Please read this document carefullybefore installing theGNU Compiler Collection on your machine.
Note that this list of install notes isnot a list of supportedhosts or targets. Not all supported hosts and targets are listedhere, only the ones that require host-specific or target-specificinformation have to.
To enable a workaround for the Cortex-A53 erratum number 835769 by default(for all CPUs regardless of -mcpu option given) at configure time use the--enable-fix-cortex-a53-835769 option. This will enable the fix bydefault and can be explicitly disabled during compilation by passing the-mno-fix-cortex-a53-835769 option. Conversely,--disable-fix-cortex-a53-835769 will disable the workaround bydefault. The workaround is disabled by default if neither of--enable-fix-cortex-a53-835769 or--disable-fix-cortex-a53-835769 is given at configure time.
To enable a workaround for the Cortex-A53 erratum number 843419 by default(for all CPUs regardless of -mcpu option given) at configure time use the--enable-fix-cortex-a53-843419 option. This workaround is applied atlink time. Enabling the workaround will cause GCC to pass the relevant optionto the linker. It can be explicitly disabled during compilation by passing the-mno-fix-cortex-a53-843419 option. Conversely,--disable-fix-cortex-a53-843419 will disable the workaround by default.The workaround is disabled by default if neither of--enable-fix-cortex-a53-843419 or--disable-fix-cortex-a53-843419 is given at configure time.
To enable Branch Target Identification Mechanism and Return Address Signing bydefault at configure time use the--enable-standard-branch-protectionoption. This is equivalent to having-mbranch-protection=standardduring compilation. This can be explicitly disabled during compilation bypassing the-mbranch-protection=none option which turns off alltypes of branch protections. Conversely,--disable-standard-branch-protection will disable both theprotections by default. This mechanism is turned off by default if neitherof the options are given at configure time.
This is a synonym for ‘x86_64-*-solaris2*’.
AMD GCN GPU target.
Instead of GNU Binutils, you need to install LLVM and copybin/llvm-mc toamdgcn-amdhsa/bin/as,bin/lld toamdgcn-amdhsa/bin/ld,bin/llvm-objdump toamdgcn-amdhsa/bin/objdump,bin/llvm-nm toamdgcn-amdhsa/bin/nm, andbin/llvm-ar to bothbin/amdgcn-amdhsa-ar andbin/amdgcn-amdhsa-ranlib. The LLVM version is required for theassembler (llvm-mc) and linker (lld); however, for the others, the respectiveGNU Binutils counterpart can be used instead. While all mentioned programs arerequired when building GCC, the installed GCC compiler only needs the assemblerand linker;nm,ar, andranlib are required when installinggcc-nm,gcc-ar, andgcc-ranlib.
The required version of LLVM depends on the devices that you want to support.As the list of ISAs is long, GCC by default only builds a subset of thesupported ISAs as multilib; use--with-multilib-list= to tailor the builtmultilibs. Note that mixing ISAs in the same binary is not supported and givesa linker error.
By default, multilib support is built forgfx908,gfx90a,gfx9-generic,gfx9-4-generic,gfx10-3-generic, andgfx11-generic, which covers all supported archs. The default multilibconfiguration requires LLVM 19 or newer. LLVM 13.0.1 or LLVM 14 can be used byspecifying a--with-multilib-list= that only lists GFX9 or GFX10-3devices, while LLVM 15 is required for GFX 11 device andgfx1036.At least LLVM 16 is required forgfx1150 andgfx1151,LLVM 18 forgfx942, LLVM 19 for the genericgfx9-generic,gfx10-3-generic, andgfx11-generic targets and forgfx1152,while LLVM 20 is required forgfx950,gfx1153, andgfx9-4-generic.
The supported ISA architectures are listed in the GCC manual. The genericISA targetsgfx9-generic,gfx10-3-generic, andgfx11-generic reduce the number of required multilibs but notethatgfx9-generic does not includegfx908 orgfx90a,that linking specific ISA code with generic code is currently not supported,and that only ROCm 6.4.0 or newer is able to execute generic code.
Use Newlib (4.6.0 or newer).
To run the binaries, install the HSA Runtime from theROCm Platform, and uselibexec/gcc/amdhsa-amdhsa/version/gcn-run to launch themon the GPU.
Use ‘configure --target=arc-elf32 --with-cpu=cpu --enable-languages="c,c++"’to configure GCC, withcpu being one of ‘arc600’, ‘arc601’,or ‘arc700’.
Use ‘configure --target=arc-linux-uclibc --with-cpu=arc700 --enable-languages="c,c++"’ to configure GCC.
ARM-family processors.
Building the Ada frontend commonly fails (an infinite loop executingxsinfo) if the host compiler is GNAT 4.8. Host compilers built from theGNAT 4.6, 4.9 or 5 release branches are known to succeed.
AVR 8-bit microcontrollers. These are used in embeddedapplications. There are no standard Unix configurations.SeeAVR Optionsin the main manualfor the list of supported MCU types.
Use ‘configure --target=avr --enable-languages="c,c++"’ to configure GCC.
Further installation notes and other useful information about AVR toolscan also be obtained from:
The Blackfin processor, an Analog Devices DSP.See “Blackfin Options” in the main manual
More information, and a version of binutils with support for this processor,are available athttps://sourceforge.net/projects/adi-toolchain/.
CRIS is a CPU architecture in Axis Communications systems-on-a-chip, forexample the ETRAX series. These are used in embedded applications.
See “CRIS Options” in the main manualfor a list of CRIS-specific options.
Use ‘configure --target=cris-elf’ to configure GCC for buildinga cross-compiler for CRIS.
Please have a look at thebinaries page.
You cannot install GCC by itself on MSDOS; it will not compile underany MSDOS compiler except itself. You need to get the completecompilation package DJGPP, which includes binaries as well as sources,and includes all the necessary compilation tools and libraries.
Adapteva Epiphany.This configuration is intended for embedded systems.
FreeBSD using the ELF file format with DWARF 2 debugging.
We recommend bootstrapping against the latest GNU Binutils or theversion found in thedevel/binutils port. This also has beenknown to enable additional features and improve overall testsuiteresults.
Ada and D (or rather their respective libraries) are broken onFreeBSD/i386. This also affects building 32-bit libraries onFreeBSD/amd64, so configure with--disable-multilibthere in case you are building one of these front ends.
Go (or rather libgo) is generally broken on FreeBSD.
The FT32 processor.This configuration is intended for embedded systems.
Renesas H8/300 series of processors.
Only the 64-bit ‘hppa’ target is supported on HP-UX. Supportfor 32-bit ‘hppa’ was discontinued in GCC 13.
We require using gas on all hppa platforms.
It may be helpful to configure GCC with the--with-gnu-as and--with-as=… options to ensure that GCC can find GAS.
Only the HP linker is supported. Thus, it is best to explicitlyconfigure the target with the--with-ld=… option.
The DCE thread library is not supported, so--enable-threads=dcedoes not work.
Currently, there are no precompiled binaries for 64-bit HP-UX.
Binutils and other required tools can be built using the HP tools.Then, the GCC distribution can be built. This is challenging dueto the many dependencies.
Versions of libstdc++-v3 starting with 3.2.1 require bug fixes presentin glibc 2.2.5 and later. More information is available in thelibstdc++-v3 documentation.
If you receive Signal 11 errors when building on GNU/Linux, then it ispossible you have a hardware problem. Further information on this can befound onwww.bitwizard.nl.
Use this for Solaris 11.4 on x86 and x86-64 systems. Startingwith GCC 4.7, there is also a 64-bit ‘amd64-*-solaris2*’ or‘x86_64-*-solaris2*’ configuration that corresponds to‘sparcv9-sun-solaris2*’.
IA-64 processor (also known as IPF, or Itanium Processor Family)running GNU/Linux.
If you are using the installed system libunwind library with--with-system-libunwind, then you must use libunwind 0.98 orlater.
Building GCC on this target requires the GNU Assembler. The bundled HPassembler will not work. To prevent GCC from using the wrong assembler,the option--with-gnu-as may be necessary.
Support for AIX version 3 and older was discontinued in GCC 3.4.Support for AIX version 4.2 and older was discontinued in GCC 4.5.
“out of memory” bootstrap failures may indicate a problem withprocess resource limits (ulimit). Hard limits are configured in the/etc/security/limits system configuration file.
GCC 4.9 and above require a C++ compiler for bootstrap. IBM VAC++ / xlCcannot bootstrap GCC. xlc can bootstrap an older version of GCC andG++ can bootstrap recent releases of GCC.
GCC can bootstrap with recent versions of IBM XLC, but bootstrappingwith an earlier release of GCC is recommended. Bootstrapping with XLCrequires a larger data segment, which can be enabled through theLDR_CNTRL environment variable, e.g.,
% LDR_CNTRL=MAXDATA=0x50000000% export LDR_CNTRL
One can start with a pre-compiled version of GCC to build fromsources. One may delete GCC’s “fixed” header files when startingwith a version of GCC built for an earlier release of AIX.
To speed up the configuration phases of bootstrapping and installing GCC,one may use GNU Bash instead of AIX/bin/sh, e.g.,
% CONFIG_SHELL=/opt/freeware/bin/bash% export CONFIG_SHELL
and then proceed as described inthe buildinstructions, where we strongly recommend specifying an absolute pathto invokesrcdir/configure.
Because GCC on AIX is built as a 32-bit executable by default,(although it can generate 64-bit programs) the GMP and MPFR librariesrequired by gfortran must be 32-bit libraries. Building GMP and MPFRas static archive libraries works better than shared libraries.
Errors involvingalloca when building GCC generally are dueto an incorrect definition ofCC in the Makefile or mixing filescompiled with the native C compiler and GCC. During the stage1 phase ofthe build, the native AIX compilermust be invoked ascc(notxlc). Onceconfigure has been informed ofxlc, one needs to use ‘make distclean’ to remove theconfigure cache files and ensure thatCC environment variabledoes not provide a definition that will confuseconfigure.If this error occurs during stage2 or later, then the problem most likelyis the version of Make (see above).
The nativeas andld are recommended forbootstrapping on AIX. The GNU Assembler, GNU Linker, and GNUBinutils version 2.20 is the minimum level that supports bootstrap onAIX 5. The GNU Assembler has not been updated to support AIX 6 orAIX 7. The native AIX tools do interoperate with GCC.
AIX 7.1 added partial support for DWARF debugging, but full supportrequires AIX 7.1 TL03 SP7 that supports additional DWARF sections andfixes a bug in the assembler. AIX 7.1 TL03 SP5 distributed a versionof libm.a missing important symbols; a fix for IV77796 will beincluded in SP6.
AIX 5.3 TL10, AIX 6.1 TL05 and AIX 7.1 TL00 introduced an AIXassembler change that sometimes produces corrupt assembly filescausing AIX linker errors. The bug breaks GCC bootstrap on AIX andcan cause compilation failures with existing GCC installations. AnAIX iFix for AIX 5.3 is available (APAR IZ98385 for AIX 5.3 TL10, APARIZ98477 for AIX 5.3 TL11 and IZ98134 for AIX 5.3 TL12). AIX 5.3 TL11 SP8,AIX 5.3 TL12 SP5, AIX 6.1 TL04 SP11, AIX 6.1 TL05 SP7, AIX 6.1 TL06 SP6,AIX 6.1 TL07 and AIX 7.1 TL01 should include the fix.
Buildinglibstdc++.a requires a fix for an AIX Assembler bugAPAR IY26685 (AIX 4.3) or APAR IY25528 (AIX 5.1). It also requires afix for another AIX Assembler bug and a co-dependent AIX Archiver fixreferenced as APAR IY53606 (AIX 5.2) or as APAR IY54774 (AIX 5.1)
‘libstdc++’ in GCC 3.4 increments the major version number of theshared object and GCC installation places thelibstdc++.ashared library in a common location which will overwrite the and GCC3.3 version of the shared library. Applications either need to bere-linked against the new shared library or the GCC 3.1 and GCC 3.3versions of the ‘libstdc++’ shared object needs to be availableto the AIX runtime loader. The GCC 3.1 ‘libstdc++.so.4’, ifpresent, and GCC 3.3 ‘libstdc++.so.5’ shared objects can beinstalled for runtime dynamic loading using the following steps to setthe ‘F_LOADONLY’ flag in the shared object foreachmultiliblibstdc++.a installed:
Extract the shared objects from the currently installedlibstdc++.a archive:
% ar -x libstdc++.a libstdc++.so.4 libstdc++.so.5
Enable the ‘F_LOADONLY’ flag so that the shared object will beavailable for runtime dynamic loading, but not linking:
% strip -e libstdc++.so.4 libstdc++.so.5
Archive the runtime-only shared object in the GCC 3.4libstdc++.a archive:
% ar -q libstdc++.a libstdc++.so.4 libstdc++.so.5
Eventually, the--with-aix-soname=svr4configure option may drop the need for this procedure for libraries thatsupport it.
Linking executables and shared libraries may produce warnings ofduplicate symbols. The assembly files generated by GCC for AIX alwayshave included multiple symbol definitions for certain global variableand function declarations in the original program. The warnings shouldnot prevent the linker from producing a correct library or runnableexecutable.
AIX 4.3 utilizes a “large format” archive to support both 32-bit and64-bit object modules. The routines provided in AIX 4.3.0 and AIX 4.3.1to parse archive libraries did not handle the new format correctly.These routines are used by GCC and result in error messages duringlinking such as “not a COFF file”. The version of the routines shippedwith AIX 4.3.1 should work for a 32-bit environment. The-goption of the archive command may be used to create archives of 32-bitobjects using the original “small format”. A correct version of theroutines is shipped with AIX 4.3.2 and above.
Some versions of the AIX binder (linker) can fail with a relocationoverflow severe error when the-bbigtoc option is used to linkGCC-produced object files into an executable that overflows the TOC. A fixfor APAR IX75823 (OVERFLOW DURING LINK WHEN USING GCC AND -BBIGTOC) isavailable as PTF U455193.
The AIX 4.3.2.1 linker (bos.rte.bind_cmds Level 4.3.2.1) will dump corewith a segmentation fault when invoked by any version of GCC. A fix forAPAR IX87327 is available as PTF U461879. This fix is incorporated inAIX 4.3.3 and above.
The initial assembler shipped with AIX 4.3.0 generates incorrect objectfiles. A fix for APAR IX74254 (64BIT DISASSEMBLED OUTPUT FROM COMPILER FAILSTO ASSEMBLE/BIND) is available as PTF U453956. This fix is incorporated inAIX 4.3.1 and above.
AIX provides National Language Support (NLS). Compilers and assemblersuse NLS to support locale-specific representations of various dataformats including floating-point numbers (e.g., ‘.’ vs ‘,’ forseparating decimal fractions). There have been problems reported whereGCC does not produce the same floating-point formats that the assemblerexpects. If one encounters this problem, set theLANGenvironment variable to ‘C’ or ‘En_US’.
A default can be specified with the-mcpu=cpu_typeswitch and using the configure option--with-cpu-cpu_type.
Vitesse IQ2000 processors. These are used in embeddedapplications. There are no standard Unix configurations.
Lattice Mico32 processor.This configuration is intended for embedded systems.
Lattice Mico32 processor.This configuration is intended for embedded systems running uClinux.
LoongArch processor.The following LoongArch targets are available:
loongarch64-linux-gnu*LoongArch processor running GNU/Linux. This target triplet may be coupledwith a small set of possible suffixes to identify their default ABI type:
f64Useslp64d/base ABI by default.
f32Useslp64f/base ABI by default.
sfUseslp64s/base ABI by default.
loongarch64-linux-gnuSame asloongarch64-linux-gnuf64 for legacy support.
More information about LoongArch can be found athttps://github.com/loongson/LoongArch-Documentation.
Renesas M32C processor.This configuration is intended for embedded systems.
Renesas M32R processor.This configuration is intended for embedded systems.
By default,‘m68k-*-elf*’, ‘m68k-*-rtems’, ‘m68k-*-uclinux’ and‘m68k-*-linux’build libraries for both M680x0 and ColdFire processors. If you onlyneed the M680x0 libraries, you can omit the ColdFire ones by passing--with-arch=m68k toconfigure. Alternatively, youcan omit the M680x0 libraries by passing--with-arch=cf toconfigure. These targets default to 5206 or 5475 code asappropriate for the target system whenconfigured with--with-arch=cf and 68020 code otherwise.
The ‘m68k-*-netbsd’ and‘m68k-*-openbsd’ targets also support the--with-archoption. They will generate ColdFire CFV4e code when configured with--with-arch=cf and 68020 code otherwise.
You can override the default processors listed above by configuringwith--with-cpu=target. Thistarget can eitherbe a-mcpu argument or one of the following values:‘m68000’, ‘m68010’, ‘m68020’, ‘m68030’,‘m68040’, ‘m68060’, ‘m68020-40’ and ‘m68020-60’.
GCC requires at least binutils version 2.17 on these targets.
GCC 4.3 changed the uClinux configuration so that it uses the‘m68k-linux-gnu’ ABI rather than the ‘m68k-elf’ ABI.It also added improved support for C++ and flat shared libraries,both of which were ABI changes.
Xilinx MicroBlaze processor.This configuration is intended for embedded systems.
If on a MIPS system you get an error message saying “does not have gpsections for all it’s [sic] sectons [sic]”, don’t worry about it. Thishappens whenever you use GAS with the MIPS linker, but there is notreally anything wrong, and it is okay to use the output file. You canstop such warnings by installing the GNU linker.
It would be nice to extend GAS to produce the gp tables, but they areoptional, and there should not be a warning about their absence.
The libstdc++ atomic locking routines for MIPS targets requires MIPS IIand later. A patch went in just after the GCC 3.3 release tomake ‘mips*-*-*’ use the generic implementation instead. You can alsoconfigure for ‘mipsel-elf’ as a workaround. The‘mips*-*-linux*’ target continues to use the MIPS II routines. Morework on this is expected in future releases.
The built-in__sync_* functions are available on MIPS II andlater systems and others that support the ‘ll’, ‘sc’ and‘sync’ instructions. This can be overridden by passing--with-llsc or--without-llsc when configuring GCC.Since the Linux kernel emulates these instructions if they aremissing, the default for ‘mips*-*-linux*’ targets is--with-llsc. The--with-llsc and--without-llsc configure options may be overridden at compiletime by passing the-mllsc or-mno-llsc options tothe compiler.
MIPS systems check for division by zero (unless-mno-check-zero-division is passed to the compiler) bygenerating either a conditional trap or a break instruction. Usingtrap results in smaller code, but is only supported on MIPS II andlater. Also, some versions of the Linux kernel have a bug thatprevents trap from generating the proper signal (SIGFPE). To enablethe use of break, use the--with-divide=breaksconfigure option when configuring GCC. The default is touse traps on systems that support them.
The moxie processor.
TI MSP430 processor.This configuration is intended for embedded systems.
‘msp430-*-elf’ is the standard configuration with most GCCfeatures enabled by default.
‘msp430-*-elfbare’ is tuned for a bare-metal environment, and disablesfeatures related to shared libraries and other functionality not used forthis device. This reduces code and data usage of the GCC libraries, resultingin a minimal run-time environment by default.
Features disabled by default include:
Andes NDS32 target in little endian mode.
Andes NDS32 target in big endian mode.
Nvidia PTX target.
Instead of GNU Binutils, you will need to installnvptx-tools.Tell GCC where to find it:--with-build-time-tools=[install-nvptx-tools]/nvptx-none/bin.
You will need newlib 4.3.0 or later. It can beautomatically built together with GCC. For this, add a symbolic linkto nvptx-newlib’snewlib directory to the directory containingthe GCC sources.
Use the--disable-sjlj-exceptions and--enable-newlib-io-long-long options when configuring.
The--with-arch option may be specified to override thedefault value for the-march option, and to also buildcorresponding target libraries.The default is--with-arch=sm_52.
For example, if--with-arch=sm_70 is specified,code generation defaults to-march=sm_70 andcorresponding target libraries are built, in addition to thosemandated by--with-multilib-list, if any,see--with-multilib-list.
The OpenRISC 1000 32-bit processor with delay slots.This configuration is intended for embedded systems.
The OpenRISC 1000 32-bit processor with delay slots.
You can specify a default version for the-mcpu=cpu_typeswitch by using the configure option--with-cpu-cpu_type.
You will need GNU Binutils 2.20 or newer.
PowerPC running Darwin (Mac OS X kernel).
Pre-installed versions of Mac OS X may not include any developer tools,meaning that you will not be able to build GCC from source. Toolbinaries are available athttps://opensource.apple.com.
This version of GCC requires at least cctools-590.36. Thecctools-590.36 package referenced fromhttps://gcc.gnu.org/ml/gcc/2006-03/msg00507.html will not workon systems older than 10.3.9 (aka darwin7.9.0).
PowerPC system in big endian mode, running System V.4.
PowerPC system in big endian mode running Linux.
PowerPC system in big endian mode running NetBSD.
Embedded PowerPC system in big endian mode for use in running under thePSIM simulator.
Embedded PowerPC system in big endian mode.
PowerPC system in little endian mode, running System V.4.
Embedded PowerPC system in little endian mode for use in running underthe PSIM simulator.
Embedded PowerPC system in little endian mode.
The Renesas RL78 processor.This configuration is intended for embedded systems.
The RISC-V RV32 instruction set.This configuration is intended for embedded systems.This (and all other RISC-V) targets require the binutils 2.30 release.
The RISC-V RV32 instruction set running GNU/Linux.This (and all other RISC-V) targets require the binutils 2.30 release.
The RISC-V RV64 instruction set.This configuration is intended for embedded systems.This (and all other RISC-V) targets require the binutils 2.30 release.
The RISC-V RV64 instruction set running GNU/Linux.This (and all other RISC-V) targets require the binutils 2.30 release.
The Renesas RX processor.
S/390 system running GNU/Linux for S/390.
zSeries system (64-bit) running GNU/Linux for zSeries.
zSeries system (64-bit) running TPF. This platform issupported as cross-compilation target only.
Support for Solaris 11.3 and earlier has been removed in GCC 15.Support for Solaris 10 has been removed in GCC 10. Support for Solaris9 has been removed in GCC 5. Support for Solaris 8 has been removed inGCC 4.8. Support for Solaris 7 has been removed in GCC 4.6.
Solaris 11.4 provides one or more of GCC 5, 7, 9, 10, 11, 12, 13, 14,and 15.
You need to install thesystem/header,system/linker, anddeveloper/assembler packages.
Trying to use the compatibility tools in/usr/ucb, from thecompatibility/ucb package, to install GCC has been observed tocause trouble. The fix is to remove/usr/ucb from yourPATH.
The build process works more smoothly with the legacy Solaris tools so,if you have/usr/xpg4/bin in yourPATH, we recommend thatyou place/usr/bin before/usr/xpg4/bin for the durationof the build.
We recommend the use of the Solaris assembler or the GNU assembler, inconjunction with the Solaris linker.
The GNUas versions included in Solaris 11.4, from GNUbinutils 2.30.1 or newer (in/usr/bin/gas and/usr/gnu/bin/as), are known to work. This is the minimumsupported version. The version from GNUbinutils 2.45 is known to work as well. Recent versions of the Solarisassembler in/usr/bin/as work almost as well, though. To use GNUas, configure with the options--with-gnu-as--with-as=/usr/gnu/bin/as.
For linking, the Solaris linker is preferred. If you want to use theGNU linker instead, the version in Solaris 11.4, from GNU Binutils2.30.1 or newer (in/usr/gnu/bin/ld and/usr/bin/gld),works. However, itgenerally lacks platform specific features, so better stay with Solarisld. To use the LTO linker plugin(-fuse-linker-plugin) with GNUld, GNU Binutilsmust be configured with--enable-largefile. To useSolarisld, we recommend to configure with--without-gnu-ld --with-ld=/usr/bin/ld to guarantee theright linker is found irrespective of the user’sPATH.
Note that your mileage may vary if you use a combination of the GNUtools and the Solaris tools: while the combination GNUas andSolarisld works well, the reverse combination Solarisas with GNUld may fail to build or cause memorycorruption at runtime in some cases for C++ programs.
To enable symbol versioning in ‘libstdc++’ and other runtimelibraries with the Solaris linker, you need to have any version of GNUc++filt, which is part of GNU Binutils. Symbol versioningwill be disabled if no appropriate version is found. Solarisc++filt from the Solaris Studio compilers doesnotwork.
In order to build the GNU Ada compiler, GNAT, a working GNAT is needed.Since Solaris 11.4 SRU 39, GNAT 11, 12, 13, 14, or 15 is bundled in thedeveloper/gcc/gcc-gnat package.
In order to build the GNU D compiler, GDC, a working ‘libphobos’ isneeded. That library wasn’t built by default in GCC 9–11 on SPARC, oron x86 when the Solaris assembler is used, but can be enabled byconfiguring with--enable-libphobos. Also, GDC 9.4.0 isrequired on x86, while GDC 9.3.0 is known to work on SPARC.
The versions of the GNU Multiple Precision Library (GMP), the MPFRlibrary and the MPC library bundled with Solaris 11.4 arerecent enough to match GCC’s requirements.
This section contains general configuration information for allSPARC-based platforms. In addition to reading this section, pleaseread all other sections that match your target.
Newer versions of the GNU Multiple Precision Library (GMP), the MPFRlibrary and the MPC library are known to be miscompiled by earlierversions of GCC on these platforms. We therefore recommend the useof the exact versions of these libraries listed as minimal versionsinthe prerequisites.
When GCC is configured to use GNU Binutils 2.14 or later, the binariesproduced are smaller than the ones produced using Solaris native tools;this difference is quite significant for binaries containing debugginginformation.
Starting with Solaris 7, the operating system is capable of executing64-bit SPARC V9 binaries. GCC 3.1 and later properly supportsthis; the-m64 option enables 64-bit code generation.
When configuring the GNU Multiple Precision Library (GMP), the MPFRlibrary or the MPC library on Solaris, the canonical target triplet mustbe specified as thebuild parameter on theconfigureline. This target triplet can be obtained by invoking./config.guess in the toplevel source directory of GCC (andnot that of GMP or MPFR or MPC). For example:
%srcdir/configure --build=sparc-sun-solaris2.11 --prefix=dirname
This is a synonym for ‘sparcv9-*-solaris2*’.
When configuring a 64-bit-default GCC on Solaris/SPARC, you must use abuild compiler that generates 64-bit code, either by default or byspecifying ‘CC='gcc -m64' CXX='g++ -m64' GDC='gdc -m64'’ toconfigure.Additionally, youmust pass--build=sparcv9-sun-solaris2.11or--build=sparc64-sun-solaris2.11 becauseconfig.guessmisdetects this situation, which can cause build failures.
When configuring the GNU Multiple Precision Library (GMP), the MPFRlibrary or the MPC library, the canonical target triplet must be specifiedas thebuild parameter on theconfigure line. For example:
%srcdir/configure --build=sparcv9-sun-solaris2.11 --prefix=dirname
The C6X family of processors. This port requires binutils 2.22 or newer.
CDS VISIUMcore processor.This configuration is intended for embedded systems.
Support for VxWorks is in flux. At present GCC supportsonly thevery recent VxWorks 5.5 (aka Tornado 2.2) release, and only on PowerPC.We welcome patches for other architectures supported by VxWorks 5.5.Support for VxWorks AE would also be welcome; we believe this is merelya matter of writing an appropriate “configlette” (see below). We arenot interested in supporting older, a.out or COFF-based, versions ofVxWorks in GCC 3.
VxWorks comes with an older version of GCC installed in$WIND_BASE/host; we recommend you do not overwrite it.Choose an installationprefix entirely outside$WIND_BASE.Before runningconfigure, create the directoriesprefixandprefix/bin. Link or copy the appropriate assembler,linker, etc. intoprefix/bin, and set yourPATH toinclude that directory while running bothconfigure andmake.
You must giveconfigure the--with-headers=$WIND_BASE/target/h switch so that it canfind the VxWorks system headers. Since VxWorks is a cross compilationtarget only, you must also specify--target=target.configure will attempt to create the directoryprefix/target/sys-include and copy files into it;make sure the user runningconfigure has sufficient privilegeto do so.
GCC’s exception handling runtime requires a special “configlette”module,contrib/gthr_supp_vxw_5x.c. Follow the instructions inthat file to add the module to your kernel build. (Future versions ofVxWorks will incorporate this module.)
GCC supports the x86-64 architecture implemented by the AMD64 processor(amd64-*-* is an alias for x86_64-*-*) on GNU/Linux, FreeBSD and NetBSD.On GNU/Linux the default is a bi-arch compiler which is able to generateboth 64-bit x86-64 and 32-bit x86 code (via the-m32 switch).
Unlike other systems, without special options abi-arch compiler is built which generates 32-bit code by default, butcan generate 64-bit x86-64 code with the-m64 switch. SinceGCC 4.7, there is also a configuration that defaults to 64-bit code, butcan generate 32-bit code with-m32. To configure and buildthis way, you have to provide all support libraries likelibgmpas 64-bit code, configure with--target=x86_64-pc-solaris2.11and ‘CC=gcc -m64’.
This target is intended for embedded Xtensa systems using the‘newlib’ C library. It uses ELF but does not support sharedobjects. Designed-defined instructions specified via theTensilica Instruction Extension (TIE) language are only supportedthrough inline assembly.
The Xtensa configuration information must be specified prior tobuilding GCC. Theinclude/xtensa-config.h headerfile contains the configuration information. If you created yourown Xtensa configuration with the Xtensa Processor Generator, thedownloaded files include a customized copy of this header file,which you can use to replace the default header file.
This target is for Xtensa systems running GNU/Linux. It supports ELFshared objects and the GNU C library (glibc). It also generatesposition-independent code (PIC) regardless of whether the-fpic or-fPIC options are used. In otherrespects, this target is the same as the‘xtensa*-*-elf’ target.
The 32-bit versions of Windows, including Windows 95, Windows NT, WindowsXP, and Windows Vista, are supported by several different targetplatforms. These targets differ in which Windows subsystem they targetand which C libraries are used.
GCC contains support for x86-64 using the mingw-w64runtime library, available fromhttps://www.mingw-w64.org/downloads/.This library should be used with the target triple x86_64-pc-mingw32.
Windows CE is supported as a target only on HitachiSuperH (sh-wince-pe), and MIPS (mips-wince-pe).
GCC no longer supports Windows NT on the Alpha or PowerPC.
GCC no longer supports the Windows POSIX subsystem.
Old target names including *-*-winnt and *-*-windowsnt are no longer used.
UWIN support has been removed due to a lack of maintenance.
Ports of GCC are included with theCygwin environment.
GCC will build under Cygwin without modification; it does not buildwith Microsoft’s C++ compiler and there are no plans to make it do so.
The Cygwin native compiler can be configured to target any 32-bit x86cpu architecture desired; the default is i686-pc-cygwin. It should beused with as up-to-date a version of binutils as possible; use eitherthe latest official GNU Binutils release in the Cygwin distribution,or version 2.20 or above if building your own.
GCC will build with and support only MinGW runtime 3.12 and later.Earlier versions of headers are incompatible with the new default semanticsofextern inline in-std=c99 and-std=gnu99 modes.
GCC contains support files for many older (1980s and early1990s) Unix variants. For the most part, support for these systemshas not been deliberately removed, but it has not been maintained forseveral years and may suffer from bitrot.
Starting with GCC 3.1, each release has a list of “obsoleted” systems.Support for these systems is still present in that release, butconfigure will fail unless the--enable-obsoleteoption is given. Unless a maintainer steps forward, support for thesesystems will be removed from the next release of GCC.
Support for old systems as hosts for GCC can cause problems if theworkarounds for compiler, library and operating system bugs affect thecleanliness or maintainability of the rest of GCC. In some cases, tobring GCC up on such a system, if still possible with current GCC, mayrequire first installing an old version of GCC which did work on thatsystem, and using it to compile a more recent GCC, to avoid bugs in thevendor compiler. Old releases of GCC 1 and GCC 2 are available in theold-releases directory on theGCC mirrorsites. Header bugs may generally be avoided usingfixincludes, but bugs or deficiencies in libraries and theoperating system may still cause problems.
Support for older systems as targets for cross-compilation is lessproblematic than support for them as hosts for GCC; if an enthusiastwishes to make such a target work again (including resurrecting any ofthe targets that never worked with GCC 2, starting from the lastversion before they were removed), patchesfollowing the usual requirements would belikely to be accepted, since they should not affect the support for moremodern targets.
Some of the information on specific systems above relates tosuch older systems, but much of the informationabout GCC on such systems (which may no longer be applicable tocurrent GCC) is to be found in the GCC texinfo manual.
C++ support is significantly better on ELF targets if you use theGNU linker; duplicate copies ofinlines, vtables and template instantiations will be discardedautomatically.
Return to the GCC Installation page
Copyright (C)Free Software Foundation, Inc.Verbatim copying and distribution of this entire article ispermitted in any medium, provided this notice is preserved.
These pages aremaintained by the GCC team.Last modified 2026-01-28.