This manual is for Libffi, a portable foreign-function interface library.

   Copyright (C) 2008, 2010, 2011 Red Hat, Inc.

   Permission is granted to copy, distribute and/or modify this
   document under the terms of the GNU General Public License as
   published by the Free Software Foundation; either version 2, or
   (at your option) any later version.  A copy of the license is
   included in the section entitled "GNU General Public License".

• Menu:
  • Introduction:: What is libffi?
  • Using libffi:: How to use libffi.
  • Missing Features:: Things libffi can't do.
  • Index:: Index.

File: libffi.info, Node: Introduction, Next: Using libffi, Prev: Top, Up: Top

Compilers for high level languages generate code that follow certain conventions. These conventions are necessary, in part, for separate compilation to work. One such convention is the „calling convention“. The calling convention is a set of assumptions made by the compiler about where function arguments will be found on entry to a function. A calling convention also specifies where the return value for a function is found. The calling convention is also sometimes called the „ABI“ or „Application Binary Interface“.

Some programs may not know at the time of compilation what arguments are to be passed to a function. For instance, an interpreter may be told at run-time about the number and types of arguments used to call a given function. `Libffi' can be used in such programs to provide a bridge from the interpreter program to compiled code.

The `libffi' library provides a portable, high level programming interface to various calling conventions. This allows a programmer to call any function specified by a call interface description at run time.

FFI stands for Foreign Function Interface. A foreign function interface is the popular name for the interface that allows code written in one language to call code written in another language. The `libffi' library really only provides the lowest, machine dependent layer of a fully featured foreign function interface. A layer must exist above `libffi' that handles type conversions for values passed between the two languages. 

• Menu:

• The Basics:: The basic libffi API.
• Simple Example:: A simple example.
• Types:: libffi type descriptions.
• Multiple ABIs:: Different passing styles on one platform.
• The Closure API:: Writing a generic function.
• Closure Example:: A closure example.

File: libffi.info, Node: The Basics, Next: Simple Example, Up: Using libffi

`Libffi' assumes that you have a pointer to the function you wish to call and that you know the number and types of arguments to pass it, as well as the return type of the function.

 The first thing you must do is create an `ffi_cif' object that

matches the signature of the function you wish to call. This is a separate step because it is common to make multiple calls using a single `ffi_cif'. The „cif“ in `ffi_cif' stands for Call InterFace. To prepare a call interface object, use the function `ffi_prep_cif'.

This initializes CIF according to the given parameters.

• ABI is the ABI to use; normally `FFI_DEFAULT_ABI' is what you want. *note Multiple ABIs:: for more information.

• NARGS is the number of arguments that this function accepts.
• RTYPE is a pointer to an `ffi_type' structure that describes the return type of the function. *Note Types::.

• ARGTYPES is a vector of `ffi_type' pointers. ARGTYPES must have
• NARGS elements. If NARGS is 0, this argument is ignored.

`ffi_prep_cif' returns a `libffi' status code, of type `ffi_status'. This will be either `FFI_OK' if everything worked properly; `FFI_BAD_TYPEDEF' if one of the `ffi_type' objects is incorrect; or `FFI_BAD_ABI' if the ABI parameter is invalid.

If the function being called is variadic (varargs) then `ffi_prep_cif_var' must be used instead of `ffi_prep_cif'.

This initializes CIF according to the given parameters for a call to a variadic function. In general it's operation is the same as for `ffi_prep_cif' except that:

• NFIXEDARGS is the number of fixed arguments, prior to any variadic arguments. It must be greater than zero.

• NTOTALARGS the total number of arguments, including variadic and fixed arguments.

Note that, different cif's must be prepped for calls to the same function when different numbers of arguments are passed.

Also note that a call to `ffi_prep_cif_var' with NFIXEDARGS=NOTOTALARGS is NOT equivalent to a call to `ffi_prep_cif'.

To call a function using an initialized `ffi_cif', use the ffi_call' function:

This calls the function FN according to the description given in CIF. CIF must have already been prepared using `ffi_prep_cif'.

• RVALUE is a pointer to a chunk of memory that will hold the result of the function call. This must be large enough to hold the result and must be suitably aligned; it is the caller's responsibility to ensure this. If CIF declares that the function returns `void' (using `ffi_type_void'), then RVALUE is ignored. If RVALUE is `NULL', then the return value is discarded.

• AVALUES is a vector of `void *' pointers that point to the memory locations holding the argument values for a call. If CIF declares that the function has no arguments (i.e., NARGS was 0), then AVALUES is ignored. Note that argument values may be modified by the callee (for instance, structs passed by value); the burden of copying pass-by-value arguments is placed on the caller.

Here is a trivial example that calls `puts' a few times.

     #include <stdio.h>
     #include <ffi.h>
 
     int main()
     {
       ffi_cif cif;
       ffi_type *args[1];
       void *values[1];
       char *s;
       int rc;
 
       /* Initialize the argument info vectors */
       args[0] = &ffi_type_pointer;
       values[0] = &s;
 
       /* Initialize the cif */
       if (ffi_prep_cif(&cif, FFI_DEFAULT_ABI, 1,
     		       &ffi_type_uint, args) == FFI_OK)
         {
           s = "Hello World!";
           ffi_call(&cif, puts, &rc, values);
           /* rc now holds the result of the call to puts */
 
           /* values holds a pointer to the function's arg, so to
              call puts() again all we need to do is change the
              value of s */
           s = "This is cool!";
           ffi_call(&cif, puts, &rc, values);
         }
 
       return 0;
     }

• Menu:
  • Primitive Types:: Built-in types.
  • Structures:: Structure types.
  • Type Example:: Structure type example.

`Libffi' provides a number of built-in type descriptors that can be used to describe argument and return types:

• `ffi_type_void'
  The type `void'. This cannot be used for argument types, only for return values.
• `ffi_type_uint8'
  An unsigned, 8-bit integer type.
• `ffi_type_sint8'
  A signed, 8-bit integer type.
• `ffi_type_uint16'
  An unsigned, 16-bit integer type.
• `ffi_type_sint16'
  A signed, 16-bit integer type.
• `ffi_type_uint32'
  An unsigned, 32-bit integer type.
• `ffi_type_sint32'
  A signed, 32-bit integer type.
• `ffi_type_uint64'
  An unsigned, 64-bit integer type.
• `ffi_type_sint64'
  A signed, 64-bit integer type.
• `ffi_type_float'
  The C `float' type.
• `ffi_type_double'
  The C `double' type.
• `ffi_type_uchar'
  The C `unsigned char' type.
• `ffi_type_schar'
  The C `signed char' type. (Note that there is not an exact equivalent to the C `char' type in `libffi'; ordinarily you should either use `ffi_type_schar' or `ffi_type_uchar' depending on whether `char' is signed.)
• `ffi_type_ushort'
  The C `unsigned short' type.
• `ffi_type_sshort'
  The C `short' type.
• `ffi_type_uint'
  The C `unsigned int' type.
• `ffi_type_sint'
  The C `int' type.
• `ffi_type_ulong'
  The C `unsigned long' type.
• `ffi_type_slong'
  The C `long' type.
• `ffi_type_longdouble'
  On platforms that have a C `long double' type, this is defined. On other platforms, it is not.
• `ffi_type_pointer'
  A generic `void *' pointer. You should use this for all pointers, regardless of their real type.

Each of these is of type `ffi_type', so you must take the address when passing to `ffi_prep_cif'.

Although `libffi' has no special support for unions or bit-fields, it is perfectly happy passing structures back and forth. You must first describe the structure to `libffi' by creating a new `ffi_type' object for it.

The `ffi_type' has the following members:

* `size_t size' \\ This is set by `libffi'; you should initialize it to zero.
* `unsigned short alignment' \\ This is set by `libffi'; you should initialize it to zero.
* `unsigned short type' \\ For a structure, this should be set to `FFI_TYPE_STRUCT'.
* `ffi_type **elements' \\ This is a `NULL'-terminated array of pointers to `ffi_type' objects. There is one element per field of the struct.

The following example initializes a `ffi_type' object representing the `tm' struct from Linux's `time.h'.

Here is how the struct is defined:

     struct tm {
         int tm_sec;
         int tm_min;
         int tm_hour;
         int tm_mday;
         int tm_mon;
         int tm_year;
         int tm_wday;
         int tm_yday;
         int tm_isdst;
         /* Those are for future use. */
         long int __tm_gmtoff__;
         __const char *__tm_zone__;
     };

Here is the corresponding code to describe this struct to `libffi':

         {
           ffi_type tm_type;
           ffi_type *tm_type_elements[12];
           int i;
 
           tm_type.size = tm_type.alignment = 0;
           tm_type.elements = &tm_type_elements;
 
           for (i = 0; i < 9; i++)
               tm_type_elements[i] = &ffi_type_sint;
 
           tm_type_elements[9] = &ffi_type_slong;
           tm_type_elements[10] = &ffi_type_pointer;
           tm_type_elements[11] = NULL;
 
           /* tm_type can now be used to represent tm argument types and
     	 return types for ffi_prep_cif() */
         }

A given platform may provide multiple different ABIs at once. For instance, the x86 platform has both `stdcall' and `fastcall' functions.

`libffi' provides some support for this. However, this is necessarily platform-specific.

`libffi' also provides a way to write a generic function - a function that can accept and decode any combination of arguments. This can be useful when writing an interpreter, or to provide wrappers for arbitrary functions.

This facility is called the „closure API“. Closures are not supported on all platforms; you can check the `FFI_CLOSURES' define to determine whether they are supported on the current platform.

Because closures work by assembling a tiny function at runtime, they require special allocation on platforms that have a non-executable heap. Memory management for closures is handled by a pair of functions:

Allocate a chunk of memory holding SIZE bytes. This returns a pointer to the writable address, and sets *CODE to the corresponding executable address.

• SIZE should be sufficient to hold a `ffi_closure' object.

Free memory allocated using `ffi_closure_alloc'. The argument is the writable address that was returned.

Once you have allocated the memory for a closure, you must construct a `ffi_cif' describing the function call. Finally you can prepare the closure function:

Prepare a closure function.

• CLOSURE is the address of a `ffi_closure' object; this is the writable address returned by `ffi_closure_alloc'.
• CIF is the `ffi_cif' describing the function parameters.
• USER_DATA is an arbitrary datum that is passed, uninterpreted, to your closure function.
• CODELOC is the executable address returned by `ffi_closure_alloc'.
• FUN is the function which will be called when the closure is invoked. It is called with the arguments:
  • CIF The `ffi_cif' passed to `ffi_prep_closure_loc'.
  • RET A pointer to the memory used for the function's return value. FUN must fill this, unless the function is declared as returning `void'.
  • ARGS A vector of pointers to memory holding the arguments to the function.
  • USER_DATA The same USER_DATA that was passed to `ffi_prep_closure_loc'.

`ffi_prep_closure_loc' will return `FFI_OK' if everything went ok, and something else on error.

After calling `ffi_prep_closure_loc', you can cast CODELOC to the appropriate pointer-to-function type.

You may see old code referring to `ffi_prep_closure'. This function is deprecated, as it cannot handle the need for separate writable and executable addresses.

A trivial example that creates a new `puts' by binding `fputs' with `stdin'.

     #include <stdio.h>
     #include <ffi.h>
 
     /* Acts like puts with the file given at time of enclosure. */
     void puts_binding(ffi_cif *cif, unsigned int *ret, void* args[],
                       FILE *stream)
     {
       *ret = fputs(*(char **)args[0], stream);
     }
 
     int main()
     {
       ffi_cif cif;
       ffi_type *args[1];
       ffi_closure *closure;
 
       int (*bound_puts)(char *);
       int rc;
 
       /* Allocate closure and bound_puts */
       closure = ffi_closure_alloc(sizeof(ffi_closure), &bound_puts);
 
       if (closure)
         {
           /* Initialize the argument info vectors */
           args[0] = &ffi_type_pointer;
 
           /* Initialize the cif */
           if (ffi_prep_cif(&cif, FFI_DEFAULT_ABI, 1,
                            &ffi_type_uint, args) == FFI_OK)
             {
               /* Initialize the closure, setting stream to stdout */
               if (ffi_prep_closure_loc(closure, &cif, puts_binding,
                                        stdout, bound_puts) == FFI_OK)
                 {
                   rc = bound_puts("Hello World!");
                   /* rc now holds the result of the call to fputs */
                 }
             }
         }
 
       /* Deallocate both closure, and bound_puts */
       ffi_closure_free(closure);
 
       return 0;
     }

`libffi' is missing a few features. We welcome patches to add support for these.

• Variadic closures.
• There is no support for bit fields in structures.
• The closure API is
• The „raw“ API is undocumented.

Note that variadic support is very new and tested on a relatively small number of platforms.