GNU Libtool

For version 1.4.2, 10 September 2001

Gordon Matzigkeit
Alexandre Oliva
Thomas Tanner
Gary V. Vaughan


Table of Contents


Introduction

In the past, if a source code package developer wanted to take advantage of the power of shared libraries, he needed to write custom support code for each platform on which his package ran. He also had to design a configuration interface so that the package installer could choose what sort of libraries were built.

GNU Libtool simplifies the developer's job by encapsulating both the platform-specific dependencies, and the user interface, in a single script. GNU Libtool is designed so that the complete functionality of each host type is available via a generic interface, but nasty quirks are hidden from the programmer.

GNU Libtool's consistent interface is reassuring... users don't need to read obscure documentation in order to have their favorite source package build shared libraries. They just run your package configure script (or equivalent), and libtool does all the dirty work.

There are several examples throughout this document. All assume the same environment: we want to build a library, `libhello', in a generic way.

`libhello' could be a shared library, a static library, or both... whatever is available on the host system, as long as libtool has been ported to it.

This chapter explains the original design philosophy of libtool. Feel free to skip to the next chapter, unless you are interested in history, or want to write code to extend libtool in a consistent way.

Motivation for writing libtool

Since early 1995, several different GNU developers have recognized the importance of having shared library support for their packages. The primary motivation for such a change is to encourage modularity and reuse of code (both conceptually and physically) in GNU programs.

Such a demand means that the way libraries are built in GNU packages needs to be general, to allow for any library type the package installer might want. The problem is compounded by the absence of a standard procedure for creating shared libraries on different platforms.

The following sections outline the major issues facing shared library support in GNU, and how shared library support could be standardized with libtool.

The following specifications were used in developing and evaluating this system:

  1. The system must be as elegant as possible.
  2. The system must be fully integrated with the GNU Autoconf and Automake utilities, so that it will be easy for GNU maintainers to use. However, the system must not require these tools, so that it can be used by non-GNU packages.
  3. Portability to other (non-GNU) architectures and tools is desirable.

Implementation issues

The following issues need to be addressed in any reusable shared library system, specifically libtool:

  1. The package installer should be able to control what sort of libraries are built.
  2. It can be tricky to run dynamically linked programs whose libraries have not yet been installed. LD_LIBRARY_PATH must be set properly (if it is supported), or programs fail to run.
  3. The system must operate consistently even on hosts which don't support shared libraries.
  4. The commands required to build shared libraries may differ wildly from host to host. These need to be determined at configure time in a consistent way.
  5. It is not always obvious with which suffix a shared library should be installed. This makes it difficult for `Makefile' rules, since they generally assume that file names are the same from host to host.
  6. The system needs a simple library version number abstraction, so that shared libraries can be upgraded in place. The programmer should be informed how to design the interfaces to the library to maximize binary compatibility.
  7. The install `Makefile' target should warn the package installer to set the proper environment variables (LD_LIBRARY_PATH or equivalent), or run ldconfig.

Other implementations

Even before libtool was developed, many free software packages built and installed their own shared libraries. At first, these packages were examined to avoid reinventing existing features.

Now it is clear that none of these packages have documented the details of shared library systems that libtool requires. So, other packages have been more or less abandoned as influences.

A postmortem analysis of other implementations

In all fairness, each of the implementations that were examined do the job that they were intended to do, for a number of different host systems. However, none of these solutions seem to function well as a generalized, reusable component.

Most were too complex to use (much less modify) without understanding exactly what the implementation does, and they were generally not documented.

The main difficulty is that different vendors have different views of what libraries are, and none of the packages which were examined seemed to be confident enough to settle on a single paradigm that just works.

Ideally, libtool would be a standard that would be implemented as series of extensions and modifications to existing library systems to make them work consistently. However, it is not an easy task to convince operating system developers to mend their evil ways, and people want to build shared libraries right now, even on buggy, broken, confused operating systems.

For this reason, libtool was designed as an independent shell script. It isolates the problems and inconsistencies in library building that plague `Makefile' writers by wrapping the compiler suite on different platforms with a consistent, powerful interface.

With luck, libtool will be useful to and used by the GNU community, and that the lessons that were learned in writing it will be taken up by designers of future library systems.

The libtool paradigm

At first, libtool was designed to support an arbitrary number of library object types. After libtool was ported to more platforms, a new paradigm gradually developed for describing the relationship between libraries and programs.

In summary, "libraries are programs with multiple entry points, and more formally defined interfaces."

Version 0.7 of libtool was a complete redesign and rewrite of libtool to reflect this new paradigm. So far, it has proved to be successful: libtool is simpler and more useful than before.

The best way to introduce the libtool paradigm is to contrast it with the paradigm of existing library systems, with examples from each. It is a new way of thinking, so it may take a little time to absorb, but when you understand it, the world becomes simpler.

Using libtool

It makes little sense to talk about using libtool in your own packages until you have seen how it makes your life simpler. The examples in this chapter introduce the main features of libtool by comparing the standard library building procedure to libtool's operation on two different platforms:

`a23'
An Ultrix 4.2 platform with only static libraries.
`burger'
A NetBSD/i386 1.2 platform with shared libraries.

You can follow these examples on your own platform, using the preconfigured libtool script that was installed with libtool (see section Configuring libtool).

Source files for the following examples are taken from the `demo' subdirectory of the libtool distribution. Assume that we are building a library, `libhello', out of the files `foo.c' and `hello.c'.

Note that the `foo.c' source file uses the cos math library function, which is usually found in the standalone math library, and not the C library (see section `Trigonometric Functions' in The GNU C Library Reference Manual). So, we need to add -lm to the end of the link line whenever we link `foo.o' or `foo.lo' into an executable or a library (see section Inter-library dependencies).

The same rule applies whenever you use functions that don't appear in the standard C library... you need to add the appropriate -lname flag to the end of the link line when you link against those objects.

After we have built that library, we want to create a program by linking `main.o' against `libhello'.

Creating object files

To create an object file from a source file, the compiler is invoked with the `-c' flag (and any other desired flags):

burger$ gcc -g -O -c main.c
burger$

The above compiler command produces an object file, `main.o', from the source file `main.c'.

For most library systems, creating object files that become part of a static library is as simple as creating object files that are linked to form an executable:

burger$ gcc -g -O -c foo.c
burger$ gcc -g -O -c hello.c
burger$

Shared libraries, however, may only be built from position-independent code (PIC). So, special flags must be passed to the compiler to tell it to generate PIC rather than the standard position-dependent code.

Since this is a library implementation detail, libtool hides the complexity of PIC compiler flags by using separate library object files (which end in `.lo' instead of `.o'). On systems without shared libraries (or without special PIC compiler flags), these library object files are identical to "standard" object files.

To create library object files for `foo.c' and `hello.c', simply invoke libtool with the standard compilation command as arguments (see section Compile mode):

a23$ libtool gcc -g -O -c foo.c
gcc -g -O -c foo.c
echo timestamp > foo.lo
a23$ libtool gcc -g -O -c hello.c
gcc -g -O -c hello.c
echo timestamp > hello.lo
a23$

Note that libtool creates two files for each invocation. The `.lo' file is a library object, which may be built into a shared library, and the `.o' file is a standard object file. On `a23', the library objects are just timestamps, because only static libraries are supported.

On shared library systems, libtool automatically inserts the PIC generation flags into the compilation command, so that the library object and the standard object differ:

burger$ libtool gcc -g -O -c foo.c
gcc -g -O -c -fPIC -DPIC foo.c
mv -f foo.o foo.lo
gcc -g -O -c foo.c >/dev/null 2>&1
burger$ libtool gcc -g -O -c hello.c
gcc -g -O -c -fPIC -DPIC hello.c
mv -f hello.o hello.lo
gcc -g -O -c hello.c >/dev/null 2>&1
burger$

Notice that the second run of GCC has its output discarded. This is done so that compiler warnings aren't annoyingly duplicated.

Linking libraries

Without libtool, the programmer would invoke the ar command to create a static library:

burger$ ar cru libhello.a hello.o foo.o
burger$

But of course, that would be too simple, so many systems require that you run the ranlib command on the resulting library (to give it better karma, or something):

burger$ ranlib libhello.a
burger$

It seems more natural to use the C compiler for this task, given libtool's "libraries are programs" approach. So, on platforms without shared libraries, libtool simply acts as a wrapper for the system ar (and possibly ranlib) commands.

Again, the libtool library name differs from the standard name (it has a `.la' suffix instead of a `.a' suffix). The arguments to libtool are the same ones you would use to produce an executable named `libhello.la' with your compiler (see section Link mode):

a23$ libtool gcc -g -O -o libhello.la foo.o hello.o
libtool: cannot build libtool library `libhello.la' from non-libtool \
                objects
a23$

Aha! Libtool caught a common error... trying to build a library from standard objects instead of library objects. This doesn't matter for static libraries, but on shared library systems, it is of great importance.

So, let's try again, this time with the library object files. Remember also that we need to add -lm to the link command line because `foo.c' uses the cos math library function (see section Using libtool).

Another complication in building shared libraries is that we need to specify the path to the directory in which they (eventually) will be installed (in this case, `/usr/local/lib')(1):

a23$ libtool gcc -g -O -o libhello.la foo.lo hello.lo \
                -rpath /usr/local/lib -lm
mkdir .libs
ar cru .libs/libhello.a foo.o hello.o
ranlib .libs/libhello.a
creating libhello.la
a23$

Now, let's try the same trick on the shared library platform:

burger$ libtool gcc -g -O -o libhello.la foo.lo hello.lo \
                -rpath /usr/local/lib -lm
mkdir .libs
ld -Bshareable -o .libs/libhello.so.0.0 foo.lo hello.lo -lm
ar cru .libs/libhello.a foo.o hello.o
ranlib .libs/libhello.a
creating libhello.la
burger$

Now that's significantly cooler... libtool just ran an obscure ld command to create a shared library, as well as the static library.

Note how libtool creates extra files in the `.libs' subdirectory, rather than the current directory. This feature is to make it easier to clean up the build directory, and to help ensure that other programs fail horribly if you accidentally forget to use libtool when you should.

Linking executables

If you choose at this point to install the library (put it in a permanent location) before linking executables against it, then you don't need to use libtool to do the linking. Simply use the appropriate `-L' and `-l' flags to specify the library's location.

Some system linkers insist on encoding the full directory name of each shared library in the resulting executable. Libtool has to work around this misfeature by special magic to ensure that only permanent directory names are put into installed executables.

The importance of this bug must not be overlooked: it won't cause programs to crash in obvious ways. It creates a security hole, and possibly even worse, if you are modifying the library source code after you have installed the package, you will change the behaviour of the installed programs!

So, if you want to link programs against the library before you install it, you must use libtool to do the linking.

Here's the old way of linking against an uninstalled library:

burger$ gcc -g -O -o hell.old main.o libhello.a -lm
burger$

Libtool's way is almost the same(2) (see section Link mode):

a23$ libtool gcc -g -O -o hell main.o libhello.la -lm
gcc -g -O -o hell main.o ./.libs/libhello.a -lm
a23$

That looks too simple to be true. All libtool did was transform `libhello.la' to `./.libs/libhello.a', but remember that `a23' has no shared libraries.

On `burger' the situation is different:

burger$ libtool gcc -g -O -o hell main.o libhello.la -lm
gcc -g -O -o .libs/hell main.o -L./.libs -R/usr/local/lib -lhello -lm
creating hell
burger$

Now assume `libhello.la' had already been installed, and you want to link a new program with it. You could figure out where it lives by yourself, then run:

burger$ gcc -g -O -o test test.o -L/usr/local/lib -lhello

However, unless `/usr/local/lib' is in the standard library search path, you won't be able to run test. However, if you use libtool to link the already-installed libtool library, it will do The Right Thing (TM) for you:

burger$ libtool gcc -g -O -o test test.o /usr/local/lib/libhello.la
gcc -g -O -o .libs/test test.o -Wl,--rpath
-Wl,/usr/local/lib /usr/local/lib/libhello.a -lm
creating test
burger$

Note that libtool added the necessary run-time path flag, as well as `-lm', the library libhello.la depended upon. Nice, huh?

Since libtool created a wrapper script, you should use libtool to install it and debug it too. However, since the program does not depend on any uninstalled libtool library, it is probably usable even without the wrapper script. Libtool could probably be made smarter to avoid the creation of the wrapper script in this case, but this is left as an exercise for the reader.

Notice that the executable, hell, was actually created in the `.libs' subdirectory. Then, a wrapper script was created in the current directory.

On NetBSD 1.2, libtool encodes the installation directory of `libhello', by using the `-R/usr/local/lib' compiler flag. Then, the wrapper script guarantees that the executable finds the correct shared library (the one in `./.libs') until it is properly installed.

Let's compare the two different programs:

burger$ time ./hell.old
Welcome to GNU Hell!
** This is not GNU Hello.  There is no built-in mail reader. **
        0.21 real         0.02 user         0.08 sys
burger$ time ./hell
Welcome to GNU Hell!
** This is not GNU Hello.  There is no built-in mail reader. **
        0.63 real         0.09 user         0.59 sys
burger$

The wrapper script takes significantly longer to execute, but at least the results are correct, even though the shared library hasn't been installed yet.

So, what about all the space savings that shared libraries are supposed to yield?

burger$ ls -l hell.old libhello.a
-rwxr-xr-x  1 gord  gord  15481 Nov 14 12:11 hell.old
-rw-r--r--  1 gord  gord   4274 Nov 13 18:02 libhello.a
burger$ ls -l .libs/hell .libs/libhello.*
-rwxr-xr-x  1 gord  gord  11647 Nov 14 12:10 .libs/hell
-rw-r--r--  1 gord  gord   4274 Nov 13 18:44 .libs/libhello.a
-rwxr-xr-x  1 gord  gord  12205 Nov 13 18:44 .libs/libhello.so.0.0
burger$

Well, that sucks. Maybe I should just scrap this project and take up basket weaving.

Actually, it just proves an important point: shared libraries incur overhead because of their (relative) complexity. In this situation, the price of being dynamic is eight kilobytes, and the payoff is about four kilobytes. So, having a shared `libhello' won't be an advantage until we link it against at least a few more programs.

Debugging executables

If `hell' was a complicated program, you would certainly want to test and debug it before installing it on your system. In the above section, you saw how the libtool wrapper script makes it possible to run the program directly, but unfortunately, this mechanism interferes with the debugger:

burger$ gdb hell
GDB is free software and you are welcome to distribute copies of it
 under certain conditions; type "show copying" to see the conditions.
There is no warranty for GDB; type "show warranty" for details.
GDB 4.16 (i386-unknown-netbsd), (C) 1996 Free Software Foundation, Inc.

"hell": not in executable format: File format not recognized

(gdb) quit
burger$

Sad. It doesn't work because GDB doesn't know where the executable lives. So, let's try again, by invoking GDB directly on the executable:

burger$ gdb .libs/hell
trick:/home/src/libtool/demo$ gdb .libs/hell
GDB is free software and you are welcome to distribute copies of it
 under certain conditions; type "show copying" to see the conditions.
There is no warranty for GDB; type "show warranty" for details.
GDB 4.16 (i386-unknown-netbsd), (C) 1996 Free Software Foundation, Inc.
(gdb) break main
Breakpoint 1 at 0x8048547: file main.c, line 29.
(gdb) run
Starting program: /home/src/libtool/demo/.libs/hell
/home/src/libtool/demo/.libs/hell: can't load library 'libhello.so.2'

Program exited with code 020.
(gdb) quit
burger$

Argh. Now GDB complains because it cannot find the shared library that `hell' is linked against. So, we must use libtool in order to properly set the library path and run the debugger. Fortunately, we can forget all about the `.libs' directory, and just run it on the executable wrapper (see section Execute mode):

burger$ libtool gdb hell
GDB is free software and you are welcome to distribute copies of it
 under certain conditions; type "show copying" to see the conditions.
There is no warranty for GDB; type "show warranty" for details.
GDB 4.16 (i386-unknown-netbsd), (C) 1996 Free Software Foundation, Inc.
(gdb) break main
Breakpoint 1 at 0x8048547: file main.c, line 29.
(gdb) run
Starting program: /home/src/libtool/demo/.libs/hell

Breakpoint 1, main (argc=1, argv=0xbffffc40) at main.c:29
29	  printf ("Welcome to GNU Hell!\n");
(gdb) quit
The program is running.  Quit anyway (and kill it)? (y or n) y
burger$

Installing libraries

Installing libraries on a non-libtool system is quite straightforward... just copy them into place:(3)

burger$ su
Password: ********
burger# cp libhello.a /usr/local/lib/libhello.a
burger#

Oops, don't forget the ranlib command:

burger# ranlib /usr/local/lib/libhello.a
burger#

Libtool installation is quite simple, as well. Just use the install or cp command that you normally would (see section Install mode):

a23# libtool cp libhello.la /usr/local/lib/libhello.la
cp libhello.la /usr/local/lib/libhello.la
cp .libs/libhello.a /usr/local/lib/libhello.a
ranlib /usr/local/lib/libhello.a
a23#

Note that the libtool library `libhello.la' is also installed, to help libtool with uninstallation (see section Uninstall mode) and linking (see section Linking executables) and to help programs with dlopening (see section Dlopened modules).

Here is the shared library example:

burger# libtool install -c libhello.la /usr/local/lib/libhello.la
install -c .libs/libhello.so.0.0 /usr/local/lib/libhello.so.0.0
install -c libhello.la /usr/local/lib/libhello.la
install -c .libs/libhello.a /usr/local/lib/libhello.a
ranlib /usr/local/lib/libhello.a
burger#

It is safe to specify the `-s' (strip symbols) flag if you use a BSD-compatible install program when installing libraries. Libtool will either ignore the `-s' flag, or will run a program that will strip only debugging and compiler symbols from the library.

Once the libraries have been put in place, there may be some additional configuration that you need to do before using them. First, you must make sure that where the library is installed actually agrees with the `-rpath' flag you used to build it.

Then, running `libtool -n --finish libdir' can give you further hints on what to do (see section Finish mode):

burger# libtool -n --finish /usr/local/lib
PATH="$PATH:/sbin" ldconfig -m /usr/local/lib
-----------------------------------------------------------------
Libraries have been installed in:
   /usr/local/lib

To link against installed libraries in a given directory, LIBDIR,
you must use the `-LLIBDIR' flag during linking.

 You will also need to do one of the following:
   - add LIBDIR to the `LD_LIBRARY_PATH' environment variable
     during execution
   - add LIBDIR to the `LD_RUN_PATH' environment variable
     during linking
   - use the `-RLIBDIR' linker flag

See any operating system documentation about shared libraries for
more information, such as the ld and ld.so manual pages.
-----------------------------------------------------------------
burger#

After you have completed these steps, you can go on to begin using the installed libraries. You may also install any executables that depend on libraries you created.

Installing executables

If you used libtool to link any executables against uninstalled libtool libraries (see section Linking executables), you need to use libtool to install the executables after the libraries have been installed (see section Installing libraries).

So, for our Ultrix example, we would run:

a23# libtool install -c hell /usr/local/bin/hell
install -c hell /usr/local/bin/hell
a23#

On shared library systems, libtool just ignores the wrapper script and installs the correct binary:

burger# libtool install -c hell /usr/local/bin/hell
install -c .libs/hell /usr/local/bin/hell
burger#

Linking static libraries

Why return to ar and ranlib silliness when you've had a taste of libtool? Well, sometimes it is desirable to create a static archive that can never be shared. The most frequent case is when you have a set of object files that you use to build several different programs. You can create a "convenience library" out of those objects, and link programs with the library, instead of listing all object files for every program. This technique is often used to overcome GNU automake's lack of support for linking object files built from sources in other directories, because it supports linking with libraries from other directories. This limitation applies to GNU automake up to release 1.4; newer releases should support sources in other directories.

If you just want to link this convenience library into programs, then you could just ignore libtool entirely, and use the old ar and ranlib commands (or the corresponding GNU automake `_LIBRARIES' rules). You can even install a convenience library (but you probably don't want to) using libtool:

burger$ libtool ./install-sh -c libhello.a /local/lib/libhello.a
./install-sh -c libhello.a /local/lib/libhello.a
ranlib /local/lib/libhello.a
burger$

Using libtool for static library installation protects your library from being accidentally stripped (if the installer used the `-s' flag), as well as automatically running the correct ranlib command.

But libtool libraries are more than just collections of object files: they can also carry library dependency information, which old archives do not. If you want to create a libtool static convenience library, you can omit the `-rpath' flag and use `-static' to indicate that you're only interested in a static library. When you link a program with such a library, libtool will actually link all object files and dependency libraries into the program.

If you omit both `-rpath' and `-static', libtool will create a convenience library that can be used to create other libtool libraries, even shared ones. Just like in the static case, the library behaves as an alias to a set of object files and dependency libraries, but in this case the object files are suitable for inclusion in shared libraries. But be careful not to link a single convenience library, directly or indirectly, into a single program or library, otherwise you may get errors about symbol redefinitions.

When GNU automake is used, you should use noinst_LTLIBRARIES instead of lib_LTLIBRARIES for convenience libraries, so that the `-rpath' option is not passed when they are linked.

As a rule of thumb, link a libtool convenience library into at most one libtool library, and never into a program, and link libtool static convenience libraries only into programs, and only if you need to carry library dependency information to the user of the static convenience library.

Another common situation where static linking is desirable is in creating a standalone binary. Use libtool to do the linking and add the `-all-static' flag.

Invoking libtool

The libtool program has the following synopsis:

libtool [option]... [mode-arg]...

and accepts the following options:

`--config'
Display libtool configuration variables and exit.
`--debug'
Dump a trace of shell script execution to standard output. This produces a lot of output, so you may wish to pipe it to less (or more) or redirect to a file.
`-n'
`--dry-run'
Don't create, modify, or delete any files, just show what commands would be executed by libtool.
`--features'
Display basic configuration options. This provides a way for packages to determine whether shared or static libraries will be built.
`--finish'
Same as `--mode=finish'.
`--help'
Display a help message and exit. If `--mode=mode' is specified, then detailed help for mode is displayed.
`--mode=mode'
Use mode as the operation mode. By default, the operation mode is inferred from the mode-args. If mode is specified, it must be one of the following:
`compile'
Compile a source file into a libtool object.
`execute'
Automatically set the library path so that another program can use uninstalled libtool-generated programs or libraries.
`finish'
Complete the installation of libtool libraries on the system.
`install'
Install libraries or executables.
`link'
Create a library or an executable.
`uninstall'
Delete installed libraries or executables.
`clean'
Delete uninstalled libraries or executables.
`--version'
Print libtool version information and exit.

The mode-args are a variable number of arguments, depending on the selected operation mode. In general, each mode-arg is interpreted by programs libtool invokes, rather than libtool itself.

Compile mode

For compile mode, mode-args is a compiler command to be used in creating a `standard' object file. These arguments should begin with the name of the C compiler, and contain the `-c' compiler flag so that only an object file is created.

Libtool determines the name of the output file by removing the directory component from the source file name, then substituting the source code suffix (e.g. `.c' for C source code) with the library object suffix, `.lo'.

If shared libraries are being built, any necessary PIC generation flags are substituted into the compilation command. You can pass compiler and linker specific flags using `-Wc,flag' and `-Xcompiler flag' or `-Wl,flag' and `-Xlinker flag', respectively.

If the `-static' option is given, then a `.o' file is built, even if libtool was configured with `--disable-static'.

Note that the `-o' option is now fully supported. It is emulated on the platforms that don't support it (by locking and moving the objects), so it is really easy to use libtool, just with minor modifications to your Makefiles. Typing for example

libtool gcc -c foo/x.c -o foo/x.lo

will do what you expect.

Note, however, that, if the compiler does not support `-c' and `-o', it is impossible to compile `foo/x.c' without overwriting an existing `./x.o'. Therefore, if you do have a source file `./x.c', make sure you introduce dependencies in your `Makefile' to make sure `./x.o' (or `./x.lo') is re-created after any sub-directory's `x.lo':

x.o x.lo: foo/x.lo bar/x.lo

This will also ensure that make won't try to use a temporarily corrupted `x.o' to create a program or library. It may cause needless recompilation on platforms that support `-c' and `-o' together, but it's the only way to make it safe for those that don't.

Link mode

Link mode links together object files (including library objects) to form another library or to create an executable program.

mode-args consist of a command using the C compiler to create an output file (with the `-o' flag) from several object files.

The following components of mode-args are treated specially:

`-all-static'
If output-file is a program, then do not link it against any shared libraries at all. If output-file is a library, then only create a static library.
`-avoid-version'
Tries to avoid versioning (see section Library interface versions) for libraries and modules, i.e. no version information is stored and no symbolic links are created. If the platform requires versioning, this option has no effect.
`-dlopen file'
Same as `-dlpreopen file', if native dlopening is not supported on the host platform (see section Dlopened modules) or if the program is linked with `-static' or `-all-static'. Otherwise, no effect. If file is self libtool will make sure that the program can dlopen itself, either by enabling -export-dynamic or by falling back to `-dlpreopen self'.
`-dlpreopen file'
Link file into the output program, and add its symbols to lt_preloaded_symbols (see section Dlpreopening). If file is self, the symbols of the program itself will be added to lt_preloaded_symbols. If file is force libtool will make sure that lt_preloaded_symbols is always defined, regardless of whether it's empty or not.
`-export-dynamic'
Allow symbols from output-file to be resolved with dlsym (see section Dlopened modules).
`-export-symbols symfile'
Tells the linker to export only the symbols listed in symfile. The symbol file should end in `.sym' and must contain the name of one symbol per line. This option has no effect on some platforms. By default all symbols are exported.
`-export-symbols-regex regex'
Same as `-export-symbols', except that only symbols matching the regular expression regex are exported. By default all symbols are exported.
`-Llibdir'
Search libdir for required libraries that have already been installed.
`-lname'
output-file requires the installed library `libname'. This option is required even when output-file is not an executable.
`-module'
Creates a library that can be dlopened (see section Dlopened modules). This option doesn't work for programs. Module names don't need to be prefixed with 'lib'. In order to prevent name clashes, however, 'libname' and 'name' must not be used at the same time in your package.
`-no-fast-install'
Disable fast-install mode for the executable output-file. Useful if the program won't be necessarily installed.
`-no-install'
Link an executable output-file that can't be installed and therefore doesn't need a wrapper script. Useful if the program is only used in the build tree, e.g., for testing or generating other files.
`-no-undefined'
Declare that output-file does not depend on any other libraries. Some platforms cannot create shared libraries that depend on other libraries (see section Inter-library dependencies).
`-o output-file'
Create output-file from the specified objects and libraries.
`-release release'
Specify that the library was generated by release release of your package, so that users can easily tell which versions are newer than others. Be warned that no two releases of your package will be binary compatible if you use this flag. If you want binary compatibility, use the `-version-info' flag instead (see section Library interface versions).
`-rpath libdir'
If output-file is a library, it will eventually be installed in libdir. If output-file is a program, add libdir to the run-time path of the program.
`-R libdir'
If output-file is a program, add libdir to its run-time path. If output-file is a library, add -Rlibdir to its dependency_libs, so that, whenever the library is linked into a program, libdir will be added to its run-time path.
`-static'
If output-file is a program, then do not link it against any uninstalled shared libtool libraries. If output-file is a library, then only create a static library.
`-version-info current[:revision[:age]]'
If output-file is a libtool library, use interface version information current, revision, and age to build it (see section Library interface versions). Do not use this flag to specify package release information, rather see the `-release' flag.
`-Wl,flag'
`-Xlinker flag'
Pass a linker specific flag directly to the linker.

If the output-file ends in `.la', then a libtool library is created, which must be built only from library objects (`.lo' files). The `-rpath' option is required. In the current implementation, libtool libraries may not depend on other uninstalled libtool libraries (see section Inter-library dependencies).

If the output-file ends in `.a', then a standard library is created using ar and possibly ranlib.

If output-file ends in `.o' or `.lo', then a reloadable object file is created from the input files (generally using `ld -r'). This method is often called partial linking.

Otherwise, an executable program is created.

Execute mode

For execute mode, the library path is automatically set, then a program is executed.

The first of the mode-args is treated as a program name, with the rest as arguments to that program.

The following components of mode-args are treated specially:

`-dlopen file'
Add the directory containing file to the library path.

This mode sets the library path environment variable according to any `-dlopen' flags.

If any of the args are libtool executable wrappers, then they are translated into the name of their corresponding uninstalled binary, and any of their required library directories are added to the library path.

Install mode

In install mode, libtool interprets mode-args as an installation command beginning with cp, or a BSD-compatible install program.

The rest of the mode-args are interpreted as arguments to that command.

The command is run, and any necessary unprivileged post-installation commands are also completed.

Finish mode

Finish mode helps system administrators install libtool libraries so that they can be located and linked into user programs.

Each mode-arg is interpreted as the name of a library directory. Running this command may require superuser privileges, so the `--dry-run' option may be useful.

Uninstall mode

Uninstall mode deletes installed libraries, executables and objects.

The first mode-arg is the name of the program to use to delete files (typically `/bin/rm').

The remaining mode-args are either flags for the deletion program (beginning with a `-'), or the names of files to delete.

Clean mode

Clean mode deletes uninstalled libraries, executables, objects and libtool's temporary files associated with them.

The first mode-arg is the name of the program to use to delete files (typically `/bin/rm').

The remaining mode-args are either flags for the deletion program (beginning with a `-'), or the names of files to delete.

Integrating libtool with your package

This chapter describes how to integrate libtool with your packages so that your users can install hassle-free shared libraries.

Writing `Makefile' rules for libtool

Libtool is fully integrated with Automake (see section `Introduction' in The Automake Manual), starting with Automake version 1.2.

If you want to use libtool in a regular `Makefile' (or `Makefile.in'), you are on your own. If you're not using Automake 1.2, and you don't know how to incorporate libtool into your package you need to do one of the following:

  1. Download Automake (version 1.2 or later) from your nearest GNU mirror, install it, and start using it.
  2. Learn how to write `Makefile' rules by hand. They're sometimes complex, but if you're clever enough to write rules for compiling your old libraries, then you should be able to figure out new rules for libtool libraries (hint: examine the `Makefile.in' in the `demo' subdirectory of the libtool distribution... note especially that it was automatically generated from the `Makefile.am' by Automake).

Using Automake with libtool

Libtool library support is implemented under the `LTLIBRARIES' primary.

Here are some samples from the Automake `Makefile.am' in the libtool distribution's `demo' subdirectory.

First, to link a program against a libtool library, just use the `program_LDADD' variable:

bin_PROGRAMS = hell hell.debug

# Build hell from main.c and libhello.la
hell_SOURCES = main.c
hell_LDADD = libhello.la

# Create an easier-to-debug version of hell.
hell_debug_SOURCES = main.c
hell_debug_LDADD = libhello.la
hell_debug_LDFLAGS = -static

The flags `-dlopen' or `-dlpreopen' (see section Link mode) would fit better in the program_LDADD variable. Unfortunately, GNU automake, up to release 1.4, doesn't accept these flags in a program_LDADD variable, so you have the following alternatives:

You may use the `program_LDFLAGS' variable to stuff in any flags you want to pass to libtool while linking `program' (such as `-static' to avoid linking uninstalled shared libtool libraries).

Building a libtool library is almost as trivial... note the use of `libhello_la_LDFLAGS' to pass the `-version-info' (see section Library interface versions) option to libtool:

# Build a libtool library, libhello.la for installation in libdir.
lib_LTLIBRARIES = libhello.la
libhello_la_SOURCES = hello.c foo.c
libhello_la_LDFLAGS = -version-info 3:12:1

The `-rpath' option is passed automatically by Automake (except for libraries listed as noinst_LTLIBRARIES), so you should not specify it.

See section `The Automake Manual' in The Automake Manual, for more information.

Configuring libtool

Libtool requires intimate knowledge of your compiler suite and operating system in order to be able to create shared libraries and link against them properly. When you install the libtool distribution, a system-specific libtool script is installed into your binary directory.

However, when you distribute libtool with your own packages (see section Including libtool in your package), you do not always know which compiler suite and operating system are used to compile your package.

For this reason, libtool must be configured before it can be used. This idea should be familiar to anybody who has used a GNU configure script. configure runs a number of tests for system features, then generates the `Makefiles' (and possibly a `config.h' header file), after which you can run make and build the package.

Libtool adds its own tests to your configure script in order to generate a libtool script for the installer's host machine.

The AC_PROG_LIBTOOL macro

If you are using GNU Autoconf (or Automake), you should add a call to AC_PROG_LIBTOOL to your `configure.in' file. This macro adds many new tests to the configure script so that the generated libtool script will understand the characteristics of the host:

Macro: AC_PROG_LIBTOOL
Macro: AM_PROG_LIBTOOL
Add support for the `--enable-shared' and `--disable-shared' configure flags.(4) AM_PROG_LIBTOOL was the old name for this macro, and although supported at the moment is deprecated.

By default, this macro turns on shared libraries if they are available, and also enables static libraries if they don't conflict with the shared libraries. You can modify these defaults by calling either the AC_DISABLE_SHARED or AC_DISABLE_STATIC macros:

# Turn off shared libraries during beta-testing, since they
# make the build process take too long.
AC_DISABLE_SHARED
AC_PROG_LIBTOOL

The user may specify modified forms of the configure flags `--enable-shared' and `--enable-static' to choose whether shared or static libraries are built based on the name of the package. For example, to have shared `bfd' and `gdb' libraries built, but not shared `libg++', you can run all three configure scripts as follows:

trick$ ./configure --enable-shared=bfd,gdb

In general, specifying `--enable-shared=pkgs' is the same as configuring with `--enable-shared' every package named in the comma-separated pkgs list, and every other package with `--disable-shared'. The `--enable-static=pkgs' flag behaves similarly, but it uses `--enable-static' and `--disable-static'. The same applies to the `--enable-fast-install=pkgs' flag, which uses `--enable-fast-install' and `--disable-fast-install'.

The package name `default' matches any packages which have not set their name in the PACKAGE environment variable.

This macro also sets the shell variable LIBTOOL_DEPS, that you can use to automatically update the libtool script if it becomes out-of-date. In order to do that, add to your `configure.in':

AC_PROG_LIBTOOL
AC_SUBST(LIBTOOL_DEPS)

and, to `Makefile.in' or `Makefile.am':

LIBTOOL_DEPS = @LIBTOOL_DEPS@
libtool: $(LIBTOOL_DEPS)
        $(SHELL) ./config.status --recheck

If you are using GNU automake, you can omit the assignment, as automake will take care of it. You'll obviously have to create some dependency on `libtool'.

Macro: AC_LIBTOOL_DLOPEN
Enable checking for dlopen support. This macro should be used if the package makes use of the `-dlopen' and `-dlpreopen' flags, otherwise libtool will assume that the system does not support dlopening. The macro must be called before AC_PROG_LIBTOOL.

Macro: AC_LIBTOOL_WIN32_DLL
This macro should be used if the package has been ported to build clean dlls on win32 platforms. Usually this means that any library data items are exported with __declspec(dllexport) and imported with __declspec(dllimport). If this macro is not used, libtool will assume that the package libraries are not dll clean and will build only static libraries on win32 hosts.

This macro must be called before AC_PROG_LIBTOOL, and provision must be made to pass `-no-undefined' to libtool in link mode from the package Makefile. Naturally, if you pass `-no-undefined', you must ensure that all the library symbols really are defined at link time!

Macro: AC_DISABLE_FAST_INSTALL
Change the default behaviour for AC_PROG_LIBTOOL to disable optimization for fast installation. The user may still override this default, depending on platform support, by specifying `--enable-fast-install'.

Macro: AC_DISABLE_SHARED
Macro: AM_DISABLE_SHARED
Change the default behaviour for AC_PROG_LIBTOOL to disable shared libraries. The user may still override this default by specifying `--enable-shared'.

Macro: AC_DISABLE_STATIC
Macro: AM_DISABLE_STATIC
Change the default behaviour for AC_PROG_LIBTOOL to disable static libraries. The user may still override this default by specifying `--enable-static'.

The tests in AC_PROG_LIBTOOL also recognize the following environment variables:

Variable: CC
The C compiler that will be used by the generated libtool. If this is not set, AC_PROG_LIBTOOL will look for gcc or cc.

Variable: CFLAGS
Compiler flags used to generate standard object files. If this is not set, AC_PROG_LIBTOOL will not use any such flags. It affects only the way AC_PROG_LIBTOOL runs tests, not the produced libtool.

Variable: CPPFLAGS
C preprocessor flags. If this is not set, AC_PROG_LIBTOOL will not use any such flags. It affects only the way AC_PROG_LIBTOOL runs tests, not the produced libtool.

Variable: LD
The system linker to use (if the generated libtool requires one). If this is not set, AC_PROG_LIBTOOL will try to find out what is the linker used by CC.

Variable: LDFLAGS
The flags to be used by libtool when it links a program. If this is not set, AC_PROG_LIBTOOL will not use any such flags. It affects only the way AC_PROG_LIBTOOL runs tests, not the produced libtool.

Variable: LIBS
The libraries to be used by AC_PROG_LIBTOOL when it links a program. If this is not set, AC_PROG_LIBTOOL will not use any such flags. It affects only the way AC_PROG_LIBTOOL runs tests, not the produced libtool.

Variable: NM
Program to use rather than checking for nm.

Variable: RANLIB
Program to use rather than checking for ranlib.

Variable: LN_S
A command that creates a link of a program, a soft-link if possible, a hard-link otherwise. AC_PROG_LIBTOOL will check for a suitable program if this variable is not set.

Variable: DLLTOOL
Program to use rather than checking for dlltool. Only meaningful for Cygwin/MS-Windows.

Variable: OBJDUMP
Program to use rather than checking for objdump. Only meaningful for Cygwin/MS-Windows.

Variable: AS
Program to use rather than checking for as. Only used on Cygwin/MS-Windows at the moment.

When you invoke the libtoolize program (see section Invoking libtoolize), it will tell you where to find a definition of AC_PROG_LIBTOOL. If you use Automake, the aclocal program will automatically add AC_PROG_LIBTOOL support to your configure script.

Nevertheless, it is advisable to include a copy of `libtool.m4' in `acinclude.m4', so that, even if `aclocal.m4' and `configure' are rebuilt for any reason, the appropriate libtool macros will be used. The alternative is to hope the user will have a compatible version of `libtool.m4' installed and accessible for aclocal. This may lead to weird errors when versions don't match.

Including libtool in your package

In order to use libtool, you need to include the following files with your package:

`config.guess'
Attempt to guess a canonical system name.
`config.sub'
Canonical system name validation subroutine script.
`ltmain.sh'
A generic script implementing basic libtool functionality.

Note that the libtool script itself should not be included with your package. See section Configuring libtool.

You should use the libtoolize program, rather than manually copying these files into your package.

Invoking libtoolize

The libtoolize program provides a standard way to add libtool support to your package. In the future, it may implement better usage checking, or other features to make libtool even easier to use.

The libtoolize program has the following synopsis:

libtoolize [option]...

and accepts the following options:

`--automake'
Work silently, and assume that Automake libtool support is used. `libtoolize --automake' is used by Automake to add libtool files to your package, when AC_PROG_LIBTOOL appears in your `configure.in'.
`--copy'
`-c'
Copy files from the libtool data directory rather than creating symlinks.
`--debug'
Dump a trace of shell script execution to standard output. This produces a lot of output, so you may wish to pipe it to less (or more) or redirect to a file.
`--dry-run'
`-n'
Don't run any commands that modify the file system, just print them out.
`--force'
`-f'
Replace existing libtool files. By default, libtoolize won't overwrite existing files.
`--help'
Display a help message and exit.
`--ltdl'
Install libltdl in a subdirectory of your package.
`--ltdl-tar'
Add the file libltdl.tar.gz to your package.
`--version'
Print libtoolize version information and exit.

If libtoolize detects an explicit call to AC_CONFIG_AUX_DIR (see section `The Autoconf Manual' in The Autoconf Manual) in your `configure.in', it will put the files in the specified directory.

libtoolize displays hints for adding libtool support to your package, as well.

Autoconf `.o' macros

The Autoconf package comes with a few macros that run tests, then set a variable corresponding to the name of an object file. Sometimes it is necessary to use corresponding names for libtool objects.

Here are the names of variables that list libtool objects:

Variable: LTALLOCA
Substituted by AC_FUNC_ALLOCA (see section `The Autoconf Manual' in The Autoconf Manual). Is either empty, or contains `alloca.lo'.

Variable: LTLIBOBJS
Substituted by AC_REPLACE_FUNCS (see section `The Autoconf Manual' in The Autoconf Manual), and a few other functions.

Unfortunately, the stable release of Autoconf (2.13, at the time of this writing) does not have any way for libtool to provide support for these variables. So, if you depend on them, use the following code immediately before the call to AC_OUTPUT in your `configure.in':

LTLIBOBJS=`echo "$LIBOBJS" | sed 's/\.[^.]* /.lo /g;s/\.[^.]*$/.lo/'`
AC_SUBST(LTLIBOBJS)
LTALLOCA=`echo "$ALLOCA" | sed 's/\.[^.]* /.lo /g;s/\.[^.]*$/.lo/'`
AC_SUBST(LTALLOCA)
AC_OUTPUT(...)

Static-only libraries

When you are developing a package, it is often worthwhile to configure your package with the `--disable-shared' flag, or to override the defaults for AC_PROG_LIBTOOL by using the AC_DISABLE_SHARED Autoconf macro (see section The AC_PROG_LIBTOOL macro). This prevents libtool from building shared libraries, which has several advantages:

You may want to put a small note in your package `README' to let other developers know that `--disable-shared' can save them time. The following example note is taken from the GIMP(5) distribution `README':

The GIMP uses GNU Libtool in order to build shared libraries on a
variety of systems. While this is very nice for making usable
binaries, it can be a pain when trying to debug a program. For that
reason, compilation of shared libraries can be turned off by
specifying the `--disable-shared' option to `configure'.

Library interface versions

The most difficult issue introduced by shared libraries is that of creating and resolving runtime dependencies. Dependencies on programs and libraries are often described in terms of a single name, such as sed. So, one may say "libtool depends on sed," and that is good enough for most purposes.

However, when an interface changes regularly, we need to be more specific: "Gnus 5.1 requires Emacs 19.28 or above." Here, the description of an interface consists of a name, and a "version number."

Even that sort of description is not accurate enough for some purposes. What if Emacs 20 changes enough to break Gnus 5.1?

The same problem exists in shared libraries: we require a formal version system to describe the sorts of dependencies that programs have on shared libraries, so that the dynamic linker can guarantee that programs are linked only against libraries that provide the interface they require.

What are library interfaces?

Interfaces for libraries may be any of the following (and more):

Note that static functions do not count as interfaces, because they are not directly available to the user of the library.

Libtool's versioning system

Libtool has its own formal versioning system. It is not as flexible as some, but it is definitely the simplest of the more powerful versioning systems.

Think of a library as exporting several sets of interfaces, arbitrarily represented by integers. When a program is linked against a library, it may use any subset of those interfaces.

Libtool's description of the interfaces that a program uses is simple: it encodes the least and the greatest interface numbers in the resulting binary (first-interface, last-interface).

The dynamic linker is guaranteed that if a library supports every interface number between first-interface and last-interface, then the program can be relinked against that library.

Note that this can cause problems because libtool's compatibility requirements are actually stricter than is necessary.

Say `libhello' supports interfaces 5, 16, 17, 18, and 19, and that libtool is used to link `test' against `libhello'.

Libtool encodes the numbers 5 and 19 in `test', and the dynamic linker will only link `test' against libraries that support every interface between 5 and 19. So, the dynamic linker refuses to link `test' against `libhello'!

In order to eliminate this problem, libtool only allows libraries to declare consecutive interface numbers. So, `libhello' can declare at most that it supports interfaces 16 through 19. Then, the dynamic linker will link `test' against `libhello'.

So, libtool library versions are described by three integers:

current
The most recent interface number that this library implements.
revision
The implementation number of the current interface.
age
The difference between the newest and oldest interfaces that this library implements. In other words, the library implements all the interface numbers in the range from number current - age to current.

If two libraries have identical current and age numbers, then the dynamic linker chooses the library with the greater revision number.

Updating library version information

If you want to use libtool's versioning system, then you must specify the version information to libtool using the `-version-info' flag during link mode (see section Link mode).

This flag accepts an argument of the form `current[:revision[:age]]'. So, passing `-version-info 3:12:1' sets current to 3, revision to 12, and age to 1.

If either revision or age are omitted, they default to 0. Also note that age must be less than or equal to the current interface number.

Here are a set of rules to help you update your library version information:

  1. Start with version information of `0:0:0' for each libtool library.
  2. Update the version information only immediately before a public release of your software. More frequent updates are unnecessary, and only guarantee that the current interface number gets larger faster.
  3. If the library source code has changed at all since the last update, then increment revision (`c:r:a' becomes `c:r+1:a').
  4. If any interfaces have been added, removed, or changed since the last update, increment current, and set revision to 0.
  5. If any interfaces have been added since the last public release, then increment age.
  6. If any interfaces have been removed since the last public release, then set age to 0.

Never try to set the interface numbers so that they correspond to the release number of your package. This is an abuse that only fosters misunderstanding of the purpose of library versions. Instead, use the `-release' flag (see section Managing release information), but be warned that every release of your package will not be binary compatible with any other release.

Managing release information

Often, people want to encode the name of the package release into the shared library so that it is obvious to the user which package their programs are linked against. This convention is used especially on GNU/Linux:

trick$ ls /usr/lib/libbfd*
/usr/lib/libbfd.a	    /usr/lib/libbfd.so.2.7.0.2
/usr/lib/libbfd.so
trick$

On `trick', `/usr/lib/libbfd.so' is a symbolic link to `libbfd.so.2.7.0.2', which was distributed as a part of `binutils-2.7.0.2'.

Unfortunately, this convention conflicts directly with libtool's idea of library interface versions, because the library interface rarely changes at the same time that the release number does, and the library suffix is never the same across all platforms.

So, in order to accommodate both views, you can use the `-release' flag in order to set release information for libraries which you do not want to use `-version-info'. For the `libbfd' example, the next release which uses libtool should be built with `-release 2.9.0', which will produce the following files on GNU/Linux:

trick$ ls /usr/lib/libbfd*
/usr/lib/libbfd-2.9.0.so     /usr/lib/libbfd.a
/usr/lib/libbfd.so
trick$

In this case, `/usr/lib/libbfd.so' is a symbolic link to `libbfd-2.9.0.so'. This makes it obvious that the user is dealing with `binutils-2.9.0', without compromising libtool's idea of interface versions.

Note that this option causes a modification of the library name, so do not use it unless you want to break binary compatibility with any past library releases. In general, you should only use `-release' for package-internal libraries or for ones whose interfaces change very frequently.

Tips for interface design

Writing a good library interface takes a lot of practice and thorough understanding of the problem that the library is intended to solve.

If you design a good interface, it won't have to change often, you won't have to keep updating documentation, and users won't have to keep relearning how to use the library.

Here is a brief list of tips for library interface design, which may help you in your exploits:

Plan ahead
Try to make every interface truly minimal, so that you won't need to delete entry points very often.
Avoid interface changes
Some people love redesigning and changing entry points just for the heck of it (note: renaming a function is considered changing an entry point). Don't be one of those people. If you must redesign an interface, then try to leave compatibility functions behind so that users don't need to rewrite their existing code.
Use opaque data types
The fewer data type definitions a library user has access to, the better. If possible, design your functions to accept a generic pointer (which you can cast to an internal data type), and provide access functions rather than allowing the library user to directly manipulate the data. That way, you have the freedom to change the data structures without changing the interface. This is essentially the same thing as using abstract data types and inheritance in an object-oriented system.
Use header files
If you are careful to document each of your library's global functions and variables in header files, and include them in your library source files, then the compiler will let you know if you make any interface changes by accident (see section Writing C header files).
Use the static keyword (or equivalent) whenever possible
The fewer global functions your library has, the more flexibility you'll have in changing them. Static functions and variables may change forms as often as you like... your users cannot access them, so they aren't interface changes.

Writing C header files

Writing portable C header files can be difficult, since they may be read by different types of compilers:

C++ compilers
C++ compilers require that functions be declared with full prototypes, since C++ is more strongly typed than C. C functions and variables also need to be declared with the extern "C" directive, so that the names aren't mangled. See section Writing libraries for C++, for other issues relevant to using C++ with libtool.
ANSI C compilers
ANSI C compilers are not as strict as C++ compilers, but functions should be prototyped to avoid unnecessary warnings when the header file is #included.
non-ANSI C compilers
Non-ANSI compilers will report errors if functions are prototyped.

These complications mean that your library interface headers must use some C preprocessor magic in order to be usable by each of the above compilers.

`foo.h' in the `demo' subdirectory of the libtool distribution serves as an example for how to write a header file that can be safely installed in a system directory.

Here are the relevant portions of that file:

/* BEGIN_C_DECLS should be used at the beginning of your declarations,
   so that C++ compilers don't mangle their names.  Use END_C_DECLS at
   the end of C declarations. */
#undef BEGIN_C_DECLS
#undef END_C_DECLS
#ifdef __cplusplus
# define BEGIN_C_DECLS extern "C" {
# define END_C_DECLS }
#else
# define BEGIN_C_DECLS /* empty */
# define END_C_DECLS /* empty */
#endif

/* PARAMS is a macro used to wrap function prototypes, so that
   compilers that don't understand ANSI C prototypes still work,
   and ANSI C compilers can issue warnings about type mismatches. */
#undef PARAMS
#if defined (__STDC__) || defined (_AIX) \
        || (defined (__mips) && defined (_SYSTYPE_SVR4)) \
        || defined(WIN32) || defined(__cplusplus)
# define PARAMS(protos) protos
#else
# define PARAMS(protos) ()
#endif

These macros are used in `foo.h' as follows:

#ifndef FOO_H
#define FOO_H 1

/* The above macro definitions. */
#include "..."

BEGIN_C_DECLS

int foo PARAMS((void));
int hello PARAMS((void));

END_C_DECLS

#endif /* !FOO_H */

Note that the `#ifndef FOO_H' prevents the body of `foo.h' from being read more than once in a given compilation.

Also the only thing that must go outside the BEGIN_C_DECLS/END_C_DECLS pair are #include lines. Strictly speaking it is only C symbol names that need to be protected, but your header files will be more maintainable if you have a single pair of of these macros around the majority of the header contents.

You should use these definitions of PARAMS, BEGIN_C_DECLS, and END_C_DECLS into your own headers. Then, you may use them to create header files that are valid for C++, ANSI, and non-ANSI compilers(6).

Do not be naive about writing portable code. Following the tips given above will help you miss the most obvious problems, but there are definitely other subtle portability issues. You may need to cope with some of the following issues:

Inter-library dependencies

By definition, every shared library system provides a way for executables to depend on libraries, so that symbol resolution is deferred until runtime.

An inter-library dependency is one in which a library depends on other libraries. For example, if the libtool library `libhello' uses the cos function, then it has an inter-library dependency on `libm', the math library that implements cos.

Some shared library systems provide this feature in an internally-consistent way: these systems allow chains of dependencies of potentially infinite length.

However, most shared library systems are restricted in that they only allow a single level of dependencies. In these systems, programs may depend on shared libraries, but shared libraries may not depend on other shared libraries.

In any event, libtool provides a simple mechanism for you to declare inter-library dependencies: for every library `libname' that your own library depends on, simply add a corresponding -lname option to the link line when you create your library. To make an example of our `libhello' that depends on `libm':

burger$ libtool gcc -g -O -o libhello.la foo.lo hello.lo \
                -rpath /usr/local/lib -lm
burger$

When you link a program against `libhello', you don't need to specify the same `-l' options again: libtool will do that for you, in order to guarantee that all the required libraries are found. This restriction is only necessary to preserve compatibility with static library systems and simple dynamic library systems.

Some platforms, such as AIX, do not even allow you this flexibility. In order to build a shared library, it must be entirely self-contained (that is, have references only to symbols that are found in the `.lo' files or the specified `-l' libraries), and you need to specify the -no-undefined flag. By default, libtool builds only static libraries on these kinds of platforms.

The simple-minded inter-library dependency tracking code of libtool releases prior to 1.2 was disabled because it was not clear when it was possible to link one library with another, and complex failures would occur. A more complex implementation of this concept was re-introduced before release 1.3, but it has not been ported to all platforms that libtool supports. The default, conservative behavior is to avoid linking one library with another, introducing their inter-dependencies only when a program is linked with them.

Dlopened modules

It can sometimes be confusing to discuss dynamic linking, because the term is used to refer to two different concepts:

  1. Compiling and linking a program against a shared library, which is resolved automatically at run time by the dynamic linker. In this process, dynamic linking is transparent to the application.
  2. The application calling functions such as dlopen,(7) which load arbitrary, user-specified modules at runtime. This type of dynamic linking is explicitly controlled by the application.

To mitigate confusion, this manual refers to the second type of dynamic linking as dlopening a module.

The main benefit to dlopening object modules is the ability to access compiled object code to extend your program, rather than using an interpreted language. In fact, dlopen calls are frequently used in language interpreters to provide an efficient way to extend the language.

As of version 1.4.2, libtool provides support for dlopened modules. However, you should indicate that your package is willing to use such support, by using the macro `AC_LIBTOOL_DLOPEN' in `configure.in'. If this macro is not used (or it is used after `AC_PROG_LIBTOOL'), libtool will assume no dlopening mechanism is available, and will try to simulate it.

This chapter discusses how you as a dlopen application developer might use libtool to generate dlopen-accessible modules.

Building modules to dlopen

On some operating systems, a program symbol must be specially declared in order to be dynamically resolved with the dlsym (or equivalent) function.

Libtool provides the `-export-dynamic' and `-module' link flags (see section Link mode), which do this declaration. You need to use these flags if you are linking an application program that dlopens other modules or a libtool library that will also be dlopened.

For example, if we wanted to build a shared library, `libhello', that would later be dlopened by an application, we would add `-module' to the other link flags:

burger$ libtool gcc -module -o libhello.la foo.lo \
                hello.lo -rpath /usr/local/lib -lm
burger$

If symbols from your executable are needed to satisfy unresolved references in a library you want to dlopen you will have to use the flag `-export-dynamic'. You should use `-export-dynamic' while linking the executable that calls dlopen:

burger$ libtool gcc -export-dynamic -o hell-dlopener main.o
burger$

Dlpreopening

Libtool provides special support for dlopening libtool object and libtool library files, so that their symbols can be resolved even on platforms without any dlopen and dlsym functions.

Consider the following alternative ways of loading code into your program, in order of increasing "laziness":

  1. Linking against object files that become part of the program executable, whether or not they are referenced. If an object file cannot be found, then the linker refuses to create the executable.
  2. Declaring a static library to the linker, so that it is searched at link time in order to satisfy any undefined references in the above object files. If the static library cannot be found, then the linker refuses to link the executable.
  3. Declaring a shared library to the runtime linker, so that it is searched at runtime in order to satisfy any undefined references in the above files. If the shared library cannot be found, then the dynamic linker aborts the program before it runs.
  4. Dlopening a module, so that the application can resolve its own, dynamically-computed references. If there is an error opening the module, or the module is not found, then the application can recover without crashing.

Libtool emulates `-dlopen' on static platforms by linking objects into the program at compile time, and creating data structures that represent the program's symbol table.

In order to use this feature, you must declare the objects you want your application to dlopen by using the `-dlopen' or `-dlpreopen' flags when you link your program (see section Link mode).

Structure: struct lt_dlsymlist { const char *name; lt_ptr address; }
The name attribute is a null-terminated character string of the symbol name, such as "fprintf". The address attribute is a generic pointer to the appropriate object, such as &fprintf.

Variable: const lt_dlsymlist * lt_preloaded_symbols
An array of lt_symbol structures, representing all the preloaded symbols linked into the program. For each `-dlpreloaded' file there is an element with the name of the file and a address of 0, followed by all symbols exported from this file. For the executable itself the special name @PROGRAM@ is used. The last element has a name and address of 0.

Some compilers may allow identifiers which are not valid in ANSI C, such as dollar signs. Libtool only recognizes valid ANSI C symbols (an initial ASCII letter or underscore, followed by zero or more ASCII letters, digits, and underscores), so non-ANSI symbols will not appear in lt_preloaded_symbols.

Finding the correct name to dlopen

After a library has been linked with `-module', it can be dlopened. Unfortunately, because of the variation in library names, your package needs to determine the correct file to dlopen.

The most straightforward and flexible implementation is to determine the name at runtime, by finding the installed `.la' file, and searching it for the following lines:

# The name that we can dlopen.
dlname='dlname'

If dlname is empty, then the library cannot be dlopened. Otherwise, it gives the dlname of the library. So, if the library was installed as `/usr/local/lib/libhello.la', and the dlname was `libhello.so.3', then `/usr/local/lib/libhello.so.3' should be dlopened.

If your program uses this approach, then it should search the directories listed in the LD_LIBRARY_PATH(8) environment variable, as well as the directory where libraries will eventually be installed. Searching this variable (or equivalent) will guarantee that your program can find its dlopened modules, even before installation, provided you have linked them using libtool.

Unresolved dlopen issues

The following problems are not solved by using libtool's dlopen support:

Using libltdl

Libtool provides a small library, called `libltdl', that aims at hiding the various difficulties of dlopening libraries from programmers. It consists of a header-file and a small C source file that can be distributed with applications that need dlopening functionality. On some platforms, whose dynamic linkers are too limited for a simple implementation of `libltdl' services, it requires GNU DLD, or it will only emulate dynamic linking with libtool's dlpreopening mechanism.

libltdl supports currently the following dynamic linking mechanisms:

libltdl is licensed under the terms of the GNU Library General Public License, with the following exception:

As a special exception to the GNU Lesser General Public License, if you distribute this file as part of a program or library that is built using GNU libtool, you may include it under the same distribution terms that you use for the rest of that program.

How to use libltdl in your programs

The libltdl API is similar to the dlopen interface of Solaris and Linux, which is very simple but powerful.

To use libltdl in your program you have to include the header file `ltdl.h':

#include <ltdl.h>

The last release of libltdl used some symbols that violated the POSIX namespace conventions. These symbols are now deprecated, and have been replaced by those described here. If you have code that relies on the old deprecated symbol names, defining `LT_NON_POSIX_NAMESPACE' before you include `ltdl.h' provides conversion macros. Whichever set of symbols you use, the new api is not binary compatible with the last, so you will need to recompile your application in order to use this version of libltdl.

Note that libltdl is not threadsafe, i.e. a multithreaded application has to use a mutex for libltdl. It was reported that GNU/Linux's glibc 2.0's dlopen with `RTLD_LAZY' (which libltdl uses by default) is not thread-safe, but this problem is supposed to be fixed in glibc 2.1. On the other hand, `RTLD_NOW' was reported to introduce problems in multi-threaded applications on FreeBSD. Working around these problems is left as an exercise for the reader; contributions are certainly welcome.

The following types are defined in `ltdl.h':

Type: lt_ptr
lt_ptr is a generic pointer.

Type: lt_dlhandle
lt_dlhandle is a module "handle". Every lt_dlopened module has a handle associated with it.

Type: lt_dlsymlist
lt_dlsymlist is a symbol list for dlpreopened modules. This structure is described in see section Dlpreopening.

libltdl provides the following functions:

Function: int lt_dlinit (void)
Initialize libltdl. This function must be called before using libltdl and may be called several times. Return 0 on success, otherwise the number of errors.

Function: int lt_dlexit (void)
Shut down libltdl and close all modules. This function will only then shut down libltdl when it was called as many times as lt_dlinit has been successfully called. Return 0 on success, otherwise the number of errors.

Function: lt_dlhandle lt_dlopen (const char *filename)
Open the module with the file name filename and return a handle for it. lt_dlopen is able to open libtool dynamic modules, preloaded static modules, the program itself and native dynamic libraries.

Unresolved symbols in the module are resolved using its dependency libraries (not implemented yet) and previously dlopened modules. If the executable using this module was linked with the -export-dynamic flag, then the global symbols in the executable will also be used to resolve references in the module.

If filename is NULL and the program was linked with -export-dynamic or -dlopen self, lt_dlopen will return a handle for the program itself, which can be used to access its symbols.

If libltdl cannot find the library and the file name filename does not have a directory component it will additionally search in the following search paths for the module (in the order as follows):

  1. user-defined search path: This search path can be set by the program using the functions lt_dlsetsearchpath and lt_dladdsearchdir.
  2. libltdl's search path: This search path is the value of the environment variable LTDL_LIBRARY_PATH.
  3. system library search path: The system dependent library search path (e.g. on Linux it is LD_LIBRARY_PATH).

Each search path must be a colon-separated list of absolute directories, for example, "/usr/lib/mypkg:/lib/foo".

If the same module is loaded several times, the same handle is returned. If lt_dlopen fails for any reason, it returns NULL.

Function: lt_dlhandle lt_dlopenext (const char *filename)
The same as lt_dlopen, except that it tries to append different file name extensions to the file name. If the file with the file name filename cannot be found libltdl tries to append the following extensions:

  1. the libtool archive extension `.la'
  2. the extension used for native dynamic libraries on the host platform, e.g., `.so', `.sl', etc.

This lookup strategy was designed to allow programs that don't have knowledge about native dynamic libraries naming conventions to be able to dlopen such libraries as well as libtool modules transparently.

Function: int lt_dlclose (lt_dlhandle handle)
Decrement the reference count on the module handle. If it drops to zero and no other module depends on this module, then the module is unloaded. Return 0 on success.

Function: lt_ptr lt_dlsym (lt_dlhandle handle, const char *name)
Return the address in the module handle, where the symbol given by the null-terminated string name is loaded. If the symbol cannot be found, NULL is returned.

Function: {const char *}lt_dlerror (void)
Return a human readable string describing the most recent error that occurred from any of libltdl's functions. Return NULL if no errors have occurred since initialization or since it was last called.

Function: int lt_dlpreload (const lt_dlsymlist *preloaded)
Register the list of preloaded modules preloaded. If preloaded is NULL, then all previously registered symbol lists, except the list set by lt_dlpreload_default, are deleted. Return 0 on success.

Function: int lt_dlpreload_default (const lt_dlsymlist *preloaded)
Set the default list of preloaded modules to preloaded, which won't be deleted by lt_dlpreload. Note that this function does not require libltdl to be initialized using lt_dlinit and can be used in the program to register the default preloaded modules. Instead of calling this function directly, most programs will use the macro LTDL_SET_PRELOADED_SYMBOLS.

Return 0 on success.

Macro: LTDL_SET_PRELOADED_SYMBOLS()
Set the default list of preloaded symbols. Should be used in your program to initialize libltdl's list of preloaded modules.

#include <ltdl.h>

int main() {
  /* ... */
  LTDL_SET_PRELOADED_SYMBOLS();
  /* ... */
}

Function: int lt_dladdsearchdir (const char *search_dir)
Add the search directory search_dir to the user-defined library search path. Return 0 on success.

Function: int lt_dlsetsearchpath (const char *search_path)
Replace the current user-defined library search path with search_path, which must be a colon-separated list of absolute directories. Return 0 on success.

Function: {const char *}lt_dlgetsearchpath (void)
Return the current user-defined library search path.

Function: int lt_dlmakeresident (lt_dlhandle handle)
Mark a module so that it cannot be `lt_dlclose'd. This can be useful if a module implements some core functionality in your project, which would cause your code to crash if removed. Return 0 on success.

If you use `lt_dlopen (NULL)' to get a handle for the running binary, that handle will always be marked as resident, and consequently cannot be successfully `lt_dlclose'd.

Function: int lt_dlisresident (lt_dlhandle handle)
Check whether a particular module has been marked as resident, returning 1 if it has or 0 otherwise. If there is an error while executing this function, return -1 and set an error message for retrieval with lt_dlerror.

Variable: lt_ptr (*) (size_t size) lt_dlmalloc
Variable: void (*) (lt_ptr ptr) lt_dlfree
These variables are set to malloc and free, by default, but you can set them to any other functions that provides equivalent functionality. However, you must not modify their values after calling any libltdl function other than lt_dlpreopen_default or the macro LTDL_SET_PRELOADED_SYMBOLS.

Creating modules that can be dlopened

Libtool modules are like normal libtool libraries with a few exceptions:

You have to link the module with libtool's `-module' switch, and you should link any program that is intended to dlopen the module with `-dlopen modulename.la' so that libtool can dlpreopen the module on platforms which don't support dlopening. If the module depends on any other libraries, make sure you specify them either when you link the module or when you link programs that dlopen it. If you want to disable see section Library interface versions for a specific module you should link it with the `-avoid-version' switch. Note that libtool modules don't need to have a "lib" prefix. However, automake 1.4 or higher is required to build such modules.

Usually a set of modules provide the same interface, i.e, exports the same symbols, so that a program can dlopen them without having to know more about their internals. In order to avoid symbol conflicts all exported symbols must be prefixed with "modulename_LTX_" (`modulename' is the name of the module). Internal symbols must be named in such a way that they won't conflict with other modules, for example, by prefixing them with "_modulename_". Although some platforms support having the same symbols defined more than once it is generally not portable and it makes it impossible to dlpreopen such modules. libltdl will automatically cut the prefix off to get the real name of the symbol. Additionally, it supports modules which don't use a prefix so that you can also dlopen non-libtool modules.

`foo1.c' gives an example of a portable libtool module. Exported symbols are prefixed with "foo1_LTX_", internal symbols with "_foo1_". Aliases are defined at the beginning so that the code is more readable.

/* aliases for the exported symbols */
#define foo	foo1_LTX_foo
#define bar	foo1_LTX_bar

/* a global variable definition */
int bar = 1;

/* a private function */
int _foo1_helper() {
  return bar;
}

/* an exported function */
int foo() {
  return _foo1_helper();
}

The `Makefile.am' contains the necessary rules to build the module `foo1.la':

...
lib_LTLIBRARIES = foo1.la

foo1_la_SOURCES = foo1.c
foo1_la_LDFLAGS = -module
...

Using libtldl in a multi threaded environment

Using the lt_dlmutex_register() function, and by providing some appropriate callback function definitions, libltdl can be used in a multi-threaded environment.

Type: void lt_dlmutex_lock (void)
This is the type of a function pointer holding the address of a function which will be called at the start of parts of the libltdl implementation code which require a mutex lock.

Because libltdl is inherantly recursive, it is important that the locking mechanism employed by these callback functions are reentrant, or else strange problems will occur.

Type: void lt_dlmutex_unlock (void)
The type of a matching unlock function.

Type: void lt_dlmutex_seterror (const char *error);
Many of the functions in the libltdl API have a special return value to indicate to the client that an error has occured. Normally (in single threaded applications) a string describing that error can be retrieved from internal storage with lt_dlerror().

A function of this type must be registered with the library in order for it to work in a multi-threaded context. The function should store any error message passed in thread local storage.

Type: const char * lt_dlmutex_geterror (void)
The type of a matching callback function to retrieve the last stored error message from thread local storage.

When regeistered correctly this function will be used by lt_dlerror()) from all threads to retrieve error messages for the client.

Function: int lt_dlmutex_register (lt_dlmutex_lock *lock, lt_dlmutex_unlock *unlock, lt_dlmutex_set_error *seterror, lt_dlmutex_geterror *geterror)
Use this function to register one of each of function ttypes described above in preparation for multi-threaded use of libltdl. All arguments must be valid non-NULL function addresses, or else all NULL to return to single threaded operation.

Data associated with loaded modules

Some of the internal information about each loaded module that is maintained by libltdl is available to the user, in the form of this structure:

Type: struct lt_dlinfo { char *filename; char *name; int ref_count; }
lt_dlinfo is used to store information about a module. The filename attribute is a null-terminated character string of the real module file name. If the module is a libtool module then name is its module name (e.g. "libfoo" for "dir/libfoo.la"), otherwise it is set to NULL. The ref_count attribute is a reference counter that describes how often the same module is currently loaded.

The following function will return a pointer to libltdl's internal copy of this structure for the given handle:

Function: const lt_dlinfo * lt_dlgetinfo (lt_dlhandle handle)
Return a pointer to a struct that contains some information about the module handle. The contents of the struct must not be modified. Return NULL on failure.

Furthermore, in order to save you from having to keep a list of the handles of all the modules you have loaded, these functions allow you to iterate over libltdl's list of loaded modules:

Function: int lt_dlforeach (int (*func) (lt_dlhandle handle, lt_ptr data), lt_ptr data)
For each loaded module call the function func. The argument handle is the handle of one of the loaded modules, data is the data argument passed to lt_dlforeach. As soon as func returns a non-zero value for one of the handles, lt_dlforeach will stop calling func and immediately return 1. Otherwise 0 is returned.

Function: lt_dlhandle lt_dlhandle_next (lt_dlhandle place)
Iterate over the loaded module handles, returning the first handle in the list if place is NULL, and the next one on subsequent calls. If place is the last element in the list of loaded modules, this function returns NULL.

Of course, you would still need to maintain your own list of loaded module handles to parallel the list maintained by libltdl if there are any other data that you need to associate with each handle for the purposes of your application. However, if you use the following API calls to associate your application data with individual module handles as they are loaded there is actually no need to do that. You must first obtain a unique caller id from libltdl which you subsequently use to retrieve the data you stored earlier. This allows for different libraries that each wish to store their own data against loaded modules to do so without interfering with one another's data.

Type: lt_dlcaller_id
The opaque type used to hold individual data set keys.

Function: lt_dlcaller_id lt_dlcaller_register (void)
Use this to obtain a unique key to store and retrieve individual sets of per module data.

Function: lt_ptr lt_dlcaller_set_data (lt_dlcaller_id key, lt_dlhandle handle, lt_ptr data)
Set data as the set of data uniquely associated with key and handle for later retrieval. This function returns the data previously associated with key and handle if any. A result of 0, may indicate that a diagnostic for the last error (if any) is available from lt_dlerror().

For example, to correctly remove some associated data:

    lt_ptr stale = lt_dlcaller_set_data (key, handle, 0);
    if (stale == NULL)
      {
        char *error_msg = lt_dlerror ();

        if (error_msg != NULL)
          {
            my_error_handler (error_msg);
            return STATUS_FAILED;
          }
      }
    else
      {
        free (stale);
      }

Function: lt_ptr lt_dlcaller_get_data (lt_dlcaller_id key, lt_dlhandle handle)
Return the address of the data associated with key and handle, or else NULL if there is none.

The preceding functions can be combined with lt_dlforeach to implement search and apply operations without the need for your application to track the modules that have been loaded and unloaded:

int
my_dlcaller_callback (lt_dlhandle handle, lt_ptr key_ptr)
{
  struct my_module_data *my_data;

  my_data = lt_dlcaller_get_data (handle, (lt_dlcaller_id) *key_ptr);

  return process (my_data);
}

int
my_dlcaller_foreach (lt_dlcaller_id key)
{
  lt_dlforeach (my_dlcaller_callback, (lt_ptr) &key);
}

How to create and register new module loaders

Sometimes libltdl's many ways of gaining access to modules are not sufficient for the purposes of a project. You can write your own loader, and register it with libltdl so that lt_dlopen will be able to use it.

Writing a loader involves writing at least three functions which can be called by lt_dlopen, lt_dlsym and lt_dlclose. Optionally, you can provide a finalisation function to perform any cleanup operations when lt_dlexit executes, and a symbol prefix string which will be prepended to any symbols passed to lt_dlsym. These functions must match the function pointer types below, after which they can be allocated to an instance of lt_user_dlloader and registered.

Registering the loader requires that you choose a name for it, so that it can be recognised by lt_dlloader_find and removed with lt_dlloader_remove. The name you choose must be unique, and not already in use by libltdl's builtin loaders:

"dlopen"
The system dynamic library loader, if one exists.
"dld"
The GNU dld loader, if `libdld' was installed when libltdl was built.
"dlpreload"
The loader for lt_dlopening of preloaded static modules.

The prefix "dl" is reserved for loaders supplied with future versions of libltdl, so you should not use that for your own loader names.

The following types are defined in `ltdl.h':

Type: lt_module
lt_module is a dlloader dependent module. The dynamic module loader extensions communicate using these low level types.

Type: lt_dlloader
lt_dlloader is a handle for module loader types.

Type: lt_dlloader_data
lt_dlloader_data is used for specifying loader instance data.

Type: struct lt_user_dlloader {const char *sym_prefix; lt_module_open *module_open;lt_module_close *module_close; lt_find_sym *find_sym; lt_dlloader_exit *dlloader_exit; lt_dlloader_data dlloader_data; }
If you want to define a new way to open dynamic modules, and have the lt_dlopen API use it, you need to instantiate one of these structures and pass it to lt_dlloader_add. You can pass whatever you like in the dlloader_data field, and it will be passed back as the value of the first parameter to each of the functions specified in the function pointer fields.

Type: lt_module lt_module_open (lt_user_data loader_data, const char *filename)
The type of the loader function for an lt_dlloader module loader. The value set in the dlloader_data field of the struct lt_user_dlloader structure will be passed into this function in the loader_data parameter. Implementation of such a function should attempt to load the named module, and return an lt_module suitable for passing in to the associated lt_module_close and lt_sym_find function pointers. If the function fails it should return NULL, and set the error message with lt_dlseterror.

Type: int lt_module_close (lt_dlloader_data loader_data, lt_module module)
The type of the unloader function for a user defined module loader. Implementatation of such a function should attempt to release any resources tied up by the module module, and then unload it from memory. If the function fails for some reason, set the error message with lt_dlseterror and return non-zero.

Type: lt_ptr lt_find_sym (lt_user_data loader_data, lt_module module, const char *symbol)
The type of the symbol lookup function for a user defined module loader. Implementation of such a function should return the address of the named symbol in the module module, or else set the error message with lt_dlseterror and return NULL if lookup fails.

Type: int lt_dlloader_exit (lt_user_data loader_data)
The type of the finalisation function for a user defined module loader. Implementation of such a function should free any resources associated with the loader, including any user specified data in the dlloader_data field of the lt_user_dlloader. If non-NULL, the function will be called by lt_dlexit, and lt_dlloader_remove.

For example:

int
register_myloader (void)
{
  lt_user_dlloader dlloader;

  /* User modules are responsible for their own initialisation. */
  if (myloader_init () != 0)
    return MYLOADER_INIT_ERROR;

  dlloader.sym_prefix    = NULL;
  dlloader.module_open   = myloader_open;
  dlloader.module_close  = myloader_close;
  dlloader.find_sym      = myloader_find_sym.
  dlloader.dlloader_exit = myloader_exit;
  dlloader.dlloader_data = (lt_user_data)myloader_function;

  /* Add my loader as the default module loader. */
  if (lt_dlloader_add (lt_dlloader_next (NULL), &dlloader, "myloader") != 0)
    return ERROR;

  return OK;
}

Note that if there is any initialisation required for the loader, it must be performed manually before the loader is registered -- libltdl doesn't handle user loader initialisation.

Finalisation is handled by libltdl however, and it is important to ensure the dlloader_exit callback releases any resources claimed during the initialisation phase.

libltdl provides the following functions for writing your own module loaders:

Function: int lt_dlloader_add (lt_dlloader *place, lt_user_dlloader *dlloader, const char *loader_name)
Add a new module loader to the list of all loaders, either as the last loader (if place is NULL), else immediately before the loader passed as place. loader_name will be returned by lt_dlloader_name if it is subsequently passed a newly registered loader. These loader_names must be unique, or lt_dlloader_remove and lt_dlloader_find cannot work. Returns 0 for success.

{
  /* Make myloader be the last one. */
  if (lt_dlloader_add (NULL, myloader) != 0)
    perror (lt_dlerror ());
}

Function: int lt_dlloader_remove (const char *loader_name)
Remove the loader identified by the unique name, loader_name. Before this can succeed, all modules opened by the named loader must have been closed. Returns 0 for success, otherwise an error message can be obtained from lt_dlerror.

{
  /* Remove myloader. */
  if (lt_dlloader_remove ("myloader") != 0)
    perror (lt_dlerror ());
}

Function: {lt_dlloader *}lt_dlloader_next (lt_dlloader *place)
Iterate over the module loaders, returning the first loader if place is NULL, and the next one on subsequent calls. The handle is for use with lt_dlloader_add.

{
  /* Make myloader be the first one. */
  if (lt_dlloader_add (lt_dlloader_next (NULL), myloader) != 0)
    return ERROR;
}

Function: {lt_dlloader *}lt_dlloader_find (const char *loader_name)
Return the first loader with a matching loader_name identifier, or else NULL, if the identifier is not found.

The identifiers which may be used by libltdl itself, if the host architecture supports them are dlopen(9), dld and dlpreload.

{
  /* Add a user loader as the next module loader to be tried if
     the standard dlopen loader were to fail when lt_dlopening. */
  if (lt_dlloader_add (lt_dlloader_find ("dlopen"), myloader) != 0)