|
1 Status |
|
2 ====== |
|
3 |
|
4 libffi-3.0.5 was released on April 3, 2008. Check the libffi web |
|
5 page for updates: <URL:http://sourceware.org/libffi/>. |
|
6 |
|
7 |
|
8 What is libffi? |
|
9 =============== |
|
10 |
|
11 Compilers for high level languages generate code that follow certain |
|
12 conventions. These conventions are necessary, in part, for separate |
|
13 compilation to work. One such convention is the "calling convention". |
|
14 The "calling convention" is a set of assumptions made by the compiler |
|
15 about where function arguments will be found on entry to a function. |
|
16 A "calling convention" also specifies where the return value for a |
|
17 function is found. |
|
18 |
|
19 Some programs may not know at the time of compilation what arguments |
|
20 are to be passed to a function. For instance, an interpreter may be |
|
21 told at run-time about the number and types of arguments used to call |
|
22 a given function. Libffi can be used in such programs to provide a |
|
23 bridge from the interpreter program to compiled code. |
|
24 |
|
25 The libffi library provides a portable, high level programming |
|
26 interface to various calling conventions. This allows a programmer to |
|
27 call any function specified by a call interface description at run |
|
28 time. |
|
29 |
|
30 FFI stands for Foreign Function Interface. A foreign function |
|
31 interface is the popular name for the interface that allows code |
|
32 written in one language to call code written in another language. The |
|
33 libffi library really only provides the lowest, machine dependent |
|
34 layer of a fully featured foreign function interface. A layer must |
|
35 exist above libffi that handles type conversions for values passed |
|
36 between the two languages. |
|
37 |
|
38 |
|
39 Supported Platforms |
|
40 =================== |
|
41 |
|
42 Libffi has been ported to many different platforms, although this |
|
43 release was only tested on: |
|
44 |
|
45 arm oabi linux |
|
46 arm eabi linux |
|
47 hppa linux |
|
48 mips o32 linux (little endian) |
|
49 powerpc darwin |
|
50 powerpc64 linux |
|
51 sparc solaris |
|
52 sparc64 solaris |
|
53 x86 cygwin |
|
54 x86 darwin |
|
55 x86 freebsd |
|
56 x86 linux |
|
57 x86 openbsd |
|
58 x86-64 darwin |
|
59 x86-64 linux |
|
60 x86-64 OS X |
|
61 x86-64 freebsd |
|
62 |
|
63 Please send additional platform test results to |
|
64 libffi-discuss@sourceware.org. |
|
65 |
|
66 Installing libffi |
|
67 ================= |
|
68 |
|
69 [Note: before actually performing any of these installation steps, |
|
70 you may wish to read the "Platform Specific Notes" below.] |
|
71 |
|
72 First you must configure the distribution for your particular |
|
73 system. Go to the directory you wish to build libffi in and run the |
|
74 "configure" program found in the root directory of the libffi source |
|
75 distribution. |
|
76 |
|
77 You may want to tell configure where to install the libffi library and |
|
78 header files. To do that, use the --prefix configure switch. Libffi |
|
79 will install under /usr/local by default. |
|
80 |
|
81 If you want to enable extra run-time debugging checks use the the |
|
82 --enable-debug configure switch. This is useful when your program dies |
|
83 mysteriously while using libffi. |
|
84 |
|
85 Another useful configure switch is --enable-purify-safety. Using this |
|
86 will add some extra code which will suppress certain warnings when you |
|
87 are using Purify with libffi. Only use this switch when using |
|
88 Purify, as it will slow down the library. |
|
89 |
|
90 Configure has many other options. Use "configure --help" to see them all. |
|
91 |
|
92 Once configure has finished, type "make". Note that you must be using |
|
93 GNU make. You can ftp GNU make from prep.ai.mit.edu:/pub/gnu. |
|
94 |
|
95 To ensure that libffi is working as advertised, type "make check". |
|
96 This will require that you have DejaGNU installed. |
|
97 |
|
98 To install the library and header files, type "make install". |
|
99 |
|
100 |
|
101 Platform Specific Notes |
|
102 ======================= |
|
103 |
|
104 MIPS - Irix 5.3 & 6.x |
|
105 --------------------- |
|
106 |
|
107 Irix 6.2 and better supports three different calling conventions: o32, |
|
108 n32 and n64. Currently, libffi only supports both o32 and n32 under |
|
109 Irix 6.x, but only o32 under Irix 5.3. Libffi will automatically be |
|
110 configured for whichever calling convention it was built for. |
|
111 |
|
112 By default, the configure script will try to build libffi with the GNU |
|
113 development tools. To build libffi with the SGI development tools, set |
|
114 the environment variable CC to either "cc -32" or "cc -n32" before |
|
115 running configure under Irix 6.x (depending on whether you want an o32 |
|
116 or n32 library), or just "cc" for Irix 5.3. |
|
117 |
|
118 With the n32 calling convention, when returning structures smaller |
|
119 than 16 bytes, be sure to provide an RVALUE that is 8 byte aligned. |
|
120 Here's one way of forcing this: |
|
121 |
|
122 double struct_storage[2]; |
|
123 my_small_struct *s = (my_small_struct *) struct_storage; |
|
124 /* Use s for RVALUE */ |
|
125 |
|
126 If you don't do this you are liable to get spurious bus errors. |
|
127 |
|
128 "long long" values are not supported yet. |
|
129 |
|
130 You must use GNU Make to build libffi on SGI platforms. |
|
131 |
|
132 |
|
133 PowerPC System V ABI |
|
134 -------------------- |
|
135 |
|
136 There are two `System V ABI's which libffi implements for PowerPC. |
|
137 They differ only in how small structures are returned from functions. |
|
138 |
|
139 In the FFI_SYSV version, structures that are 8 bytes or smaller are |
|
140 returned in registers. This is what GCC does when it is configured |
|
141 for solaris, and is what the System V ABI I have (dated September |
|
142 1995) says. |
|
143 |
|
144 In the FFI_GCC_SYSV version, all structures are returned the same way: |
|
145 by passing a pointer as the first argument to the function. This is |
|
146 what GCC does when it is configured for linux or a generic sysv |
|
147 target. |
|
148 |
|
149 EGCS 1.0.1 (and probably other versions of EGCS/GCC) also has a |
|
150 inconsistency with the SysV ABI: When a procedure is called with many |
|
151 floating-point arguments, some of them get put on the stack. They are |
|
152 all supposed to be stored in double-precision format, even if they are |
|
153 only single-precision, but EGCS stores single-precision arguments as |
|
154 single-precision anyway. This causes one test to fail (the `many |
|
155 arguments' test). |
|
156 |
|
157 |
|
158 History |
|
159 ======= |
|
160 |
|
161 3.0.5 Apr-3-08 |
|
162 Fix libffi.pc file. |
|
163 Fix #define ARM for IcedTea users. |
|
164 Fix x86 closure bug. |
|
165 |
|
166 3.0.4 Feb-24-08 |
|
167 Fix x86 OpenBSD configury. |
|
168 |
|
169 3.0.3 Feb-22-08 |
|
170 Enable x86 OpenBSD thanks to Thomas Heller, and |
|
171 x86-64 FreeBSD thanks to Björn König and Andreas Tobler. |
|
172 Clean up test instruction in README. |
|
173 |
|
174 3.0.2 Feb-21-08 |
|
175 Improved x86 FreeBSD support. |
|
176 Thanks to Björn König. |
|
177 |
|
178 3.0.1 Feb-15-08 |
|
179 Fix instruction cache flushing bug on MIPS. |
|
180 Thanks to David Daney. |
|
181 |
|
182 3.0.0 Feb-15-08 |
|
183 Many changes, mostly thanks to the GCC project. |
|
184 Cygnus Solutions is now Red Hat. |
|
185 |
|
186 [10 years go by...] |
|
187 |
|
188 1.20 Oct-5-98 |
|
189 Raffaele Sena produces ARM port. |
|
190 |
|
191 1.19 Oct-5-98 |
|
192 Fixed x86 long double and long long return support. |
|
193 m68k bug fixes from Andreas Schwab. |
|
194 Patch for DU assembler compatibility for the Alpha from Richard |
|
195 Henderson. |
|
196 |
|
197 1.18 Apr-17-98 |
|
198 Bug fixes and MIPS configuration changes. |
|
199 |
|
200 1.17 Feb-24-98 |
|
201 Bug fixes and m68k port from Andreas Schwab. PowerPC port from |
|
202 Geoffrey Keating. Various bug x86, Sparc and MIPS bug fixes. |
|
203 |
|
204 1.16 Feb-11-98 |
|
205 Richard Henderson produces Alpha port. |
|
206 |
|
207 1.15 Dec-4-97 |
|
208 Fixed an n32 ABI bug. New libtool, auto* support. |
|
209 |
|
210 1.14 May-13-97 |
|
211 libtool is now used to generate shared and static libraries. |
|
212 Fixed a minor portability problem reported by Russ McManus |
|
213 <mcmanr@eq.gs.com>. |
|
214 |
|
215 1.13 Dec-2-96 |
|
216 Added --enable-purify-safety to keep Purify from complaining |
|
217 about certain low level code. |
|
218 Sparc fix for calling functions with < 6 args. |
|
219 Linux x86 a.out fix. |
|
220 |
|
221 1.12 Nov-22-96 |
|
222 Added missing ffi_type_void, needed for supporting void return |
|
223 types. Fixed test case for non MIPS machines. Cygnus Support |
|
224 is now Cygnus Solutions. |
|
225 |
|
226 1.11 Oct-30-96 |
|
227 Added notes about GNU make. |
|
228 |
|
229 1.10 Oct-29-96 |
|
230 Added configuration fix for non GNU compilers. |
|
231 |
|
232 1.09 Oct-29-96 |
|
233 Added --enable-debug configure switch. Clean-ups based on LCLint |
|
234 feedback. ffi_mips.h is always installed. Many configuration |
|
235 fixes. Fixed ffitest.c for sparc builds. |
|
236 |
|
237 1.08 Oct-15-96 |
|
238 Fixed n32 problem. Many clean-ups. |
|
239 |
|
240 1.07 Oct-14-96 |
|
241 Gordon Irlam rewrites v8.S again. Bug fixes. |
|
242 |
|
243 1.06 Oct-14-96 |
|
244 Gordon Irlam improved the sparc port. |
|
245 |
|
246 1.05 Oct-14-96 |
|
247 Interface changes based on feedback. |
|
248 |
|
249 1.04 Oct-11-96 |
|
250 Sparc port complete (modulo struct passing bug). |
|
251 |
|
252 1.03 Oct-10-96 |
|
253 Passing struct args, and returning struct values works for |
|
254 all architectures/calling conventions. Expanded tests. |
|
255 |
|
256 1.02 Oct-9-96 |
|
257 Added SGI n32 support. Fixed bugs in both o32 and Linux support. |
|
258 Added "make test". |
|
259 |
|
260 1.01 Oct-8-96 |
|
261 Fixed float passing bug in mips version. Restructured some |
|
262 of the code. Builds cleanly with SGI tools. |
|
263 |
|
264 1.00 Oct-7-96 |
|
265 First release. No public announcement. |
|
266 |
|
267 |
|
268 Authors & Credits |
|
269 ================= |
|
270 |
|
271 libffi was originally written by Anthony Green <green@redhat.com>. |
|
272 |
|
273 The developers of the GNU Compiler Collection project have made |
|
274 innumerable valuable contributions. See the ChangeLog file for |
|
275 details. |
|
276 |
|
277 Some of the ideas behind libffi were inspired by Gianni Mariani's free |
|
278 gencall library for Silicon Graphics machines. |
|
279 |
|
280 The closure mechanism was designed and implemented by Kresten Krab |
|
281 Thorup. |
|
282 |
|
283 Major processor architecture ports were contributed by the following |
|
284 developers: |
|
285 |
|
286 alpha Richard Henderson |
|
287 arm Raffaele Sena |
|
288 cris Simon Posnjak, Hans-Peter Nilsson |
|
289 frv Anthony Green |
|
290 ia64 Hans Boehm |
|
291 m32r Kazuhiro Inaoka |
|
292 m68k Andreas Schwab |
|
293 mips Anthony Green, Casey Marshall |
|
294 mips64 David Daney |
|
295 pa Randolph Chung, Dave Anglin, Andreas Tobler |
|
296 powerpc Geoffrey Keating, Andreas Tobler, |
|
297 David Edelsohn, John Hornkvist |
|
298 powerpc64 Jakub Jelinek |
|
299 s390 Gerhard Tonn, Ulrich Weigand |
|
300 sh Kaz Kojima |
|
301 sh64 Kaz Kojima |
|
302 sparc Anthony Green, Gordon Irlam |
|
303 x86 Anthony Green, Jon Beniston |
|
304 x86-64 Bo Thorsen |
|
305 |
|
306 Jesper Skov and Andrew Haley both did more than their fair share of |
|
307 stepping through the code and tracking down bugs. |
|
308 |
|
309 Thanks also to Tom Tromey for bug fixes, documentation and |
|
310 configuration help. |
|
311 |
|
312 Thanks to Jim Blandy, who provided some useful feedback on the libffi |
|
313 interface. |
|
314 |
|
315 Andreas Tobler has done a tremendous amount of work on the testsuite. |
|
316 |
|
317 Alex Oliva solved the executable page problem for SElinux. |
|
318 |
|
319 The list above is almost certainly incomplete and inaccurate. I'm |
|
320 happy to make corrections or additions upon request. |
|
321 |
|
322 If you have a problem, or have found a bug, please send a note to |
|
323 green@redhat.com. |