Go to the previous, next section.

Library Maintenance

How to Install the GNU C Library

Installation of the GNU C library is relatively simple.

You need the latest version of GNU make. Modifying the GNU C Library to work with other make programs would be so hard that we recommend you port GNU make instead. Really.

To configure the GNU C library for your system, run the shell script `configure' with sh. Use an argument which is the conventional GNU name for your system configuration--for example, `sparc-sun-sunos4.1', for a Sun 4 running Sunos 4.1. See section 'Installing GNU CC' in Using and Porting GNU CC, for a full description of standard GNU configuration names.

The GNU C Library currently supports configurations that match the following patterns:

sparc-sun-sunos4.n
m68k-hp-bsd4.3
m68k-sun-sunos4.n
m68k-sony-bsd4.3
mips-dec-ultrix4.n
i386-bsd4.3
i386-sysv
i386-sysv4

While no other configurations are supported, there are handy aliases for these few. (These aliases work in other GNU software as well.)

sun4-sunos4.n
hp320-bsd4.3
sun3-sunos4.n
news
decstation-ultrix
i386-svr4

Here are some options that you should specify (if appropriate) when you run configure:

`--with-gnu-ld'
Use this option if you plan to use GNU ld to link programs with the GNU C Library. (We strongly recommend that you do.)

`--with-gnu-as'
Use this option if you plan to use the GNU assembler, gas, when building the GNU C Library. On some systems, the library may not build properly if you do not use gas.

`--nfp'

Use this option if your computer lacks hardware floating point support.

`--prefix=directory'
Install machine-independent data files in subdirectories of `directory'. (You can also set this in `configparms'; see below.)

`--exec-prefix=directory'
Install the library and other machine-dependent files in subdirectories of `directory'. (You can also set this in `configparms'; see below.)

The simplest way to run configure is to do it in the directory that contains the library sources. This prepares to build the library in that very directory.

You can prepare to build the library in some other directory by going to that other directory to run configure. In order to run configure, you will have to specify a directory for it, like this:

mkdir ../hp320
cd ../hp320
../src/configure hp320-bsd4.3

configure looks for the sources in whatever directory you specified for finding configure itself. It does not matter where in the file system the source and build directories are--as long as you specify the source directory when you run configure, you will get the proper results.

This feature lets you keep sources and binaries in different directories, and that makes it easy to build the library for several different machines from the same set of sources. Simply create a build directory for each target machine, and run configure in that directory specifying the target machine's configuration name.

The library has a number of special-purpose configuration parameters. These are defined in the file `Makeconfig'; see the comments in that file for the details.

But don't edit the file `Makeconfig' yourself--instead, create a file `configparms' in the directory where you are building the library, and define in that file the parameters you want to specify. `configparms' should not be an edited copy of `Makeconfig'; specify only the parameters that you want to override.

Some of the machine-dependent code for some machines uses extensions in the GNU C compiler, so you may need to compile the library with GCC. (In fact, all of the existing complete ports require GCC.)

The current release of the C library contains some header files that the compiler normally provides: `stddef.h', `stdarg.h', and several files with names of the form `va-machine.h'. The versions of these files that came with older releases of GCC do not work properly with the GNU C library. The `stddef.h' file in release 2.2 and later of GCC is correct. If you have release 2.2 or later of GCC, use its version of `stddef.h' instead of the C library's. To do this, put the line `override stddef.h =' in `configparms'. The other files are corrected in release 2.3 and later of GCC. `configure' will automatically detect whether the installed `stdarg.h' and `va-machine.h' files are compatible with the C library, and use its own if not.

There is a potential problem with the size_t type and versions of GCC prior to release 2.4. ANSI C requires that size_t always be an unsigned type. For compatibility with existing systems' header files, GCC defines size_t in `stddef.h' to be whatever type the system's `sys/types.h' defines it to be. Most Unix systems that define size_t in `sys/types.h', define it to be a signed type. Some code in the library depends on size_t being an unsigned type, and will not work correctly if it is signed.

The GNU C library code which expects size_t to be unsigned is correct. The definition of size_t as a signed type is incorrect. We plan that in version 2.4, GCC will always define size_t as an unsigned type, and the `fixincludes' script will massage the system's `sys/types.h' so as not to conflict with this.

In the meantime, we work around this problem by telling GCC explicitly to use an unsigned type for size_t when compiling the GNU C library. `configure' will automatically detect what type GCC uses for size_t arrange to override it if necessary.

To build the library, type make lib. This will produce a lot of output, some of which looks like errors from make (but isn't). Look for error messages from make containing `***'. Those indicate that something is really wrong. Using the `-w' option to make may make the output easier to understand (this option tells make to print messages telling you what subdirectories it is working on).

To install the library and header files, type make install, after setting the installation directories in `configparms'. This will build things if necessary, before installing them.

Reporting Bugs

There are probably bugs in the GNU C library. If you report them, they will get fixed. If you don't, no one will ever know about them and they will remain unfixed for all eternity, if not longer.

To report a bug, first you must find it. Hopefully, this will be the hard part. Once you've found a bug, make sure it's really a bug. A good way to do this is to see if the GNU C library behaves the same way some other C library does. If so, probably you are wrong and the libraries are right (but not necessarily). If not, one of the libraries is probably wrong.

Once you're sure you've found a bug, try to narrow it down to the smallest test case that reproduces the problem. In the case of a C library, you really only need to narrow it down to one library function call, if possible. This should not be too difficult.

The final step when you have a simple test case is to report the bug. When reporting a bug, send your test case, the results you got, the results you expected, what you think the problem might be (if you've thought of anything), your system type, and the version of the GNU C library which you are using.

If you think you have found some way in which the GNU C library does not conform to the ANSI and POSIX standards (see section Standards and Portability), that is definitely a bug. Report it!

Send bug reports to the Internet address `bug-glibc@prep.ai.mit.edu' or the UUCP path `mit-eddie!prep.ai.mit.edu!bug-glibc'. If you have other problems with installation, use, or the documentation, please report those as well.

Adding New Functions

The process of building the library is driven by the makefiles, which make heavy use of special features of GNU make. The makefiles are very complex, and you probably don't want to try to understand them. But what they do is fairly straightforward, and only requires that you define a few variables in the right places.

The library sources are divided into subdirectories, grouped by topic. The `string' subdirectory has all the string-manipulation functions, `stdio' has all the standard I/O functions, etc.

Each subdirectory contains a simple makefile, called `Makefile', which defines a few make variables and then includes the global makefile `Rules' with a line like:

include ../Rules

The basic variables that a subdirectory makefile defines are:

subdir
The name of the subdirectory, for example `stdio'. This variable must be defined.

headers
The names of the header files in this section of the library, such as `stdio.h'.

routines
aux
The names of the modules (source files) in this section of the library. These should be simple names, such as `strlen' (rather than complete file names, such as `strlen.c'). Use routines for modules that define functions in the library, and aux for auxiliary modules containing things like data definitions. But the values of routines and aux are just concatenated, so there really is no practical difference.

tests
The names of test programs for this section of the library. These should be simple names, such as `tester' (rather than complete file names, such as `tester.c'). `make tests' will build and run all the test programs. If a test program needs input, put the test data in a file called `test-program.input'; it will be given to the test program on its standard input. If a test program wants to be run with arguments, put the arguments (all on a single line) in a file called `test-program.args'.

others
The names of "other" programs associated with this section of the library. These are programs which are not tests per se, but are other small programs included with the library. They are built by `make others'.

install-lib
install-data
install
Files to be installed by `make install'. Things listed in `install-lib' are installed in the directory specified by `libdir' in `Makeconfig' (see section How to Install the GNU C Library). Files listed in install-data are installed in the directory specified by `datadir' in `configparms' or `Makeconfig'. Files listed in install are installed in the directory specified by `bindir' in `Makeconfig'.

distribute
Other files from this subdirectory which should be put into a distribution tar file. You need not list here the makefile itself or the source and header files listed in the other standard variables. Only define distribute if there are files used in an unusual way that should go into the distribution.

generated
Files which are generated by `Makefile' in this subdirectory. These files will be removed by `make clean', and they will never go into a distribution.

extra-objs
Extra object files which are built by `Makefile' in this subdirectory. This should be a list of file names like `foo.o'; the files will actually be found in whatever directory object files are being built in. These files will be removed by `make clean'. This variable is used for secondary object files needed to build others or tests.

Porting the GNU C Library

The GNU C library is written to be easily portable to a variety of machines and operating systems. Machine- and operating system-dependent functions are well separated to make it easy to add implementations for new machines or operating systems. This section describes the layout of the library source tree and explains the mechanisms used to select machine-dependent code to use.

All the machine-dependent and operating system-dependent files in the library are in the subdirectory `sysdeps' under the top-level library source directory. This directory contains a hierarchy of subdirectories (see section The Layout of the `sysdeps' Directory Hierarchy).

Each subdirectory of `sysdeps' contains source files for a particular machine or operating system, or for a class of machine or operating system (for example, systems by a particular vendor, or all machines that use IEEE 754 floating-point format). A configuration specifies an ordered list of these subdirectories. Each subdirectory implicitly appends its parent directory to the list. For example, specifying the list `unix/bsd/vax' is equivalent to specifying the list `unix/bsd/vax unix/bsd unix'. A subdirectory can also specify that it implies other subdirectories which are not directly above it in the directory hierarchy. If the file `Implies' exists in a subdirectory, it lists other subdirectories of `sysdeps' which are appended to the list, appearing after the subdirectory containing the `Implies' file. Lines in an `Implies' file that begin with a `#' character are ignored as comments. For example, `unix/bsd/Implies' contains:

# BSD has Internet-related things.
unix/inet
and `unix/Implies' contains:
posix

So the final list is `unix/bsd/vax unix/bsd vax unix/inet unix posix'.

`sysdeps' has two "special" subdirectories, called `generic' and `stub'. These two are always implicitly appended to the list of subdirectories (in that order), so you needn't put them in an `Implies' file, and you should not create any subdirectories under them. `generic' is for things that can be implemented in machine-independent C, using only other machine-independent functions in the C library. `stub' is for stub versions of functions which cannot be implemented on a particular machine or operating system. The stub functions always return an error, and set errno to ENOSYS (Function not implemented). See section Error Reporting.

A source file is known to be system-dependent by its having a version in `generic' or `stub'; every system-dependent function should have either a generic or stub implementation (there is no point in having both).

If you come across a file that is in one of the main source directories (`string', `stdio', etc.), and you want to write a machine- or operating system-dependent version of it, move the file into `sysdeps/generic' and write your new implementation in the appropriate system-specific subdirectory. Note that if a file is to be system-dependent, it must not appear in one of the main source directories.

There are a few special files that may exist in each subdirectory of `sysdeps':

`Makefile'
A makefile for this machine or operating system, or class of machine or operating system. This file is included by the library makefile `Makerules', which is used by the top-level makefile and the subdirectory makefiles. It can change the variables set in the including makefile or add new rules. It can use GNU make conditional directives based on the variable `subdir' (see above) to select different sets of variables and rules for different sections of the library. It can also set the make variable `sysdep-routines', to specify extra modules to be included in the library. You should use `sysdep-routines' rather than adding modules to `routines' because the latter is used in determining what to distribute for each subdirectory of the main source tree.

Each makefile in a subdirectory in the ordered list of subdirectories to be searched is included in order. Since several system-dependent makefiles may be included, each should append to `sysdep-routines' rather than simply setting it:

sysdep-routines := $(sysdep-routines) foo bar

`Subdirs'
This file contains the names of new whole subdirectories under the top-level library source tree that should be included for this system. These subdirectories are treated just like the system-independent subdirectories in the library source tree, such as `stdio' and `math'.

Use this when there are whole new sets of routines and header files that should go into the library for the system this subdirectory of `sysdeps' implements. For example, `sysdeps/unix/inet/Subdirs' contains `inet'; the `inet' directory contains various network-oriented operations which only make sense to put in the library on systems that support the Internet.

`Dist'
This file contains the names of files (relative the the subdirectory of `sysdeps' in which it appears) which should be included in the distribution. List any new files used by rules in the `Makefile' in the same directory, or header files used by the source files in that directory. You don't need to list files that are implementations (either C or assembly source) of routines whose names are given in the machine-independent makefiles in the main source tree.

That is the general system for how system-dependencies are isolated. The next section explains how to decide what directories in `sysdeps' to use. section Porting the GNU C Library to Unix Systems, has some tips on porting the library to Unix variants.

The Layout of the `sysdeps' Directory Hierarchy

A GNU configuration name has three parts: the CPU type, the manufacturer's name, and the operating system. `configure' uses these to pick the list of system-dependent directories to look for. If the `--nfp' option is not passed to `configure', the directory `machine/fpu' is also used. The operating system often has a base operating system; for example, if the operating system is `sunos4.1', the base operating system is `unix/bsd'. The algorithm used to pick the list of directories is simple: `configure' makes a list of the base operating system, manufacturer, CPU type, and operating system, in that order. It then concatenates all these together with slashes in between, to produce a directory name; for example, the configuration `sparc-sun-sunos4.1' results in `unix/bsd/sun/sparc/sunos4.1'. `configure' then tries removing each element of the list in turn, so `unix/bsd/sparc' and `sun/sparc' are also tried, among others. Since the precise version number of the operating system is often not important, and it would be very inconvenient, for example, to have identical `sunos4.1.1' and `sunos4.1.2' directories, `configure' tries successively less specific operating system names by removing trailing suffixes starting with a period.

Here is the complete list of directories that would be tried for the configuration `sparc-sun-sunos4.1':

sparc/fpu
unix/bsd/sun/sunos4.1/sparc
unix/bsd/sun/sunos4.1
unix/bsd/sun/sunos4/sparc
unix/bsd/sun/sunos4
unix/bsd/sun/sparc
unix/bsd/sun
unix/bsd/sunos4.1/sparc
unix/bsd/sunos4.1
unix/bsd/sunos4/sparc
unix/bsd/sunos4
unix/bsd/sparc
unix/bsd
sun/sunos4.1/sparc
sun/sunos4.1
sun/sunos4/sparc
sun/sunos4
sun/sparc
sun
sunos4.1/sparc
sunos4.1
sunos4/sparc
sunos4
sparc

Different machine architectures are generally at the top level of the `sysdeps' directory tree. For example, `sysdeps/sparc' and `sysdeps/m68k'. These contain files specific to those machine architectures, but not specific to any particular operating system. There might be subdirectories for specializations of those architectures, such as `sysdeps/m68k/68020'. Code which is specific to the floating-point coprocessor used with a particular machine should go in `sysdeps/machine/fpu'.

There are a few directories at the top level of the `sysdeps' hierarchy that are not for particular machine architectures.

`generic'
`stub'
As described above (see section Porting the GNU C Library), these are the two subdirectories that every configuration implicitly uses after all others.

`ieee754'
This directory is for code using the IEEE 754 floating-point format, where the C type float is IEEE 754 single-precision format, and double is IEEE 754 double-precision format. Usually this directory is referred to in the `Implies' file in a machine architecture-specific directory, such as `m68k/Implies'.

`posix'
This directory contains implementations of things in the library in terms of POSIX.1 functions. This includes some of the POSIX.1 functions themselves. Of course, POSIX.1 cannot be completely implemented in terms of itself, so a configuration using just `posix' cannot be complete.

`unix'
This is the directory for Unix-like things. See See section Porting the GNU C Library to Unix Systems. `unix' implies `posix'. There are some special-purpose subdirectories of `unix':

`unix/common'
This directory is for things common to both BSD and System V release 4. Both `unix/bsd' and `unix/sysv/sysv4' imply `unix/common'.

`unix/inet'
This directory is for socket and related functions on Unix systems. The `inet' top-level subdirectory is enabled by `unix/inet/Subdirs'. `unix/common' implies `unix/inet'.

  • mach This is the directory for things based on the Mach microkernel from CMU (including the GNU operating system). Other basic operating systems (VMS, for example) would have their own directories at the top level of the `sysdeps' hierarchy, parallel to `unix' and `mach'.
  • Porting the GNU C Library to Unix Systems

    Most Unix systems are fundamentally very similar. There are variations between different machines, and variations in what facilities are provided by the kernel. But the interface to the operating system facilities is, for the most part, pretty uniform and simple.

    The code for Unix systems is in the directory `unix', at the top level of the `sysdeps' hierarchy. This directory contains subdirectories (and subdirectory trees) for various Unix variants.

    The functions which are system calls in most Unix systems are implemented in assembly code in files in `sysdeps/unix'. These files are named with a suffix of `.S'; for example, `__open.S'. Files ending in `.S' are run through the C preprocessor before being fed to the assembler.

    These files all use a set of macros that should be defined in `sysdep.h'. The `sysdep.h' file in `sysdeps/unix' partially defines them; a `sysdep.h' file in another directory must finish defining them for the particular machine and operating system variant. See `sysdeps/unix/sysdep.h' and the machine-specific `sysdep.h' implementations to see what these macros are and what they should do.

    The system-specific makefile for the `unix' directory, `sysdeps/unix/Makefile', gives rules to generate several files from the Unix system you are building the library on (which is assumed to be the target system you are building the library for). All the generated files are put in the directory where the object files are kept; they should not affect the source tree itself. The files generated are `ioctls.h', `errnos.h', `sys/param.h', and `errlist.c' (for the `stdio' section of the library).

    Contributors to the GNU C Library

    The GNU C library was written almost entirely by Roland McGrath. Some parts of the library were contributed by other people.

    Go to the previous, next section.