Installing GCC

Now that GCC is configured, you are ready to build the compiler and runtime libraries.

Some commands executed when making the compiler may fail (return a nonzero status) and be ignored by make. These failures, which are often due to files that were not found, are expected, and can safely be ignored.

It is normal to have compiler warnings when compiling certain files. Unless you are a GCC developer, you can generally ignore these warnings unless they cause compilation to fail. Developers should attempt to fix any warnings encountered, however they can temporarily continue past warnings-as-errors by specifying the configure flag --disable-werror.

On certain old systems, defining certain environment variables such as CC can interfere with the functioning of make.

If you encounter seemingly strange errors when trying to build the compiler in a directory other than the source directory, it could be because you have previously configured the compiler in the source directory. Make sure you have done all the necessary preparations.

If you build GCC on a BSD system using a directory stored in an old System V file system, problems may occur in running fixincludes if the System V file system doesn’t support symbolic links. These problems result in a failure to fix the declaration of size_t in sys/types.h. If you find that size_t is a signed type and that type mismatches occur, this could be the cause.

The solution is not to use such a directory for building GCC.

Similarly, when building from SVN or snapshots, or if you modify *.l files, you need the Flex lexical analyzer generator installed. If you do not modify *.l files, releases contain the Flex-generated files and you do not need Flex installed to build them. There is still one Flex-based lexical analyzer (part of the build machinery, not of GCC itself) that is used even if you only build the C front end.

When building from SVN or snapshots, or if you modify Texinfo documentation, you need version 4.7 or later of Texinfo installed if you want Info documentation to be regenerated. Releases contain Info documentation pre-built for the unmodified documentation in the release.

Building a native compiler

For a native build, the default configuration is to perform a 3-stage bootstrap of the compiler when ‘make’ is invoked. This will build the entire GCC system and ensure that it compiles itself correctly. It can be disabled with the --disable-bootstrap parameter to ‘configure’, but bootstrapping is suggested because the compiler will be tested more completely and could also have better performance.

The bootstrapping process will complete the following steps:

If you are short on disk space you might consider ‘make bootstrap-lean’ instead. The sequence of compilation is the same described above, but object files from the stage1 and stage2 of the 3-stage bootstrap of the compiler are deleted as soon as they are no longer needed.

If you wish to use non-default GCC flags when compiling the stage2 and stage3 compilers, set BOOT_CFLAGS on the command line when doing ‘make’. For example, if you want to save additional space during the bootstrap and in the final installation as well, you can build the compiler binaries without debugging information as in the following example. This will save roughly 40% of disk space both for the bootstrap and the final installation. (Libraries will still contain debugging information.)

make BOOT_CFLAGS='-O' bootstrap

You can place non-default optimization flags into BOOT_CFLAGS; they are less well tested here than the default of ‘-g -O2’, but should still work. In a few cases, you may find that you need to specify special flags such as -msoft-float here to complete the bootstrap; or, if the native compiler miscompiles the stage1 compiler, you may need to work around this, by choosing BOOT_CFLAGS to avoid the parts of the stage1 compiler that were miscompiled, or by using ‘make bootstrap4’ to increase the number of stages of bootstrap.

BOOT_CFLAGS does not apply to bootstrapped target libraries. Since these are always compiled with the compiler currently being bootstrapped, you can use CFLAGS_FOR_TARGET to modify their compilation flags, as for non-bootstrapped target libraries. Again, if the native compiler miscompiles the stage1 compiler, you may need to work around this by avoiding non-working parts of the stage1 compiler. Use STAGE1_TFLAGS to this end.

If you used the flag --enable-languages=… to restrict the compilers to be built, only those you’ve actually enabled will be built. This will of course only build those runtime libraries, for which the particular compiler has been built. Please note, that re-defining LANGUAGES when calling ‘makedoes not work anymore!

If the comparison of stage2 and stage3 fails, this normally indicates that the stage2 compiler has compiled GCC incorrectly, and is therefore a potentially serious bug which you should investigate and report. (On a few systems, meaningful comparison of object files is impossible; they always appear “different”. If you encounter this problem, you will need to disable comparison in the Makefile.)

If you do not want to bootstrap your compiler, you can configure with --disable-bootstrap. In particular cases, you may want to bootstrap your compiler even if the target system is not the same as the one you are building on: for example, you could build a powerpc-unknown-linux-gnu toolchain on a powerpc64-unknown-linux-gnu host. In this case, pass --enable-bootstrap to the configure script.

BUILD_CONFIG can be used to bring in additional customization to the build. It can be set to a whitespace-separated list of names. For each such NAME, top-level config/NAME.mk will be included by the top-level Makefile, bringing in any settings it contains. The default BUILD_CONFIG can be set using the configure option --with-build-config=NAME.... Some examples of supported build configurations are:

bootstrap-O1

Removes any -O-started option from BOOT_CFLAGS, and adds -O1 to it. ‘BUILD_CONFIG=bootstrap-O1’ is equivalent to ‘BOOT_CFLAGS='-g -O1'’.

bootstrap-O3

Analogous to bootstrap-O1.

bootstrap-lto

Enables Link-Time Optimization for host tools during bootstrapping. ‘BUILD_CONFIG=bootstrap-lto’ is equivalent to adding -flto to ‘BOOT_CFLAGS’. This option assumes that the host supports the linker plugin (e.g. GNU ld version 2.21 or later or GNU gold version 2.21 or later).

bootstrap-lto-noplugin

This option is similar to bootstrap-lto, but is intended for hosts that do not support the linker plugin. Without the linker plugin static libraries are not compiled with link-time optimizations. Since the GCC middle end and back end are in libbackend.a this means that only the front end is actually LTO optimized.

bootstrap-lto-lean

This option is similar to bootstrap-lto, but is intended for faster build by only using LTO in the final bootstrap stage. With ‘make profiledbootstrap’ the LTO frontend is trained only on generator files.

bootstrap-debug

Verifies that the compiler generates the same executable code, whether or not it is asked to emit debug information. To this end, this option builds stage2 host programs without debug information, and uses contrib/compare-debug to compare them with the stripped stage3 object files. If BOOT_CFLAGS is overridden so as to not enable debug information, stage2 will have it, and stage3 won’t. This option is enabled by default when GCC bootstrapping is enabled, if strip can turn object files compiled with and without debug info into identical object files. In addition to better test coverage, this option makes default bootstraps faster and leaner.

bootstrap-debug-big

Rather than comparing stripped object files, as in bootstrap-debug, this option saves internal compiler dumps during stage2 and stage3 and compares them as well, which helps catch additional potential problems, but at a great cost in terms of disk space. It can be specified in addition to ‘bootstrap-debug’.

bootstrap-debug-lean

This option saves disk space compared with bootstrap-debug-big, but at the expense of some recompilation. Instead of saving the dumps of stage2 and stage3 until the final compare, it uses -fcompare-debug to generate, compare and remove the dumps during stage3, repeating the compilation that already took place in stage2, whose dumps were not saved.

bootstrap-debug-lib

This option tests executable code invariance over debug information generation on target libraries, just like bootstrap-debug-lean tests it on host programs. It builds stage3 libraries with -fcompare-debug, and it can be used along with any of the bootstrap-debug options above.

There aren’t -lean or -big counterparts to this option because most libraries are only build in stage3, so bootstrap compares would not get significant coverage. Moreover, the few libraries built in stage2 are used in stage3 host programs, so we wouldn’t want to compile stage2 libraries with different options for comparison purposes.

bootstrap-debug-ckovw

Arranges for error messages to be issued if the compiler built on any stage is run without the option -fcompare-debug. This is useful to verify the full -fcompare-debug testing coverage. It must be used along with bootstrap-debug-lean and bootstrap-debug-lib.

bootstrap-cet

This option enables Intel CET for host tools during bootstrapping. ‘BUILD_CONFIG=bootstrap-cet’ is equivalent to adding -fcf-protection to ‘BOOT_CFLAGS’. This option assumes that the host supports Intel CET (e.g. GNU assembler version 2.30 or later).

bootstrap-time

Arranges for the run time of each program started by the GCC driver, built in any stage, to be logged to time.log, in the top level of the build tree.

Building a cross compiler

When building a cross compiler, it is not generally possible to do a 3-stage bootstrap of the compiler. This makes for an interesting problem as parts of GCC can only be built with GCC.

To build a cross compiler, we recommend first building and installing a native compiler. You can then use the native GCC compiler to build the cross compiler. The installed native compiler needs to be GCC version 2.95 or later.

Assuming you have already installed a native copy of GCC and configured your cross compiler, issue the command make, which performs the following steps:

Note that if an error occurs in any step the make process will exit.

If you are not building GNU binutils in the same source tree as GCC, you will need a cross-assembler and cross-linker installed before configuring GCC. Put them in the directory prefix/target/bin. Here is a table of the tools you should put in this directory:

as

This should be the cross-assembler.

ld

This should be the cross-linker.

ar

This should be the cross-archiver: a program which can manipulate archive files (linker libraries) in the target machine’s format.

ranlib

This should be a program to construct a symbol table in an archive file.

The installation of GCC will find these programs in that directory, and copy or link them to the proper place to for the cross-compiler to find them when run later.

The easiest way to provide these files is to build the Binutils package. Configure it with the same --host and --target options that you use for configuring GCC, then build and install them. They install their executables automatically into the proper directory. Alas, they do not support all the targets that GCC supports.

If you are not building a C library in the same source tree as GCC, you should also provide the target libraries and headers before configuring GCC, specifying the directories with --with-sysroot or --with-headers and --with-libs. Many targets also require “start files” such as crt0.o and crtn.o which are linked into each executable. There may be several alternatives for crt0.o, for use with profiling or other compilation options. Check your target’s definition of STARTFILE_SPEC to find out what start files it uses.

Building in parallel

GNU Make 3.80 and above, which is necessary to build GCC, support building in parallel. To activate this, you can use ‘make -j 2’ instead of ‘make’. You can also specify a bigger number, and in most cases using a value greater than the number of processors in your machine will result in fewer and shorter I/O latency hits, thus improving overall throughput; this is especially true for slow drives and network filesystems.

Building the Ada compiler

In order to build GNAT, the Ada compiler, you need a working GNAT compiler (GCC version 4.0 or later). This includes GNAT tools such as gnatmake and gnatlink, since the Ada front end is written in Ada and uses some GNAT-specific extensions.

In order to build a cross compiler, it is suggested to install the new compiler as native first, and then use it to build the cross compiler.

configure does not test whether the GNAT installation works and has a sufficiently recent version; if too old a GNAT version is installed, the build will fail unless --enable-languages is used to disable building the Ada front end.

ADA_INCLUDE_PATH and ADA_OBJECT_PATH environment variables must not be set when building the Ada compiler, the Ada tools, or the Ada runtime libraries. You can check that your build environment is clean by verifying that ‘gnatls -v’ lists only one explicit path in each section.

Building with profile feedback

It is possible to use profile feedback to optimize the compiler itself. This should result in a faster compiler binary. Experiments done on x86 using gcc 3.3 showed approximately 7 percent speedup on compiling C programs. To bootstrap the compiler with profile feedback, use make profiledbootstrap.

When ‘make profiledbootstrap’ is run, it will first build a stage1 compiler. This compiler is used to build a stageprofile compiler instrumented to collect execution counts of instruction and branch probabilities. Training run is done by building stagetrain compiler. Finally a stagefeedback compiler is built using the information collected.

Unlike standard bootstrap, several additional restrictions apply. The compiler used to build stage1 needs to support a 64-bit integral type. It is recommended to only use GCC for this.

On Linux/x86_64 hosts with some restrictions (no virtualization) it is also possible to do autofdo build with ‘make autoprofiledback’. This uses Linux perf to sample branches in the binary and then rebuild it with feedback derived from the profile. Linux perf and the autofdo toolkit needs to be installed for this.

Only the profile from the current build is used, so when an error occurs it is recommended to clean before restarting. Otherwise the code quality may be much worse.


Return to the GCC Installation page