configure.html revision 1.16
1<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd"> 2<html> 3<!-- Copyright (C) 1988-2022 Free Software Foundation, Inc. 4 5Permission is granted to copy, distribute and/or modify this document 6under the terms of the GNU Free Documentation License, Version 1.3 or 7any later version published by the Free Software Foundation; with no 8Invariant Sections, the Front-Cover texts being (a) (see below), and 9with the Back-Cover Texts being (b) (see below). A copy of the 10license is included in the section entitled "GNU 11Free Documentation License". 12 13(a) The FSF's Front-Cover Text is: 14 15A GNU Manual 16 17(b) The FSF's Back-Cover Text is: 18 19You have freedom to copy and modify this GNU Manual, like GNU 20 software. Copies published by the Free Software Foundation raise 21 funds for GNU development. --> 22<!-- Created by GNU Texinfo 6.5, http://www.gnu.org/software/texinfo/ --> 23<head> 24<meta http-equiv="Content-Type" content="text/html; charset=utf-8"> 25<title>Installing GCC: Configuration</title> 26 27<meta name="description" content="Installing GCC: Configuration"> 28<meta name="keywords" content="Installing GCC: Configuration"> 29<meta name="resource-type" content="document"> 30<meta name="distribution" content="global"> 31<meta name="Generator" content="makeinfo"> 32<style type="text/css"> 33<!-- 34a.summary-letter {text-decoration: none} 35blockquote.indentedblock {margin-right: 0em} 36blockquote.smallindentedblock {margin-right: 0em; font-size: smaller} 37blockquote.smallquotation {font-size: smaller} 38div.display {margin-left: 3.2em} 39div.example {margin-left: 3.2em} 40div.lisp {margin-left: 3.2em} 41div.smalldisplay {margin-left: 3.2em} 42div.smallexample {margin-left: 3.2em} 43div.smalllisp {margin-left: 3.2em} 44kbd {font-style: oblique} 45pre.display {font-family: inherit} 46pre.format {font-family: inherit} 47pre.menu-comment {font-family: serif} 48pre.menu-preformatted {font-family: serif} 49pre.smalldisplay {font-family: inherit; font-size: smaller} 50pre.smallexample {font-size: smaller} 51pre.smallformat {font-family: inherit; font-size: smaller} 52pre.smalllisp {font-size: smaller} 53span.nolinebreak {white-space: nowrap} 54span.roman {font-family: initial; font-weight: normal} 55span.sansserif {font-family: sans-serif; font-weight: normal} 56ul.no-bullet {list-style: none} 57--> 58</style> 59 60 61</head> 62 63<body lang="en"> 64<h1 class="settitle" align="center">Installing GCC: Configuration</h1> 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84<a name="index-Configuration"></a> 85<a name="index-Installing-GCC_003a-Configuration"></a> 86 87<p>Like most GNU software, GCC must be configured before it can be built. 88This document describes the recommended configuration procedure 89for both native and cross targets. 90</p> 91<p>We use <var>srcdir</var> to refer to the toplevel source directory for 92GCC; we use <var>objdir</var> to refer to the toplevel build/object directory. 93</p> 94<p>If you obtained the sources by cloning the repository, <var>srcdir</var> 95must refer to the top <samp>gcc</samp> directory, the one where the 96<samp>MAINTAINERS</samp> file can be found, and not its <samp>gcc</samp> 97subdirectory, otherwise the build will fail. 98</p> 99<p>If either <var>srcdir</var> or <var>objdir</var> is located on an automounted NFS 100file system, the shell’s built-in <code>pwd</code> command will return 101temporary pathnames. Using these can lead to various sorts of build 102problems. To avoid this issue, set the <code>PWDCMD</code> environment 103variable to an automounter-aware <code>pwd</code> command, e.g., 104<code>pawd</code> or ‘<samp>amq -w</samp>’, during the configuration and build 105phases. 106</p> 107<p>First, we <strong>highly</strong> recommend that GCC be built into a 108separate directory from the sources which does <strong>not</strong> reside 109within the source tree. This is how we generally build GCC; building 110where <var>srcdir</var> == <var>objdir</var> should still work, but doesn’t 111get extensive testing; building where <var>objdir</var> is a subdirectory 112of <var>srcdir</var> is unsupported. 113</p> 114<p>If you have previously built GCC in the same directory for a 115different target machine, do ‘<samp>make distclean</samp>’ to delete all files 116that might be invalid. One of the files this deletes is <samp>Makefile</samp>; 117if ‘<samp>make distclean</samp>’ complains that <samp>Makefile</samp> does not exist 118or issues a message like “don’t know how to make distclean” it probably 119means that the directory is already suitably clean. However, with the 120recommended method of building in a separate <var>objdir</var>, you should 121simply use a different <var>objdir</var> for each target. 122</p> 123<p>Second, when configuring a native system, either <code>cc</code> or 124<code>gcc</code> must be in your path or you must set <code>CC</code> in 125your environment before running configure. Otherwise the configuration 126scripts may fail. 127</p> 128 129<p>To configure GCC: 130</p> 131<div class="smallexample"> 132<pre class="smallexample">% mkdir <var>objdir</var> 133% cd <var>objdir</var> 134% <var>srcdir</var>/configure [<var>options</var>] [<var>target</var>] 135</pre></div> 136 137<a name="Distributor-options"></a> 138<h3 class="heading">Distributor options</h3> 139 140<p>If you will be distributing binary versions of GCC, with modifications 141to the source code, you should use the options described in this 142section to make clear that your version contains modifications. 143</p> 144<dl compact="compact"> 145<dt><code>--with-pkgversion=<var>version</var></code></dt> 146<dd><p>Specify a string that identifies your package. You may wish 147to include a build number or build date. This version string will be 148included in the output of <code>gcc --version</code>. This suffix does 149not replace the default version string, only the ‘<samp>GCC</samp>’ part. 150</p> 151<p>The default value is ‘<samp>GCC</samp>’. 152</p> 153</dd> 154<dt><code>--with-bugurl=<var>url</var></code></dt> 155<dd><p>Specify the URL that users should visit if they wish to report a bug. 156You are of course welcome to forward bugs reported to you to the FSF, 157if you determine that they are not bugs in your modifications. 158</p> 159<p>The default value refers to the FSF’s GCC bug tracker. 160</p> 161</dd> 162<dt><code>--with-documentation-root-url=<var>url</var></code></dt> 163<dd><p>Specify the URL root that contains GCC option documentation. The <var>url</var> 164should end with a <code>/</code> character. 165</p> 166<p>The default value is <a href="https://gcc.gnu.org/onlinedocs/">https://gcc.gnu.org/onlinedocs/</a>. 167</p> 168</dd> 169<dt><code>--with-changes-root-url=<var>url</var></code></dt> 170<dd><p>Specify the URL root that contains information about changes in GCC 171releases like <code>gcc-<var>version</var>/changes.html</code>. 172The <var>url</var> should end with a <code>/</code> character. 173</p> 174<p>The default value is <a href="https://gcc.gnu.org/">https://gcc.gnu.org/</a>. 175</p> 176</dd> 177</dl> 178 179<a name="Host_002c-Build-and-Target-specification"></a> 180<h3 class="heading">Host, Build and Target specification</h3> 181 182<p>Specify the host, build and target machine configurations. You do this 183when you run the <samp>configure</samp> script. 184</p> 185<p>The <em>build</em> machine is the system which you are using, the 186<em>host</em> machine is the system where you want to run the resulting 187compiler (normally the build machine), and the <em>target</em> machine is 188the system for which you want the compiler to generate code. 189</p> 190<p>If you are building a compiler to produce code for the machine it runs 191on (a native compiler), you normally do not need to specify any operands 192to <samp>configure</samp>; it will try to guess the type of machine you are on 193and use that as the build, host and target machines. So you don’t need 194to specify a configuration when building a native compiler unless 195<samp>configure</samp> cannot figure out what your configuration is or guesses 196wrong. 197</p> 198<p>In those cases, specify the build machine’s <em>configuration name</em> 199with the <samp>--host</samp> option; the host and target will default to be 200the same as the host machine. 201</p> 202<p>Here is an example: 203</p> 204<div class="smallexample"> 205<pre class="smallexample">./configure --host=x86_64-pc-linux-gnu 206</pre></div> 207 208<p>A configuration name may be canonical or it may be more or less 209abbreviated (<samp>config.sub</samp> script produces canonical versions). 210</p> 211<p>A canonical configuration name has three parts, separated by dashes. 212It looks like this: ‘<samp><var>cpu</var>-<var>company</var>-<var>system</var></samp>’. 213</p> 214<p>Here are the possible CPU types: 215</p> 216<blockquote> 217<p>aarch64, aarch64_be, alpha, alpha64, amdgcn, arc, arceb, arm, armeb, avr, bfin, 218bpf, cr16, cris, csky, epiphany, fido, fr30, frv, ft32, h8300, hppa, hppa2.0, 219hppa64, i486, i686, ia64, iq2000, lm32, loongarch64, m32c, m32r, m32rle, m68k, 220mcore, microblaze, microblazeel, mips, mips64, mips64el, mips64octeon, 221mips64orion, mips64vr, mipsel, mipsisa32, mipsisa32r2, mipsisa64, mipsisa64r2, 222mipsisa64r2el, mipsisa64sb1, mipsisa64sr71k, mipstx39, mmix, mn10300, moxie, 223msp430, nds32be, nds32le, nios2, nvptx, or1k, pdp11, powerpc, powerpc64, 224powerpc64le, powerpcle, pru, riscv32, riscv32be, riscv64, riscv64be, rl78, rx, 225s390, s390x, sh, shle, sparc, sparc64, tic6x, tilegx, tilegxbe, tilepro, v850, 226v850e, v850e1, vax, visium, x86_64, xstormy16, xtensa 227</p></blockquote> 228 229<p>Here is a list of system types: 230</p> 231<blockquote> 232<p>aix<var>version</var>, amdhsa, aout, cygwin, darwin<var>version</var>, 233eabi, eabialtivec, eabisim, eabisimaltivec, elf, elf32, 234elfbare, elfoabi, freebsd<var>version</var>, gnu, hpux, hpux<var>version</var>, 235kfreebsd-gnu, kopensolaris-gnu, linux-androideabi, linux-gnu, 236linux-gnu_altivec, linux-musl, linux-uclibc, lynxos, mingw32, mingw32crt, 237mmixware, msdosdjgpp, netbsd, netbsdelf<var>version</var>, nto-qnx, openbsd, 238rtems, solaris<var>version</var>, symbianelf, tpf, uclinux, uclinux_eabi, vms, 239vxworks, vxworksae, vxworksmils 240</p></blockquote> 241 242<a name="Options-specification"></a> 243<h3 class="heading">Options specification</h3> 244 245<p>Use <var>options</var> to override several configure time options for 246GCC. A list of supported <var>options</var> follows; ‘<samp>configure 247--help</samp>’ may list other options, but those not listed below may not 248work and should not normally be used. 249</p> 250<p>Note that each <samp>--enable</samp> option has a corresponding 251<samp>--disable</samp> option and that each <samp>--with</samp> option has a 252corresponding <samp>--without</samp> option. 253</p> 254<dl compact="compact"> 255<dt><code>--prefix=<var>dirname</var></code></dt> 256<dd><p>Specify the toplevel installation 257directory. This is the recommended way to install the tools into a directory 258other than the default. The toplevel installation directory defaults to 259<samp>/usr/local</samp>. 260</p> 261<p>We <strong>highly</strong> recommend against <var>dirname</var> being the same or a 262subdirectory of <var>objdir</var> or vice versa. If specifying a directory 263beneath a user’s home directory tree, some shells will not expand 264<var>dirname</var> correctly if it contains the ‘<samp>~</samp>’ metacharacter; use 265<code>$HOME</code> instead. 266</p> 267<p>The following standard <code>autoconf</code> options are supported. Normally you 268should not need to use these options. 269</p><dl compact="compact"> 270<dt><code>--exec-prefix=<var>dirname</var></code></dt> 271<dd><p>Specify the toplevel installation directory for architecture-dependent 272files. The default is <samp><var>prefix</var></samp>. 273</p> 274</dd> 275<dt><code>--bindir=<var>dirname</var></code></dt> 276<dd><p>Specify the installation directory for the executables called by users 277(such as <code>gcc</code> and <code>g++</code>). The default is 278<samp><var>exec-prefix</var>/bin</samp>. 279</p> 280</dd> 281<dt><code>--libdir=<var>dirname</var></code></dt> 282<dd><p>Specify the installation directory for object code libraries and 283internal data files of GCC. The default is <samp><var>exec-prefix</var>/lib</samp>. 284</p> 285</dd> 286<dt><code>--libexecdir=<var>dirname</var></code></dt> 287<dd><p>Specify the installation directory for internal executables of GCC. 288The default is <samp><var>exec-prefix</var>/libexec</samp>. 289</p> 290</dd> 291<dt><code>--with-slibdir=<var>dirname</var></code></dt> 292<dd><p>Specify the installation directory for the shared libgcc library. The 293default is <samp><var>libdir</var></samp>. 294</p> 295</dd> 296<dt><code>--datarootdir=<var>dirname</var></code></dt> 297<dd><p>Specify the root of the directory tree for read-only architecture-independent 298data files referenced by GCC. The default is <samp><var>prefix</var>/share</samp>. 299</p> 300</dd> 301<dt><code>--infodir=<var>dirname</var></code></dt> 302<dd><p>Specify the installation directory for documentation in info format. 303The default is <samp><var>datarootdir</var>/info</samp>. 304</p> 305</dd> 306<dt><code>--datadir=<var>dirname</var></code></dt> 307<dd><p>Specify the installation directory for some architecture-independent 308data files referenced by GCC. The default is <samp><var>datarootdir</var></samp>. 309</p> 310</dd> 311<dt><code>--docdir=<var>dirname</var></code></dt> 312<dd><p>Specify the installation directory for documentation files (other 313than Info) for GCC. The default is <samp><var>datarootdir</var>/doc</samp>. 314</p> 315</dd> 316<dt><code>--htmldir=<var>dirname</var></code></dt> 317<dd><p>Specify the installation directory for HTML documentation files. 318The default is <samp><var>docdir</var></samp>. 319</p> 320</dd> 321<dt><code>--pdfdir=<var>dirname</var></code></dt> 322<dd><p>Specify the installation directory for PDF documentation files. 323The default is <samp><var>docdir</var></samp>. 324</p> 325</dd> 326<dt><code>--mandir=<var>dirname</var></code></dt> 327<dd><p>Specify the installation directory for manual pages. The default is 328<samp><var>datarootdir</var>/man</samp>. (Note that the manual pages are only extracts 329from the full GCC manuals, which are provided in Texinfo format. The manpages 330are derived by an automatic conversion process from parts of the full 331manual.) 332</p> 333</dd> 334<dt><code>--with-gxx-include-dir=<var>dirname</var></code></dt> 335<dd><p>Specify 336the installation directory for G++ header files. The default depends 337on other configuration options, and differs between cross and native 338configurations. 339</p> 340</dd> 341<dt><code>--with-specs=<var>specs</var></code></dt> 342<dd><p>Specify additional command line driver SPECS. 343This can be useful if you need to turn on a non-standard feature by 344default without modifying the compiler’s source code, for instance 345<samp>--with-specs=%{!fcommon:%{!fno-common:-fno-common}}</samp>. 346See “Spec Files” in the main manual 347</p> 348</dd> 349</dl> 350 351</dd> 352<dt><code>--program-prefix=<var>prefix</var></code></dt> 353<dd><p>GCC supports some transformations of the names of its programs when 354installing them. This option prepends <var>prefix</var> to the names of 355programs to install in <var>bindir</var> (see above). For example, specifying 356<samp>--program-prefix=foo-</samp> would result in ‘<samp>gcc</samp>’ 357being installed as <samp>/usr/local/bin/foo-gcc</samp>. 358</p> 359</dd> 360<dt><code>--program-suffix=<var>suffix</var></code></dt> 361<dd><p>Appends <var>suffix</var> to the names of programs to install in <var>bindir</var> 362(see above). For example, specifying <samp>--program-suffix=-3.1</samp> 363would result in ‘<samp>gcc</samp>’ being installed as 364<samp>/usr/local/bin/gcc-3.1</samp>. 365</p> 366</dd> 367<dt><code>--program-transform-name=<var>pattern</var></code></dt> 368<dd><p>Applies the ‘<samp>sed</samp>’ script <var>pattern</var> to be applied to the names 369of programs to install in <var>bindir</var> (see above). <var>pattern</var> has to 370consist of one or more basic ‘<samp>sed</samp>’ editing commands, separated by 371semicolons. For example, if you want the ‘<samp>gcc</samp>’ program name to be 372transformed to the installed program <samp>/usr/local/bin/myowngcc</samp> and 373the ‘<samp>g++</samp>’ program name to be transformed to 374<samp>/usr/local/bin/gspecial++</samp> without changing other program names, 375you could use the pattern 376<samp>--program-transform-name='s/^gcc$/myowngcc/; s/^g++$/gspecial++/'</samp> 377to achieve this effect. 378</p> 379<p>All three options can be combined and used together, resulting in more 380complex conversion patterns. As a basic rule, <var>prefix</var> (and 381<var>suffix</var>) are prepended (appended) before further transformations 382can happen with a special transformation script <var>pattern</var>. 383</p> 384<p>As currently implemented, this option only takes effect for native 385builds; cross compiler binaries’ names are not transformed even when a 386transformation is explicitly asked for by one of these options. 387</p> 388<p>For native builds, some of the installed programs are also installed 389with the target alias in front of their name, as in 390‘<samp>i686-pc-linux-gnu-gcc</samp>’. All of the above transformations happen 391before the target alias is prepended to the name—so, specifying 392<samp>--program-prefix=foo-</samp> and <samp>program-suffix=-3.1</samp>, the 393resulting binary would be installed as 394<samp>/usr/local/bin/i686-pc-linux-gnu-foo-gcc-3.1</samp>. 395</p> 396<p>As a last shortcoming, none of the installed Ada programs are 397transformed yet, which will be fixed in some time. 398</p> 399</dd> 400<dt><code>--with-local-prefix=<var>dirname</var></code></dt> 401<dd><p>Specify the 402installation directory for local include files. The default is 403<samp>/usr/local</samp>. Specify this option if you want the compiler to 404search directory <samp><var>dirname</var>/include</samp> for locally installed 405header files <em>instead</em> of <samp>/usr/local/include</samp>. 406</p> 407<p>You should specify <samp>--with-local-prefix</samp> <strong>only</strong> if your 408site has a different convention (not <samp>/usr/local</samp>) for where to put 409site-specific files. 410</p> 411<p>The default value for <samp>--with-local-prefix</samp> is <samp>/usr/local</samp> 412regardless of the value of <samp>--prefix</samp>. Specifying 413<samp>--prefix</samp> has no effect on which directory GCC searches for 414local header files. This may seem counterintuitive, but actually it is 415logical. 416</p> 417<p>The purpose of <samp>--prefix</samp> is to specify where to <em>install 418GCC</em>. The local header files in <samp>/usr/local/include</samp>—if you put 419any in that directory—are not part of GCC. They are part of other 420programs—perhaps many others. (GCC installs its own header files in 421another directory which is based on the <samp>--prefix</samp> value.) 422</p> 423<p>Both the local-prefix include directory and the GCC-prefix include 424directory are part of GCC’s “system include” directories. Although these 425two directories are not fixed, they need to be searched in the proper 426order for the correct processing of the include_next directive. The 427local-prefix include directory is searched before the GCC-prefix 428include directory. Another characteristic of system include directories 429is that pedantic warnings are turned off for headers in these directories. 430</p> 431<p>Some autoconf macros add <samp>-I <var>directory</var></samp> options to the 432compiler command line, to ensure that directories containing installed 433packages’ headers are searched. When <var>directory</var> is one of GCC’s 434system include directories, GCC will ignore the option so that system 435directories continue to be processed in the correct order. This 436may result in a search order different from what was specified but the 437directory will still be searched. 438</p> 439<p>GCC automatically searches for ordinary libraries using 440<code>GCC_EXEC_PREFIX</code>. Thus, when the same installation prefix is 441used for both GCC and packages, GCC will automatically search for 442both headers and libraries. This provides a configuration that is 443easy to use. GCC behaves in a manner similar to that when it is 444installed as a system compiler in <samp>/usr</samp>. 445</p> 446<p>Sites that need to install multiple versions of GCC may not want to 447use the above simple configuration. It is possible to use the 448<samp>--program-prefix</samp>, <samp>--program-suffix</samp> and 449<samp>--program-transform-name</samp> options to install multiple versions 450into a single directory, but it may be simpler to use different prefixes 451and the <samp>--with-local-prefix</samp> option to specify the location of the 452site-specific files for each version. It will then be necessary for 453users to specify explicitly the location of local site libraries 454(e.g., with <code>LIBRARY_PATH</code>). 455</p> 456<p>The same value can be used for both <samp>--with-local-prefix</samp> and 457<samp>--prefix</samp> provided it is not <samp>/usr</samp>. This can be used 458to avoid the default search of <samp>/usr/local/include</samp>. 459</p> 460<p><strong>Do not</strong> specify <samp>/usr</samp> as the <samp>--with-local-prefix</samp>! 461The directory you use for <samp>--with-local-prefix</samp> <strong>must not</strong> 462contain any of the system’s standard header files. If it did contain 463them, certain programs would be miscompiled (including GNU Emacs, on 464certain targets), because this would override and nullify the header 465file corrections made by the <code>fixincludes</code> script. 466</p> 467<p>Indications are that people who use this option use it based on mistaken 468ideas of what it is for. People use it as if it specified where to 469install part of GCC. Perhaps they make this assumption because 470installing GCC creates the directory. 471</p> 472</dd> 473<dt><code>--with-gcc-major-version-only</code></dt> 474<dd><p>Specifies that GCC should use only the major number rather than 475<var>major</var>.<var>minor</var>.<var>patchlevel</var> in filesystem paths. 476</p> 477</dd> 478<dt><code>--with-native-system-header-dir=<var>dirname</var></code></dt> 479<dd><p>Specifies that <var>dirname</var> is the directory that contains native system 480header files, rather than <samp>/usr/include</samp>. This option is most useful 481if you are creating a compiler that should be isolated from the system 482as much as possible. It is most commonly used with the 483<samp>--with-sysroot</samp> option and will cause GCC to search 484<var>dirname</var> inside the system root specified by that option. 485</p> 486</dd> 487<dt><code>--enable-shared[=<var>package</var>[,…]]</code></dt> 488<dd><p>Build shared versions of libraries, if shared libraries are supported on 489the target platform. Unlike GCC 2.95.x and earlier, shared libraries 490are enabled by default on all platforms that support shared libraries. 491</p> 492<p>If a list of packages is given as an argument, build shared libraries 493only for the listed packages. For other packages, only static libraries 494will be built. Package names currently recognized in the GCC tree are 495‘<samp>libgcc</samp>’ (also known as ‘<samp>gcc</samp>’), ‘<samp>libstdc++</samp>’ (not 496‘<samp>libstdc++-v3</samp>’), ‘<samp>libffi</samp>’, ‘<samp>zlib</samp>’, ‘<samp>boehm-gc</samp>’, 497‘<samp>ada</samp>’, ‘<samp>libada</samp>’, ‘<samp>libgo</samp>’, ‘<samp>libobjc</samp>’, and ‘<samp>libphobos</samp>’. 498Note ‘<samp>libiberty</samp>’ does not support shared libraries at all. 499</p> 500<p>Use <samp>--disable-shared</samp> to build only static libraries. Note that 501<samp>--disable-shared</samp> does not accept a list of package names as 502argument, only <samp>--enable-shared</samp> does. 503</p> 504<p>Contrast with <samp>--enable-host-shared</samp>, which affects <em>host</em> 505code. 506</p> 507</dd> 508<dt><code>--enable-host-shared</code></dt> 509<dd><p>Specify that the <em>host</em> code should be built into position-independent 510machine code (with -fPIC), allowing it to be used within shared libraries, 511but yielding a slightly slower compiler. 512</p> 513<p>This option is required when building the libgccjit.so library. 514</p> 515<p>Contrast with <samp>--enable-shared</samp>, which affects <em>target</em> 516libraries. 517</p> 518</dd> 519<dt><code><a name="with-gnu-as"></a>--with-gnu-as</code></dt> 520<dd><p>Specify that the compiler should assume that the 521assembler it finds is the GNU assembler. However, this does not modify 522the rules to find an assembler and will result in confusion if the 523assembler found is not actually the GNU assembler. (Confusion may also 524result if the compiler finds the GNU assembler but has not been 525configured with <samp>--with-gnu-as</samp>.) If you have more than one 526assembler installed on your system, you may want to use this option in 527connection with <samp>--with-as=<var>pathname</var></samp> or 528<samp>--with-build-time-tools=<var>pathname</var></samp>. 529</p> 530<p>The following systems are the only ones where it makes a difference 531whether you use the GNU assembler. On any other system, 532<samp>--with-gnu-as</samp> has no effect. 533</p> 534<ul> 535<li> ‘<samp>hppa1.0-<var>any</var>-<var>any</var></samp>’ 536</li><li> ‘<samp>hppa1.1-<var>any</var>-<var>any</var></samp>’ 537</li><li> ‘<samp>sparc-sun-solaris2.<var>any</var></samp>’ 538</li><li> ‘<samp>sparc64-<var>any</var>-solaris2.<var>any</var></samp>’ 539</li></ul> 540 541</dd> 542<dt><code><a name="with-as"></a>--with-as=<var>pathname</var></code></dt> 543<dd><p>Specify that the compiler should use the assembler pointed to by 544<var>pathname</var>, rather than the one found by the standard rules to find 545an assembler, which are: 546</p><ul> 547<li> Unless GCC is being built with a cross compiler, check the 548<samp><var>libexec</var>/gcc/<var>target</var>/<var>version</var></samp> directory. 549<var>libexec</var> defaults to <samp><var>exec-prefix</var>/libexec</samp>; 550<var>exec-prefix</var> defaults to <var>prefix</var>, which 551defaults to <samp>/usr/local</samp> unless overridden by the 552<samp>--prefix=<var>pathname</var></samp> switch described above. <var>target</var> 553is the target system triple, such as ‘<samp>sparc-sun-solaris2.7</samp>’, and 554<var>version</var> denotes the GCC version, such as 3.0. 555 556</li><li> If the target system is the same that you are building on, check 557operating system specific directories (e.g. <samp>/usr/ccs/bin</samp> on 558Solaris 2). 559 560</li><li> Check in the <code>PATH</code> for a tool whose name is prefixed by the 561target system triple. 562 563</li><li> Check in the <code>PATH</code> for a tool whose name is not prefixed by the 564target system triple, if the host and target system triple are 565the same (in other words, we use a host tool if it can be used for 566the target as well). 567</li></ul> 568 569<p>You may want to use <samp>--with-as</samp> if no assembler 570is installed in the directories listed above, or if you have multiple 571assemblers installed and want to choose one that is not found by the 572above rules. 573</p> 574</dd> 575<dt><code><a name="with-gnu-ld"></a>--with-gnu-ld</code></dt> 576<dd><p>Same as <a href="#with-gnu-as"><samp>--with-gnu-as</samp></a> 577but for the linker. 578</p> 579</dd> 580<dt><code>--with-ld=<var>pathname</var></code></dt> 581<dd><p>Same as <a href="#with-as"><samp>--with-as</samp></a> 582but for the linker. 583</p> 584</dd> 585<dt><code>--with-dsymutil=<var>pathname</var></code></dt> 586<dd><p>Same as <a href="#with-as"><samp>--with-as</samp></a> 587but for the debug linker (only used on Darwin platforms so far). 588</p> 589</dd> 590<dt><code>--with-tls=<var>dialect</var></code></dt> 591<dd><p>Specify the default TLS dialect, for systems were there is a choice. 592For ARM targets, possible values for <var>dialect</var> are <code>gnu</code> or 593<code>gnu2</code>, which select between the original GNU dialect and the GNU TLS 594descriptor-based dialect. 595</p> 596</dd> 597<dt><code>--enable-multiarch</code></dt> 598<dd><p>Specify whether to enable or disable multiarch support. The default is 599to check for glibc start files in a multiarch location, and enable it 600if the files are found. The auto detection is enabled for native builds, 601and for cross builds configured with <samp>--with-sysroot</samp>, and without 602<samp>--with-native-system-header-dir</samp>. 603More documentation about multiarch can be found at 604<a href="https://wiki.debian.org/Multiarch">https://wiki.debian.org/Multiarch</a>. 605</p> 606</dd> 607<dt><code>--enable-sjlj-exceptions</code></dt> 608<dd><p>Force use of the <code>setjmp</code>/<code>longjmp</code>-based scheme for exceptions. 609‘<samp>configure</samp>’ ordinarily picks the correct value based on the platform. 610Only use this option if you are sure you need a different setting. 611</p> 612</dd> 613<dt><code>--enable-vtable-verify</code></dt> 614<dd><p>Specify whether to enable or disable the vtable verification feature. 615Enabling this feature causes libstdc++ to be built with its virtual calls 616in verifiable mode. This means that, when linked with libvtv, every 617virtual call in libstdc++ will verify the vtable pointer through which the 618call will be made before actually making the call. If not linked with libvtv, 619the verifier will call stub functions (in libstdc++ itself) and do nothing. 620If vtable verification is disabled, then libstdc++ is not built with its 621virtual calls in verifiable mode at all. However the libvtv library will 622still be built (see <samp>--disable-libvtv</samp> to turn off building libvtv). 623<samp>--disable-vtable-verify</samp> is the default. 624</p> 625</dd> 626<dt><code>--disable-gcov</code></dt> 627<dd><p>Specify that the run-time library used for coverage analysis 628and associated host tools should not be built. 629</p> 630</dd> 631<dt><code>--disable-multilib</code></dt> 632<dd><p>Specify that multiple target 633libraries to support different target variants, calling 634conventions, etc. should not be built. The default is to build a 635predefined set of them. 636</p> 637<p>Some targets provide finer-grained control over which multilibs are built 638(e.g., <samp>--disable-softfloat</samp>): 639</p><dl compact="compact"> 640<dt><code>arm-*-*</code></dt> 641<dd><p>fpu, 26bit, underscore, interwork, biendian, nofmult. 642</p> 643</dd> 644<dt><code>m68*-*-*</code></dt> 645<dd><p>softfloat, m68881, m68000, m68020. 646</p> 647</dd> 648<dt><code>mips*-*-*</code></dt> 649<dd><p>single-float, biendian, softfloat. 650</p> 651</dd> 652<dt><code>msp430-*-*</code></dt> 653<dd><p>no-exceptions 654</p> 655</dd> 656<dt><code>powerpc*-*-*, rs6000*-*-*</code></dt> 657<dd><p>aix64, pthread, softfloat, powercpu, powerpccpu, powerpcos, biendian, 658sysv, aix. 659</p> 660</dd> 661</dl> 662 663</dd> 664<dt><code>--with-multilib-list=<var>list</var></code></dt> 665<dt><code>--without-multilib-list</code></dt> 666<dd><p>Specify what multilibs to build. <var>list</var> is a comma separated list of 667values, possibly consisting of a single value. Currently only implemented 668for aarch64*-*-*, arm*-*-*, loongarch64-*-*, riscv*-*-*, sh*-*-* and 669x86-64-*-linux*. The accepted values and meaning for each target is given 670below. 671</p> 672<dl compact="compact"> 673<dt><code>aarch64*-*-*</code></dt> 674<dd><p><var>list</var> is a comma separated list of <code>ilp32</code>, and <code>lp64</code> 675to enable ILP32 and LP64 run-time libraries, respectively. If 676<var>list</var> is empty, then there will be no multilibs and only the 677default run-time library will be built. If <var>list</var> is 678<code>default</code> or –with-multilib-list= is not specified, then the 679default set of libraries is selected based on the value of 680<samp>--target</samp>. 681</p> 682</dd> 683<dt><code>arm*-*-*</code></dt> 684<dd><p><var>list</var> is a comma separated list of <code>aprofile</code> and 685<code>rmprofile</code> to build multilibs for A or R and M architecture 686profiles respectively. Note that, due to some limitation of the current 687multilib framework, using the combined <code>aprofile,rmprofile</code> 688multilibs selects in some cases a less optimal multilib than when using 689the multilib profile for the architecture targetted. The special value 690<code>default</code> is also accepted and is equivalent to omitting the 691option, i.e., only the default run-time library will be enabled. 692</p> 693<p><var>list</var> may instead contain <code>@name</code>, to use the multilib 694configuration Makefile fragment <samp>name</samp> in <samp>gcc/config/arm</samp> in 695the source tree (it is part of the corresponding sources, after all). 696It is recommended, but not required, that files used for this purpose to 697be named starting with <samp>t-ml-</samp>, to make their intended purpose 698self-evident, in line with GCC conventions. Such files enable custom, 699user-chosen multilib lists to be configured. Whether multiple such 700files can be used together depends on the contents of the supplied 701files. See <samp>gcc/config/arm/t-multilib</samp> and its supplementary 702<samp>gcc/config/arm/t-*profile</samp> files for an example of what such 703Makefile fragments might look like for this version of GCC. The macros 704expected to be defined in these fragments are not stable across GCC 705releases, so make sure they define the <code>MULTILIB</code>-related macros 706expected by the version of GCC you are building. 707See “Target Makefile Fragments” in the internals manual. 708</p> 709<p>The table below gives the combination of ISAs, architectures, FPUs and 710floating-point ABIs for which multilibs are built for each predefined 711profile. The union of these options is considered when specifying both 712<code>aprofile</code> and <code>rmprofile</code>. 713</p> 714<table> 715<tr><td width="15%">Option</td><td width="28%">aprofile</td><td width="30%">rmprofile</td></tr> 716<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> 717<tr><td width="15%">Architectures<br><br><br><br><br><br></td><td width="28%">default architecture<br> 718<code>-march=armv7-a</code><br> 719<code>-march=armv7ve</code><br> 720<code>-march=armv8-a</code><br><br><br></td><td width="30%">default architecture<br> 721<code>-march=armv6s-m</code><br> 722<code>-march=armv7-m</code><br> 723<code>-march=armv7e-m</code><br> 724<code>-march=armv8-m.base</code><br> 725<code>-march=armv8-m.main</code><br> 726<code>-march=armv7</code></td></tr> 727<tr><td width="15%">FPUs<br><br><br><br><br></td><td width="28%">none<br> 728<code>-mfpu=vfpv3-d16</code><br> 729<code>-mfpu=neon</code><br> 730<code>-mfpu=vfpv4-d16</code><br> 731<code>-mfpu=neon-vfpv4</code><br> 732<code>-mfpu=neon-fp-armv8</code></td><td width="30%">none<br> 733<code>-mfpu=vfpv3-d16</code><br> 734<code>-mfpu=fpv4-sp-d16</code><br> 735<code>-mfpu=fpv5-sp-d16</code><br> 736<code>-mfpu=fpv5-d16</code><br></td></tr> 737<tr><td width="15%">floating-point ABIs<br><br></td><td width="28%"><code>-mfloat-abi=soft</code><br> 738<code>-mfloat-abi=softfp</code><br> 739<code>-mfloat-abi=hard</code></td><td width="30%"><code>-mfloat-abi=soft</code><br> 740<code>-mfloat-abi=softfp</code><br> 741<code>-mfloat-abi=hard</code></td></tr> 742</table> 743 744</dd> 745<dt><code>loongarch*-*-*</code></dt> 746<dd><p><var>list</var> is a comma-separated list of the following ABI identifiers: 747<code>lp64d[/base]</code> <code>lp64f[/base]</code> <code>lp64d[/base]</code>, where the 748<code>/base</code> suffix may be omitted, to enable their respective run-time 749libraries. If <var>list</var> is empty or <code>default</code>, 750or if <samp>--with-multilib-list</samp> is not specified, then the default ABI 751as specified by <samp>--with-abi</samp> or implied by <samp>--target</samp> is selected. 752</p> 753</dd> 754<dt><code>riscv*-*-*</code></dt> 755<dd><p><var>list</var> is a single ABI name. The target architecture must be either 756<code>rv32gc</code> or <code>rv64gc</code>. This will build a single multilib for the 757specified architecture and ABI pair. If <code>--with-multilib-list</code> is not 758given, then a default set of multilibs is selected based on the value of 759<samp>--target</samp>. This is usually a large set of multilibs. 760</p> 761</dd> 762<dt><code>sh*-*-*</code></dt> 763<dd><p><var>list</var> is a comma separated list of CPU names. These must be of the 764form <code>sh*</code> or <code>m*</code> (in which case they match the compiler option 765for that processor). The list should not contain any endian options - 766these are handled by <samp>--with-endian</samp>. 767</p> 768<p>If <var>list</var> is empty, then there will be no multilibs for extra 769processors. The multilib for the secondary endian remains enabled. 770</p> 771<p>As a special case, if an entry in the list starts with a <code>!</code> 772(exclamation point), then it is added to the list of excluded multilibs. 773Entries of this sort should be compatible with ‘<samp>MULTILIB_EXCLUDES</samp>’ 774(once the leading <code>!</code> has been stripped). 775</p> 776<p>If <samp>--with-multilib-list</samp> is not given, then a default set of 777multilibs is selected based on the value of <samp>--target</samp>. This is 778usually the complete set of libraries, but some targets imply a more 779specialized subset. 780</p> 781<p>Example 1: to configure a compiler for SH4A only, but supporting both 782endians, with little endian being the default: 783</p><div class="smallexample"> 784<pre class="smallexample">--with-cpu=sh4a --with-endian=little,big --with-multilib-list= 785</pre></div> 786 787<p>Example 2: to configure a compiler for both SH4A and SH4AL-DSP, but with 788only little endian SH4AL: 789</p><div class="smallexample"> 790<pre class="smallexample">--with-cpu=sh4a --with-endian=little,big \ 791--with-multilib-list=sh4al,!mb/m4al 792</pre></div> 793 794</dd> 795<dt><code>x86-64-*-linux*</code></dt> 796<dd><p><var>list</var> is a comma separated list of <code>m32</code>, <code>m64</code> and 797<code>mx32</code> to enable 32-bit, 64-bit and x32 run-time libraries, 798respectively. If <var>list</var> is empty, then there will be no multilibs 799and only the default run-time library will be enabled. 800</p> 801<p>If <samp>--with-multilib-list</samp> is not given, then only 32-bit and 80264-bit run-time libraries will be enabled. 803</p></dd> 804</dl> 805 806</dd> 807<dt><code>--with-multilib-generator=<var>config</var></code></dt> 808<dd><p>Specify what multilibs to build. <var>config</var> is a semicolon separated list of 809values, possibly consisting of a single value. Currently only implemented 810for riscv*-*-elf*. The accepted values and meanings are given below. 811</p> 812 813<p>Every config is constructed with four components: architecture string, ABI, 814reuse rule with architecture string and reuse rule with sub-extension. 815</p> 816<p>Example 1: Add multi-lib suppport for rv32i with ilp32. 817</p><div class="smallexample"> 818<pre class="smallexample">rv32i-ilp32-- 819</pre></div> 820 821<p>Example 2: Add multi-lib suppport for rv32i with ilp32 and rv32imafd with ilp32. 822</p><div class="smallexample"> 823<pre class="smallexample">rv32i-ilp32--;rv32imafd-ilp32-- 824</pre></div> 825 826<p>Example 3: Add multi-lib suppport for rv32i with ilp32; rv32im with ilp32 and 827rv32ic with ilp32 will reuse this multi-lib set. 828</p><div class="smallexample"> 829<pre class="smallexample">rv32i-ilp32-rv32im-c 830</pre></div> 831 832<p>Example 4: Add multi-lib suppport for rv64ima with lp64; rv64imaf with lp64, 833rv64imac with lp64 and rv64imafc with lp64 will reuse this multi-lib set. 834</p><div class="smallexample"> 835<pre class="smallexample">rv64ima-lp64--f,c,fc 836</pre></div> 837 838<p><samp>--with-multilib-generator</samp> have an optional configuration argument 839<samp>--cmodel=val</samp> for code model, this option will expand with other 840config options, <var>val</var> is a comma separated list of possible code model, 841currently we support medlow and medany. 842</p> 843<p>Example 5: Add multi-lib suppport for rv64ima with lp64; rv64ima with lp64 and 844medlow code model 845</p><div class="smallexample"> 846<pre class="smallexample">rv64ima-lp64--;--cmodel=medlow 847</pre></div> 848 849<p>Example 6: Add multi-lib suppport for rv64ima with lp64; rv64ima with lp64 and 850medlow code model; rv64ima with lp64 and medany code model 851</p><div class="smallexample"> 852<pre class="smallexample">rv64ima-lp64--;--cmodel=medlow,medany 853</pre></div> 854 855</dd> 856<dt><code>--with-endian=<var>endians</var></code></dt> 857<dd><p>Specify what endians to use. 858Currently only implemented for sh*-*-*. 859</p> 860<p><var>endians</var> may be one of the following: 861</p><dl compact="compact"> 862<dt><code>big</code></dt> 863<dd><p>Use big endian exclusively. 864</p></dd> 865<dt><code>little</code></dt> 866<dd><p>Use little endian exclusively. 867</p></dd> 868<dt><code>big,little</code></dt> 869<dd><p>Use big endian by default. Provide a multilib for little endian. 870</p></dd> 871<dt><code>little,big</code></dt> 872<dd><p>Use little endian by default. Provide a multilib for big endian. 873</p></dd> 874</dl> 875 876</dd> 877<dt><code>--enable-threads</code></dt> 878<dd><p>Specify that the target 879supports threads. This affects the Objective-C compiler and runtime 880library, and exception handling for other languages like C++. 881On some systems, this is the default. 882</p> 883<p>In general, the best (and, in many cases, the only known) threading 884model available will be configured for use. Beware that on some 885systems, GCC has not been taught what threading models are generally 886available for the system. In this case, <samp>--enable-threads</samp> is an 887alias for <samp>--enable-threads=single</samp>. 888</p> 889</dd> 890<dt><code>--disable-threads</code></dt> 891<dd><p>Specify that threading support should be disabled for the system. 892This is an alias for <samp>--enable-threads=single</samp>. 893</p> 894</dd> 895<dt><code>--enable-threads=<var>lib</var></code></dt> 896<dd><p>Specify that 897<var>lib</var> is the thread support library. This affects the Objective-C 898compiler and runtime library, and exception handling for other languages 899like C++. The possibilities for <var>lib</var> are: 900</p> 901<dl compact="compact"> 902<dt><code>aix</code></dt> 903<dd><p>AIX thread support. 904</p></dd> 905<dt><code>dce</code></dt> 906<dd><p>DCE thread support. 907</p></dd> 908<dt><code>lynx</code></dt> 909<dd><p>LynxOS thread support. 910</p></dd> 911<dt><code>mipssde</code></dt> 912<dd><p>MIPS SDE thread support. 913</p></dd> 914<dt><code>no</code></dt> 915<dd><p>This is an alias for ‘<samp>single</samp>’. 916</p></dd> 917<dt><code>posix</code></dt> 918<dd><p>Generic POSIX/Unix98 thread support. 919</p></dd> 920<dt><code>rtems</code></dt> 921<dd><p>RTEMS thread support. 922</p></dd> 923<dt><code>single</code></dt> 924<dd><p>Disable thread support, should work for all platforms. 925</p></dd> 926<dt><code>tpf</code></dt> 927<dd><p>TPF thread support. 928</p></dd> 929<dt><code>vxworks</code></dt> 930<dd><p>VxWorks thread support. 931</p></dd> 932<dt><code>win32</code></dt> 933<dd><p>Microsoft Win32 API thread support. 934</p></dd> 935</dl> 936 937</dd> 938<dt><code>--enable-tls</code></dt> 939<dd><p>Specify that the target supports TLS (Thread Local Storage). Usually 940configure can correctly determine if TLS is supported. In cases where 941it guesses incorrectly, TLS can be explicitly enabled or disabled with 942<samp>--enable-tls</samp> or <samp>--disable-tls</samp>. This can happen if 943the assembler supports TLS but the C library does not, or if the 944assumptions made by the configure test are incorrect. 945</p> 946</dd> 947<dt><code>--disable-tls</code></dt> 948<dd><p>Specify that the target does not support TLS. 949This is an alias for <samp>--enable-tls=no</samp>. 950</p> 951</dd> 952<dt><code>--disable-tm-clone-registry</code></dt> 953<dd><p>Disable TM clone registry in libgcc. It is enabled in libgcc by default. 954This option helps to reduce code size for embedded targets which do 955not use transactional memory. 956</p> 957</dd> 958<dt><code>--with-cpu=<var>cpu</var></code></dt> 959<dt><code>--with-cpu-32=<var>cpu</var></code></dt> 960<dt><code>--with-cpu-64=<var>cpu</var></code></dt> 961<dd><p>Specify which cpu variant the compiler should generate code for by default. 962<var>cpu</var> will be used as the default value of the <samp>-mcpu=</samp> switch. 963This option is only supported on some targets, including ARC, ARM, i386, M68k, 964PowerPC, and SPARC. It is mandatory for ARC. The <samp>--with-cpu-32</samp> and 965<samp>--with-cpu-64</samp> options specify separate default CPUs for 96632-bit and 64-bit modes; these options are only supported for aarch64, i386, 967x86-64, PowerPC, and SPARC. 968</p> 969</dd> 970<dt><code>--with-schedule=<var>cpu</var></code></dt> 971<dt><code>--with-arch=<var>cpu</var></code></dt> 972<dt><code>--with-arch-32=<var>cpu</var></code></dt> 973<dt><code>--with-arch-64=<var>cpu</var></code></dt> 974<dt><code>--with-tune=<var>cpu</var></code></dt> 975<dt><code>--with-tune-32=<var>cpu</var></code></dt> 976<dt><code>--with-tune-64=<var>cpu</var></code></dt> 977<dt><code>--with-abi=<var>abi</var></code></dt> 978<dt><code>--with-fpu=<var>type</var></code></dt> 979<dt><code>--with-float=<var>type</var></code></dt> 980<dd><p>These configure options provide default values for the <samp>-mschedule=</samp>, 981<samp>-march=</samp>, <samp>-mtune=</samp>, <samp>-mabi=</samp>, and <samp>-mfpu=</samp> 982options and for <samp>-mhard-float</samp> or <samp>-msoft-float</samp>. As with 983<samp>--with-cpu</samp>, which switches will be accepted and acceptable values 984of the arguments depend on the target. 985</p> 986</dd> 987<dt><code>--with-mode=<var>mode</var></code></dt> 988<dd><p>Specify if the compiler should default to <samp>-marm</samp> or <samp>-mthumb</samp>. 989This option is only supported on ARM targets. 990</p> 991</dd> 992<dt><code>--with-stack-offset=<var>num</var></code></dt> 993<dd><p>This option sets the default for the -mstack-offset=<var>num</var> option, 994and will thus generally also control the setting of this option for 995libraries. This option is only supported on Epiphany targets. 996</p> 997</dd> 998<dt><code>--with-fpmath=<var>isa</var></code></dt> 999<dd><p>This options sets <samp>-mfpmath=sse</samp> by default and specifies the default 1000ISA for floating-point arithmetics. You can select either ‘<samp>sse</samp>’ which 1001enables <samp>-msse2</samp> or ‘<samp>avx</samp>’ which enables <samp>-mavx</samp> by default. 1002This option is only supported on i386 and x86-64 targets. 1003</p> 1004</dd> 1005<dt><code>--with-fp-32=<var>mode</var></code></dt> 1006<dd><p>On MIPS targets, set the default value for the <samp>-mfp</samp> option when using 1007the o32 ABI. The possibilities for <var>mode</var> are: 1008</p><dl compact="compact"> 1009<dt><code>32</code></dt> 1010<dd><p>Use the o32 FP32 ABI extension, as with the <samp>-mfp32</samp> command-line 1011option. 1012</p></dd> 1013<dt><code>xx</code></dt> 1014<dd><p>Use the o32 FPXX ABI extension, as with the <samp>-mfpxx</samp> command-line 1015option. 1016</p></dd> 1017<dt><code>64</code></dt> 1018<dd><p>Use the o32 FP64 ABI extension, as with the <samp>-mfp64</samp> command-line 1019option. 1020</p></dd> 1021</dl> 1022<p>In the absence of this configuration option the default is to use the o32 1023FP32 ABI extension. 1024</p> 1025</dd> 1026<dt><code>--with-odd-spreg-32</code></dt> 1027<dd><p>On MIPS targets, set the <samp>-modd-spreg</samp> option by default when using 1028the o32 ABI. 1029</p> 1030</dd> 1031<dt><code>--without-odd-spreg-32</code></dt> 1032<dd><p>On MIPS targets, set the <samp>-mno-odd-spreg</samp> option by default when using 1033the o32 ABI. This is normally used in conjunction with 1034<samp>--with-fp-32=64</samp> in order to target the o32 FP64A ABI extension. 1035</p> 1036</dd> 1037<dt><code>--with-nan=<var>encoding</var></code></dt> 1038<dd><p>On MIPS targets, set the default encoding convention to use for the 1039special not-a-number (NaN) IEEE 754 floating-point data. The 1040possibilities for <var>encoding</var> are: 1041</p><dl compact="compact"> 1042<dt><code>legacy</code></dt> 1043<dd><p>Use the legacy encoding, as with the <samp>-mnan=legacy</samp> command-line 1044option. 1045</p></dd> 1046<dt><code>2008</code></dt> 1047<dd><p>Use the 754-2008 encoding, as with the <samp>-mnan=2008</samp> command-line 1048option. 1049</p></dd> 1050</dl> 1051<p>To use this configuration option you must have an assembler version 1052installed that supports the <samp>-mnan=</samp> command-line option too. 1053In the absence of this configuration option the default convention is 1054the legacy encoding, as when neither of the <samp>-mnan=2008</samp> and 1055<samp>-mnan=legacy</samp> command-line options has been used. 1056</p> 1057</dd> 1058<dt><code>--with-divide=<var>type</var></code></dt> 1059<dd><p>Specify how the compiler should generate code for checking for 1060division by zero. This option is only supported on the MIPS target. 1061The possibilities for <var>type</var> are: 1062</p><dl compact="compact"> 1063<dt><code>traps</code></dt> 1064<dd><p>Division by zero checks use conditional traps (this is the default on 1065systems that support conditional traps). 1066</p></dd> 1067<dt><code>breaks</code></dt> 1068<dd><p>Division by zero checks use the break instruction. 1069</p></dd> 1070</dl> 1071 1072 1073</dd> 1074<dt><code>--with-llsc</code></dt> 1075<dd><p>On MIPS targets, make <samp>-mllsc</samp> the default when no 1076<samp>-mno-llsc</samp> option is passed. This is the default for 1077Linux-based targets, as the kernel will emulate them if the ISA does 1078not provide them. 1079</p> 1080</dd> 1081<dt><code>--without-llsc</code></dt> 1082<dd><p>On MIPS targets, make <samp>-mno-llsc</samp> the default when no 1083<samp>-mllsc</samp> option is passed. 1084</p> 1085</dd> 1086<dt><code>--with-synci</code></dt> 1087<dd><p>On MIPS targets, make <samp>-msynci</samp> the default when no 1088<samp>-mno-synci</samp> option is passed. 1089</p> 1090</dd> 1091<dt><code>--without-synci</code></dt> 1092<dd><p>On MIPS targets, make <samp>-mno-synci</samp> the default when no 1093<samp>-msynci</samp> option is passed. This is the default. 1094</p> 1095</dd> 1096<dt><code>--with-lxc1-sxc1</code></dt> 1097<dd><p>On MIPS targets, make <samp>-mlxc1-sxc1</samp> the default when no 1098<samp>-mno-lxc1-sxc1</samp> option is passed. This is the default. 1099</p> 1100</dd> 1101<dt><code>--without-lxc1-sxc1</code></dt> 1102<dd><p>On MIPS targets, make <samp>-mno-lxc1-sxc1</samp> the default when no 1103<samp>-mlxc1-sxc1</samp> option is passed. The indexed load/store 1104instructions are not directly a problem but can lead to unexpected 1105behaviour when deployed in an application intended for a 32-bit address 1106space but run on a 64-bit processor. The issue is seen because all 1107known MIPS 64-bit Linux kernels execute o32 and n32 applications 1108with 64-bit addressing enabled which affects the overflow behaviour 1109of the indexed addressing mode. GCC will assume that ordinary 111032-bit arithmetic overflow behaviour is the same whether performed 1111as an <code>addu</code> instruction or as part of the address calculation 1112in <code>lwxc1</code> type instructions. This assumption holds true in a 1113pure 32-bit environment and can hold true in a 64-bit environment if 1114the address space is accurately set to be 32-bit for o32 and n32. 1115</p> 1116</dd> 1117<dt><code>--with-madd4</code></dt> 1118<dd><p>On MIPS targets, make <samp>-mmadd4</samp> the default when no 1119<samp>-mno-madd4</samp> option is passed. This is the default. 1120</p> 1121</dd> 1122<dt><code>--without-madd4</code></dt> 1123<dd><p>On MIPS targets, make <samp>-mno-madd4</samp> the default when no 1124<samp>-mmadd4</samp> option is passed. The <code>madd4</code> instruction 1125family can be problematic when targeting a combination of cores that 1126implement these instructions differently. There are two known cores 1127that implement these as fused operations instead of unfused (where 1128unfused is normally expected). Disabling these instructions is the 1129only way to ensure compatible code is generated; this will incur 1130a performance penalty. 1131</p> 1132</dd> 1133<dt><code>--with-mips-plt</code></dt> 1134<dd><p>On MIPS targets, make use of copy relocations and PLTs. 1135These features are extensions to the traditional 1136SVR4-based MIPS ABIs and require support from GNU binutils 1137and the runtime C library. 1138</p> 1139</dd> 1140<dt><code>--with-stack-clash-protection-guard-size=<var>size</var></code></dt> 1141<dd><p>On certain targets this option sets the default stack clash protection guard 1142size as a power of two in bytes. On AArch64 <var>size</var> is required to be either 114312 (4KB) or 16 (64KB). 1144</p> 1145</dd> 1146<dt><code>--with-isa-spec=<var>ISA-spec-string</var></code></dt> 1147<dd><p>On RISC-V targets specify the default version of the RISC-V Unprivileged 1148(formerly User-Level) ISA specification to produce code conforming to. 1149The possibilities for <var>ISA-spec-string</var> are: 1150</p><dl compact="compact"> 1151<dt><code>2.2</code></dt> 1152<dd><p>Produce code conforming to version 2.2. 1153</p></dd> 1154<dt><code>20190608</code></dt> 1155<dd><p>Produce code conforming to version 20190608. 1156</p></dd> 1157<dt><code>20191213</code></dt> 1158<dd><p>Produce code conforming to version 20191213. 1159</p></dd> 1160</dl> 1161<p>In the absence of this configuration option the default version is 20191213. 1162</p> 1163</dd> 1164<dt><code>--enable-__cxa_atexit</code></dt> 1165<dd><p>Define if you want to use __cxa_atexit, rather than atexit, to 1166register C++ destructors for local statics and global objects. 1167This is essential for fully standards-compliant handling of 1168destructors, but requires __cxa_atexit in libc. This option is currently 1169only available on systems with GNU libc. When enabled, this will cause 1170<samp>-fuse-cxa-atexit</samp> to be passed by default. 1171</p> 1172</dd> 1173<dt><code>--enable-gnu-indirect-function</code></dt> 1174<dd><p>Define if you want to enable the <code>ifunc</code> attribute. This option is 1175currently only available on systems with GNU libc on certain targets. 1176</p> 1177</dd> 1178<dt><code>--enable-target-optspace</code></dt> 1179<dd><p>Specify that target 1180libraries should be optimized for code space instead of code speed. 1181This is the default for the m32r platform. 1182</p> 1183</dd> 1184<dt><code>--with-cpp-install-dir=<var>dirname</var></code></dt> 1185<dd><p>Specify that the user visible <code>cpp</code> program should be installed 1186in <samp><var>prefix</var>/<var>dirname</var>/cpp</samp>, in addition to <var>bindir</var>. 1187</p> 1188</dd> 1189<dt><code>--enable-comdat</code></dt> 1190<dd><p>Enable COMDAT group support. This is primarily used to override the 1191automatically detected value. 1192</p> 1193</dd> 1194<dt><code>--enable-initfini-array</code></dt> 1195<dd><p>Force the use of sections <code>.init_array</code> and <code>.fini_array</code> 1196(instead of <code>.init</code> and <code>.fini</code>) for constructors and 1197destructors. Option <samp>--disable-initfini-array</samp> has the 1198opposite effect. If neither option is specified, the configure script 1199will try to guess whether the <code>.init_array</code> and 1200<code>.fini_array</code> sections are supported and, if they are, use them. 1201</p> 1202</dd> 1203<dt><code>--enable-link-mutex</code></dt> 1204<dd><p>When building GCC, use a mutex to avoid linking the compilers for 1205multiple languages at the same time, to avoid thrashing on build 1206systems with limited free memory. The default is not to use such a mutex. 1207</p> 1208</dd> 1209<dt><code>--enable-link-serialization</code></dt> 1210<dd><p>When building GCC, use make dependencies to serialize linking the compilers for 1211multiple languages, to avoid thrashing on build 1212systems with limited free memory. The default is not to add such 1213dependencies and thus with parallel make potentially link different 1214compilers concurrently. If the argument is a positive integer, allow 1215that number of concurrent link processes for the large binaries. 1216</p> 1217</dd> 1218<dt><code>--enable-maintainer-mode</code></dt> 1219<dd><p>The build rules that regenerate the Autoconf and Automake output files as 1220well as the GCC master message catalog <samp>gcc.pot</samp> are normally 1221disabled. This is because it can only be rebuilt if the complete source 1222tree is present. If you have changed the sources and want to rebuild the 1223catalog, configuring with <samp>--enable-maintainer-mode</samp> will enable 1224this. Note that you need a recent version of the <code>gettext</code> tools 1225to do so. 1226</p> 1227</dd> 1228<dt><code>--disable-bootstrap</code></dt> 1229<dd><p>For a native build, the default configuration is to perform 1230a 3-stage bootstrap of the compiler when ‘<samp>make</samp>’ is invoked, 1231testing that GCC can compile itself correctly. If you want to disable 1232this process, you can configure with <samp>--disable-bootstrap</samp>. 1233</p> 1234</dd> 1235<dt><code>--enable-bootstrap</code></dt> 1236<dd><p>In special cases, you may want to perform a 3-stage build 1237even if the target and host triplets are different. 1238This is possible when the host can run code compiled for 1239the target (e.g. host is i686-linux, target is i486-linux). 1240Starting from GCC 4.2, to do this you have to configure explicitly 1241with <samp>--enable-bootstrap</samp>. 1242</p> 1243</dd> 1244<dt><code>--enable-generated-files-in-srcdir</code></dt> 1245<dd><p>Neither the .c and .h files that are generated from Bison and flex nor the 1246info manuals and man pages that are built from the .texi files are present 1247in the repository development tree. When building GCC from that development tree, 1248or from one of our snapshots, those generated files are placed in your 1249build directory, which allows for the source to be in a readonly 1250directory. 1251</p> 1252<p>If you configure with <samp>--enable-generated-files-in-srcdir</samp> then those 1253generated files will go into the source directory. This is mainly intended 1254for generating release or prerelease tarballs of the GCC sources, since it 1255is not a requirement that the users of source releases to have flex, Bison, 1256or makeinfo. 1257</p> 1258</dd> 1259<dt><code>--enable-version-specific-runtime-libs</code></dt> 1260<dd><p>Specify 1261that runtime libraries should be installed in the compiler specific 1262subdirectory (<samp><var>libdir</var>/gcc</samp>) rather than the usual places. In 1263addition, ‘<samp>libstdc++</samp>’’s include files will be installed into 1264<samp><var>libdir</var></samp> unless you overruled it by using 1265<samp>--with-gxx-include-dir=<var>dirname</var></samp>. Using this option is 1266particularly useful if you intend to use several versions of GCC in 1267parallel. The default is ‘<samp>yes</samp>’ for ‘<samp>libada</samp>’, and ‘<samp>no</samp>’ for 1268the remaining libraries. 1269</p> 1270</dd> 1271<dt><code><a name="WithAixSoname"></a>--with-aix-soname=‘<samp>aix</samp>’, ‘<samp>svr4</samp>’ or ‘<samp>both</samp>’</code></dt> 1272<dd><p>Traditional AIX shared library versioning (versioned <code>Shared Object</code> 1273files as members of unversioned <code>Archive Library</code> files named 1274‘<samp>lib.a</samp>’) causes numerous headaches for package managers. However, 1275<code>Import Files</code> as members of <code>Archive Library</code> files allow for 1276<strong>filename-based versioning</strong> of shared libraries as seen on Linux/SVR4, 1277where this is called the "SONAME". But as they prevent static linking, 1278<code>Import Files</code> may be used with <code>Runtime Linking</code> only, where the 1279linker does search for ‘<samp>libNAME.so</samp>’ before ‘<samp>libNAME.a</samp>’ library 1280filenames with the ‘<samp>-lNAME</samp>’ linker flag. 1281</p> 1282<a name="AixLdCommand"></a><p>For detailed information please refer to the AIX 1283<a href="https://www.ibm.com/support/knowledgecenter/search/%22the%20ld%20command%2C%20also%20called%20the%20linkage%20editor%20or%20binder%22">ld 1284Command</a> reference. 1285</p> 1286<p>As long as shared library creation is enabled, upon: 1287</p><dl compact="compact"> 1288<dt><code>--with-aix-soname=aix</code></dt> 1289<dt><code>--with-aix-soname=both</code></dt> 1290<dd><p>A (traditional AIX) <code>Shared Archive Library</code> file is created: 1291 </p><ul> 1292<li> using the ‘<samp>libNAME.a</samp>’ filename scheme 1293 </li><li> with the <code>Shared Object</code> file as archive member named 1294 ‘<samp>libNAME.so.V</samp>’ (except for ‘<samp>libgcc_s</samp>’, where the <code>Shared 1295 Object</code> file is named ‘<samp>shr.o</samp>’ for backwards compatibility), which 1296 <ul class="no-bullet"> 1297<li>- is used for runtime loading from inside the ‘<samp>libNAME.a</samp>’ file 1298 </li><li>- is used for dynamic loading via 1299 <code>dlopen("libNAME.a(libNAME.so.V)", RTLD_MEMBER)</code> 1300 </li><li>- is used for shared linking 1301 </li><li>- is used for static linking, so no separate <code>Static Archive 1302 Library</code> file is needed 1303 </li></ul> 1304</li></ul> 1305</dd> 1306<dt><code>--with-aix-soname=both</code></dt> 1307<dt><code>--with-aix-soname=svr4</code></dt> 1308<dd><p>A (second) <code>Shared Archive Library</code> file is created: 1309 </p><ul> 1310<li> using the ‘<samp>libNAME.so.V</samp>’ filename scheme 1311 </li><li> with the <code>Shared Object</code> file as archive member named 1312 ‘<samp>shr.o</samp>’, which 1313 <ul class="no-bullet"> 1314<li>- is created with the <code>-G linker flag</code> 1315 </li><li>- has the <code>F_LOADONLY</code> flag set 1316 </li><li>- is used for runtime loading from inside the ‘<samp>libNAME.so.V</samp>’ file 1317 </li><li>- is used for dynamic loading via <code>dlopen("libNAME.so.V(shr.o)", 1318 RTLD_MEMBER)</code> 1319 </li></ul> 1320</li><li> with the <code>Import File</code> as archive member named ‘<samp>shr.imp</samp>’, 1321 which 1322 <ul class="no-bullet"> 1323<li>- refers to ‘<samp>libNAME.so.V(shr.o)</samp>’ as the "SONAME", to be recorded 1324 in the <code>Loader Section</code> of subsequent binaries 1325 </li><li>- indicates whether ‘<samp>libNAME.so.V(shr.o)</samp>’ is 32 or 64 bit 1326 </li><li>- lists all the public symbols exported by ‘<samp>lib.so.V(shr.o)</samp>’, 1327 eventually decorated with the <code>‘<samp>weak</samp>’ Keyword</code> 1328 </li><li>- is necessary for shared linking against ‘<samp>lib.so.V(shr.o)</samp>’ 1329 </li></ul> 1330</li></ul> 1331<p>A symbolic link using the ‘<samp>libNAME.so</samp>’ filename scheme is created: 1332 </p><ul> 1333<li> pointing to the ‘<samp>libNAME.so.V</samp>’ <code>Shared Archive Library</code> file 1334 </li><li> to permit the <code>ld Command</code> to find ‘<samp>lib.so.V(shr.imp)</samp>’ via 1335 the ‘<samp>-lNAME</samp>’ argument (requires <code>Runtime Linking</code> to be enabled) 1336 </li><li> to permit dynamic loading of ‘<samp>lib.so.V(shr.o)</samp>’ without the need 1337 to specify the version number via <code>dlopen("libNAME.so(shr.o)", 1338 RTLD_MEMBER)</code> 1339 </li></ul> 1340</dd> 1341</dl> 1342 1343<p>As long as static library creation is enabled, upon: 1344</p><dl compact="compact"> 1345<dt><code>--with-aix-soname=svr4</code></dt> 1346<dd><p>A <code>Static Archive Library</code> is created: 1347 </p><ul> 1348<li> using the ‘<samp>libNAME.a</samp>’ filename scheme 1349 </li><li> with all the <code>Static Object</code> files as archive members, which 1350 <ul class="no-bullet"> 1351<li>- are used for static linking 1352 </li></ul> 1353</li></ul> 1354</dd> 1355</dl> 1356 1357<p>While the aix-soname=‘<samp>svr4</samp>’ option does not create <code>Shared Object</code> 1358files as members of unversioned <code>Archive Library</code> files any more, package 1359managers still are responsible to 1360<a href="./specific.html#TransferAixShobj">transfer</a> <code>Shared Object</code> files 1361found as member of a previously installed unversioned <code>Archive Library</code> 1362file into the newly installed <code>Archive Library</code> file with the same 1363filename. 1364</p> 1365<p><em>WARNING:</em> Creating <code>Shared Object</code> files with <code>Runtime Linking</code> 1366enabled may bloat the TOC, eventually leading to <code>TOC overflow</code> errors, 1367requiring the use of either the <samp>-Wl,-bbigtoc</samp> linker flag (seen to 1368break with the <code>GDB</code> debugger) or some of the TOC-related compiler flags, 1369see “RS/6000 and PowerPC Options” in the main manual. 1370</p> 1371<p><samp>--with-aix-soname</samp> is currently supported by ‘<samp>libgcc_s</samp>’ only, so 1372this option is still experimental and not for normal use yet. 1373</p> 1374<p>Default is the traditional behavior <samp>--with-aix-soname=‘<samp>aix</samp>’</samp>. 1375</p> 1376</dd> 1377<dt><code>--enable-languages=<var>lang1</var>,<var>lang2</var>,…</code></dt> 1378<dd><p>Specify that only a particular subset of compilers and 1379their runtime libraries should be built. For a list of valid values for 1380<var>langN</var> you can issue the following command in the 1381<samp>gcc</samp> directory of your GCC source tree:<br> 1382</p><div class="smallexample"> 1383<pre class="smallexample">grep ^language= */config-lang.in 1384</pre></div> 1385<p>Currently, you can use any of the following: 1386<code>all</code>, <code>default</code>, <code>ada</code>, <code>c</code>, <code>c++</code>, <code>d</code>, 1387<code>fortran</code>, <code>go</code>, <code>jit</code>, <code>lto</code>, <code>objc</code>, <code>obj-c++</code>. 1388Building the Ada compiler has special requirements, see below. 1389If you do not pass this flag, or specify the option <code>default</code>, then the 1390default languages available in the <samp>gcc</samp> sub-tree will be configured. 1391Ada, D, Go, Jit, and Objective-C++ are not default languages. LTO is not a 1392default language, but is built by default because <samp>--enable-lto</samp> is 1393enabled by default. The other languages are default languages. If 1394<code>all</code> is specified, then all available languages are built. An 1395exception is <code>jit</code> language, which requires 1396<samp>--enable-host-shared</samp> to be included with <code>all</code>. 1397</p> 1398</dd> 1399<dt><code>--enable-stage1-languages=<var>lang1</var>,<var>lang2</var>,…</code></dt> 1400<dd><p>Specify that a particular subset of compilers and their runtime 1401libraries should be built with the system C compiler during stage 1 of 1402the bootstrap process, rather than only in later stages with the 1403bootstrapped C compiler. The list of valid values is the same as for 1404<samp>--enable-languages</samp>, and the option <code>all</code> will select all 1405of the languages enabled by <samp>--enable-languages</samp>. This option is 1406primarily useful for GCC development; for instance, when a development 1407version of the compiler cannot bootstrap due to compiler bugs, or when 1408one is debugging front ends other than the C front end. When this 1409option is used, one can then build the target libraries for the 1410specified languages with the stage-1 compiler by using <code>make 1411stage1-bubble all-target</code>, or run the testsuite on the stage-1 compiler 1412for the specified languages using <code>make stage1-start check-gcc</code>. 1413</p> 1414</dd> 1415<dt><code>--disable-libada</code></dt> 1416<dd><p>Specify that the run-time libraries and tools used by GNAT should not 1417be built. This can be useful for debugging, or for compatibility with 1418previous Ada build procedures, when it was required to explicitly 1419do a ‘<samp>make -C gcc gnatlib_and_tools</samp>’. 1420</p> 1421</dd> 1422<dt><code>--disable-libsanitizer</code></dt> 1423<dd><p>Specify that the run-time libraries for the various sanitizers should 1424not be built. 1425</p> 1426</dd> 1427<dt><code>--disable-libssp</code></dt> 1428<dd><p>Specify that the run-time libraries for stack smashing protection 1429should not be built or linked against. On many targets library support 1430is provided by the C library instead. 1431</p> 1432</dd> 1433<dt><code>--disable-libquadmath</code></dt> 1434<dd><p>Specify that the GCC quad-precision math library should not be built. 1435On some systems, the library is required to be linkable when building 1436the Fortran front end, unless <samp>--disable-libquadmath-support</samp> 1437is used. 1438</p> 1439</dd> 1440<dt><code>--disable-libquadmath-support</code></dt> 1441<dd><p>Specify that the Fortran front end and <code>libgfortran</code> do not add 1442support for <code>libquadmath</code> on systems supporting it. 1443</p> 1444</dd> 1445<dt><code>--disable-libgomp</code></dt> 1446<dd><p>Specify that the GNU Offloading and Multi Processing Runtime Library 1447should not be built. 1448</p> 1449</dd> 1450<dt><code>--disable-libvtv</code></dt> 1451<dd><p>Specify that the run-time libraries used by vtable verification 1452should not be built. 1453</p> 1454</dd> 1455<dt><code>--with-dwarf2</code></dt> 1456<dd><p>Specify that the compiler should 1457use DWARF 2 debugging information as the default. 1458</p> 1459</dd> 1460<dt><code>--with-advance-toolchain=<var>at</var></code></dt> 1461<dd><p>On 64-bit PowerPC Linux systems, configure the compiler to use the 1462header files, library files, and the dynamic linker from the Advance 1463Toolchain release <var>at</var> instead of the default versions that are 1464provided by the Linux distribution. In general, this option is 1465intended for the developers of GCC, and it is not intended for general 1466use. 1467</p> 1468</dd> 1469<dt><code>--enable-targets=all</code></dt> 1470<dt><code>--enable-targets=<var>target_list</var></code></dt> 1471<dd><p>Some GCC targets, e.g. powerpc64-linux, build bi-arch compilers. 1472These are compilers that are able to generate either 64-bit or 32-bit 1473code. Typically, the corresponding 32-bit target, e.g. 1474powerpc-linux for powerpc64-linux, only generates 32-bit code. This 1475option enables the 32-bit target to be a bi-arch compiler, which is 1476useful when you want a bi-arch compiler that defaults to 32-bit, and 1477you are building a bi-arch or multi-arch binutils in a combined tree. 1478On mips-linux, this will build a tri-arch compiler (ABI o32/n32/64), 1479defaulted to o32. 1480Currently, this option only affects sparc-linux, powerpc-linux, x86-linux, 1481mips-linux and s390-linux. 1482</p> 1483</dd> 1484<dt><code>--enable-default-pie</code></dt> 1485<dd><p>Turn on <samp>-fPIE</samp> and <samp>-pie</samp> by default. 1486</p> 1487</dd> 1488<dt><code>--enable-secureplt</code></dt> 1489<dd><p>This option enables <samp>-msecure-plt</samp> by default for powerpc-linux. 1490See “RS/6000 and PowerPC Options” in the main manual 1491</p> 1492</dd> 1493<dt><code>--enable-default-ssp</code></dt> 1494<dd><p>Turn on <samp>-fstack-protector-strong</samp> by default. 1495</p> 1496</dd> 1497<dt><code>--enable-cld</code></dt> 1498<dd><p>This option enables <samp>-mcld</samp> by default for 32-bit x86 targets. 1499See “i386 and x86-64 Options” in the main manual 1500</p> 1501</dd> 1502<dt><code>--enable-large-address-aware</code></dt> 1503<dd><p>The <samp>--enable-large-address-aware</samp> option arranges for MinGW 1504executables to be linked using the <samp>--large-address-aware</samp> 1505option, that enables the use of more than 2GB of memory. If GCC is 1506configured with this option, its effects can be reversed by passing the 1507<samp>-Wl,--disable-large-address-aware</samp> option to the so-configured 1508compiler driver. 1509</p> 1510</dd> 1511<dt><code>--enable-win32-registry</code></dt> 1512<dt><code>--enable-win32-registry=<var>key</var></code></dt> 1513<dt><code>--disable-win32-registry</code></dt> 1514<dd><p>The <samp>--enable-win32-registry</samp> option enables Microsoft Windows-hosted GCC 1515to look up installations paths in the registry using the following key: 1516</p> 1517<div class="smallexample"> 1518<pre class="smallexample"><code>HKEY_LOCAL_MACHINE\SOFTWARE\Free Software Foundation\<var>key</var></code> 1519</pre></div> 1520 1521<p><var>key</var> defaults to GCC version number, and can be overridden by the 1522<samp>--enable-win32-registry=<var>key</var></samp> option. Vendors and distributors 1523who use custom installers are encouraged to provide a different key, 1524perhaps one comprised of vendor name and GCC version number, to 1525avoid conflict with existing installations. This feature is enabled 1526by default, and can be disabled by <samp>--disable-win32-registry</samp> 1527option. This option has no effect on the other hosts. 1528</p> 1529</dd> 1530<dt><code>--nfp</code></dt> 1531<dd><p>Specify that the machine does not have a floating point unit. This 1532option only applies to ‘<samp>m68k-sun-sunos<var>n</var></samp>’. On any other 1533system, <samp>--nfp</samp> has no effect. 1534</p> 1535</dd> 1536<dt><code>--enable-werror</code></dt> 1537<dt><code>--disable-werror</code></dt> 1538<dt><code>--enable-werror=yes</code></dt> 1539<dt><code>--enable-werror=no</code></dt> 1540<dd><p>When you specify this option, it controls whether certain files in the 1541compiler are built with <samp>-Werror</samp> in bootstrap stage2 and later. 1542If you don’t specify it, <samp>-Werror</samp> is turned on for the main 1543development trunk. However it defaults to off for release branches and 1544final releases. The specific files which get <samp>-Werror</samp> are 1545controlled by the Makefiles. 1546</p> 1547</dd> 1548<dt><code>--enable-checking</code></dt> 1549<dt><code>--disable-checking</code></dt> 1550<dt><code>--enable-checking=<var>list</var></code></dt> 1551<dd><p>This option controls performing internal consistency checks in the compiler. 1552It does not change the generated code, but adds error checking of the 1553requested complexity. This slows down the compiler and may only work 1554properly if you are building the compiler with GCC. 1555</p> 1556<p>When the option is not specified, the active set of checks depends on context. 1557Namely, bootstrap stage 1 defaults to ‘<samp>--enable-checking=yes</samp>’, builds 1558from release branches or release archives default to 1559‘<samp>--enable-checking=release</samp>’, and otherwise 1560‘<samp>--enable-checking=yes,extra</samp>’ is used. When the option is 1561specified without a <var>list</var>, the result is the same as 1562‘<samp>--enable-checking=yes</samp>’. Likewise, ‘<samp>--disable-checking</samp>’ is 1563equivalent to ‘<samp>--enable-checking=no</samp>’. 1564</p> 1565<p>The categories of checks available in <var>list</var> are ‘<samp>yes</samp>’ (most common 1566checks ‘<samp>assert,misc,gc,gimple,rtlflag,runtime,tree,types</samp>’), ‘<samp>no</samp>’ 1567(no checks at all), ‘<samp>all</samp>’ (all but ‘<samp>valgrind</samp>’), ‘<samp>release</samp>’ 1568(cheapest checks ‘<samp>assert,runtime</samp>’) or ‘<samp>none</samp>’ (same as ‘<samp>no</samp>’). 1569‘<samp>release</samp>’ checks are always on and to disable them 1570‘<samp>--disable-checking</samp>’ or ‘<samp>--enable-checking=no[,<other checks>]</samp>’ 1571must be explicitly requested. Disabling assertions makes the compiler and 1572runtime slightly faster but increases the risk of undetected internal errors 1573causing wrong code to be generated. 1574</p> 1575<p>Individual checks can be enabled with these flags: ‘<samp>assert</samp>’, ‘<samp>df</samp>’, 1576‘<samp>extra</samp>’, ‘<samp>fold</samp>’, ‘<samp>gc</samp>’, ‘<samp>gcac</samp>’, ‘<samp>gimple</samp>’, 1577‘<samp>misc</samp>’, ‘<samp>rtl</samp>’, ‘<samp>rtlflag</samp>’, ‘<samp>runtime</samp>’, ‘<samp>tree</samp>’, 1578‘<samp>types</samp>’ and ‘<samp>valgrind</samp>’. ‘<samp>extra</samp>’ extends ‘<samp>misc</samp>’ 1579checking with extra checks that might affect code generation and should 1580therefore not differ between stage1 and later stages in bootstrap. 1581</p> 1582<p>The ‘<samp>valgrind</samp>’ check requires the external <code>valgrind</code> simulator, 1583available from <a href="https://valgrind.org">https://valgrind.org</a>. The ‘<samp>rtl</samp>’ checks are 1584expensive and the ‘<samp>df</samp>’, ‘<samp>gcac</samp>’ and ‘<samp>valgrind</samp>’ checks are very 1585expensive. 1586</p> 1587</dd> 1588<dt><code>--disable-stage1-checking</code></dt> 1589<dt><code>--enable-stage1-checking</code></dt> 1590<dt><code>--enable-stage1-checking=<var>list</var></code></dt> 1591<dd><p>This option affects only bootstrap build. If no <samp>--enable-checking</samp> 1592option is specified the stage1 compiler is built with ‘<samp>yes</samp>’ checking 1593enabled, otherwise the stage1 checking flags are the same as specified by 1594<samp>--enable-checking</samp>. To build the stage1 compiler with 1595different checking options use <samp>--enable-stage1-checking</samp>. 1596The list of checking options is the same as for <samp>--enable-checking</samp>. 1597If your system is too slow or too small to bootstrap a released compiler 1598with checking for stage1 enabled, you can use ‘<samp>--disable-stage1-checking</samp>’ 1599to disable checking for the stage1 compiler. 1600</p> 1601</dd> 1602<dt><code>--enable-coverage</code></dt> 1603<dt><code>--enable-coverage=<var>level</var></code></dt> 1604<dd><p>With this option, the compiler is built to collect self coverage 1605information, every time it is run. This is for internal development 1606purposes, and only works when the compiler is being built with gcc. The 1607<var>level</var> argument controls whether the compiler is built optimized or 1608not, values are ‘<samp>opt</samp>’ and ‘<samp>noopt</samp>’. For coverage analysis you 1609want to disable optimization, for performance analysis you want to 1610enable optimization. When coverage is enabled, the default level is 1611without optimization. 1612</p> 1613</dd> 1614<dt><code>--enable-gather-detailed-mem-stats</code></dt> 1615<dd><p>When this option is specified more detailed information on memory 1616allocation is gathered. This information is printed when using 1617<samp>-fmem-report</samp>. 1618</p> 1619</dd> 1620<dt><code>--enable-valgrind-annotations</code></dt> 1621<dd><p>Mark selected memory related operations in the compiler when run under 1622valgrind to suppress false positives. 1623</p> 1624</dd> 1625<dt><code>--enable-nls</code></dt> 1626<dt><code>--disable-nls</code></dt> 1627<dd><p>The <samp>--enable-nls</samp> option enables Native Language Support (NLS), 1628which lets GCC output diagnostics in languages other than American 1629English. Native Language Support is enabled by default if not doing a 1630canadian cross build. The <samp>--disable-nls</samp> option disables NLS. 1631</p> 1632</dd> 1633<dt><code>--with-included-gettext</code></dt> 1634<dd><p>If NLS is enabled, the <samp>--with-included-gettext</samp> option causes the build 1635procedure to prefer its copy of GNU <code>gettext</code>. 1636</p> 1637</dd> 1638<dt><code>--with-catgets</code></dt> 1639<dd><p>If NLS is enabled, and if the host lacks <code>gettext</code> but has the 1640inferior <code>catgets</code> interface, the GCC build procedure normally 1641ignores <code>catgets</code> and instead uses GCC’s copy of the GNU 1642<code>gettext</code> library. The <samp>--with-catgets</samp> option causes the 1643build procedure to use the host’s <code>catgets</code> in this situation. 1644</p> 1645</dd> 1646<dt><code>--with-libiconv-prefix=<var>dir</var></code></dt> 1647<dd><p>Search for libiconv header files in <samp><var>dir</var>/include</samp> and 1648libiconv library files in <samp><var>dir</var>/lib</samp>. 1649</p> 1650</dd> 1651<dt><code>--enable-obsolete</code></dt> 1652<dd><p>Enable configuration for an obsoleted system. If you attempt to 1653configure GCC for a system (build, host, or target) which has been 1654obsoleted, and you do not specify this flag, configure will halt with an 1655error message. 1656</p> 1657<p>All support for systems which have been obsoleted in one release of GCC 1658is removed entirely in the next major release, unless someone steps 1659forward to maintain the port. 1660</p> 1661</dd> 1662<dt><code>--enable-decimal-float</code></dt> 1663<dt><code>--enable-decimal-float=yes</code></dt> 1664<dt><code>--enable-decimal-float=no</code></dt> 1665<dt><code>--enable-decimal-float=bid</code></dt> 1666<dt><code>--enable-decimal-float=dpd</code></dt> 1667<dt><code>--disable-decimal-float</code></dt> 1668<dd><p>Enable (or disable) support for the C decimal floating point extension 1669that is in the IEEE 754-2008 standard. This is enabled by default only 1670on PowerPC, i386, and x86_64 GNU/Linux systems. Other systems may also 1671support it, but require the user to specifically enable it. You can 1672optionally control which decimal floating point format is used (either 1673‘<samp>bid</samp>’ or ‘<samp>dpd</samp>’). The ‘<samp>bid</samp>’ (binary integer decimal) 1674format is default on i386 and x86_64 systems, and the ‘<samp>dpd</samp>’ 1675(densely packed decimal) format is default on PowerPC systems. 1676</p> 1677</dd> 1678<dt><code>--enable-fixed-point</code></dt> 1679<dt><code>--disable-fixed-point</code></dt> 1680<dd><p>Enable (or disable) support for C fixed-point arithmetic. 1681This option is enabled by default for some targets (such as MIPS) which 1682have hardware-support for fixed-point operations. On other targets, you 1683may enable this option manually. 1684</p> 1685</dd> 1686<dt><code>--with-long-double-128</code></dt> 1687<dd><p>Specify if <code>long double</code> type should be 128-bit by default on selected 1688GNU/Linux architectures. If using <code>--without-long-double-128</code>, 1689<code>long double</code> will be by default 64-bit, the same as <code>double</code> type. 1690When neither of these configure options are used, the default will be 1691128-bit <code>long double</code> when built against GNU C Library 2.4 and later, 169264-bit <code>long double</code> otherwise. 1693</p> 1694</dd> 1695<dt><code>--with-long-double-format=ibm</code></dt> 1696<dt><code>--with-long-double-format=ieee</code></dt> 1697<dd><p>Specify whether <code>long double</code> uses the IBM extended double format 1698or the IEEE 128-bit floating point format on PowerPC Linux systems. 1699This configuration switch will only work on little endian PowerPC 1700Linux systems and on big endian 64-bit systems where the default cpu 1701is at least power7 (i.e. <samp>--with-cpu=power7</samp>, 1702<samp>--with-cpu=power8</samp>, or <samp>--with-cpu=power9</samp> is used). 1703</p> 1704<p>If you use the <samp>--with-long-double-64</samp> configuration option, 1705the <samp>--with-long-double-format=ibm</samp> and 1706<samp>--with-long-double-format=ieee</samp> options are ignored. 1707</p> 1708<p>The default <code>long double</code> format is to use IBM extended double. 1709Until all of the libraries are converted to use IEEE 128-bit floating 1710point, it is not recommended to use 1711<samp>--with-long-double-format=ieee</samp>. 1712</p> 1713</dd> 1714<dt><code>--enable-fdpic</code></dt> 1715<dd><p>On SH Linux systems, generate ELF FDPIC code. 1716</p> 1717</dd> 1718<dt><code>--with-gmp=<var>pathname</var></code></dt> 1719<dt><code>--with-gmp-include=<var>pathname</var></code></dt> 1720<dt><code>--with-gmp-lib=<var>pathname</var></code></dt> 1721<dt><code>--with-mpfr=<var>pathname</var></code></dt> 1722<dt><code>--with-mpfr-include=<var>pathname</var></code></dt> 1723<dt><code>--with-mpfr-lib=<var>pathname</var></code></dt> 1724<dt><code>--with-mpc=<var>pathname</var></code></dt> 1725<dt><code>--with-mpc-include=<var>pathname</var></code></dt> 1726<dt><code>--with-mpc-lib=<var>pathname</var></code></dt> 1727<dd><p>If you want to build GCC but do not have the GMP library, the MPFR 1728library and/or the MPC library installed in a standard location and 1729do not have their sources present in the GCC source tree then you 1730can explicitly specify the directory where they are installed 1731(‘<samp>--with-gmp=<var>gmpinstalldir</var></samp>’, 1732‘<samp>--with-mpfr=<var>mpfrinstalldir</var></samp>’, 1733‘<samp>--with-mpc=<var>mpcinstalldir</var></samp>’). The 1734<samp>--with-gmp=<var>gmpinstalldir</var></samp> option is shorthand for 1735<samp>--with-gmp-lib=<var>gmpinstalldir</var>/lib</samp> and 1736<samp>--with-gmp-include=<var>gmpinstalldir</var>/include</samp>. Likewise the 1737<samp>--with-mpfr=<var>mpfrinstalldir</var></samp> option is shorthand for 1738<samp>--with-mpfr-lib=<var>mpfrinstalldir</var>/lib</samp> and 1739<samp>--with-mpfr-include=<var>mpfrinstalldir</var>/include</samp>, also the 1740<samp>--with-mpc=<var>mpcinstalldir</var></samp> option is shorthand for 1741<samp>--with-mpc-lib=<var>mpcinstalldir</var>/lib</samp> and 1742<samp>--with-mpc-include=<var>mpcinstalldir</var>/include</samp>. If these 1743shorthand assumptions are not correct, you can use the explicit 1744include and lib options directly. You might also need to ensure the 1745shared libraries can be found by the dynamic linker when building and 1746using GCC, for example by setting the runtime shared library path 1747variable (<code>LD_LIBRARY_PATH</code> on GNU/Linux and Solaris systems). 1748</p> 1749<p>These flags are applicable to the host platform only. When building 1750a cross compiler, they will not be used to configure target libraries. 1751</p> 1752</dd> 1753<dt><code>--with-isl=<var>pathname</var></code></dt> 1754<dt><code>--with-isl-include=<var>pathname</var></code></dt> 1755<dt><code>--with-isl-lib=<var>pathname</var></code></dt> 1756<dd><p>If you do not have the isl library installed in a standard location and you 1757want to build GCC, you can explicitly specify the directory where it is 1758installed (‘<samp>--with-isl=<var>islinstalldir</var></samp>’). The 1759<samp>--with-isl=<var>islinstalldir</var></samp> option is shorthand for 1760<samp>--with-isl-lib=<var>islinstalldir</var>/lib</samp> and 1761<samp>--with-isl-include=<var>islinstalldir</var>/include</samp>. If this 1762shorthand assumption is not correct, you can use the explicit 1763include and lib options directly. 1764</p> 1765<p>These flags are applicable to the host platform only. When building 1766a cross compiler, they will not be used to configure target libraries. 1767</p> 1768</dd> 1769<dt><code>--with-stage1-ldflags=<var>flags</var></code></dt> 1770<dd><p>This option may be used to set linker flags to be used when linking 1771stage 1 of GCC. These are also used when linking GCC if configured with 1772<samp>--disable-bootstrap</samp>. If <samp>--with-stage1-libs</samp> is not set to a 1773value, then the default is ‘<samp>-static-libstdc++ -static-libgcc</samp>’, if 1774supported. 1775</p> 1776</dd> 1777<dt><code>--with-stage1-libs=<var>libs</var></code></dt> 1778<dd><p>This option may be used to set libraries to be used when linking stage 1 1779of GCC. These are also used when linking GCC if configured with 1780<samp>--disable-bootstrap</samp>. 1781</p> 1782</dd> 1783<dt><code>--with-boot-ldflags=<var>flags</var></code></dt> 1784<dd><p>This option may be used to set linker flags to be used when linking 1785stage 2 and later when bootstrapping GCC. If –with-boot-libs 1786is not is set to a value, then the default is 1787‘<samp>-static-libstdc++ -static-libgcc</samp>’. 1788</p> 1789</dd> 1790<dt><code>--with-boot-libs=<var>libs</var></code></dt> 1791<dd><p>This option may be used to set libraries to be used when linking stage 2 1792and later when bootstrapping GCC. 1793</p> 1794</dd> 1795<dt><code>--with-debug-prefix-map=<var>map</var></code></dt> 1796<dd><p>Convert source directory names using <samp>-fdebug-prefix-map</samp> when 1797building runtime libraries. ‘<samp><var>map</var></samp>’ is a space-separated 1798list of maps of the form ‘<samp><var>old</var>=<var>new</var></samp>’. 1799</p> 1800</dd> 1801<dt><code>--enable-linker-build-id</code></dt> 1802<dd><p>Tells GCC to pass <samp>--build-id</samp> option to the linker for all final 1803links (links performed without the <samp>-r</samp> or <samp>--relocatable</samp> 1804option), if the linker supports it. If you specify 1805<samp>--enable-linker-build-id</samp>, but your linker does not 1806support <samp>--build-id</samp> option, a warning is issued and the 1807<samp>--enable-linker-build-id</samp> option is ignored. The default is off. 1808</p> 1809</dd> 1810<dt><code>--with-linker-hash-style=<var>choice</var></code></dt> 1811<dd><p>Tells GCC to pass <samp>--hash-style=<var>choice</var></samp> option to the 1812linker for all final links. <var>choice</var> can be one of 1813‘<samp>sysv</samp>’, ‘<samp>gnu</samp>’, and ‘<samp>both</samp>’ where ‘<samp>sysv</samp>’ is the default. 1814</p> 1815</dd> 1816<dt><code>--enable-gnu-unique-object</code></dt> 1817<dt><code>--disable-gnu-unique-object</code></dt> 1818<dd><p>Tells GCC to use the gnu_unique_object relocation for C++ template 1819static data members and inline function local statics. Enabled by 1820default for a toolchain with an assembler that accepts it and 1821GLIBC 2.11 or above, otherwise disabled. 1822</p> 1823</dd> 1824<dt><code>--with-diagnostics-color=<var>choice</var></code></dt> 1825<dd><p>Tells GCC to use <var>choice</var> as the default for <samp>-fdiagnostics-color=</samp> 1826option (if not used explicitly on the command line). <var>choice</var> 1827can be one of ‘<samp>never</samp>’, ‘<samp>auto</samp>’, ‘<samp>always</samp>’, and ‘<samp>auto-if-env</samp>’ 1828where ‘<samp>auto</samp>’ is the default. ‘<samp>auto-if-env</samp>’ makes 1829<samp>-fdiagnostics-color=auto</samp> the default if <code>GCC_COLORS</code> 1830is present and non-empty in the environment of the compiler, and 1831<samp>-fdiagnostics-color=never</samp> otherwise. 1832</p> 1833</dd> 1834<dt><code>--with-diagnostics-urls=<var>choice</var></code></dt> 1835<dd><p>Tells GCC to use <var>choice</var> as the default for <samp>-fdiagnostics-urls=</samp> 1836option (if not used explicitly on the command line). <var>choice</var> 1837can be one of ‘<samp>never</samp>’, ‘<samp>auto</samp>’, ‘<samp>always</samp>’, and ‘<samp>auto-if-env</samp>’ 1838where ‘<samp>auto</samp>’ is the default. ‘<samp>auto-if-env</samp>’ makes 1839<samp>-fdiagnostics-urls=auto</samp> the default if <code>GCC_URLS</code> 1840or <code>TERM_URLS</code> is present and non-empty in the environment of the 1841compiler, and <samp>-fdiagnostics-urls=never</samp> otherwise. 1842</p> 1843</dd> 1844<dt><code>--enable-lto</code></dt> 1845<dt><code>--disable-lto</code></dt> 1846<dd><p>Enable support for link-time optimization (LTO). This is enabled by 1847default, and may be disabled using <samp>--disable-lto</samp>. 1848</p> 1849</dd> 1850<dt><code>--enable-linker-plugin-configure-flags=FLAGS</code></dt> 1851<dt><code>--enable-linker-plugin-flags=FLAGS</code></dt> 1852<dd><p>By default, linker plugins (such as the LTO plugin) are built for the 1853host system architecture. For the case that the linker has a 1854different (but run-time compatible) architecture, these flags can be 1855specified to build plugins that are compatible to the linker. For 1856example, if you are building GCC for a 64-bit x86_64 1857(‘<samp>x86_64-pc-linux-gnu</samp>’) host system, but have a 32-bit x86 1858GNU/Linux (‘<samp>i686-pc-linux-gnu</samp>’) linker executable (which is 1859executable on the former system), you can configure GCC as follows for 1860getting compatible linker plugins: 1861</p> 1862<div class="smallexample"> 1863<pre class="smallexample">% <var>srcdir</var>/configure \ 1864 --host=x86_64-pc-linux-gnu \ 1865 --enable-linker-plugin-configure-flags=--host=i686-pc-linux-gnu \ 1866 --enable-linker-plugin-flags='CC=gcc\ -m32\ -Wl,-rpath,[...]/i686-pc-linux-gnu/lib' 1867</pre></div> 1868 1869</dd> 1870<dt><code>--with-plugin-ld=<var>pathname</var></code></dt> 1871<dd><p>Enable an alternate linker to be used at link-time optimization (LTO) 1872link time when <samp>-fuse-linker-plugin</samp> is enabled. 1873This linker should have plugin support such as gold starting with 1874version 2.20 or GNU ld starting with version 2.21. 1875See <samp>-fuse-linker-plugin</samp> for details. 1876</p> 1877</dd> 1878<dt><code>--enable-canonical-system-headers</code></dt> 1879<dt><code>--disable-canonical-system-headers</code></dt> 1880<dd><p>Enable system header path canonicalization for <samp>libcpp</samp>. This can 1881produce shorter header file paths in diagnostics and dependency output 1882files, but these changed header paths may conflict with some compilation 1883environments. Enabled by default, and may be disabled using 1884<samp>--disable-canonical-system-headers</samp>. 1885</p> 1886</dd> 1887<dt><code>--with-glibc-version=<var>major</var>.<var>minor</var></code></dt> 1888<dd><p>Tell GCC that when the GNU C Library (glibc) is used on the target it 1889will be version <var>major</var>.<var>minor</var> or later. Normally this can 1890be detected from the C library’s header files, but this option may be 1891needed when bootstrapping a cross toolchain without the header files 1892available for building the initial bootstrap compiler. 1893</p> 1894<p>If GCC is configured with some multilibs that use glibc and some that 1895do not, this option applies only to the multilibs that use glibc. 1896However, such configurations may not work well as not all the relevant 1897configuration in GCC is on a per-multilib basis. 1898</p> 1899</dd> 1900<dt><code>--enable-as-accelerator-for=<var>target</var></code></dt> 1901<dd><p>Build as offload target compiler. Specify offload host triple by <var>target</var>. 1902</p> 1903</dd> 1904<dt><code>--enable-offload-targets=<var>target1</var>[=<var>path1</var>],…,<var>targetN</var>[=<var>pathN</var>]</code></dt> 1905<dd><p>Enable offloading to targets <var>target1</var>, …, <var>targetN</var>. 1906Offload compilers are expected to be already installed. Default search 1907path for them is <samp><var>exec-prefix</var></samp>, but it can be changed by 1908specifying paths <var>path1</var>, …, <var>pathN</var>. 1909</p> 1910<div class="smallexample"> 1911<pre class="smallexample">% <var>srcdir</var>/configure \ 1912 --enable-offload-targets=x86_64-intelmicemul-linux-gnu=/path/to/x86_64/compiler,nvptx-none 1913</pre></div> 1914 1915</dd> 1916<dt><code>--enable-offload-defaulted</code></dt> 1917<dd> 1918<p>Tell GCC that configured but not installed offload compilers and libgomp 1919plugins are silently ignored. Useful for distribution compilers where 1920those are in separate optional packages and where the presence or absence 1921of those optional packages should determine the actual supported offloading 1922target set rather than the GCC configure-time selection. 1923</p> 1924</dd> 1925<dt><code>--with-hsa-runtime=<var>pathname</var></code></dt> 1926<dt><code>--with-hsa-runtime-include=<var>pathname</var></code></dt> 1927<dt><code>--with-hsa-runtime-lib=<var>pathname</var></code></dt> 1928<dd> 1929<p>If you configure GCC with offloading which uses an HSA run-time such as 1930AMDGCN but do not have the HSA run-time library installed in a standard 1931location then you can explicitly specify the directory where they are 1932installed. The <samp>--with-hsa-runtime=<var>hsainstalldir</var></samp> option 1933is a shorthand for 1934<samp>--with-hsa-runtime-lib=<var>hsainstalldir</var>/lib</samp> and 1935<samp>--with-hsa-runtime-include=<var>hsainstalldir</var>/include</samp>. 1936</p> 1937</dd> 1938<dt><code>--enable-cet</code></dt> 1939<dt><code>--disable-cet</code></dt> 1940<dd><p>Enable building target run-time libraries with control-flow 1941instrumentation, see <samp>-fcf-protection</samp> option. When 1942<code>--enable-cet</code> is specified target libraries are configured 1943to add <samp>-fcf-protection</samp> and, if needed, other target 1944specific options to a set of building options. 1945</p> 1946<p><code>--enable-cet=auto</code> is default. CET is enabled on Linux/x86 if 1947target binutils supports <code>Intel CET</code> instructions and disabled 1948otherwise. In this case, the target libraries are configured to get 1949additional <samp>-fcf-protection</samp> option. 1950</p> 1951</dd> 1952<dt><code>--with-riscv-attribute=‘<samp>yes</samp>’, ‘<samp>no</samp>’ or ‘<samp>default</samp>’</code></dt> 1953<dd><p>Generate RISC-V attribute by default, in order to record extra build 1954information in object. 1955</p> 1956<p>The option is disabled by default. It is enabled on RISC-V/ELF (bare-metal) 1957target if target binutils supported. 1958</p> 1959</dd> 1960<dt><code>--enable-s390-excess-float-precision</code></dt> 1961<dt><code>--disable-s390-excess-float-precision</code></dt> 1962<dd><p>On s390(x) targets, enable treatment of float expressions with double precision 1963when in standards-compliant mode (e.g., when <code>--std=c99</code> or 1964<code>-fexcess-precision=standard</code> are given). 1965</p> 1966<p>For a native build and cross compiles that have target headers, the option’s 1967default is derived from glibc’s behavior. When glibc clamps float_t to double, 1968GCC follows and enables the option. For other cross compiles, the default is 1969disabled. 1970</p> 1971</dd> 1972<dt><code>--with-zstd=<var>pathname</var></code></dt> 1973<dt><code>--with-zstd-include=<var>pathname</var></code></dt> 1974<dt><code>--with-zstd-lib=<var>pathname</var></code></dt> 1975<dd><p>If you do not have the <code>zstd</code> library installed in a standard 1976location and you want to build GCC, you can explicitly specify the 1977directory where it is installed (‘<samp>--with-zstd=<var>zstdinstalldir</var></samp>’). 1978The <samp>--with-zstd=<var>zstdinstalldir</var></samp> option is shorthand for 1979<samp>--with-zstd-lib=<var>zstdinstalldir</var>/lib</samp> and 1980<samp>--with-zstd-include=<var>zstdinstalldir</var>/include</samp>. If this 1981shorthand assumption is not correct, you can use the explicit 1982include and lib options directly. 1983</p> 1984<p>These flags are applicable to the host platform only. When building 1985a cross compiler, they will not be used to configure target libraries. 1986</p></dd> 1987</dl> 1988 1989<a name="Cross-Compiler-Specific-Options"></a> 1990<h4 class="subheading">Cross-Compiler-Specific Options</h4> 1991<p>The following options only apply to building cross compilers. 1992</p> 1993<dl compact="compact"> 1994<dt><code>--with-toolexeclibdir=<var>dir</var></code></dt> 1995<dd><p>Specify the installation directory for libraries built with a cross compiler. 1996The default is <samp>${gcc_tooldir}/lib</samp>. 1997</p> 1998</dd> 1999<dt><code>--with-sysroot</code></dt> 2000<dt><code>--with-sysroot=<var>dir</var></code></dt> 2001<dd><p>Tells GCC to consider <var>dir</var> as the root of a tree that contains 2002(a subset of) the root filesystem of the target operating system. 2003Target system headers, libraries and run-time object files will be 2004searched for in there. More specifically, this acts as if 2005<samp>--sysroot=<var>dir</var></samp> was added to the default options of the built 2006compiler. The specified directory is not copied into the 2007install tree, unlike the options <samp>--with-headers</samp> and 2008<samp>--with-libs</samp> that this option obsoletes. The default value, 2009in case <samp>--with-sysroot</samp> is not given an argument, is 2010<samp>${gcc_tooldir}/sys-root</samp>. If the specified directory is a 2011subdirectory of <samp>${exec_prefix}</samp>, then it will be found relative to 2012the GCC binaries if the installation tree is moved. 2013</p> 2014<p>This option affects the system root for the compiler used to build 2015target libraries (which runs on the build system) and the compiler newly 2016installed with <code>make install</code>; it does not affect the compiler which is 2017used to build GCC itself. 2018</p> 2019<p>If you specify the <samp>--with-native-system-header-dir=<var>dirname</var></samp> 2020option then the compiler will search that directory within <var>dirname</var> for 2021native system headers rather than the default <samp>/usr/include</samp>. 2022</p> 2023</dd> 2024<dt><code>--with-build-sysroot</code></dt> 2025<dt><code>--with-build-sysroot=<var>dir</var></code></dt> 2026<dd><p>Tells GCC to consider <var>dir</var> as the system root (see 2027<samp>--with-sysroot</samp>) while building target libraries, instead of 2028the directory specified with <samp>--with-sysroot</samp>. This option is 2029only useful when you are already using <samp>--with-sysroot</samp>. You 2030can use <samp>--with-build-sysroot</samp> when you are configuring with 2031<samp>--prefix</samp> set to a directory that is different from the one in 2032which you are installing GCC and your target libraries. 2033</p> 2034<p>This option affects the system root for the compiler used to build 2035target libraries (which runs on the build system); it does not affect 2036the compiler which is used to build GCC itself. 2037</p> 2038<p>If you specify the <samp>--with-native-system-header-dir=<var>dirname</var></samp> 2039option then the compiler will search that directory within <var>dirname</var> for 2040native system headers rather than the default <samp>/usr/include</samp>. 2041</p> 2042</dd> 2043<dt><code>--with-headers</code></dt> 2044<dt><code>--with-headers=<var>dir</var></code></dt> 2045<dd><p>Deprecated in favor of <samp>--with-sysroot</samp>. 2046Specifies that target headers are available when building a cross compiler. 2047The <var>dir</var> argument specifies a directory which has the target include 2048files. These include files will be copied into the <samp>gcc</samp> install 2049directory. <em>This option with the <var>dir</var> argument is required</em> when 2050building a cross compiler, if <samp><var>prefix</var>/<var>target</var>/sys-include</samp> 2051doesn’t pre-exist. If <samp><var>prefix</var>/<var>target</var>/sys-include</samp> does 2052pre-exist, the <var>dir</var> argument may be omitted. <code>fixincludes</code> 2053will be run on these files to make them compatible with GCC. 2054</p> 2055</dd> 2056<dt><code>--without-headers</code></dt> 2057<dd><p>Tells GCC not use any target headers from a libc when building a cross 2058compiler. When crossing to GNU/Linux, you need the headers so GCC 2059can build the exception handling for libgcc. 2060</p> 2061</dd> 2062<dt><code>--with-libs</code></dt> 2063<dt><code>--with-libs="<var>dir1</var> <var>dir2</var> … <var>dirN</var>"</code></dt> 2064<dd><p>Deprecated in favor of <samp>--with-sysroot</samp>. 2065Specifies a list of directories which contain the target runtime 2066libraries. These libraries will be copied into the <samp>gcc</samp> install 2067directory. If the directory list is omitted, this option has no 2068effect. 2069</p> 2070</dd> 2071<dt><code>--with-newlib</code></dt> 2072<dd><p>Specifies that ‘<samp>newlib</samp>’ is 2073being used as the target C library. This causes <code>__eprintf</code> to be 2074omitted from <samp>libgcc.a</samp> on the assumption that it will be provided by 2075‘<samp>newlib</samp>’. 2076</p> 2077<a name="avr"></a> 2078</dd> 2079<dt><code>--with-avrlibc</code></dt> 2080<dd><p>Only supported for the AVR target. Specifies that ‘<samp>AVR-Libc</samp>’ is 2081being used as the target C library. This causes float support 2082functions like <code>__addsf3</code> to be omitted from <samp>libgcc.a</samp> on 2083the assumption that it will be provided by <samp>libm.a</samp>. For more 2084technical details, cf. <a href="https://gcc.gnu.org/PR54461">PR54461</a>. 2085It is not supported for 2086RTEMS configurations, which currently use newlib. The option is 2087supported since version 4.7.2 and is the default in 4.8.0 and newer. 2088</p> 2089</dd> 2090<dt><code>--with-double={32|64|32,64|64,32}</code></dt> 2091<dt><code>--with-long-double={32|64|32,64|64,32|double}</code></dt> 2092<dd><p>Only supported for the AVR target since version 10. 2093Specify the default layout available for the C/C++ ‘<samp>double</samp>’ 2094and ‘<samp>long double</samp>’ type, respectively. The following rules apply: 2095</p><ul> 2096<li> The first value after the ‘<samp>=</samp>’ specifies the default layout (in bits) 2097of the type and also the default for the <samp>-mdouble=</samp> resp. 2098<samp>-mlong-double=</samp> compiler option. 2099</li><li> If more than one value is specified, respective multilib variants are 2100available, and <samp>-mdouble=</samp> resp. <samp>-mlong-double=</samp> acts 2101as a multilib option. 2102</li><li> If <samp>--with-long-double=double</samp> is specified, ‘<samp>double</samp>’ and 2103‘<samp>long double</samp>’ will have the same layout. 2104</li><li> The defaults are <samp>--with-long-double=64,32</samp> and 2105<samp>--with-double=32,64</samp>. The default ‘<samp>double</samp>’ layout imposed by 2106the latter is compatible with older versions of the compiler that implement 2107‘<samp>double</samp>’ as a 32-bit type, which does not comply to the language standard. 2108</li></ul> 2109<p>Not all combinations of <samp>--with-double=</samp> and 2110<samp>--with-long-double=</samp> are valid. For example, the combination 2111<samp>--with-double=32,64</samp> <samp>--with-long-double=32</samp> will be 2112rejected because the first option specifies the availability of 2113multilibs for ‘<samp>double</samp>’, whereas the second option implies 2114that ‘<samp>long double</samp>’ — and hence also ‘<samp>double</samp>’ — is always 211532 bits wide. 2116</p> 2117</dd> 2118<dt><code>--with-double-comparison={tristate|bool|libf7}</code></dt> 2119<dd><p>Only supported for the AVR target since version 10. 2120Specify what result format is returned by library functions that 2121compare 64-bit floating point values (<code>DFmode</code>). 2122The GCC default is ‘<samp>tristate</samp>’. If the floating point 2123implementation returns a boolean instead, set it to ‘<samp>bool</samp>’. 2124</p> 2125</dd> 2126<dt><code>--with-libf7={libgcc|math|math-symbols|no}</code></dt> 2127<dd><p>Only supported for the AVR target since version 10. 2128Specify to which degree code from LibF7 is included in libgcc. 2129LibF7 is an ad-hoc, AVR-specific, 64-bit floating point emulation 2130written in C and (inline) assembly. ‘<samp>libgcc</samp>’ adds support 2131for functions that one would usually expect in libgcc like double addition, 2132double comparisons and double conversions. ‘<samp>math</samp>’ also adds routines 2133that one would expect in <samp>libm.a</samp>, but with <code>__</code> (two underscores) 2134prepended to the symbol names as specified by <samp>math.h</samp>. 2135‘<samp>math-symbols</samp>’ also defines weak aliases for the functions 2136declared in <samp>math.h</samp>. However, <code>--with-libf7</code> won’t 2137install no <samp>math.h</samp> header file whatsoever, this file must come 2138from elsewhere. This option sets <samp>--with-double-comparison</samp> 2139to ‘<samp>bool</samp>’. 2140</p> 2141</dd> 2142<dt><code>--with-nds32-lib=<var>library</var></code></dt> 2143<dd><p>Specifies that <var>library</var> setting is used for building <samp>libgcc.a</samp>. 2144Currently, the valid <var>library</var> is ‘<samp>newlib</samp>’ or ‘<samp>mculib</samp>’. 2145This option is only supported for the NDS32 target. 2146</p> 2147</dd> 2148<dt><code>--with-build-time-tools=<var>dir</var></code></dt> 2149<dd><p>Specifies where to find the set of target tools (assembler, linker, etc.) 2150that will be used while building GCC itself. This option can be useful 2151if the directory layouts are different between the system you are building 2152GCC on, and the system where you will deploy it. 2153</p> 2154<p>For example, on an ‘<samp>ia64-hp-hpux</samp>’ system, you may have the GNU 2155assembler and linker in <samp>/usr/bin</samp>, and the native tools in a 2156different path, and build a toolchain that expects to find the 2157native tools in <samp>/usr/bin</samp>. 2158</p> 2159<p>When you use this option, you should ensure that <var>dir</var> includes 2160<code>ar</code>, <code>as</code>, <code>ld</code>, <code>nm</code>, 2161<code>ranlib</code> and <code>strip</code> if necessary, and possibly 2162<code>objdump</code>. Otherwise, GCC may use an inconsistent set of 2163tools. 2164</p></dd> 2165</dl> 2166 2167<a name="Overriding-configure-test-results"></a> 2168<h4 class="subsubheading">Overriding <code>configure</code> test results</h4> 2169 2170<p>Sometimes, it might be necessary to override the result of some 2171<code>configure</code> test, for example in order to ease porting to a new 2172system or work around a bug in a test. The toplevel <code>configure</code> 2173script provides three variables for this: 2174</p> 2175<dl compact="compact"> 2176<dt><code>build_configargs</code></dt> 2177<dd><a name="index-build_005fconfigargs"></a> 2178<p>The contents of this variable is passed to all build <code>configure</code> 2179scripts. 2180</p> 2181</dd> 2182<dt><code>host_configargs</code></dt> 2183<dd><a name="index-host_005fconfigargs"></a> 2184<p>The contents of this variable is passed to all host <code>configure</code> 2185scripts. 2186</p> 2187</dd> 2188<dt><code>target_configargs</code></dt> 2189<dd><a name="index-target_005fconfigargs"></a> 2190<p>The contents of this variable is passed to all target <code>configure</code> 2191scripts. 2192</p> 2193</dd> 2194</dl> 2195 2196<p>In order to avoid shell and <code>make</code> quoting issues for complex 2197overrides, you can pass a setting for <code>CONFIG_SITE</code> and set 2198variables in the site file. 2199</p> 2200<a name="Objective-C-Specific-Options"></a> 2201<h4 class="subheading">Objective-C-Specific Options</h4> 2202 2203<p>The following options apply to the build of the Objective-C runtime library. 2204</p> 2205<dl compact="compact"> 2206<dt><code>--enable-objc-gc</code></dt> 2207<dd><p>Specify that an additional variant of the GNU Objective-C runtime library 2208is built, using an external build of the Boehm-Demers-Weiser garbage 2209collector (<a href="https://www.hboehm.info/gc/">https://www.hboehm.info/gc/</a>). This library needs to be 2210available for each multilib variant, unless configured with 2211<samp>--enable-objc-gc=‘<samp>auto</samp>’</samp> in which case the build of the 2212additional runtime library is skipped when not available and the build 2213continues. 2214</p> 2215</dd> 2216<dt><code>--with-target-bdw-gc=<var>list</var></code></dt> 2217<dt><code>--with-target-bdw-gc-include=<var>list</var></code></dt> 2218<dt><code>--with-target-bdw-gc-lib=<var>list</var></code></dt> 2219<dd><p>Specify search directories for the garbage collector header files and 2220libraries. <var>list</var> is a comma separated list of key value pairs of the 2221form ‘<samp><var>multilibdir</var>=<var>path</var></samp>’, where the default multilib key 2222is named as ‘<samp>.</samp>’ (dot), or is omitted (e.g. 2223‘<samp>--with-target-bdw-gc=/opt/bdw-gc,32=/opt-bdw-gc32</samp>’). 2224</p> 2225<p>The options <samp>--with-target-bdw-gc-include</samp> and 2226<samp>--with-target-bdw-gc-lib</samp> must always be specified together 2227for each multilib variant and they take precedence over 2228<samp>--with-target-bdw-gc</samp>. If <samp>--with-target-bdw-gc-include</samp> 2229is missing values for a multilib, then the value for the default 2230multilib is used (e.g. ‘<samp>--with-target-bdw-gc-include=/opt/bdw-gc/include</samp>’ 2231‘<samp>--with-target-bdw-gc-lib=/opt/bdw-gc/lib64,32=/opt-bdw-gc/lib32</samp>’). 2232If none of these options are specified, the library is assumed in 2233default locations. 2234</p></dd> 2235</dl> 2236 2237<a name="D-Specific-Options"></a> 2238<h4 class="subheading">D-Specific Options</h4> 2239 2240<p>The following options apply to the build of the D runtime library. 2241</p> 2242<dl compact="compact"> 2243<dt><code>--enable-libphobos-checking</code></dt> 2244<dt><code>--disable-libphobos-checking</code></dt> 2245<dt><code>--enable-libphobos-checking=<var>list</var></code></dt> 2246<dd><p>This option controls whether run-time checks and contracts are compiled into 2247the D runtime library. When the option is not specified, the library is built 2248with ‘<samp>release</samp>’ checking. When the option is specified without a 2249<var>list</var>, the result is the same as ‘<samp>--enable-libphobos-checking=yes</samp>’. 2250Likewise, ‘<samp>--disable-libphobos-checking</samp>’ is equivalent to 2251‘<samp>--enable-libphobos-checking=no</samp>’. 2252</p> 2253<p>The categories of checks available in <var>list</var> are ‘<samp>yes</samp>’ (compiles 2254libphobos with <samp>-fno-release</samp>), ‘<samp>no</samp>’ (compiles libphobos with 2255<samp>-frelease</samp>), ‘<samp>all</samp>’ (same as ‘<samp>yes</samp>’), ‘<samp>none</samp>’ or 2256‘<samp>release</samp>’ (same as ‘<samp>no</samp>’). 2257</p> 2258<p>Individual checks available in <var>list</var> are ‘<samp>assert</samp>’ (compiles libphobos 2259with an extra option <samp>-fassert</samp>). 2260</p> 2261</dd> 2262<dt><code>--with-libphobos-druntime-only</code></dt> 2263<dt><code>--with-libphobos-druntime-only=<var>choice</var></code></dt> 2264<dd><p>Specify whether to build only the core D runtime library (druntime), or both 2265the core and standard library (phobos) into libphobos. This is useful for 2266targets that have full support in druntime, but no or incomplete support 2267in phobos. <var>choice</var> can be one of ‘<samp>auto</samp>’, ‘<samp>yes</samp>’, and ‘<samp>no</samp>’ 2268where ‘<samp>auto</samp>’ is the default. 2269</p> 2270<p>When the option is not specified, the default choice ‘<samp>auto</samp>’ means that it 2271is inferred whether the target has support for the phobos standard library. 2272When the option is specified without a <var>choice</var>, the result is the same as 2273‘<samp>--with-libphobos-druntime-only=yes</samp>’. 2274</p> 2275</dd> 2276<dt><code>--with-target-system-zlib</code></dt> 2277<dd><p>Use installed ‘<samp>zlib</samp>’ rather than that included with GCC. This needs 2278to be available for each multilib variant, unless configured with 2279<samp>--with-target-system-zlib=‘<samp>auto</samp>’</samp> in which case the GCC included 2280‘<samp>zlib</samp>’ is only used when the system installed library is not available. 2281</p></dd> 2282</dl> 2283 2284<hr /> 2285<p> 2286<p><a href="./index.html">Return to the GCC Installation page</a> 2287</p> 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298</body> 2299</html> 2300