1 \input texinfo.tex @c -*-texinfo-*-
4 @setfilename gccinstall.info
5 @settitle Installing GCC
10 @c Specify title for specific html page
12 @settitle Installing GCC
15 @settitle Host/Target specific installation notes for GCC
18 @settitle Downloading GCC
21 @settitle Installing GCC: Configuration
24 @settitle Installing GCC: Building
27 @settitle Installing GCC: Testing
29 @ifset finalinstallhtml
30 @settitle Installing GCC: Final installation
33 @settitle Installing GCC: Binaries
36 @settitle Installing GCC: Old documentation
39 @settitle Installing GCC: GNU Free Documentation License
42 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
43 @c 1999, 2000, 2001, 2002 Free Software Foundation, Inc.
44 @c *** Converted to texinfo by Dean Wakerley, dean@wakerley.com
46 @c Include everything if we're not making html
60 @c Part 2 Summary Description and Copyright
61 @macro copyrightnotice
62 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
63 1999, 2000, 2001, 2002 Free Software Foundation, Inc.
65 Permission is granted to copy, distribute and/or modify this document
66 under the terms of the GNU Free Documentation License, Version 1.1 or
67 any later version published by the Free Software Foundation; with no
68 Invariant Sections, the Front-Cover texts being (a) (see below), and
69 with the Back-Cover Texts being (b) (see below). A copy of the
70 license is included in the section entitled ``@uref{./gfdl.html,,GNU
71 Free Documentation License}''.
73 (a) The FSF's Front-Cover Text is:
77 (b) The FSF's Back-Cover Text is:
79 You have freedom to copy and modify this GNU Manual, like GNU
80 software. Copies published by the Free Software Foundation raise
81 funds for GNU development.
87 @c Part 3 Titlepage and Copyright
90 @comment The title is printed in a large font.
91 @center @titlefont{Installing GCC}
93 @c The following two commands start the copyright page.
95 @vskip 0pt plus 1filll
99 @c Part 4 Top node and Master Menu
102 @comment node-name, next, Previous, up
105 * Installing GCC:: This document describes the generic installation
106 procedure for GCC as well as detailing some target
107 specific installation instructions.
109 * Specific:: Host/target specific installation notes for GCC.
110 * Binaries:: Where to get pre-compiled binaries.
112 * Old:: Old installation documentation.
114 * GNU Free Documentation License:: How you can copy and share this manual.
115 * Concept Index:: This index has two entries.
119 @c Part 5 The Body of the Document
120 @c ***Installing GCC**********************************************************
122 @comment node-name, next, previous, up
123 @node Installing GCC, Binaries, , Top
127 @chapter Installing GCC
130 The latest version of this document is always available at
131 @uref{http://gcc.gnu.org/install/,,http://gcc.gnu.org/install/}.
133 This document describes the generic installation procedure for GCC as well
134 as detailing some target specific installation instructions.
136 GCC includes several components that previously were separate distributions
137 with their own installation instructions. This document supersedes all
138 package specific installation instructions.
140 @emph{Before} starting the build/install procedure please check the
142 @ref{Specific, host/target specific installation notes}.
145 @uref{specific.html,,host/target specific installation notes}.
147 We recommend you browse the entire generic installation instructions before
150 Lists of successful builds for released versions of GCC are
151 available at @uref{http://gcc.gnu.org/buildstat.html}.
152 These lists are updated as new information becomes available.
154 The installation procedure itself is broken into five steps.
158 * Downloading the source::
161 * Testing:: (optional)
168 @uref{download.html,,Downloading the source}
170 @uref{configure.html,,Configuration}
172 @uref{build.html,,Building}
174 @uref{test.html,,Testing} (optional)
176 @uref{finalinstall.html,,Final install}
180 Please note that GCC does not support @samp{make uninstall} and probably
181 won't do so in the near future as this would open a can of worms. Instead,
182 we suggest that you install GCC into a directory of its own and simply
183 remove that directory when you do not need that specific version of GCC
184 any longer, and, if shared libraries are installed there as well, no
185 more binaries exist that use them.
188 There are also some @uref{old.html,,old installation instructions},
189 which are mostly obsolete but still contain some information which has
190 not yet been merged into the main part of this manual.
198 @uref{./index.html,,Return to the GCC Installation page}
204 @c ***Downloading the source**************************************************
206 @comment node-name, next, previous, up
207 @node Downloading the source, Configuration, , Installing GCC
211 @chapter Downloading GCC
213 @cindex Downloading GCC
214 @cindex Downloading the Source
216 GCC is distributed via @uref{http://gcc.gnu.org/cvs.html,,CVS} and FTP
217 tarballs compressed with @command{gzip} or
218 @command{bzip2}. It is possible to download a full distribution or specific
221 Please refer to our @uref{http://gcc.gnu.org/releases.html,,releases web page}
222 for information on how to obtain GCC@.
224 The full distribution includes the C, C++, Objective-C, Fortran, Java,
225 and Ada (in case of GCC 3.1 and later) compilers. The full distribution
226 also includes runtime libraries for C++, Objective-C, Fortran, and Java.
227 In GCC 3.0 and later versions, GNU compiler testsuites are also included
228 in the full distribution.
230 If you choose to download specific components, you must download the core
231 GCC distribution plus any language specific distributions you wish to
232 use. The core distribution includes the C language front end as well as the
233 shared components. Each language has a tarball which includes the language
234 front end as well as the language runtime (when appropriate).
236 Unpack the core distribution as well as any language specific
237 distributions in the same directory.
239 If you also intend to build binutils (either to upgrade an existing
240 installation or for use in place of the corresponding tools of your
241 OS), unpack the binutils distribution either in the same directory or
242 a separate one. In the latter case, add symbolic links to any
243 components of the binutils you intend to build alongside the compiler
244 (@file{bfd}, @file{binutils}, @file{gas}, @file{gprof}, @file{ld},
245 @file{opcodes}, @dots{}) to the directory containing the GCC sources.
252 @uref{./index.html,,Return to the GCC Installation page}
256 @c ***Configuration***********************************************************
258 @comment node-name, next, previous, up
259 @node Configuration, Building, Downloading the source, Installing GCC
263 @chapter Installing GCC: Configuration
265 @cindex Configuration
266 @cindex Installing GCC: Configuration
268 Like most GNU software, GCC must be configured before it can be built.
269 This document describes the recommended configuration procedure
270 for both native and cross targets.
272 We use @var{srcdir} to refer to the toplevel source directory for
273 GCC; we use @var{objdir} to refer to the toplevel build/object directory.
275 If you obtained the sources via CVS, @var{srcdir} must refer to the top
276 @file{gcc} directory, the one where the @file{MAINTAINERS} can be found,
277 and not its @file{gcc} subdirectory, otherwise the build will fail.
279 If either @var{srcdir} or @var{objdir} is located on an automounted NFS
280 file system, the shell's built-in @command{pwd} command will return
281 temporary pathnames. Using these can lead to various sorts of build
282 problems. To avoid this issue, set the @env{PWDCMD} environment
283 variable to an automounter-aware @command{pwd} command, e.g.,
284 @command{pawd} or @command{amq -w}, during the configuration and build
287 First, in general, GCC @strong{must} be built into a separate directory
288 than the sources which does @strong{not} reside within the source tree.
289 This is how almost all developers build GCC; building where @var{srcdir}
290 == @var{objdir} is completely unsupported; building where @var{objdir}
291 is a subdirectory of @var{srcdir} is completely unsupported.
293 If you have previously built GCC in the same directory for a
294 different target machine, do @samp{make distclean} to delete all files
295 that might be invalid. One of the files this deletes is
296 @file{Makefile}; if @samp{make distclean} complains that @file{Makefile}
297 does not exist, it probably means that the directory is already suitably
298 clean. However, with the recommended method of building in a separate
299 @var{objdir}, you should simply use a different @var{objdir} for each
302 Second, when configuring a native system, either @command{cc} or
303 @command{gcc} must be in your path or you must set @env{CC} in
304 your environment before running configure. Otherwise the configuration
307 Note that the bootstrap compiler and the resulting GCC must be link
308 compatible, else the bootstrap will fail with linker errors about
309 incompatible object file formats. Several multilibed targets are
310 affected by this requirement, see
312 @ref{Specific, host/target specific installation notes}.
315 @uref{specific.html,,host/target specific installation notes}.
323 % @var{srcdir}/configure [@var{options}] [@var{target}]
327 @heading Target specification
330 GCC has code to correctly determine the correct value for @var{target}
331 for nearly all native systems. Therefore, we highly recommend you not
332 provide a configure target when configuring a native compiler.
335 @var{target} must be specified as @option{--target=@var{target}}
336 when configuring a cross compiler; examples of valid targets would be
337 i960-rtems, m68k-coff, sh-elf, etc.
340 Specifying just @var{target} instead of @option{--target=@var{target}}
341 implies that the host defaults to @var{target}.
345 @heading Options specification
347 Use @var{options} to override several configure time options for
348 GCC@. A list of supported @var{options} follows; @command{configure
349 --help} may list other options, but those not listed below may not
350 work and should not normally be used.
353 @item --prefix=@var{dirname}
354 Specify the toplevel installation
355 directory. This is the recommended way to install the tools into a directory
356 other than the default. The toplevel installation directory defaults to
359 We @strong{highly} recommend against @var{dirname} being the same or a
360 subdirectory of @var{objdir} or vice versa.
362 These additional options control where certain parts of the distribution
363 are installed. Normally you should not need to use these options.
365 @item --exec-prefix=@var{dirname}
366 Specify the toplevel installation directory for architecture-dependent
367 files. The default is @file{@var{prefix}}.
369 @item --bindir=@var{dirname}
370 Specify the installation directory for the executables called by users
371 (such as @command{gcc} and @command{g++}). The default is
372 @file{@var{exec-prefix}/bin}.
374 @item --libdir=@var{dirname}
375 Specify the installation directory for object code libraries and
376 internal parts of GCC@. The default is @file{@var{exec-prefix}/lib}.
378 @item --with-slibdir=@var{dirname}
379 Specify the installation directory for the shared libgcc library. The
380 default is @file{@var{libdir}}.
382 @item --infodir=@var{dirname}
383 Specify the installation directory for documentation in info format.
384 The default is @file{@var{prefix}/info}.
386 @item --datadir=@var{dirname}
387 Specify the installation directory for some architecture-independent
388 data files referenced by GCC@. The default is @file{@var{prefix}/share}.
390 @item --mandir=@var{dirname}
391 Specify the installation directory for manual pages. The default is
392 @file{@var{prefix}/man}. (Note that the manual pages are only extracts from
393 the full GCC manuals, which are provided in Texinfo format. The
394 @command{g77} manpage is unmaintained and may be out of date; the others
395 are derived by an automatic conversion process from parts of the full
398 @item --with-gxx-include-dir=@var{dirname}
400 the installation directory for G++ header files. The default is
401 @file{@var{prefix}/include/g++-v3}.
405 @item --program-prefix=@var{prefix}
406 GCC supports some transformations of the names of its programs when
407 installing them. This option prepends @var{prefix} to the names of
408 programs to install in @var{bindir} (see above). For example, specifying
409 @option{--program-prefix=foo-} would result in @samp{gcc}
410 being installed as @file{/usr/local/bin/foo-gcc}.
412 @item --program-suffix=@var{suffix}
413 Appends @var{suffix} to the names of programs to install in @var{bindir}
414 (see above). For example, specifying @option{--program-suffix=-3.1}
415 would result in @samp{gcc} being installed as
416 @file{/usr/local/bin/gcc-3.1}.
418 @item --program-transform-name=@var{pattern}
419 Applies the @samp{sed} script @var{pattern} to be applied to the names
420 of programs to install in @var{bindir} (see above). @var{pattern} has to
421 consist of one or more basic @samp{sed} editing commands, separated by
422 semicolons. For example, if you want the @samp{gcc} program name to be
423 transformed to the installed program @file{/usr/local/bin/myowngcc} and
424 the @samp{g++} program name to be transformed to
425 @file{/usr/local/bin/gspecial++} without changing other program names,
426 you could use the pattern
427 @option{--program-transform-name='s/^gcc$/myowngcc/; s/^g++$/gspecial++/'}
428 to achieve this effect.
430 All three options can be combined and used together, resulting in more
431 complex conversion patterns. As a basic rule, @var{prefix} (and
432 @var{suffix}) are prepended (appended) before further transformations
433 can happen with a special transformation script @var{pattern}.
435 As currently implemented, this option only takes effect for native
436 builds; cross compiler binaries' names are not transformed even when a
437 transformation is explicitly asked for by one of these options.
439 For native builds, some of the installed programs are also installed
440 with the target alias in front of their name, as in
441 @samp{i686-pc-linux-gnu-gcc}. All of the above transformations happen
442 before the target alias is prepended to the name - so, specifying
443 @option{--program-prefix=foo-} and @option{program-suffix=-3.1}, the
444 resulting binary would be installed as
445 @file{/usr/local/bin/i686-pc-linux-gnu-foo-gcc-3.1}.
447 As a last shortcoming, none of the installed Ada programs are
448 transformed yet, which will be fixed in some time.
450 @item --with-local-prefix=@var{dirname}
452 installation directory for local include files. The default is
453 @file{/usr/local}. Specify this option if you want the compiler to
454 search directory @file{@var{dirname}/include} for locally installed
455 header files @emph{instead} of @file{/usr/local/include}.
457 You should specify @option{--with-local-prefix} @strong{only} if your
458 site has a different convention (not @file{/usr/local}) for where to put
461 The default value for @option{--with-local-prefix} is @file{/usr/local}
462 regardless of the value of @option{--prefix}. Specifying
463 @option{--prefix} has no effect on which directory GCC searches for
464 local header files. This may seem counterintuitive, but actually it is
467 The purpose of @option{--prefix} is to specify where to @emph{install
468 GCC}. The local header files in @file{/usr/local/include}---if you put
469 any in that directory---are not part of GCC@. They are part of other
470 programs---perhaps many others. (GCC installs its own header files in
471 another directory which is based on the @option{--prefix} value.)
473 Both the local-prefix include directory and the GCC-prefix include
474 directory are part of GCC's "system include" directories. Although these
475 two directories are not fixed, they need to be searched in the proper
476 order for the correct processing of the include_next directive. The
477 local-prefix include directory is searched before the GCC-prefix
478 include directory. Another characteristic of system include directories
479 is that pedantic warnings are turned off for headers in these directories.
481 Some autoconf macros add @option{-I @var{directory}} options to the
482 compiler command line, to ensure that directories containing installed
483 packages' headers are searched. When @var{directory} is one of GCC's
484 system include directories, GCC will ignore the option so that system
485 directories continue to be processed in the correct order. This
486 may result in a search order different from what was specified but the
487 directory will still be searched.
489 GCC automatically searches for ordinary libraries using
490 @env{GCC_EXEC_PREFIX}. Thus, when the same installation prefix is
491 used for both GCC and packages, GCC will automatically search for
492 both headers and libraries. This provides a configuration that is
493 easy to use. GCC behaves in a manner similar to that when it is
494 installed as a system compiler in @file{/usr}.
496 Sites that need to install multiple versions of GCC may not want to
497 use the above simple configuration. It is possible to use the
498 @option{--program-prefix}, @option{--program-suffix} and
499 @option{--program-transform-name} options to install multiple versions
500 into a single directory, but it may be simpler to use different prefixes
501 and the @option{--with-local-prefix} option to specify the location of the
502 site-specific files for each version. It will then be necessary for
503 users to specify explicitly the location of local site libraries
504 (e.g., with @env{LIBRARY_PATH}).
506 The same value can be used for both @option{--with-local-prefix} and
507 @option{--prefix} provided it is not @file{/usr}. This can be used
508 to avoid the default search of @file{/usr/local/include}.
510 @strong{Do not} specify @file{/usr} as the @option{--with-local-prefix}!
511 The directory you use for @option{--with-local-prefix} @strong{must not}
512 contain any of the system's standard header files. If it did contain
513 them, certain programs would be miscompiled (including GNU Emacs, on
514 certain targets), because this would override and nullify the header
515 file corrections made by the @command{fixincludes} script.
517 Indications are that people who use this option use it based on mistaken
518 ideas of what it is for. People use it as if it specified where to
519 install part of GCC@. Perhaps they make this assumption because
520 installing GCC creates the directory.
522 @item --enable-shared[=@var{package}[,@dots{}]]
523 Build shared versions of libraries, if shared libraries are supported on
524 the target platform. Unlike GCC 2.95.x and earlier, shared libraries
525 are enabled by default on all platforms that support shared libraries,
526 except for @samp{libobjc} which is built as a static library only by
529 If a list of packages is given as an argument, build shared libraries
530 only for the listed packages. For other packages, only static libraries
531 will be built. Package names currently recognized in the GCC tree are
532 @samp{libgcc} (also known as @samp{gcc}), @samp{libstdc++} (not
533 @samp{libstdc++-v3}), @samp{libffi}, @samp{zlib}, @samp{boehm-gc} and
534 @samp{libjava}. Note that @samp{libobjc} does not recognize itself by
535 any name, so, if you list package names in @option{--enable-shared},
536 you will only get static Objective-C libraries. @samp{libf2c} and
537 @samp{libiberty} do not support shared libraries at all.
539 Use @option{--disable-shared} to build only static libraries. Note that
540 @option{--disable-shared} does not accept a list of package names as
541 argument, only @option{--enable-shared} does.
543 @item @anchor{with-gnu-as}--with-gnu-as
544 Specify that the compiler should assume that the
545 assembler it finds is the GNU assembler. However, this does not modify
546 the rules to find an assembler and will result in confusion if found
547 assembler is not actually the GNU assembler. (Confusion will also
548 result if the compiler finds the GNU assembler but has not been
549 configured with @option{--with-gnu-as}.) If you have more than one
550 assembler installed on your system, you may want to use this option in
551 connection with @option{--with-as=@var{pathname}}.
553 The following systems are the only ones where it makes a difference
554 whether you use the GNU assembler. On any other system,
555 @option{--with-gnu-as} has no effect.
558 @item @samp{hppa1.0-@var{any}-@var{any}}
559 @item @samp{hppa1.1-@var{any}-@var{any}}
560 @item @samp{i386-@var{any}-sysv}
561 @item @samp{i386-@var{any}-isc}
562 @item @samp{i860-@var{any}-bsd}
563 @item @samp{m68k-bull-sysv}
564 @item @samp{m68k-hp-hpux}
565 @item @samp{m68k-sony-bsd}
566 @item @samp{m68k-altos-sysv}
567 @item @samp{m68000-hp-hpux}
568 @item @samp{m68000-att-sysv}
569 @item @samp{@var{any}-lynx-lynxos}
570 @item @samp{mips-@var{any}}
573 On the systems listed above (except for the HP-PA, for ISC on the
574 386, and for @samp{mips-sgi-irix5.*}), if you use the GNU assembler,
575 you should also use the GNU linker (and specify @option{--with-gnu-ld}).
577 @item --with-as=@var{pathname}
579 compiler should use the assembler pointed to by @var{pathname}, rather
580 than the one found by the standard rules to find an assembler, which
585 @file{@var{exec_prefix}/lib/gcc-lib/@var{target}/@var{version}}
586 directory, where @var{exec_prefix} defaults to @var{prefix} which
587 defaults to @file{/usr/local} unless overridden by the
588 @option{--prefix=@var{pathname}} switch described above. @var{target} is the
589 target system triple, such as @samp{sparc-sun-solaris2.7}, and
590 @var{version} denotes the GCC version, such as 3.0.
592 Check operating system specific directories (e.g.@: @file{/usr/ccs/bin} on
595 Note that these rules do not check for the value of @env{PATH}. You may
596 want to use @option{--with-as} if no assembler is installed in the
597 directories listed above, or if you have multiple assemblers installed
598 and want to choose one that is not found by the above rules.
600 @item @anchor{with-gnu-ld}--with-gnu-ld
601 Same as @uref{#with-gnu-as,,@option{--with-gnu-as}}
605 @item --with-ld=@var{pathname}
607 @option{--with-as}, but for the linker.
610 Specify that stabs debugging
611 information should be used instead of whatever format the host normally
612 uses. Normally GCC uses the same debug format as the host system.
614 On MIPS based systems and on Alphas, you must specify whether you want
615 GCC to create the normal ECOFF debugging format, or to use BSD-style
616 stabs passed through the ECOFF symbol table. The normal ECOFF debug
617 format cannot fully handle languages other than C@. BSD stabs format can
618 handle other languages, but it only works with the GNU debugger GDB@.
620 Normally, GCC uses the ECOFF debugging format by default; if you
621 prefer BSD stabs, specify @option{--with-stabs} when you configure GCC@.
623 No matter which default you choose when you configure GCC, the user
624 can use the @option{-gcoff} and @option{-gstabs+} options to specify explicitly
625 the debug format for a particular compilation.
627 @option{--with-stabs} is meaningful on the ISC system on the 386, also, if
628 @option{--with-gas} is used. It selects use of stabs debugging
629 information embedded in COFF output. This kind of debugging information
630 supports C++ well; ordinary COFF debugging information does not.
632 @option{--with-stabs} is also meaningful on 386 systems running SVR4. It
633 selects use of stabs debugging information embedded in ELF output. The
634 C++ compiler currently (2.6.0) does not support the DWARF debugging
635 information normally used on 386 SVR4 platforms; stabs provide a
636 workable alternative. This requires gas and gdb, as the normal SVR4
637 tools can not generate or interpret stabs.
639 @item --disable-multilib
640 Specify that multiple target
641 libraries to support different target variants, calling
642 conventions, etc should not be built. The default is to build a
643 predefined set of them.
645 Some targets provide finer-grained control over which multilibs are built
646 (e.g., @option{--disable-softfloat}):
652 fpu, 26bit, underscore, interwork, biendian, nofmult.
655 softfloat, m68881, m68000, m68020.
658 single-float, biendian, softfloat.
660 @item powerpc*-*-*, rs6000*-*-*
661 aix64, pthread, softfloat, powercpu, powerpccpu, powerpcos, biendian,
666 @item --enable-threads
667 Specify that the target
668 supports threads. This affects the Objective-C compiler and runtime
669 library, and exception handling for other languages like C++ and Java.
670 On some systems, this is the default.
672 In general, the best (and, in many cases, the only known) threading
673 model available will be configured for use. Beware that on some
674 systems, gcc has not been taught what threading models are generally
675 available for the system. In this case, @option{--enable-threads} is an
676 alias for @option{--enable-threads=single}.
678 @item --disable-threads
679 Specify that threading support should be disabled for the system.
680 This is an alias for @option{--enable-threads=single}.
682 @item --enable-threads=@var{lib}
684 @var{lib} is the thread support library. This affects the Objective-C
685 compiler and runtime library, and exception handling for other languages
686 like C++ and Java. The possibilities for @var{lib} are:
694 Generic MACH thread support, known to work on NeXTSTEP@. (Please note
695 that the file needed to support this configuration, @file{gthr-mach.h}, is
696 missing and thus this setting will cause a known bootstrap failure.)
698 This is an alias for @samp{single}.
700 Generic POSIX thread support.
702 Same as @samp{posix} on arm*-*-linux*, *-*-chorusos* and *-*-freebsd*
703 only. A future release of gcc might remove this alias or extend it
706 RTEMS thread support.
708 Disable thread support, should work for all platforms.
710 Sun Solaris 2 thread support.
712 VxWorks thread support.
714 Microsoft Win32 API thread support.
717 @item --with-cpu=@var{cpu}
718 Specify which cpu variant the
719 compiler should generate code for by default. This is currently
720 only supported on the some ports, specifically arm, powerpc, and
721 SPARC@. If configure does not recognize the model name (e.g.@: arm700,
722 603e, or ultrasparc) you provide, please check the configure script
723 for a complete list of supported models.
725 @item --enable-altivec
726 Specify that the target supports AltiVec vector enhancements. This
727 option will adjust the ABI for AltiVec enhancements, as well as generate
728 AltiVec code when appropriate. This option is only available for
731 @item --enable-target-optspace
733 libraries should be optimized for code space instead of code speed.
734 This is the default for the m32r platform.
737 Specify that a user visible @command{cpp} program should not be installed.
739 @item --with-cpp-install-dir=@var{dirname}
740 Specify that the user visible @command{cpp} program should be installed
741 in @file{@var{prefix}/@var{dirname}/cpp}, in addition to @var{bindir}.
743 @item --enable-maintainer-mode
745 regenerate the GCC master message catalog @file{gcc.pot} are normally
746 disabled. This is because it can only be rebuilt if the complete source
747 tree is present. If you have changed the sources and want to rebuild the
748 catalog, configuring with @option{--enable-maintainer-mode} will enable
749 this. Note that you need a recent version of the @code{gettext} tools
752 @item --enable-version-specific-runtime-libs
754 that runtime libraries should be installed in the compiler specific
755 subdirectory (@file{@var{libsubdir}}) rather than the usual places. In
756 addition, @samp{libstdc++}'s include files will be installed in
757 @file{@var{libsubdir}/include/g++} unless you overruled it by using
758 @option{--with-gxx-include-dir=@var{dirname}}. Using this option is
759 particularly useful if you intend to use several versions of GCC in
760 parallel. This is currently supported by @samp{libf2c} and
761 @samp{libstdc++}, and is the default for @samp{libobjc} which cannot be
762 changed in this case.
764 @item --enable-languages=@var{lang1},@var{lang2},@dots{}
765 Specify that only a particular subset of compilers and
766 their runtime libraries should be built. For a list of valid values for
767 @var{langN} you can issue the following command in the
768 @file{gcc} directory of your GCC source tree:@*
770 grep language= */config-lang.in
772 Currently, you can use any of the following:
773 @code{ada}, @code{c}, @code{c++}, @code{f77}, @code{java}, @code{objc}.
774 Building the Ada compiler has special requirements, see below.@*
775 If you do not pass this flag, all languages available in the @file{gcc}
776 sub-tree will be configured. Re-defining @code{LANGUAGES} when calling
777 @samp{make bootstrap} @strong{does not} work anymore, as those
778 language sub-directories might not have been configured!
780 @item --disable-libgcj
781 Specify that the run-time libraries
782 used by GCJ should not be built. This is useful in case you intend
783 to use GCJ with some other run-time, or you're going to install it
784 separately, or it just happens not to build on your particular
785 machine. In general, if the Java front end is enabled, the GCJ
786 libraries will be enabled too, unless they're known to not work on
787 the target platform. If GCJ is enabled but @samp{libgcj} isn't built, you
788 may need to port it; in this case, before modifying the top-level
789 @file{configure.in} so that @samp{libgcj} is enabled by default on this platform,
790 you may use @option{--enable-libgcj} to override the default.
793 Specify that the compiler should
794 use DWARF 2 debugging information as the default.
796 @item --enable-win32-registry
797 @itemx --enable-win32-registry=@var{key}
798 @itemx --disable-win32-registry
799 The @option{--enable-win32-registry} option enables Windows-hosted GCC
800 to look up installations paths in the registry using the following key:
803 @code{HKEY_LOCAL_MACHINE\SOFTWARE\Free Software Foundation\@var{key}}
806 @var{key} defaults to GCC version number, and can be overridden by the
807 @option{--enable-win32-registry=@var{key}} option. Vendors and distributors
808 who use custom installers are encouraged to provide a different key,
809 perhaps one comprised of vendor name and GCC version number, to
810 avoid conflict with existing installations. This feature is enabled
811 by default, and can be disabled by @option{--disable-win32-registry}
812 option. This option has no effect on the other hosts.
815 Specify that the machine does not have a floating point unit. This
816 option only applies to @samp{m68k-sun-sunos@var{n}} and
817 @samp{m68k-isi-bsd}. On any other system, @option{--nfp} has no effect.
819 @item --enable-checking
820 @itemx --enable-checking=@var{list}
821 When you specify this option, the compiler is built to perform checking
822 of tree node types when referencing fields of that node, and some other
823 internal consistency checks. This does not change the generated code,
824 but adds error checking within the compiler. This will slow down the
825 compiler and may only work properly if you are building the compiler
826 with GCC@. This is on by default when building from CVS or snapshots,
827 but off for releases. More control over the checks may be had by
828 specifying @var{list}; the categories of checks available are
829 @samp{misc}, @samp{tree}, @samp{gc}, @samp{rtl}, @samp{rtlflag} and
831 default when @var{list} is not specified is @samp{misc,tree,gc,rtlflag}; the
832 checks @samp{rtl} and @samp{gcac} are very expensive.
834 @item --enable-coverage
835 @item --enable-coverage=@var{level}
836 With this option, the compiler is built to collect self coverage
837 information, every time it is run. This is for internal development
838 purposes, and only works when the compiler is being built with gcc. The
839 @var{level} argument controls whether the compiler is built optimized or
840 not, values are @samp{opt} and @samp{noopt}. For coverage analysis you
841 want to disable optimization, for performance analysis you want to
842 enable optimization. When coverage is enabled, the default level is
843 without optimization.
847 The @option{--enable-nls} option enables Native Language Support (NLS),
848 which lets GCC output diagnostics in languages other than American
849 English. Native Language Support is enabled by default if not doing a
850 canadian cross build. The @option{--disable-nls} option disables NLS@.
852 @item --with-included-gettext
853 If NLS is enabled, the @option{--with-included-gettext} option causes the build
854 procedure to prefer its copy of GNU @command{gettext}.
857 If NLS is enabled, and if the host lacks @code{gettext} but has the
858 inferior @code{catgets} interface, the GCC build procedure normally
859 ignores @code{catgets} and instead uses GCC's copy of the GNU
860 @code{gettext} library. The @option{--with-catgets} option causes the
861 build procedure to use the host's @code{catgets} in this situation.
863 @item --with-libiconv-prefix=@var{dir}
864 Search for libiconv header files in @file{@var{dir}/include} and
865 libiconv library files in @file{@var{dir}/lib}.
867 @item --with-system-zlib
868 Use installed zlib rather than that included with GCC@. This option
869 only applies if the Java front end is being built.
871 @item --enable-obsolete
872 Enable configuration for an obsoleted system. If you attempt to
873 configure GCC for a system (build, host, or target) which has been
874 obsoleted, and you do not specify this flag, configure will halt with an
877 All support for systems which have been obsoleted in one release of GCC
878 is removed entirely in the next major release, unless someone steps
879 forward to maintain the port.
882 Some options which only apply to building cross compilers:
885 @itemx --with-headers=@var{dir}
886 Specifies that target headers are available when building a cross compiler.
887 The @var{dir} argument specifies a directory which has the target include
888 files. These include files will be copied into the @file{gcc} install
889 directory. @emph{This option with the @var{dir} argument is required} when
890 building a cross compiler, if @file{@var{prefix}/@var{target}/sys-include}
891 doesn't pre-exist. If @file{@var{prefix}/@var{target}/sys-include} does
892 pre-exist, the @var{dir} argument may be omitted. @command{fixincludes}
893 will be run on these files to make them compatible with GCC.
895 @itemx --with-libs=``@var{dir1} @var{dir2} @dots{} @var{dirN}''
896 Specifies a list of directories which contain the target runtime
897 libraries. These libraries will be copied into the @file{gcc} install
898 directory. If the directory list is omitted, this option has no
901 Specifies that @samp{newlib} is
902 being used as the target C library. This causes @code{__eprintf} to be
903 omitted from @file{libgcc.a} on the assumption that it will be provided by
907 Note that each @option{--enable} option has a corresponding
908 @option{--disable} option and that each @option{--with} option has a
909 corresponding @option{--without} option.
916 @uref{./index.html,,Return to the GCC Installation page}
920 @c ***Building****************************************************************
922 @comment node-name, next, previous, up
923 @node Building, Testing, Configuration, Installing GCC
929 @cindex Installing GCC: Building
931 Now that GCC is configured, you are ready to build the compiler and
934 We @strong{highly} recommend that GCC be built using GNU make;
935 other versions may work, then again they might not.
936 GNU make is required for compiling GNAT (the Ada compiler) and the Java
939 (For example, many broken versions of make will fail if you use the
940 recommended setup where @var{objdir} is different from @var{srcdir}.
941 Other broken versions may recompile parts of the compiler when
942 installing the compiler.)
944 Some commands executed when making the compiler may fail (return a
945 nonzero status) and be ignored by @code{make}. These failures, which
946 are often due to files that were not found, are expected, and can safely
949 It is normal to have compiler warnings when compiling certain files.
950 Unless you are a GCC developer, you can generally ignore these warnings
951 unless they cause compilation to fail.
953 On certain old systems, defining certain environment variables such as
954 @env{CC} can interfere with the functioning of @command{make}.
956 If you encounter seemingly strange errors when trying to build the
957 compiler in a directory other than the source directory, it could be
958 because you have previously configured the compiler in the source
959 directory. Make sure you have done all the necessary preparations.
961 If you build GCC on a BSD system using a directory stored in an old System
962 V file system, problems may occur in running @command{fixincludes} if the
963 System V file system doesn't support symbolic links. These problems
964 result in a failure to fix the declaration of @code{size_t} in
965 @file{sys/types.h}. If you find that @code{size_t} is a signed type and
966 that type mismatches occur, this could be the cause.
968 The solution is not to use such a directory for building GCC@.
970 When building from CVS or snapshots, or if you modify parser sources,
971 you need the Bison parser generator installed. Any version 1.25 or
972 later should work; older versions may also work. If you do not modify
973 parser sources, releases contain the Bison-generated files and you do
974 not need Bison installed to build them.
976 When building from CVS or snapshots, or if you modify Texinfo
977 documentation, you need version 4.1 or later of Texinfo installed if you
978 want Info documentation to be regenerated. Releases contain Info
979 documentation pre-built for the unmodified documentation in the release.
981 @section Building a native compiler
983 For a native build issue the command @samp{make bootstrap}. This
984 will build the entire GCC system, which includes the following steps:
988 Build host tools necessary to build the compiler such as texinfo, bison,
992 Build target tools for use by the compiler such as binutils (bfd,
993 binutils, gas, gprof, ld, and opcodes)
994 if they have been individually linked
995 or moved into the top level GCC source tree before configuring.
998 Perform a 3-stage bootstrap of the compiler.
1001 Perform a comparison test of the stage2 and stage3 compilers.
1004 Build runtime libraries using the stage3 compiler from the previous step.
1008 If you are short on disk space you might consider @samp{make
1009 bootstrap-lean} instead. This is identical to @samp{make
1010 bootstrap} except that object files from the stage1 and
1011 stage2 of the 3-stage bootstrap of the compiler are deleted as
1012 soon as they are no longer needed.
1014 If you want to save additional space during the bootstrap and in
1015 the final installation as well, you can build the compiler binaries
1016 without debugging information as in the following example. This will save
1017 roughly 40% of disk space both for the bootstrap and the final installation.
1018 (Libraries will still contain debugging information.)
1021 make CFLAGS='-O' LIBCFLAGS='-g -O2' \
1022 LIBCXXFLAGS='-g -O2 -fno-implicit-templates' bootstrap
1025 If you wish to use non-default GCC flags when compiling the stage2 and
1026 stage3 compilers, set @code{BOOT_CFLAGS} on the command line when doing
1027 @samp{make bootstrap}. Non-default optimization flags are less well
1028 tested here than the default of @samp{-g -O2}, but should still work.
1029 In a few cases, you may find that you need to specify special flags such
1030 as @option{-msoft-float} here to complete the bootstrap; or, if the
1031 native compiler miscompiles the stage1 compiler, you may need to work
1032 around this, by choosing @code{BOOT_CFLAGS} to avoid the parts of the
1033 stage1 compiler that were miscompiled, or by using @samp{make
1034 bootstrap4} to increase the number of stages of bootstrap.
1036 If you used the flag @option{--enable-languages=@dots{}} to restrict
1037 the compilers to be built, only those you've actually enabled will be
1038 built. This will of course only build those runtime libraries, for
1039 which the particular compiler has been built. Please note,
1040 that re-defining @env{LANGUAGES} when calling @samp{make bootstrap}
1041 @strong{does not} work anymore!
1043 If the comparison of stage2 and stage3 fails, this normally indicates
1044 that the stage2 compiler has compiled GCC incorrectly, and is therefore
1045 a potentially serious bug which you should investigate and report. (On
1046 a few systems, meaningful comparison of object files is impossible; they
1047 always appear ``different''. If you encounter this problem, you will
1048 need to disable comparison in the @file{Makefile}.)
1050 @section Building a cross compiler
1052 We recommend reading the
1053 @uref{http://www.objsw.com/CrossGCC/,,crossgcc FAQ}
1054 for information about building cross compilers.
1056 When building a cross compiler, it is not generally possible to do a
1057 3-stage bootstrap of the compiler. This makes for an interesting problem
1058 as parts of GCC can only be built with GCC@.
1060 To build a cross compiler, we first recommend building and installing a
1061 native compiler. You can then use the native GCC compiler to build the
1062 cross compiler. The installed native compiler needs to be GCC version
1065 Assuming you have already installed a native copy of GCC and configured
1066 your cross compiler, issue the command @command{make}, which performs the
1071 Build host tools necessary to build the compiler such as texinfo, bison,
1075 Build target tools for use by the compiler such as binutils (bfd,
1076 binutils, gas, gprof, ld, and opcodes)
1077 if they have been individually linked or moved into the top level GCC source
1078 tree before configuring.
1081 Build the compiler (single stage only).
1084 Build runtime libraries using the compiler from the previous step.
1087 Note that if an error occurs in any step the make process will exit.
1089 @section Building in parallel
1091 If you have a multiprocessor system you can use @samp{make bootstrap
1092 MAKE="make -j 2" -j 2} or just @samp{make -j 2 bootstrap}
1093 for GNU Make 3.79 and above instead of just @samp{make bootstrap}
1094 when building GCC@. You can use a bigger number instead of two if
1095 you like. In most cases, it won't help to use a number bigger than
1096 the number of processors in your machine.
1098 @section Building the Ada compiler
1100 In order to build GNAT, the Ada compiler, you need a working GNAT
1101 compiler (GNAT version 3.13 or later, or GCC version 3.1 or later),
1102 since the Ada front end is written in Ada (with some
1103 GNAT-specific extensions), and GNU make.
1105 However, you do not need a full installation of GNAT, just the GNAT
1106 binary @file{gnat1}, a copy of @file{gnatbind}, and a compiler driver
1107 which can deal with Ada input (by invoking the @file{gnat1} binary).
1108 You can specify this compiler driver by setting the @env{ADAC}
1109 environment variable at the configure step. @command{configure} can
1110 detect the driver automatically if it has got a common name such as
1111 @command{gcc} or @command{gnatgcc}. Of course, you still need a working
1112 C compiler (the compiler driver can be different or not).
1113 @command{configure} does not test whether the GNAT installation works
1114 and has a sufficiently recent version; if too old a GNAT version is
1115 installed, the build will fail unless @option{--enable-languages} is
1116 used to disable building the Ada front end.
1118 Additional build tools (such as @command{gnatmake}) or a working GNAT
1119 run-time library installation are usually @emph{not} required. However,
1120 if you want to bootstrap the compiler using a minimal version of GNAT,
1121 you have to issue the following commands before invoking @samp{make
1122 bootstrap} (this assumes that you start with an unmodified and consistent
1123 source distribution):
1126 cd @var{srcdir}/gcc/ada
1127 touch treeprs.ads [es]info.h nmake.ad[bs]
1130 At the moment, the GNAT library and several tools for GNAT are not built
1131 by @samp{make bootstrap}. You have to invoke
1132 @samp{make gnatlib_and_tools} in the @file{@var{objdir}/gcc}
1133 subdirectory before proceeding with the next steps.
1135 For example, you can build a native Ada compiler by issuing the
1136 following commands (assuming @command{make} is GNU make):
1140 @var{srcdir}/configure --enable-languages=c,ada
1141 cd @var{srcdir}/gcc/ada
1142 touch treeprs.ads [es]info.h nmake.ad[bs]
1146 make gnatlib_and_tools
1150 Currently, when compiling the Ada front end, you cannot use the parallel
1151 build feature described in the previous section.
1158 @uref{./index.html,,Return to the GCC Installation page}
1162 @c ***Testing*****************************************************************
1164 @comment node-name, next, previous, up
1165 @node Testing, Final install, Building, Installing GCC
1169 @chapter Installing GCC: Testing
1172 @cindex Installing GCC: Testing
1175 Before you install GCC, we encourage you to run the testsuites and to
1176 compare your results with results from a similar configuration that have
1177 been submitted to the
1178 @uref{http://gcc.gnu.org/ml/gcc-testresults/,,gcc-testresults mailing list}.
1179 This step is optional and may require you to download additional software,
1180 but it can give you confidence in your new GCC installation or point out
1181 problems before you install and start using your new GCC.
1183 First, you must have @uref{download.html,,downloaded the testsuites}.
1184 These are part of the full distribution, but if you downloaded the
1185 ``core'' compiler plus any front ends, you must download the testsuites
1188 Second, you must have the testing tools installed. This includes
1189 a @uref{http://www.gnu.org/software/dejagnu/,,current version of DejaGnu};
1190 dejagnu 1.3 is not sufficient.
1191 It also includes Tcl and Expect; the DejaGnu site has links to these.
1193 Now you may need specific preparations:
1198 The following environment variables may need to be set appropriately, as in
1199 the following example (which assumes that DejaGnu has been installed
1200 under @file{/usr/local}):
1203 TCL_LIBRARY = /usr/local/share/tcl8.0
1204 DEJAGNULIBS = /usr/local/share/dejagnu
1207 On systems such as Cygwin, these paths are required to be actual
1208 paths, not mounts or links; presumably this is due to some lack of
1209 portability in the DejaGnu code.
1211 If the directories where @command{runtest} and @command{expect} were
1212 installed are in the @env{PATH}, it should not be necessary to set these
1213 environment variables.
1217 Finally, you can run the testsuite (which may take a long time):
1219 cd @var{objdir}; make -k check
1222 The testing process will try to test as many components in the GCC
1223 distribution as possible, including the C, C++, Objective-C and Fortran
1224 compilers as well as the C++ and Java runtime libraries.
1226 While running the testsuite, DejaGnu might emit messages resembling
1227 @samp{WARNING: Couldn't find the global config file.} or
1228 @samp{WARNING: Couldn't find tool init file}.
1229 These messages are harmless and do not affect the validity of the tests.
1231 @section How can I run the test suite on selected tests?
1233 As a first possibility to cut down the number of tests that are run it is
1234 possible to use @samp{make check-gcc} or @samp{make check-g++}
1235 in the @file{gcc} subdirectory of the object directory. To further cut down the
1236 tests the following is possible:
1239 make check-gcc RUNTESTFLAGS="execute.exp @var{other-options}"
1242 This will run all @command{gcc} execute tests in the testsuite.
1245 make check-g++ RUNTESTFLAGS="old-deja.exp=9805* @var{other-options}"
1248 This will run the @command{g++} ``old-deja'' tests in the testsuite where the filename
1249 matches @samp{9805*}.
1251 The @file{*.exp} files are located in the testsuite directories of the GCC
1252 source, the most important ones being @file{compile.exp},
1253 @file{execute.exp}, @file{dg.exp} and @file{old-deja.exp}.
1254 To get a list of the possible @file{*.exp} files, pipe the
1255 output of @samp{make check} into a file and look at the
1256 @samp{Running @dots{} .exp} lines.
1258 @section How to interpret test results
1260 After the testsuite has run you'll find various @file{*.sum} and @file{*.log}
1261 files in the testsuite subdirectories. The @file{*.log} files contain a
1262 detailed log of the compiler invocations and the corresponding
1263 results, the @file{*.sum} files summarize the results. These summaries list
1264 all the tests that have been run with a corresponding status code:
1268 PASS: the test passed as expected
1270 XPASS: the test unexpectedly passed
1272 FAIL: the test unexpectedly failed
1274 XFAIL: the test failed as expected
1276 UNSUPPORTED: the test is not supported on this platform
1278 ERROR: the testsuite detected an error
1280 WARNING: the testsuite detected a possible problem
1283 It is normal for some tests to report unexpected failures. At the
1284 current time our testing harness does not allow fine grained control
1285 over whether or not a test is expected to fail. We expect to fix this
1286 problem in future releases.
1289 @section Submitting test results
1291 If you want to report the results to the GCC project, use the
1292 @file{contrib/test_summary} shell script. Start it in the @var{objdir} with
1295 @var{srcdir}/contrib/test_summary -p your_commentary.txt \
1296 -m gcc-testresults@@gcc.gnu.org |sh
1299 This script uses the @command{Mail} program to send the results, so
1300 make sure it is in your @env{PATH}. The file @file{your_commentary.txt} is
1301 prepended to the testsuite summary and should contain any special
1302 remarks you have on your results or your build environment. Please
1303 do not edit the testsuite result block or the subject line, as these
1304 messages are automatically parsed and presented at the
1305 @uref{http://gcc.gnu.org/testresults/,,GCC testresults} web
1306 page. Here you can also gather information on how specific tests
1307 behave on different platforms and compare them with your results. A
1308 few failing testcases are possible even on released versions and you
1309 should look here first if you think your results are unreasonable.
1316 @uref{./index.html,,Return to the GCC Installation page}
1320 @c ***Final install***********************************************************
1322 @comment node-name, next, previous, up
1323 @node Final install, , Testing, Installing GCC
1325 @ifset finalinstallhtml
1327 @chapter Installing GCC: Final installation
1330 Now that GCC has been built (and optionally tested), you can install it with
1332 cd @var{objdir}; make install
1335 We strongly recommend to install into a target directory where there is
1336 no previous version of GCC present.
1338 That step completes the installation of GCC; user level binaries can
1339 be found in @file{@var{prefix}/bin} where @var{prefix} is the value you
1340 specified with the @option{--prefix} to configure (or @file{/usr/local}
1341 by default). (If you specified @option{--bindir}, that directory will
1342 be used instead; otherwise, if you specified @option{--exec-prefix},
1343 @file{@var{exec-prefix}/bin} will be used.) Headers for the C++ and
1344 Java libraries are installed in @file{@var{prefix}/include}; libraries
1345 in @file{@var{libdir}} (normally @file{@var{prefix}/lib}); internal
1346 parts of the compiler in @file{@var{libdir}/gcc-lib}; documentation in
1347 info format in @file{@var{infodir}} (normally @file{@var{prefix}/info}).
1349 If you built a released version of GCC using @samp{make bootstrap} then please
1350 quickly review the build status page for your release, available from
1351 @uref{http://gcc.gnu.org/buildstat.html}.
1352 If your system is not listed for the version of GCC that you built,
1354 @email{gcc@@gcc.gnu.org} indicating
1355 that you successfully built and installed GCC.
1356 Include the following information:
1360 Output from running @file{@var{srcdir}/config.guess}. Do not send us
1361 that file itself, just the one-line output from running it.
1364 The output of @samp{gcc -v} for your newly installed gcc.
1365 This tells us which version of GCC you built and the options you passed to
1369 Whether you enabled all languages or a subset of them. If you used a
1370 full distribution then this information is part of the configure
1371 options in the output of @samp{gcc -v}, but if you downloaded the
1372 ``core'' compiler plus additional front ends then it isn't apparent
1373 which ones you built unless you tell us about it.
1376 If the build was for GNU/Linux, also include:
1379 The distribution name and version (e.g., Red Hat 7.1 or Debian 2.2.3);
1380 this information should be available from @file{/etc/issue}.
1383 The version of the Linux kernel, available from @samp{uname --version}
1387 The version of glibc you used; for RPM-based systems like Red Hat,
1388 Mandrake, and SuSE type @samp{rpm -q glibc} to get the glibc version,
1389 and on systems like Debian and Progeny use @samp{dpkg -l libc6}.
1391 For other systems, you can include similar information if you think it is
1395 Any other information that you think would be useful to people building
1396 GCC on the same configuration. The new entry in the build status list
1397 will include a link to the archived copy of your message.
1400 We'd also like to know if the
1402 @ref{Specific, host/target specific installation notes}
1405 @uref{specific.html,,host/target specific installation notes}
1407 didn't include your host/target information or if that information is
1408 incomplete or out of date. Send a note to
1409 @email{gcc@@gcc.gnu.org} telling us how the information should be changed.
1411 If you find a bug, please report it following our
1412 @uref{../bugs.html,,bug reporting guidelines}.
1414 If you want to print the GCC manuals, do @samp{cd @var{objdir}; make
1415 dvi}. You will need to have @command{texi2dvi} (version at least 4.1)
1416 and @TeX{} installed. This creates a number of @file{.dvi} files in
1417 subdirectories of @file{@var{objdir}}; these may be converted for
1418 printing with programs such as @command{dvips}. You can also
1419 @uref{http://www.gnu.org/order/order.html,,buy printed manuals from the
1420 Free Software Foundation}, though such manuals may not be for the most
1421 recent version of GCC@.
1428 @uref{./index.html,,Return to the GCC Installation page}
1432 @c ***Binaries****************************************************************
1434 @comment node-name, next, previous, up
1435 @node Binaries, Specific, Installing GCC, Top
1439 @chapter Installing GCC: Binaries
1442 @cindex Installing GCC: Binaries
1444 We are often asked about pre-compiled versions of GCC@. While we cannot
1445 provide these for all platforms, below you'll find links to binaries for
1446 various platforms where creating them by yourself is not easy due to various
1449 Please note that we did not create these binaries, nor do we
1450 support them. If you have any problems installing them, please
1451 contact their makers.
1458 @uref{http://www.bullfreeware.com,,Bull's Freeware and Shareware Archive for AIX};
1461 @uref{http://aixpdslib.seas.ucla.edu,,UCLA Software Library for AIX}.
1465 DOS---@uref{http://www.delorie.com/djgpp/,,DJGPP}.
1468 Hitachi H8/300[HS]---@uref{http://h8300-hms.sourceforge.net/,,GNU
1469 Development Tools for the Hitachi H8/300[HS] Series}.
1475 @uref{http://hpux.cae.wisc.edu/,,HP-UX Porting Center};
1478 @uref{ftp://sunsite.informatik.rwth-aachen.de/pub/packages/gcc_hpux/,,Binaries for HP-UX 11.00 at Aachen University of Technology}.
1482 @uref{http://www.sco.com/skunkware/devtools/index.html#gcc,,SCO
1483 OpenServer/Unixware}.
1486 Sinix/Reliant Unix---@uref{ftp://ftp.siemens.de/sni/mr/pd/gnu/gcc,,Siemens}.
1489 Solaris 2 (SPARC, Intel)---@uref{http://www.sunfreeware.com/,,Sunfreeware}.
1492 SGI---@uref{http://freeware.sgi.com/,,SGI Freeware}.
1495 Windows 95, 98, and NT:
1498 The @uref{http://sources.redhat.com/cygwin/,,Cygwin} project;
1500 The @uref{http://www.mingw.org/,,MinGW} project.
1504 @uref{ftp://ftp.thewrittenword.com/packages/free/by-name/,,The
1505 Written Word} offers binaries for Solaris 2.5.1, 2.6, 2.7/SPARC, 2.7/Intel,
1506 IRIX 6.2, 6.5, Digital UNIX 4.0D, HP-UX 10.20, and HP-UX 11.00.
1510 In addition to those specific offerings, you can get a binary
1511 distribution CD-ROM from the
1512 @uref{http://www.fsf.org/order/order.html,,Free Software Foundation}.
1513 It contains binaries for a number of platforms, and
1514 includes not only GCC, but other stuff as well. The current CD does
1515 not contain the latest version of GCC, but it should allow
1516 bootstrapping the compiler. An updated version of that disk is in the
1524 @uref{./index.html,,Return to the GCC Installation page}
1528 @c ***Specific****************************************************************
1530 @comment node-name, next, previous, up
1531 @node Specific, Old, Binaries, Top
1535 @chapter Host/target specific installation notes for GCC
1538 @cindex Specific installation notes
1539 @cindex Target specific installation
1540 @cindex Host specific installation
1541 @cindex Target specific installation notes
1543 Please read this document carefully @emph{before} installing the
1544 GNU Compiler Collection on your machine.
1549 @uref{#alpha*-*-*,,alpha*-*-*}
1551 @uref{#alpha*-dec-osf*,,alpha*-dec-osf*}
1553 @uref{#alphaev5-cray-unicosmk*,,alphaev5-cray-unicosmk*}
1555 @uref{#arc-*-elf,,arc-*-elf}
1557 @uref{#arm-*-aout,,arm-*-aout}
1559 @uref{#arm-*-elf,,arm-*-elf}
1561 @uref{#arm*-*-linux-gnu,,arm*-*-linux-gnu}
1569 @uref{#dsp16xx,,dsp16xx}
1571 @uref{#*-*-freebsd*,,*-*-freebsd*}
1573 @uref{#h8300-hms,,h8300-hms}
1575 @uref{#hppa*-hp-hpux*,,hppa*-hp-hpux*}
1577 @uref{#hppa*-hp-hpux9,,hppa*-hp-hpux9}
1579 @uref{#hppa*-hp-hpux10,,hppa*-hp-hpux10}
1581 @uref{#hppa*-hp-hpux11,,hppa*-hp-hpux11}
1583 @uref{#i370-*-*,,i370-*-*}
1585 @uref{#*-*-linux-gnu,,*-*-linux-gnu}
1587 @uref{#ix86-*-linux*aout,,i?86-*-linux*aout}
1589 @uref{#ix86-*-linux*,,i?86-*-linux*}
1591 @uref{#ix86-*-sco,,i?86-*-sco}
1593 @uref{#ix86-*-sco3.2v4,,i?86-*-sco3.2v4}
1595 @uref{#ix86-*-sco3.2v5*,,i?86-*-sco3.2v5*}
1597 @uref{#ix86-*-udk,,i?86-*-udk}
1599 @uref{#ix86-*-esix,,i?86-*-esix}
1601 @uref{#ia64-*-linux,,ia64-*-linux}
1603 @uref{#*-lynx-lynxos,,*-lynx-lynxos}
1605 @uref{#*-ibm-aix*,,*-ibm-aix*}
1607 @uref{#ip2k-*-elf,,ip2k-*-elf}
1609 @uref{#m32r-*-elf,,m32r-*-elf}
1611 @uref{#m68000-hp-bsd,,m68000-hp-bsd}
1613 @uref{#m6811-elf,,m6811-elf}
1615 @uref{#m6812-elf,,m6812-elf}
1617 @uref{#m68k-att-sysv,,m68k-att-sysv}
1619 @uref{#m68k-crds-unos,,m68k-crds-unos}
1621 @uref{#m68k-hp-hpux,,m68k-hp-hpux}
1623 @uref{#m68k-ncr-*,,m68k-ncr-*}
1625 @uref{#m68k-sun,,m68k-sun}
1627 @uref{#m68k-sun-sunos4.1.1,,m68k-sun-sunos4.1.1}
1629 @uref{#mips-*-*,,mips-*-*}
1631 @uref{#mips-sgi-irix5,,mips-sgi-irix5}
1633 @uref{#mips-sgi-irix6,,mips-sgi-irix6}
1635 @uref{#powerpc*-*-*,,powerpc*-*-*, powerpc-*-sysv4}
1637 @uref{#powerpc-*-darwin*,,powerpc-*-darwin*}
1639 @uref{#powerpc-*-elf,,powerpc-*-elf, powerpc-*-sysv4}
1641 @uref{#powerpc-*-linux-gnu*,,powerpc-*-linux-gnu*}
1643 @uref{#powerpc-*-netbsd*,,powerpc-*-netbsd*}
1645 @uref{#powerpc-*-eabiaix,,powerpc-*-eabiaix}
1647 @uref{#powerpc-*-eabisim,,powerpc-*-eabisim}
1649 @uref{#powerpc-*-eabi,,powerpc-*-eabi}
1651 @uref{#powerpcle-*-elf,,powerpcle-*-elf, powerpcle-*-sysv4}
1653 @uref{#powerpcle-*-eabisim,,powerpcle-*-eabisim}
1655 @uref{#powerpcle-*-eabi,,powerpcle-*-eabi}
1657 @uref{#powerpcle-*-winnt,,powerpcle-*-winnt, powerpcle-*-pe}
1659 @uref{#s390-*-linux*,,s390-*-linux*}
1661 @uref{#s390x-*-linux*,,s390x-*-linux*}
1663 @uref{#*-*-solaris2*,,*-*-solaris2*}
1665 @uref{#sparc-sun-solaris2*,,sparc-sun-solaris2*}
1667 @uref{#sparc-sun-solaris2.7,,sparc-sun-solaris2.7}
1669 @uref{#sparc-sun-sunos4*,,sparc-sun-sunos4*}
1671 @uref{#sparc-unknown-linux-gnulibc1,,sparc-unknown-linux-gnulibc1}
1673 @uref{#sparc-*-linux*,,sparc-*-linux*}
1675 @uref{#sparc64-*-*,,sparc64-*-*}
1677 @uref{#sparcv9-*-solaris2*,,sparcv9-*-solaris2*}
1679 @uref{#*-*-sysv*,,*-*-sysv*}
1681 @uref{#vax-dec-ultrix,,vax-dec-ultrix}
1683 @uref{#xtensa-*-elf,,xtensa-*-elf}
1685 @uref{#xtensa-*-linux*,,xtensa-*-linux*}
1687 @uref{#windows,,Microsoft Windows}
1691 @uref{#older,,Older systems}
1696 @uref{#elf_targets,,all ELF targets} (SVR4, Solaris 2, etc.)
1702 <!-- -------- host/target specific issues start here ---------------- -->
1705 @heading @anchor{alpha*-*-*}alpha*-*-*
1707 This section contains general configuration information for all
1708 alpha-based platforms using ELF (in particular, ignore this section for
1709 DEC OSF/1, Digital UNIX and Tru64 UNIX)@. In addition to reading this
1710 section, please read all other sections that match your target.
1712 We require binutils 2.11.2 or newer.
1713 Previous binutils releases had a number of problems with DWARF 2
1714 debugging information, not the least of which is incorrect linking of
1720 @heading @anchor{alpha*-dec-osf*}alpha*-dec-osf*
1721 Systems using processors that implement the DEC Alpha architecture and
1722 are running the DEC/Compaq Unix (DEC OSF/1, Digital UNIX, or Compaq
1723 Tru64 UNIX) operating system, for example the DEC Alpha AXP systems.
1725 As of GCC 3.2, versions before @code{alpha*-dec-osf4} are no longer
1726 supported. (These are the versions which identify themselves as DEC
1729 In Digital Unix V4.0, virtual memory exhausted bootstrap failures
1730 may be fixed by configuring with @option{--with-gc=simple},
1731 reconfiguring Kernel Virtual Memory and Swap parameters
1732 per the @command{/usr/sbin/sys_check} Tuning Suggestions,
1733 or applying the patch in
1734 @uref{http://gcc.gnu.org/ml/gcc/2002-08/msg00822.html}.
1736 In Tru64 UNIX V5.1, Compaq introduced a new assembler that does not
1737 currently (2001-06-13) work with @command{mips-tfile}. As a workaround,
1738 we need to use the old assembler, invoked via the barely documented
1739 @option{-oldas} option. To bootstrap GCC, you either need to use the
1743 % CC=cc @var{srcdir}/configure [@var{options}] [@var{target}]
1746 or you can use a copy of GCC 2.95.3 or higher built on Tru64 UNIX V4.0:
1749 % CC=gcc -Wa,-oldas @var{srcdir}/configure [@var{options}] [@var{target}]
1752 As of GNU binutils 2.11.2, neither GNU @command{as} nor GNU @command{ld}
1753 are supported on Tru64 UNIX, so you must not configure GCC with
1754 @option{--with-gnu-as} or @option{--with-gnu-ld}.
1756 The @option{--enable-threads} options isn't supported yet. A patch is
1757 in preparation for a future release.
1759 GCC writes a @samp{.verstamp} directive to the assembler output file
1760 unless it is built as a cross-compiler. It gets the version to use from
1761 the system header file @file{/usr/include/stamp.h}. If you install a
1762 new version of DEC Unix, you should rebuild GCC to pick up the new version
1765 Note that since the Alpha is a 64-bit architecture, cross-compilers from
1766 32-bit machines will not generate code as efficient as that generated
1767 when the compiler is running on a 64-bit machine because many
1768 optimizations that depend on being able to represent a word on the
1769 target in an integral value on the host cannot be performed. Building
1770 cross-compilers on the Alpha for 32-bit machines has only been tested in
1771 a few cases and may not work properly.
1773 @code{make compare} may fail on old versions of DEC Unix unless you add
1774 @option{-save-temps} to @code{CFLAGS}. On these systems, the name of the
1775 assembler input file is stored in the object file, and that makes
1776 comparison fail if it differs between the @code{stage1} and
1777 @code{stage2} compilations. The option @option{-save-temps} forces a
1778 fixed name to be used for the assembler input file, instead of a
1779 randomly chosen name in @file{/tmp}. Do not add @option{-save-temps}
1780 unless the comparisons fail without that option. If you add
1781 @option{-save-temps}, you will have to manually delete the @samp{.i} and
1782 @samp{.s} files after each series of compilations.
1784 GCC now supports both the native (ECOFF) debugging format used by DBX
1785 and GDB and an encapsulated STABS format for use only with GDB@. See the
1786 discussion of the @option{--with-stabs} option of @file{configure} above
1787 for more information on these formats and how to select them.
1789 There is a bug in DEC's assembler that produces incorrect line numbers
1790 for ECOFF format when the @samp{.align} directive is used. To work
1791 around this problem, GCC will not emit such alignment directives
1792 while writing ECOFF format debugging information even if optimization is
1793 being performed. Unfortunately, this has the very undesirable
1794 side-effect that code addresses when @option{-O} is specified are
1795 different depending on whether or not @option{-g} is also specified.
1797 To avoid this behavior, specify @option{-gstabs+} and use GDB instead of
1798 DBX@. DEC is now aware of this problem with the assembler and hopes to
1799 provide a fix shortly.
1804 @heading @anchor{alphaev5-cray-unicosmk*}alphaev5-cray-unicosmk*
1805 Cray T3E systems running Unicos/Mk.
1807 This port is incomplete and has many known bugs. We hope to improve the
1808 support for this target soon. Currently, only the C front end is supported,
1809 and it is not possible to build parallel applications. Cray modules are not
1810 supported; in particular, Craylibs are assumed to be in
1811 @file{/opt/ctl/craylibs/craylibs}.
1813 You absolutely @strong{must} use GNU make on this platform. Also, you
1814 need to tell GCC where to find the assembler and the linker. The
1815 simplest way to do so is by providing @option{--with-as} and
1816 @option{--with-ld} to @file{configure}, e.g.@:
1819 configure --with-as=/opt/ctl/bin/cam --with-ld=/opt/ctl/bin/cld \
1820 --enable-languages=c
1823 The comparison test during @samp{make bootstrap} fails on Unicos/Mk
1824 because the assembler inserts timestamps into object files. You should
1825 be able to work around this by doing @samp{make all} after getting this
1831 @heading @anchor{arc-*-elf}arc-*-elf
1832 Argonaut ARC processor.
1833 This configuration is intended for embedded systems.
1838 @heading @anchor{arm-*-aout}arm-*-aout
1839 Advanced RISC Machines ARM-family processors. These are often used in
1840 embedded applications. There are no standard Unix configurations.
1841 This configuration corresponds to the basic instruction sequences and will
1842 produce @file{a.out} format object modules.
1844 You may need to make a variant of the file @file{arm.h} for your particular
1850 @heading @anchor{arm-*-elf}arm-*-elf
1851 This configuration is intended for embedded systems.
1856 @heading @anchor{arm*-*-linux-gnu}arm*-*-linux-gnu
1858 We require GNU binutils 2.10 or newer.
1863 @heading @anchor{avr}avr
1865 ATMEL AVR-family micro controllers. These are used in embedded
1866 applications. There are no standard Unix configurations.
1868 @xref{AVR Options,, AVR Options, gcc, Using and Porting the GNU Compiler
1872 See ``AVR Options'' in the main manual
1874 for the list of supported MCU types.
1876 Use @samp{configure --target=avr --enable-languages="c"} to configure GCC@.
1878 Further installation notes and other useful information about AVR tools
1879 can also be obtained from:
1883 @uref{http://home.overta.ru/users/denisc,,http://home.overta.ru/users/denisc}
1885 @uref{http://www.amelek.gda.pl/avr,,http://www.amelek.gda.pl/avr}
1888 We @emph{strongly} recommend using binutils 2.11 or newer.
1890 The following error:
1892 Error: register required
1895 indicates that you should upgrade to a newer version of the binutils.
1900 @heading @anchor{c4x}c4x
1902 Texas Instruments TMS320C3x and TMS320C4x Floating Point Digital Signal
1903 Processors. These are used in embedded applications. There are no
1904 standard Unix configurations.
1906 @xref{TMS320C3x/C4x Options,, TMS320C3x/C4x Options, gcc, Using and
1907 Porting the GNU Compiler Collection (GCC)},
1910 See ``TMS320C3x/C4x Options'' in the main manual
1912 for the list of supported MCU types.
1914 GCC can be configured as a cross compiler for both the C3x and C4x
1915 architectures on the same system. Use @samp{configure --target=c4x
1916 --enable-languages="c,c++"} to configure.
1919 Further installation notes and other useful information about C4x tools
1920 can also be obtained from:
1924 @uref{http://www.elec.canterbury.ac.nz/c4x/,,http://www.elec.canterbury.ac.nz/c4x/}
1930 @heading @anchor{cris}CRIS
1932 CRIS is the CPU architecture in Axis Communications ETRAX system-on-a-chip
1933 series. These are used in embedded applications.
1936 @xref{CRIS Options,, CRIS Options, gcc, Using and Porting the GNU Compiler
1940 See ``CRIS Options'' in the main manual
1942 for a list of CRIS-specific options.
1944 There are a few different CRIS targets:
1946 @item cris-axis-aout
1947 Old target. Includes a multilib for the @samp{elinux} a.out-based
1948 target. No multilibs for newer architecture variants.
1950 Mainly for monolithic embedded systems. Includes a multilib for the
1951 @samp{v10} core used in @samp{ETRAX 100 LX}.
1952 @item cris-axis-linux-gnu
1953 A GNU/Linux port for the CRIS architecture, currently targeting
1954 @samp{ETRAX 100 LX} by default.
1957 For @code{cris-axis-aout} and @code{cris-axis-elf} you need binutils 2.11
1958 or newer. For @code{cris-axis-linux-gnu} you need binutils 2.12 or newer.
1960 Pre-packaged tools can be obtained from
1961 @uref{ftp://ftp.axis.com/pub/axis/tools/cris/compiler-kit/}. More
1962 information about this platform is available at
1963 @uref{http://developer.axis.com/}.
1968 @heading @anchor{dos}DOS
1970 Please have a look at our @uref{binaries.html,,binaries page}.
1972 You cannot install GCC by itself on MSDOS; it will not compile under
1973 any MSDOS compiler except itself. You need to get the complete
1974 compilation package DJGPP, which includes binaries as well as sources,
1975 and includes all the necessary compilation tools and libraries.
1980 @heading @anchor{dsp16xx}dsp16xx
1981 A port to the AT&T DSP1610 family of processors.
1986 @heading @anchor{*-*-freebsd*}*-*-freebsd*
1988 The version of binutils installed in @file{/usr/bin} is known to work unless
1989 otherwise specified in any per-architecture notes. However, binutils
1990 2.12.1 or greater is known to improve overall testsuite results.
1992 FreeBSD 1 is no longer supported in GCC 3.2.
1994 For FreeBSD 2 or any mutant a.out versions of FreeBSD 3: All
1995 configuration support and files as shipped with GCC 2.95 are still in
1996 place. FreeBSD 2.2.7 has been known to bootstrap completely; however,
1997 it is unknown which version of binutils was used (it is assumed that it
1998 was the system copy in @file{/usr/bin}) and C++ EH failures were noted.
2000 For FreeBSD using the ELF file format: DWARF 2 debugging is now the
2001 default for all CPU architectures. It had been the default on
2002 FreeBSD/alpha since its inception. You may use @option{-gstabs} instead
2003 of @option{-g}, if you really want the old debugging format. There are
2004 no known issues with mixing object files and libraries with different
2005 debugging formats. Otherwise, this release of GCC should now match more
2006 of the configuration used in the stock FreeBSD configuration of GCC. In
2007 particular, @option{--enable-threads} is now configured by default.
2008 However, as a general user, do not attempt to replace the system
2009 compiler with this release. Known to bootstrap and check with good
2010 results on FreeBSD 3.0, 3.4, 4.0, 4.2, 4.3, 4.4, 4.5-STABLE and 5-CURRENT@.
2012 In principle, @option{--enable-threads} is now compatible with
2013 @option{--enable-libgcj} on FreeBSD@. However, it has only been built
2014 and tested on @samp{i386-*-freebsd4.5} and @samp{alpha-*-freebsd5.0}.
2016 library may be incorrectly built (symbols are missing at link time).
2017 There is a rare timing-based startup hang (probably involves an
2018 assupmtion about the thread library). Multi-threaded boehm-gc (required for
2019 libjava) exposes severe threaded signal-handling bugs on FreeBSD before
2020 4.5-RELEASE. The alpha port may not fully bootstrap without some manual
2021 intervention: @command{gcjh} will crash with a floating-point exception while
2022 generating @file{java/lang/Double.h} (just copy the version built on
2023 @samp{i386-*-freebsd*} and rerun the top-level @command{gmake} with no
2025 should properly complete the bootstrap). Other CPU architectures
2026 supported by FreeBSD will require additional configuration tuning in, at
2027 the very least, both boehm-gc and libffi.
2029 Shared @file{libgcc_s.so} is now built and installed by default.
2034 @heading @anchor{h8300-hms}h8300-hms
2035 Hitachi H8/300 series of processors.
2037 Please have a look at our @uref{binaries.html,,binaries page}.
2039 The calling convention and structure layout has changed in release 2.6.
2040 All code must be recompiled. The calling convention now passes the
2041 first three arguments in function calls in registers. Structures are no
2042 longer a multiple of 2 bytes.
2047 @heading @anchor{hppa*-hp-hpux*}hppa*-hp-hpux*
2049 We @emph{highly} recommend using gas/binutils 2.8 or newer on all hppa
2050 platforms; you may encounter a variety of problems when using the HP
2051 assembler. The HP assembler does not work with the @samp{hppa64-hp-hpux11*}
2054 Specifically, @option{-g} does not work on HP-UX (since that system
2055 uses a peculiar debugging format which GCC does not know about), unless you
2056 use GAS and GDB and configure GCC with the
2057 @uref{./configure.html#with-gnu-as,,@option{--with-gnu-as}} and
2058 @option{--with-as=@dots{}} options.
2060 If you wish to use the pa-risc 2.0 architecture support with a 32-bit
2061 runtime, you must use either the HP assembler, gas/binutils 2.11 or newer,
2063 @uref{ftp://sources.redhat.com/pub/binutils/snapshots,,snapshot of gas}.
2065 There are two default scheduling models for instructions. These are
2066 PROCESSOR_7100LC and PROCESSOR_8000. They are selected from the pa-risc
2067 architecture specified for the target machine when configuring.
2068 PROCESSOR_8000 is the default. PROCESSOR_7100LC is selected when
2069 the target is a @samp{hppa1*} machine.
2071 The PROCESSOR_8000 model is not well suited to older processors. Thus,
2072 it is important to completely specify the machine architecture when
2073 configuring if you want a model other than PROCESSOR_8000. The macro
2074 TARGET_SCHED_DEFAULT can be defined in BOOT_CFLAGS if a different
2075 default scheduling model is desired.
2077 More specific information to @samp{hppa*-hp-hpux*} targets follows.
2082 @heading @anchor{hppa*-hp-hpux9}hppa*-hp-hpux9
2084 The HP assembler has major problems on this platform. We've tried to work
2085 around the worst of the problems. However, those workarounds may be causing
2086 linker crashes in some circumstances; the workarounds also probably prevent
2087 shared libraries from working. Use the GNU assembler to avoid these problems.
2090 The configuration scripts for GCC will also trigger a bug in the hpux9
2091 shell. To avoid this problem set @env{CONFIG_SHELL} to @file{/bin/ksh}
2092 and @env{SHELL} to @file{/bin/ksh} in your environment.
2098 @heading @anchor{hppa*-hp-hpux10}hppa*-hp-hpux10
2100 For hpux10.20, we @emph{highly} recommend you pick up the latest sed patch
2101 @code{PHCO_19798} from HP@. HP has two sites which provide patches free of
2107 <a href="http://us-support.external.hp.com">US, Canada, Asia-Pacific, and
2111 @uref{http://us-support.external.hp.com,,}US, Canada, Asia-Pacific, and
2115 @uref{http://europe-support.external.hp.com,,Europe}
2118 The HP assembler on these systems is much better than the hpux9 assembler,
2119 but still has some problems. Most notably the assembler inserts timestamps
2120 into each object file it creates, causing the 3-stage comparison test to fail
2121 during a @samp{make bootstrap}. You should be able to continue by
2122 saying @samp{make all} after getting the failure from @samp{make
2129 @heading @anchor{hppa*-hp-hpux11}hppa*-hp-hpux11
2131 GCC 3.0 and up support HP-UX 11. On 64-bit capable systems, there
2132 are two distinct ports. The @samp{hppa2.0w-hp-hpux11*} port generates
2133 code for the 32-bit pa-risc runtime architecture. It uses the HP
2134 linker and is currently the default selected by config.guess. The
2135 optional @samp{hppa64-hp-hpux11*} port generates 64-bit code for the
2136 pa-risc 2.0 architecture. It must be explicitly selected using the
2137 @samp{--host=hppa64-hp-hpux11*} configure option. Different prefixes
2138 must be used if both ports are to be installed on the same system.
2140 You must use GNU binutils 2.11 or above with the 32-bit port. Thread
2141 support is not currently implemented, so @option{--enable-threads} does
2145 @item @uref{http://gcc.gnu.org/ml/gcc-prs/2002-01/msg00551.html}
2146 @item @uref{http://gcc.gnu.org/ml/gcc-bugs/2002-01/msg00663.html}
2149 GCC 2.95.x is not supported under HP-UX 11 and cannot be used to
2150 compile GCC 3.0 and up. Refer to @uref{binaries.html,,binaries} for
2151 information about obtaining precompiled GCC binaries for HP-UX.
2153 GNU binutils 2.13 or later is recommended with the 64-bit port.
2154 The HP assembler is not supported. It is @emph{highly} recommended
2155 that the GNU linker be used as well. Either binutils must be built
2156 prior to gcc, or a binary distribution of gcc or binutils must be
2157 obtained for the initial builds. When starting with a HP compiler,
2158 it is preferable to use the ANSI compiler as the bundled compiler
2159 only supports traditional C. Bootstrapping with the bundled compiler
2160 is tested infrequently and problems often arise because of the subtle
2161 differences in semantics between traditional and ISO C. There also
2162 have been problems reported with various binary distributions. This
2163 port still is undergoing significant development.
2168 @heading @anchor{i370-*-*}i370-*-*
2169 This port is very preliminary and has many known bugs. We hope to
2170 have a higher-quality port for this machine soon.
2175 @heading @anchor{*-*-linux-gnu}*-*-linux-gnu
2177 If you use glibc 2.2 (or 2.1.9x), GCC 2.95.2 won't install
2178 out-of-the-box. You'll get compile errors while building @samp{libstdc++}.
2179 The patch @uref{glibc-2.2.patch,,glibc-2.2.patch}, that is to be
2180 applied in the GCC source tree, fixes the compatibility problems.
2189 Currently Glibc 2.2.3 (and older releases) and GCC 3.0 are out of sync
2190 since the latest exception handling changes for GCC@. Compiling glibc
2191 with GCC 3.0 will give a binary incompatible glibc and therefore cause
2192 lots of problems and might make your system completly unusable. This
2193 will definitly need fixes in glibc but might also need fixes in GCC@. We
2194 strongly advise to wait for glibc 2.2.4 and to read the release notes of
2195 glibc 2.2.4 whether patches for GCC 3.0 are needed. You can use glibc
2196 2.2.3 with GCC 3.0, just do not try to recompile it.
2201 @heading @anchor{ix86-*-linux*aout}i?86-*-linux*aout
2202 Use this configuration to generate @file{a.out} binaries on Linux-based
2203 GNU systems. This configuration is being superseded. You must use
2204 gas/binutils version 2.5.2 or later.
2209 @heading @anchor{ix86-*-linux*}i?86-*-linux*
2211 You will need binutils 2.9.1.0.15 or newer for exception handling to work.
2213 If you receive Signal 11 errors when building on GNU/Linux, then it is
2214 possible you have a hardware problem. Further information on this can be
2215 found on @uref{http://www.bitwizard.nl/sig11/,,www.bitwizard.nl}.
2220 @heading @anchor{ix86-*-sco}i?86-*-sco
2221 Compilation with RCC is recommended. Also, it may be a good idea to
2222 link with GNU malloc instead of the malloc that comes with the system.
2227 @heading @anchor{ix86-*-sco3.2v4}i?86-*-sco3.2v4
2228 Use this configuration for SCO release 3.2 version 4.
2233 @heading @anchor{ix86-*-sco3.2v5*}i?86-*-sco3.2v5*
2234 Use this for the SCO OpenServer Release 5 family of operating systems.
2236 Unlike earlier versions of GCC, the ability to generate COFF with this
2237 target is no longer provided.
2239 Earlier versions of GCC emitted DWARF 1 when generating ELF to allow
2240 the system debugger to be used. That support was too burdensome to
2241 maintain. GCC now emits only DWARF 2 for this target. This means you
2242 may use either the UDK debugger or GDB to debug programs built by this
2245 Use of the @option{-march=pentiumpro} flag can result in
2246 unrecognized opcodes when using the native assembler on OS versions before
2247 5.0.6. (Support for P6 opcodes was added to the native ELF assembler in
2248 that version.) While it's rather rare to see these emitted by GCC yet,
2249 errors of the basic form:
2252 /usr/tmp/ccaNlqBc.s:22:unknown instruction: fcomip
2253 /usr/tmp/ccaNlqBc.s:50:unknown instruction: fucomip
2256 are symptoms of this problem. You may work around this by not
2257 building affected files with that flag, by using the GNU assembler, or
2258 by using the assembler provided with the current version of the OS@.
2259 Users of GNU assembler should see the note below for hazards on doing
2262 The native SCO assembler that is provided with the OS at no
2263 charge is normally required. If, however, you must be able to use
2264 the GNU assembler (perhaps you're compiling code with asms that
2265 require GAS syntax) you may configure this package using the flags
2266 @uref{./configure.html#with-gnu-as,,@option{--with-gnu-as}}. You must
2267 use a recent version of GNU binutils; versions past 2.9.1 seem to work
2270 In general, the @option{--with-gnu-as} option isn't as well tested
2271 as the native assembler.
2273 Look in @file{gcc/config/i386/sco5.h} (search for ``messy'') for
2274 additional OpenServer-specific flags.
2276 Systems based on OpenServer before 5.0.4 (@samp{uname -X}
2277 will tell you what you're running) require TLS597 from
2278 @uref{ftp://ftp.sco.com/TLS/,,ftp://ftp.sco.com/TLS/}
2279 for C++ constructors and destructors to work right.
2281 The system linker in (at least) 5.0.4 and 5.0.5 will sometimes
2282 do the wrong thing for a construct that GCC will emit for PIC
2283 code. This can be seen as execution testsuite failures when using
2284 @option{-fPIC} on @file{921215-1.c}, @file{931002-1.c}, @file{nestfunc-1.c}, and @file{gcov-1.c}.
2285 For 5.0.5, an updated linker that will cure this problem is
2286 available. You must install both
2287 @uref{ftp://ftp.sco.com/Supplements/rs505a/,,ftp://ftp.sco.com/Supplements/rs505a/}
2288 and @uref{ftp://ftp.sco.com/SLS/,,OSS499A}.
2290 The dynamic linker in OpenServer 5.0.5 (earlier versions may show
2291 the same problem) aborts on certain G77-compiled programs. It's particularly
2292 likely to be triggered by building Fortran code with the @option{-fPIC} flag.
2293 Although it's conceivable that the error could be triggered by other
2294 code, only G77-compiled code has been observed to cause this abort.
2295 If you are getting core dumps immediately upon execution of your
2296 G77 program---and especially if it's compiled with @option{-fPIC}---try applying
2297 @uref{sco_osr5_g77.patch,,@file{sco_osr5_g77.patch}} to your @samp{libf2c} and
2299 Affected faults, when analyzed in a debugger, will show a stack
2300 backtrace with a fault occurring in @code{rtld()} and the program
2301 running as @file{/usr/lib/ld.so.1}. This problem has been reported to SCO
2302 engineering and will hopefully be addressed in later releases.
2308 @heading @anchor{ix86-*-udk}i?86-*-udk
2310 This target emulates the SCO Universal Development Kit and requires that
2311 package be installed. (If it is installed, you will have a
2312 @file{/udk/usr/ccs/bin/cc} file present.) It's very much like the
2313 @samp{i?86-*-unixware7*} target
2314 but is meant to be used when hosting on a system where UDK isn't the
2315 default compiler such as OpenServer 5 or Unixware 2. This target will
2316 generate binaries that will run on OpenServer, Unixware 2, or Unixware 7,
2317 with the same warnings and caveats as the SCO UDK@.
2319 This target is a little tricky to build because we have to distinguish
2320 it from the native tools (so it gets headers, startups, and libraries
2321 from the right place) while making the tools not think we're actually
2322 building a cross compiler. The easiest way to do this is with a configure
2326 CC=/udk/usr/ccs/bin/cc @var{/your/path/to}/gcc/configure \
2327 --host=i686-pc-udk --target=i686-pc-udk --program-prefix=udk-
2330 @emph{You should substitute @samp{i686} in the above command with the appropriate
2331 processor for your host.}
2333 After the usual @samp{make bootstrap} and
2334 @samp{make install}, you can then access the UDK-targeted GCC
2335 tools by adding @command{udk-} before the commonly known name. For
2336 example, to invoke the C compiler, you would use @command{udk-gcc}.
2337 They will coexist peacefully with any native-target GCC tools you may
2344 @heading @anchor{ia64-*-linux}ia64-*-linux
2345 IA-64 processor (also known as IPF, or Itanium Processor Family)
2348 The toolchain is not completely finished, so requirements will continue
2350 GCC 3.0.1 and later require glibc 2.2.4.
2351 GCC 3.0.2 requires binutils from 2001-09-05 or later.
2352 GCC 3.0.1 requires binutils 2.11.1 or later.
2354 None of the following versions of GCC has an ABI that is compatible
2355 with any of the other versions in this list, with the exception that
2356 Red Hat 2.96 and Trillian 000171 are compatible with each other:
2357 3.0.2, 3.0.1, 3.0, Red Hat 2.96, and Trillian 000717.
2358 This primarily affects C++ programs and programs that create shared libraries.
2359 Because of these ABI incompatibilities, GCC 3.0.2 is not recommended for
2360 user programs on GNU/Linux systems built using earlier compiler releases.
2361 GCC 3.0.2 is recommended for compiling linux, the kernel.
2362 GCC 3.0.2 is believed to be fully ABI compliant, and hence no more major
2363 ABI changes are expected.
2368 @heading @anchor{*-lynx-lynxos}*-lynx-lynxos
2369 LynxOS 2.2 and earlier comes with GCC 1.x already installed as
2370 @file{/bin/gcc}. You should compile with this instead of @file{/bin/cc}.
2371 You can tell GCC to use the GNU assembler and linker, by specifying
2372 @samp{--with-gnu-as --with-gnu-ld} when configuring. These will produce
2373 COFF format object files and executables; otherwise GCC will use the
2374 installed tools, which produce @file{a.out} format executables.
2378 <!-- rs6000-ibm-aix*, powerpc-ibm-aix* -->
2380 @heading @anchor{*-ibm-aix*}*-ibm-aix*
2382 AIX Make frequently has problems with GCC makefiles. GNU Make 3.76 or
2383 newer is recommended to build on this platform.
2385 Errors involving @code{alloca} when building GCC generally are due
2386 to an incorrect definition of @code{CC} in the Makefile or mixing files
2387 compiled with the native C compiler and GCC@. During the stage1 phase of
2388 the build, the native AIX compiler @strong{must} be invoked as @command{cc}
2389 (not @command{xlc}). Once @command{configure} has been informed of
2390 @command{xlc}, one needs to use @samp{make distclean} to remove the
2391 configure cache files and ensure that @env{CC} environment variable
2392 does not provide a definition that will confuse @command{configure}.
2393 If this error occurs during stage2 or later, then the problem most likely
2394 is the version of Make (see above).
2396 The GNU Assembler incorrectly reports that it supports WEAK symbols on
2397 AIX which causes GCC to try to utilize weak symbol functionality although
2398 it is not supported on the platform. The native @command{as} and
2399 @command{ld} still are recommended. The native AIX tools do
2400 interoperate with GCC@.
2402 Building @file{libstdc++.a} requires a fix for an AIX Assembler bug
2403 APAR IY26685 (AIX 4.3) or APAR IY25528 (AIX 5.1).
2405 @samp{libstdc++} in GCC 3.2 increments the major version number of the
2406 shared object and GCC installation places the @file{libstdc++.a}
2407 shared library in a common location which will overwrite the GCC 3.1
2408 version of the shared library. Applications either need to be
2409 re-linked against the new shared library or the GCC 3.1 version of the
2410 @samp{libstdc++} shared object needs to be available to the AIX
2411 runtime loader. The GCC 3.1 @samp{libstdc++.so.4} shared object can
2412 be installed for runtime dynamic loading using the following steps to
2413 set the @samp{F_LOADONLY} flag in the shared object for @emph{each}
2414 multilib @file{libstdc++.a} installed:
2416 Extract the shared object from each the GCC 3.1 @file{libstdc++.a}
2419 % ar -x libstdc++.a libstdc++.so.4
2422 Enable the @samp{F_LOADONLY} flag so that the shared object will be
2423 available for runtime dynamic loading, but not linking:
2425 % strip -e libstdc++.so.4
2428 Archive the runtime-only shared object in the GCC 3.2
2429 @file{libstdc++.a} archive:
2431 % ar -q libstdc++.a libstdc++.so.4
2434 Linking executables and shared libraries may produce warnings of
2435 duplicate symbols. The assembly files generated by GCC for AIX always
2436 have included multiple symbol definitions for certain global variable
2437 and function declarations in the original program. The warnings should
2438 not prevent the linker from producing a correct library or runnable
2441 AIX 4.3 utilizes a ``large format'' archive to support both 32-bit and
2442 64-bit object modules. The routines provided in AIX 4.3.0 and AIX 4.3.1
2443 to parse archive libraries did not handle the new format correctly.
2444 These routines are used by GCC and result in error messages during
2445 linking such as ``not a COFF file''. The version of the routines shipped
2446 with AIX 4.3.1 should work for a 32-bit environment. The @option{-g}
2447 option of the archive command may be used to create archives of 32-bit
2448 objects using the original ``small format''. A correct version of the
2449 routines is shipped with AIX 4.3.2 and above.
2451 Some versions of the AIX binder (linker) can fail with a relocation
2452 overflow severe error when the @option{-bbigtoc} option is used to link
2453 GCC-produced object files into an executable that overflows the TOC@. A fix
2454 for APAR IX75823 (OVERFLOW DURING LINK WHEN USING GCC AND -BBIGTOC) is
2455 available from IBM Customer Support and from its
2456 @uref{http://techsupport.services.ibm.com/,,techsupport.services.ibm.com}
2457 website as PTF U455193.
2459 The AIX 4.3.2.1 linker (bos.rte.bind_cmds Level 4.3.2.1) will dump core
2460 with a segmentation fault when invoked by any version of GCC@. A fix for
2461 APAR IX87327 is available from IBM Customer Support and from its
2462 @uref{http://techsupport.services.ibm.com/,,techsupport.services.ibm.com}
2463 website as PTF U461879. This fix is incorporated in AIX 4.3.3 and above.
2465 The initial assembler shipped with AIX 4.3.0 generates incorrect object
2466 files. A fix for APAR IX74254 (64BIT DISASSEMBLED OUTPUT FROM COMPILER FAILS
2467 TO ASSEMBLE/BIND) is available from IBM Customer Support and from its
2468 @uref{http://techsupport.services.ibm.com/,,techsupport.services.ibm.com}
2469 website as PTF U453956. This fix is incorporated in AIX 4.3.1 and above.
2471 AIX provides National Language Support (NLS)@. Compilers and assemblers
2472 use NLS to support locale-specific representations of various data
2473 formats including floating-point numbers (e.g., @samp{.} vs @samp{,} for
2474 separating decimal fractions). There have been problems reported where
2475 GCC does not produce the same floating-point formats that the assembler
2476 expects. If one encounters this problem, set the @env{LANG}
2477 environment variable to @samp{C} or @samp{En_US}.
2479 By default, GCC for AIX 4.1 and above produces code that can be used on
2480 both Power or PowerPC processors.
2482 A default can be specified with the @option{-mcpu=@var{cpu_type}}
2483 switch and using the configure option @option{--with-cpu-@var{cpu_type}}.
2488 @heading @anchor{ip2k-*-elf}ip2k-*-elf
2489 Ubicom IP2022 micro controller.
2490 This configuration is intended for embedded systems.
2491 There are no standard Unix configurations.
2493 Use @samp{configure --target=ip2k-elf --enable-languages=c} to configure GCC@.
2498 @heading @anchor{m32r-*-elf}m32r-*-elf
2499 Mitsubishi M32R processor.
2500 This configuration is intended for embedded systems.
2505 @heading @anchor{m68000-hp-bsd}m68000-hp-bsd
2506 HP 9000 series 200 running BSD@. Note that the C compiler that comes
2507 with this system cannot compile GCC; contact @email{law@@cygnus.com}
2508 to get binaries of GCC for bootstrapping.
2513 @heading @anchor{m6811-elf}m6811-elf
2514 Motorola 68HC11 family micro controllers. These are used in embedded
2515 applications. There are no standard Unix configurations.
2520 @heading @anchor{m6812-elf}m6812-elf
2521 Motorola 68HC12 family micro controllers. These are used in embedded
2522 applications. There are no standard Unix configurations.
2527 @heading @anchor{m68k-att-sysv}m68k-att-sysv
2528 AT&T 3b1, a.k.a.@: 7300 PC@. This version of GCC cannot
2529 be compiled with the system C compiler, which is too buggy.
2530 You will need to get a previous version of GCC and use it to
2531 bootstrap. Binaries are available from the OSU-CIS archive, at
2532 @uref{ftp://archive.cis.ohio-state.edu/pub/att7300/}.
2537 @heading @anchor{m68k-crds-unos}m68k-crds-unos
2538 Use @samp{configure unos} for building on Unos.
2540 The Unos assembler is named @code{casm} instead of @code{as}. For some
2541 strange reason linking @file{/bin/as} to @file{/bin/casm} changes the
2542 behavior, and does not work. So, when installing GCC, you should
2543 install the following script as @file{as} in the subdirectory where
2544 the passes of GCC are installed:
2551 The default Unos library is named @file{libunos.a} instead of
2552 @file{libc.a}. To allow GCC to function, either change all
2553 references to @option{-lc} in @file{gcc.c} to @option{-lunos} or link
2554 @file{/lib/libc.a} to @file{/lib/libunos.a}.
2556 @cindex @code{alloca}, for Unos
2557 When compiling GCC with the standard compiler, to overcome bugs in
2558 the support of @code{alloca}, do not use @option{-O} when making stage 2.
2559 Then use the stage 2 compiler with @option{-O} to make the stage 3
2560 compiler. This compiler will have the same characteristics as the usual
2561 stage 2 compiler on other systems. Use it to make a stage 4 compiler
2562 and compare that with stage 3 to verify proper compilation.
2564 (Perhaps simply defining @code{ALLOCA} in @file{x-crds} as described in
2565 the comments there will make the above paragraph superfluous. Please
2566 inform us of whether this works.)
2568 Unos uses memory segmentation instead of demand paging, so you will need
2569 a lot of memory. 5 Mb is barely enough if no other tasks are running.
2570 If linking @file{cc1} fails, try putting the object files into a library
2571 and linking from that library.
2576 @heading @anchor{m68k-hp-hpux}m68k-hp-hpux
2577 HP 9000 series 300 or 400 running HP-UX@. HP-UX version 8.0 has a bug in
2578 the assembler that prevents compilation of GCC@. This
2579 bug manifests itself during the first stage of compilation, while
2580 building @file{libgcc2.a}:
2584 cc1: warning: `-g' option not supported on this version of GCC
2585 cc1: warning: `-g1' option not supported on this version of GCC
2586 ./xgcc: Internal compiler error: program as got fatal signal 11
2589 A patched version of the assembler is available as the file
2590 @uref{ftp://altdorf.ai.mit.edu/archive/cph/hpux-8.0-assembler}. If you
2591 have HP software support, the patch can also be obtained directly from
2592 HP, as described in the following note:
2595 This is the patched assembler, to patch SR#1653-010439, where the
2596 assembler aborts on floating point constants.
2598 The bug is not really in the assembler, but in the shared library
2599 version of the function ``cvtnum(3c)''. The bug on ``cvtnum(3c)'' is
2600 SR#4701-078451. Anyway, the attached assembler uses the archive
2601 library version of ``cvtnum(3c)'' and thus does not exhibit the bug.
2604 This patch is also known as PHCO_4484.
2606 In addition, if you wish to use gas, you must use
2607 gas version 2.1 or later, and you must use the GNU linker version 2.1 or
2608 later. Earlier versions of gas relied upon a program which converted the
2609 gas output into the native HP-UX format, but that program has not been
2610 kept up to date. gdb does not understand that native HP-UX format, so
2611 you must use gas if you wish to use gdb.
2613 On HP-UX version 8.05, but not on 8.07 or more recent versions, the
2614 @code{fixproto} shell script triggers a bug in the system shell. If you
2615 encounter this problem, upgrade your operating system or use BASH (the
2616 GNU shell) to run @code{fixproto}. This bug will cause the fixproto
2617 program to report an error of the form:
2620 ./fixproto: sh internal 1K buffer overflow
2623 To fix this, you can also change the first line of the fixproto script
2634 @heading @anchor{m68k-ncr-*}m68k-ncr-*
2635 On the Tower models 4@var{n}0 and 6@var{n}0, by default a process is not
2636 allowed to have more than one megabyte of memory. GCC cannot compile
2637 itself (or many other programs) with @option{-O} in that much memory.
2639 To solve this problem, reconfigure the kernel adding the following line
2640 to the configuration file:
2650 @heading @anchor{m68k-sun}m68k-sun
2651 Sun 3. We do not provide a configuration file to use the Sun FPA by
2652 default, because programs that establish signal handlers for floating
2653 point traps inherently cannot work with the FPA@.
2658 @heading @anchor{m68k-sun-sunos4.1.1}m68k-sun-sunos4.1.1
2660 It is reported that you may need the GNU assembler on this platform.
2666 @heading @anchor{mips-*-*}mips-*-*
2667 If on a MIPS system you get an error message saying ``does not have gp
2668 sections for all it's [sic] sectons [sic]'', don't worry about it. This
2669 happens whenever you use GAS with the MIPS linker, but there is not
2670 really anything wrong, and it is okay to use the output file. You can
2671 stop such warnings by installing the GNU linker.
2673 It would be nice to extend GAS to produce the gp tables, but they are
2674 optional, and there should not be a warning about their absence.
2676 @heading @anchor{mips-sgi-irix5}mips-sgi-irix5
2678 This configuration has considerable problems, which will be fixed in a
2681 In order to compile GCC on an SGI running IRIX 5, the ``compiler_dev.hdr''
2682 subsystem must be installed from the IDO CD-ROM supplied by Silicon
2683 Graphics. It is also available for download from
2684 @uref{http://www.sgi.com/developers/devtools/apis/ido.html,,http://www.sgi.com/developers/devtools/apis/ido.html}.
2686 @code{make compare} may fail on version 5 of IRIX unless you add
2687 @option{-save-temps} to @code{CFLAGS}. On these systems, the name of the
2688 assembler input file is stored in the object file, and that makes
2689 comparison fail if it differs between the @code{stage1} and
2690 @code{stage2} compilations. The option @option{-save-temps} forces a
2691 fixed name to be used for the assembler input file, instead of a
2692 randomly chosen name in @file{/tmp}. Do not add @option{-save-temps}
2693 unless the comparisons fail without that option. If you do you
2694 @option{-save-temps}, you will have to manually delete the @samp{.i} and
2695 @samp{.s} files after each series of compilations.
2697 If you use the MIPS C compiler to bootstrap, it may be necessary
2698 to increase its table size for switch statements with the
2699 @option{-Wf,-XNg1500} option. If you use the @option{-O2}
2700 optimization option, you also need to use @option{-Olimit 3000}.
2702 To enable debugging under IRIX 5, you must use GNU @command{as} 2.11.2
2704 and use the @option{--with-gnu-as} configure option when configuring GCC.
2705 GNU @command{as} is distributed as part of the binutils package.
2706 When using release 2.11.2, you need to apply a patch
2707 @uref{http://sources.redhat.com/ml/binutils/2001-07/msg00352.html,,http://sources.redhat.com/ml/binutils/2001-07/msg00352.html}
2708 which will be included in the next release of binutils.
2710 When building GCC, the build process loops rebuilding @command{cc1} over
2711 and over again. This happens on @samp{mips-sgi-irix5.2}, and possibly
2712 other platforms. It has been reported that this is a known bug in the
2713 @command{make} shipped with IRIX 5.2. We recommend you use GNU
2714 @command{make} instead of the vendor supplied @command{make} program;
2715 however, you may have success with @command{smake} on IRIX 5.2 if you do
2716 not have GNU @command{make} available.
2721 @heading @anchor{mips-sgi-irix6}mips-sgi-irix6
2723 If you are using IRIX @command{cc} as your bootstrap compiler, you must
2724 ensure that the N32 ABI is in use. To test this, compile a simple C
2725 file with @command{cc} and then run @command{file} on the
2726 resulting object file. The output should look like:
2729 test.o: ELF N32 MSB @dots{}
2735 test.o: ELF 32-bit MSB @dots{}
2741 test.o: ELF 64-bit MSB @dots{}
2744 then your version of @command{cc} uses the O32 or N64 ABI by default. You
2745 should set the environment variable @env{CC} to @samp{cc -n32}
2746 before configuring GCC@.
2748 If you want the resulting @command{gcc} to run on old 32-bit systems
2749 with the MIPS R4400 CPU, you need to ensure that only code for the mips3
2750 instruction set architecture (ISA) is generated. While GCC 3.x does
2751 this correctly, both GCC 2.95 and SGI's MIPSpro @command{cc} may change
2752 the ISA depending on the machine where GCC is built. Using one of them
2753 as the bootstrap compiler may result in mips4 code, which won't run at
2754 all on mips3-only systems. For the test program above, you should see:
2757 test.o: ELF N32 MSB mips-3 @dots{}
2763 test.o: ELF N32 MSB mips-4 @dots{}
2766 instead, you should set the environment variable @env{CC} to @samp{cc
2767 -n32 -mips3} or @samp{gcc -mips3} respectively before configuring GCC@.
2769 GCC on IRIX 6 is usually built to support both the N32 and N64 ABIs. If
2770 you build GCC on a system that doesn't have the N64 libraries installed,
2771 you need to configure with @option{--disable-multilib} so GCC doesn't
2772 try to use them. Look for @file{/usr/lib64/libc.so.1} to see if you
2773 have the 64-bit libraries installed.
2775 You must @emph{not} use GNU @command{as} (which isn't built anyway as of
2776 binutils 2.11.2) on IRIX 6 platforms; doing so will only cause problems.
2778 GCC does not currently support generating O32 ABI binaries in the
2779 @samp{mips-sgi-irix6} configurations. It is possible to create a GCC
2780 with O32 ABI only support by configuring it for the @samp{mips-sgi-irix5}
2781 target and using a patched GNU @command{as} 2.11.2 as documented in the
2782 @uref{#mips-sgi-irix5,,@samp{mips-sgi-irix5}} section above. Using the
2783 native assembler requires patches to GCC which will be included in a
2784 future release. It is
2785 expected that O32 ABI support will be available again in a future release.
2787 The @option{--enable-threads} option doesn't currently work, a patch is
2788 in preparation for a future release. The @option{--enable-libgcj}
2789 option is disabled by default: IRIX 6 uses a very low default limit
2790 (20480) for the command line length. Although libtool contains a
2791 workaround for this problem, at least the N64 @samp{libgcj} is known not
2792 to build despite this, running into an internal error of the native
2793 @command{ld}. A sure fix is to increase this limit (@samp{ncargs}) to
2794 its maximum of 262144 bytes. If you have root access, you can use the
2795 @command{systune} command to do this.
2797 GCC does not correctly pass/return structures which are
2798 smaller than 16 bytes and which are not 8 bytes. The problem is very
2799 involved and difficult to fix. It affects a number of other targets also,
2800 but IRIX 6 is affected the most, because it is a 64-bit target, and 4 byte
2801 structures are common. The exact problem is that structures are being padded
2802 at the wrong end, e.g.@: a 4 byte structure is loaded into the lower 4 bytes
2803 of the register when it should be loaded into the upper 4 bytes of the
2806 GCC is consistent with itself, but not consistent with the SGI C compiler
2807 (and the SGI supplied runtime libraries), so the only failures that can
2808 happen are when there are library functions that take/return such
2809 structures. There are very few such library functions. Currently this
2810 is known to affect @code{inet_ntoa}, @code{inet_lnaof},
2811 @code{inet_netof}, @code{inet_makeaddr}, and @code{semctl}. Until the
2812 bug is fixed, GCC contains workarounds for the known affected functions.
2814 See @uref{http://freeware.sgi.com/,,http://freeware.sgi.com/} for more
2815 information about using GCC on IRIX platforms.
2820 @heading @anchor{powerpc*-*-*}powerpc-*-*
2822 You can specify a default version for the @option{-mcpu=@var{cpu_type}}
2823 switch by using the configure option @option{--with-cpu-@var{cpu_type}}.
2828 @heading @anchor{powerpc-*-darwin*}powerpc-*-darwin*
2829 PowerPC running Darwin (Mac OS X kernel).
2831 GCC 3.0 does not support Darwin, but 3.1 and later releases will work.
2833 Pre-installed versions of Mac OS X may not include any developer tools,
2834 meaning that you will not be able to build GCC from source. Tool
2835 binaries are available at
2836 @uref{http://www.opensource.apple.com/projects/darwin} (free
2837 registration required).
2839 Versions of the assembler prior to ``cctools-364'' cannot handle the
2840 4-argument form of @code{rlwinm} and related mask-using instructions. Darwin
2841 1.3 (Mac OS X 10.0) uses cctools-353 for instance. To get cctools-364,
2842 check out @file{cctools} with tag @samp{Apple-364}, build it, and
2843 install the assembler as @file{usr/bin/as}. See
2844 @uref{http://www.opensource.apple.com/tools/cvs/docs.html} for details.
2846 Also, the default stack limit of 512K is too small, and a bootstrap will
2847 typically fail when self-compiling @file{expr.c}. Set the stack to 800K
2848 or more, for instance by doing @samp{limit stack 800}. It's also
2849 convenient to use the GNU preprocessor instead of Apple's during the
2850 first stage of bootstrapping; this is automatic when doing @samp{make
2851 bootstrap}, but to do it from the toplevel objdir you will need to say
2852 @samp{make CC='cc -no-cpp-precomp' bootstrap}.
2854 Note that the version of GCC shipped by Apple typically includes a
2855 number of extensions not available in a standard GCC release. These
2856 extensions are generally specific to Mac programming.
2861 @heading @anchor{powerpc-*-elf}powerpc-*-elf, powerpc-*-sysv4
2862 PowerPC system in big endian mode, running System V.4.
2867 @heading @anchor{powerpc-*-linux-gnu*}powerpc-*-linux-gnu*
2870 @uref{ftp://ftp.kernel.org/pub/linux/devel/binutils,,binutils 2.12.90.0.7}
2871 or newer for a working GCC@.
2876 @heading @anchor{powerpc-*-netbsd*}powerpc-*-netbsd*
2877 PowerPC system in big endian mode running NetBSD@. To build the
2878 documentation you will need Texinfo version 4.1 (NetBSD 1.5.1 included
2879 Texinfo version 3.12).
2884 @heading @anchor{powerpc-*-eabiaix}powerpc-*-eabiaix
2885 Embedded PowerPC system in big endian mode with @option{-mcall-aix} selected as
2891 @heading @anchor{powerpc-*-eabisim}powerpc-*-eabisim
2892 Embedded PowerPC system in big endian mode for use in running under the
2898 @heading @anchor{powerpc-*-eabi}powerpc-*-eabi
2899 Embedded PowerPC system in big endian mode.
2904 @heading @anchor{powerpcle-*-elf}powerpcle-*-elf, powerpcle-*-sysv4
2905 PowerPC system in little endian mode, running System V.4.
2910 @heading @anchor{powerpcle-*-eabisim}powerpcle-*-eabisim
2911 Embedded PowerPC system in little endian mode for use in running under
2917 @heading @anchor{powerpcle-*-eabi}powerpcle-*-eabi
2918 Embedded PowerPC system in little endian mode.
2923 @heading @anchor{powerpcle-*-winnt}powerpcle-*-winnt, powerpcle-*-pe
2924 PowerPC system in little endian mode running Windows NT@.
2929 @heading @anchor{s390-*-linux*}s390-*-linux*
2930 S/390 system running Linux for S/390@.
2935 @heading @anchor{s390x-*-linux*}s390x-*-linux*
2936 zSeries system (64-bit) running Linux for zSeries@.
2941 @c Please use Solaris 2 to refer to all release of Solaris, starting
2942 @c with 2.0 until 2.6, 7, and 8. Solaris 1 was a marketing name for
2943 @c SunOS 4 releases which we don't use to avoid confusion. Solaris
2944 @c alone is too unspecific and must be avoided.
2945 @heading @anchor{*-*-solaris2*}*-*-solaris2*
2947 Sun does not ship a C compiler with Solaris 2. To bootstrap and install
2948 GCC you first have to install a pre-built compiler, see our
2949 @uref{binaries.html,,binaries page} for details.
2951 The Solaris 2 @command{/bin/sh} will often fail to configure
2952 @file{libstdc++-v3}, @file{boehm-gc} or
2953 @file{libjava}. If you encounter this problem, set @env{CONFIG_SHELL} to
2954 @command{/bin/ksh} in your environment before running @command{configure}.
2956 Solaris 2 comes with a number of optional OS packages. Some of these
2957 packages are needed to use GCC fully, namely @code{SUNWarc},
2958 @code{SUNWbtool}, @code{SUNWesu}, @code{SUNWhea}, @code{SUNWlibm},
2959 @code{SUNWsprot}, and @code{SUNWtoo}. If you did not install all
2960 optional packages when installing Solaris 2, you will need to verify that
2961 the packages that GCC needs are installed.
2963 To check whether an optional package is installed, use
2964 the @command{pkginfo} command. To add an optional package, use the
2965 @command{pkgadd} command. For further details, see the Solaris 2
2968 Trying to use the linker and other tools in
2969 @file{/usr/ucb} to install GCC has been observed to cause trouble.
2970 For example, the linker may hang indefinitely. The fix is to remove
2971 @file{/usr/ucb} from your @env{PATH}.
2973 All releases of GNU binutils prior to 2.11.2 have known bugs on this
2974 platform. We recommend the use of GNU binutils 2.11.2 or the vendor
2975 tools (Sun @command{as}, Sun @command{ld}).
2977 Sun bug 4296832 turns up when compiling X11 headers with GCC 2.95 or
2978 newer: @command{g++} will complain that types are missing. These headers assume
2979 that omitting the type means @code{int}; this assumption worked for C89 but
2980 is wrong for C++, and is now wrong for C99 also.
2982 @command{g++} accepts such (invalid) constructs with the option
2983 @option{-fpermissive}; it
2984 will assume that any missing type is @code{int} (as defined by C89).
2986 There are patches for Solaris 2.6 (105633-56 or newer for SPARC,
2987 106248-42 or newer for Intel), Solaris 7 (108376-21 or newer for SPARC,
2988 108377-20 for Intel), and Solaris 8 (108652-24 or newer for SPARC,
2989 108653-22 for Intel) that fix this bug.
2994 @heading @anchor{sparc-sun-solaris2*}sparc-sun-solaris2*
2996 When GCC is configured to use binutils 2.11.2 or later the binaries
2997 produced are smaller than the ones produced using Sun's native tools;
2998 this difference is quite significant for binaries containing debugging
3001 Sun @command{as} 4.x is broken in that it cannot cope with long symbol names.
3002 A typical error message might look similar to the following:
3005 /usr/ccs/bin/as: "/var/tmp/ccMsw135.s", line 11041: error:
3006 can't compute value of an expression involving an external symbol.
3009 This is Sun bug 4237974. This is fixed with patch 108908-02 for Solaris
3010 2.6 and has been fixed in later (5.x) versions of the assembler,
3011 starting with Solaris 7.
3013 Starting with Solaris 7, the operating system is capable of executing
3014 64-bit SPARC V9 binaries. GCC 3.1 and later properly supports
3015 this; the @option{-m64} option enables 64-bit code generation.
3016 However, if all you want is code tuned for the UltraSPARC CPU, you
3017 should try the @option{-mtune=ultrasparc} option instead, which produces
3018 code that, unlike full 64-bit code, can still run on non-UltraSPARC
3021 When configuring on a Solaris 7 or later system that is running a kernel
3022 that supports only 32-bit binaries, one must configure with
3023 @option{--disable-multilib}, since we will not be able to build the
3024 64-bit target libraries.
3029 @heading @anchor{sparc-sun-solaris2.7}sparc-sun-solaris2.7
3031 Sun patch 107058-01 (1999-01-13) for Solaris 7/SPARC triggers a bug in
3032 the dynamic linker. This problem (Sun bug 4210064) affects GCC 2.8
3033 and later, including all EGCS releases. Sun formerly recommended
3034 107058-01 for all Solaris 7 users, but around 1999-09-01 it started to
3035 recommend it only for people who use Sun's compilers.
3037 Here are some workarounds to this problem:
3040 Do not install Sun patch 107058-01 until after Sun releases a
3041 complete patch for bug 4210064. This is the simplest course to take,
3042 unless you must also use Sun's C compiler. Unfortunately 107058-01
3043 is preinstalled on some new Solaris 7-based hosts, so you may have to
3047 Copy the original, unpatched Solaris 7
3048 @command{/usr/ccs/bin/as} into
3049 @command{/usr/local/lib/gcc-lib/sparc-sun-solaris2.7/3.1/as},
3050 adjusting the latter name to fit your local conventions and software
3054 Install Sun patch 106950-03 (1999-05-25) or later. Nobody with
3055 both 107058-01 and 106950-03 installed has reported the bug with GCC
3056 and Sun's dynamic linker. This last course of action is riskiest,
3057 for two reasons. First, you must install 106950 on all hosts that
3058 run code generated by GCC; it doesn't suffice to install it only on
3059 the hosts that run GCC itself. Second, Sun says that 106950-03 is
3060 only a partial fix for bug 4210064, but Sun doesn't know whether the
3061 partial fix is adequate for GCC@. Revision -08 or later should fix
3062 the bug. The current (as of 2001-09-24) revision is -14, and is included in
3063 the Solaris 7 Recommended Patch Cluster.
3071 @heading @anchor{sparc-sun-sunos4*}sparc-sun-sunos4*
3073 A bug in the SunOS 4 linker will cause it to crash when linking
3074 @option{-fPIC} compiled objects (and will therefore not allow you to build
3077 To fix this problem you can either use the most recent version of
3078 binutils or get the latest SunOS 4 linker patch (patch ID 100170-10)
3079 from Sun's patch site.
3081 Sometimes on a Sun 4 you may observe a crash in the program
3082 @command{genflags} or @command{genoutput} while building GCC. This is said to
3083 be due to a bug in @command{sh}. You can probably get around it by running
3084 @command{genflags} or @command{genoutput} manually and then retrying the
3090 @heading @anchor{sparc-unknown-linux-gnulibc1}sparc-unknown-linux-gnulibc1
3092 It has been reported that you might need
3093 @uref{ftp://ftp.yggdrasil.com/private/hjl,,binutils 2.8.1.0.23}
3094 for this platform, too.
3100 @heading @anchor{sparc-*-linux*}sparc-*-linux*
3102 GCC versions 3.0 and higher require binutils 2.11.2 and glibc 2.2.4
3103 or newer on this platform. All earlier binutils and glibc
3104 releases mishandled unaligned relocations on @code{sparc-*-*} targets.
3110 @heading @anchor{sparc64-*-*}sparc64-*-*
3112 GCC version 2.95 is not able to compile code correctly for
3113 @code{sparc64} targets. Users of the Linux kernel, at least,
3114 can use the @code{sparc32} program to start up a new shell
3115 invocation with an environment that causes @command{configure} to
3116 recognize (via @samp{uname -a}) the system as @samp{sparc-*-*} instead.
3121 @heading @anchor{sparcv9-*-solaris2*}sparcv9-*-solaris2*
3123 The following compiler flags must be specified in the configure
3124 step in order to bootstrap this target with the Sun compiler:
3127 % CC="cc -xildoff -xarch=v9" @var{srcdir}/configure [@var{options}] [@var{target}]
3130 @option{-xildoff} turns off the incremental linker, and @option{-xarch=v9}
3131 specifies the v9 architecture to the Sun linker and assembler.
3136 @heading @anchor{#*-*-sysv*}*-*-sysv*
3137 On System V release 3, you may get this error message
3141 ld fatal: failed to write symbol name @var{something}
3142 in strings table for file @var{whatever}
3145 This probably indicates that the disk is full or your ulimit won't allow
3146 the file to be as large as it needs to be.
3148 This problem can also result because the kernel parameter @code{MAXUMEM}
3149 is too small. If so, you must regenerate the kernel and make the value
3150 much larger. The default value is reported to be 1024; a value of 32768
3151 is said to work. Smaller values may also work.
3153 On System V, if you get an error like this,
3156 /usr/local/lib/bison.simple: In function `yyparse':
3157 /usr/local/lib/bison.simple:625: virtual memory exhausted
3161 that too indicates a problem with disk space, ulimit, or @code{MAXUMEM}.
3163 On a System V release 4 system, make sure @file{/usr/bin} precedes
3164 @file{/usr/ucb} in @code{PATH}. The @code{cc} command in
3165 @file{/usr/ucb} uses libraries which have bugs.
3170 @heading @anchor{vax-dec-ultrix}vax-dec-ultrix
3171 Don't try compiling with VAX C (@code{vcc}). It produces incorrect code
3172 in some cases (for example, when @code{alloca} is used).
3177 @heading @anchor{xtensa-*-elf}xtensa-*-elf
3179 This target is intended for embedded Xtensa systems using the
3180 @samp{newlib} C library. It uses ELF but does not support shared
3181 objects. Designed-defined instructions specified via the
3182 Tensilica Instruction Extension (TIE) language are only supported
3183 through inline assembly.
3185 The Xtensa configuration information must be specified prior to
3186 building GCC@. The @file{gcc/config/xtensa/xtensa-config.h} header
3187 file contains the configuration information. If you created your
3188 own Xtensa configuration with the Xtensa Processor Generator, the
3189 downloaded files include a customized copy of this header file,
3190 which you can use to replace the default header file.
3195 @heading @anchor{xtensa-*-linux*}xtensa-*-linux*
3197 This target is for Xtensa systems running GNU/Linux. It supports ELF
3198 shared objects and the GNU C library (glibc). It also generates
3199 position-independent code (PIC) regardless of whether the
3200 @option{-fpic} or @option{-fPIC} options are used. In other
3201 respects, this target is the same as the
3202 @uref{#xtensa-*-elf,,@samp{xtensa-*-elf}} target.
3207 @heading @anchor{windows}Microsoft Windows (32-bit)
3209 A port of GCC 2.95.x is included with the
3210 @uref{http://www.cygwin.com/,,Cygwin environment}.
3212 Current (as of early 2001) snapshots of GCC will build under Cygwin
3213 without modification.
3218 @heading @anchor{os2}OS/2
3220 GCC does not currently support OS/2. However, Andrew Zabolotny has been
3221 working on a generic OS/2 port with pgcc. The current code can be found
3222 at @uref{http://www.goof.com/pcg/os2/,,http://www.goof.com/pcg/os2/}.
3224 An older copy of GCC 2.8.1 is included with the EMX tools available at
3225 @uref{ftp://ftp.leo.org/pub/comp/os/os2/leo/devtools/emx+gcc/,,
3226 ftp://ftp.leo.org/pub/comp/os/os2/leo/devtools/emx+gcc/}.
3231 @heading @anchor{older}Older systems
3233 GCC contains support files for many older (1980s and early
3234 1990s) Unix variants. For the most part, support for these systems
3235 has not been deliberately removed, but it has not been maintained for
3236 several years and may suffer from bitrot.
3238 Starting with GCC 3.1, each release has a list of ``obsoleted'' systems.
3239 Support for these systems is still present in that release, but
3240 @command{configure} will fail unless the @option{--enable-obsolete}
3241 option is given. Unless a maintainer steps forward, support for these
3242 systems will be removed from the next release of GCC@.
3244 Support for old systems as hosts for GCC can cause problems if the
3245 workarounds for compiler, library and operating system bugs affect the
3246 cleanliness or maintainability of the rest of GCC@. In some cases, to
3247 bring GCC up on such a system, if still possible with current GCC, may
3248 require first installing an old version of GCC which did work on that
3249 system, and using it to compile a more recent GCC, to avoid bugs in the
3250 vendor compiler. Old releases of GCC 1 and GCC 2 are available in the
3251 @file{old-releases} directory on the @uref{../mirrors.html,,GCC mirror
3252 sites}. Header bugs may generally be avoided using
3253 @command{fixincludes}, but bugs or deficiencies in libraries and the
3254 operating system may still cause problems.
3256 Support for older systems as targets for cross-compilation is less
3257 problematic than support for them as hosts for GCC; if an enthusiast
3258 wishes to make such a target work again (including resurrecting any of
3259 the targets that never worked with GCC 2, starting from the last CVS
3260 version before they were removed), patches
3261 @uref{../contribute.html,,following the usual requirements} would be
3262 likely to be accepted, since they should not affect the support for more
3265 For some systems, old versions of GNU binutils may also be useful,
3266 and are available from @file{pub/binutils/old-releases} on
3267 @uref{http://sources.redhat.com/mirrors.html,,sources.redhat.com mirror sites}.
3269 Some of the information on specific systems above relates to
3270 such older systems, but much of the information
3271 about GCC on such systems (which may no longer be applicable to
3272 current GCC) is to be found in the GCC texinfo manual.
3277 @heading @anchor{elf_targets}all ELF targets (SVR4, Solaris 2, etc.)
3279 C++ support is significantly better on ELF targets if you use the
3280 @uref{./configure.html#with-gnu-ld,,GNU linker}; duplicate copies of
3281 inlines, vtables and template instantiations will be discarded
3290 @uref{./index.html,,Return to the GCC Installation page}
3294 @c ***Old documentation******************************************************
3296 @include install-old.texi
3302 @uref{./index.html,,Return to the GCC Installation page}
3306 @c ***GFDL********************************************************************
3314 @uref{./index.html,,Return to the GCC Installation page}
3318 @c ***************************************************************************
3319 @c Part 6 The End of the Document
3321 @comment node-name, next, previous, up
3322 @node Concept Index, , GNU Free Documentation License, Top
3326 @unnumbered Concept Index