Like most GNU software, GCC must be configured before it can be built.This document describes the recommended configuration procedurefor both native and cross targets.
We use srcdir to refer to the toplevel source directory forGCC; we use objdir to refer to the toplevel build/object directory.
If you obtained the sources by cloning the repository, srcdirmust refer to the top gcc directory, the one where theMAINTAINERS file can be found, and not its gccsubdirectory, otherwise the build will fail.
If either srcdir or objdir is located on an automounted NFSfile system, the shell’s built-in pwd
command will returntemporary pathnames. Using these can lead to various sorts of buildproblems. To avoid this issue, set the PWDCMD
environmentvariable to an automounter-aware pwd
command, e.g.,pawd
or ‘amq -w’, during the configuration and buildphases.
First, we highly recommend that GCC be built into aseparate directory from the sources which does not residewithin the source tree. This is how we generally build GCC; buildingwhere objdir is a subdirectory of srcdir should work as well;building where objdir == srcdir is unsupported.
If you have previously built GCC in the same directory for adifferent target machine, do ‘make distclean’ to delete all filesthat might be invalid. One of the files this deletes is Makefile;if ‘make distclean’ complains that Makefile does not existor issues a message like “don’t know how to make distclean” it probablymeans that the directory is already suitably clean. However, with therecommended method of building in a separate objdir, you shouldsimply use a different objdir for each target.
Second, when configuring a native system, either cc
orgcc
must be in your path or you must set CC
inyour environment before running configure. Otherwise the configurationscripts may fail.
If you will be distributing binary versions of GCC, with modificationsto the source code, you should use the options described in thissection to make clear that your version contains modifications.
Specify the host, build and target machine configurations. You do thiswhen you run the configure script.
The build machine is the system which you are using, thehost machine is the system where you want to run the resultingcompiler (normally the build machine), and the target machine isthe system for which you want the compiler to generate code.
If you are building a compiler to produce code for the machine it runson (a native compiler), you normally do not need to specify any operandsto configure; it will try to guess the type of machine you are onand use that as the build, host and target machines. So you don’t needto specify a configuration when building a native compiler unlessconfigure cannot figure out what your configuration is or guesseswrong.
In those cases, specify the build machine’s configuration namewith the --host option; the host and target will default to bethe same as the host machine.
A configuration name may be canonical or it may be more or lessabbreviated (config.sub script produces canonical versions).
A canonical configuration name has three parts, separated by dashes.It looks like this: ‘cpu-company-system’.
Use options to override several configure time options forGCC. A list of supported options follows; ‘configure--help’ may list other options, but those not listed below may notwork and should not normally be used.
Note that each --enable option has a corresponding--disable option and that each --with option has acorresponding --without option.
--prefix=dirname
Specify the toplevel installationdirectory. This is the recommended way to install the tools into a directoryother than the default. The toplevel installation directory defaults to/usr/local.
We highly recommend against dirname being the same or asubdirectory of objdir or vice versa. If specifying a directorybeneath a user’s home directory tree, some shells will not expanddirname correctly if it contains the ‘~’ metacharacter; use$HOME
instead.
The following standard autoconf
options are supported. Normally youshould not need to use these options.
--exec-prefix=dirname
Specify the toplevel installation directory for architecture-dependentfiles. The default is prefix.
--bindir=dirname
Specify the installation directory for the executables called by users(such as
gcc
andg++
). The default isexec-prefix/bin.--libdir=dirname
Specify the installation directory for object code libraries andinternal data files of GCC. The default is exec-prefix/lib.
--libexecdir=dirname
Specify the installation directory for internal executables of GCC.The default is exec-prefix/libexec.
--with-slibdir=dirname
Specify the installation directory for the shared libgcc library. Thedefault is libdir.
--datarootdir=dirname
Specify the root of the directory tree for read-only architecture-independentdata files referenced by GCC. The default is prefix/share.
--infodir=dirname
Specify the installation directory for documentation in info format.The default is datarootdir/info.
--datadir=dirname
Specify the installation directory for some architecture-independentdata files referenced by GCC. The default is datarootdir.
--docdir=dirname
Specify the installation directory for documentation files (otherthan Info) for GCC. The default is datarootdir/doc.
--htmldir=dirname
Specify the installation directory for HTML documentation files.The default is docdir.
--pdfdir=dirname
Specify the installation directory for PDF documentation files.The default is docdir.
--mandir=dirname
Specify the installation directory for manual pages. The default isdatarootdir/man. (Note that the manual pages are only extractsfrom the full GCC manuals, which are provided in Texinfo format. The manpagesare derived by an automatic conversion process from parts of the fullmanual.)
--with-gxx-include-dir=dirname
Specifythe installation directory for G++ header files. The default dependson other configuration options, and differs between cross and nativeconfigurations.
--with-specs=specs
Specify additional command line driver SPECS.This can be useful if you need to turn on a non-standard feature bydefault without modifying the compiler’s source code, for instance--with-specs=%{!fcommon:%{!fno-common:-fno-common}}.See “Spec Files” in the main manual
--program-prefix=prefix
GCC supports some transformations of the names of its programs wheninstalling them. This option prepends prefix to the names ofprograms to install in bindir (see above). For example, specifying--program-prefix=foo- would result in ‘gcc’being installed as /usr/local/bin/foo-gcc.
--program-suffix=suffix
Appends suffix to the names of programs to install in bindir(see above). For example, specifying --program-suffix=-3.1would result in ‘gcc’ being installed as/usr/local/bin/gcc-3.1.
--program-transform-name=pattern
Applies the ‘sed’ script pattern to be applied to the namesof programs to install in bindir (see above). pattern has toconsist of one or more basic ‘sed’ editing commands, separated bysemicolons. For example, if you want the ‘gcc’ program name to betransformed to the installed program /usr/local/bin/myowngcc andthe ‘g++’ program name to be transformed to/usr/local/bin/gspecial++ without changing other program names,you could use the pattern--program-transform-name='s/^gcc$/myowngcc/; s/^g++$/gspecial++/'to achieve this effect.
All three options can be combined and used together, resulting in morecomplex conversion patterns. As a basic rule, prefix (andsuffix) are prepended (appended) before further transformationscan happen with a special transformation script pattern.
As currently implemented, this option only takes effect for nativebuilds; cross compiler binaries’ names are not transformed even when atransformation is explicitly asked for by one of these options.
For native builds, some of the installed programs are also installedwith the target alias in front of their name, as in‘i686-pc-linux-gnu-gcc’. All of the above transformations happenbefore the target alias is prepended to the name—so, specifying--program-prefix=foo- and program-suffix=-3.1, theresulting binary would be installed as/usr/local/bin/i686-pc-linux-gnu-foo-gcc-3.1.
As a last shortcoming, none of the installed Ada programs aretransformed yet, which will be fixed in some time.
--with-local-prefix=dirname
Specify theinstallation directory for local include files. The default is/usr/local. Specify this option if you want the compiler tosearch directory dirname/include for locally installedheader files instead of /usr/local/include.
You should specify --with-local-prefix only if yoursite has a different convention (not /usr/local) for where to putsite-specific files.
The default value for --with-local-prefix is /usr/localregardless of the value of --prefix. Specifying--prefix has no effect on which directory GCC searches forlocal header files. This may seem counterintuitive, but actually it islogical.
The purpose of --prefix is to specify where to installGCC. The local header files in /usr/local/include—if you putany in that directory—are not part of GCC. They are part of otherprograms—perhaps many others. (GCC installs its own header files inanother directory which is based on the --prefix value.)
Both the local-prefix include directory and the GCC-prefix includedirectory are part of GCC’s “system include” directories. Although thesetwo directories are not fixed, they need to be searched in the properorder for the correct processing of the include_next directive. Thelocal-prefix include directory is searched before the GCC-prefixinclude directory. Another characteristic of system include directoriesis that pedantic warnings are turned off for headers in these directories.
Some autoconf macros add -I directory options to thecompiler command line, to ensure that directories containing installedpackages’ headers are searched. When directory is one of GCC’ssystem include directories, GCC will ignore the option so that systemdirectories continue to be processed in the correct order. Thismay result in a search order different from what was specified but thedirectory will still be searched.
GCC automatically searches for ordinary libraries usingGCC_EXEC_PREFIX
. Thus, when the same installation prefix isused for both GCC and packages, GCC will automatically search forboth headers and libraries. This provides a configuration that iseasy to use. GCC behaves in a manner similar to that when it isinstalled as a system compiler in /usr.
Sites that need to install multiple versions of GCC may not want touse the above simple configuration. It is possible to use the--program-prefix, --program-suffix and--program-transform-name options to install multiple versionsinto a single directory, but it may be simpler to use different prefixesand the --with-local-prefix option to specify the location of thesite-specific files for each version. It will then be necessary forusers to specify explicitly the location of local site libraries(e.g., with LIBRARY_PATH
).
The same value can be used for both --with-local-prefix and--prefix provided it is not /usr. This can be usedto avoid the default search of /usr/local/include.
Do not specify /usr as the --with-local-prefix!The directory you use for --with-local-prefix must notcontain any of the system’s standard header files. If it did containthem, certain programs would be miscompiled (including GNU Emacs, oncertain targets), because this would override and nullify the headerfile corrections made by the fixincludes
script.
Indications are that people who use this option use it based on mistakenideas of what it is for. People use it as if it specified where toinstall part of GCC. Perhaps they make this assumption becauseinstalling GCC creates the directory.
--with-gcc-major-version-only
Specifies that GCC should use only the major number rather thanmajor.minor.patchlevel in filesystem paths.
--with-native-system-header-dir=dirname
Specifies that dirname is the directory that contains native systemheader files, rather than /usr/include. This option is most usefulif you are creating a compiler that should be isolated from the systemas much as possible. It is most commonly used with the--with-sysroot option and will cause GCC to searchdirname inside the system root specified by that option.
--enable-shared[=package[,…]]
Build shared versions of libraries, if shared libraries are supported onthe target platform. Unlike GCC 2.95.x and earlier, shared librariesare enabled by default on all platforms that support shared libraries.
If a list of packages is given as an argument, build shared librariesonly for the listed packages. For other packages, only static librarieswill be built. Package names currently recognized in the GCC tree are‘libgcc’ (also known as ‘gcc’), ‘libstdc++’ (not‘libstdc++-v3’), ‘libffi’, ‘zlib’, ‘boehm-gc’,‘ada’, ‘libada’, ‘libgo’, ‘libobjc’, and ‘libphobos’.Note ‘libiberty’ does not support shared libraries at all.
Use --disable-shared to build only static libraries. Note that--disable-shared does not accept a list of package names asargument, only --enable-shared does.
Contrast with --enable-host-shared, which affects hostcode.
--enable-host-shared
Specify that the host code should be built into position-independentmachine code (with -fPIC), allowing it to be used within sharedlibraries, but yielding a slightly slower compiler.
This option is required when building the libgccjit.so library.
Contrast with --enable-shared, which affects targetlibraries.
--enable-host-pie
Specify that the host executables should be built intoposition-independent executables (with -fPIE and -pie),yielding a slightly slower compiler (but faster than--enable-host-shared). Position-independent executables are loadedat random addresses each time they are executed, therefore provide additionalprotection against Return Oriented Programming (ROP) attacks.
--enable-host-pie) may be used with --enable-host-shared),in which case -fPIC is used when compiling, and -pie whenlinking.
--enable-host-bind-now
Specify that the host executables should be linked with the option-Wl,-z,now, which means that the dynamic linker will resolve allsymbols when the executables are started, and that in turn allows RELRO tomark the GOT read-only, resulting in better security.
--with-gnu-as
Specify that the compiler should assume that theassembler it finds is the GNU assembler. However, this does not modifythe rules to find an assembler and will result in confusion if theassembler found is not actually the GNU assembler. (Confusion may alsoresult if the compiler finds the GNU assembler but has not beenconfigured with --with-gnu-as.) If you have more than oneassembler installed on your system, you may want to use this option inconnection with --with-as=pathname or--with-build-time-tools=pathname.
The following systems are the only ones where it makes a differencewhether you use the GNU assembler. On any other system,--with-gnu-as has no effect.
- ‘hppa1.0-any-any’
- ‘hppa1.1-any-any’
- ‘*-*-solaris2.11’
--with-as=pathname
Specify that the compiler should use the assembler pointed to bypathname, rather than the one found by the standard rules to findan assembler, which are:
- Unless GCC is being built with a cross compiler, check thelibexec/gcc/target/version directory.libexec defaults to exec-prefix/libexec;exec-prefix defaults to prefix, whichdefaults to /usr/local unless overridden by the--prefix=pathname switch described above. targetis the target system triple, such as ‘sparc-sun-solaris2.11’, andversion denotes the GCC version, such as 3.0.
- If the target system is the same that you are building on, checkoperating system specific directories.
- Check in the
PATH
for a tool whose name is prefixed by thetarget system triple. - Check in the
PATH
for a tool whose name is not prefixed by thetarget system triple, if the host and target system triple arethe same (in other words, we use a host tool if it can be used forthe target as well).
You may want to use --with-as if no assembleris installed in the directories listed above, or if you have multipleassemblers installed and want to choose one that is not found by theabove rules.
--with-gnu-ld
Same as --with-gnu-asbut for the linker.
--with-ld=pathname
Same as --with-asbut for the linker.
--with-dsymutil=pathname
Same as --with-asbut for the debug linker (only used on Darwin platforms so far).
--with-tls=dialect
Specify the default TLS dialect, for systems were there is a choice.For ARM targets, possible values for dialect are gnu
orgnu2
, which select between the original GNU dialect and the GNU TLSdescriptor-based dialect.For RISC-V targets, possible values for dialect are trad
ordesc
, which select between the traditional GNU dialect and the GNU TLSdescriptor-based dialect.
--enable-multiarch
Specify whether to enable or disable multiarch support. The default isto check for glibc start files in a multiarch location, and enable itif the files are found. The auto detection is enabled for native builds,and for cross builds configured with --with-sysroot, and without--with-native-system-header-dir.More documentation about multiarch can be found athttps://wiki.debian.org/Multiarch.
--enable-sjlj-exceptions
Force use of the setjmp
/longjmp
-based scheme for exceptions.‘configure’ ordinarily picks the correct value based on the platform.Only use this option if you are sure you need a different setting.
--enable-vtable-verify
Specify whether to enable or disable the vtable verification feature.Enabling this feature causes libstdc++ to be built with its virtual callsin verifiable mode. This means that, when linked with libvtv, everyvirtual call in libstdc++ will verify the vtable pointer through which thecall will be made before actually making the call. If not linked with libvtv,the verifier will call stub functions (in libstdc++ itself) and do nothing.If vtable verification is disabled, then libstdc++ is not built with itsvirtual calls in verifiable mode at all. However the libvtv library willstill be built (see --disable-libvtv to turn off building libvtv).--disable-vtable-verify is the default.
--disable-gcov
Specify that the run-time library used for coverage analysisand associated host tools should not be built.
--disable-multilib
Specify that multiple targetlibraries to support different target variants, callingconventions, etc. should not be built. The default is to build apredefined set of them.
Some targets provide finer-grained control over which multilibs are built(e.g., --disable-softfloat):
arm-*-*
fpu, 26bit, underscore, interwork, biendian, nofmult.
m68*-*-*
softfloat, m68881, m68000, m68020.
mips*-*-*
single-float, biendian, softfloat.
msp430-*-*
no-exceptions
powerpc*-*-*, rs6000*-*-*
aix64, pthread, softfloat, powercpu, powerpccpu, powerpcos, biendian,sysv, aix.
--with-multilib-list=list
--without-multilib-list
Specify what multilibs to build. list is a comma separated list ofvalues, possibly consisting of a single value. Currently only implementedfor aarch64*-*-*, amdgcn*-*-*, arm*-*-*, loongarch*-*-*, riscv*-*-*, sh*-*-*and x86-64-*-linux*. The accepted values and meaning for each target is givenbelow.
aarch64*-*-*
list is a comma separated list of
ilp32
, andlp64
to enable ILP32 and LP64 run-time libraries, respectively. Iflist is empty, then there will be no multilibs and only thedefault run-time library will be built. If list isdefault
or –with-multilib-list= is not specified, then thedefault set of libraries is selected based on the value of--target.amdgcn*-*-*
list is a comma separated list of ISA names (allowed values:
fiji
,gfx900
,gfx906
,gfx908
,gfx90a
,gfx90c
,gfx1030
,gfx1036
,gfx1100
,gfx1103
).It ought not include the name of the defaultISA, specified via --with-arch. If list is empty, then therewill be no multilibs and only the default run-time library will be built. Iflist isdefault
or --with-multilib-list= is notspecified, then the default set of libraries is selected.arm*-*-*
list is a comma separated list of
aprofile
andrmprofile
to build multilibs for A or R and M architectureprofiles respectively. Note that, due to some limitation of the currentmultilib framework, using the combinedaprofile,rmprofile
multilibs selects in some cases a less optimal multilib than when usingthe multilib profile for the architecture targetted. The special valuedefault
is also accepted and is equivalent to omitting theoption, i.e., only the default run-time library will be enabled.list may instead contain
@name
, to use the multilibconfiguration Makefile fragment name in gcc/config/arm inthe source tree (it is part of the corresponding sources, after all).It is recommended, but not required, that files used for this purpose tobe named starting with t-ml-, to make their intended purposeself-evident, in line with GCC conventions. Such files enable custom,user-chosen multilib lists to be configured. Whether multiple suchfiles can be used together depends on the contents of the suppliedfiles. See gcc/config/arm/t-multilib and its supplementarygcc/config/arm/t-*profile files for an example of what suchMakefile fragments might look like for this version of GCC. The macrosexpected to be defined in these fragments are not stable across GCCreleases, so make sure they define theMULTILIB
-related macrosexpected by the version of GCC you are building.See “Target Makefile Fragments” in the internals manual.The table below gives the combination of ISAs, architectures, FPUs andfloating-point ABIs for which multilibs are built for each predefinedprofile. The union of these options is considered when specifying both
aprofile
andrmprofile
.Option aprofile rmprofile ISAs -marm
and-mthumb
-mthumb
Architectures default architecture -march=armv7-a
-march=armv7ve
-march=armv8-a
default architecture -march=armv6s-m
-march=armv7-m
-march=armv7e-m
-march=armv8-m.base
-march=armv8-m.main
-march=armv7
FPUs none -mfpu=vfpv3-d16
-mfpu=neon
-mfpu=vfpv4-d16
-mfpu=neon-vfpv4
-mfpu=neon-fp-armv8
none -mfpu=vfpv3-d16
-mfpu=fpv4-sp-d16
-mfpu=fpv5-sp-d16
-mfpu=fpv5-d16
floating-point ABIs -mfloat-abi=soft
-mfloat-abi=softfp
-mfloat-abi=hard
-mfloat-abi=soft
-mfloat-abi=softfp
-mfloat-abi=hard
loongarch*-*-*
list is a comma-separated list, with each of the element starting withthe following ABI identifiers:
lp64d[/base]
lp64f[/base]
lp64d[/base]
(the/base
suffix may be omitted)to enable their respective run-time libraries.A suffix
[/arch][/option/…]
may follow immediatelyafter the ABI identifier to customize the compiler options for building thegiven set of libraries. arch denotes the architecture name recognizedby the -march=arch compiler option, which acts as a basic targetISA configuration that can be adjusted using the subsequent optionsuffixes, where each option is a compiler option without a leading dash(’-’).If no such suffix is present for a given multilib variant, theconfigured value of --with-multilib-default is appended as a defaultsuffix. If --with-multilib-default is not given, the default buildoption -march=abi-default is applied when building the variantswithout a suffix.
As a special case,
fixed
may be used in the position of arch,which means using the architecture configured with--with-arch=arch, or its default value (e.g.loongarch64
forloongarch64-*
targets).If list is empty or
default
, or if --with-multilib-listis not specified, then only the default variant of the libraries are built,where the default ABI is implied by the configured target triplet.riscv*-*-*
list is a single ABI name. The target architecture must be either
rv32gc
orrv64gc
. This will build a single multilib for thespecified architecture and ABI pair. If--with-multilib-list
is notgiven, then a default set of multilibs is selected based on the value of--target. This is usually a large set of multilibs.sh*-*-*
list is a comma separated list of CPU names. These must be of theform
sh*
orm*
(in which case they match the compiler optionfor that processor). The list should not contain any endian options -these are handled by --with-endian.If list is empty, then there will be no multilibs for extraprocessors. The multilib for the secondary endian remains enabled.
As a special case, if an entry in the list starts with a
!
(exclamation point), then it is added to the list of excluded multilibs.Entries of this sort should be compatible with ‘MULTILIB_EXCLUDES’(once the leading!
has been stripped).If --with-multilib-list is not given, then a default set ofmultilibs is selected based on the value of --target. This isusually the complete set of libraries, but some targets imply a morespecialized subset.
Example 1: to configure a compiler for SH4A only, but supporting bothendians, with little endian being the default:
--with-cpu=sh4a --with-endian=little,big --with-multilib-list=
Example 2: to configure a compiler for both SH4A and SH4AL-DSP, but withonly little endian SH4AL:
--with-cpu=sh4a --with-endian=little,big \--with-multilib-list=sh4al,!mb/m4al
x86-64-*-linux*
list is a comma separated list of
m32
,m64
andmx32
to enable 32-bit, 64-bit and x32 run-time libraries,respectively. If list is empty, then there will be no multilibsand only the default run-time library will be enabled.If --with-multilib-list is not given, then only 32-bit and64-bit run-time libraries will be enabled.
--with-multilib-default
On LoongArch targets, set the default build options for enabled multilibswithout build options appended to their corresponding--with-multilib-list items. The format of this value is[/arch][/option/…]
, where arch is anarchitecture name recognized by -march=arch compiler option,and subsequent option suffixes are compiler options minus a leadingdash (’-’).
Multiple options may appear consecutively while arch may onlyappear in the beginning or be omitted (which means -march=abi-defaultis applied when building the libraries).
--with-strict-align-lib
On LoongArch targets, build all enabled multilibs with -mstrict-align(Not enabled by default).
--with-multilib-generator=config
Specify what multilibs to build. config is a semicolon separated list ofvalues, possibly consisting of a single value. Currently only implementedfor riscv*-*-elf*. The accepted values and meanings are given below.
Every config is constructed with four components: architecture string, ABI,reuse rule with architecture string and reuse rule with sub-extension.
Example 1: Add multi-lib suppport for rv32i with ilp32.
rv32i-ilp32--
Example 2: Add multi-lib suppport for rv32i with ilp32 and rv32imafd with ilp32.
rv32i-ilp32--;rv32imafd-ilp32--
Example 3: Add multi-lib suppport for rv32i with ilp32; rv32im with ilp32 andrv32ic with ilp32 will reuse this multi-lib set.
rv32i-ilp32-rv32im-c
Example 4: Add multi-lib suppport for rv64ima with lp64; rv64imaf with lp64,rv64imac with lp64 and rv64imafc with lp64 will reuse this multi-lib set.
rv64ima-lp64--f,c,fc
--with-multilib-generator have an optional configuration argument--cmodel=val for code model, this option will expand with otherconfig options, val is a comma separated list of possible code model,currently we support medlow and medany.
Example 5: Add multi-lib suppport for rv64ima with lp64; rv64ima with lp64 andmedlow code model
rv64ima-lp64--;--cmodel=medlow
Example 6: Add multi-lib suppport for rv64ima with lp64; rv64ima with lp64 andmedlow code model; rv64ima with lp64 and medany code model
rv64ima-lp64--;--cmodel=medlow,medany
--with-endian=endians
Specify what endians to use.Currently only implemented for sh*-*-*.
endians may be one of the following:
big
Use big endian exclusively.
little
Use little endian exclusively.
big,little
Use big endian by default. Provide a multilib for little endian.
little,big
Use little endian by default. Provide a multilib for big endian.
--enable-threads
Specify that the targetsupports threads. This affects the Objective-C compiler and runtimelibrary, and exception handling for other languages like C++.On some systems, this is the default.
In general, the best (and, in many cases, the only known) threadingmodel available will be configured for use. Beware that on somesystems, GCC has not been taught what threading models are generallyavailable for the system. In this case, --enable-threads is analias for --enable-threads=single.
--disable-threads
Specify that threading support should be disabled for the system.This is an alias for --enable-threads=single.
--enable-threads=lib
Specify thatlib is the thread support library. This affects the Objective-Ccompiler and runtime library, and exception handling for other languageslike C++. The possibilities for lib are:
aix
AIX thread support.
dce
DCE thread support.
lynx
LynxOS thread support.
mipssde
MIPS SDE thread support.
no
This is an alias for ‘single’.
posix
Generic POSIX/Unix98 thread support.
rtems
RTEMS thread support.
single
Disable thread support, should work for all platforms.
tpf
TPF thread support.
vxworks
VxWorks thread support.
win32
Microsoft Win32 API thread support.
--enable-tls
Specify that the target supports TLS (Thread Local Storage). Usuallyconfigure can correctly determine if TLS is supported. In cases whereit guesses incorrectly, TLS can be explicitly enabled or disabled with--enable-tls or --disable-tls. This can happen ifthe assembler supports TLS but the C library does not, or if theassumptions made by the configure test are incorrect.
--disable-tls
Specify that the target does not support TLS.This is an alias for --enable-tls=no.
--disable-tm-clone-registry
Disable TM clone registry in libgcc. It is enabled in libgcc by default.This option helps to reduce code size for embedded targets which donot use transactional memory.
--with-cpu=cpu
--with-cpu-32=cpu
--with-cpu-64=cpu
Specify which cpu variant the compiler should generate code for by default.cpu will be used as the default value of the -mcpu= switch.This option is only supported on some targets, including ARC, ARM, i386, M68k,PowerPC, and SPARC. It is mandatory for ARC. The --with-cpu-32 and--with-cpu-64 options specify separate default CPUs for32-bit and 64-bit modes; these options are only supported for aarch64, i386,x86-64, PowerPC, and SPARC.
--with-schedule=cpu
--with-arch=cpu
--with-arch-32=cpu
--with-arch-64=cpu
--with-tune=cpu
--with-tune-32=cpu
--with-tune-64=cpu
--with-abi=abi
--with-fpu=type
--with-float=type
--with-simd=type
These configure options provide default values for the -mschedule=,-march=, -mtune=, -mabi=, and -mfpu=options and for -mhard-float or -msoft-float. As with--with-cpu, which switches will be accepted and acceptable valuesof the arguments depend on the target.
--with-mode=mode
Specify if the compiler should default to -marm or -mthumb.This option is only supported on ARM targets.
--with-stack-offset=num
This option sets the default for the -mstack-offset=num option,and will thus generally also control the setting of this option forlibraries. This option is only supported on Epiphany targets.
--with-fpmath=isa
This options sets -mfpmath=sse by default and specifies the defaultISA for floating-point arithmetics. You can select either ‘sse’ whichenables -msse2 or ‘avx’ which enables -mavx by default.This option is only supported on i386 and x86-64 targets.
--with-fp-32=mode
On MIPS targets, set the default value for the -mfp option when usingthe o32 ABI. The possibilities for mode are:
32
Use the o32 FP32 ABI extension, as with the -mfp32 command-lineoption.
xx
Use the o32 FPXX ABI extension, as with the -mfpxx command-lineoption.
64
Use the o32 FP64 ABI extension, as with the -mfp64 command-lineoption.
In the absence of this configuration option the default is to use the o32FP32 ABI extension.
--with-odd-spreg-32
On MIPS targets, set the -modd-spreg option by default when usingthe o32 ABI.
--without-odd-spreg-32
On MIPS targets, set the -mno-odd-spreg option by default when usingthe o32 ABI. This is normally used in conjunction with--with-fp-32=64 in order to target the o32 FP64A ABI extension.
--with-nan=encoding
On MIPS targets, set the default encoding convention to use for thespecial not-a-number (NaN) IEEE 754 floating-point data. Thepossibilities for encoding are:
legacy
Use the legacy encoding, as with the -mnan=legacy command-lineoption.
2008
Use the 754-2008 encoding, as with the -mnan=2008 command-lineoption.
To use this configuration option you must have an assembler versioninstalled that supports the -mnan= command-line option too.In the absence of this configuration option the default convention isthe legacy encoding, as when neither of the -mnan=2008 and-mnan=legacy command-line options has been used.
--with-divide=type
Specify how the compiler should generate code for checking fordivision by zero. This option is only supported on the MIPS target.The possibilities for type are:
traps
Division by zero checks use conditional traps (this is the default onsystems that support conditional traps).
breaks
Division by zero checks use the break instruction.
--with-compact-branches=policy
Specify how the compiler should generate branch instructions.This option is only supported on the MIPS target.The possibilities for type are:
optimal
Cause a delay slot branch to be used if one is available in thecurrent ISA and the delay slot is successfully filled. If the delay slotis not filled, a compact branch will be chosen if one is available.
never
Ensures that compact branch instructions will never be generated.
always
Ensures that a compact branch instruction will be generated if available.If a compact branch instruction is not available,a delay slot form of the branch will be used instead.This option is supported from MIPS Release 6 onwards.For pre-R6/microMIPS/MIPS16, this option is just same as never/optimal.
--with-llsc
On MIPS targets, make -mllsc the default when no-mno-llsc option is passed. This is the default forLinux-based targets, as the kernel will emulate them if the ISA doesnot provide them.
--without-llsc
On MIPS targets, make -mno-llsc the default when no-mllsc option is passed.
--with-synci
On MIPS targets, make -msynci the default when no-mno-synci option is passed.
--without-synci
On MIPS targets, make -mno-synci the default when no-msynci option is passed. This is the default.
--with-lxc1-sxc1
On MIPS targets, make -mlxc1-sxc1 the default when no-mno-lxc1-sxc1 option is passed. This is the default.
--without-lxc1-sxc1
On MIPS targets, make -mno-lxc1-sxc1 the default when no-mlxc1-sxc1 option is passed. The indexed load/storeinstructions are not directly a problem but can lead to unexpectedbehaviour when deployed in an application intended for a 32-bit addressspace but run on a 64-bit processor. The issue is seen because allknown MIPS 64-bit Linux kernels execute o32 and n32 applicationswith 64-bit addressing enabled which affects the overflow behaviourof the indexed addressing mode. GCC will assume that ordinary32-bit arithmetic overflow behaviour is the same whether performedas an addu
instruction or as part of the address calculationin lwxc1
type instructions. This assumption holds true in apure 32-bit environment and can hold true in a 64-bit environment ifthe address space is accurately set to be 32-bit for o32 and n32.
--with-madd4
On MIPS targets, make -mmadd4 the default when no-mno-madd4 option is passed. This is the default.
--without-madd4
On MIPS targets, make -mno-madd4 the default when no-mmadd4 option is passed. The madd4
instructionfamily can be problematic when targeting a combination of cores thatimplement these instructions differently. There are two known coresthat implement these as fused operations instead of unfused (whereunfused is normally expected). Disabling these instructions is theonly way to ensure compatible code is generated; this will incura performance penalty.
--with-msa
On MIPS targets, make -mmsa the default when no-mno-msa option is passed.
--without-msa
On MIPS targets, make -mno-msa the default when no-mmsa option is passed. This is the default.
--with-mips-plt
On MIPS targets, make use of copy relocations and PLTs.These features are extensions to the traditionalSVR4-based MIPS ABIs and require support from GNU binutilsand the runtime C library.
--with-stack-clash-protection-guard-size=size
On certain targets this option sets the default stack clash protection guardsize as a power of two in bytes. On AArch64 size is required to be either12 (4KB) or 16 (64KB).
--with-isa-spec=ISA-spec-string
On RISC-V targets specify the default version of the RISC-V Unprivileged(formerly User-Level) ISA specification to produce code conforming to.The possibilities for ISA-spec-string are:
2.2
Produce code conforming to version 2.2.
20190608
Produce code conforming to version 20190608.
20191213
Produce code conforming to version 20191213.
In the absence of this configuration option the default version is 20191213.
--enable-__cxa_atexit
Define if you want to use __cxa_atexit
, rather than atexit, toregister C++ destructors for local statics and global objects.This is essential for fully standards-compliant handling ofdestructors, but requires __cxa_atexit
in libc. This option iscurrently only available on systems with GNU libc. When enabled, thiswill cause -fuse-cxa-atexit to be passed by default.
--enable-gnu-indirect-function
Define if you want to enable the ifunc
attribute. This option iscurrently only available on systems with GNU libc on certain targets.
--enable-target-optspace
Specify that targetlibraries should be optimized for code space instead of code speed.This is the default for the m32r platform.
--with-cpp-install-dir=dirname
Specify that the user visible cpp
program should be installedin prefix/dirname/cpp, in addition to bindir.
--enable-comdat
Enable COMDAT group support. This is primarily used to override theautomatically detected value.
--enable-initfini-array
Force the use of sections .init_array
and .fini_array
(instead of .init
and .fini
) for constructors anddestructors. Option --disable-initfini-array has theopposite effect. If neither option is specified, the configure scriptwill try to guess whether the .init_array
and.fini_array
sections are supported and, if they are, use them.
--enable-link-mutex
When building GCC, use a mutex to avoid linking the compilers formultiple languages at the same time, to avoid thrashing on buildsystems with limited free memory. The default is not to use such a mutex.
--enable-link-serialization
When building GCC, use make dependencies to serialize linking the compilers formultiple languages, to avoid thrashing on buildsystems with limited free memory. The default is not to add suchdependencies and thus with parallel make potentially link differentcompilers concurrently. If the argument is a positive integer, allowthat number of concurrent link processes for the large binaries.
--enable-maintainer-mode
The build rules that regenerate the Autoconf and Automake output files aswell as the GCC master message catalog gcc.pot are normallydisabled. This is because it can only be rebuilt if the complete sourcetree is present. If you have changed the sources and want to rebuild thecatalog, configuring with --enable-maintainer-mode will enablethis. Note that you need a recent version of the gettext
toolsto do so.
--disable-bootstrap
For a native build, the default configuration is to performa 3-stage bootstrap of the compiler when ‘make’ is invoked,testing that GCC can compile itself correctly. If you want to disablethis process, you can configure with --disable-bootstrap.
--enable-bootstrap
In special cases, you may want to perform a 3-stage buildeven if the target and host triplets are different.This is possible when the host can run code compiled forthe target (e.g. host is i686-linux, target is i486-linux).Starting from GCC 4.2, to do this you have to configure explicitlywith --enable-bootstrap.
--enable-generated-files-in-srcdir
Neither the .c and .h files that are generated from Bison and flex nor theinfo manuals and man pages that are built from the .texi files are presentin the repository development tree. When building GCC from that development tree,or from one of our snapshots, those generated files are placed in yourbuild directory, which allows for the source to be in a readonlydirectory.
If you configure with --enable-generated-files-in-srcdir then thosegenerated files will go into the source directory. This is mainly intendedfor generating release or prerelease tarballs of the GCC sources, since itis not a requirement that the users of source releases to have flex, Bison,or makeinfo.
--enable-version-specific-runtime-libs
Specifythat runtime libraries should be installed in the compiler specificsubdirectory (libdir/gcc) rather than the usual places. Inaddition, ‘libstdc++’’s include files will be installed intolibdir unless you overruled it by using--with-gxx-include-dir=dirname. Using this option isparticularly useful if you intend to use several versions of GCC inparallel. The default is ‘yes’ for ‘libada’, and ‘no’ forthe remaining libraries.
--with-darwin-extra-rpath
This is provided to allow distributions to add a single additionalrunpath on Darwin / macOS systems. This allows for cases where theinstalled GCC library directories are then symlinked to a commondirectory outside of the GCC installation.
--with-aix-soname=‘aix’, ‘svr4’ or ‘both’
Traditional AIX shared library versioning (versioned Shared Object
files as members of unversioned Archive Library
files named‘lib.a’) causes numerous headaches for package managers. However,Import Files
as members of Archive Library
files allow forfilename-based versioning of shared libraries as seen on Linux/SVR4,where this is called the "SONAME". But as they prevent static linking,Import Files
may be used with Runtime Linking
only, where thelinker does search for ‘libNAME.so’ before ‘libNAME.a’ libraryfilenames with the ‘-lNAME’ linker flag.
For detailed information please refer to the AIXldCommand reference.
As long as shared library creation is enabled, upon:
--with-aix-soname=aix
--with-aix-soname=both
A (traditional AIX)
Shared Archive Library
file is created:- using the ‘libNAME.a’ filename scheme
- with the
Shared Object
file as archive member named ‘libNAME.so.V’ (except for ‘libgcc_s’, where theShared Object
file is named ‘shr.o’ for backwards compatibility), which- - is used for runtime loading from inside the ‘libNAME.a’ file
- - is used for dynamic loading via
dlopen("libNAME.a(libNAME.so.V)", RTLD_MEMBER)
- - is used for shared linking
- - is used for static linking, so no separate
Static Archive Library
file is needed
--with-aix-soname=both
--with-aix-soname=svr4
A (second)
Shared Archive Library
file is created:- using the ‘libNAME.so.V’ filename scheme
- with the
Shared Object
file as archive member named ‘shr.o’, which- - is created with the
-G linker flag
- - has the
F_LOADONLY
flag set - - is used for runtime loading from inside the ‘libNAME.so.V’ file
- - is used for dynamic loading via
dlopen("libNAME.so.V(shr.o)", RTLD_MEMBER)
- - is created with the
- with the
Import File
as archive member named ‘shr.imp’, which- - refers to ‘libNAME.so.V(shr.o)’ as the "SONAME", to be recorded in the
Loader Section
of subsequent binaries - - indicates whether ‘libNAME.so.V(shr.o)’ is 32 or 64 bit
- - lists all the public symbols exported by ‘lib.so.V(shr.o)’, eventually decorated with the
‘weak’ Keyword
- - is necessary for shared linking against ‘lib.so.V(shr.o)’
- - refers to ‘libNAME.so.V(shr.o)’ as the "SONAME", to be recorded in the
A symbolic link using the ‘libNAME.so’ filename scheme is created:
- pointing to the ‘libNAME.so.V’
Shared Archive Library
file - to permit the
ld Command
to find ‘lib.so.V(shr.imp)’ via the ‘-lNAME’ argument (requiresRuntime Linking
to be enabled) - to permit dynamic loading of ‘lib.so.V(shr.o)’ without the need to specify the version number via
dlopen("libNAME.so(shr.o)", RTLD_MEMBER)
As long as static library creation is enabled, upon:
--with-aix-soname=svr4
A
Static Archive Library
is created:- using the ‘libNAME.a’ filename scheme
- with all the
Static Object
files as archive members, which- - are used for static linking
While the aix-soname=‘svr4’ option does not create Shared Object
files as members of unversioned Archive Library
files any more, packagemanagers still are responsible totransfer Shared Object
filesfound as member of a previously installed unversioned Archive Library
file into the newly installed Archive Library
file with the samefilename.
WARNING: Creating Shared Object
files with Runtime Linking
enabled may bloat the TOC, eventually leading to TOC overflow
errors,requiring the use of either the -Wl,-bbigtoc linker flag (seen tobreak with the GDB
debugger) or some of the TOC-related compiler flags,see “RS/6000 and PowerPC Options” in the main manual.
--with-aix-soname is currently supported by ‘libgcc_s’ only, sothis option is still experimental and not for normal use yet.
Default is the traditional behavior --with-aix-soname=‘aix’.
--enable-languages=lang1,lang2,…
Specify that only a particular subset of compilers andtheir runtime libraries should be built. For a list of valid values forlangN you can issue the following command in thegcc directory of your GCC source tree:
grep ^language= */config-lang.in
Currently, you can use any of the following:all
, default
, ada
, c
, c++
, d
,fortran
, go
, jit
, lto
, m2
,objc
, obj-c++
.Building the Ada compiler has special requirements, see below.If you do not pass this flag, or specify the option default
, then thedefault languages available in the gcc sub-tree will be configured.Ada, D, Go, Jit, Objective-C++ and Modula-2 are not default languages.LTO is not adefault language, but is built by default because --enable-lto isenabled by default. The other languages are default languages. Ifall
is specified, then all available languages are built. Anexception is jit
language, which requires--enable-host-shared to be included with all
.
--enable-stage1-languages=lang1,lang2,…
Specify that a particular subset of compilers and their runtimelibraries should be built with the system C compiler during stage 1 ofthe bootstrap process, rather than only in later stages with thebootstrapped C compiler. The list of valid values is the same as for--enable-languages, and the option all
will select allof the languages enabled by --enable-languages. This option isprimarily useful for GCC development; for instance, when a developmentversion of the compiler cannot bootstrap due to compiler bugs, or whenone is debugging front ends other than the C front end. When thisoption is used, one can then build the target libraries for thespecified languages with the stage-1 compiler by using makestage1-bubble all-target
, or run the testsuite on the stage-1 compilerfor the specified languages using make stage1-start check-gcc
.
--disable-libada
Specify that the run-time libraries and tools used by GNAT should notbe built. This can be useful for debugging, or for compatibility withprevious Ada build procedures, when it was required to explicitlydo a ‘make -C gcc gnatlib_and_tools’.
--disable-libgm2
Specify that the run-time libraries and tools used by Modula-2 should notbe built. This can be useful for debugging.
--disable-libsanitizer
Specify that the run-time libraries for the various sanitizers shouldnot be built.
--disable-libssp
Specify that the run-time libraries for stack smashing protectionshould not be built or linked against. On many targets library supportis provided by the C library instead.
--disable-libquadmath
Specify that the GCC quad-precision math library should not be built.On some systems, the library is required to be linkable when buildingthe Fortran front end, unless --disable-libquadmath-supportis used.
--disable-libquadmath-support
Specify that the Fortran front end and libgfortran
do not addsupport for libquadmath
on systems supporting it.
--disable-libgomp
Specify that the GNU Offloading and Multi Processing Runtime Libraryshould not be built.
--disable-libvtv
Specify that the run-time libraries used by vtable verificationshould not be built.
--with-dwarf2
Specify that the compiler shoulduse DWARF debugging information as the default; the exactDWARF version that is the default is target-specific.
--with-advance-toolchain=at
On 64-bit PowerPC Linux systems, configure the compiler to use theheader files, library files, and the dynamic linker from the AdvanceToolchain release at instead of the default versions that areprovided by the Linux distribution. In general, this option isintended for the developers of GCC, and it is not intended for generaluse.
--enable-targets=all
--enable-targets=target_list
Some GCC targets, e.g. powerpc64-linux, build bi-arch compilers.These are compilers that are able to generate either 64-bit or 32-bitcode. Typically, the corresponding 32-bit target, e.g.powerpc-linux for powerpc64-linux, only generates 32-bit code. Thisoption enables the 32-bit target to be a bi-arch compiler, which isuseful when you want a bi-arch compiler that defaults to 32-bit, andyou are building a bi-arch or multi-arch binutils in a combined tree.On mips-linux, this will build a tri-arch compiler (ABI o32/n32/64),defaulted to o32.Currently, this option only affects sparc-linux, powerpc-linux, x86-linux,mips-linux and s390-linux.
--enable-default-pie
Turn on -fPIE and -pie by default.
--enable-secureplt
This option enables -msecure-plt by default for powerpc-linux.See “RS/6000 and PowerPC Options” in the main manual
--enable-default-ssp
Turn on -fstack-protector-strong by default.
--enable-cld
This option enables -mcld by default for 32-bit x86 targets.See “i386 and x86-64 Options” in the main manual
--enable-large-address-aware
The --enable-large-address-aware option arranges for MinGWexecutables to be linked using the --large-address-awareoption, that enables the use of more than 2GB of memory. If GCC isconfigured with this option, its effects can be reversed by passing the-Wl,--disable-large-address-aware option to the so-configuredcompiler driver.
--enable-win32-registry
--enable-win32-registry=key
--disable-win32-registry
The --enable-win32-registry option enables Microsoft Windows-hosted GCCto look up installations paths in the registry using the following key:
HKEY_LOCAL_MACHINE\SOFTWARE\Free Software Foundation\key
key defaults to GCC version number, and can be overridden by the--enable-win32-registry=key option. Vendors and distributorswho use custom installers are encouraged to provide a different key,perhaps one comprised of vendor name and GCC version number, toavoid conflict with existing installations. This feature is enabledby default, and can be disabled by --disable-win32-registryoption. This option has no effect on the other hosts.
--nfp
Specify that the machine does not have a floating point unit. Thisoption only applies to ‘m68k-sun-sunosn’. On any othersystem, --nfp has no effect.
--enable-werror
--disable-werror
--enable-werror=yes
--enable-werror=no
When you specify this option, it controls whether certain files in thecompiler are built with -Werror in bootstrap stage2 and later.If you don’t specify it, -Werror is turned on for the maindevelopment trunk. However it defaults to off for release branches andfinal releases. The specific files which get -Werror arecontrolled by the Makefiles.
--enable-checking
--disable-checking
--enable-checking=list
This option controls performing internal consistency checks in the compiler.It does not change the generated code, but adds error checking of therequested complexity. This slows down the compiler and may only workproperly if you are building the compiler with GCC.
When the option is not specified, the active set of checks depends on context.Namely, bootstrap stage 1 defaults to ‘--enable-checking=yes’, buildsfrom release branches or release archives default to‘--enable-checking=release’, and otherwise‘--enable-checking=yes,extra’ is used. When the option isspecified without a list, the result is the same as‘--enable-checking=yes’. Likewise, ‘--disable-checking’ isequivalent to ‘--enable-checking=no’.
The categories of checks available in list are ‘yes’ (most commonchecks ‘assert,misc,gc,gimple,rtlflag,runtime,tree,types’), ‘no’(no checks at all), ‘all’ (all but ‘valgrind’), ‘release’(cheapest checks ‘assert,runtime’) or ‘none’ (same as ‘no’).‘release’ checks are always on and to disable them‘--disable-checking’ or ‘--enable-checking=no[,<other checks>]’must be explicitly requested. Disabling assertions makes the compiler andruntime slightly faster but increases the risk of undetected internal errorscausing wrong code to be generated.
Individual checks can be enabled with these flags: ‘assert’, ‘df’,‘extra’, ‘fold’, ‘gc’, ‘gcac’, ‘gimple’,‘misc’, ‘rtl’, ‘rtlflag’, ‘runtime’, ‘tree’,‘types’ and ‘valgrind’. ‘extra’ extends ‘misc’checking with extra checks that might affect code generation and shouldtherefore not differ between stage1 and later stages in bootstrap.
The ‘valgrind’ check requires the external valgrind
simulator,available from https://valgrind.org. The ‘rtl’ checks areexpensive and the ‘df’, ‘gcac’ and ‘valgrind’ checks are veryexpensive.
--disable-stage1-checking
--enable-stage1-checking
--enable-stage1-checking=list
This option affects only bootstrap build. If no --enable-checkingoption is specified the stage1 compiler is built with ‘yes’ checkingenabled, otherwise the stage1 checking flags are the same as specified by--enable-checking. To build the stage1 compiler withdifferent checking options use --enable-stage1-checking.The list of checking options is the same as for --enable-checking.If your system is too slow or too small to bootstrap a released compilerwith checking for stage1 enabled, you can use ‘--disable-stage1-checking’to disable checking for the stage1 compiler.
--enable-coverage
--enable-coverage=level
With this option, the compiler is built to collect self coverageinformation, every time it is run. This is for internal developmentpurposes, and only works when the compiler is being built with gcc. Thelevel argument controls whether the compiler is built optimized ornot, values are ‘opt’ and ‘noopt’. For coverage analysis youwant to disable optimization, for performance analysis you want toenable optimization. When coverage is enabled, the default level iswithout optimization.
--enable-gather-detailed-mem-stats
When this option is specified more detailed information on memoryallocation is gathered. This information is printed when using-fmem-report.
--enable-valgrind-annotations
Mark selected memory related operations in the compiler when run undervalgrind to suppress false positives.
--enable-nls
--disable-nls
The --enable-nls option enables Native Language Support (NLS),which lets GCC output diagnostics in languages other than AmericanEnglish. Native Language Support is enabled by default if not doing acanadian cross build. The --disable-nls option disables NLS.
Note that this functionality requires either libintl (provided by GNUgettext) or C standard library that contains support for gettext (suchas the GNU C Library).See –with-included-gettext for moreinformation on the conditions required to get gettext support.
--with-libintl-prefix=dir
--without-libintl-prefix
Searches for libintl in dir/include anddir/lib, or disables manual searching for it, letting thelinker handle it.
--with-libintl-type=type
Specifies the type of library to search for when looking for libintl.type can be one of auto
, static
or shared
.
--with-included-gettext
Only available if gettext is present in the source tree.
Forces the gettext tree to be configured to build and use a new staticlibintl, overriding the system libintl. Results in GCC being builtagainst the newly built libintl rather than the system libintl.
The build system makes a somewhat complicated choice when picking whereto get gettext routines from. The following table is a summary of thepossible options:
GNU gettext present in sources | libintl installed on the system | gettext present in libc | --with-included-gettext | Effects on localization |
---|---|---|---|---|
No | No | No | (ignored) | No localization |
No | No | Yes | (ignored) | Localized, libc gettext |
No | Yes | No | (ignored) | Localized, libintl |
No | Yes | Yes | (ignored) | Localized, libintl |
Yes | No | No | No | Localized, new, static libintl |
Yes | No | No | Yes | Localized, new, static libintl |
Yes | No | Yes | No | Localized, libc gettext |
Yes | No | Yes | Yes | Localized, new, static libintl |
Yes | Yes | No | No | Localized, libintl |
Yes | Yes | No | Yes | Localized, new, static libintl |
Yes | Yes | Yes | No | Localized, libintl |
Yes | Yes | Yes | Yes | Localized, new, static libintl |
--with-catgets
If NLS is enabled, and if the host lacks gettext
but has theinferior catgets
interface, the GCC build procedure normallyignores catgets
and instead uses GCC’s copy of the GNUgettext
library. The --with-catgets option causes thebuild procedure to use the host’s catgets
in this situation.
--with-libiconv-prefix=dir
Search for libiconv header files in dir/include andlibiconv library files in dir/lib.
--enable-obsolete
Enable configuration for an obsoleted system. If you attempt toconfigure GCC for a system (build, host, or target) which has beenobsoleted, and you do not specify this flag, configure will halt with anerror message.
All support for systems which have been obsoleted in one release of GCCis removed entirely in the next major release, unless someone stepsforward to maintain the port.
--enable-decimal-float
--enable-decimal-float=yes
--enable-decimal-float=no
--enable-decimal-float=bid
--enable-decimal-float=dpd
--disable-decimal-float
Enable (or disable) support for the C decimal floating point extensionthat is in the IEEE 754-2008 standard. This is enabled by defaultonly on AArch64, PowerPC, i386, and x86_64 GNU/Linux systems. Othersystems may also support it, but require the user to specificallyenable it. You can optionally control which decimal floating pointformat is used (either ‘bid’ or ‘dpd’). The ‘bid’(binary integer decimal) format is default on AArch64, i386 and x86_64systems, and the ‘dpd’ (densely packed decimal) format is defaulton PowerPC systems.
--enable-fixed-point
--disable-fixed-point
Enable (or disable) support for C fixed-point arithmetic.This option is enabled by default for some targets (such as MIPS) whichhave hardware-support for fixed-point operations. On other targets, youmay enable this option manually.
--with-long-double-128
Specify if long double
type should be 128-bit by default on selectedGNU/Linux architectures. If using --without-long-double-128
,long double
will be by default 64-bit, the same as double
type.When neither of these configure options are used, the default will be128-bit long double
when built against GNU C Library 2.4 and later,64-bit long double
otherwise.
--with-long-double-format=ibm
--with-long-double-format=ieee
Specify whether long double
uses the IBM extended double formator the IEEE 128-bit floating point format on PowerPC Linux systems.This configuration switch will only work on little endian PowerPCLinux systems and on big endian 64-bit systems where the default cpuis at least power7 (i.e. --with-cpu=power7,--with-cpu=power8, or --with-cpu=power9 is used).
If you use the --with-long-double-64 configuration option,the --with-long-double-format=ibm and--with-long-double-format=ieee options are ignored.
The default long double
format is to use IBM extended double.Until all of the libraries are converted to use IEEE 128-bit floatingpoint, it is not recommended to use--with-long-double-format=ieee.
--enable-fdpic
On SH Linux systems, generate ELF FDPIC code.
--with-gmp=pathname
--with-gmp-include=pathname
--with-gmp-lib=pathname
--with-mpfr=pathname
--with-mpfr-include=pathname
--with-mpfr-lib=pathname
--with-mpc=pathname
--with-mpc-include=pathname
--with-mpc-lib=pathname
If you want to build GCC but do not have the GMP library, the MPFRlibrary and/or the MPC library installed in a standard location anddo not have their sources present in the GCC source tree then youcan explicitly specify the directory where they are installed(‘--with-gmp=gmpinstalldir’,‘--with-mpfr=mpfrinstalldir’,‘--with-mpc=mpcinstalldir’). The--with-gmp=gmpinstalldir option is shorthand for--with-gmp-lib=gmpinstalldir/lib and--with-gmp-include=gmpinstalldir/include. Likewise the--with-mpfr=mpfrinstalldir option is shorthand for--with-mpfr-lib=mpfrinstalldir/lib and--with-mpfr-include=mpfrinstalldir/include, also the--with-mpc=mpcinstalldir option is shorthand for--with-mpc-lib=mpcinstalldir/lib and--with-mpc-include=mpcinstalldir/include. If theseshorthand assumptions are not correct, you can use the explicitinclude and lib options directly. You might also need to ensure theshared libraries can be found by the dynamic linker when building andusing GCC, for example by setting the runtime shared library pathvariable (LD_LIBRARY_PATH
on GNU/Linux and Solaris systems).
These flags are applicable to the host platform only. When buildinga cross compiler, they will not be used to configure target libraries.
--with-isl=pathname
--with-isl-include=pathname
--with-isl-lib=pathname
If you do not have the isl library installed in a standard location and youwant to build GCC, you can explicitly specify the directory where it isinstalled (‘--with-isl=islinstalldir’). The--with-isl=islinstalldir option is shorthand for--with-isl-lib=islinstalldir/lib and--with-isl-include=islinstalldir/include. If thisshorthand assumption is not correct, you can use the explicitinclude and lib options directly.
These flags are applicable to the host platform only. When buildinga cross compiler, they will not be used to configure target libraries.
--with-stage1-ldflags=flags
This option may be used to set linker flags to be used when linkingstage 1 of GCC. These are also used when linking GCC if configured with--disable-bootstrap. If --with-stage1-libs is not set to avalue, then the default is ‘-static-libstdc++ -static-libgcc’, ifsupported.
--with-stage1-libs=libs
This option may be used to set libraries to be used when linking stage 1of GCC. These are also used when linking GCC if configured with--disable-bootstrap.
--with-boot-ldflags=flags
This option may be used to set linker flags to be used when linkingstage 2 and later when bootstrapping GCC. If –with-boot-libsis not is set to a value, then the default is‘-static-libstdc++ -static-libgcc’.
--with-boot-libs=libs
This option may be used to set libraries to be used when linking stage 2and later when bootstrapping GCC.
--with-debug-prefix-map=map
Convert source directory names using -fdebug-prefix-map whenbuilding runtime libraries. ‘map’ is a space-separatedlist of maps of the form ‘old=new’.
--enable-linker-build-id
Tells GCC to pass --build-id option to the linker for all finallinks (links performed without the -r or --relocatableoption), if the linker supports it. If you specify--enable-linker-build-id, but your linker does notsupport --build-id option, a warning is issued and the--enable-linker-build-id option is ignored. The default is off.
--with-linker-hash-style=choice
Tells GCC to pass --hash-style=choice option to thelinker for all final links. choice can be one of‘sysv’, ‘gnu’, and ‘both’ where ‘sysv’ is the default.
--enable-gnu-unique-object
--disable-gnu-unique-object
Tells GCC to use the gnu_unique_object relocation for C++ templatestatic data members and inline function local statics. Enabled bydefault for a toolchain with an assembler that accepts it andGLIBC 2.11 or above, otherwise disabled.
--with-diagnostics-color=choice
Tells GCC to use choice as the default for -fdiagnostics-color=option (if not used explicitly on the command line). choicecan be one of ‘never’, ‘auto’, ‘always’, and ‘auto-if-env’where ‘auto’ is the default. ‘auto-if-env’ makes-fdiagnostics-color=auto the default if GCC_COLORS
is present and non-empty in the environment of the compiler, and-fdiagnostics-color=never otherwise.
--with-diagnostics-urls=choice
Tells GCC to use choice as the default for -fdiagnostics-urls=option (if not used explicitly on the command line). choicecan be one of ‘never’, ‘auto’, ‘always’, and ‘auto-if-env’where ‘auto’ is the default. ‘auto-if-env’ makes-fdiagnostics-urls=auto the default if GCC_URLS
or TERM_URLS
is present and non-empty in the environment of thecompiler, and -fdiagnostics-urls=never otherwise.
--enable-lto
--disable-lto
Enable support for link-time optimization (LTO). This is enabled bydefault, and may be disabled using --disable-lto.
--enable-linker-plugin-configure-flags=FLAGS
--enable-linker-plugin-flags=FLAGS
By default, linker plugins (such as the LTO plugin) are built for thehost system architecture. For the case that the linker has adifferent (but run-time compatible) architecture, these flags can bespecified to build plugins that are compatible to the linker. Forexample, if you are building GCC for a 64-bit x86_64(‘x86_64-pc-linux-gnu’) host system, but have a 32-bit x86GNU/Linux (‘i686-pc-linux-gnu’) linker executable (which isexecutable on the former system), you can configure GCC as follows forgetting compatible linker plugins:
% srcdir/configure \ --host=x86_64-pc-linux-gnu \ --enable-linker-plugin-configure-flags=--host=i686-pc-linux-gnu \ --enable-linker-plugin-flags='CC=gcc\ -m32\ -Wl,-rpath,[...]/i686-pc-linux-gnu/lib'
--with-plugin-ld=pathname
Enable an alternate linker to be used at link-time optimization (LTO)link time when -fuse-linker-plugin is enabled.This linker should have plugin support such as gold starting withversion 2.20 or GNU ld starting with version 2.21.See -fuse-linker-plugin for details.
--enable-canonical-system-headers
--disable-canonical-system-headers
Enable system header path canonicalization for libcpp. This canproduce shorter header file paths in diagnostics and dependency outputfiles, but these changed header paths may conflict with some compilationenvironments. Enabled by default, and may be disabled using--disable-canonical-system-headers.
--with-glibc-version=major.minor
Tell GCC that when the GNU C Library (glibc) is used on the target itwill be version major.minor or later. Normally this canbe detected from the C library’s header files, but this option may beneeded when bootstrapping a cross toolchain without the header filesavailable for building the initial bootstrap compiler.
If GCC is configured with some multilibs that use glibc and some thatdo not, this option applies only to the multilibs that use glibc.However, such configurations may not work well as not all the relevantconfiguration in GCC is on a per-multilib basis.
--enable-as-accelerator-for=target
Build as offload target compiler. Specify offload host triple by target.
--enable-offload-targets=target1[=path1],…,targetN[=pathN]
Enable offloading to targets target1, …, targetN.Offload compilers are expected to be already installed. Default searchpath for them is exec-prefix, but it can be changed byspecifying paths path1, …, pathN.
% srcdir/configure \ --enable-offload-targets=amdgcn-amdhsa,nvptx-none
--enable-offload-defaulted
Tell GCC that configured but not installed offload compilers and libgompplugins are silently ignored. Useful for distribution compilers wherethose are in separate optional packages and where the presence or absenceof those optional packages should determine the actual supported offloadingtarget set rather than the GCC configure-time selection.
--enable-cet
--disable-cet
Enable building target run-time libraries with control-flowinstrumentation, see -fcf-protection option. When--enable-cet
is specified target libraries are configuredto add -fcf-protection and, if needed, other targetspecific options to a set of building options.
--enable-cet=auto
is default. CET is enabled on Linux/x86 iftarget binutils supports Intel CET
instructions and disabledotherwise. In this case, the target libraries are configured to getadditional -fcf-protection option.
--with-riscv-attribute=‘yes’, ‘no’ or ‘default’
Generate RISC-V attribute by default, in order to record extra buildinformation in object.
The option is disabled by default. It is enabled on RISC-V/ELF (bare-metal)target if target binutils supported.
--enable-s390-excess-float-precision
--disable-s390-excess-float-precision
On s390(x) targets, enable treatment of float expressions with double precisionwhen in standards-compliant mode (e.g., when --std=c99
or-fexcess-precision=standard
are given).
For a native build and cross compiles that have target headers, the option’sdefault is derived from glibc’s behavior. When glibc clamps float_t to double,GCC follows and enables the option. For other cross compiles, the default isdisabled.
--with-zstd=pathname
--with-zstd-include=pathname
--with-zstd-lib=pathname
If you do not have the zstd
library installed in a standardlocation and you want to build GCC, you can explicitly specify thedirectory where it is installed (‘--with-zstd=zstdinstalldir’).The --with-zstd=zstdinstalldir option is shorthand for--with-zstd-lib=zstdinstalldir/lib and--with-zstd-include=zstdinstalldir/include. If thisshorthand assumption is not correct, you can use the explicitinclude and lib options directly.
These flags are applicable to the host platform only. When buildinga cross compiler, they will not be used to configure target libraries.
The following options only apply to building cross compilers.
Sometimes, it might be necessary to override the result of someconfigure
test, for example in order to ease porting to a newsystem or work around a bug in a test. The toplevel configure
script provides three variables for this:
In order to avoid shell and make
quoting issues for complexoverrides, you can pass a setting for CONFIG_SITE
and setvariables in the site file.
The following options apply to the build of the Objective-C runtime library.
The following options apply to the build of the D runtime library.