1 \input texinfo.tex @c -*-texinfo-*-
4 @setfilename gccinstall.info
9 @include gcc-common.texi
11 @c Specify title for specific html page
13 @settitle Installing GCC
16 @settitle Host/Target specific installation notes for GCC
18 @ifset prerequisiteshtml
19 @settitle Prerequisites for GCC
22 @settitle Downloading GCC
25 @settitle Installing GCC: Configuration
28 @settitle Installing GCC: Building
31 @settitle Installing GCC: Testing
33 @ifset finalinstallhtml
34 @settitle Installing GCC: Final installation
37 @settitle Installing GCC: Binaries
40 @settitle Installing GCC: Old documentation
43 @settitle Installing GCC: GNU Free Documentation License
46 @c Copyright (C) 1988-2019 Free Software Foundation, Inc.
47 @c *** Converted to texinfo by Dean Wakerley, dean@wakerley.com
49 @c IMPORTANT: whenever you modify this file, run `install.texi2html' to
50 @c test the generation of HTML documents for the gcc.gnu.org web pages.
52 @c Do not use @footnote{} in this file as it breaks install.texi2html!
54 @c Include everything if we're not making html
58 @set prerequisiteshtml
69 @c Part 2 Summary Description and Copyright
71 Copyright @copyright{} 1988-2019 Free Software Foundation, Inc.
73 Permission is granted to copy, distribute and/or modify this document
74 under the terms of the GNU Free Documentation License, Version 1.3 or
75 any later version published by the Free Software Foundation; with no
76 Invariant Sections, the Front-Cover texts being (a) (see below), and
77 with the Back-Cover Texts being (b) (see below). A copy of the
78 license is included in the section entitled ``@uref{./gfdl.html,,GNU
79 Free Documentation License}''.
81 (a) The FSF's Front-Cover Text is:
85 (b) The FSF's Back-Cover Text is:
87 You have freedom to copy and modify this GNU Manual, like GNU
88 software. Copies published by the Free Software Foundation raise
89 funds for GNU development.
94 @dircategory Software development
96 * gccinstall: (gccinstall). Installing the GNU Compiler Collection.
99 @c Part 3 Titlepage and Copyright
101 @title Installing GCC
104 @c The following two commands start the copyright page.
106 @vskip 0pt plus 1filll
110 @c Part 4 Top node, Master Menu, and/or Table of Contents
113 @comment node-name, next, Previous, up
116 * Installing GCC:: This document describes the generic installation
117 procedure for GCC as well as detailing some target
118 specific installation instructions.
120 * Specific:: Host/target specific installation notes for GCC.
121 * Binaries:: Where to get pre-compiled binaries.
123 * Old:: Old installation documentation.
125 * GNU Free Documentation License:: How you can copy and share this manual.
126 * Concept Index:: This index has two entries.
134 @c Part 5 The Body of the Document
135 @c ***Installing GCC**********************************************************
137 @comment node-name, next, previous, up
138 @node Installing GCC, Binaries, , Top
142 @chapter Installing GCC
145 The latest version of this document is always available at
146 @uref{http://gcc.gnu.org/install/,,http://gcc.gnu.org/install/}.
147 It refers to the current development sources, instructions for
148 specific released versions are included with the sources.
150 This document describes the generic installation procedure for GCC as well
151 as detailing some target specific installation instructions.
153 GCC includes several components that previously were separate distributions
154 with their own installation instructions. This document supersedes all
155 package-specific installation instructions.
157 @emph{Before} starting the build/install procedure please check the
159 @ref{Specific, host/target specific installation notes}.
162 @uref{specific.html,,host/target specific installation notes}.
164 We recommend you browse the entire generic installation instructions before
167 Lists of successful builds for released versions of GCC are
168 available at @uref{http://gcc.gnu.org/buildstat.html}.
169 These lists are updated as new information becomes available.
171 The installation procedure itself is broken into five steps.
176 * Downloading the source::
179 * Testing:: (optional)
186 @uref{prerequisites.html,,Prerequisites}
188 @uref{download.html,,Downloading the source}
190 @uref{configure.html,,Configuration}
192 @uref{build.html,,Building}
194 @uref{test.html,,Testing} (optional)
196 @uref{finalinstall.html,,Final install}
200 Please note that GCC does not support @samp{make uninstall} and probably
201 won't do so in the near future as this would open a can of worms. Instead,
202 we suggest that you install GCC into a directory of its own and simply
203 remove that directory when you do not need that specific version of GCC
204 any longer, and, if shared libraries are installed there as well, no
205 more binaries exist that use them.
208 There are also some @uref{old.html,,old installation instructions},
209 which are mostly obsolete but still contain some information which has
210 not yet been merged into the main part of this manual.
218 @uref{./index.html,,Return to the GCC Installation page}
224 @c ***Prerequisites**************************************************
226 @comment node-name, next, previous, up
227 @node Prerequisites, Downloading the source, , Installing GCC
229 @ifset prerequisiteshtml
231 @chapter Prerequisites
233 @cindex Prerequisites
235 GCC requires that various tools and packages be available for use in the
236 build procedure. Modifying GCC sources requires additional tools
239 @heading Tools/packages necessary for building GCC
241 @item ISO C++98 compiler
242 Necessary to bootstrap GCC, although versions of GCC prior
243 to 4.8 also allow bootstrapping with a ISO C89 compiler and versions
244 of GCC prior to 3.4 also allow bootstrapping with a traditional
247 To build all languages in a cross-compiler or other configuration where
248 3-stage bootstrap is not performed, you need to start with an existing
249 GCC binary (version 3.4 or later) because source code for language
250 frontends other than C might use GCC extensions.
252 Note that to bootstrap GCC with versions of GCC earlier than 3.4, you
253 may need to use @option{--disable-stage1-checking}, though
254 bootstrapping the compiler with such earlier compilers is strongly
257 @item C standard library and headers
259 In order to build GCC, the C standard library and headers must be present
260 for all target variants for which target libraries will be built (and not
261 only the variant of the host C++ compiler).
263 This affects the popular @samp{x86_64-pc-linux-gnu} platform (among
264 other multilib targets), for which 64-bit (@samp{x86_64}) and 32-bit
265 (@samp{i386}) libc headers are usually packaged separately. If you do a
266 build of a native compiler on @samp{x86_64-pc-linux-gnu}, make sure you
267 either have the 32-bit libc developer package properly installed (the exact
268 name of the package depends on your distro) or you must build GCC as a
269 64-bit only compiler by configuring with the option
270 @option{--disable-multilib}. Otherwise, you may encounter an error such as
271 @samp{fatal error: gnu/stubs-32.h: No such file}
273 @item @anchor{GNAT-prerequisite}GNAT
275 In order to build GNAT, the Ada compiler, you need a working GNAT
276 compiler (GCC version 4.7 or later).
278 This includes GNAT tools such as @command{gnatmake} and
279 @command{gnatlink}, since the Ada front end is written in Ada and
280 uses some GNAT-specific extensions.
282 In order to build a cross compiler, it is strongly recommended to install
283 the new compiler as native first, and then use it to build the cross
284 compiler. Other native compiler versions may work but this is not guaranteed and
285 will typically fail with hard to understand compilation errors during the
288 Similarly, it is strongly recommended to use an older version of GNAT to build
289 GNAT. More recent versions of GNAT than the version built are not guaranteed
290 to work and will often fail during the build with compilation errors.
292 Note that @command{configure} does not test whether the GNAT installation works
293 and has a sufficiently recent version; if too old a GNAT version is
294 installed and @option{--enable-languages=ada} is used, the build will fail.
296 @env{ADA_INCLUDE_PATH} and @env{ADA_OBJECT_PATH} environment variables
297 must not be set when building the Ada compiler, the Ada tools, or the
298 Ada runtime libraries. You can check that your build environment is clean
299 by verifying that @samp{gnatls -v} lists only one explicit path in each
302 @item A ``working'' POSIX compatible shell, or GNU bash
304 Necessary when running @command{configure} because some
305 @command{/bin/sh} shells have bugs and may crash when configuring the
306 target libraries. In other cases, @command{/bin/sh} or @command{ksh}
307 have disastrous corner-case performance problems. This
308 can cause target @command{configure} runs to literally take days to
309 complete in some cases.
311 So on some platforms @command{/bin/ksh} is sufficient, on others it
312 isn't. See the host/target specific instructions for your platform, or
313 use @command{bash} to be sure. Then set @env{CONFIG_SHELL} in your
314 environment to your ``good'' shell prior to running
315 @command{configure}/@command{make}.
317 @command{zsh} is not a fully compliant POSIX shell and will not
318 work when configuring GCC@.
320 @item A POSIX or SVR4 awk
322 Necessary for creating some of the generated source files for GCC@.
323 If in doubt, use a recent GNU awk version, as some of the older ones
324 are broken. GNU awk version 3.1.5 is known to work.
328 Necessary in some circumstances, optional in others. See the
329 host/target specific instructions for your platform for the exact
332 @item gzip version 1.2.4 (or later) or
333 @itemx bzip2 version 1.0.2 (or later)
335 Necessary to uncompress GCC @command{tar} files when source code is
336 obtained via FTP mirror sites.
338 @item GNU make version 3.80 (or later)
340 You must have GNU make installed to build GCC@.
342 @item GNU tar version 1.14 (or later)
344 Necessary (only on some platforms) to untar the source code. Many
345 systems' @command{tar} programs will also work, only try GNU
346 @command{tar} if you have problems.
348 @item Perl version between 5.6.1 and 5.6.24
350 Necessary when targeting Darwin, building @samp{libstdc++},
351 and not using @option{--disable-symvers}.
352 Necessary when targeting Solaris 2 with Sun @command{ld} and not using
353 @option{--disable-symvers}. The bundled @command{perl} in Solaris@tie{}8
356 Necessary when regenerating @file{Makefile} dependencies in libiberty.
357 Necessary when regenerating @file{libiberty/functions.texi}.
358 Necessary when generating manpages from Texinfo manuals.
359 Used by various scripts to generate some files included in SVN (mainly
360 Unicode-related and rarely changing) from source tables.
362 Used by @command{automake}.
366 Several support libraries are necessary to build GCC, some are required,
367 others optional. While any sufficiently new version of required tools
368 usually work, library requirements are generally stricter. Newer
369 versions may work in some cases, but it's safer to use the exact
370 versions documented. We appreciate bug reports about problems with
371 newer versions, though. If your OS vendor provides packages for the
372 support libraries then using those packages may be the simplest way to
373 install the libraries.
376 @item GNU Multiple Precision Library (GMP) version 4.3.2 (or later)
378 Necessary to build GCC@. If a GMP source distribution is found in a
379 subdirectory of your GCC sources named @file{gmp}, it will be built
380 together with GCC. Alternatively, if GMP is already installed but it
381 is not in your library search path, you will have to configure with the
382 @option{--with-gmp} configure option. See also @option{--with-gmp-lib}
383 and @option{--with-gmp-include}.
384 The in-tree build is only supported with the GMP version that
385 download_prerequisites installs.
387 @item MPFR Library version 2.4.2 (or later)
389 Necessary to build GCC@. It can be downloaded from
390 @uref{https://www.mpfr.org}. If an MPFR source distribution is found
391 in a subdirectory of your GCC sources named @file{mpfr}, it will be
392 built together with GCC. Alternatively, if MPFR is already installed
393 but it is not in your default library search path, the
394 @option{--with-mpfr} configure option should be used. See also
395 @option{--with-mpfr-lib} and @option{--with-mpfr-include}.
396 The in-tree build is only supported with the MPFR version that
397 download_prerequisites installs.
399 @item MPC Library version 0.8.1 (or later)
401 Necessary to build GCC@. It can be downloaded from
402 @uref{http://www.multiprecision.org/mpc/}. If an MPC source distribution
403 is found in a subdirectory of your GCC sources named @file{mpc}, it
404 will be built together with GCC. Alternatively, if MPC is already
405 installed but it is not in your default library search path, the
406 @option{--with-mpc} configure option should be used. See also
407 @option{--with-mpc-lib} and @option{--with-mpc-include}.
408 The in-tree build is only supported with the MPC version that
409 download_prerequisites installs.
411 @item isl Library version 0.15 or later.
413 Necessary to build GCC with the Graphite loop optimizations.
414 It can be downloaded from @uref{ftp://gcc.gnu.org/pub/gcc/infrastructure/}.
415 If an isl source distribution is found
416 in a subdirectory of your GCC sources named @file{isl}, it will be
417 built together with GCC. Alternatively, the @option{--with-isl} configure
418 option should be used if isl is not installed in your default library
423 Necessary to build GCC with zstd compression used for LTO bytecode.
424 The library is searched in your default library patch search.
425 Alternatively, the @option{--with-zstd} configure option should be used.
429 @heading Tools/packages necessary for modifying GCC
431 @item autoconf version 2.69
432 @itemx GNU m4 version 1.4.6 (or later)
434 Necessary when modifying @file{configure.ac}, @file{aclocal.m4}, etc.@:
435 to regenerate @file{configure} and @file{config.in} files.
437 @item automake version 1.15.1
439 Necessary when modifying a @file{Makefile.am} file to regenerate its
440 associated @file{Makefile.in}.
442 Much of GCC does not use automake, so directly edit the @file{Makefile.in}
443 file. Specifically this applies to the @file{gcc}, @file{intl},
444 @file{libcpp}, @file{libiberty}, @file{libobjc} directories as well
445 as any of their subdirectories.
447 For directories that use automake, GCC requires the latest release in
448 the 1.15 series, which is currently 1.15.1. When regenerating a directory
449 to a newer version, please update all the directories using an older 1.15
450 to the latest released version.
452 @item gettext version 0.14.5 (or later)
454 Needed to regenerate @file{gcc.pot}.
456 @item gperf version 2.7.2 (or later)
458 Necessary when modifying @command{gperf} input files, e.g.@:
459 @file{gcc/cp/cfns.gperf} to regenerate its associated header file, e.g.@:
460 @file{gcc/cp/cfns.h}.
466 Necessary to run the GCC testsuite; see the section on testing for
469 @item autogen version 5.5.4 (or later) and
470 @itemx guile version 1.4.1 (or later)
472 Necessary to regenerate @file{fixinc/fixincl.x} from
473 @file{fixinc/inclhack.def} and @file{fixinc/*.tpl}.
475 Necessary to run @samp{make check} for @file{fixinc}.
477 Necessary to regenerate the top level @file{Makefile.in} file from
478 @file{Makefile.tpl} and @file{Makefile.def}.
480 @item Flex version 2.5.4 (or later)
482 Necessary when modifying @file{*.l} files.
484 Necessary to build GCC during development because the generated output
485 files are not included in the SVN repository. They are included in
488 @item Texinfo version 4.7 (or later)
490 Necessary for running @command{makeinfo} when modifying @file{*.texi}
491 files to test your changes.
493 Necessary for running @command{make dvi} or @command{make pdf} to
494 create printable documentation in DVI or PDF format. Texinfo version
495 4.8 or later is required for @command{make pdf}.
497 Necessary to build GCC documentation during development because the
498 generated output files are not included in the SVN repository. They are
499 included in releases.
501 @item @TeX{} (any working version)
503 Necessary for running @command{texi2dvi} and @command{texi2pdf}, which
504 are used when running @command{make dvi} or @command{make pdf} to create
505 DVI or PDF files, respectively.
507 @item Sphinx version 1.0 (or later)
509 Necessary to regenerate @file{jit/docs/_build/texinfo} from the @file{.rst}
510 files in the directories below @file{jit/docs}.
512 @item SVN (any version)
513 @itemx SSH (any version)
515 Necessary to access the SVN repository. Public releases and weekly
516 snapshots of the development sources are also available via FTP@.
518 @item GNU diffutils version 2.7 (or later)
520 Useful when submitting patches for the GCC source code.
522 @item patch version 2.5.4 (or later)
524 Necessary when applying patches, created with @command{diff}, to one's
534 @uref{./index.html,,Return to the GCC Installation page}
538 @c ***Downloading the source**************************************************
540 @comment node-name, next, previous, up
541 @node Downloading the source, Configuration, Prerequisites, Installing GCC
545 @chapter Downloading GCC
547 @cindex Downloading GCC
548 @cindex Downloading the Source
550 GCC is distributed via @uref{http://gcc.gnu.org/svn.html,,SVN} and FTP
551 tarballs compressed with @command{gzip} or
554 Please refer to the @uref{http://gcc.gnu.org/releases.html,,releases web page}
555 for information on how to obtain GCC@.
557 The source distribution includes the C, C++, Objective-C, Fortran,
558 and Ada (in the case of GCC 3.1 and later) compilers, as well as
559 runtime libraries for C++, Objective-C, and Fortran.
560 For previous versions these were downloadable as separate components such
561 as the core GCC distribution, which included the C language front end and
562 shared components, and language-specific distributions including the
563 language front end and the language runtime (where appropriate).
565 If you also intend to build binutils (either to upgrade an existing
566 installation or for use in place of the corresponding tools of your
567 OS), unpack the binutils distribution either in the same directory or
568 a separate one. In the latter case, add symbolic links to any
569 components of the binutils you intend to build alongside the compiler
570 (@file{bfd}, @file{binutils}, @file{gas}, @file{gprof}, @file{ld},
571 @file{opcodes}, @dots{}) to the directory containing the GCC sources.
573 Likewise the GMP, MPFR and MPC libraries can be automatically built
574 together with GCC. You may simply run the
575 @command{contrib/download_prerequisites} script in the GCC source directory
576 to set up everything.
577 Otherwise unpack the GMP, MPFR and/or MPC source
578 distributions in the directory containing the GCC sources and rename
579 their directories to @file{gmp}, @file{mpfr} and @file{mpc},
580 respectively (or use symbolic links with the same name).
587 @uref{./index.html,,Return to the GCC Installation page}
591 @c ***Configuration***********************************************************
593 @comment node-name, next, previous, up
594 @node Configuration, Building, Downloading the source, Installing GCC
598 @chapter Installing GCC: Configuration
600 @cindex Configuration
601 @cindex Installing GCC: Configuration
603 Like most GNU software, GCC must be configured before it can be built.
604 This document describes the recommended configuration procedure
605 for both native and cross targets.
607 We use @var{srcdir} to refer to the toplevel source directory for
608 GCC; we use @var{objdir} to refer to the toplevel build/object directory.
610 If you obtained the sources via SVN, @var{srcdir} must refer to the top
611 @file{gcc} directory, the one where the @file{MAINTAINERS} file can be
612 found, and not its @file{gcc} subdirectory, otherwise the build will fail.
614 If either @var{srcdir} or @var{objdir} is located on an automounted NFS
615 file system, the shell's built-in @command{pwd} command will return
616 temporary pathnames. Using these can lead to various sorts of build
617 problems. To avoid this issue, set the @env{PWDCMD} environment
618 variable to an automounter-aware @command{pwd} command, e.g.,
619 @command{pawd} or @samp{amq -w}, during the configuration and build
622 First, we @strong{highly} recommend that GCC be built into a
623 separate directory from the sources which does @strong{not} reside
624 within the source tree. This is how we generally build GCC; building
625 where @var{srcdir} == @var{objdir} should still work, but doesn't
626 get extensive testing; building where @var{objdir} is a subdirectory
627 of @var{srcdir} is unsupported.
629 If you have previously built GCC in the same directory for a
630 different target machine, do @samp{make distclean} to delete all files
631 that might be invalid. One of the files this deletes is @file{Makefile};
632 if @samp{make distclean} complains that @file{Makefile} does not exist
633 or issues a message like ``don't know how to make distclean'' it probably
634 means that the directory is already suitably clean. However, with the
635 recommended method of building in a separate @var{objdir}, you should
636 simply use a different @var{objdir} for each target.
638 Second, when configuring a native system, either @command{cc} or
639 @command{gcc} must be in your path or you must set @env{CC} in
640 your environment before running configure. Otherwise the configuration
644 Note that the bootstrap compiler and the resulting GCC must be link
645 compatible, else the bootstrap will fail with linker errors about
646 incompatible object file formats. Several multilibed targets are
647 affected by this requirement, see
649 @ref{Specific, host/target specific installation notes}.
652 @uref{specific.html,,host/target specific installation notes}.
661 % @var{srcdir}/configure [@var{options}] [@var{target}]
664 @heading Distributor options
666 If you will be distributing binary versions of GCC, with modifications
667 to the source code, you should use the options described in this
668 section to make clear that your version contains modifications.
671 @item --with-pkgversion=@var{version}
672 Specify a string that identifies your package. You may wish
673 to include a build number or build date. This version string will be
674 included in the output of @command{gcc --version}. This suffix does
675 not replace the default version string, only the @samp{GCC} part.
677 The default value is @samp{GCC}.
679 @item --with-bugurl=@var{url}
680 Specify the URL that users should visit if they wish to report a bug.
681 You are of course welcome to forward bugs reported to you to the FSF,
682 if you determine that they are not bugs in your modifications.
684 The default value refers to the FSF's GCC bug tracker.
688 @heading Target specification
691 GCC has code to correctly determine the correct value for @var{target}
692 for nearly all native systems. Therefore, we highly recommend you do
693 not provide a configure target when configuring a native compiler.
696 @var{target} must be specified as @option{--target=@var{target}}
697 when configuring a cross compiler; examples of valid targets would be
698 m68k-elf, sh-elf, etc.
701 Specifying just @var{target} instead of @option{--target=@var{target}}
702 implies that the host defaults to @var{target}.
706 @heading Options specification
708 Use @var{options} to override several configure time options for
709 GCC@. A list of supported @var{options} follows; @samp{configure
710 --help} may list other options, but those not listed below may not
711 work and should not normally be used.
713 Note that each @option{--enable} option has a corresponding
714 @option{--disable} option and that each @option{--with} option has a
715 corresponding @option{--without} option.
718 @item --prefix=@var{dirname}
719 Specify the toplevel installation
720 directory. This is the recommended way to install the tools into a directory
721 other than the default. The toplevel installation directory defaults to
724 We @strong{highly} recommend against @var{dirname} being the same or a
725 subdirectory of @var{objdir} or vice versa. If specifying a directory
726 beneath a user's home directory tree, some shells will not expand
727 @var{dirname} correctly if it contains the @samp{~} metacharacter; use
730 The following standard @command{autoconf} options are supported. Normally you
731 should not need to use these options.
733 @item --exec-prefix=@var{dirname}
734 Specify the toplevel installation directory for architecture-dependent
735 files. The default is @file{@var{prefix}}.
737 @item --bindir=@var{dirname}
738 Specify the installation directory for the executables called by users
739 (such as @command{gcc} and @command{g++}). The default is
740 @file{@var{exec-prefix}/bin}.
742 @item --libdir=@var{dirname}
743 Specify the installation directory for object code libraries and
744 internal data files of GCC@. The default is @file{@var{exec-prefix}/lib}.
746 @item --libexecdir=@var{dirname}
747 Specify the installation directory for internal executables of GCC@.
748 The default is @file{@var{exec-prefix}/libexec}.
750 @item --with-slibdir=@var{dirname}
751 Specify the installation directory for the shared libgcc library. The
752 default is @file{@var{libdir}}.
754 @item --datarootdir=@var{dirname}
755 Specify the root of the directory tree for read-only architecture-independent
756 data files referenced by GCC@. The default is @file{@var{prefix}/share}.
758 @item --infodir=@var{dirname}
759 Specify the installation directory for documentation in info format.
760 The default is @file{@var{datarootdir}/info}.
762 @item --datadir=@var{dirname}
763 Specify the installation directory for some architecture-independent
764 data files referenced by GCC@. The default is @file{@var{datarootdir}}.
766 @item --docdir=@var{dirname}
767 Specify the installation directory for documentation files (other
768 than Info) for GCC@. The default is @file{@var{datarootdir}/doc}.
770 @item --htmldir=@var{dirname}
771 Specify the installation directory for HTML documentation files.
772 The default is @file{@var{docdir}}.
774 @item --pdfdir=@var{dirname}
775 Specify the installation directory for PDF documentation files.
776 The default is @file{@var{docdir}}.
778 @item --mandir=@var{dirname}
779 Specify the installation directory for manual pages. The default is
780 @file{@var{datarootdir}/man}. (Note that the manual pages are only extracts
781 from the full GCC manuals, which are provided in Texinfo format. The manpages
782 are derived by an automatic conversion process from parts of the full
785 @item --with-gxx-include-dir=@var{dirname}
787 the installation directory for G++ header files. The default depends
788 on other configuration options, and differs between cross and native
791 @item --with-specs=@var{specs}
792 Specify additional command line driver SPECS.
793 This can be useful if you need to turn on a non-standard feature by
794 default without modifying the compiler's source code, for instance
795 @option{--with-specs=%@{!fcommon:%@{!fno-common:-fno-common@}@}}.
797 @xref{Spec Files,, Specifying subprocesses and the switches to pass to them,
798 gcc, Using the GNU Compiler Collection (GCC)},
801 See ``Spec Files'' in the main manual
806 @item --program-prefix=@var{prefix}
807 GCC supports some transformations of the names of its programs when
808 installing them. This option prepends @var{prefix} to the names of
809 programs to install in @var{bindir} (see above). For example, specifying
810 @option{--program-prefix=foo-} would result in @samp{gcc}
811 being installed as @file{/usr/local/bin/foo-gcc}.
813 @item --program-suffix=@var{suffix}
814 Appends @var{suffix} to the names of programs to install in @var{bindir}
815 (see above). For example, specifying @option{--program-suffix=-3.1}
816 would result in @samp{gcc} being installed as
817 @file{/usr/local/bin/gcc-3.1}.
819 @item --program-transform-name=@var{pattern}
820 Applies the @samp{sed} script @var{pattern} to be applied to the names
821 of programs to install in @var{bindir} (see above). @var{pattern} has to
822 consist of one or more basic @samp{sed} editing commands, separated by
823 semicolons. For example, if you want the @samp{gcc} program name to be
824 transformed to the installed program @file{/usr/local/bin/myowngcc} and
825 the @samp{g++} program name to be transformed to
826 @file{/usr/local/bin/gspecial++} without changing other program names,
827 you could use the pattern
828 @option{--program-transform-name='s/^gcc$/myowngcc/; s/^g++$/gspecial++/'}
829 to achieve this effect.
831 All three options can be combined and used together, resulting in more
832 complex conversion patterns. As a basic rule, @var{prefix} (and
833 @var{suffix}) are prepended (appended) before further transformations
834 can happen with a special transformation script @var{pattern}.
836 As currently implemented, this option only takes effect for native
837 builds; cross compiler binaries' names are not transformed even when a
838 transformation is explicitly asked for by one of these options.
840 For native builds, some of the installed programs are also installed
841 with the target alias in front of their name, as in
842 @samp{i686-pc-linux-gnu-gcc}. All of the above transformations happen
843 before the target alias is prepended to the name---so, specifying
844 @option{--program-prefix=foo-} and @option{program-suffix=-3.1}, the
845 resulting binary would be installed as
846 @file{/usr/local/bin/i686-pc-linux-gnu-foo-gcc-3.1}.
848 As a last shortcoming, none of the installed Ada programs are
849 transformed yet, which will be fixed in some time.
851 @item --with-local-prefix=@var{dirname}
853 installation directory for local include files. The default is
854 @file{/usr/local}. Specify this option if you want the compiler to
855 search directory @file{@var{dirname}/include} for locally installed
856 header files @emph{instead} of @file{/usr/local/include}.
858 You should specify @option{--with-local-prefix} @strong{only} if your
859 site has a different convention (not @file{/usr/local}) for where to put
862 The default value for @option{--with-local-prefix} is @file{/usr/local}
863 regardless of the value of @option{--prefix}. Specifying
864 @option{--prefix} has no effect on which directory GCC searches for
865 local header files. This may seem counterintuitive, but actually it is
868 The purpose of @option{--prefix} is to specify where to @emph{install
869 GCC}. The local header files in @file{/usr/local/include}---if you put
870 any in that directory---are not part of GCC@. They are part of other
871 programs---perhaps many others. (GCC installs its own header files in
872 another directory which is based on the @option{--prefix} value.)
874 Both the local-prefix include directory and the GCC-prefix include
875 directory are part of GCC's ``system include'' directories. Although these
876 two directories are not fixed, they need to be searched in the proper
877 order for the correct processing of the include_next directive. The
878 local-prefix include directory is searched before the GCC-prefix
879 include directory. Another characteristic of system include directories
880 is that pedantic warnings are turned off for headers in these directories.
882 Some autoconf macros add @option{-I @var{directory}} options to the
883 compiler command line, to ensure that directories containing installed
884 packages' headers are searched. When @var{directory} is one of GCC's
885 system include directories, GCC will ignore the option so that system
886 directories continue to be processed in the correct order. This
887 may result in a search order different from what was specified but the
888 directory will still be searched.
890 GCC automatically searches for ordinary libraries using
891 @env{GCC_EXEC_PREFIX}. Thus, when the same installation prefix is
892 used for both GCC and packages, GCC will automatically search for
893 both headers and libraries. This provides a configuration that is
894 easy to use. GCC behaves in a manner similar to that when it is
895 installed as a system compiler in @file{/usr}.
897 Sites that need to install multiple versions of GCC may not want to
898 use the above simple configuration. It is possible to use the
899 @option{--program-prefix}, @option{--program-suffix} and
900 @option{--program-transform-name} options to install multiple versions
901 into a single directory, but it may be simpler to use different prefixes
902 and the @option{--with-local-prefix} option to specify the location of the
903 site-specific files for each version. It will then be necessary for
904 users to specify explicitly the location of local site libraries
905 (e.g., with @env{LIBRARY_PATH}).
907 The same value can be used for both @option{--with-local-prefix} and
908 @option{--prefix} provided it is not @file{/usr}. This can be used
909 to avoid the default search of @file{/usr/local/include}.
911 @strong{Do not} specify @file{/usr} as the @option{--with-local-prefix}!
912 The directory you use for @option{--with-local-prefix} @strong{must not}
913 contain any of the system's standard header files. If it did contain
914 them, certain programs would be miscompiled (including GNU Emacs, on
915 certain targets), because this would override and nullify the header
916 file corrections made by the @command{fixincludes} script.
918 Indications are that people who use this option use it based on mistaken
919 ideas of what it is for. People use it as if it specified where to
920 install part of GCC@. Perhaps they make this assumption because
921 installing GCC creates the directory.
923 @item --with-gcc-major-version-only
924 Specifies that GCC should use only the major number rather than
925 @var{major}.@var{minor}.@var{patchlevel} in filesystem paths.
927 @item --with-native-system-header-dir=@var{dirname}
928 Specifies that @var{dirname} is the directory that contains native system
929 header files, rather than @file{/usr/include}. This option is most useful
930 if you are creating a compiler that should be isolated from the system
931 as much as possible. It is most commonly used with the
932 @option{--with-sysroot} option and will cause GCC to search
933 @var{dirname} inside the system root specified by that option.
935 @item --enable-shared[=@var{package}[,@dots{}]]
936 Build shared versions of libraries, if shared libraries are supported on
937 the target platform. Unlike GCC 2.95.x and earlier, shared libraries
938 are enabled by default on all platforms that support shared libraries.
940 If a list of packages is given as an argument, build shared libraries
941 only for the listed packages. For other packages, only static libraries
942 will be built. Package names currently recognized in the GCC tree are
943 @samp{libgcc} (also known as @samp{gcc}), @samp{libstdc++} (not
944 @samp{libstdc++-v3}), @samp{libffi}, @samp{zlib}, @samp{boehm-gc},
945 @samp{ada}, @samp{libada}, @samp{libgo}, @samp{libobjc}, and @samp{libphobos}.
946 Note @samp{libiberty} does not support shared libraries at all.
948 Use @option{--disable-shared} to build only static libraries. Note that
949 @option{--disable-shared} does not accept a list of package names as
950 argument, only @option{--enable-shared} does.
952 Contrast with @option{--enable-host-shared}, which affects @emph{host}
955 @item --enable-host-shared
956 Specify that the @emph{host} code should be built into position-independent
957 machine code (with -fPIC), allowing it to be used within shared libraries,
958 but yielding a slightly slower compiler.
960 This option is required when building the libgccjit.so library.
962 Contrast with @option{--enable-shared}, which affects @emph{target}
965 @item @anchor{with-gnu-as}--with-gnu-as
966 Specify that the compiler should assume that the
967 assembler it finds is the GNU assembler. However, this does not modify
968 the rules to find an assembler and will result in confusion if the
969 assembler found is not actually the GNU assembler. (Confusion may also
970 result if the compiler finds the GNU assembler but has not been
971 configured with @option{--with-gnu-as}.) If you have more than one
972 assembler installed on your system, you may want to use this option in
973 connection with @option{--with-as=@var{pathname}} or
974 @option{--with-build-time-tools=@var{pathname}}.
976 The following systems are the only ones where it makes a difference
977 whether you use the GNU assembler. On any other system,
978 @option{--with-gnu-as} has no effect.
981 @item @samp{hppa1.0-@var{any}-@var{any}}
982 @item @samp{hppa1.1-@var{any}-@var{any}}
983 @item @samp{sparc-sun-solaris2.@var{any}}
984 @item @samp{sparc64-@var{any}-solaris2.@var{any}}
987 @item @anchor{with-as}--with-as=@var{pathname}
988 Specify that the compiler should use the assembler pointed to by
989 @var{pathname}, rather than the one found by the standard rules to find
990 an assembler, which are:
993 Unless GCC is being built with a cross compiler, check the
994 @file{@var{libexec}/gcc/@var{target}/@var{version}} directory.
995 @var{libexec} defaults to @file{@var{exec-prefix}/libexec};
996 @var{exec-prefix} defaults to @var{prefix}, which
997 defaults to @file{/usr/local} unless overridden by the
998 @option{--prefix=@var{pathname}} switch described above. @var{target}
999 is the target system triple, such as @samp{sparc-sun-solaris2.7}, and
1000 @var{version} denotes the GCC version, such as 3.0.
1003 If the target system is the same that you are building on, check
1004 operating system specific directories (e.g.@: @file{/usr/ccs/bin} on
1008 Check in the @env{PATH} for a tool whose name is prefixed by the
1009 target system triple.
1012 Check in the @env{PATH} for a tool whose name is not prefixed by the
1013 target system triple, if the host and target system triple are
1014 the same (in other words, we use a host tool if it can be used for
1015 the target as well).
1018 You may want to use @option{--with-as} if no assembler
1019 is installed in the directories listed above, or if you have multiple
1020 assemblers installed and want to choose one that is not found by the
1023 @item @anchor{with-gnu-ld}--with-gnu-ld
1024 Same as @uref{#with-gnu-as,,@option{--with-gnu-as}}
1027 @item --with-ld=@var{pathname}
1028 Same as @uref{#with-as,,@option{--with-as}}
1032 Specify that stabs debugging
1033 information should be used instead of whatever format the host normally
1034 uses. Normally GCC uses the same debug format as the host system.
1036 @item --with-tls=@var{dialect}
1037 Specify the default TLS dialect, for systems were there is a choice.
1038 For ARM targets, possible values for @var{dialect} are @code{gnu} or
1039 @code{gnu2}, which select between the original GNU dialect and the GNU TLS
1040 descriptor-based dialect.
1042 @item --enable-multiarch
1043 Specify whether to enable or disable multiarch support. The default is
1044 to check for glibc start files in a multiarch location, and enable it
1045 if the files are found. The auto detection is enabled for native builds,
1046 and for cross builds configured with @option{--with-sysroot}, and without
1047 @option{--with-native-system-header-dir}.
1048 More documentation about multiarch can be found at
1049 @uref{https://wiki.debian.org/Multiarch}.
1051 @item --enable-sjlj-exceptions
1052 Force use of the @code{setjmp}/@code{longjmp}-based scheme for exceptions.
1053 @samp{configure} ordinarily picks the correct value based on the platform.
1054 Only use this option if you are sure you need a different setting.
1056 @item --enable-vtable-verify
1057 Specify whether to enable or disable the vtable verification feature.
1058 Enabling this feature causes libstdc++ to be built with its virtual calls
1059 in verifiable mode. This means that, when linked with libvtv, every
1060 virtual call in libstdc++ will verify the vtable pointer through which the
1061 call will be made before actually making the call. If not linked with libvtv,
1062 the verifier will call stub functions (in libstdc++ itself) and do nothing.
1063 If vtable verification is disabled, then libstdc++ is not built with its
1064 virtual calls in verifiable mode at all. However the libvtv library will
1065 still be built (see @option{--disable-libvtv} to turn off building libvtv).
1066 @option{--disable-vtable-verify} is the default.
1068 @item --disable-gcov
1069 Specify that the run-time library used for coverage analysis
1070 and associated host tools should not be built.
1072 @item --disable-multilib
1073 Specify that multiple target
1074 libraries to support different target variants, calling
1075 conventions, etc.@: should not be built. The default is to build a
1076 predefined set of them.
1078 Some targets provide finer-grained control over which multilibs are built
1079 (e.g., @option{--disable-softfloat}):
1082 fpu, 26bit, underscore, interwork, biendian, nofmult.
1085 softfloat, m68881, m68000, m68020.
1088 single-float, biendian, softfloat.
1090 @item powerpc*-*-*, rs6000*-*-*
1091 aix64, pthread, softfloat, powercpu, powerpccpu, powerpcos, biendian,
1096 @item --with-multilib-list=@var{list}
1097 @itemx --without-multilib-list
1098 Specify what multilibs to build. @var{list} is a comma separated list of
1099 values, possibly consisting of a single value. Currently only implemented
1100 for aarch64*-*-*, arm*-*-*, riscv*-*-*, sh*-*-* and x86-64-*-linux*. The
1101 accepted values and meaning for each target is given below.
1105 @var{list} is a comma separated list of @code{ilp32}, and @code{lp64}
1106 to enable ILP32 and LP64 run-time libraries, respectively. If
1107 @var{list} is empty, then there will be no multilibs and only the
1108 default run-time library will be built. If @var{list} is
1109 @code{default} or --with-multilib-list= is not specified, then the
1110 default set of libraries is selected based on the value of
1114 @var{list} is a comma separated list of @code{aprofile} and
1115 @code{rmprofile} to build multilibs for A or R and M architecture
1116 profiles respectively. Note that, due to some limitation of the current
1117 multilib framework, using the combined @code{aprofile,rmprofile}
1118 multilibs selects in some cases a less optimal multilib than when using
1119 the multilib profile for the architecture targetted. The special value
1120 @code{default} is also accepted and is equivalent to omitting the
1121 option, i.e., only the default run-time library will be enabled.
1123 @var{list} may instead contain @code{@@name}, to use the multilib
1124 configuration Makefile fragment @file{name} in @file{gcc/config/arm} in
1125 the source tree (it is part of the corresponding sources, after all).
1126 It is recommended, but not required, that files used for this purpose to
1127 be named starting with @file{t-ml-}, to make their intended purpose
1128 self-evident, in line with GCC conventions. Such files enable custom,
1129 user-chosen multilib lists to be configured. Whether multiple such
1130 files can be used together depends on the contents of the supplied
1131 files. See @file{gcc/config/arm/t-multilib} and its supplementary
1132 @file{gcc/config/arm/t-*profile} files for an example of what such
1133 Makefile fragments might look like for this version of GCC. The macros
1134 expected to be defined in these fragments are not stable across GCC
1135 releases, so make sure they define the @code{MULTILIB}-related macros
1136 expected by the version of GCC you are building.
1138 @xref{Target Fragment,, Target Makefile Fragments, gccint, GNU Compiler
1139 Collection (GCC) Internals}.
1142 See ``Target Makefile Fragments'' in the internals manual.
1145 The table below gives the combination of ISAs, architectures, FPUs and
1146 floating-point ABIs for which multilibs are built for each predefined
1147 profile. The union of these options is considered when specifying both
1148 @code{aprofile} and @code{rmprofile}.
1150 @multitable @columnfractions .15 .28 .30
1151 @item Option @tab aprofile @tab rmprofile
1153 @tab @code{-marm} and @code{-mthumb}
1155 @item Architectures@*@*@*@*@*@*
1156 @tab default architecture@*
1157 @code{-march=armv7-a}@*
1158 @code{-march=armv7ve}@*
1159 @code{-march=armv8-a}@*@*@*
1160 @tab default architecture@*
1161 @code{-march=armv6s-m}@*
1162 @code{-march=armv7-m}@*
1163 @code{-march=armv7e-m}@*
1164 @code{-march=armv8-m.base}@*
1165 @code{-march=armv8-m.main}@*
1167 @item FPUs@*@*@*@*@*
1169 @code{-mfpu=vfpv3-d16}@*
1171 @code{-mfpu=vfpv4-d16}@*
1172 @code{-mfpu=neon-vfpv4}@*
1173 @code{-mfpu=neon-fp-armv8}
1175 @code{-mfpu=vfpv3-d16}@*
1176 @code{-mfpu=fpv4-sp-d16}@*
1177 @code{-mfpu=fpv5-sp-d16}@*
1178 @code{-mfpu=fpv5-d16}@*
1179 @item floating-point@/ ABIs@*@*
1180 @tab @code{-mfloat-abi=soft}@*
1181 @code{-mfloat-abi=softfp}@*
1182 @code{-mfloat-abi=hard}
1183 @tab @code{-mfloat-abi=soft}@*
1184 @code{-mfloat-abi=softfp}@*
1185 @code{-mfloat-abi=hard}
1189 @var{list} is a single ABI name. The target architecture must be either
1190 @code{rv32gc} or @code{rv64gc}. This will build a single multilib for the
1191 specified architecture and ABI pair. If @code{--with-multilib-list} is not
1192 given, then a default set of multilibs is selected based on the value of
1193 @option{--target}. This is usually a large set of multilibs.
1196 @var{list} is a comma separated list of CPU names. These must be of the
1197 form @code{sh*} or @code{m*} (in which case they match the compiler option
1198 for that processor). The list should not contain any endian options -
1199 these are handled by @option{--with-endian}.
1201 If @var{list} is empty, then there will be no multilibs for extra
1202 processors. The multilib for the secondary endian remains enabled.
1204 As a special case, if an entry in the list starts with a @code{!}
1205 (exclamation point), then it is added to the list of excluded multilibs.
1206 Entries of this sort should be compatible with @samp{MULTILIB_EXCLUDES}
1207 (once the leading @code{!} has been stripped).
1209 If @option{--with-multilib-list} is not given, then a default set of
1210 multilibs is selected based on the value of @option{--target}. This is
1211 usually the complete set of libraries, but some targets imply a more
1214 Example 1: to configure a compiler for SH4A only, but supporting both
1215 endians, with little endian being the default:
1217 --with-cpu=sh4a --with-endian=little,big --with-multilib-list=
1220 Example 2: to configure a compiler for both SH4A and SH4AL-DSP, but with
1221 only little endian SH4AL:
1223 --with-cpu=sh4a --with-endian=little,big \
1224 --with-multilib-list=sh4al,!mb/m4al
1227 @item x86-64-*-linux*
1228 @var{list} is a comma separated list of @code{m32}, @code{m64} and
1229 @code{mx32} to enable 32-bit, 64-bit and x32 run-time libraries,
1230 respectively. If @var{list} is empty, then there will be no multilibs
1231 and only the default run-time library will be enabled.
1233 If @option{--with-multilib-list} is not given, then only 32-bit and
1234 64-bit run-time libraries will be enabled.
1237 @item --with-endian=@var{endians}
1238 Specify what endians to use.
1239 Currently only implemented for sh*-*-*.
1241 @var{endians} may be one of the following:
1244 Use big endian exclusively.
1246 Use little endian exclusively.
1248 Use big endian by default. Provide a multilib for little endian.
1250 Use little endian by default. Provide a multilib for big endian.
1253 @item --enable-threads
1254 Specify that the target
1255 supports threads. This affects the Objective-C compiler and runtime
1256 library, and exception handling for other languages like C++.
1257 On some systems, this is the default.
1259 In general, the best (and, in many cases, the only known) threading
1260 model available will be configured for use. Beware that on some
1261 systems, GCC has not been taught what threading models are generally
1262 available for the system. In this case, @option{--enable-threads} is an
1263 alias for @option{--enable-threads=single}.
1265 @item --disable-threads
1266 Specify that threading support should be disabled for the system.
1267 This is an alias for @option{--enable-threads=single}.
1269 @item --enable-threads=@var{lib}
1271 @var{lib} is the thread support library. This affects the Objective-C
1272 compiler and runtime library, and exception handling for other languages
1273 like C++. The possibilities for @var{lib} are:
1281 LynxOS thread support.
1283 MIPS SDE thread support.
1285 This is an alias for @samp{single}.
1287 Generic POSIX/Unix98 thread support.
1289 RTEMS thread support.
1291 Disable thread support, should work for all platforms.
1295 VxWorks thread support.
1297 Microsoft Win32 API thread support.
1301 Specify that the target supports TLS (Thread Local Storage). Usually
1302 configure can correctly determine if TLS is supported. In cases where
1303 it guesses incorrectly, TLS can be explicitly enabled or disabled with
1304 @option{--enable-tls} or @option{--disable-tls}. This can happen if
1305 the assembler supports TLS but the C library does not, or if the
1306 assumptions made by the configure test are incorrect.
1309 Specify that the target does not support TLS.
1310 This is an alias for @option{--enable-tls=no}.
1312 @item --disable-tm-clone-registry
1313 Disable TM clone registry in libgcc. It is enabled in libgcc by default.
1314 This option helps to reduce code size for embedded targets which do
1315 not use transactional memory.
1317 @item --with-cpu=@var{cpu}
1318 @itemx --with-cpu-32=@var{cpu}
1319 @itemx --with-cpu-64=@var{cpu}
1320 Specify which cpu variant the compiler should generate code for by default.
1321 @var{cpu} will be used as the default value of the @option{-mcpu=} switch.
1322 This option is only supported on some targets, including ARC, ARM, i386, M68k,
1323 PowerPC, and SPARC@. It is mandatory for ARC@. The @option{--with-cpu-32} and
1324 @option{--with-cpu-64} options specify separate default CPUs for
1325 32-bit and 64-bit modes; these options are only supported for i386,
1326 x86-64, PowerPC, and SPARC@.
1328 @item --with-schedule=@var{cpu}
1329 @itemx --with-arch=@var{cpu}
1330 @itemx --with-arch-32=@var{cpu}
1331 @itemx --with-arch-64=@var{cpu}
1332 @itemx --with-tune=@var{cpu}
1333 @itemx --with-tune-32=@var{cpu}
1334 @itemx --with-tune-64=@var{cpu}
1335 @itemx --with-abi=@var{abi}
1336 @itemx --with-fpu=@var{type}
1337 @itemx --with-float=@var{type}
1338 These configure options provide default values for the @option{-mschedule=},
1339 @option{-march=}, @option{-mtune=}, @option{-mabi=}, and @option{-mfpu=}
1340 options and for @option{-mhard-float} or @option{-msoft-float}. As with
1341 @option{--with-cpu}, which switches will be accepted and acceptable values
1342 of the arguments depend on the target.
1344 @item --with-mode=@var{mode}
1345 Specify if the compiler should default to @option{-marm} or @option{-mthumb}.
1346 This option is only supported on ARM targets.
1348 @item --with-stack-offset=@var{num}
1349 This option sets the default for the -mstack-offset=@var{num} option,
1350 and will thus generally also control the setting of this option for
1351 libraries. This option is only supported on Epiphany targets.
1353 @item --with-fpmath=@var{isa}
1354 This options sets @option{-mfpmath=sse} by default and specifies the default
1355 ISA for floating-point arithmetics. You can select either @samp{sse} which
1356 enables @option{-msse2} or @samp{avx} which enables @option{-mavx} by default.
1357 This option is only supported on i386 and x86-64 targets.
1359 @item --with-fp-32=@var{mode}
1360 On MIPS targets, set the default value for the @option{-mfp} option when using
1361 the o32 ABI. The possibilities for @var{mode} are:
1364 Use the o32 FP32 ABI extension, as with the @option{-mfp32} command-line
1367 Use the o32 FPXX ABI extension, as with the @option{-mfpxx} command-line
1370 Use the o32 FP64 ABI extension, as with the @option{-mfp64} command-line
1373 In the absence of this configuration option the default is to use the o32
1376 @item --with-odd-spreg-32
1377 On MIPS targets, set the @option{-modd-spreg} option by default when using
1380 @item --without-odd-spreg-32
1381 On MIPS targets, set the @option{-mno-odd-spreg} option by default when using
1382 the o32 ABI. This is normally used in conjunction with
1383 @option{--with-fp-32=64} in order to target the o32 FP64A ABI extension.
1385 @item --with-nan=@var{encoding}
1386 On MIPS targets, set the default encoding convention to use for the
1387 special not-a-number (NaN) IEEE 754 floating-point data. The
1388 possibilities for @var{encoding} are:
1391 Use the legacy encoding, as with the @option{-mnan=legacy} command-line
1394 Use the 754-2008 encoding, as with the @option{-mnan=2008} command-line
1397 To use this configuration option you must have an assembler version
1398 installed that supports the @option{-mnan=} command-line option too.
1399 In the absence of this configuration option the default convention is
1400 the legacy encoding, as when neither of the @option{-mnan=2008} and
1401 @option{-mnan=legacy} command-line options has been used.
1403 @item --with-divide=@var{type}
1404 Specify how the compiler should generate code for checking for
1405 division by zero. This option is only supported on the MIPS target.
1406 The possibilities for @var{type} are:
1409 Division by zero checks use conditional traps (this is the default on
1410 systems that support conditional traps).
1412 Division by zero checks use the break instruction.
1415 @c If you make --with-llsc the default for additional targets,
1416 @c update the --with-llsc description in the MIPS section below.
1419 On MIPS targets, make @option{-mllsc} the default when no
1420 @option{-mno-llsc} option is passed. This is the default for
1421 Linux-based targets, as the kernel will emulate them if the ISA does
1424 @item --without-llsc
1425 On MIPS targets, make @option{-mno-llsc} the default when no
1426 @option{-mllsc} option is passed.
1429 On MIPS targets, make @option{-msynci} the default when no
1430 @option{-mno-synci} option is passed.
1432 @item --without-synci
1433 On MIPS targets, make @option{-mno-synci} the default when no
1434 @option{-msynci} option is passed. This is the default.
1436 @item --with-lxc1-sxc1
1437 On MIPS targets, make @option{-mlxc1-sxc1} the default when no
1438 @option{-mno-lxc1-sxc1} option is passed. This is the default.
1440 @item --without-lxc1-sxc1
1441 On MIPS targets, make @option{-mno-lxc1-sxc1} the default when no
1442 @option{-mlxc1-sxc1} option is passed. The indexed load/store
1443 instructions are not directly a problem but can lead to unexpected
1444 behaviour when deployed in an application intended for a 32-bit address
1445 space but run on a 64-bit processor. The issue is seen because all
1446 known MIPS 64-bit Linux kernels execute o32 and n32 applications
1447 with 64-bit addressing enabled which affects the overflow behaviour
1448 of the indexed addressing mode. GCC will assume that ordinary
1449 32-bit arithmetic overflow behaviour is the same whether performed
1450 as an @code{addu} instruction or as part of the address calculation
1451 in @code{lwxc1} type instructions. This assumption holds true in a
1452 pure 32-bit environment and can hold true in a 64-bit environment if
1453 the address space is accurately set to be 32-bit for o32 and n32.
1456 On MIPS targets, make @option{-mmadd4} the default when no
1457 @option{-mno-madd4} option is passed. This is the default.
1459 @item --without-madd4
1460 On MIPS targets, make @option{-mno-madd4} the default when no
1461 @option{-mmadd4} option is passed. The @code{madd4} instruction
1462 family can be problematic when targeting a combination of cores that
1463 implement these instructions differently. There are two known cores
1464 that implement these as fused operations instead of unfused (where
1465 unfused is normally expected). Disabling these instructions is the
1466 only way to ensure compatible code is generated; this will incur
1467 a performance penalty.
1469 @item --with-mips-plt
1470 On MIPS targets, make use of copy relocations and PLTs.
1471 These features are extensions to the traditional
1472 SVR4-based MIPS ABIs and require support from GNU binutils
1473 and the runtime C library.
1475 @item --with-stack-clash-protection-guard-size=@var{size}
1476 On certain targets this option sets the default stack clash protection guard
1477 size as a power of two in bytes. On AArch64 @var{size} is required to be either
1478 12 (4KB) or 16 (64KB).
1480 @item --enable-__cxa_atexit
1481 Define if you want to use __cxa_atexit, rather than atexit, to
1482 register C++ destructors for local statics and global objects.
1483 This is essential for fully standards-compliant handling of
1484 destructors, but requires __cxa_atexit in libc. This option is currently
1485 only available on systems with GNU libc. When enabled, this will cause
1486 @option{-fuse-cxa-atexit} to be passed by default.
1488 @item --enable-gnu-indirect-function
1489 Define if you want to enable the @code{ifunc} attribute. This option is
1490 currently only available on systems with GNU libc on certain targets.
1492 @item --enable-target-optspace
1494 libraries should be optimized for code space instead of code speed.
1495 This is the default for the m32r platform.
1497 @item --with-cpp-install-dir=@var{dirname}
1498 Specify that the user visible @command{cpp} program should be installed
1499 in @file{@var{prefix}/@var{dirname}/cpp}, in addition to @var{bindir}.
1501 @item --enable-comdat
1502 Enable COMDAT group support. This is primarily used to override the
1503 automatically detected value.
1505 @item --enable-initfini-array
1506 Force the use of sections @code{.init_array} and @code{.fini_array}
1507 (instead of @code{.init} and @code{.fini}) for constructors and
1508 destructors. Option @option{--disable-initfini-array} has the
1509 opposite effect. If neither option is specified, the configure script
1510 will try to guess whether the @code{.init_array} and
1511 @code{.fini_array} sections are supported and, if they are, use them.
1513 @item --enable-link-mutex
1514 When building GCC, use a mutex to avoid linking the compilers for
1515 multiple languages at the same time, to avoid thrashing on build
1516 systems with limited free memory. The default is not to use such a mutex.
1518 @item --enable-maintainer-mode
1519 The build rules that regenerate the Autoconf and Automake output files as
1520 well as the GCC master message catalog @file{gcc.pot} are normally
1521 disabled. This is because it can only be rebuilt if the complete source
1522 tree is present. If you have changed the sources and want to rebuild the
1523 catalog, configuring with @option{--enable-maintainer-mode} will enable
1524 this. Note that you need a recent version of the @code{gettext} tools
1527 @item --disable-bootstrap
1528 For a native build, the default configuration is to perform
1529 a 3-stage bootstrap of the compiler when @samp{make} is invoked,
1530 testing that GCC can compile itself correctly. If you want to disable
1531 this process, you can configure with @option{--disable-bootstrap}.
1533 @item --enable-bootstrap
1534 In special cases, you may want to perform a 3-stage build
1535 even if the target and host triplets are different.
1536 This is possible when the host can run code compiled for
1537 the target (e.g.@: host is i686-linux, target is i486-linux).
1538 Starting from GCC 4.2, to do this you have to configure explicitly
1539 with @option{--enable-bootstrap}.
1541 @item --enable-generated-files-in-srcdir
1542 Neither the .c and .h files that are generated from Bison and flex nor the
1543 info manuals and man pages that are built from the .texi files are present
1544 in the SVN development tree. When building GCC from that development tree,
1545 or from one of our snapshots, those generated files are placed in your
1546 build directory, which allows for the source to be in a readonly
1549 If you configure with @option{--enable-generated-files-in-srcdir} then those
1550 generated files will go into the source directory. This is mainly intended
1551 for generating release or prerelease tarballs of the GCC sources, since it
1552 is not a requirement that the users of source releases to have flex, Bison,
1555 @item --enable-version-specific-runtime-libs
1557 that runtime libraries should be installed in the compiler specific
1558 subdirectory (@file{@var{libdir}/gcc}) rather than the usual places. In
1559 addition, @samp{libstdc++}'s include files will be installed into
1560 @file{@var{libdir}} unless you overruled it by using
1561 @option{--with-gxx-include-dir=@var{dirname}}. Using this option is
1562 particularly useful if you intend to use several versions of GCC in
1563 parallel. This is currently supported by @samp{libgfortran},
1564 @samp{libstdc++}, and @samp{libobjc}.
1566 @item @anchor{WithAixSoname}--with-aix-soname=@samp{aix}, @samp{svr4} or @samp{both}
1567 Traditional AIX shared library versioning (versioned @code{Shared Object}
1568 files as members of unversioned @code{Archive Library} files named
1569 @samp{lib.a}) causes numerous headaches for package managers. However,
1570 @code{Import Files} as members of @code{Archive Library} files allow for
1571 @strong{filename-based versioning} of shared libraries as seen on Linux/SVR4,
1572 where this is called the "SONAME". But as they prevent static linking,
1573 @code{Import Files} may be used with @code{Runtime Linking} only, where the
1574 linker does search for @samp{libNAME.so} before @samp{libNAME.a} library
1575 filenames with the @samp{-lNAME} linker flag.
1577 @anchor{AixLdCommand}For detailed information please refer to the AIX
1578 @uref{https://www.ibm.com/support/knowledgecenter/search/%22the%20ld%20command%2C%20also%20called%20the%20linkage%20editor%20or%20binder%22,,ld
1581 As long as shared library creation is enabled, upon:
1583 @item --with-aix-soname=aix
1584 @item --with-aix-soname=both
1585 A (traditional AIX) @code{Shared Archive Library} file is created:
1587 @item using the @samp{libNAME.a} filename scheme
1588 @item with the @code{Shared Object} file as archive member named
1589 @samp{libNAME.so.V} (except for @samp{libgcc_s}, where the @code{Shared
1590 Object} file is named @samp{shr.o} for backwards compatibility), which
1592 @item is used for runtime loading from inside the @samp{libNAME.a} file
1593 @item is used for dynamic loading via
1594 @code{dlopen("libNAME.a(libNAME.so.V)", RTLD_MEMBER)}
1595 @item is used for shared linking
1596 @item is used for static linking, so no separate @code{Static Archive
1597 Library} file is needed
1600 @item --with-aix-soname=both
1601 @item --with-aix-soname=svr4
1602 A (second) @code{Shared Archive Library} file is created:
1604 @item using the @samp{libNAME.so.V} filename scheme
1605 @item with the @code{Shared Object} file as archive member named
1608 @item is created with the @code{-G linker flag}
1609 @item has the @code{F_LOADONLY} flag set
1610 @item is used for runtime loading from inside the @samp{libNAME.so.V} file
1611 @item is used for dynamic loading via @code{dlopen("libNAME.so.V(shr.o)",
1614 @item with the @code{Import File} as archive member named @samp{shr.imp},
1617 @item refers to @samp{libNAME.so.V(shr.o)} as the "SONAME", to be recorded
1618 in the @code{Loader Section} of subsequent binaries
1619 @item indicates whether @samp{libNAME.so.V(shr.o)} is 32 or 64 bit
1620 @item lists all the public symbols exported by @samp{lib.so.V(shr.o)},
1621 eventually decorated with the @code{@samp{weak} Keyword}
1622 @item is necessary for shared linking against @samp{lib.so.V(shr.o)}
1625 A symbolic link using the @samp{libNAME.so} filename scheme is created:
1627 @item pointing to the @samp{libNAME.so.V} @code{Shared Archive Library} file
1628 @item to permit the @code{ld Command} to find @samp{lib.so.V(shr.imp)} via
1629 the @samp{-lNAME} argument (requires @code{Runtime Linking} to be enabled)
1630 @item to permit dynamic loading of @samp{lib.so.V(shr.o)} without the need
1631 to specify the version number via @code{dlopen("libNAME.so(shr.o)",
1636 As long as static library creation is enabled, upon:
1638 @item --with-aix-soname=svr4
1639 A @code{Static Archive Library} is created:
1641 @item using the @samp{libNAME.a} filename scheme
1642 @item with all the @code{Static Object} files as archive members, which
1644 @item are used for static linking
1649 While the aix-soname=@samp{svr4} option does not create @code{Shared Object}
1650 files as members of unversioned @code{Archive Library} files any more, package
1651 managers still are responsible to
1652 @uref{./specific.html#TransferAixShobj,,transfer} @code{Shared Object} files
1653 found as member of a previously installed unversioned @code{Archive Library}
1654 file into the newly installed @code{Archive Library} file with the same
1657 @emph{WARNING:} Creating @code{Shared Object} files with @code{Runtime Linking}
1658 enabled may bloat the TOC, eventually leading to @code{TOC overflow} errors,
1659 requiring the use of either the @option{-Wl,-bbigtoc} linker flag (seen to
1660 break with the @code{GDB} debugger) or some of the TOC-related compiler flags,
1662 @xref{RS/6000 and PowerPC Options,, RS/6000 and PowerPC Options, gcc,
1663 Using the GNU Compiler Collection (GCC)}.
1666 see ``RS/6000 and PowerPC Options'' in the main manual.
1669 @option{--with-aix-soname} is currently supported by @samp{libgcc_s} only, so
1670 this option is still experimental and not for normal use yet.
1672 Default is the traditional behavior @option{--with-aix-soname=@samp{aix}}.
1674 @item --enable-languages=@var{lang1},@var{lang2},@dots{}
1675 Specify that only a particular subset of compilers and
1676 their runtime libraries should be built. For a list of valid values for
1677 @var{langN} you can issue the following command in the
1678 @file{gcc} directory of your GCC source tree:@*
1680 grep ^language= */config-lang.in
1682 Currently, you can use any of the following:
1683 @code{all}, @code{default}, @code{ada}, @code{c}, @code{c++}, @code{d},
1684 @code{fortran}, @code{go}, @code{jit}, @code{lto}, @code{objc}, @code{obj-c++}.
1685 Building the Ada compiler has special requirements, see below.
1686 If you do not pass this flag, or specify the option @code{default}, then the
1687 default languages available in the @file{gcc} sub-tree will be configured.
1688 Ada, D, Go, Jit, and Objective-C++ are not default languages. LTO is not a
1689 default language, but is built by default because @option{--enable-lto} is
1690 enabled by default. The other languages are default languages. If
1691 @code{all} is specified, then all available languages are built. An
1692 exception is @code{jit} language, which requires
1693 @option{--enable-host-shared} to be included with @code{all}.
1695 @item --enable-stage1-languages=@var{lang1},@var{lang2},@dots{}
1696 Specify that a particular subset of compilers and their runtime
1697 libraries should be built with the system C compiler during stage 1 of
1698 the bootstrap process, rather than only in later stages with the
1699 bootstrapped C compiler. The list of valid values is the same as for
1700 @option{--enable-languages}, and the option @code{all} will select all
1701 of the languages enabled by @option{--enable-languages}. This option is
1702 primarily useful for GCC development; for instance, when a development
1703 version of the compiler cannot bootstrap due to compiler bugs, or when
1704 one is debugging front ends other than the C front end. When this
1705 option is used, one can then build the target libraries for the
1706 specified languages with the stage-1 compiler by using @command{make
1707 stage1-bubble all-target}, or run the testsuite on the stage-1 compiler
1708 for the specified languages using @command{make stage1-start check-gcc}.
1710 @item --disable-libada
1711 Specify that the run-time libraries and tools used by GNAT should not
1712 be built. This can be useful for debugging, or for compatibility with
1713 previous Ada build procedures, when it was required to explicitly
1714 do a @samp{make -C gcc gnatlib_and_tools}.
1716 @item --disable-libsanitizer
1717 Specify that the run-time libraries for the various sanitizers should
1720 @item --disable-libssp
1721 Specify that the run-time libraries for stack smashing protection
1722 should not be built or linked against. On many targets library support
1723 is provided by the C library instead.
1725 @item --disable-libquadmath
1726 Specify that the GCC quad-precision math library should not be built.
1727 On some systems, the library is required to be linkable when building
1728 the Fortran front end, unless @option{--disable-libquadmath-support}
1731 @item --disable-libquadmath-support
1732 Specify that the Fortran front end and @code{libgfortran} do not add
1733 support for @code{libquadmath} on systems supporting it.
1735 @item --disable-libgomp
1736 Specify that the GNU Offloading and Multi Processing Runtime Library
1737 should not be built.
1739 @item --disable-libvtv
1740 Specify that the run-time libraries used by vtable verification
1741 should not be built.
1744 Specify that the compiler should
1745 use DWARF 2 debugging information as the default.
1747 @item --with-advance-toolchain=@var{at}
1748 On 64-bit PowerPC Linux systems, configure the compiler to use the
1749 header files, library files, and the dynamic linker from the Advance
1750 Toolchain release @var{at} instead of the default versions that are
1751 provided by the Linux distribution. In general, this option is
1752 intended for the developers of GCC, and it is not intended for general
1755 @item --enable-targets=all
1756 @itemx --enable-targets=@var{target_list}
1757 Some GCC targets, e.g.@: powerpc64-linux, build bi-arch compilers.
1758 These are compilers that are able to generate either 64-bit or 32-bit
1759 code. Typically, the corresponding 32-bit target, e.g.@:
1760 powerpc-linux for powerpc64-linux, only generates 32-bit code. This
1761 option enables the 32-bit target to be a bi-arch compiler, which is
1762 useful when you want a bi-arch compiler that defaults to 32-bit, and
1763 you are building a bi-arch or multi-arch binutils in a combined tree.
1764 On mips-linux, this will build a tri-arch compiler (ABI o32/n32/64),
1766 Currently, this option only affects sparc-linux, powerpc-linux, x86-linux,
1767 mips-linux and s390-linux.
1769 @item --enable-default-pie
1770 Turn on @option{-fPIE} and @option{-pie} by default.
1772 @item --enable-secureplt
1773 This option enables @option{-msecure-plt} by default for powerpc-linux.
1775 @xref{RS/6000 and PowerPC Options,, RS/6000 and PowerPC Options, gcc,
1776 Using the GNU Compiler Collection (GCC)},
1779 See ``RS/6000 and PowerPC Options'' in the main manual
1782 @item --enable-default-ssp
1783 Turn on @option{-fstack-protector-strong} by default.
1786 This option enables @option{-mcld} by default for 32-bit x86 targets.
1788 @xref{i386 and x86-64 Options,, i386 and x86-64 Options, gcc,
1789 Using the GNU Compiler Collection (GCC)},
1792 See ``i386 and x86-64 Options'' in the main manual
1795 @item --enable-large-address-aware
1796 The @option{--enable-large-address-aware} option arranges for MinGW
1797 executables to be linked using the @option{--large-address-aware}
1798 option, that enables the use of more than 2GB of memory. If GCC is
1799 configured with this option, its effects can be reversed by passing the
1800 @option{-Wl,--disable-large-address-aware} option to the so-configured
1803 @item --enable-win32-registry
1804 @itemx --enable-win32-registry=@var{key}
1805 @itemx --disable-win32-registry
1806 The @option{--enable-win32-registry} option enables Microsoft Windows-hosted GCC
1807 to look up installations paths in the registry using the following key:
1810 @code{HKEY_LOCAL_MACHINE\SOFTWARE\Free Software Foundation\@var{key}}
1813 @var{key} defaults to GCC version number, and can be overridden by the
1814 @option{--enable-win32-registry=@var{key}} option. Vendors and distributors
1815 who use custom installers are encouraged to provide a different key,
1816 perhaps one comprised of vendor name and GCC version number, to
1817 avoid conflict with existing installations. This feature is enabled
1818 by default, and can be disabled by @option{--disable-win32-registry}
1819 option. This option has no effect on the other hosts.
1822 Specify that the machine does not have a floating point unit. This
1823 option only applies to @samp{m68k-sun-sunos@var{n}}. On any other
1824 system, @option{--nfp} has no effect.
1826 @item --enable-werror
1827 @itemx --disable-werror
1828 @itemx --enable-werror=yes
1829 @itemx --enable-werror=no
1830 When you specify this option, it controls whether certain files in the
1831 compiler are built with @option{-Werror} in bootstrap stage2 and later.
1832 If you don't specify it, @option{-Werror} is turned on for the main
1833 development trunk. However it defaults to off for release branches and
1834 final releases. The specific files which get @option{-Werror} are
1835 controlled by the Makefiles.
1837 @item --enable-checking
1838 @itemx --enable-checking=@var{list}
1839 When you specify this option, the compiler is built to perform internal
1840 consistency checks of the requested complexity. This does not change the
1841 generated code, but adds error checking within the compiler. This will
1842 slow down the compiler and may only work properly if you are building
1843 the compiler with GCC@. This is @samp{yes,extra} by default when building
1844 from SVN or snapshots, but @samp{release} for releases. The default
1845 for building the stage1 compiler is @samp{yes}. More control
1846 over the checks may be had by specifying @var{list}. The categories of
1847 checks available are @samp{yes} (most common checks
1848 @samp{assert,misc,tree,gc,rtlflag,runtime}), @samp{no} (no checks at
1849 all), @samp{all} (all but @samp{valgrind}), @samp{release} (cheapest
1850 checks @samp{assert,runtime}) or @samp{none} (same as @samp{no}).
1851 Individual checks can be enabled with these flags @samp{assert},
1852 @samp{df}, @samp{fold}, @samp{gc}, @samp{gcac}, @samp{misc}, @samp{rtl},
1853 @samp{rtlflag}, @samp{runtime}, @samp{tree}, @samp{extra} and @samp{valgrind}.
1854 @samp{extra} adds for @samp{misc} checking extra checks that might affect
1855 code generation and should therefore not differ between stage1 and later
1858 The @samp{valgrind} check requires the external @command{valgrind}
1859 simulator, available from @uref{http://valgrind.org/}. The
1860 @samp{df}, @samp{rtl}, @samp{gcac} and @samp{valgrind} checks are very expensive.
1861 To disable all checking, @samp{--disable-checking} or
1862 @samp{--enable-checking=none} must be explicitly requested. Disabling
1863 assertions will make the compiler and runtime slightly faster but
1864 increase the risk of undetected internal errors causing wrong code to be
1867 @item --disable-stage1-checking
1868 @itemx --enable-stage1-checking
1869 @itemx --enable-stage1-checking=@var{list}
1870 If no @option{--enable-checking} option is specified the stage1
1871 compiler will be built with @samp{yes} checking enabled, otherwise
1872 the stage1 checking flags are the same as specified by
1873 @option{--enable-checking}. To build the stage1 compiler with
1874 different checking options use @option{--enable-stage1-checking}.
1875 The list of checking options is the same as for @option{--enable-checking}.
1876 If your system is too slow or too small to bootstrap a released compiler
1877 with checking for stage1 enabled, you can use @samp{--disable-stage1-checking}
1878 to disable checking for the stage1 compiler.
1880 @item --enable-coverage
1881 @itemx --enable-coverage=@var{level}
1882 With this option, the compiler is built to collect self coverage
1883 information, every time it is run. This is for internal development
1884 purposes, and only works when the compiler is being built with gcc. The
1885 @var{level} argument controls whether the compiler is built optimized or
1886 not, values are @samp{opt} and @samp{noopt}. For coverage analysis you
1887 want to disable optimization, for performance analysis you want to
1888 enable optimization. When coverage is enabled, the default level is
1889 without optimization.
1891 @item --enable-gather-detailed-mem-stats
1892 When this option is specified more detailed information on memory
1893 allocation is gathered. This information is printed when using
1894 @option{-fmem-report}.
1896 @item --enable-valgrind-annotations
1897 Mark selected memory related operations in the compiler when run under
1898 valgrind to suppress false positives.
1901 @itemx --disable-nls
1902 The @option{--enable-nls} option enables Native Language Support (NLS),
1903 which lets GCC output diagnostics in languages other than American
1904 English. Native Language Support is enabled by default if not doing a
1905 canadian cross build. The @option{--disable-nls} option disables NLS@.
1907 @item --with-included-gettext
1908 If NLS is enabled, the @option{--with-included-gettext} option causes the build
1909 procedure to prefer its copy of GNU @command{gettext}.
1911 @item --with-catgets
1912 If NLS is enabled, and if the host lacks @code{gettext} but has the
1913 inferior @code{catgets} interface, the GCC build procedure normally
1914 ignores @code{catgets} and instead uses GCC's copy of the GNU
1915 @code{gettext} library. The @option{--with-catgets} option causes the
1916 build procedure to use the host's @code{catgets} in this situation.
1918 @item --with-libiconv-prefix=@var{dir}
1919 Search for libiconv header files in @file{@var{dir}/include} and
1920 libiconv library files in @file{@var{dir}/lib}.
1922 @item --enable-obsolete
1923 Enable configuration for an obsoleted system. If you attempt to
1924 configure GCC for a system (build, host, or target) which has been
1925 obsoleted, and you do not specify this flag, configure will halt with an
1928 All support for systems which have been obsoleted in one release of GCC
1929 is removed entirely in the next major release, unless someone steps
1930 forward to maintain the port.
1932 @item --enable-decimal-float
1933 @itemx --enable-decimal-float=yes
1934 @itemx --enable-decimal-float=no
1935 @itemx --enable-decimal-float=bid
1936 @itemx --enable-decimal-float=dpd
1937 @itemx --disable-decimal-float
1938 Enable (or disable) support for the C decimal floating point extension
1939 that is in the IEEE 754-2008 standard. This is enabled by default only
1940 on PowerPC, i386, and x86_64 GNU/Linux systems. Other systems may also
1941 support it, but require the user to specifically enable it. You can
1942 optionally control which decimal floating point format is used (either
1943 @samp{bid} or @samp{dpd}). The @samp{bid} (binary integer decimal)
1944 format is default on i386 and x86_64 systems, and the @samp{dpd}
1945 (densely packed decimal) format is default on PowerPC systems.
1947 @item --enable-fixed-point
1948 @itemx --disable-fixed-point
1949 Enable (or disable) support for C fixed-point arithmetic.
1950 This option is enabled by default for some targets (such as MIPS) which
1951 have hardware-support for fixed-point operations. On other targets, you
1952 may enable this option manually.
1954 @item --with-long-double-128
1955 Specify if @code{long double} type should be 128-bit by default on selected
1956 GNU/Linux architectures. If using @code{--without-long-double-128},
1957 @code{long double} will be by default 64-bit, the same as @code{double} type.
1958 When neither of these configure options are used, the default will be
1959 128-bit @code{long double} when built against GNU C Library 2.4 and later,
1960 64-bit @code{long double} otherwise.
1962 @item --with-long-double-format=ibm
1963 @itemx --with-long-double-format=ieee
1964 Specify whether @code{long double} uses the IBM extended double format
1965 or the IEEE 128-bit floating point format on PowerPC Linux systems.
1966 This configuration switch will only work on little endian PowerPC
1967 Linux systems and on big endian 64-bit systems where the default cpu
1968 is at least power7 (i.e.@: @option{--with-cpu=power7},
1969 @option{--with-cpu=power8}, or @option{--with-cpu=power9} is used).
1971 If you use the @option{--with-long-double-64} configuration option,
1972 the @option{--with-long-double-format=ibm} and
1973 @option{--with-long-double-format=ieee} options are ignored.
1975 The default @code{long double} format is to use IBM extended double.
1976 Until all of the libraries are converted to use IEEE 128-bit floating
1977 point, it is not recommended to use
1978 @option{--with-long-double-format=ieee}.
1980 On little endian PowerPC Linux systems, if you explicitly set the
1981 @code{long double} type, it will build multilibs to allow you to
1982 select either @code{long double} format, unless you disable multilibs
1983 with the @code{--disable-multilib} option. At present,
1984 @code{long double} multilibs are not built on big endian PowerPC Linux
1985 systems. If you are building multilibs, you will need to configure
1986 the compiler using the @option{--with-system-zlib} option.
1988 If you do not set the @code{long double} type explicitly, no multilibs
1991 @item --enable-fdpic
1992 On SH Linux systems, generate ELF FDPIC code.
1994 @item --with-gmp=@var{pathname}
1995 @itemx --with-gmp-include=@var{pathname}
1996 @itemx --with-gmp-lib=@var{pathname}
1997 @itemx --with-mpfr=@var{pathname}
1998 @itemx --with-mpfr-include=@var{pathname}
1999 @itemx --with-mpfr-lib=@var{pathname}
2000 @itemx --with-mpc=@var{pathname}
2001 @itemx --with-mpc-include=@var{pathname}
2002 @itemx --with-mpc-lib=@var{pathname}
2003 If you want to build GCC but do not have the GMP library, the MPFR
2004 library and/or the MPC library installed in a standard location and
2005 do not have their sources present in the GCC source tree then you
2006 can explicitly specify the directory where they are installed
2007 (@samp{--with-gmp=@var{gmpinstalldir}},
2008 @samp{--with-mpfr=@/@var{mpfrinstalldir}},
2009 @samp{--with-mpc=@/@var{mpcinstalldir}}). The
2010 @option{--with-gmp=@/@var{gmpinstalldir}} option is shorthand for
2011 @option{--with-gmp-lib=@/@var{gmpinstalldir}/lib} and
2012 @option{--with-gmp-include=@/@var{gmpinstalldir}/include}. Likewise the
2013 @option{--with-mpfr=@/@var{mpfrinstalldir}} option is shorthand for
2014 @option{--with-mpfr-lib=@/@var{mpfrinstalldir}/lib} and
2015 @option{--with-mpfr-include=@/@var{mpfrinstalldir}/include}, also the
2016 @option{--with-mpc=@/@var{mpcinstalldir}} option is shorthand for
2017 @option{--with-mpc-lib=@/@var{mpcinstalldir}/lib} and
2018 @option{--with-mpc-include=@/@var{mpcinstalldir}/include}. If these
2019 shorthand assumptions are not correct, you can use the explicit
2020 include and lib options directly. You might also need to ensure the
2021 shared libraries can be found by the dynamic linker when building and
2022 using GCC, for example by setting the runtime shared library path
2023 variable (@env{LD_LIBRARY_PATH} on GNU/Linux and Solaris systems).
2025 These flags are applicable to the host platform only. When building
2026 a cross compiler, they will not be used to configure target libraries.
2028 @item --with-isl=@var{pathname}
2029 @itemx --with-isl-include=@var{pathname}
2030 @itemx --with-isl-lib=@var{pathname}
2031 If you do not have the isl library installed in a standard location and you
2032 want to build GCC, you can explicitly specify the directory where it is
2033 installed (@samp{--with-isl=@/@var{islinstalldir}}). The
2034 @option{--with-isl=@/@var{islinstalldir}} option is shorthand for
2035 @option{--with-isl-lib=@/@var{islinstalldir}/lib} and
2036 @option{--with-isl-include=@/@var{islinstalldir}/include}. If this
2037 shorthand assumption is not correct, you can use the explicit
2038 include and lib options directly.
2040 These flags are applicable to the host platform only. When building
2041 a cross compiler, they will not be used to configure target libraries.
2043 @item --with-stage1-ldflags=@var{flags}
2044 This option may be used to set linker flags to be used when linking
2045 stage 1 of GCC. These are also used when linking GCC if configured with
2046 @option{--disable-bootstrap}. If @option{--with-stage1-libs} is not set to a
2047 value, then the default is @samp{-static-libstdc++ -static-libgcc}, if
2050 @item --with-stage1-libs=@var{libs}
2051 This option may be used to set libraries to be used when linking stage 1
2052 of GCC. These are also used when linking GCC if configured with
2053 @option{--disable-bootstrap}.
2055 @item --with-boot-ldflags=@var{flags}
2056 This option may be used to set linker flags to be used when linking
2057 stage 2 and later when bootstrapping GCC. If --with-boot-libs
2058 is not is set to a value, then the default is
2059 @samp{-static-libstdc++ -static-libgcc}.
2061 @item --with-boot-libs=@var{libs}
2062 This option may be used to set libraries to be used when linking stage 2
2063 and later when bootstrapping GCC.
2065 @item --with-debug-prefix-map=@var{map}
2066 Convert source directory names using @option{-fdebug-prefix-map} when
2067 building runtime libraries. @samp{@var{map}} is a space-separated
2068 list of maps of the form @samp{@var{old}=@var{new}}.
2070 @item --enable-linker-build-id
2071 Tells GCC to pass @option{--build-id} option to the linker for all final
2072 links (links performed without the @option{-r} or @option{--relocatable}
2073 option), if the linker supports it. If you specify
2074 @option{--enable-linker-build-id}, but your linker does not
2075 support @option{--build-id} option, a warning is issued and the
2076 @option{--enable-linker-build-id} option is ignored. The default is off.
2078 @item --with-linker-hash-style=@var{choice}
2079 Tells GCC to pass @option{--hash-style=@var{choice}} option to the
2080 linker for all final links. @var{choice} can be one of
2081 @samp{sysv}, @samp{gnu}, and @samp{both} where @samp{sysv} is the default.
2083 @item --enable-gnu-unique-object
2084 @itemx --disable-gnu-unique-object
2085 Tells GCC to use the gnu_unique_object relocation for C++ template
2086 static data members and inline function local statics. Enabled by
2087 default for a toolchain with an assembler that accepts it and
2088 GLIBC 2.11 or above, otherwise disabled.
2090 @item --with-diagnostics-color=@var{choice}
2091 Tells GCC to use @var{choice} as the default for @option{-fdiagnostics-color=}
2092 option (if not used explicitly on the command line). @var{choice}
2093 can be one of @samp{never}, @samp{auto}, @samp{always}, and @samp{auto-if-env}
2094 where @samp{auto} is the default. @samp{auto-if-env} means that
2095 @option{-fdiagnostics-color=auto} will be the default if @code{GCC_COLORS}
2096 is present and non-empty in the environment, and
2097 @option{-fdiagnostics-color=never} otherwise.
2100 @itemx --disable-lto
2101 Enable support for link-time optimization (LTO). This is enabled by
2102 default, and may be disabled using @option{--disable-lto}.
2104 @item --enable-linker-plugin-configure-flags=FLAGS
2105 @itemx --enable-linker-plugin-flags=FLAGS
2106 By default, linker plugins (such as the LTO plugin) are built for the
2107 host system architecture. For the case that the linker has a
2108 different (but run-time compatible) architecture, these flags can be
2109 specified to build plugins that are compatible to the linker. For
2110 example, if you are building GCC for a 64-bit x86_64
2111 (@samp{x86_64-pc-linux-gnu}) host system, but have a 32-bit x86
2112 GNU/Linux (@samp{i686-pc-linux-gnu}) linker executable (which is
2113 executable on the former system), you can configure GCC as follows for
2114 getting compatible linker plugins:
2117 % @var{srcdir}/configure \
2118 --host=x86_64-pc-linux-gnu \
2119 --enable-linker-plugin-configure-flags=--host=i686-pc-linux-gnu \
2120 --enable-linker-plugin-flags='CC=gcc\ -m32\ -Wl,-rpath,[...]/i686-pc-linux-gnu/lib'
2123 @item --with-plugin-ld=@var{pathname}
2124 Enable an alternate linker to be used at link-time optimization (LTO)
2125 link time when @option{-fuse-linker-plugin} is enabled.
2126 This linker should have plugin support such as gold starting with
2127 version 2.20 or GNU ld starting with version 2.21.
2128 See @option{-fuse-linker-plugin} for details.
2130 @item --enable-canonical-system-headers
2131 @itemx --disable-canonical-system-headers
2132 Enable system header path canonicalization for @file{libcpp}. This can
2133 produce shorter header file paths in diagnostics and dependency output
2134 files, but these changed header paths may conflict with some compilation
2135 environments. Enabled by default, and may be disabled using
2136 @option{--disable-canonical-system-headers}.
2138 @item --with-glibc-version=@var{major}.@var{minor}
2139 Tell GCC that when the GNU C Library (glibc) is used on the target it
2140 will be version @var{major}.@var{minor} or later. Normally this can
2141 be detected from the C library's header files, but this option may be
2142 needed when bootstrapping a cross toolchain without the header files
2143 available for building the initial bootstrap compiler.
2145 If GCC is configured with some multilibs that use glibc and some that
2146 do not, this option applies only to the multilibs that use glibc.
2147 However, such configurations may not work well as not all the relevant
2148 configuration in GCC is on a per-multilib basis.
2150 @item --enable-as-accelerator-for=@var{target}
2151 Build as offload target compiler. Specify offload host triple by @var{target}.
2153 @item --enable-offload-targets=@var{target1}[=@var{path1}],@dots{},@var{targetN}[=@var{pathN}]
2154 Enable offloading to targets @var{target1}, @dots{}, @var{targetN}.
2155 Offload compilers are expected to be already installed. Default search
2156 path for them is @file{@var{exec-prefix}}, but it can be changed by
2157 specifying paths @var{path1}, @dots{}, @var{pathN}.
2160 % @var{srcdir}/configure \
2161 --enable-offload-targets=x86_64-intelmicemul-linux-gnu=/path/to/x86_64/compiler,nvptx-none,hsa
2164 If @samp{hsa} is specified as one of the targets, the compiler will be
2165 built with support for HSA GPU accelerators. Because the same
2166 compiler will emit the accelerator code, no path should be specified.
2168 @item --with-hsa-runtime=@var{pathname}
2169 @itemx --with-hsa-runtime-include=@var{pathname}
2170 @itemx --with-hsa-runtime-lib=@var{pathname}
2172 If you configure GCC with HSA offloading but do not have the HSA
2173 run-time library installed in a standard location then you can
2174 explicitly specify the directory where they are installed. The
2175 @option{--with-hsa-runtime=@/@var{hsainstalldir}} option is a
2177 @option{--with-hsa-runtime-lib=@/@var{hsainstalldir}/lib} and
2178 @option{--with-hsa-runtime-include=@/@var{hsainstalldir}/include}.
2181 @itemx --disable-cet
2182 Enable building target run-time libraries with control-flow
2183 instrumentation, see @option{-fcf-protection} option. When
2184 @code{--enable-cet} is specified target libraries are configured
2185 to add @option{-fcf-protection} and, if needed, other target
2186 specific options to a set of building options.
2188 The option is disabled by default. When @code{--enable-cet=auto}
2189 is used, it is enabled on Linux/x86 if target binutils
2190 supports @code{Intel CET} instructions and disabled otherwise.
2191 In this case the target libraries are configured to get additional
2192 @option{-fcf-protection} option.
2194 @item --with-riscv-attribute=@samp{yes}, @samp{no} or @samp{default}
2195 Generate RISC-V attribute by default, in order to record extra build
2196 information in object.
2198 The option is disabled by default. It is enabled on RISC-V/ELF (bare-metal)
2199 target if target binutils supported.
2202 @subheading Cross-Compiler-Specific Options
2203 The following options only apply to building cross compilers.
2206 @item --with-sysroot
2207 @itemx --with-sysroot=@var{dir}
2208 Tells GCC to consider @var{dir} as the root of a tree that contains
2209 (a subset of) the root filesystem of the target operating system.
2210 Target system headers, libraries and run-time object files will be
2211 searched for in there. More specifically, this acts as if
2212 @option{--sysroot=@var{dir}} was added to the default options of the built
2213 compiler. The specified directory is not copied into the
2214 install tree, unlike the options @option{--with-headers} and
2215 @option{--with-libs} that this option obsoletes. The default value,
2216 in case @option{--with-sysroot} is not given an argument, is
2217 @option{$@{gcc_tooldir@}/sys-root}. If the specified directory is a
2218 subdirectory of @option{$@{exec_prefix@}}, then it will be found relative to
2219 the GCC binaries if the installation tree is moved.
2221 This option affects the system root for the compiler used to build
2222 target libraries (which runs on the build system) and the compiler newly
2223 installed with @code{make install}; it does not affect the compiler which is
2224 used to build GCC itself.
2226 If you specify the @option{--with-native-system-header-dir=@var{dirname}}
2227 option then the compiler will search that directory within @var{dirname} for
2228 native system headers rather than the default @file{/usr/include}.
2230 @item --with-build-sysroot
2231 @itemx --with-build-sysroot=@var{dir}
2232 Tells GCC to consider @var{dir} as the system root (see
2233 @option{--with-sysroot}) while building target libraries, instead of
2234 the directory specified with @option{--with-sysroot}. This option is
2235 only useful when you are already using @option{--with-sysroot}. You
2236 can use @option{--with-build-sysroot} when you are configuring with
2237 @option{--prefix} set to a directory that is different from the one in
2238 which you are installing GCC and your target libraries.
2240 This option affects the system root for the compiler used to build
2241 target libraries (which runs on the build system); it does not affect
2242 the compiler which is used to build GCC itself.
2244 If you specify the @option{--with-native-system-header-dir=@var{dirname}}
2245 option then the compiler will search that directory within @var{dirname} for
2246 native system headers rather than the default @file{/usr/include}.
2248 @item --with-headers
2249 @itemx --with-headers=@var{dir}
2250 Deprecated in favor of @option{--with-sysroot}.
2251 Specifies that target headers are available when building a cross compiler.
2252 The @var{dir} argument specifies a directory which has the target include
2253 files. These include files will be copied into the @file{gcc} install
2254 directory. @emph{This option with the @var{dir} argument is required} when
2255 building a cross compiler, if @file{@var{prefix}/@var{target}/sys-include}
2256 doesn't pre-exist. If @file{@var{prefix}/@var{target}/sys-include} does
2257 pre-exist, the @var{dir} argument may be omitted. @command{fixincludes}
2258 will be run on these files to make them compatible with GCC@.
2260 @item --without-headers
2261 Tells GCC not use any target headers from a libc when building a cross
2262 compiler. When crossing to GNU/Linux, you need the headers so GCC
2263 can build the exception handling for libgcc.
2266 @itemx --with-libs="@var{dir1} @var{dir2} @dots{} @var{dirN}"
2267 Deprecated in favor of @option{--with-sysroot}.
2268 Specifies a list of directories which contain the target runtime
2269 libraries. These libraries will be copied into the @file{gcc} install
2270 directory. If the directory list is omitted, this option has no
2274 Specifies that @samp{newlib} is
2275 being used as the target C library. This causes @code{__eprintf} to be
2276 omitted from @file{libgcc.a} on the assumption that it will be provided by
2282 @item --with-avrlibc
2283 Only supported for the AVR target. Specifies that @samp{AVR-Libc} is
2284 being used as the target C@tie{} library. This causes float support
2285 functions like @code{__addsf3} to be omitted from @file{libgcc.a} on
2286 the assumption that it will be provided by @file{libm.a}. For more
2287 technical details, cf. @uref{http://gcc.gnu.org/PR54461,,PR54461}.
2288 It is not supported for
2289 RTEMS configurations, which currently use newlib. The option is
2290 supported since version 4.7.2 and is the default in 4.8.0 and newer.
2292 @item --with-double=@{32|64|32,64|64,32@}
2293 @itemx --with-long-double=@{32|64|32,64|64,32|double@}
2294 Only supported for the AVR target since version@tie{}10.
2295 Specify the default layout available for the C/C++ @samp{double}
2296 and @samp{long double} type, respectively. The following rules apply:
2299 The first value after the @samp{=} specifies the default layout (in bits)
2300 of the type and also the default for the @option{-mdouble=} resp.
2301 @option{-mlong-double=} compiler option.
2303 If more than one value is specified, respective multilib variants are
2304 available, and @option{-mdouble=} resp. @option{-mlong-double=} acts
2305 as a multilib option.
2307 If @option{--with-long-double=double} is specified, @samp{double} and
2308 @samp{long double} will have the same layout.
2310 If the configure option is not set, it defaults to @samp{32} which
2311 is compatible with older versions of the compiler that use non-standard
2312 32-bit types for @samp{double} and @samp{long double}.
2314 Not all combinations of @option{--with-double=} and
2315 @option{--with-long-double=} are valid. For example, the combination
2316 @option{--with-double=32,64} @option{--with-long-double=32} will be
2317 rejected because the first option specifies the availability of
2318 multilibs for @samp{double}, whereas the second option implies
2319 that @samp{long double} --- and hence also @samp{double} --- is always
2322 @item --with-nds32-lib=@var{library}
2323 Specifies that @var{library} setting is used for building @file{libgcc.a}.
2324 Currently, the valid @var{library} is @samp{newlib} or @samp{mculib}.
2325 This option is only supported for the NDS32 target.
2327 @item --with-build-time-tools=@var{dir}
2328 Specifies where to find the set of target tools (assembler, linker, etc.)
2329 that will be used while building GCC itself. This option can be useful
2330 if the directory layouts are different between the system you are building
2331 GCC on, and the system where you will deploy it.
2333 For example, on an @samp{ia64-hp-hpux} system, you may have the GNU
2334 assembler and linker in @file{/usr/bin}, and the native tools in a
2335 different path, and build a toolchain that expects to find the
2336 native tools in @file{/usr/bin}.
2338 When you use this option, you should ensure that @var{dir} includes
2339 @command{ar}, @command{as}, @command{ld}, @command{nm},
2340 @command{ranlib} and @command{strip} if necessary, and possibly
2341 @command{objdump}. Otherwise, GCC may use an inconsistent set of
2345 @subsubheading Overriding @command{configure} test results
2347 Sometimes, it might be necessary to override the result of some
2348 @command{configure} test, for example in order to ease porting to a new
2349 system or work around a bug in a test. The toplevel @command{configure}
2350 script provides three variables for this:
2354 @item build_configargs
2355 @cindex @code{build_configargs}
2356 The contents of this variable is passed to all build @command{configure}
2359 @item host_configargs
2360 @cindex @code{host_configargs}
2361 The contents of this variable is passed to all host @command{configure}
2364 @item target_configargs
2365 @cindex @code{target_configargs}
2366 The contents of this variable is passed to all target @command{configure}
2371 In order to avoid shell and @command{make} quoting issues for complex
2372 overrides, you can pass a setting for @env{CONFIG_SITE} and set
2373 variables in the site file.
2375 @subheading Objective-C-Specific Options
2377 The following options apply to the build of the Objective-C runtime library.
2380 @item --enable-objc-gc
2381 Specify that an additional variant of the GNU Objective-C runtime library
2382 is built, using an external build of the Boehm-Demers-Weiser garbage
2383 collector (@uref{https://www.hboehm.info/gc/}). This library needs to be
2384 available for each multilib variant, unless configured with
2385 @option{--enable-objc-gc=@samp{auto}} in which case the build of the
2386 additional runtime library is skipped when not available and the build
2389 @item --with-target-bdw-gc=@var{list}
2390 @itemx --with-target-bdw-gc-include=@var{list}
2391 @itemx --with-target-bdw-gc-lib=@var{list}
2392 Specify search directories for the garbage collector header files and
2393 libraries. @var{list} is a comma separated list of key value pairs of the
2394 form @samp{@var{multilibdir}=@var{path}}, where the default multilib key
2395 is named as @samp{.} (dot), or is omitted (e.g.@:
2396 @samp{--with-target-bdw-gc=/opt/bdw-gc,32=/opt-bdw-gc32}).
2398 The options @option{--with-target-bdw-gc-include} and
2399 @option{--with-target-bdw-gc-lib} must always be specified together
2400 for each multilib variant and they take precedence over
2401 @option{--with-target-bdw-gc}. If @option{--with-target-bdw-gc-include}
2402 is missing values for a multilib, then the value for the default
2403 multilib is used (e.g.@: @samp{--with-target-bdw-gc-include=/opt/bdw-gc/include}
2404 @samp{--with-target-bdw-gc-lib=/opt/bdw-gc/lib64,32=/opt-bdw-gc/lib32}).
2405 If none of these options are specified, the library is assumed in
2409 @subheading D-Specific Options
2411 The following options apply to the build of the D runtime library.
2414 @item --with-target-system-zlib
2415 Use installed @samp{zlib} rather than that included with GCC@. This needs
2416 to be available for each multilib variant, unless configured with
2417 @option{--with-target-system-zlib=@samp{auto}} in which case the GCC@ included
2418 @samp{zlib} is only used when the system installed library is not available.
2426 @uref{./index.html,,Return to the GCC Installation page}
2430 @c ***Building****************************************************************
2432 @comment node-name, next, previous, up
2433 @node Building, Testing, Configuration, Installing GCC
2439 @cindex Installing GCC: Building
2441 Now that GCC is configured, you are ready to build the compiler and
2444 Some commands executed when making the compiler may fail (return a
2445 nonzero status) and be ignored by @command{make}. These failures, which
2446 are often due to files that were not found, are expected, and can safely
2449 It is normal to have compiler warnings when compiling certain files.
2450 Unless you are a GCC developer, you can generally ignore these warnings
2451 unless they cause compilation to fail. Developers should attempt to fix
2452 any warnings encountered, however they can temporarily continue past
2453 warnings-as-errors by specifying the configure flag
2454 @option{--disable-werror}.
2456 On certain old systems, defining certain environment variables such as
2457 @env{CC} can interfere with the functioning of @command{make}.
2459 If you encounter seemingly strange errors when trying to build the
2460 compiler in a directory other than the source directory, it could be
2461 because you have previously configured the compiler in the source
2462 directory. Make sure you have done all the necessary preparations.
2464 If you build GCC on a BSD system using a directory stored in an old System
2465 V file system, problems may occur in running @command{fixincludes} if the
2466 System V file system doesn't support symbolic links. These problems
2467 result in a failure to fix the declaration of @code{size_t} in
2468 @file{sys/types.h}. If you find that @code{size_t} is a signed type and
2469 that type mismatches occur, this could be the cause.
2471 The solution is not to use such a directory for building GCC@.
2473 Similarly, when building from SVN or snapshots, or if you modify
2474 @file{*.l} files, you need the Flex lexical analyzer generator
2475 installed. If you do not modify @file{*.l} files, releases contain
2476 the Flex-generated files and you do not need Flex installed to build
2477 them. There is still one Flex-based lexical analyzer (part of the
2478 build machinery, not of GCC itself) that is used even if you only
2479 build the C front end.
2481 When building from SVN or snapshots, or if you modify Texinfo
2482 documentation, you need version 4.7 or later of Texinfo installed if you
2483 want Info documentation to be regenerated. Releases contain Info
2484 documentation pre-built for the unmodified documentation in the release.
2486 @section Building a native compiler
2488 For a native build, the default configuration is to perform
2489 a 3-stage bootstrap of the compiler when @samp{make} is invoked.
2490 This will build the entire GCC system and ensure that it compiles
2491 itself correctly. It can be disabled with the @option{--disable-bootstrap}
2492 parameter to @samp{configure}, but bootstrapping is suggested because
2493 the compiler will be tested more completely and could also have
2496 The bootstrapping process will complete the following steps:
2500 Build tools necessary to build the compiler.
2503 Perform a 3-stage bootstrap of the compiler. This includes building
2504 three times the target tools for use by the compiler such as binutils
2505 (bfd, binutils, gas, gprof, ld, and opcodes) if they have been
2506 individually linked or moved into the top level GCC source tree before
2510 Perform a comparison test of the stage2 and stage3 compilers.
2513 Build runtime libraries using the stage3 compiler from the previous step.
2517 If you are short on disk space you might consider @samp{make
2518 bootstrap-lean} instead. The sequence of compilation is the
2519 same described above, but object files from the stage1 and
2520 stage2 of the 3-stage bootstrap of the compiler are deleted as
2521 soon as they are no longer needed.
2523 If you wish to use non-default GCC flags when compiling the stage2
2524 and stage3 compilers, set @code{BOOT_CFLAGS} on the command line when
2525 doing @samp{make}. For example, if you want to save additional space
2526 during the bootstrap and in the final installation as well, you can
2527 build the compiler binaries without debugging information as in the
2528 following example. This will save roughly 40% of disk space both for
2529 the bootstrap and the final installation. (Libraries will still contain
2530 debugging information.)
2533 make BOOT_CFLAGS='-O' bootstrap
2536 You can place non-default optimization flags into @code{BOOT_CFLAGS}; they
2537 are less well tested here than the default of @samp{-g -O2}, but should
2538 still work. In a few cases, you may find that you need to specify special
2539 flags such as @option{-msoft-float} here to complete the bootstrap; or,
2540 if the native compiler miscompiles the stage1 compiler, you may need
2541 to work around this, by choosing @code{BOOT_CFLAGS} to avoid the parts
2542 of the stage1 compiler that were miscompiled, or by using @samp{make
2543 bootstrap4} to increase the number of stages of bootstrap.
2545 @code{BOOT_CFLAGS} does not apply to bootstrapped target libraries.
2546 Since these are always compiled with the compiler currently being
2547 bootstrapped, you can use @code{CFLAGS_FOR_TARGET} to modify their
2548 compilation flags, as for non-bootstrapped target libraries.
2549 Again, if the native compiler miscompiles the stage1 compiler, you may
2550 need to work around this by avoiding non-working parts of the stage1
2551 compiler. Use @code{STAGE1_TFLAGS} to this end.
2553 If you used the flag @option{--enable-languages=@dots{}} to restrict
2554 the compilers to be built, only those you've actually enabled will be
2555 built. This will of course only build those runtime libraries, for
2556 which the particular compiler has been built. Please note,
2557 that re-defining @env{LANGUAGES} when calling @samp{make}
2558 @strong{does not} work anymore!
2560 If the comparison of stage2 and stage3 fails, this normally indicates
2561 that the stage2 compiler has compiled GCC incorrectly, and is therefore
2562 a potentially serious bug which you should investigate and report. (On
2563 a few systems, meaningful comparison of object files is impossible; they
2564 always appear ``different''. If you encounter this problem, you will
2565 need to disable comparison in the @file{Makefile}.)
2567 If you do not want to bootstrap your compiler, you can configure with
2568 @option{--disable-bootstrap}. In particular cases, you may want to
2569 bootstrap your compiler even if the target system is not the same as
2570 the one you are building on: for example, you could build a
2571 @code{powerpc-unknown-linux-gnu} toolchain on a
2572 @code{powerpc64-unknown-linux-gnu} host. In this case, pass
2573 @option{--enable-bootstrap} to the configure script.
2575 @code{BUILD_CONFIG} can be used to bring in additional customization
2576 to the build. It can be set to a whitespace-separated list of names.
2577 For each such @code{NAME}, top-level @file{config/@code{NAME}.mk} will
2578 be included by the top-level @file{Makefile}, bringing in any settings
2579 it contains. The default @code{BUILD_CONFIG} can be set using the
2580 configure option @option{--with-build-config=@code{NAME}...}. Some
2581 examples of supported build configurations are:
2584 @item @samp{bootstrap-O1}
2585 Removes any @option{-O}-started option from @code{BOOT_CFLAGS}, and adds
2586 @option{-O1} to it. @samp{BUILD_CONFIG=bootstrap-O1} is equivalent to
2587 @samp{BOOT_CFLAGS='-g -O1'}.
2589 @item @samp{bootstrap-O3}
2590 @itemx @samp{bootstrap-Og}
2591 Analogous to @code{bootstrap-O1}.
2593 @item @samp{bootstrap-lto}
2594 Enables Link-Time Optimization for host tools during bootstrapping.
2595 @samp{BUILD_CONFIG=bootstrap-lto} is equivalent to adding
2596 @option{-flto} to @samp{BOOT_CFLAGS}. This option assumes that the host
2597 supports the linker plugin (e.g.@: GNU ld version 2.21 or later or GNU gold
2598 version 2.21 or later).
2600 @item @samp{bootstrap-lto-noplugin}
2601 This option is similar to @code{bootstrap-lto}, but is intended for
2602 hosts that do not support the linker plugin. Without the linker plugin
2603 static libraries are not compiled with link-time optimizations. Since
2604 the GCC middle end and back end are in @file{libbackend.a} this means
2605 that only the front end is actually LTO optimized.
2607 @item @samp{bootstrap-lto-lean}
2608 This option is similar to @code{bootstrap-lto}, but is intended for
2609 faster build by only using LTO in the final bootstrap stage.
2610 With @samp{make profiledbootstrap} the LTO frontend
2611 is trained only on generator files.
2613 @item @samp{bootstrap-debug}
2614 Verifies that the compiler generates the same executable code, whether
2615 or not it is asked to emit debug information. To this end, this
2616 option builds stage2 host programs without debug information, and uses
2617 @file{contrib/compare-debug} to compare them with the stripped stage3
2618 object files. If @code{BOOT_CFLAGS} is overridden so as to not enable
2619 debug information, stage2 will have it, and stage3 won't. This option
2620 is enabled by default when GCC bootstrapping is enabled, if
2621 @code{strip} can turn object files compiled with and without debug
2622 info into identical object files. In addition to better test
2623 coverage, this option makes default bootstraps faster and leaner.
2625 @item @samp{bootstrap-debug-big}
2626 Rather than comparing stripped object files, as in
2627 @code{bootstrap-debug}, this option saves internal compiler dumps
2628 during stage2 and stage3 and compares them as well, which helps catch
2629 additional potential problems, but at a great cost in terms of disk
2630 space. It can be specified in addition to @samp{bootstrap-debug}.
2632 @item @samp{bootstrap-debug-lean}
2633 This option saves disk space compared with @code{bootstrap-debug-big},
2634 but at the expense of some recompilation. Instead of saving the dumps
2635 of stage2 and stage3 until the final compare, it uses
2636 @option{-fcompare-debug} to generate, compare and remove the dumps
2637 during stage3, repeating the compilation that already took place in
2638 stage2, whose dumps were not saved.
2640 @item @samp{bootstrap-debug-lib}
2641 This option tests executable code invariance over debug information
2642 generation on target libraries, just like @code{bootstrap-debug-lean}
2643 tests it on host programs. It builds stage3 libraries with
2644 @option{-fcompare-debug}, and it can be used along with any of the
2645 @code{bootstrap-debug} options above.
2647 There aren't @code{-lean} or @code{-big} counterparts to this option
2648 because most libraries are only build in stage3, so bootstrap compares
2649 would not get significant coverage. Moreover, the few libraries built
2650 in stage2 are used in stage3 host programs, so we wouldn't want to
2651 compile stage2 libraries with different options for comparison purposes.
2653 @item @samp{bootstrap-debug-ckovw}
2654 Arranges for error messages to be issued if the compiler built on any
2655 stage is run without the option @option{-fcompare-debug}. This is
2656 useful to verify the full @option{-fcompare-debug} testing coverage. It
2657 must be used along with @code{bootstrap-debug-lean} and
2658 @code{bootstrap-debug-lib}.
2660 @item @samp{bootstrap-cet}
2661 This option enables Intel CET for host tools during bootstrapping.
2662 @samp{BUILD_CONFIG=bootstrap-cet} is equivalent to adding
2663 @option{-fcf-protection} to @samp{BOOT_CFLAGS}. This option
2664 assumes that the host supports Intel CET (e.g.@: GNU assembler version
2667 @item @samp{bootstrap-time}
2668 Arranges for the run time of each program started by the GCC driver,
2669 built in any stage, to be logged to @file{time.log}, in the top level of
2674 @section Building a cross compiler
2676 When building a cross compiler, it is not generally possible to do a
2677 3-stage bootstrap of the compiler. This makes for an interesting problem
2678 as parts of GCC can only be built with GCC@.
2680 To build a cross compiler, we recommend first building and installing a
2681 native compiler. You can then use the native GCC compiler to build the
2682 cross compiler. The installed native compiler needs to be GCC version
2685 Assuming you have already installed a native copy of GCC and configured
2686 your cross compiler, issue the command @command{make}, which performs the
2691 Build host tools necessary to build the compiler.
2694 Build target tools for use by the compiler such as binutils (bfd,
2695 binutils, gas, gprof, ld, and opcodes)
2696 if they have been individually linked or moved into the top level GCC source
2697 tree before configuring.
2700 Build the compiler (single stage only).
2703 Build runtime libraries using the compiler from the previous step.
2706 Note that if an error occurs in any step the make process will exit.
2708 If you are not building GNU binutils in the same source tree as GCC,
2709 you will need a cross-assembler and cross-linker installed before
2710 configuring GCC@. Put them in the directory
2711 @file{@var{prefix}/@var{target}/bin}. Here is a table of the tools
2712 you should put in this directory:
2716 This should be the cross-assembler.
2719 This should be the cross-linker.
2722 This should be the cross-archiver: a program which can manipulate
2723 archive files (linker libraries) in the target machine's format.
2726 This should be a program to construct a symbol table in an archive file.
2729 The installation of GCC will find these programs in that directory,
2730 and copy or link them to the proper place to for the cross-compiler to
2731 find them when run later.
2733 The easiest way to provide these files is to build the Binutils package.
2734 Configure it with the same @option{--host} and @option{--target}
2735 options that you use for configuring GCC, then build and install
2736 them. They install their executables automatically into the proper
2737 directory. Alas, they do not support all the targets that GCC
2740 If you are not building a C library in the same source tree as GCC,
2741 you should also provide the target libraries and headers before
2742 configuring GCC, specifying the directories with
2743 @option{--with-sysroot} or @option{--with-headers} and
2744 @option{--with-libs}. Many targets also require ``start files'' such
2745 as @file{crt0.o} and
2746 @file{crtn.o} which are linked into each executable. There may be several
2747 alternatives for @file{crt0.o}, for use with profiling or other
2748 compilation options. Check your target's definition of
2749 @code{STARTFILE_SPEC} to find out what start files it uses.
2751 @section Building in parallel
2753 GNU Make 3.80 and above, which is necessary to build GCC, support
2754 building in parallel. To activate this, you can use @samp{make -j 2}
2755 instead of @samp{make}. You can also specify a bigger number, and
2756 in most cases using a value greater than the number of processors in
2757 your machine will result in fewer and shorter I/O latency hits, thus
2758 improving overall throughput; this is especially true for slow drives
2759 and network filesystems.
2761 @section Building the Ada compiler
2764 @ref{GNAT-prerequisite}.
2767 @uref{prerequisites.html#GNAT-prerequisite,,GNAT prerequisites}.
2770 @section Building with profile feedback
2772 It is possible to use profile feedback to optimize the compiler itself. This
2773 should result in a faster compiler binary. Experiments done on x86 using gcc
2774 3.3 showed approximately 7 percent speedup on compiling C programs. To
2775 bootstrap the compiler with profile feedback, use @code{make profiledbootstrap}.
2777 When @samp{make profiledbootstrap} is run, it will first build a @code{stage1}
2778 compiler. This compiler is used to build a @code{stageprofile} compiler
2779 instrumented to collect execution counts of instruction and branch
2780 probabilities. Training run is done by building @code{stagetrain}
2781 compiler. Finally a @code{stagefeedback} compiler is built
2782 using the information collected.
2784 Unlike standard bootstrap, several additional restrictions apply. The
2785 compiler used to build @code{stage1} needs to support a 64-bit integral type.
2786 It is recommended to only use GCC for this.
2788 On Linux/x86_64 hosts with some restrictions (no virtualization) it is
2789 also possible to do autofdo build with @samp{make
2790 autoprofiledback}. This uses Linux perf to sample branches in the
2791 binary and then rebuild it with feedback derived from the profile.
2792 Linux perf and the @code{autofdo} toolkit needs to be installed for
2795 Only the profile from the current build is used, so when an error
2796 occurs it is recommended to clean before restarting. Otherwise
2797 the code quality may be much worse.
2804 @uref{./index.html,,Return to the GCC Installation page}
2808 @c ***Testing*****************************************************************
2810 @comment node-name, next, previous, up
2811 @node Testing, Final install, Building, Installing GCC
2815 @chapter Installing GCC: Testing
2818 @cindex Installing GCC: Testing
2821 Before you install GCC, we encourage you to run the testsuites and to
2822 compare your results with results from a similar configuration that have
2823 been submitted to the
2824 @uref{http://gcc.gnu.org/ml/gcc-testresults/,,gcc-testresults mailing list}.
2825 Some of these archived results are linked from the build status lists
2826 at @uref{http://gcc.gnu.org/buildstat.html}, although not everyone who
2827 reports a successful build runs the testsuites and submits the results.
2828 This step is optional and may require you to download additional software,
2829 but it can give you confidence in your new GCC installation or point out
2830 problems before you install and start using your new GCC@.
2832 First, you must have @uref{download.html,,downloaded the testsuites}.
2833 These are part of the full distribution, but if you downloaded the
2834 ``core'' compiler plus any front ends, you must download the testsuites
2837 Second, you must have the testing tools installed. This includes
2838 @uref{http://www.gnu.org/software/dejagnu/,,DejaGnu}, Tcl, and Expect;
2839 the DejaGnu site has links to these. For running the BRIG frontend
2840 tests, a tool to assemble the binary BRIGs from HSAIL text,
2841 @uref{https://github.com/HSAFoundation/HSAIL-Tools/,,HSAILasm} must
2844 If the directories where @command{runtest} and @command{expect} were
2845 installed are not in the @env{PATH}, you may need to set the following
2846 environment variables appropriately, as in the following example (which
2847 assumes that DejaGnu has been installed under @file{/usr/local}):
2850 TCL_LIBRARY = /usr/local/share/tcl8.0
2851 DEJAGNULIBS = /usr/local/share/dejagnu
2854 (On systems such as Cygwin, these paths are required to be actual
2855 paths, not mounts or links; presumably this is due to some lack of
2856 portability in the DejaGnu code.)
2859 Finally, you can run the testsuite (which may take a long time):
2861 cd @var{objdir}; make -k check
2864 This will test various components of GCC, such as compiler
2865 front ends and runtime libraries. While running the testsuite, DejaGnu
2866 might emit some harmless messages resembling
2867 @samp{WARNING: Couldn't find the global config file.} or
2868 @samp{WARNING: Couldn't find tool init file} that can be ignored.
2870 If you are testing a cross-compiler, you may want to run the testsuite
2871 on a simulator as described at @uref{http://gcc.gnu.org/simtest-howto.html}.
2873 @section How can you run the testsuite on selected tests?
2875 In order to run sets of tests selectively, there are targets
2876 @samp{make check-gcc} and language specific @samp{make check-c},
2877 @samp{make check-c++}, @samp{make check-d} @samp{make check-fortran},
2878 @samp{make check-ada}, @samp{make check-objc}, @samp{make check-obj-c++},
2879 @samp{make check-lto}
2880 in the @file{gcc} subdirectory of the object directory. You can also
2881 just run @samp{make check} in a subdirectory of the object directory.
2884 A more selective way to just run all @command{gcc} execute tests in the
2888 make check-gcc RUNTESTFLAGS="execute.exp @var{other-options}"
2891 Likewise, in order to run only the @command{g++} ``old-deja'' tests in
2892 the testsuite with filenames matching @samp{9805*}, you would use
2895 make check-g++ RUNTESTFLAGS="old-deja.exp=9805* @var{other-options}"
2898 The file-matching expression following @var{filename}@command{.exp=} is treated
2899 as a series of whitespace-delimited glob expressions so that multiple patterns
2900 may be passed, although any whitespace must either be escaped or surrounded by
2901 single quotes if multiple expressions are desired. For example,
2904 make check-g++ RUNTESTFLAGS="old-deja.exp=9805*\ virtual2.c @var{other-options}"
2905 make check-g++ RUNTESTFLAGS="'old-deja.exp=9805* virtual2.c' @var{other-options}"
2908 The @file{*.exp} files are located in the testsuite directories of the GCC
2909 source, the most important ones being @file{compile.exp},
2910 @file{execute.exp}, @file{dg.exp} and @file{old-deja.exp}.
2911 To get a list of the possible @file{*.exp} files, pipe the
2912 output of @samp{make check} into a file and look at the
2913 @samp{Running @dots{} .exp} lines.
2915 @section Passing options and running multiple testsuites
2917 You can pass multiple options to the testsuite using the
2918 @samp{--target_board} option of DejaGNU, either passed as part of
2919 @samp{RUNTESTFLAGS}, or directly to @command{runtest} if you prefer to
2920 work outside the makefiles. For example,
2923 make check-g++ RUNTESTFLAGS="--target_board=unix/-O3/-fmerge-constants"
2926 will run the standard @command{g++} testsuites (``unix'' is the target name
2927 for a standard native testsuite situation), passing
2928 @samp{-O3 -fmerge-constants} to the compiler on every test, i.e.,
2929 slashes separate options.
2931 You can run the testsuites multiple times using combinations of options
2932 with a syntax similar to the brace expansion of popular shells:
2935 @dots{}"--target_board=arm-sim\@{-mhard-float,-msoft-float\@}\@{-O1,-O2,-O3,\@}"
2938 (Note the empty option caused by the trailing comma in the final group.)
2939 The following will run each testsuite eight times using the @samp{arm-sim}
2940 target, as if you had specified all possible combinations yourself:
2943 --target_board='arm-sim/-mhard-float/-O1 \
2944 arm-sim/-mhard-float/-O2 \
2945 arm-sim/-mhard-float/-O3 \
2946 arm-sim/-mhard-float \
2947 arm-sim/-msoft-float/-O1 \
2948 arm-sim/-msoft-float/-O2 \
2949 arm-sim/-msoft-float/-O3 \
2950 arm-sim/-msoft-float'
2953 They can be combined as many times as you wish, in arbitrary ways. This
2957 @dots{}"--target_board=unix/-Wextra\@{-O3,-fno-strength\@}\@{-fomit-frame,\@}"
2960 will generate four combinations, all involving @samp{-Wextra}.
2962 The disadvantage to this method is that the testsuites are run in serial,
2963 which is a waste on multiprocessor systems. For users with GNU Make and
2964 a shell which performs brace expansion, you can run the testsuites in
2965 parallel by having the shell perform the combinations and @command{make}
2966 do the parallel runs. Instead of using @samp{--target_board}, use a
2967 special makefile target:
2970 make -j@var{N} check-@var{testsuite}//@var{test-target}/@var{option1}/@var{option2}/@dots{}
2976 make -j3 check-gcc//sh-hms-sim/@{-m1,-m2,-m3,-m3e,-m4@}/@{,-nofpu@}
2979 will run three concurrent ``make-gcc'' testsuites, eventually testing all
2980 ten combinations as described above. Note that this is currently only
2981 supported in the @file{gcc} subdirectory. (To see how this works, try
2982 typing @command{echo} before the example given here.)
2985 @section How to interpret test results
2987 The result of running the testsuite are various @file{*.sum} and @file{*.log}
2988 files in the testsuite subdirectories. The @file{*.log} files contain a
2989 detailed log of the compiler invocations and the corresponding
2990 results, the @file{*.sum} files summarize the results. These summaries
2991 contain status codes for all tests:
2995 PASS: the test passed as expected
2997 XPASS: the test unexpectedly passed
2999 FAIL: the test unexpectedly failed
3001 XFAIL: the test failed as expected
3003 UNSUPPORTED: the test is not supported on this platform
3005 ERROR: the testsuite detected an error
3007 WARNING: the testsuite detected a possible problem
3010 It is normal for some tests to report unexpected failures. At the
3011 current time the testing harness does not allow fine grained control
3012 over whether or not a test is expected to fail. This problem should
3013 be fixed in future releases.
3016 @section Submitting test results
3018 If you want to report the results to the GCC project, use the
3019 @file{contrib/test_summary} shell script. Start it in the @var{objdir} with
3022 @var{srcdir}/contrib/test_summary -p your_commentary.txt \
3023 -m gcc-testresults@@gcc.gnu.org |sh
3026 This script uses the @command{Mail} program to send the results, so
3027 make sure it is in your @env{PATH}. The file @file{your_commentary.txt} is
3028 prepended to the testsuite summary and should contain any special
3029 remarks you have on your results or your build environment. Please
3030 do not edit the testsuite result block or the subject line, as these
3031 messages may be automatically processed.
3038 @uref{./index.html,,Return to the GCC Installation page}
3042 @c ***Final install***********************************************************
3044 @comment node-name, next, previous, up
3045 @node Final install, , Testing, Installing GCC
3047 @ifset finalinstallhtml
3049 @chapter Installing GCC: Final installation
3052 Now that GCC has been built (and optionally tested), you can install it with
3054 cd @var{objdir} && make install
3057 We strongly recommend to install into a target directory where there is
3058 no previous version of GCC present. Also, the GNAT runtime should not
3059 be stripped, as this would break certain features of the debugger that
3060 depend on this debugging information (catching Ada exceptions for
3063 That step completes the installation of GCC; user level binaries can
3064 be found in @file{@var{prefix}/bin} where @var{prefix} is the value
3065 you specified with the @option{--prefix} to configure (or
3066 @file{/usr/local} by default). (If you specified @option{--bindir},
3067 that directory will be used instead; otherwise, if you specified
3068 @option{--exec-prefix}, @file{@var{exec-prefix}/bin} will be used.)
3069 Headers for the C++ library are installed in
3070 @file{@var{prefix}/include}; libraries in @file{@var{libdir}}
3071 (normally @file{@var{prefix}/lib}); internal parts of the compiler in
3072 @file{@var{libdir}/gcc} and @file{@var{libexecdir}/gcc}; documentation
3073 in info format in @file{@var{infodir}} (normally
3074 @file{@var{prefix}/info}).
3076 When installing cross-compilers, GCC's executables
3077 are not only installed into @file{@var{bindir}}, that
3078 is, @file{@var{exec-prefix}/bin}, but additionally into
3079 @file{@var{exec-prefix}/@var{target-alias}/bin}, if that directory
3080 exists. Typically, such @dfn{tooldirs} hold target-specific
3081 binutils, including assembler and linker.
3083 Installation into a temporary staging area or into a @command{chroot}
3084 jail can be achieved with the command
3087 make DESTDIR=@var{path-to-rootdir} install
3091 where @var{path-to-rootdir} is the absolute path of
3092 a directory relative to which all installation paths will be
3093 interpreted. Note that the directory specified by @code{DESTDIR}
3094 need not exist yet; it will be created if necessary.
3096 There is a subtle point with tooldirs and @code{DESTDIR}:
3097 If you relocate a cross-compiler installation with
3098 e.g.@: @samp{DESTDIR=@var{rootdir}}, then the directory
3099 @file{@var{rootdir}/@var{exec-prefix}/@var{target-alias}/bin} will
3100 be filled with duplicated GCC executables only if it already exists,
3101 it will not be created otherwise. This is regarded as a feature,
3102 not as a bug, because it gives slightly more control to the packagers
3103 using the @code{DESTDIR} feature.
3105 You can install stripped programs and libraries with
3111 If you are bootstrapping a released version of GCC then please
3112 quickly review the build status page for your release, available from
3113 @uref{http://gcc.gnu.org/buildstat.html}.
3114 If your system is not listed for the version of GCC that you built,
3116 @email{gcc@@gcc.gnu.org} indicating
3117 that you successfully built and installed GCC@.
3118 Include the following information:
3122 Output from running @file{@var{srcdir}/config.guess}. Do not send
3123 that file itself, just the one-line output from running it.
3126 The output of @samp{gcc -v} for your newly installed @command{gcc}.
3127 This tells us which version of GCC you built and the options you passed to
3131 Whether you enabled all languages or a subset of them. If you used a
3132 full distribution then this information is part of the configure
3133 options in the output of @samp{gcc -v}, but if you downloaded the
3134 ``core'' compiler plus additional front ends then it isn't apparent
3135 which ones you built unless you tell us about it.
3138 If the build was for GNU/Linux, also include:
3141 The distribution name and version (e.g., Red Hat 7.1 or Debian 2.2.3);
3142 this information should be available from @file{/etc/issue}.
3145 The version of the Linux kernel, available from @samp{uname --version}
3149 The version of glibc you used; for RPM-based systems like Red Hat,
3150 Mandrake, and SuSE type @samp{rpm -q glibc} to get the glibc version,
3151 and on systems like Debian and Progeny use @samp{dpkg -l libc6}.
3153 For other systems, you can include similar information if you think it is
3157 Any other information that you think would be useful to people building
3158 GCC on the same configuration. The new entry in the build status list
3159 will include a link to the archived copy of your message.
3162 We'd also like to know if the
3164 @ref{Specific, host/target specific installation notes}
3167 @uref{specific.html,,host/target specific installation notes}
3169 didn't include your host/target information or if that information is
3170 incomplete or out of date. Send a note to
3171 @email{gcc@@gcc.gnu.org} detailing how the information should be changed.
3173 If you find a bug, please report it following the
3174 @uref{../bugs/,,bug reporting guidelines}.
3176 If you want to print the GCC manuals, do @samp{cd @var{objdir}; make
3177 dvi}. You will need to have @command{texi2dvi} (version at least 4.7)
3178 and @TeX{} installed. This creates a number of @file{.dvi} files in
3179 subdirectories of @file{@var{objdir}}; these may be converted for
3180 printing with programs such as @command{dvips}. Alternately, by using
3181 @samp{make pdf} in place of @samp{make dvi}, you can create documentation
3182 in the form of @file{.pdf} files; this requires @command{texi2pdf}, which
3183 is included with Texinfo version 4.8 and later. You can also
3184 @uref{https://shop.fsf.org/,,buy printed manuals from the
3185 Free Software Foundation}, though such manuals may not be for the most
3186 recent version of GCC@.
3188 If you would like to generate online HTML documentation, do @samp{cd
3189 @var{objdir}; make html} and HTML will be generated for the gcc manuals in
3190 @file{@var{objdir}/gcc/HTML}.
3197 @uref{./index.html,,Return to the GCC Installation page}
3201 @c ***Binaries****************************************************************
3203 @comment node-name, next, previous, up
3204 @node Binaries, Specific, Installing GCC, Top
3208 @chapter Installing GCC: Binaries
3211 @cindex Installing GCC: Binaries
3213 We are often asked about pre-compiled versions of GCC@. While we cannot
3214 provide these for all platforms, below you'll find links to binaries for
3215 various platforms where creating them by yourself is not easy due to various
3218 Please note that we did not create these binaries, nor do we
3219 support them. If you have any problems installing them, please
3220 contact their makers.
3227 @uref{http://www.bullfreeware.com,,Bull's Open Source Software Archive for
3228 for AIX 5L and AIX 6};
3231 @uref{http://www.perzl.org/aix/,,AIX Open Source Packages (AIX5L AIX 6.1
3236 DOS---@uref{http://www.delorie.com/djgpp/,,DJGPP}.
3242 @uref{http://hpux.connect.org.uk/,,HP-UX Porting Center};
3246 Solaris 2 (SPARC, Intel):
3249 @uref{https://www.opencsw.org/,,OpenCSW}
3252 @uref{http://jupiterrise.com/tgcware/,,TGCware}
3259 The @uref{https://brew.sh,,Homebrew} package manager;
3261 @uref{https://www.macports.org,,MacPorts}.
3268 The @uref{https://sourceware.org/cygwin/,,Cygwin} project;
3270 The @uref{http://www.mingw.org/,,MinGW} and
3271 @uref{http://mingw-w64.org/doku.php,,mingw-w64} projects.
3275 @uref{http://www.openpkg.org/,,OpenPKG} offers binaries for quite a
3276 number of platforms.
3279 The @uref{http://gcc.gnu.org/wiki/GFortranBinaries,,GFortran Wiki} has
3280 links to GNU Fortran binaries for several platforms.
3288 @uref{./index.html,,Return to the GCC Installation page}
3292 @c ***Specific****************************************************************
3294 @comment node-name, next, previous, up
3295 @node Specific, Old, Binaries, Top
3299 @chapter Host/target specific installation notes for GCC
3302 @cindex Specific installation notes
3303 @cindex Target specific installation
3304 @cindex Host specific installation
3305 @cindex Target specific installation notes
3307 Please read this document carefully @emph{before} installing the
3308 GNU Compiler Collection on your machine.
3310 Note that this list of install notes is @emph{not} a list of supported
3311 hosts or targets. Not all supported hosts and targets are listed
3312 here, only the ones that require host-specific or target-specific
3313 information have to.
3318 @uref{#aarch64-x-x,,aarch64*-*-*}
3320 @uref{#alpha-x-x,,alpha*-*-*}
3322 @uref{#amd64-x-solaris2,,amd64-*-solaris2*}
3324 @uref{#arm-x-eabi,,arm-*-eabi}
3328 @uref{#bfin,,Blackfin}
3332 @uref{#x-x-freebsd,,*-*-freebsd*}
3334 @uref{#h8300-hms,,h8300-hms}
3336 @uref{#hppa-hp-hpux,,hppa*-hp-hpux*}
3338 @uref{#hppa-hp-hpux10,,hppa*-hp-hpux10}
3340 @uref{#hppa-hp-hpux11,,hppa*-hp-hpux11}
3342 @uref{#x-x-linux-gnu,,*-*-linux-gnu}
3344 @uref{#ix86-x-linux,,i?86-*-linux*}
3346 @uref{#ix86-x-solaris2,,i?86-*-solaris2*}
3348 @uref{#ia64-x-linux,,ia64-*-linux}
3350 @uref{#ia64-x-hpux,,ia64-*-hpux*}
3352 @uref{#x-ibm-aix,,*-ibm-aix*}
3354 @uref{#iq2000-x-elf,,iq2000-*-elf}
3356 @uref{#lm32-x-elf,,lm32-*-elf}
3358 @uref{#lm32-x-uclinux,,lm32-*-uclinux}
3360 @uref{#m32c-x-elf,,m32c-*-elf}
3362 @uref{#m32r-x-elf,,m32r-*-elf}
3364 @uref{#m68k-x-x,,m68k-*-*}
3366 @uref{#m68k-uclinux,,m68k-uclinux}
3368 @uref{#microblaze-x-elf,,microblaze-*-elf}
3370 @uref{#mips-x-x,,mips-*-*}
3372 @uref{#nds32le-x-elf,,nds32le-*-elf}
3374 @uref{#nds32be-x-elf,,nds32be-*-elf}
3376 @uref{#nvptx-x-none,,nvptx-*-none}
3378 @uref{#or1k-x-elf,,or1k-*-elf}
3380 @uref{#or1k-x-linux,,or1k-*-linux}
3382 @uref{#powerpc-x-x,,powerpc*-*-*}
3384 @uref{#powerpc-x-darwin,,powerpc-*-darwin*}
3386 @uref{#powerpc-x-elf,,powerpc-*-elf}
3388 @uref{#powerpc-x-linux-gnu,,powerpc*-*-linux-gnu*}
3390 @uref{#powerpc-x-netbsd,,powerpc-*-netbsd*}
3392 @uref{#powerpc-x-eabisim,,powerpc-*-eabisim}
3394 @uref{#powerpc-x-eabi,,powerpc-*-eabi}
3396 @uref{#powerpcle-x-elf,,powerpcle-*-elf}
3398 @uref{#powerpcle-x-eabisim,,powerpcle-*-eabisim}
3400 @uref{#powerpcle-x-eabi,,powerpcle-*-eabi}
3402 @uref{#riscv32-x-elf,,riscv32-*-elf}
3404 @uref{#riscv32-x-linux,,riscv32-*-linux}
3406 @uref{#riscv64-x-elf,,riscv64-*-elf}
3408 @uref{#riscv64-x-linux,,riscv64-*-linux}
3410 @uref{#s390-x-linux,,s390-*-linux*}
3412 @uref{#s390x-x-linux,,s390x-*-linux*}
3414 @uref{#s390x-ibm-tpf,,s390x-ibm-tpf*}
3416 @uref{#x-x-solaris2,,*-*-solaris2*}
3418 @uref{#sparc-x-x,,sparc*-*-*}
3420 @uref{#sparc-sun-solaris2,,sparc-sun-solaris2*}
3422 @uref{#sparc-x-linux,,sparc-*-linux*}
3424 @uref{#sparc64-x-solaris2,,sparc64-*-solaris2*}
3426 @uref{#sparcv9-x-solaris2,,sparcv9-*-solaris2*}
3428 @uref{#c6x-x-x,,c6x-*-*}
3430 @uref{#tilegx-x-linux,,tilegx-*-linux*}
3432 @uref{#tilegxbe-x-linux,,tilegxbe-*-linux*}
3434 @uref{#tilepro-x-linux,,tilepro-*-linux*}
3436 @uref{#visium-x-elf, visium-*-elf}
3438 @uref{#x-x-vxworks,,*-*-vxworks*}
3440 @uref{#x86-64-x-x,,x86_64-*-*, amd64-*-*}
3442 @uref{#x86-64-x-solaris2,,x86_64-*-solaris2*}
3444 @uref{#xtensa-x-elf,,xtensa*-*-elf}
3446 @uref{#xtensa-x-linux,,xtensa*-*-linux*}
3448 @uref{#windows,,Microsoft Windows}
3450 @uref{#x-x-cygwin,,*-*-cygwin}
3452 @uref{#x-x-mingw32,,*-*-mingw32}
3456 @uref{#older,,Older systems}
3461 @uref{#elf,,all ELF targets} (SVR4, Solaris 2, etc.)
3467 <!-- -------- host/target specific issues start here ---------------- -->
3470 @anchor{aarch64-x-x}
3471 @heading aarch64*-*-*
3472 Binutils pre 2.24 does not have support for selecting @option{-mabi} and
3473 does not support ILP32. If it is used to build GCC 4.9 or later, GCC will
3474 not support option @option{-mabi=ilp32}.
3476 To enable a workaround for the Cortex-A53 erratum number 835769 by default
3477 (for all CPUs regardless of -mcpu option given) at configure time use the
3478 @option{--enable-fix-cortex-a53-835769} option. This will enable the fix by
3479 default and can be explicitly disabled during compilation by passing the
3480 @option{-mno-fix-cortex-a53-835769} option. Conversely,
3481 @option{--disable-fix-cortex-a53-835769} will disable the workaround by
3482 default. The workaround is disabled by default if neither of
3483 @option{--enable-fix-cortex-a53-835769} or
3484 @option{--disable-fix-cortex-a53-835769} is given at configure time.
3486 To enable a workaround for the Cortex-A53 erratum number 843419 by default
3487 (for all CPUs regardless of -mcpu option given) at configure time use the
3488 @option{--enable-fix-cortex-a53-843419} option. This workaround is applied at
3489 link time. Enabling the workaround will cause GCC to pass the relevant option
3490 to the linker. It can be explicitly disabled during compilation by passing the
3491 @option{-mno-fix-cortex-a53-843419} option. Conversely,
3492 @option{--disable-fix-cortex-a53-843419} will disable the workaround by default.
3493 The workaround is disabled by default if neither of
3494 @option{--enable-fix-cortex-a53-843419} or
3495 @option{--disable-fix-cortex-a53-843419} is given at configure time.
3497 To enable Branch Target Identification Mechanism and Return Address Signing by
3498 default at configure time use the @option{--enable-standard-branch-protection}
3499 option. This is equivalent to having @option{-mbranch-protection=standard}
3500 during compilation. This can be explicitly disabled during compilation by
3501 passing the @option{-mbranch-protection=none} option which turns off all
3502 types of branch protections. Conversely,
3503 @option{--disable-standard-branch-protection} will disable both the
3504 protections by default. This mechanism is turned off by default if neither
3505 of the options are given at configure time.
3512 This section contains general configuration information for all
3513 Alpha-based platforms using ELF@. In addition to reading this
3514 section, please read all other sections that match your target.
3519 @anchor{amd64-x-solaris2}
3520 @heading amd64-*-solaris2*
3521 This is a synonym for @samp{x86_64-*-solaris2*}.
3526 @anchor{amdgcn-unknown-amdhsa}
3527 @heading amdgcn-unknown-amdhsa
3530 Instead of GNU Binutils, you will need to install LLVM 6, or later, and copy
3531 @file{bin/llvm-mc} to @file{amdgcn-unknown-amdhsa/bin/as},
3532 @file{bin/lld} to @file{amdgcn-unknown-amdhsa/bin/ld},
3533 @file{bin/llvm-nm} to @file{amdgcn-unknown-amdhsa/bin/nm}, and
3534 @file{bin/llvm-ar} to both @file{bin/amdgcn-unknown-amdhsa-ar} and
3535 @file{bin/amdgcn-unknown-amdhsa-ranlib}.
3537 Use Newlib (2019-01-16, or newer).
3539 To run the binaries, install the HSA Runtime from the
3540 @uref{https://rocm.github.io,,ROCm Platform}, and use
3541 @file{libexec/gcc/amdhsa-unknown-amdhsa/@var{version}/gcn-run} to launch them
3547 @anchor{arc-x-elf32}
3548 @heading arc-*-elf32
3550 Use @samp{configure --target=arc-elf32 --with-cpu=@var{cpu} --enable-languages="c,c++"}
3551 to configure GCC, with @var{cpu} being one of @samp{arc600}, @samp{arc601},
3557 @anchor{arc-linux-uclibc}
3558 @heading arc-linux-uclibc
3560 Use @samp{configure --target=arc-linux-uclibc --with-cpu=arc700 --enable-languages="c,c++"} to configure GCC@.
3567 ARM-family processors.
3569 Building the Ada frontend commonly fails (an infinite loop executing
3570 @code{xsinfo}) if the host compiler is GNAT 4.8. Host compilers built from the
3571 GNAT 4.6, 4.9 or 5 release branches are known to succeed.
3578 ATMEL AVR-family micro controllers. These are used in embedded
3579 applications. There are no standard Unix configurations.
3581 @xref{AVR Options,, AVR Options, gcc, Using the GNU Compiler
3585 See ``AVR Options'' in the main manual
3587 for the list of supported MCU types.
3589 Use @samp{configure --target=avr --enable-languages="c"} to configure GCC@.
3591 Further installation notes and other useful information about AVR tools
3592 can also be obtained from:
3596 @uref{http://www.nongnu.org/avr/,,http://www.nongnu.org/avr/}
3598 @uref{http://www.amelek.gda.pl/avr/,,http://www.amelek.gda.pl/avr/}
3601 The following error:
3603 Error: register required
3606 indicates that you should upgrade to a newer version of the binutils.
3613 The Blackfin processor, an Analog Devices DSP.
3615 @xref{Blackfin Options,, Blackfin Options, gcc, Using the GNU Compiler
3619 See ``Blackfin Options'' in the main manual
3622 More information, and a version of binutils with support for this processor,
3623 are available at @uref{https://sourceforge.net/projects/adi-toolchain/}.
3630 The CR16 CompactRISC architecture is a 16-bit architecture. This
3631 architecture is used in embedded applications.
3634 @xref{CR16 Options,, CR16 Options, gcc, Using and Porting the GNU Compiler
3639 See ``CR16 Options'' in the main manual for a list of CR16-specific options.
3642 Use @samp{configure --target=cr16-elf --enable-languages=c,c++} to configure
3643 GCC@ for building a CR16 elf cross-compiler.
3645 Use @samp{configure --target=cr16-uclinux --enable-languages=c,c++} to
3646 configure GCC@ for building a CR16 uclinux cross-compiler.
3653 CRIS is the CPU architecture in Axis Communications ETRAX system-on-a-chip
3654 series. These are used in embedded applications.
3657 @xref{CRIS Options,, CRIS Options, gcc, Using the GNU Compiler
3661 See ``CRIS Options'' in the main manual
3663 for a list of CRIS-specific options.
3665 There are a few different CRIS targets:
3668 Mainly for monolithic embedded systems. Includes a multilib for the
3669 @samp{v10} core used in @samp{ETRAX 100 LX}.
3670 @item cris-axis-linux-gnu
3671 A GNU/Linux port for the CRIS architecture, currently targeting
3672 @samp{ETRAX 100 LX} by default.
3675 Pre-packaged tools can be obtained from
3676 @uref{ftp://ftp.axis.com/@/pub/@/axis/@/tools/@/cris/@/compiler-kit/}. More
3677 information about this platform is available at
3678 @uref{http://developer.axis.com/}.
3685 Please have a look at the @uref{binaries.html,,binaries page}.
3687 You cannot install GCC by itself on MSDOS; it will not compile under
3688 any MSDOS compiler except itself. You need to get the complete
3689 compilation package DJGPP, which includes binaries as well as sources,
3690 and includes all the necessary compilation tools and libraries.
3695 @anchor{epiphany-x-elf}
3696 @heading epiphany-*-elf
3698 This configuration is intended for embedded systems.
3703 @anchor{x-x-freebsd}
3704 @heading *-*-freebsd*
3705 Support for FreeBSD 1 was discontinued in GCC 3.2. Support for
3706 FreeBSD 2 (and any mutant a.out variants of FreeBSD 3) was
3707 discontinued in GCC 4.0.
3709 In order to better utilize FreeBSD base system functionality and match
3710 the configuration of the system compiler, GCC 4.5 and above as well as
3711 GCC 4.4 past 2010-06-20 leverage SSP support in libc (which is present
3712 on FreeBSD 7 or later) and the use of @code{__cxa_atexit} by default
3713 (on FreeBSD 6 or later). The use of @code{dl_iterate_phdr} inside
3714 @file{libgcc_s.so.1} and boehm-gc (on FreeBSD 7 or later) is enabled
3715 by GCC 4.5 and above.
3717 We support FreeBSD using the ELF file format with DWARF 2 debugging
3718 for all CPU architectures. You may use @option{-gstabs} instead of
3719 @option{-g}, if you really want the old debugging format. There are
3720 no known issues with mixing object files and libraries with different
3721 debugging formats. Otherwise, this release of GCC should now match
3722 more of the configuration used in the stock FreeBSD configuration of
3723 GCC@. In particular, @option{--enable-threads} is now configured by
3724 default. However, as a general user, do not attempt to replace the
3725 system compiler with this release. Known to bootstrap and check with
3726 good results on FreeBSD 7.2-STABLE@. In the past, known to bootstrap
3727 and check with good results on FreeBSD 3.0, 3.4, 4.0, 4.2, 4.3, 4.4,
3728 4.5, 4.8, 4.9 and 5-CURRENT@.
3730 The version of binutils installed in @file{/usr/bin} probably works
3731 with this release of GCC@. Bootstrapping against the latest GNU
3732 binutils and/or the version found in @file{/usr/ports/devel/binutils} has
3733 been known to enable additional features and improve overall testsuite
3734 results. However, it is currently known that boehm-gc may not configure
3735 properly on FreeBSD prior to the FreeBSD 7.0 release with GNU binutils
3744 This configuration is intended for embedded systems.
3751 Renesas H8/300 series of processors.
3753 Please have a look at the @uref{binaries.html,,binaries page}.
3755 The calling convention and structure layout has changed in release 2.6.
3756 All code must be recompiled. The calling convention now passes the
3757 first three arguments in function calls in registers. Structures are no
3758 longer a multiple of 2 bytes.
3763 @anchor{hppa-hp-hpux}
3764 @heading hppa*-hp-hpux*
3765 Support for HP-UX version 9 and older was discontinued in GCC 3.4.
3767 We require using gas/binutils on all hppa platforms. Version 2.19 or
3768 later is recommended.
3770 It may be helpful to configure GCC with the
3771 @uref{./configure.html#with-gnu-as,,@option{--with-gnu-as}} and
3772 @option{--with-as=@dots{}} options to ensure that GCC can find GAS@.
3774 The HP assembler should not be used with GCC. It is rarely tested and may
3775 not work. It shouldn't be used with any languages other than C due to its
3778 Specifically, @option{-g} does not work (HP-UX uses a peculiar debugging
3779 format which GCC does not know about). It also inserts timestamps
3780 into each object file it creates, causing the 3-stage comparison test to
3781 fail during a bootstrap. You should be able to continue by saying
3782 @samp{make all-host all-target} after getting the failure from @samp{make}.
3784 Various GCC features are not supported. For example, it does not support weak
3785 symbols or alias definitions. As a result, explicit template instantiations
3786 are required when using C++. This makes it difficult if not impossible to
3787 build many C++ applications.
3789 There are two default scheduling models for instructions. These are
3790 PROCESSOR_7100LC and PROCESSOR_8000. They are selected from the pa-risc
3791 architecture specified for the target machine when configuring.
3792 PROCESSOR_8000 is the default. PROCESSOR_7100LC is selected when
3793 the target is a @samp{hppa1*} machine.
3795 The PROCESSOR_8000 model is not well suited to older processors. Thus,
3796 it is important to completely specify the machine architecture when
3797 configuring if you want a model other than PROCESSOR_8000. The macro
3798 TARGET_SCHED_DEFAULT can be defined in BOOT_CFLAGS if a different
3799 default scheduling model is desired.
3801 As of GCC 4.0, GCC uses the UNIX 95 namespace for HP-UX 10.10
3802 through 11.00, and the UNIX 98 namespace for HP-UX 11.11 and later.
3803 This namespace change might cause problems when bootstrapping with
3804 an earlier version of GCC or the HP compiler as essentially the same
3805 namespace is required for an entire build. This problem can be avoided
3806 in a number of ways. With HP cc, @env{UNIX_STD} can be set to @samp{95}
3807 or @samp{98}. Another way is to add an appropriate set of predefines
3808 to @env{CC}. The description for the @option{munix=} option contains
3809 a list of the predefines used with each standard.
3811 More specific information to @samp{hppa*-hp-hpux*} targets follows.
3816 @anchor{hppa-hp-hpux10}
3817 @heading hppa*-hp-hpux10
3818 For hpux10.20, we @emph{highly} recommend you pick up the latest sed patch
3819 @code{PHCO_19798} from HP@.
3821 The C++ ABI has changed incompatibly in GCC 4.0. COMDAT subspaces are
3822 used for one-only code and data. This resolves many of the previous
3823 problems in using C++ on this target. However, the ABI is not compatible
3824 with the one implemented under HP-UX 11 using secondary definitions.
3829 @anchor{hppa-hp-hpux11}
3830 @heading hppa*-hp-hpux11
3831 GCC 3.0 and up support HP-UX 11. GCC 2.95.x is not supported and cannot
3832 be used to compile GCC 3.0 and up.
3834 The libffi library haven't been ported to 64-bit HP-UX@ and doesn't build.
3836 Refer to @uref{binaries.html,,binaries} for information about obtaining
3837 precompiled GCC binaries for HP-UX@. Precompiled binaries must be obtained
3838 to build the Ada language as it cannot be bootstrapped using C@. Ada is
3839 only available for the 32-bit PA-RISC runtime.
3841 Starting with GCC 3.4 an ISO C compiler is required to bootstrap. The
3842 bundled compiler supports only traditional C; you will need either HP's
3843 unbundled compiler, or a binary distribution of GCC@.
3845 It is possible to build GCC 3.3 starting with the bundled HP compiler,
3846 but the process requires several steps. GCC 3.3 can then be used to
3847 build later versions.
3849 There are several possible approaches to building the distribution.
3850 Binutils can be built first using the HP tools. Then, the GCC
3851 distribution can be built. The second approach is to build GCC
3852 first using the HP tools, then build binutils, then rebuild GCC@.
3853 There have been problems with various binary distributions, so it
3854 is best not to start from a binary distribution.
3856 On 64-bit capable systems, there are two distinct targets. Different
3857 installation prefixes must be used if both are to be installed on
3858 the same system. The @samp{hppa[1-2]*-hp-hpux11*} target generates code
3859 for the 32-bit PA-RISC runtime architecture and uses the HP linker.
3860 The @samp{hppa64-hp-hpux11*} target generates 64-bit code for the
3861 PA-RISC 2.0 architecture.
3863 The script config.guess now selects the target type based on the compiler
3864 detected during configuration. You must define @env{PATH} or @env{CC} so
3865 that configure finds an appropriate compiler for the initial bootstrap.
3866 When @env{CC} is used, the definition should contain the options that are
3867 needed whenever @env{CC} is used.
3869 Specifically, options that determine the runtime architecture must be
3870 in @env{CC} to correctly select the target for the build. It is also
3871 convenient to place many other compiler options in @env{CC}. For example,
3872 @env{CC="cc -Ac +DA2.0W -Wp,-H16376 -D_CLASSIC_TYPES -D_HPUX_SOURCE"}
3873 can be used to bootstrap the GCC 3.3 branch with the HP compiler in
3874 64-bit K&R/bundled mode. The @option{+DA2.0W} option will result in
3875 the automatic selection of the @samp{hppa64-hp-hpux11*} target. The
3876 macro definition table of cpp needs to be increased for a successful
3877 build with the HP compiler. _CLASSIC_TYPES and _HPUX_SOURCE need to
3878 be defined when building with the bundled compiler, or when using the
3879 @option{-Ac} option. These defines aren't necessary with @option{-Ae}.
3881 It is best to explicitly configure the @samp{hppa64-hp-hpux11*} target
3882 with the @option{--with-ld=@dots{}} option. This overrides the standard
3883 search for ld. The two linkers supported on this target require different
3884 commands. The default linker is determined during configuration. As a
3885 result, it's not possible to switch linkers in the middle of a GCC build.
3886 This has been reported to sometimes occur in unified builds of binutils
3889 A recent linker patch must be installed for the correct operation of
3890 GCC 3.3 and later. @code{PHSS_26559} and @code{PHSS_24304} are the
3891 oldest linker patches that are known to work. They are for HP-UX
3892 11.00 and 11.11, respectively. @code{PHSS_24303}, the companion to
3893 @code{PHSS_24304}, might be usable but it hasn't been tested. These
3894 patches have been superseded. Consult the HP patch database to obtain
3895 the currently recommended linker patch for your system.
3897 The patches are necessary for the support of weak symbols on the
3898 32-bit port, and for the running of initializers and finalizers. Weak
3899 symbols are implemented using SOM secondary definition symbols. Prior
3900 to HP-UX 11, there are bugs in the linker support for secondary symbols.
3901 The patches correct a problem of linker core dumps creating shared
3902 libraries containing secondary symbols, as well as various other
3903 linking issues involving secondary symbols.
3905 GCC 3.3 uses the ELF DT_INIT_ARRAY and DT_FINI_ARRAY capabilities to
3906 run initializers and finalizers on the 64-bit port. The 32-bit port
3907 uses the linker @option{+init} and @option{+fini} options for the same
3908 purpose. The patches correct various problems with the +init/+fini
3909 options, including program core dumps. Binutils 2.14 corrects a
3910 problem on the 64-bit port resulting from HP's non-standard use of
3911 the .init and .fini sections for array initializers and finalizers.
3913 Although the HP and GNU linkers are both supported for the
3914 @samp{hppa64-hp-hpux11*} target, it is strongly recommended that the
3915 HP linker be used for link editing on this target.
3917 At this time, the GNU linker does not support the creation of long
3918 branch stubs. As a result, it cannot successfully link binaries
3919 containing branch offsets larger than 8 megabytes. In addition,
3920 there are problems linking shared libraries, linking executables
3921 with @option{-static}, and with dwarf2 unwind and exception support.
3922 It also doesn't provide stubs for internal calls to global functions
3923 in shared libraries, so these calls cannot be overloaded.
3925 The HP dynamic loader does not support GNU symbol versioning, so symbol
3926 versioning is not supported. It may be necessary to disable symbol
3927 versioning with @option{--disable-symvers} when using GNU ld.
3929 POSIX threads are the default. The optional DCE thread library is not
3930 supported, so @option{--enable-threads=dce} does not work.
3935 @anchor{x-x-linux-gnu}
3936 @heading *-*-linux-gnu
3937 Versions of libstdc++-v3 starting with 3.2.1 require bug fixes present
3938 in glibc 2.2.5 and later. More information is available in the
3939 libstdc++-v3 documentation.
3944 @anchor{ix86-x-linux}
3945 @heading i?86-*-linux*
3946 As of GCC 3.3, binutils 2.13.1 or later is required for this platform.
3947 See @uref{http://gcc.gnu.org/PR10877,,bug 10877} for more information.
3949 If you receive Signal 11 errors when building on GNU/Linux, then it is
3950 possible you have a hardware problem. Further information on this can be
3951 found on @uref{http://www.bitwizard.nl/sig11/,,www.bitwizard.nl}.
3956 @anchor{ix86-x-solaris2}
3957 @heading i?86-*-solaris2*
3958 Use this for Solaris 11 or later on x86 and x86-64 systems. Starting
3959 with GCC 4.7, there is also a 64-bit @samp{amd64-*-solaris2*} or
3960 @samp{x86_64-*-solaris2*} configuration that corresponds to
3961 @samp{sparcv9-sun-solaris2*}.
3963 It is recommended that you configure GCC to use the GNU assembler. The
3964 versions included in Solaris 11, from GNU binutils 2.19 or
3965 newer (also available as @file{/usr/bin/gas} and
3966 @file{/usr/gnu/bin/as}), work fine. The current version, from GNU
3967 binutils 2.32, is known to work, but the version from GNU binutils 2.26
3968 must be avoided. Recent versions of the Solaris assembler in
3969 @file{/usr/ccs/bin/as} work almost as well, though.
3970 @c FIXME: as patch requirements?
3972 For linking, the Solaris linker, is preferred. If you want to use the GNU
3973 linker instead, the version in Solaris 11, from GNU binutils 2.19 or
3974 newer (also in @file{/usr/gnu/bin/ld} and @file{/usr/bin/gld}), works,
3975 as does the latest version, from GNU binutils 2.32.
3977 To use GNU @command{as}, configure with the options
3978 @option{--with-gnu-as --with-as=@//usr/@/gnu/@/bin/@/as}. It may be necessary
3979 to configure with @option{--without-gnu-ld --with-ld=@//usr/@/ccs/@/bin/@/ld} to
3980 guarantee use of Sun @command{ld}.
3981 @c FIXME: why --without-gnu-ld --with-ld?
3986 @anchor{ia64-x-linux}
3987 @heading ia64-*-linux
3988 IA-64 processor (also known as IPF, or Itanium Processor Family)
3991 If you are using the installed system libunwind library with
3992 @option{--with-system-libunwind}, then you must use libunwind 0.98 or
3995 None of the following versions of GCC has an ABI that is compatible
3996 with any of the other versions in this list, with the exception that
3997 Red Hat 2.96 and Trillian 000171 are compatible with each other:
3998 3.1, 3.0.2, 3.0.1, 3.0, Red Hat 2.96, and Trillian 000717.
3999 This primarily affects C++ programs and programs that create shared libraries.
4000 GCC 3.1 or later is recommended for compiling linux, the kernel.
4001 As of version 3.1 GCC is believed to be fully ABI compliant, and hence no
4002 more major ABI changes are expected.
4007 @anchor{ia64-x-hpux}
4008 @heading ia64-*-hpux*
4009 Building GCC on this target requires the GNU Assembler. The bundled HP
4010 assembler will not work. To prevent GCC from using the wrong assembler,
4011 the option @option{--with-gnu-as} may be necessary.
4013 The GCC libunwind library has not been ported to HPUX@. This means that for
4014 GCC versions 3.2.3 and earlier, @option{--enable-libunwind-exceptions}
4015 is required to build GCC@. For GCC 3.3 and later, this is the default.
4016 For gcc 3.4.3 and later, @option{--enable-libunwind-exceptions} is
4017 removed and the system libunwind library will always be used.
4021 <!-- rs6000-ibm-aix*, powerpc-ibm-aix* -->
4025 Support for AIX version 3 and older was discontinued in GCC 3.4.
4026 Support for AIX version 4.2 and older was discontinued in GCC 4.5.
4028 ``out of memory'' bootstrap failures may indicate a problem with
4029 process resource limits (ulimit). Hard limits are configured in the
4030 @file{/etc/security/limits} system configuration file.
4032 GCC 4.9 and above require a C++ compiler for bootstrap. IBM VAC++ / xlC
4033 cannot bootstrap GCC. xlc can bootstrap an older version of GCC and
4034 G++ can bootstrap recent releases of GCC.
4036 GCC can bootstrap with recent versions of IBM XLC, but bootstrapping
4037 with an earlier release of GCC is recommended. Bootstrapping with XLC
4038 requires a larger data segment, which can be enabled through the
4039 @var{LDR_CNTRL} environment variable, e.g.,
4042 % LDR_CNTRL=MAXDATA=0x50000000
4046 One can start with a pre-compiled version of GCC to build from
4047 sources. One may delete GCC's ``fixed'' header files when starting
4048 with a version of GCC built for an earlier release of AIX.
4050 To speed up the configuration phases of bootstrapping and installing GCC,
4051 one may use GNU Bash instead of AIX @command{/bin/sh}, e.g.,
4054 % CONFIG_SHELL=/opt/freeware/bin/bash
4055 % export CONFIG_SHELL
4058 and then proceed as described in @uref{build.html,,the build
4059 instructions}, where we strongly recommend specifying an absolute path
4060 to invoke @var{srcdir}/configure.
4062 Because GCC on AIX is built as a 32-bit executable by default,
4063 (although it can generate 64-bit programs) the GMP and MPFR libraries
4064 required by gfortran must be 32-bit libraries. Building GMP and MPFR
4065 as static archive libraries works better than shared libraries.
4067 Errors involving @code{alloca} when building GCC generally are due
4068 to an incorrect definition of @code{CC} in the Makefile or mixing files
4069 compiled with the native C compiler and GCC@. During the stage1 phase of
4070 the build, the native AIX compiler @strong{must} be invoked as @command{cc}
4071 (not @command{xlc}). Once @command{configure} has been informed of
4072 @command{xlc}, one needs to use @samp{make distclean} to remove the
4073 configure cache files and ensure that @env{CC} environment variable
4074 does not provide a definition that will confuse @command{configure}.
4075 If this error occurs during stage2 or later, then the problem most likely
4076 is the version of Make (see above).
4078 The native @command{as} and @command{ld} are recommended for
4079 bootstrapping on AIX@. The GNU Assembler, GNU Linker, and GNU
4080 Binutils version 2.20 is the minimum level that supports bootstrap on
4081 AIX 5@. The GNU Assembler has not been updated to support AIX 6@ or
4082 AIX 7. The native AIX tools do interoperate with GCC@.
4084 AIX 7.1 added partial support for DWARF debugging, but full support
4085 requires AIX 7.1 TL03 SP7 that supports additional DWARF sections and
4086 fixes a bug in the assembler. AIX 7.1 TL03 SP5 distributed a version
4087 of libm.a missing important symbols; a fix for IV77796 will be
4090 AIX 5.3 TL10, AIX 6.1 TL05 and AIX 7.1 TL00 introduced an AIX
4091 assembler change that sometimes produces corrupt assembly files
4092 causing AIX linker errors. The bug breaks GCC bootstrap on AIX and
4093 can cause compilation failures with existing GCC installations. An
4094 AIX iFix for AIX 5.3 is available (APAR IZ98385 for AIX 5.3 TL10, APAR
4095 IZ98477 for AIX 5.3 TL11 and IZ98134 for AIX 5.3 TL12). AIX 5.3 TL11 SP8,
4096 AIX 5.3 TL12 SP5, AIX 6.1 TL04 SP11, AIX 6.1 TL05 SP7, AIX 6.1 TL06 SP6,
4097 AIX 6.1 TL07 and AIX 7.1 TL01 should include the fix.
4099 Building @file{libstdc++.a} requires a fix for an AIX Assembler bug
4100 APAR IY26685 (AIX 4.3) or APAR IY25528 (AIX 5.1). It also requires a
4101 fix for another AIX Assembler bug and a co-dependent AIX Archiver fix
4102 referenced as APAR IY53606 (AIX 5.2) or as APAR IY54774 (AIX 5.1)
4104 @anchor{TransferAixShobj}
4105 @samp{libstdc++} in GCC 3.4 increments the major version number of the
4106 shared object and GCC installation places the @file{libstdc++.a}
4107 shared library in a common location which will overwrite the and GCC
4108 3.3 version of the shared library. Applications either need to be
4109 re-linked against the new shared library or the GCC 3.1 and GCC 3.3
4110 versions of the @samp{libstdc++} shared object needs to be available
4111 to the AIX runtime loader. The GCC 3.1 @samp{libstdc++.so.4}, if
4112 present, and GCC 3.3 @samp{libstdc++.so.5} shared objects can be
4113 installed for runtime dynamic loading using the following steps to set
4114 the @samp{F_LOADONLY} flag in the shared object for @emph{each}
4115 multilib @file{libstdc++.a} installed:
4117 Extract the shared objects from the currently installed
4118 @file{libstdc++.a} archive:
4120 % ar -x libstdc++.a libstdc++.so.4 libstdc++.so.5
4123 Enable the @samp{F_LOADONLY} flag so that the shared object will be
4124 available for runtime dynamic loading, but not linking:
4126 % strip -e libstdc++.so.4 libstdc++.so.5
4129 Archive the runtime-only shared object in the GCC 3.4
4130 @file{libstdc++.a} archive:
4132 % ar -q libstdc++.a libstdc++.so.4 libstdc++.so.5
4136 @uref{./configure.html#WithAixSoname,,@option{--with-aix-soname=svr4}}
4137 configure option may drop the need for this procedure for libraries that
4140 Linking executables and shared libraries may produce warnings of
4141 duplicate symbols. The assembly files generated by GCC for AIX always
4142 have included multiple symbol definitions for certain global variable
4143 and function declarations in the original program. The warnings should
4144 not prevent the linker from producing a correct library or runnable
4147 AIX 4.3 utilizes a ``large format'' archive to support both 32-bit and
4148 64-bit object modules. The routines provided in AIX 4.3.0 and AIX 4.3.1
4149 to parse archive libraries did not handle the new format correctly.
4150 These routines are used by GCC and result in error messages during
4151 linking such as ``not a COFF file''. The version of the routines shipped
4152 with AIX 4.3.1 should work for a 32-bit environment. The @option{-g}
4153 option of the archive command may be used to create archives of 32-bit
4154 objects using the original ``small format''. A correct version of the
4155 routines is shipped with AIX 4.3.2 and above.
4157 Some versions of the AIX binder (linker) can fail with a relocation
4158 overflow severe error when the @option{-bbigtoc} option is used to link
4159 GCC-produced object files into an executable that overflows the TOC@. A fix
4160 for APAR IX75823 (OVERFLOW DURING LINK WHEN USING GCC AND -BBIGTOC) is
4161 available from IBM Customer Support and from its
4162 @uref{http://techsupport.services.ibm.com/,,techsupport.services.ibm.com}
4163 website as PTF U455193.
4165 The AIX 4.3.2.1 linker (bos.rte.bind_cmds Level 4.3.2.1) will dump core
4166 with a segmentation fault when invoked by any version of GCC@. A fix for
4167 APAR IX87327 is available from IBM Customer Support and from its
4168 @uref{http://techsupport.services.ibm.com/,,techsupport.services.ibm.com}
4169 website as PTF U461879. This fix is incorporated in AIX 4.3.3 and above.
4171 The initial assembler shipped with AIX 4.3.0 generates incorrect object
4172 files. A fix for APAR IX74254 (64BIT DISASSEMBLED OUTPUT FROM COMPILER FAILS
4173 TO ASSEMBLE/BIND) is available from IBM Customer Support and from its
4174 @uref{http://techsupport.services.ibm.com/,,techsupport.services.ibm.com}
4175 website as PTF U453956. This fix is incorporated in AIX 4.3.1 and above.
4177 AIX provides National Language Support (NLS)@. Compilers and assemblers
4178 use NLS to support locale-specific representations of various data
4179 formats including floating-point numbers (e.g., @samp{.} vs @samp{,} for
4180 separating decimal fractions). There have been problems reported where
4181 GCC does not produce the same floating-point formats that the assembler
4182 expects. If one encounters this problem, set the @env{LANG}
4183 environment variable to @samp{C} or @samp{En_US}.
4185 A default can be specified with the @option{-mcpu=@var{cpu_type}}
4186 switch and using the configure option @option{--with-cpu-@var{cpu_type}}.
4191 @anchor{iq2000-x-elf}
4192 @heading iq2000-*-elf
4193 Vitesse IQ2000 processors. These are used in embedded
4194 applications. There are no standard Unix configurations.
4201 Lattice Mico32 processor.
4202 This configuration is intended for embedded systems.
4207 @anchor{lm32-x-uclinux}
4208 @heading lm32-*-uclinux
4209 Lattice Mico32 processor.
4210 This configuration is intended for embedded systems running uClinux.
4217 Renesas M32C processor.
4218 This configuration is intended for embedded systems.
4225 Renesas M32R processor.
4226 This configuration is intended for embedded systems.
4234 @samp{m68k-*-elf*}, @samp{m68k-*-rtems}, @samp{m68k-*-uclinux} and
4236 build libraries for both M680x0 and ColdFire processors. If you only
4237 need the M680x0 libraries, you can omit the ColdFire ones by passing
4238 @option{--with-arch=m68k} to @command{configure}. Alternatively, you
4239 can omit the M680x0 libraries by passing @option{--with-arch=cf} to
4240 @command{configure}. These targets default to 5206 or 5475 code as
4241 appropriate for the target system when
4242 configured with @option{--with-arch=cf} and 68020 code otherwise.
4244 The @samp{m68k-*-netbsd} and
4245 @samp{m68k-*-openbsd} targets also support the @option{--with-arch}
4246 option. They will generate ColdFire CFV4e code when configured with
4247 @option{--with-arch=cf} and 68020 code otherwise.
4249 You can override the default processors listed above by configuring
4250 with @option{--with-cpu=@var{target}}. This @var{target} can either
4251 be a @option{-mcpu} argument or one of the following values:
4252 @samp{m68000}, @samp{m68010}, @samp{m68020}, @samp{m68030},
4253 @samp{m68040}, @samp{m68060}, @samp{m68020-40} and @samp{m68020-60}.
4255 GCC requires at least binutils version 2.17 on these targets.
4260 @anchor{m68k-x-uclinux}
4261 @heading m68k-*-uclinux
4262 GCC 4.3 changed the uClinux configuration so that it uses the
4263 @samp{m68k-linux-gnu} ABI rather than the @samp{m68k-elf} ABI.
4264 It also added improved support for C++ and flat shared libraries,
4265 both of which were ABI changes.
4270 @anchor{microblaze-x-elf}
4271 @heading microblaze-*-elf
4272 Xilinx MicroBlaze processor.
4273 This configuration is intended for embedded systems.
4280 If on a MIPS system you get an error message saying ``does not have gp
4281 sections for all it's [sic] sectons [sic]'', don't worry about it. This
4282 happens whenever you use GAS with the MIPS linker, but there is not
4283 really anything wrong, and it is okay to use the output file. You can
4284 stop such warnings by installing the GNU linker.
4286 It would be nice to extend GAS to produce the gp tables, but they are
4287 optional, and there should not be a warning about their absence.
4289 The libstdc++ atomic locking routines for MIPS targets requires MIPS II
4290 and later. A patch went in just after the GCC 3.3 release to
4291 make @samp{mips*-*-*} use the generic implementation instead. You can also
4292 configure for @samp{mipsel-elf} as a workaround. The
4293 @samp{mips*-*-linux*} target continues to use the MIPS II routines. More
4294 work on this is expected in future releases.
4296 @c If you make --with-llsc the default for another target, please also
4297 @c update the description of the --with-llsc option.
4299 The built-in @code{__sync_*} functions are available on MIPS II and
4300 later systems and others that support the @samp{ll}, @samp{sc} and
4301 @samp{sync} instructions. This can be overridden by passing
4302 @option{--with-llsc} or @option{--without-llsc} when configuring GCC.
4303 Since the Linux kernel emulates these instructions if they are
4304 missing, the default for @samp{mips*-*-linux*} targets is
4305 @option{--with-llsc}. The @option{--with-llsc} and
4306 @option{--without-llsc} configure options may be overridden at compile
4307 time by passing the @option{-mllsc} or @option{-mno-llsc} options to
4310 MIPS systems check for division by zero (unless
4311 @option{-mno-check-zero-division} is passed to the compiler) by
4312 generating either a conditional trap or a break instruction. Using
4313 trap results in smaller code, but is only supported on MIPS II and
4314 later. Also, some versions of the Linux kernel have a bug that
4315 prevents trap from generating the proper signal (@code{SIGFPE}). To enable
4316 the use of break, use the @option{--with-divide=breaks}
4317 @command{configure} option when configuring GCC@. The default is to
4318 use traps on systems that support them.
4323 @anchor{moxie-x-elf}
4324 @heading moxie-*-elf
4325 The moxie processor.
4330 @anchor{msp430-x-elf}
4331 @heading msp430-*-elf
4332 TI MSP430 processor.
4333 This configuration is intended for embedded systems.
4338 @anchor{nds32le-x-elf}
4339 @heading nds32le-*-elf
4340 Andes NDS32 target in little endian mode.
4345 @anchor{nds32be-x-elf}
4346 @heading nds32be-*-elf
4347 Andes NDS32 target in big endian mode.
4352 @anchor{nvptx-x-none}
4353 @heading nvptx-*-none
4356 Instead of GNU binutils, you will need to install
4357 @uref{https://github.com/MentorEmbedded/nvptx-tools/,,nvptx-tools}.
4358 Tell GCC where to find it:
4359 @option{--with-build-time-tools=[install-nvptx-tools]/nvptx-none/bin}.
4361 You will need newlib 3.0 git revision
4362 cd31fbb2aea25f94d7ecedc9db16dfc87ab0c316 or later. It can be
4363 automatically built together with GCC@. For this, add a symbolic link
4364 to nvptx-newlib's @file{newlib} directory to the directory containing
4367 Use the @option{--disable-sjlj-exceptions} and
4368 @option{--enable-newlib-io-long-long} options when configuring.
4375 The OpenRISC 1000 32-bit processor with delay slots.
4376 This configuration is intended for embedded systems.
4381 @anchor{or1k-x-linux}
4382 @heading or1k-*-linux
4383 The OpenRISC 1000 32-bit processor with delay slots.
4388 @anchor{powerpc-x-x}
4389 @heading powerpc-*-*
4390 You can specify a default version for the @option{-mcpu=@var{cpu_type}}
4391 switch by using the configure option @option{--with-cpu-@var{cpu_type}}.
4393 You will need GNU binutils 2.15 or newer.
4398 @anchor{powerpc-x-darwin}
4399 @heading powerpc-*-darwin*
4400 PowerPC running Darwin (Mac OS X kernel).
4402 Pre-installed versions of Mac OS X may not include any developer tools,
4403 meaning that you will not be able to build GCC from source. Tool
4404 binaries are available at
4405 @uref{https://opensource.apple.com}.
4407 This version of GCC requires at least cctools-590.36. The
4408 cctools-590.36 package referenced from
4409 @uref{http://gcc.gnu.org/ml/gcc/2006-03/msg00507.html} will not work
4410 on systems older than 10.3.9 (aka darwin7.9.0).
4415 @anchor{powerpc-x-elf}
4416 @heading powerpc-*-elf
4417 PowerPC system in big endian mode, running System V.4.
4422 @anchor{powerpc-x-linux-gnu}
4423 @heading powerpc*-*-linux-gnu*
4424 PowerPC system in big endian mode running Linux.
4429 @anchor{powerpc-x-netbsd}
4430 @heading powerpc-*-netbsd*
4431 PowerPC system in big endian mode running NetBSD@.
4436 @anchor{powerpc-x-eabisim}
4437 @heading powerpc-*-eabisim
4438 Embedded PowerPC system in big endian mode for use in running under the
4444 @anchor{powerpc-x-eabi}
4445 @heading powerpc-*-eabi
4446 Embedded PowerPC system in big endian mode.
4451 @anchor{powerpcle-x-elf}
4452 @heading powerpcle-*-elf
4453 PowerPC system in little endian mode, running System V.4.
4458 @anchor{powerpcle-x-eabisim}
4459 @heading powerpcle-*-eabisim
4460 Embedded PowerPC system in little endian mode for use in running under
4466 @anchor{powerpcle-x-eabi}
4467 @heading powerpcle-*-eabi
4468 Embedded PowerPC system in little endian mode.
4475 The Renesas RL78 processor.
4476 This configuration is intended for embedded systems.
4481 @anchor{riscv32-x-elf}
4482 @heading riscv32-*-elf
4483 The RISC-V RV32 instruction set.
4484 This configuration is intended for embedded systems.
4485 This (and all other RISC-V) targets are supported upstream as of the
4486 binutils 2.28 release.
4491 @anchor{riscv32-x-linux}
4492 @heading riscv32-*-linux
4493 The RISC-V RV32 instruction set running GNU/Linux.
4494 This (and all other RISC-V) targets are supported upstream as of the
4495 binutils 2.28 release.
4500 @anchor{riscv64-x-elf}
4501 @heading riscv64-*-elf
4502 The RISC-V RV64 instruction set.
4503 This configuration is intended for embedded systems.
4504 This (and all other RISC-V) targets are supported upstream as of the
4505 binutils 2.28 release.
4510 @anchor{riscv64-x-linux}
4511 @heading riscv64-*-linux
4512 The RISC-V RV64 instruction set running GNU/Linux.
4513 This (and all other RISC-V) targets are supported upstream as of the
4514 binutils 2.28 release.
4521 The Renesas RX processor.
4526 @anchor{s390-x-linux}
4527 @heading s390-*-linux*
4528 S/390 system running GNU/Linux for S/390@.
4533 @anchor{s390x-x-linux}
4534 @heading s390x-*-linux*
4535 zSeries system (64-bit) running GNU/Linux for zSeries@.
4540 @anchor{s390x-ibm-tpf}
4541 @heading s390x-ibm-tpf*
4542 zSeries system (64-bit) running TPF@. This platform is
4543 supported as cross-compilation target only.
4548 @c Please use Solaris 2 to refer to all release of Solaris, starting
4549 @c with 2.0 until 2.6, 7, 8, etc. Solaris 1 was a marketing name for
4550 @c SunOS 4 releases which we don't use to avoid confusion. Solaris
4551 @c alone is too unspecific and must be avoided.
4552 @anchor{x-x-solaris2}
4553 @heading *-*-solaris2*
4554 Support for Solaris 10 has been removed in GCC 10. Support for Solaris
4555 9 has been removed in GCC 5. Support for Solaris 8 has been removed in
4556 GCC 4.8. Support for Solaris 7 has been removed in GCC 4.6.
4558 Solaris 11 provides GCC 4.5.2, 4.7.3, and 4.8.2 as
4559 @command{/usr/gcc/4.5/bin/gcc} or similar. Alternatively,
4560 you can install a pre-built GCC to bootstrap and install GCC. See the
4561 @uref{binaries.html,,binaries page} for details.
4563 The Solaris 2 @command{/bin/sh} will often fail to configure
4564 @samp{libstdc++-v3}or @samp{boehm-gc}. We therefore recommend using the
4565 following initial sequence of commands
4568 % CONFIG_SHELL=/bin/ksh
4569 % export CONFIG_SHELL
4573 and proceed as described in @uref{configure.html,,the configure instructions}.
4574 In addition we strongly recommend specifying an absolute path to invoke
4575 @command{@var{srcdir}/configure}.
4577 In Solaris 11, you need to check for @code{system/header},
4578 @code{system/linker}, and @code{developer/assembler} packages.
4580 Trying to use the linker and other tools in
4581 @file{/usr/ucb} to install GCC has been observed to cause trouble.
4582 For example, the linker may hang indefinitely. The fix is to remove
4583 @file{/usr/ucb} from your @env{PATH}.
4585 The build process works more smoothly with the legacy Sun tools so, if you
4586 have @file{/usr/xpg4/bin} in your @env{PATH}, we recommend that you place
4587 @file{/usr/bin} before @file{/usr/xpg4/bin} for the duration of the build.
4589 We recommend the use of the Solaris assembler or the GNU assembler, in
4590 conjunction with the Solaris linker. The GNU @command{as}
4591 versions included in Solaris 11,
4592 from GNU binutils 2.19 or newer (also in @file{/usr/bin/gas} and
4593 @file{/usr/gnu/bin/as}), are known to work.
4594 The current version, from GNU binutils 2.32,
4595 is known to work as well. Note that your mileage may vary
4596 if you use a combination of the GNU tools and the Solaris tools: while the
4597 combination GNU @command{as} + Sun @command{ld} should reasonably work,
4598 the reverse combination Sun @command{as} + GNU @command{ld} may fail to
4599 build or cause memory corruption at runtime in some cases for C++ programs.
4601 GNU @command{ld} usually works as well. Again, the current
4602 version (2.32) is known to work, but generally lacks platform specific
4603 features, so better stay with Solaris @command{ld}. To use the LTO linker
4604 plugin (@option{-fuse-linker-plugin}) with GNU @command{ld}, GNU
4605 binutils @emph{must} be configured with @option{--enable-largefile}.
4607 To enable symbol versioning in @samp{libstdc++} with the Solaris linker,
4608 you need to have any version of GNU @command{c++filt}, which is part of
4609 GNU binutils. @samp{libstdc++} symbol versioning will be disabled if no
4610 appropriate version is found. Solaris @command{c++filt} from the Solaris
4611 Studio compilers does @emph{not} work.
4618 This section contains general configuration information for all
4619 SPARC-based platforms. In addition to reading this section, please
4620 read all other sections that match your target.
4622 Newer versions of the GNU Multiple Precision Library (GMP), the MPFR
4623 library and the MPC library are known to be miscompiled by earlier
4624 versions of GCC on these platforms. We therefore recommend the use
4625 of the exact versions of these libraries listed as minimal versions
4626 in @uref{prerequisites.html,,the prerequisites}.
4631 @anchor{sparc-sun-solaris2}
4632 @heading sparc-sun-solaris2*
4633 When GCC is configured to use GNU binutils 2.14 or later, the binaries
4634 produced are smaller than the ones produced using Sun's native tools;
4635 this difference is quite significant for binaries containing debugging
4638 Starting with Solaris 7, the operating system is capable of executing
4639 64-bit SPARC V9 binaries. GCC 3.1 and later properly supports
4640 this; the @option{-m64} option enables 64-bit code generation.
4641 However, if all you want is code tuned for the UltraSPARC CPU, you
4642 should try the @option{-mtune=ultrasparc} option instead, which produces
4643 code that, unlike full 64-bit code, can still run on non-UltraSPARC
4646 When configuring the GNU Multiple Precision Library (GMP), the MPFR
4647 library or the MPC library on a Solaris 7 or later system, the canonical
4648 target triplet must be specified as the @command{build} parameter on the
4649 configure line. This target triplet can be obtained by invoking @command{./config.guess} in the toplevel source directory of GCC (and
4650 not that of GMP or MPFR or MPC). For example on a Solaris 9 system:
4653 % ./configure --build=sparc-sun-solaris2.9 --prefix=xxx
4659 @anchor{sparc-x-linux}
4660 @heading sparc-*-linux*
4665 @anchor{sparc64-x-solaris2}
4666 @heading sparc64-*-solaris2*
4667 When configuring the GNU Multiple Precision Library (GMP), the MPFR
4668 library or the MPC library, the canonical target triplet must be specified
4669 as the @command{build} parameter on the configure line. For example
4670 on a Solaris 9 system:
4673 % ./configure --build=sparc64-sun-solaris2.9 --prefix=xxx
4679 @anchor{sparcv9-x-solaris2}
4680 @heading sparcv9-*-solaris2*
4681 This is a synonym for @samp{sparc64-*-solaris2*}.
4688 The C6X family of processors. This port requires binutils-2.22 or newer.
4693 @anchor{tilegx-*-linux}
4694 @heading tilegx-*-linux*
4695 The TILE-Gx processor in little endian mode, running GNU/Linux. This
4696 port requires binutils-2.22 or newer.
4701 @anchor{tilegxbe-*-linux}
4702 @heading tilegxbe-*-linux*
4703 The TILE-Gx processor in big endian mode, running GNU/Linux. This
4704 port requires binutils-2.23 or newer.
4709 @anchor{tilepro-*-linux}
4710 @heading tilepro-*-linux*
4711 The TILEPro processor running GNU/Linux. This port requires
4712 binutils-2.22 or newer.
4717 @anchor{visium-x-elf}
4718 @heading visium-*-elf
4719 CDS VISIUMcore processor.
4720 This configuration is intended for embedded systems.
4725 @anchor{x-x-vxworks}
4726 @heading *-*-vxworks*
4727 Support for VxWorks is in flux. At present GCC supports @emph{only} the
4728 very recent VxWorks 5.5 (aka Tornado 2.2) release, and only on PowerPC@.
4729 We welcome patches for other architectures supported by VxWorks 5.5.
4730 Support for VxWorks AE would also be welcome; we believe this is merely
4731 a matter of writing an appropriate ``configlette'' (see below). We are
4732 not interested in supporting older, a.out or COFF-based, versions of
4735 VxWorks comes with an older version of GCC installed in
4736 @file{@var{$WIND_BASE}/host}; we recommend you do not overwrite it.
4737 Choose an installation @var{prefix} entirely outside @var{$WIND_BASE}.
4738 Before running @command{configure}, create the directories @file{@var{prefix}}
4739 and @file{@var{prefix}/bin}. Link or copy the appropriate assembler,
4740 linker, etc.@: into @file{@var{prefix}/bin}, and set your @var{PATH} to
4741 include that directory while running both @command{configure} and
4744 You must give @command{configure} the
4745 @option{--with-headers=@var{$WIND_BASE}/target/h} switch so that it can
4746 find the VxWorks system headers. Since VxWorks is a cross compilation
4747 target only, you must also specify @option{--target=@var{target}}.
4748 @command{configure} will attempt to create the directory
4749 @file{@var{prefix}/@var{target}/sys-include} and copy files into it;
4750 make sure the user running @command{configure} has sufficient privilege
4753 GCC's exception handling runtime requires a special ``configlette''
4754 module, @file{contrib/gthr_supp_vxw_5x.c}. Follow the instructions in
4755 that file to add the module to your kernel build. (Future versions of
4756 VxWorks will incorporate this module.)
4762 @heading x86_64-*-*, amd64-*-*
4763 GCC supports the x86-64 architecture implemented by the AMD64 processor
4764 (amd64-*-* is an alias for x86_64-*-*) on GNU/Linux, FreeBSD and NetBSD@.
4765 On GNU/Linux the default is a bi-arch compiler which is able to generate
4766 both 64-bit x86-64 and 32-bit x86 code (via the @option{-m32} switch).
4771 @anchor{x86-64-x-solaris2}
4772 @heading x86_64-*-solaris2*
4773 GCC also supports the x86-64 architecture implemented by the AMD64
4774 processor (@samp{amd64-*-*} is an alias for @samp{x86_64-*-*}) on
4775 Solaris 10 or later. Unlike other systems, without special options a
4776 bi-arch compiler is built which generates 32-bit code by default, but
4777 can generate 64-bit x86-64 code with the @option{-m64} switch. Since
4778 GCC 4.7, there is also a configuration that defaults to 64-bit code, but
4779 can generate 32-bit code with @option{-m32}. To configure and build
4780 this way, you have to provide all support libraries like @file{libgmp}
4781 as 64-bit code, configure with @option{--target=x86_64-pc-solaris2.1x}
4782 and @samp{CC=gcc -m64}.
4787 @anchor{xtensa-x-elf}
4788 @heading xtensa*-*-elf
4789 This target is intended for embedded Xtensa systems using the
4790 @samp{newlib} C library. It uses ELF but does not support shared
4791 objects. Designed-defined instructions specified via the
4792 Tensilica Instruction Extension (TIE) language are only supported
4793 through inline assembly.
4795 The Xtensa configuration information must be specified prior to
4796 building GCC@. The @file{include/xtensa-config.h} header
4797 file contains the configuration information. If you created your
4798 own Xtensa configuration with the Xtensa Processor Generator, the
4799 downloaded files include a customized copy of this header file,
4800 which you can use to replace the default header file.
4805 @anchor{xtensa-x-linux}
4806 @heading xtensa*-*-linux*
4807 This target is for Xtensa systems running GNU/Linux. It supports ELF
4808 shared objects and the GNU C library (glibc). It also generates
4809 position-independent code (PIC) regardless of whether the
4810 @option{-fpic} or @option{-fPIC} options are used. In other
4811 respects, this target is the same as the
4812 @uref{#xtensa*-*-elf,,@samp{xtensa*-*-elf}} target.
4818 @heading Microsoft Windows
4820 @subheading Intel 16-bit versions
4821 The 16-bit versions of Microsoft Windows, such as Windows 3.1, are not
4824 However, the 32-bit port has limited support for Microsoft
4825 Windows 3.11 in the Win32s environment, as a target only. See below.
4827 @subheading Intel 32-bit versions
4828 The 32-bit versions of Windows, including Windows 95, Windows NT, Windows
4829 XP, and Windows Vista, are supported by several different target
4830 platforms. These targets differ in which Windows subsystem they target
4831 and which C libraries are used.
4834 @item Cygwin @uref{#x-x-cygwin,,*-*-cygwin}: Cygwin provides a user-space
4835 Linux API emulation layer in the Win32 subsystem.
4836 @item MinGW @uref{#x-x-mingw32,,*-*-mingw32}: MinGW is a native GCC port for
4837 the Win32 subsystem that provides a subset of POSIX.
4838 @item MKS i386-pc-mks: NuTCracker from MKS. See
4839 @uref{https://www.mkssoftware.com} for more information.
4842 @subheading Intel 64-bit versions
4843 GCC contains support for x86-64 using the mingw-w64
4844 runtime library, available from @uref{http://mingw-w64.org/doku.php}.
4845 This library should be used with the target triple x86_64-pc-mingw32.
4847 Presently Windows for Itanium is not supported.
4849 @subheading Windows CE
4850 Windows CE is supported as a target only on Hitachi
4851 SuperH (sh-wince-pe), and MIPS (mips-wince-pe).
4853 @subheading Other Windows Platforms
4854 GCC no longer supports Windows NT on the Alpha or PowerPC.
4856 GCC no longer supports the Windows POSIX subsystem. However, it does
4857 support the Interix subsystem. See above.
4859 Old target names including *-*-winnt and *-*-windowsnt are no longer used.
4861 PW32 (i386-pc-pw32) support was never completed, and the project seems to
4862 be inactive. See @uref{http://pw32.sourceforge.net/} for more information.
4864 UWIN support has been removed due to a lack of maintenance.
4871 Ports of GCC are included with the
4872 @uref{http://www.cygwin.com/,,Cygwin environment}.
4874 GCC will build under Cygwin without modification; it does not build
4875 with Microsoft's C++ compiler and there are no plans to make it do so.
4877 The Cygwin native compiler can be configured to target any 32-bit x86
4878 cpu architecture desired; the default is i686-pc-cygwin. It should be
4879 used with as up-to-date a version of binutils as possible; use either
4880 the latest official GNU binutils release in the Cygwin distribution,
4881 or version 2.20 or above if building your own.
4886 @anchor{x-x-mingw32}
4887 @heading *-*-mingw32
4888 GCC will build with and support only MinGW runtime 3.12 and later.
4889 Earlier versions of headers are incompatible with the new default semantics
4890 of @code{extern inline} in @code{-std=c99} and @code{-std=gnu99} modes.
4896 @heading Older systems
4897 GCC contains support files for many older (1980s and early
4898 1990s) Unix variants. For the most part, support for these systems
4899 has not been deliberately removed, but it has not been maintained for
4900 several years and may suffer from bitrot.
4902 Starting with GCC 3.1, each release has a list of ``obsoleted'' systems.
4903 Support for these systems is still present in that release, but
4904 @command{configure} will fail unless the @option{--enable-obsolete}
4905 option is given. Unless a maintainer steps forward, support for these
4906 systems will be removed from the next release of GCC@.
4908 Support for old systems as hosts for GCC can cause problems if the
4909 workarounds for compiler, library and operating system bugs affect the
4910 cleanliness or maintainability of the rest of GCC@. In some cases, to
4911 bring GCC up on such a system, if still possible with current GCC, may
4912 require first installing an old version of GCC which did work on that
4913 system, and using it to compile a more recent GCC, to avoid bugs in the
4914 vendor compiler. Old releases of GCC 1 and GCC 2 are available in the
4915 @file{old-releases} directory on the @uref{../mirrors.html,,GCC mirror
4916 sites}. Header bugs may generally be avoided using
4917 @command{fixincludes}, but bugs or deficiencies in libraries and the
4918 operating system may still cause problems.
4920 Support for older systems as targets for cross-compilation is less
4921 problematic than support for them as hosts for GCC; if an enthusiast
4922 wishes to make such a target work again (including resurrecting any of
4923 the targets that never worked with GCC 2, starting from the last
4924 version before they were removed), patches
4925 @uref{../contribute.html,,following the usual requirements} would be
4926 likely to be accepted, since they should not affect the support for more
4929 For some systems, old versions of GNU binutils may also be useful,
4930 and are available from @file{pub/binutils/old-releases} on
4931 @uref{https://sourceware.org/mirrors.html,,sourceware.org mirror sites}.
4933 Some of the information on specific systems above relates to
4934 such older systems, but much of the information
4935 about GCC on such systems (which may no longer be applicable to
4936 current GCC) is to be found in the GCC texinfo manual.
4942 @heading all ELF targets (SVR4, Solaris 2, etc.)
4943 C++ support is significantly better on ELF targets if you use the
4944 @uref{./configure.html#with-gnu-ld,,GNU linker}; duplicate copies of
4945 inlines, vtables and template instantiations will be discarded
4954 @uref{./index.html,,Return to the GCC Installation page}
4958 @c ***Old documentation******************************************************
4960 @include install-old.texi
4966 @uref{./index.html,,Return to the GCC Installation page}
4970 @c ***GFDL********************************************************************
4978 @uref{./index.html,,Return to the GCC Installation page}
4982 @c ***************************************************************************
4983 @c Part 6 The End of the Document
4985 @comment node-name, next, previous, up
4986 @node Concept Index, , GNU Free Documentation License, Top
4990 @unnumbered Concept Index