction or as part of the address calculation
in <code>lwxc1</code> type instructions. This assumption holds true in a
pure 32-bit environment and can hold true in a 64-bit environment if
the address space is accurately set to be 32-bit for o32 and n32.
</p>
</dd>
<dt><code>--with-madd4</code></dt>
<dd><p>On MIPS targets, make <samp>-mmadd4</samp> the default when no
<samp>-mno-madd4</samp> option is passed. This is the default.
</p>
</dd>
<dt><code>--without-madd4</code></dt>
<dd><p>On MIPS targets, make <samp>-mno-madd4</samp> the default when no
<samp>-mmadd4</samp> option is passed. The <code>madd4</code> instruction
family can be problematic when targeting a combination of cores that
implement these instructions differently. There are two known cores
that implement these as fused operations instead of unfused (where
unfused is normally expected). Disabling these instructions is the
only way to ensure compatible code is generated; this will incur
a performance penalty.
</p>
</dd>
<dt><code>--with-mips-plt</code></dt>
<dd><p>On MIPS targets, make use of copy relocations and PLTs.
These features are extensions to the traditional
SVR4-based MIPS ABIs and require support from GNU binutils
and the runtime C library.
</p>
</dd>
<dt><code>--enable-__cxa_atexit</code></dt>
<dd><p>Define if you want to use __cxa_atexit, rather than atexit, to
register C++ destructors for local statics and global objects.
This is essential for fully standards-compliant handling of
destructors, but requires __cxa_atexit in libc. This option is currently
only available on systems with GNU libc. When enabled, this will cause
<samp>-fuse-cxa-atexit</samp> to be passed by default.
</p>
</dd>
<dt><code>--enable-gnu-indirect-function</code></dt>
<dd><p>Define if you want to enable the <code>ifunc</code> attribute. This option is
currently only available on systems with GNU libc on certain targets.
</p>
</dd>
<dt><code>--enable-target-optspace</code></dt>
<dd><p>Specify that target
libraries should be optimized for code space instead of code speed.
This is the default for the m32r platform.
</p>
</dd>
<dt><code>--with-cpp-install-dir=<var>dirname</var></code></dt>
<dd><p>Specify that the user visible <code>cpp</code> program should be installed
in <samp><var>prefix</var>/<var>dirname</var>/cpp</samp>, in addition to <var>bindir</var>.
</p>
</dd>
<dt><code>--enable-comdat</code></dt>
<dd><p>Enable COMDAT group support. This is primarily used to override the
automatically detected value.
</p>
</dd>
<dt><code>--enable-initfini-array</code></dt>
<dd><p>Force the use of sections <code>.init_array</code> and <code>.fini_array</code>
(instead of <code>.init</code> and <code>.fini</code>) for constructors and
destructors. Option <samp>--disable-initfini-array</samp> has the
opposite effect. If neither option is specified, the configure script
will try to guess whether the <code>.init_array</code> and
<code>.fini_array</code> sections are supported and, if they are, use them.
</p>
</dd>
<dt><code>--enable-link-mutex</code></dt>
<dd><p>When building GCC, use a mutex to avoid linking the compilers for
multiple languages at the same time, to avoid thrashing on build
systems with limited free memory. The default is not to use such a mutex.
</p>
</dd>
<dt><code>--enable-maintainer-mode</code></dt>
<dd><p>The build rules that regenerate the Autoconf and Automake output files as
well as the GCC master message catalog <samp>gcc.pot</samp> are normally
disabled. This is because it can only be rebuilt if the complete source
tree is present. If you have changed the sources and want to rebuild the
catalog, configuring with <samp>--enable-maintainer-mode</samp> will enable
this. Note that you need a recent version of the <code>gettext</code> tools
to do so.
</p>
</dd>
<dt><code>--disable-bootstrap</code></dt>
<dd><p>For a native build, the default configuration is to perform
a 3-stage bootstrap of the compiler when &lsquo;<samp>make</samp>&rsquo; is invoked,
testing that GCC can compile itself correctly. If you want to disable
this process, you can configure with <samp>--disable-bootstrap</samp>.
</p>
</dd>
<dt><code>--enable-bootstrap</code></dt>
<dd><p>In special cases, you may want to perform a 3-stage build
even if the target and host triplets are different.
This is possible when the host can run code compiled for
the target (e.g. host is i686-linux, target is i486-linux).
Starting from GCC 4.2, to do this you have to configure explicitly
with <samp>--enable-bootstrap</samp>.
</p>
</dd>
<dt><code>--enable-generated-files-in-srcdir</code></dt>
<dd><p>Neither the .c and .h files that are generated from Bison and flex nor the
info manuals and man pages that are built from the .texi files are present
in the SVN development tree. When building GCC from that development tree,
or from one of our snapshots, those generated files are placed in your
build directory, which allows for the source to be in a readonly
directory.
</p>
<p>If you configure with <samp>--enable-generated-files-in-srcdir</samp> then those
generated files will go into the source directory. This is mainly intended
for generating release or prerelease tarballs of the GCC sources, since it
is not a requirement that the users of source releases to have flex, Bison,
or makeinfo.
</p>
</dd>
<dt><code>--enable-version-specific-runtime-libs</code></dt>
<dd><p>Specify
that runtime libraries should be installed in the compiler specific
subdirectory (<samp><var>libdir</var>/gcc</samp>) rather than the usual places. In
addition, &lsquo;<samp>libstdc++</samp>&rsquo;&rsquo;s include files will be installed into
<samp><var>libdir</var></samp> unless you overruled it by using
<samp>--with-gxx-include-dir=<var>dirname</var></samp>. Using this option is
particularly useful if you intend to use several versions of GCC in
parallel. This is currently supported by &lsquo;<samp>libgfortran</samp>&rsquo;,
&lsquo;<samp>libstdc++</samp>&rsquo;, and &lsquo;<samp>libobjc</samp>&rsquo;.
</p>
</dd>
<dt><code><a name="WithAixSoname"></a>--with-aix-soname=&lsquo;<samp>aix</samp>&rsquo;, &lsquo;<samp>svr4</samp>&rsquo; or &lsquo;<samp>both</samp>&rsquo;</code></dt>
<dd><p>Traditional AIX shared library versioning (versioned <code>Shared Object</code>
files as members of unversioned <code>Archive Library</code> files named
&lsquo;<samp>lib.a</samp>&rsquo;) causes numerous headaches for package managers. However,
<code>Import Files</code> as members of <code>Archive Library</code> files allow for
<strong>filename-based versioning</strong> of shared libraries as seen on Linux/SVR4,
where this is called the &quot;SONAME&quot;. But as they prevent static linking,
<code>Import Files</code> may be used with <code>Runtime Linking</code> only, where the
linker does search for &lsquo;<samp>libNAME.so</samp>&rsquo; before &lsquo;<samp>libNAME.a</samp>&rsquo; library
filenames with the &lsquo;<samp>-lNAME</samp>&rsquo; linker flag.
</p>
<a name="AixLdCommand"></a><p>For detailed information please refer to the AIX
<a href="https://www.ibm.com/support/knowledgecenter/search/%22the%20ld%20command%2C%20also%20called%20the%20linkage%20editor%20or%20binder%22">ld
Command</a> reference.
</p>
<p>As long as shared library creation is enabled, upon:
</p><dl compact="compact">
<dt><code>--with-aix-soname=aix</code></dt>
<dt><code>--with-aix-soname=both</code></dt>
<dd><p>A (traditional AIX) <code>Shared Archive Library</code> file is created:
</p><ul>
<li> using the &lsquo;<samp>libNAME.a</samp>&rsquo; filename scheme
</li><li> with the <code>Shared Object</code> file as archive member named
&lsquo;<samp>libNAME.so.V</samp>&rsquo; (except for &lsquo;<samp>libgcc_s</samp>&rsquo;, where the <code>Shared
Object</code> file is named &lsquo;<samp>shr.o</samp>&rsquo; for backwards compatibility), which
<ul class="no-bullet">
<li>- is used for runtime loading from inside the &lsquo;<samp>libNAME.a</samp>&rsquo; file
</li><li>- is used for dynamic loading via
<code>dlopen(&quot;libNAME.a(libNAME.so.V)&quot;, RTLD_MEMBER)</code>
</li><li>- is used for shared linking
</li><li>- is used for static linking, so no separate <code>Static Archive
Library</code> file is needed
</li></ul>
</li></ul>
</dd>
<dt><code>--with-aix-soname=both</code></dt>
<dt><code>--with-aix-soname=svr4</code></dt>
<dd><p>A (second) <code>Shared Archive Library</code> file is created:
</p><ul>
<li> using the &lsquo;<samp>libNAME.so.V</samp>&rsquo; filename scheme
</li><li> with the <code>Shared Object</code> file as archive member named
&lsquo;<samp>shr.o</samp>&rsquo;, which
<ul class="no-bullet">
<li>- is created with the <code>-G linker flag</code>
</li><li>- has the <code>F_LOADONLY</code> flag set
</li><li>- is used for runtime loading from inside the &lsquo;<samp>libNAME.so.V</samp>&rsquo; file
</li><li>- is used for dynamic loading via <code>dlopen(&quot;libNAME.so.V(shr.o)&quot;,
RTLD_MEMBER)</code>
</li></ul>
</li><li> with the <code>Import File</code> as archive member named &lsquo;<samp>shr.imp</samp>&rsquo;,
which
<ul class="no-bullet">
<li>- refers to &lsquo;<samp>libNAME.so.V(shr.o)</samp>&rsquo; as the &quot;SONAME&quot;, to be recorded
in the <code>Loader Section</code> of subsequent binaries
</li><li>- indicates whether &lsquo;<samp>libNAME.so.V(shr.o)</samp>&rsquo; is 32 or 64 bit
</li><li>- lists all the public symbols exported by &lsquo;<samp>lib.so.V(shr.o)</samp>&rsquo;,
eventually decorated with the <code>&lsquo;<samp>weak</samp>&rsquo; Keyword</code>
</li><li>- is necessary for shared linking against &lsquo;<samp>lib.so.V(shr.o)</samp>&rsquo;
</li></ul>
</li></ul>
<p>A symbolic link using the &lsquo;<samp>libNAME.so</samp>&rsquo; filename scheme is created:
</p><ul>
<li> pointing to the &lsquo;<samp>libNAME.so.V</samp>&rsquo; <code>Shared Archive Library</code> file
</li><li> to permit the <code>ld Command</code> to find &lsquo;<samp>lib.so.V(shr.imp)</samp>&rsquo; via
the &lsquo;<samp>-lNAME</samp>&rsquo; argument (requires <code>Runtime Linking</code> to be enabled)
</li><li> to permit dynamic loading of &lsquo;<samp>lib.so.V(shr.o)</samp>&rsquo; without the need
to specify the version number via <code>dlopen(&quot;libNAME.so(shr.o)&quot;,
RTLD_MEMBER)</code>
</li></ul>
</dd>
</dl>
<p>As long as static library creation is enabled, upon:
</p><dl compact="compact">
<dt><code>--with-aix-soname=svr4</code></dt>
<dd><p>A <code>Static Archive Library</code> is created:
</p><ul>
<li> using the &lsquo;<samp>libNAME.a</samp>&rsquo; filename scheme
</li><li> with all the <code>Static Object</code> files as archive members, which
<ul class="no-bullet">
<li>- are used for static linking
</li></ul>
</li></ul>
</dd>
</dl>
<p>While the aix-soname=&lsquo;<samp>svr4</samp>&rsquo; option does not create <code>Shared Object</code>
files as members of unversioned <code>Archive Library</code> files any more, package
managers still are responsible to
<a href="./specific.html#TransferAixShobj">transfer</a> <code>Shared Object</code> files
found as member of a previously installed unversioned <code>Archive Library</code>
file into the newly installed <code>Archive Library</code> file with the same
filename.
</p>
<p><em>WARNING:</em> Creating <code>Shared Object</code> files with <code>Runtime Linking</code>
enabled may bloat the TOC, eventually leading to <code>TOC overflow</code> errors,
requiring the use of either the <samp>-Wl,-bbigtoc</samp> linker flag (seen to
break with the <code>GDB</code> debugger) or some of the TOC-related compiler flags,
see &ldquo;RS/6000 and PowerPC Options&rdquo; in the main manual.
</p>
<p><samp>--with-aix-soname</samp> is currently supported by &lsquo;<samp>libgcc_s</samp>&rsquo; only, so
this option is still experimental and not for normal use yet.
</p>
<p>Default is the traditional behavior <samp>--with-aix-soname=&lsquo;<samp>aix</samp>&rsquo;</samp>.
</p>
</dd>
<dt><code>--enable-languages=<var>lang1</var>,<var>lang2</var>,&hellip;</code></dt>
<dd><p>Specify that only a particular subset of compilers and
their runtime libraries should be built. For a list of valid values for
<var>langN</var> you can issue the following command in the
<samp>gcc</samp> directory of your GCC source tree:<br>
</p><div class="smallexample">
<pre class="smallexample">grep ^language= */config-lang.in
</pre></div>
<p>Currently, you can use any of the following:
<code>all</code>, <code>default</code>, <code>ada</code>, <code>c</code>, <code>c++</code>, <code>fortran</code>,
<code>go</code>, <code>jit</code>, <code>lto</code>, <code>objc</code>, <code>obj-c++</code>.
Building the Ada compiler has special requirements, see below.
If you do not pass this flag, or specify the option <code>default</code>, then the
default languages available in the <samp>gcc</samp> sub-tree will be configured.
Ada, Go, Jit, and Objective-C++ are not default languages. LTO is not a
default language, but is built by default because <samp>--enable-lto</samp> is
enabled by default. The other languages are default languages. If
<code>all</code> is specified, then all available languages are built. An
exception is <code>jit</code> language, which requires
<samp>--enable-host-shared</samp> to be included with <code>all</code>.
</p>
</dd>
<dt><code>--enable-stage1-languages=<var>lang1</var>,<var>lang2</var>,&hellip;</code></dt>
<dd><p>Specify that a particular subset of compilers and their runtime
libraries should be built with the system C compiler during stage 1 of
the bootstrap process, rather than only in later stages with the
bootstrapped C compiler. The list of valid values is the same as for
<samp>--enable-languages</samp>, and the option <code>all</code> will select all
of the languages enabled by <samp>--enable-languages</samp>. This option is
primarily useful for GCC development; for instance, when a development
version of the compiler cannot bootstrap due to compiler bugs, or when
one is debugging front ends other than the C front end. When this
option is used, one can then build the target libraries for the
specified languages with the stage-1 compiler by using <code>make
stage1-bubble all-target</code>, or run the testsuite on the stage-1 compiler
for the specified languages using <code>make stage1-start check-gcc</code>.
</p>
</dd>
<dt><code>--disable-libada</code></dt>
<dd><p>Specify that the run-time libraries and tools used by GNAT should not
be built. This can be useful for debugging, or for compatibility with
previous Ada build procedures, when it was required to explicitly
do a &lsquo;<samp>make -C gcc gnatlib_and_tools</samp>&rsquo;.
</p>
</dd>
<dt><code>--disable-libsanitizer</code></dt>
<dd><p>Specify that the run-time libraries for the various sanitizers should
not be built.
</p>
</dd>
<dt><code>--disable-libssp</code></dt>
<dd><p>Specify that the run-time libraries for stack smashing protection
should not be built or linked against. On many targets library support
is provided by the C library instead.
</p>
</dd>
<dt><code>--disable-libquadmath</code></dt>
<dd><p>Specify that the GCC quad-precision math library should not be built.
On some systems, the library is required to be linkable when building
the Fortran front end, unless <samp>--disable-libquadmath-support</samp>
is used.
</p>
</dd>
<dt><code>--disable-libquadmath-support</code></dt>
<dd><p>Specify that the Fortran front end and <code>libgfortran</code> do not add
support for <code>libquadmath</code> on systems supporting it.
</p>
</dd>
<dt><code>--disable-libgomp</code></dt>
<dd><p>Specify that the GNU Offloading and Multi Processing Runtime Library
should not be built.
</p>
</dd>
<dt><code>--disable-libvtv</code></dt>
<dd><p>Specify that the run-time libraries used by vtable verification
should not be built.
</p>
</dd>
<dt><code>--with-dwarf2</code></dt>
<dd><p>Specify that the compiler should
use DWARF 2 debugging information as the default.
</p>
</dd>
<dt><code>--with-advance-toolchain=<var>at</var></code></dt>
<dd><p>On 64-bit PowerPC Linux systems, configure the compiler to use the
header files, library files, and the dynamic linker from the Advance
Toolchain release <var>at</var> instead of the default versions that are
provided by the Linux distribution. In general, this option is
intended for the developers of GCC, and it is not intended for general
use.
</p>
</dd>
<dt><code>--enable-targets=all</code></dt>
<dt><code>--enable-targets=<var>target_list</var></code></dt>
<dd><p>Some GCC targets, e.g. powerpc64-linux, build bi-arch compilers.
These are compilers that are able to generate either 64-bit or 32-bit
code. Typically, the corresponding 32-bit target, e.g.
powerpc-linux for powerpc64-linux, only generates 32-bit code. This
option enables the 32-bit target to be a bi-arch compiler, which is
useful when you want a bi-arch compiler that defaults to 32-bit, and
you are building a bi-arch or multi-arch binutils in a combined tree.
On mips-linux, this will build a tri-arch compiler (ABI o32/n32/64),
defaulted to o32.
Currently, this option only affects sparc-linux, powerpc-linux, x86-linux,
mips-linux and s390-linux.
</p>
</dd>
<dt><code>--enable-default-pie</code></dt>
<dd><p>Turn on <samp>-fPIE</samp> and <samp>-pie</samp> by default.
</p>
</dd>
<dt><code>--enable-secureplt</code></dt>
<dd><p>This option enables <samp>-msecure-plt</samp> by default for powerpc-linux.
See &ldquo;RS/6000 and PowerPC Options&rdquo; in the main manual
</p>
</dd>
<dt><code>--enable-default-ssp</code></dt>
<dd><p>Turn on <samp>-fstack-protector-strong</samp> by default.
</p>
</dd>
<dt><code>--enable-cld</code></dt>
<dd><p>This option enables <samp>-mcld</samp> by default for 32-bit x86 targets.
See &ldquo;i386 and x86-64 Options&rdquo; in the main manual
</p>
</dd>
<dt><code>--enable-win32-registry</code></dt>
<dt><code>--enable-win32-registry=<var>key</var></code></dt>
<dt><code>--disable-win32-registry</code></dt>
<dd><p>The <samp>--enable-win32-registry</samp> option enables Microsoft Windows-hosted GCC
to look up installations paths in the registry using the following key:
</p>
<div class="smallexample">
<pre class="smallexample"><code>HKEY_LOCAL_MACHINE\SOFTWARE\Free Software Foundation\<var>key</var></code>
</pre></div>
<p><var>key</var> defaults to GCC version number, and can be overridden by the
<samp>--enable-win32-registry=<var>key</var></samp> option. Vendors and distributors
who use custom installers are encouraged to provide a different key,
perhaps one comprised of vendor name and GCC version number, to
avoid conflict with existing installations. This feature is enabled
by default, and can be disabled by <samp>--disable-win32-registry</samp>
option. This option has no effect on the other hosts.
</p>
</dd>
<dt><code>--nfp</code></dt>
<dd><p>Specify that the machine does not have a floating point unit. This
option only applies to &lsquo;<samp>m68k-sun-sunos<var>n</var></samp>&rsquo;. On any other
system, <samp>--nfp</samp> has no effect.
</p>
</dd>
<dt><code>--enable-werror</code></dt>
<dt><code>--disable-werror</code></dt>
<dt><code>--enable-werror=yes</code></dt>
<dt><code>--enable-werror=no</code></dt>
<dd><p>When you specify this option, it controls whether certain files in the
compiler are built with <samp>-Werror</samp> in bootstrap stage2 and later.
If you don&rsquo;t specify it, <samp>-Werror</samp> is turned on for the main
development trunk. However it defaults to off for release branches and
final releases. The specific files which get <samp>-Werror</samp> are
controlled by the Makefiles.
</p>
</dd>
<dt><code>--enable-checking</code></dt>
<dt><code>--enable-checking=<var>list</var></code></dt>
<dd><p>When you specify this option, the compiler is built to perform internal
consistency checks of the requested complexity. This does not change the
generated code, but adds error checking within the compiler. This will
slow down the compiler and may only work properly if you are building
the compiler with GCC. This is &lsquo;<samp>yes,extra</samp>&rsquo; by default when building
from SVN or snapshots, but &lsquo;<samp>release</samp>&rsquo; for releases. The default
for building the stage1 compiler is &lsquo;<samp>yes</samp>&rsquo;. More control
over the checks may be had by specifying <var>list</var>. The categories of
checks available are &lsquo;<samp>yes</samp>&rsquo; (most common checks
&lsquo;<samp>assert,misc,tree,gc,rtlflag,runtime</samp>&rsquo;), &lsquo;<samp>no</samp>&rsquo; (no checks at
all), &lsquo;<samp>all</samp>&rsquo; (all but &lsquo;<samp>valgrind</samp>&rsquo;), &lsquo;<samp>release</samp>&rsquo; (cheapest
checks &lsquo;<samp>assert,runtime</samp>&rsquo;) or &lsquo;<samp>none</samp>&rsquo; (same as &lsquo;<samp>no</samp>&rsquo;).
Individual checks can be enabled with these flags &lsquo;<samp>assert</samp>&rsquo;,
&lsquo;<samp>df</samp>&rsquo;, &lsquo;<samp>fold</samp>&rsquo;, &lsquo;<samp>gc</samp>&rsquo;, &lsquo;<samp>gcac</samp>&rsquo;, &lsquo;<samp>misc</samp>&rsquo;, &lsquo;<samp>rtl</samp>&rsquo;,
&lsquo;<samp>rtlflag</samp>&rsquo;, &lsquo;<samp>runtime</samp>&rsquo;, &lsquo;<samp>tree</samp>&rsquo;, &lsquo;<samp>extra</samp>&rsquo; and &lsquo;<samp>valgrind</samp>&rsquo;.
&lsquo;<samp>extra</samp>&rsquo; adds for &lsquo;<samp>misc</samp>&rsquo; checking extra checks that might affect
code generation and should therefore not differ between stage1 and later
stages.
</p>
<p>The &lsquo;<samp>valgrind</samp>&rsquo; check requires the external <code>valgrind</code>
simulator, available from <a href="http://valgrind.org/">http://valgrind.org/</a>. The
&lsquo;<samp>df</samp>&rsquo;, &lsquo;<samp>rtl</samp>&rsquo;, &lsquo;<samp>gcac</samp>&rsquo; and &lsquo;<samp>valgrind</samp>&rsquo; checks are very expensive.
To disable all checking, &lsquo;<samp>--disable-checking</samp>&rsquo; or
&lsquo;<samp>--enable-checking=none</samp>&rsquo; must be explicitly requested. Disabling
assertions will make the compiler and runtime slightly faster but
increase the risk of undetected internal errors causing wrong code to be
generated.
</p>
</dd>
<dt><code>--disable-stage1-checking</code></dt>
<dt><code>--enable-stage1-checking</code></dt>
<dt><code>--enable-stage1-checking=<var>list</var></code></dt>
<dd><p>If no <samp>--enable-checking</samp> option is specified the stage1
compiler will be built with &lsquo;<samp>yes</samp>&rsquo; checking enabled, otherwise
the stage1 checking flags are the same as specified by
<samp>--enable-checking</samp>. To build the stage1 compiler with
different checking options use <samp>--enable-stage1-checking</samp>.
The list of checking options is the same as for <samp>--enable-checking</samp>.
If your system is too slow or too small to bootstrap a released compiler
with checking for stage1 enabled, you can use &lsquo;<samp>--disable-stage1-checking</samp>&rsquo;
to disable checking for the stage1 compiler.
</p>
</dd>
<dt><code>--enable-coverage</code></dt>
<dt><code>--enable-coverage=<var>level</var></code></dt>
<dd><p>With this option, the compiler is built to collect self coverage
information, every time it is run. This is for internal development
purposes, and only works when the compiler is being built with gcc. The
<var>level</var> argument controls whether the compiler is built optimized or
not, values are &lsquo;<samp>opt</samp>&rsquo; and &lsquo;<samp>noopt</samp>&rsquo;. For coverage analysis you
want to disable optimization, for performance analysis you want to
enable optimization. When coverage is enabled, the default level is
without optimization.
</p>
</dd>
<dt><code>--enable-gather-detailed-mem-stats</code></dt>
<dd><p>When this option is specified more detailed information on memory
allocation is gathered. This information is printed when using
<samp>-fmem-report</samp>.
</p>
</dd>
<dt><code>--enable-valgrind-annotations</code></dt>
<dd><p>Mark selected memory related operations in the compiler when run under
valgrind to suppress false positives.
</p>
</dd>
<dt><code>--enable-nls</code></dt>
<dt><code>--disable-nls</code></dt>
<dd><p>The <samp>--enable-nls</samp> option enables Native Language Support (NLS),
which lets GCC output diagnostics in languages other than American
English. Native Language Support is enabled by default if not doing a
canadian cross build. The <samp>--disable-nls</samp> option disables NLS.
</p>
</dd>
<dt><code>--with-included-gettext</code></dt>
<dd><p>If NLS is enabled, the <samp>--with-included-gettext</samp> option causes the build
procedure to prefer its copy of GNU <code>gettext</code>.
</p>
</dd>
<dt><code>--with-catgets</code></dt>
<dd><p>If NLS is enabled, and if the host lacks <code>gettext</code> but has the
inferior <code>catgets</code> interface, the GCC build procedure normally
ignores <code>catgets</code> and instead uses GCC&rsquo;s copy of the GNU
<code>gettext</code> library. The <samp>--with-catgets</samp> option causes the
build procedure to use the host&rsquo;s <code>catgets</code> in this situation.
</p>
</dd>
<dt><code>--with-libiconv-prefix=<var>dir</var></code></dt>
<dd><p>Search for libiconv header files in <samp><var>dir</var>/include</samp> and
libiconv library files in <samp><var>dir</var>/lib</samp>.
</p>
</dd>
<dt><code>--enable-obsolete</code></dt>
<dd><p>Enable configuration for an obsoleted system. If you attempt to
configure GCC for a system (build, host, or target) which has been
obsoleted, and you do not specify this flag, configure will halt with an
error message.
</p>
<p>All support for systems which have been obsoleted in one release of GCC
is removed entirely in the next major release, unless someone steps
forward to maintain the port.
</p>
</dd>
<dt><code>--enable-decimal-float</code></dt>
<dt><code>--enable-decimal-float=yes</code></dt>
<dt><code>--enable-decimal-float=no</code></dt>
<dt><code>--enable-decimal-float=bid</code></dt>
<dt><code>--enable-decimal-float=dpd</code></dt>
<dt><code>--disable-decimal-float</code></dt>
<dd><p>Enable (or disable) support for the C decimal floating point extension
that is in the IEEE 754-2008 standard. This is enabled by default only
on PowerPC, i386, and x86_64 GNU/Linux systems. Other systems may also
support it, but require the user to specifically enable it. You can
optionally control which decimal floating point format is used (either
&lsquo;<samp>bid</samp>&rsquo; or &lsquo;<samp>dpd</samp>&rsquo;). The &lsquo;<samp>bid</samp>&rsquo; (binary integer decimal)
format is default on i386 and x86_64 systems, and the &lsquo;<samp>dpd</samp>&rsquo;
(densely packed decimal) format is default on PowerPC systems.
</p>
</dd>
<dt><code>--enable-fixed-point</code></dt>
<dt><code>--disable-fixed-point</code></dt>
<dd><p>Enable (or disable) support for C fixed-point arithmetic.
This option is enabled by default for some targets (such as MIPS) which
have hardware-support for fixed-point operations. On other targets, you
may enable this option manually.
</p>
</dd>
<dt><code>--with-long-double-128</code></dt>
<dd><p>Specify if <code>long double</code> type should be 128-bit by default on selected
GNU/Linux architectures. If using <code>--without-long-double-128</code>,
<code>long double</code> will be by default 64-bit, the same as <code>double</code> type.
When neither of these configure options are used, the default will be
128-bit <code>long double</code> when built against GNU C Library 2.4 and later,
64-bit <code>long double</code> otherwise.
</p>
</dd>
<dt><code>--with-long-double-format=ibm</code></dt>
<dt><code>--with-long-double-format=ieee</code></dt>
<dd><p>Specify whether <code>long double</code> uses the IBM extended double format
or the IEEE 128-bit floating point format on PowerPC Linux systems.
This configuration switch will only work on little endian PowerPC
Linux systems and on big endian 64-bit systems where the default cpu
is at least power7 (i.e. <samp>--with-cpu=power7</samp>,
<samp>--with-cpu=power8</samp>, or <samp>--with-cpu=power9</samp> is used).
</p>
<p>If you use the <samp>--with-long-double-64</samp> configuration option,
the <samp>--with-long-double-format=ibm</samp> and
<samp>--with-long-double-format=ieee</samp> options are ignored.
</p>
<p>The default <code>long double</code> format is to use IBM extended double.
Until all of the libraries are converted to use IEEE 128-bit floating
point, it is not recommended to use
<samp>--with-long-double-format=ieee</samp>.
</p>
<p>On little endian PowerPC Linux systems, if you explicitly set the
<code>long double</code> type, it will build multilibs to allow you to
select either <code>long double</code> format, unless you disable multilibs
with the <code>--disable-multilib</code> option. At present,
<code>long double</code> multilibs are not built on big endian PowerPC Linux
systems. If you are building multilibs, you will need to configure
the compiler using the <samp>--with-system-zlib</samp> option.
</p>
<p>If you do not set the <code>long double</code> type explicitly, no multilibs
will be generated.
</p>
</dd>
<dt><code>--enable-fdpic</code></dt>
<dd><p>On SH Linux systems, generate ELF FDPIC code.
</p>
</dd>
<dt><code>--with-gmp=<var>pathname</var></code></dt>
<dt><code>--with-gmp-include=<var>pathname</var></code></dt>
<dt><code>--with-gmp-lib=<var>pathname</var></code></dt>
<dt><code>--with-mpfr=<var>pathname</var></code></dt>
<dt><code>--with-mpfr-include=<var>pathname</var></code></dt>
<dt><code>--with-mpfr-lib=<var>pathname</var></code></dt>
<dt><code>--with-mpc=<var>pathname</var></code></dt>
<dt><code>--with-mpc-include=<var>pathname</var></code></dt>
<dt><code>--with-mpc-lib=<var>pathname</var></code></dt>
<dd><p>If you want to build GCC but do not have the GMP library, the MPFR
library and/or the MPC library installed in a standard location and
do not have their sources present in the GCC source tree then you
can explicitly specify the directory where they are installed
(&lsquo;<samp>--with-gmp=<var>gmpinstalldir</var></samp>&rsquo;,
&lsquo;<samp>--with-mpfr=<var>mpfrinstalldir</var></samp>&rsquo;,
&lsquo;<samp>--with-mpc=<var>mpcinstalldir</var></samp>&rsquo;). The
<samp>--with-gmp=<var>gmpinstalldir</var></samp> option is shorthand for
<samp>--with-gmp-lib=<var>gmpinstalldir</var>/lib</samp> and
<samp>--with-gmp-include=<var>gmpinstalldir</var>/include</samp>. Likewise the
<samp>--with-mpfr=<var>mpfrinstalldir</var></samp> option is shorthand for
<samp>--with-mpfr-lib=<var>mpfrinstalldir</var>/lib</samp> and
<samp>--with-mpfr-include=<var>mpfrinstalldir</var>/include</samp>, also the
<samp>--with-mpc=<var>mpcinstalldir</var></samp> option is shorthand for
<samp>--with-mpc-lib=<var>mpcinstalldir</var>/lib</samp> and
<samp>--with-mpc-include=<var>mpcinstalldir</var>/include</samp>. If these
shorthand assumptions are not correct, you can use the explicit
include and lib options directly. You might also need to ensure the
shared libraries can be found by the dynamic linker when building and
using GCC, for example by setting the runtime shared library path
variable (<code>LD_LIBRARY_PATH</code> on GNU/Linux and Solaris systems).
</p>
<p>These flags are applicable to the host platform only. When building
a cross compiler, they will not be used to configure target libraries.
</p>
</dd>
<dt><code>--with-isl=<var>pathname</var></code></dt>
<dt><code>--with-isl-include=<var>pathname</var></code></dt>
<dt><code>--with-isl-lib=<var>pathname</var></code></dt>
<dd><p>If you do not have the isl library installed in a standard location and you
want to build GCC, you can explicitly specify the directory where it is
installed (&lsquo;<samp>--with-isl=<var>islinstalldir</var></samp>&rsquo;). The
<samp>--with-isl=<var>islinstalldir</var></samp> option is shorthand for
<samp>--with-isl-lib=<var>islinstalldir</var>/lib</samp> and
<samp>--with-isl-include=<var>islinstalldir</var>/include</samp>. If this
shorthand assumption is not correct, you can use the explicit
include and lib options directly.
</p>
<p>These flags are applicable to the host platform only. When building
a cross compiler, they will not be used to configure target libraries.
</p>
</dd>
<dt><code>--with-stage1-ldflags=<var>flags</var></code></dt>
<dd><p>This option may be used to set linker flags to be used when linking
stage 1 of GCC. These are also used when linking GCC if configured with
<samp>--disable-bootstrap</samp>. If <samp>--with-stage1-libs</samp> is not set to a
value, then the default is &lsquo;<samp>-static-libstdc++ -static-libgcc</samp>&rsquo;, if
supported.
</p>
</dd>
<dt><code>--with-stage1-libs=<var>libs</var></code></dt>
<dd><p>This option may be used to set libraries to be used when linking stage 1
of GCC. These are also used when linking GCC if configured with
<samp>--disable-bootstrap</samp>.
</p>
</dd>
<dt><code>--with-boot-ldflags=<var>flags</var></code></dt>
<dd><p>This option may be used to set linker flags to be used when linking
stage 2 and later when bootstrapping GCC. If &ndash;with-boot-libs
is not is set to a value, then the default is
&lsquo;<samp>-static-libstdc++ -static-libgcc</samp>&rsquo;.
</p>
</dd>
<dt><code>--with-boot-libs=<var>libs</var></code></dt>
<dd><p>This option may be used to set libraries to be used when linking stage 2
and later when bootstrapping GCC.
</p>
</dd>
<dt><code>--with-debug-prefix-map=<var>map</var></code></dt>
<dd><p>Convert source directory names using <samp>-fdebug-prefix-map</samp> when
building runtime libraries. &lsquo;<samp><var>map</var></samp>&rsquo; is a space-separated
list of maps of the form &lsquo;<samp><var>old</var>=<var>new</var></samp>&rsquo;.
</p>
</dd>
<dt><code>--enable-linker-build-id</code></dt>
<dd><p>Tells GCC to pass <samp>--build-id</samp> option to the linker for all final
links (links performed without the <samp>-r</samp> or <samp>--relocatable</samp>
option), if the linker supports it. If you specify
<samp>--enable-linker-build-id</samp>, but your linker does not
support <samp>--build-id</samp> option, a warning is issued and the
<samp>--enable-linker-build-id</samp> option is ignored. The default is off.
</p>
</dd>
<dt><code>--with-linker-hash-style=<var>choice</var></code></dt>
<dd><p>Tells GCC to pass <samp>--hash-style=<var>choice</var></samp> option to the
linker for all final links. <var>choice</var> can be one of
&lsquo;<samp>sysv</samp>&rsquo;, &lsquo;<samp>gnu</samp>&rsquo;, and &lsquo;<samp>both</samp>&rsquo; where &lsquo;<samp>sysv</samp>&rsquo; is the default.
</p>
</dd>
<dt><code>--enable-gnu-unique-object</code></dt>
<dt><code>--disable-gnu-unique-object</code></dt>
<dd><p>Tells GCC to use the gnu_unique_object relocation for C++ template
static data members and inline function local statics. Enabled by
default for a toolchain with an assembler that accepts it and
GLIBC 2.11 or above, otherwise disabled.
</p>
</dd>
<dt><code>--with-diagnostics-color=<var>choice</var></code></dt>
<dd><p>Tells GCC to use <var>choice</var> as the default for <samp>-fdiagnostics-color=</samp>
option (if not used explicitly on the command line). <var>choice</var>
can be one of &lsquo;<samp>never</samp>&rsquo;, &lsquo;<samp>auto</samp>&rsquo;, &lsquo;<samp>always</samp>&rsquo;, and &lsquo;<samp>auto-if-env</samp>&rsquo;
where &lsquo;<samp>auto</samp>&rsquo; is the default. &lsquo;<samp>auto-if-env</samp>&rsquo; means that
<samp>-fdiagnostics-color=auto</samp> will be the default if <code>GCC_COLORS</code>
is present and non-empty in the environment, and
<samp>-fdiagnostics-color=never</samp> otherwise.
</p>
</dd>
<dt><code>--enable-lto</code></dt>
<dt><code>--disable-lto</code></dt>
<dd><p>Enable support for link-time optimization (LTO). This is enabled by
default, and may be disabled using <samp>--disable-lto</samp>.
</p>
</dd>
<dt><code>--enable-linker-plugin-configure-flags=FLAGS</code></dt>
<dt><code>--enable-linker-plugin-flags=FLAGS</code></dt>
<dd><p>By default, linker plugins (such as the LTO plugin) are built for the
host system architecture. For the case that the linker has a
different (but run-time compatible) architecture, these flags can be
specified to build plugins that are compatible to the linker. For
example, if you are building GCC for a 64-bit x86_64
(&lsquo;<samp>x86_64-unknown-linux-gnu</samp>&rsquo;) host system, but have a 32-bit x86
GNU/Linux (&lsquo;<samp>i686-pc-linux-gnu</samp>&rsquo;) linker executable (which is
executable on the former system), you can configure GCC as follows for
getting compatible linker plugins:
</p>
<div class="smallexample">
<pre class="smallexample">% <var>srcdir</var>/configure \
--host=x86_64-unknown-linux-gnu \
--enable-linker-plugin-configure-flags=--host=i686-pc-linux-gnu \
--enable-linker-plugin-flags='CC=gcc\ -m32\ -Wl,-rpath,[...]/i686-pc-linux-gnu/lib'
</pre></div>
</dd>
<dt><code>--with-plugin-ld=<var>pathname</var></code></dt>
<dd><p>Enable an alternate linker to be used at link-time optimization (LTO)
link time when <samp>-fuse-linker-plugin</samp> is enabled.
This linker should have plugin support such as gold starting with
version 2.20 or GNU ld starting with version 2.21.
See <samp>-fuse-linker-plugin</samp> for details.
</p>
</dd>
<dt><code>--enable-canonical-system-headers</code></dt>
<dt><code>--disable-canonical-system-headers</code></dt>
<dd><p>Enable system header path canonicalization for <samp>libcpp</samp>. This can
produce shorter header file paths in diagnostics and dependency output
files, but these changed header paths may conflict with some compilation
environments. Enabled by default, and may be disabled using
<samp>--disable-canonical-system-headers</samp>.
</p>
</dd>
<dt><code>--with-glibc-version=<var>major</var>.<var>minor</var></code></dt>
<dd><p>Tell GCC that when the GNU C Library (glibc) is used on the target it
will be version <var>major</var>.<var>minor</var> or later. Normally this can
be detected from the C library&rsquo;s header files, but this option may be
needed when bootstrapping a cross toolchain without the header files
available for building the initial bootstrap compiler.
</p>
<p>If GCC is configured with some multilibs that use glibc and some that
do not, this option applies only to the multilibs that use glibc.
However, such configurations may not work well as not all the relevant
configuration in GCC is on a per-multilib basis.
</p>
</dd>
<dt><code>--enable-as-accelerator-for=<var>target</var></code></dt>
<dd><p>Build as offload target compiler. Specify offload host triple by <var>target</var>.
</p>
</dd>
<dt><code>--enable-offload-targets=<var>target1</var>[=<var>path1</var>],&hellip;,<var>targetN</var>[=<var>pathN</var>]</code></dt>
<dd><p>Enable offloading to targets <var>target1</var>, &hellip;, <var>targetN</var>.
Offload compilers are expected to be already installed. Default search
path for them is <samp><var>exec-prefix</var></samp>, but it can be changed by
specifying paths <var>path1</var>, &hellip;, <var>pathN</var>.
</p>
<div class="smallexample">
<pre class="smallexample">% <var>srcdir</var>/configure \
--enable-offload-target=i686-unknown-linux-gnu=/path/to/i686/compiler,x86_64-pc-linux-gnu
</pre></div>
<p>If &lsquo;<samp>hsa</samp>&rsquo; is specified as one of the targets, the compiler will be
built with support for HSA GPU accelerators. Because the same
compiler will emit the accelerator code, no path should be specified.
</p>
</dd>
<dt><code>--with-hsa-runtime=<var>pathname</var></code></dt>
<dt><code>--with-hsa-runtime-include=<var>pathname</var></code></dt>
<dt><code>--with-hsa-runtime-lib=<var>pathname</var></code></dt>
<dd>
<p>If you configure GCC with HSA offloading but do not have the HSA
run-time library installed in a standard location then you can
explicitly specify the directory where they are installed. The
<samp>--with-hsa-runtime=<var>hsainstalldir</var></samp> option is a
shorthand for
<samp>--with-hsa-runtime-lib=<var>hsainstalldir</var>/lib</samp> and
<samp>--with-hsa-runtime-include=<var>hsainstalldir</var>/include</samp>.
</p>
</dd>
<dt><code>--enable-cet</code></dt>
<dt><code>--disable-cet</code></dt>
<dd><p>Enable building target run-time libraries with control-flow
instrumentation, see <samp>-fcf-protection</samp> option. When
<code>--enable-cet</code> is specified target libraries are configured
to add <samp>-fcf-protection</samp> and, if needed, other target
specific options to a set of building options.
</p>
<p>The option is disabled by default. When <code>--enable-cet=auto</code>
is used, it is enabled on Linux/x86 if target binutils
supports <code>Intel CET</code> instructions and disabled otherwise.
In this case the target libraries are configured to get additional
<samp>-fcf-protection</samp> option.
</p>
</dd>
<dt><code>--with-riscv-attribute=&lsquo;<samp>yes</samp>&rsquo;, &lsquo;<samp>no</samp>&rsquo; or &lsquo;<samp>default</samp>&rsquo;</code></dt>
<dd><p>Generate RISC-V attribute by default, in order to record extra build
information in object.
</p>
<p>The option is disabled by default. It is enabled on RISC-V/ELF (bare-metal)
target if target binutils supported.
</p></dd>
</dl>
<a name="Cross-Compiler-Specific-Options"></a>
<h4 class="subheading">Cross-Compiler-Specific Options</h4>
<p>The following options only apply to building cross compilers.
</p>
<dl compact="compact">
<dt><code>--with-sysroot</code></dt>
<dt><code>--with-sysroot=<var>dir</var></code></dt>
<dd><p>Tells GCC to consider <var>dir</var> as the root of a tree that contains
(a subset of) the root filesystem of the target operating system.
Target system headers, libraries and run-time object files will be
searched for in there. More specifically, this acts as if
<samp>--sysroot=<var>dir</var></samp> was added to the default options of the built
compiler. The specified directory is not copied into the
install tree, unlike the options <samp>--with-headers</samp> and
<samp>--with-libs</samp> that this option obsoletes. The default value,
in case <samp>--with-sysroot</samp> is not given an argument, is
<samp>${gcc_tooldir}/sys-root</samp>. If the specified directory is a
subdirectory of <samp>${exec_prefix}</samp>, then it will be found relative to
the GCC binaries if the installation tree is moved.
</p>
<p>This option affects the system root for the compiler used to build
target libraries (which runs on the build system) and the compiler newly
installed with <code>make install</code>; it does not affect the compiler which is
used to build GCC itself.
</p>
<p>If you specify the <samp>--with-native-system-header-dir=<var>dirname</var></samp>
option then the compiler will search that directory within <var>dirname</var> for
native system headers rather than the default <samp>/usr/include</samp>.
</p>
</dd>
<dt><code>--with-build-sysroot</code></dt>
<dt><code>--with-build-sysroot=<var>dir</var></code></dt>
<dd><p>Tells GCC to consider <var>dir</var> as the system root (see
<samp>--with-sysroot</samp>) while building target libraries, instead of
the directory specified with <samp>--with-sysroot</samp>. This option is
only useful when you are already using <samp>--with-sysroot</samp>. You
can use <samp>--with-build-sysroot</samp> when you are configuring with
<samp>--prefix</samp> set to a directory that is different from the one in
which you are installing GCC and your target libraries.
</p>
<p>This option affects the system root for the compiler used to build
target libraries (which runs on the build system); it does not affect
the compiler which is used to build GCC itself.
</p>
<p>If you specify the <samp>--with-native-system-header-dir=<var>dirname</var></samp>
option then the compiler will search that directory within <var>dirname</var> for
native system headers rather than the default <samp>/usr/include</samp>.
</p>
</dd>
<dt><code>--with-headers</code></dt>
<dt><code>--with-headers=<var>dir</var></code></dt>
<dd><p>Deprecated in favor of <samp>--with-sysroot</samp>.
Specifies that target headers are available when building a cross compiler.
The <var>dir</var> argument specifies a directory which has the target include
files. These include files will be copied into the <samp>gcc</samp> install
directory. <em>This option with the <var>dir</var> argument is required</em> when
building a cross compiler, if <samp><var>prefix</var>/<var>target</var>/sys-include</samp>
doesn&rsquo;t pre-exist. If <samp><var>prefix</var>/<var>target</var>/sys-include</samp> does
pre-exist, the <var>dir</var> argument may be omitted. <code>fixincludes</code>
will be run on these files to make them compatible with GCC.
</p>
</dd>
<dt><code>--without-headers</code></dt>
<dd><p>Tells GCC not use any target headers from a libc when building a cross
compiler. When crossing to GNU/Linux, you need the headers so GCC
can build the exception handling for libgcc.
</p>
</dd>
<dt><code>--with-libs</code></dt>
<dt><code>--with-libs=&quot;<var>dir1</var> <var>dir2</var> &hellip; <var>dirN</var>&quot;</code></dt>
<dd><p>Deprecated in favor of <samp>--with-sysroot</samp>.
Specifies a list of directories which contain the target runtime
libraries. These libraries will be copied into the <samp>gcc</samp> install
directory. If the directory list is omitted, this option has no
effect.
</p>
</dd>
<dt><code>--with-newlib</code></dt>
<dd><p>Specifies that &lsquo;<samp>newlib</samp>&rsquo; is
being used as the target C library. This causes <code>__eprintf</code> to be
omitted from <samp>libgcc.a</samp> on the assumption that it will be provided by
&lsquo;<samp>newlib</samp>&rsquo;.
</p>
</dd>
<dt><code>--with-avrlibc</code></dt>
<dd><p>Specifies that &lsquo;<samp>AVR-Libc</samp>&rsquo; is
being used as the target C library. This causes float support
functions like <code>__addsf3</code> to be omitted from <samp>libgcc.a</samp> on
the assumption that it will be provided by <samp>libm.a</samp>. For more
technical details, cf. <a href="http://gcc.gnu.org/PR54461">PR54461</a>.
This option is only supported for the AVR target. It is not supported for
RTEMS configurations, which currently use newlib. The option is
supported since version 4.7.2 and is the default in 4.8.0 and newer.
</p>
</dd>
<dt><code>--with-nds32-lib=<var>library</var></code></dt>
<dd><p>Specifies that <var>library</var> setting is used for building <samp>libgcc.a</samp>.
Currently, the valid <var>library</var> is &lsquo;<samp>newlib</samp>&rsquo; or &lsquo;<samp>mculib</samp>&rsquo;.
This option is only supported for the NDS32 target.
</p>
</dd>
<dt><code>--with-build-time-tools=<var>dir</var></code></dt>
<dd><p>Specifies where to find the set of target tools (assembler, linker, etc.)
that will be used while building GCC itself. This option can be useful
if the directory layouts are different between the system you are building
GCC on, and the system where you will deploy it.
</p>
<p>For example, on an &lsquo;<samp>ia64-hp-hpux</samp>&rsquo; system, you may have the GNU
assembler and linker in <samp>/usr/bin</samp>, and the native tools in a
different path, and build a toolchain that expects to find the
native tools in <samp>/usr/bin</samp>.
</p>
<p>When you use this option, you should ensure that <var>dir</var> includes
<code>ar</code>, <code>as</code>, <code>ld</code>, <code>nm</code>,
<code>ranlib</code> and <code>strip</code> if necessary, and possibly
<code>objdump</code>. Otherwise, GCC may use an inconsistent set of
tools.
</p></dd>
</dl>
<a name="Overriding-configure-test-results"></a>
<h4 class="subsubheading">Overriding <code>configure</code> test results</h4>
<p>Sometimes, it might be necessary to override the result of some
<code>configure</code> test, for example in order to ease porting to a new
system or work around a bug in a test. The toplevel <code>configure</code>
script provides three variables for this:
</p>
<dl compact="compact">
<dt><code>build_configargs</code></dt>
<dd><a name="index-build_005fconfigargs"></a>
<p>The contents of this variable is passed to all build <code>configure</code>
scripts.
</p>
</dd>
<dt><code>host_configargs</code></dt>
<dd><a name="index-host_005fconfigargs"></a>
<p>The contents of this variable is passed to all host <code>configure</code>
scripts.
</p>
</dd>
<dt><code>target_configargs</code></dt>
<dd><a name="index-target_005fconfigargs"></a>
<p>The contents of this variable is passed to all target <code>configure</code>
scripts.
</p>
</dd>
</dl>
<p>In order to avoid shell and <code>make</code> quoting issues for complex
overrides, you can pass a setting for <code>CONFIG_SITE</code> and set
variables in the site file.
</p>
<a name="Objective-C-Specific-Options"></a>
<h4 class="subheading">Objective-C-Specific Options</h4>
<p>The following options apply to the build of the Objective-C runtime library.
</p>
<dl compact="compact">
<dt><code>--enable-objc-gc</code></dt>
<dd><p>Specify that an additional variant of the GNU Objective-C runtime library
is built, using an external build of the Boehm-Demers-Weiser garbage
collector (<a href="http://www.hboehm.info/gc/">http://www.hboehm.info/gc/</a>). This library needs to be
available for each multilib variant, unless configured with
<samp>--enable-objc-gc=&lsquo;<samp>auto</samp>&rsquo;</samp> in which case the build of the
additional runtime library is skipped when not available and the build
continues.
</p>
</dd>
<dt><code>--with-target-bdw-gc=<var>list</var></code></dt>
<dt><code>--with-target-bdw-gc-include=<var>list</var></code></dt>
<dt><code>--with-target-bdw-gc-lib=<var>list</var></code></dt>
<dd><p>Specify search directories for the garbage collector header files and
libraries. <var>list</var> is a comma separated list of key value pairs of the
form &lsquo;<samp><var>multilibdir</var>=<var>path</var></samp>&rsquo;, where the default multilib key
is named as &lsquo;<samp>.</samp>&rsquo; (dot), or is omitted (e.g.
&lsquo;<samp>--with-target-bdw-gc=/opt/bdw-gc,32=/opt-bdw-gc32</samp>&rsquo;).
</p>
<p>The options <samp>--with-target-bdw-gc-include</samp> and
<samp>--with-target-bdw-gc-lib</samp> must always be specified together
for each multilib variant and they take precedence over
<samp>--with-target-bdw-gc</samp>. If <samp>--with-target-bdw-gc-include</samp>
is missing values for a multilib, then the value for the default
multilib is used (e.g. &lsquo;<samp>--with-target-bdw-gc-include=/opt/bdw-gc/include</samp>&rsquo;
&lsquo;<samp>--with-target-bdw-gc-lib=/opt/bdw-gc/lib64,32=/opt-bdw-gc/lib32</samp>&rsquo;).
If none of these options are specified, the library is assumed in
default locations.
</p></dd>
</dl>
<hr />
<p>
<p><a href="./index.html">Return to the GCC Installation page</a>
</p>
</body>
</html>