b63463e2f83e04041621d7f229344606c3f1e197
[libffi.git] / libffi / ChangeLog
1 2008-12-18  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
2
3         PR libffi/26048
4         * configure.ac (HAVE_AS_X86_PCREL): New test.
5         * configure: Regenerate.
6         * fficonfig.h.in: Regenerate.
7         * src/x86/sysv.S [!FFI_NO_RAW_API]: Precalculate
8         RAW_CLOSURE_CIF_OFFSET, RAW_CLOSURE_FUN_OFFSET,
9         RAW_CLOSURE_USER_DATA_OFFSET for the Solaris 10/x86 assembler.
10         (.eh_frame): Only use SYMBOL-. iff HAVE_AS_X86_PCREL.
11         * src/x86/unix64.S (.Lstore_table): Move to .text section.
12         (.Lload_table): Likewise.
13         (.eh_frame): Only use SYMBOL-. iff HAVE_AS_X86_PCREL.
14
15 2008-01-24  David Edelsohn  <edelsohn@gnu.org>
16
17         * configure: Regenerate.
18
19 2008-01-06  Andreas Tobler  <a.tobler@schweiz.org>
20
21         * src/x86/ffi.c (ffi_prep_cif_machdep): Fix thinko.
22
23 2008-01-05  Andreas Tobler  <a.tobler@schweiz.org>
24
25         PR testsuite/32843
26         * src/x86/ffi.c (ffi_prep_cif_machdep): Add code for
27         signed/unsigned int8/16 for X86_DARWIN.
28         Updated copyright info.
29         Handle one and two byte structs with special cif->flags.
30         * src/x86/ffitarget.h: Add special types for one and two byte structs.
31         Updated copyright info.
32         * src/x86/darwin.S (ffi_call_SYSV): Rewrite to use a jump table like
33         sysv.S
34         Remove code to pop args from the stack after call.
35         Special-case signed/unsigned for int8/16, one and two byte structs.
36         (ffi_closure_raw_SYSV): Handle FFI_TYPE_UINT8,
37         FFI_TYPE_SINT8, FFI_TYPE_UINT16, FFI_TYPE_SINT16, FFI_TYPE_UINT32,
38         FFI_TYPE_SINT32.
39         Updated copyright info.
40
41 2007-12-08  David Daney  <ddaney@avtrex.com>
42
43         * src/mips/n32.S (ffi_call_N32):  Replace dadd with ADDU, dsub with 
44         SUBU, add with ADDU and use smaller code sequences.
45
46 2007-12-07  David Daney  <ddaney@avtrex.com>
47
48         * src/mips/ffi.c (ffi_prep_cif_machdep): Handle long double return
49         type.
50
51 2007-12-06  David Daney  <ddaney@avtrex.com>
52
53         * include/ffi.h.in (FFI_SIZEOF_JAVA_RAW): Define if not already
54         defined.
55         (ffi_java_raw): New typedef.
56         (ffi_java_raw_call, ffi_java_ptrarray_to_raw,
57         ffi_java_raw_to_ptrarray): Change parameter types from ffi_raw to
58         ffi_java_raw.
59         (ffi_java_raw_closure) : Same.
60         (ffi_prep_java_raw_closure, ffi_prep_java_raw_closure_loc): Change
61         parameter types.
62         * src/java_raw_api.c (ffi_java_raw_size):  Replace FFI_SIZEOF_ARG with
63         FFI_SIZEOF_JAVA_RAW.
64         (ffi_java_raw_to_ptrarray): Change type of raw to ffi_java_raw.
65         Replace FFI_SIZEOF_ARG with FFI_SIZEOF_JAVA_RAW. Use
66         sizeof(ffi_java_raw) for alignment calculations.
67         (ffi_java_ptrarray_to_raw): Same.
68         (ffi_java_rvalue_to_raw): Add special handling for FFI_TYPE_POINTER
69         if FFI_SIZEOF_JAVA_RAW == 4.
70         (ffi_java_raw_to_rvalue): Same.
71         (ffi_java_raw_call): Change type of raw to ffi_java_raw.
72         (ffi_java_translate_args): Same.
73         (ffi_prep_java_raw_closure_loc, ffi_prep_java_raw_closure): Change
74         parameter types.
75         * src/mips/ffitarget.h (FFI_SIZEOF_JAVA_RAW): Define for N32 ABI.
76
77 2007-12-06  David Daney  <ddaney@avtrex.com>
78
79         * src/mips/n32.S (ffi_closure_N32): Use 64-bit add instruction on 
80         pointer values.
81
82 2007-12-01  Andreas Tobler  <a.tobler@schweiz.org>
83
84         PR libffi/31937
85         * src/powerpc/ffitarget.h: Introduce new ABI FFI_LINUX_SOFT_FLOAT.
86         Add local FFI_TYPE_UINT128 to handle soft-float long-double-128.
87         * src/powerpc/ffi.c: Distinguish between __NO_FPRS__ and not and
88         set the NUM_FPR_ARG_REGISTERS according to.
89         Add support for potential soft-float support under hard-float
90         architecture.
91         (ffi_prep_args_SYSV): Set NUM_FPR_ARG_REGISTERS to 0 in case of
92         FFI_LINUX_SOFT_FLOAT, handle float, doubles and long-doubles according
93         to the FFI_LINUX_SOFT_FLOAT ABI.
94         (ffi_prep_cif_machdep): Likewise.
95         (ffi_closure_helper_SYSV): Likewise.
96         * src/powerpc/ppc_closure.S: Make sure not to store float/double
97         on archs where __NO_FPRS__ is true.
98         Add FFI_TYPE_UINT128 support.
99         * src/powerpc/sysv.S: Add support for soft-float long-double-128.
100         Adjust copyright notice.
101
102 2007-11-25  Andreas Tobler  <a.tobler@schweiz.org>
103
104         * src/closures.c: Move defintion of MAYBE_UNUSED from here to ...
105         * include/ffi_common.h: ... here.
106         Update copyright.
107
108 2007-11-17  Andreas Tobler  <a.tobler@schweiz.org>
109
110         * src/powerpc/sysv.S: Load correct cr to compare if we have long double.
111         * src/powerpc/linux64.S: Likewise.
112         * src/powerpc/ffi.c: Add a comment to show which part goes into cr6.
113         * testsuite/libffi.call/return_ldl.c: New test.
114
115 2007-09-04    <aph@redhat.com>
116
117         * src/arm/sysv.S (UNWIND): New.
118         (Whole file): Conditionally compile unwinder directives.
119         * src/arm/sysv.S: Add unwinder directives.
120
121         * src/arm/ffi.c (ffi_prep_args): Align structs by at least 4 bytes.
122         Only treat r0 as a struct address if we're actually returning a
123         struct by address.
124         Only copy the bytes that are actually within a struct.
125         (ffi_prep_cif_machdep): A Composite Type not larger than 4 bytes
126         is returned in r0, not passed by address.
127         (ffi_call): Allocate a word-sized temporary for the case where
128         a composite is returned in r0.
129         (ffi_prep_incoming_args_SYSV): Align as necessary.
130
131 2007-08-05  Steven Newbury  <s_j_newbury@yahoo.co.uk>
132
133         * src/arm/ffi.c (FFI_INIT_TRAMPOLINE): Use __clear_cache instead of 
134         directly using the sys_cacheflush syscall.
135
136 2007-07-27  Andrew Haley  <aph@redhat.com>
137
138         * src/arm/sysv.S (ffi_closure_SYSV): Add soft-float.
139
140 2007-09-03  Maciej W. Rozycki  <macro@linux-mips.org>
141
142         * Makefile.am: Unify MIPS_IRIX and MIPS_LINUX into MIPS.
143         * configure.ac: Likewise.
144         * Makefile.in: Regenerate.
145         * include/Makefile.in: Likewise.
146         * testsuite/Makefile.in: Likewise.
147         * configure: Likewise.
148
149 2007-08-24  David Daney  <ddaney@avtrex.com>
150
151         * testsuite/libffi.call/return_sl.c: New test.
152
153 2007-08-10  David Daney  <ddaney@avtrex.com>
154
155         * testsuite/libffi.call/cls_multi_ushort.c,
156         testsuite/libffi.call/cls_align_uint16.c,
157         testsuite/libffi.call/nested_struct1.c,
158         testsuite/libffi.call/nested_struct3.c,
159         testsuite/libffi.call/cls_7_1_byte.c,
160         testsuite/libffi.call/nested_struct5.c,
161         testsuite/libffi.call/cls_double.c,
162         testsuite/libffi.call/nested_struct7.c,
163         testsuite/libffi.call/cls_sint.c,
164         testsuite/libffi.call/nested_struct9.c,
165         testsuite/libffi.call/cls_20byte1.c,
166         testsuite/libffi.call/cls_multi_sshortchar.c,
167         testsuite/libffi.call/cls_align_sint64.c,
168         testsuite/libffi.call/cls_3byte2.c,
169         testsuite/libffi.call/cls_multi_schar.c,
170         testsuite/libffi.call/cls_multi_uchar.c,
171         testsuite/libffi.call/cls_19byte.c,
172         testsuite/libffi.call/cls_9byte1.c,
173         testsuite/libffi.call/cls_align_float.c,
174         testsuite/libffi.call/closure_fn1.c,
175         testsuite/libffi.call/problem1.c,
176         testsuite/libffi.call/closure_fn3.c,
177         testsuite/libffi.call/cls_sshort.c,
178         testsuite/libffi.call/closure_fn5.c,
179         testsuite/libffi.call/cls_align_double.c,
180         testsuite/libffi.call/nested_struct.c,
181         testsuite/libffi.call/cls_2byte.c,
182         testsuite/libffi.call/nested_struct10.c,
183         testsuite/libffi.call/cls_4byte.c,
184         testsuite/libffi.call/cls_6byte.c,
185         testsuite/libffi.call/cls_8byte.c,
186         testsuite/libffi.call/cls_multi_sshort.c,
187         testsuite/libffi.call/cls_align_sint16.c,
188         testsuite/libffi.call/cls_align_uint32.c,
189         testsuite/libffi.call/cls_20byte.c,
190         testsuite/libffi.call/cls_float.c,
191         testsuite/libffi.call/nested_struct2.c,
192         testsuite/libffi.call/cls_5_1_byte.c,
193         testsuite/libffi.call/nested_struct4.c,
194         testsuite/libffi.call/cls_24byte.c,
195         testsuite/libffi.call/nested_struct6.c,
196         testsuite/libffi.call/cls_64byte.c,
197         testsuite/libffi.call/nested_struct8.c,
198         testsuite/libffi.call/cls_uint.c,
199         testsuite/libffi.call/cls_multi_ushortchar.c,
200         testsuite/libffi.call/cls_schar.c,
201         testsuite/libffi.call/cls_uchar.c,
202         testsuite/libffi.call/cls_align_uint64.c,
203         testsuite/libffi.call/cls_ulonglong.c,
204         testsuite/libffi.call/cls_align_longdouble.c,
205         testsuite/libffi.call/cls_1_1byte.c,
206         testsuite/libffi.call/cls_12byte.c,
207         testsuite/libffi.call/cls_3_1byte.c,
208         testsuite/libffi.call/cls_3byte1.c,
209         testsuite/libffi.call/cls_4_1byte.c,
210         testsuite/libffi.call/cls_6_1_byte.c,
211         testsuite/libffi.call/cls_16byte.c,
212         testsuite/libffi.call/cls_18byte.c,
213         testsuite/libffi.call/closure_fn0.c,
214         testsuite/libffi.call/cls_9byte2.c,
215         testsuite/libffi.call/closure_fn2.c,
216         testsuite/libffi.call/closure_fn4.c,
217         testsuite/libffi.call/cls_ushort.c,
218         testsuite/libffi.call/closure_fn6.c,
219         testsuite/libffi.call/cls_5byte.c,
220         testsuite/libffi.call/cls_align_pointer.c,
221         testsuite/libffi.call/cls_7byte.c,
222         testsuite/libffi.call/cls_align_sint32.c,
223         testsuite/libffi.special/unwindtest_ffi_call.cc,
224         testsuite/libffi.special/unwindtest.cc: Remove xfail for mips64*-*-*.
225
226 2007-08-10  David Daney  <ddaney@avtrex.com>
227
228         PR libffi/28313
229         * configure.ac: Don't treat mips64 as a special case.
230         * Makefile.am (nodist_libffi_la_SOURCES): Add n32.S.
231         * configure: Regenerate
232         * Makefile.in: Ditto.
233         * fficonfig.h.in: Ditto.
234         * src/mips/ffitarget.h (REG_L, REG_S, SUBU, ADDU, SRL, LI): Indent.
235         (LA, EH_FRAME_ALIGN, FDE_ADDR_BYTES): New preprocessor macros.
236         (FFI_DEFAULT_ABI): Set for n64 case.
237         (FFI_CLOSURES, FFI_TRAMPOLINE_SIZE): Define for n32 and n64 cases.
238         * src/mips/n32.S (ffi_call_N32): Add debug macros and labels for FDE.
239         (ffi_closure_N32): New function.
240         (.eh_frame): New section
241         * src/mips/o32.S: Clean up comments.
242         (ffi_closure_O32): Pass ffi_closure parameter in $12.
243         * src/mips/ffi.c: Use FFI_MIPS_N32 instead of
244         _MIPS_SIM == _ABIN32 throughout.
245         (FFI_MIPS_STOP_HERE): New, use in place of
246         ffi_stop_here.
247         (ffi_prep_args): Use unsigned long to hold pointer values.  Rewrite
248         to support n32/n64 ABIs.
249         (calc_n32_struct_flags): Rewrite.
250         (calc_n32_return_struct_flags): Remove unused variable.  Reverse
251         position of flag bits.
252         (ffi_prep_cif_machdep): Rewrite n32 portion.
253         (ffi_call): Enable for n64.  Add special handling for small structure
254         return values.
255         (ffi_prep_closure_loc): Add n32 and n64 support.
256         (ffi_closure_mips_inner_O32): Add cast to silence warning.
257         (copy_struct_N32, ffi_closure_mips_inner_N32): New functions.
258
259 2007-08-08  David Daney  <ddaney@avtrex.com>
260
261         * testsuite/libffi.call/ffitest.h (ffi_type_mylong): Remove definition.
262         * testsuite/libffi.call/cls_align_uint16.c (main): Use correct type
263         specifiers.
264         * testsuite/libffi.call/nested_struct1.c (main): Ditto.
265         * testsuite/libffi.call/cls_sint.c (main): Ditto.
266         * testsuite/libffi.call/nested_struct9.c (main): Ditto.
267         * testsuite/libffi.call/cls_20byte1.c (main): Ditto.
268         * testsuite/libffi.call/cls_9byte1.c (main): Ditto.
269         * testsuite/libffi.call/closure_fn1.c (main): Ditto.
270         * testsuite/libffi.call/closure_fn3.c (main): Ditto.
271         * testsuite/libffi.call/return_dbl2.c (main): Ditto.
272         * testsuite/libffi.call/cls_sshort.c (main): Ditto.
273         * testsuite/libffi.call/return_fl3.c (main): Ditto.
274         * testsuite/libffi.call/closure_fn5.c (main): Ditto.
275         * testsuite/libffi.call/nested_struct.c (main): Ditto.
276         * testsuite/libffi.call/nested_struct10.c (main): Ditto.
277         * testsuite/libffi.call/return_ll1.c (main): Ditto.
278         * testsuite/libffi.call/cls_8byte.c (main): Ditto.
279         * testsuite/libffi.call/cls_align_uint32.c (main): Ditto.
280         * testsuite/libffi.call/cls_align_sint16.c (main): Ditto.
281         * testsuite/libffi.call/cls_20byte.c (main): Ditto.
282         * testsuite/libffi.call/nested_struct2.c (main): Ditto.
283         * testsuite/libffi.call/cls_24byte.c (main): Ditto.
284         * testsuite/libffi.call/nested_struct6.c (main): Ditto.
285         * testsuite/libffi.call/cls_uint.c (main): Ditto.
286         * testsuite/libffi.call/cls_12byte.c (main): Ditto.
287         * testsuite/libffi.call/cls_16byte.c (main): Ditto.
288         * testsuite/libffi.call/closure_fn0.c (main): Ditto.
289         * testsuite/libffi.call/cls_9byte2.c (main): Ditto.
290         * testsuite/libffi.call/closure_fn2.c (main): Ditto.
291         * testsuite/libffi.call/return_dbl1.c (main): Ditto.
292         * testsuite/libffi.call/closure_fn4.c (main): Ditto.
293         * testsuite/libffi.call/closure_fn6.c (main): Ditto.
294         * testsuite/libffi.call/cls_align_sint32.c (main): Ditto.
295
296 2007-08-07  Andrew Haley  <aph@redhat.com>
297
298         * src/x86/sysv.S (ffi_closure_raw_SYSV): Fix typo in previous
299         checkin.
300
301 2007-08-06  Andrew Haley  <aph@redhat.com>
302
303         PR testsuite/32843
304         * src/x86/sysv.S (ffi_closure_raw_SYSV): Handle FFI_TYPE_UINT8,
305         FFI_TYPE_SINT8, FFI_TYPE_UINT16, FFI_TYPE_SINT16, FFI_TYPE_UINT32,
306         FFI_TYPE_SINT32.
307
308 2007-08-02  David Daney  <ddaney@avtrex.com>
309
310         * testsuite/libffi.call/return_ul.c (main): Define return type as
311         ffi_arg.  Use proper printf conversion specifier.
312         
313 2007-07-30  Andrew Haley  <aph@redhat.com>
314
315         PR testsuite/32843
316         * src/x86/ffi.c (ffi_prep_cif_machdep): in x86 case, add code for
317         signed/unsigned int8/16.
318         * src/x86/sysv.S (ffi_call_SYSV): Rewrite to:
319         Use a jump table.
320         Remove code to pop args from the stack after call.
321         Special-case signed/unsigned int8/16.
322         * testsuite/libffi.call/return_sc.c (main): Revert.
323
324 2007-07-26  Richard Guenther  <rguenther@suse.de>
325
326         PR testsuite/32843
327         * testsuite/libffi.call/return_sc.c (main): Verify call
328         result as signed char, not ffi_arg.
329
330 2007-07-16  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
331
332         * configure.ac (i?86-*-solaris2.1[0-9]): Set TARGET to X86_64.
333         * configure: Regenerate.
334
335 2007-07-11  David Daney  <ddaney@avtrex.com>
336
337         * src/mips/ffi.c: Don't include sys/cachectl.h.
338         (ffi_prep_closure_loc): Use __builtin___clear_cache() instead of
339         cacheflush().
340
341 2007-05-18  Aurelien Jarno  <aurelien@aurel32.net>
342
343         * src/arm/ffi.c (ffi_prep_closure_loc): Renamed and ajusted
344         from (ffi_prep_closure): ... this.
345         (FFI_INIT_TRAMPOLINE): Adjust.
346
347 2005-12-31  Phil Blundell  <pb@reciva.com>
348
349         * src/arm/ffi.c (ffi_prep_incoming_args_SYSV, 
350         ffi_closure_SYSV_inner, ffi_prep_closure): New, add closure support.
351         * src/arm/sysv.S(ffi_closure_SYSV): Likewise.
352         * src/arm/ffitarget.h (FFI_TRAMPOLINE_SIZE): Likewise.
353         (FFI_CLOSURES): Enable closure support.
354
355 2007-07-03  Andrew Haley  <aph@hedges.billgatliff.com>
356
357         * testsuite/libffi.call/cls_multi_ushort.c,
358         testsuite/libffi.call/cls_align_uint16.c,
359         testsuite/libffi.call/nested_struct1.c,
360         testsuite/libffi.call/nested_struct3.c,
361         testsuite/libffi.call/cls_7_1_byte.c,
362         testsuite/libffi.call/cls_double.c,
363         testsuite/libffi.call/nested_struct5.c,
364         testsuite/libffi.call/nested_struct7.c,
365         testsuite/libffi.call/cls_sint.c,
366         testsuite/libffi.call/nested_struct9.c,
367         testsuite/libffi.call/cls_20byte1.c,
368         testsuite/libffi.call/cls_multi_sshortchar.c,
369         testsuite/libffi.call/cls_align_sint64.c,
370         testsuite/libffi.call/cls_3byte2.c,
371         testsuite/libffi.call/cls_multi_schar.c,
372         testsuite/libffi.call/cls_multi_uchar.c,
373         testsuite/libffi.call/cls_19byte.c,
374         testsuite/libffi.call/cls_9byte1.c,
375         testsuite/libffi.call/cls_align_float.c,
376         testsuite/libffi.call/closure_fn1.c,
377         testsuite/libffi.call/problem1.c,
378         testsuite/libffi.call/closure_fn3.c,
379         testsuite/libffi.call/cls_sshort.c,
380         testsuite/libffi.call/closure_fn5.c,
381         testsuite/libffi.call/cls_align_double.c,
382         testsuite/libffi.call/cls_2byte.c,
383         testsuite/libffi.call/nested_struct.c,
384         testsuite/libffi.call/nested_struct10.c,
385         testsuite/libffi.call/cls_4byte.c,
386         testsuite/libffi.call/cls_6byte.c,
387         testsuite/libffi.call/cls_8byte.c,
388         testsuite/libffi.call/cls_multi_sshort.c,
389         testsuite/libffi.call/cls_align_uint32.c,
390         testsuite/libffi.call/cls_align_sint16.c,
391         testsuite/libffi.call/cls_float.c,
392         testsuite/libffi.call/cls_20byte.c,
393         testsuite/libffi.call/cls_5_1_byte.c,
394         testsuite/libffi.call/nested_struct2.c,
395         testsuite/libffi.call/cls_24byte.c,
396         testsuite/libffi.call/nested_struct4.c,
397         testsuite/libffi.call/nested_struct6.c,
398         testsuite/libffi.call/cls_64byte.c,
399         testsuite/libffi.call/nested_struct8.c,
400         testsuite/libffi.call/cls_uint.c,
401         testsuite/libffi.call/cls_multi_ushortchar.c,
402         testsuite/libffi.call/cls_schar.c,
403         testsuite/libffi.call/cls_uchar.c,
404         testsuite/libffi.call/cls_align_uint64.c,
405         testsuite/libffi.call/cls_ulonglong.c,
406         testsuite/libffi.call/cls_align_longdouble.c,
407         testsuite/libffi.call/cls_1_1byte.c,
408         testsuite/libffi.call/cls_12byte.c,
409         testsuite/libffi.call/cls_3_1byte.c,
410         testsuite/libffi.call/cls_3byte1.c,
411         testsuite/libffi.call/cls_4_1byte.c,
412         testsuite/libffi.call/cls_6_1_byte.c,
413         testsuite/libffi.call/cls_16byte.c,
414         testsuite/libffi.call/cls_18byte.c,
415         testsuite/libffi.call/closure_fn0.c,
416         testsuite/libffi.call/cls_9byte2.c,
417         testsuite/libffi.call/closure_fn2.c,
418         testsuite/libffi.call/closure_fn4.c,
419         testsuite/libffi.call/cls_ushort.c,
420         testsuite/libffi.call/closure_fn6.c,
421         testsuite/libffi.call/cls_5byte.c,
422         testsuite/libffi.call/cls_align_pointer.c,
423         testsuite/libffi.call/cls_7byte.c,
424         testsuite/libffi.call/cls_align_sint32.c,
425         testsuite/libffi.special/unwindtest_ffi_call.cc,
426         testsuite/libffi.special/unwindtest.cc: Enable for ARM.
427
428 2007-07-05  H.J. Lu  <hongjiu.lu@intel.com>
429
430         * aclocal.m4: Regenerated.
431
432 2007-06-02  Paolo Bonzini  <bonzini@gnu.org>
433
434         * configure: Regenerate.
435
436 2007-05-23  Steve Ellcey  <sje@cup.hp.com>
437
438         * Makefile.in: Regenerate.
439         * configure: Regenerate.
440         * aclocal.m4: Regenerate.
441         * include/Makefile.in: Regenerate.
442         * testsuite/Makefile.in: Regenerate.
443
444 2007-05-10  Roman Zippel <zippel@linux-m68k.org>
445
446         * src/m68k/ffi.c (ffi_prep_incoming_args_SYSV,
447         ffi_closure_SYSV_inner,ffi_prep_closure): New, add closure support.
448         * src/m68k/sysv.S(ffi_closure_SYSV,ffi_closure_struct_SYSV): Likewise.
449         * src/m68k/ffitarget.h (FFI_TRAMPOLINE_SIZE): Likewise.
450         (FFI_CLOSURES): Enable closure support.
451
452 2007-05-10  Roman Zippel <zippel@linux-m68k.org>
453
454         * configure.ac (HAVE_AS_CFI_PSEUDO_OP): New test.
455         * configure: Regenerate.
456         * fficonfig.h.in: Regenerate.
457         * src/m68k/sysv.S (CFI_STARTPROC,CFI_ENDPROC,
458         CFI_OFFSET,CFI_DEF_CFA): New macros.
459         (ffi_call_SYSV): Add callframe annotation.
460
461 2007-05-10  Roman Zippel <zippel@linux-m68k.org>
462
463         * src/m68k/ffi.c (ffi_prep_args,ffi_prep_cif_machdep): Fix
464         numerous test suite failures.
465         * src/m68k/sysv.S (ffi_call_SYSV): Likewise.
466
467 2007-04-11  Paolo Bonzini  <bonzini@gnu.org>
468
469         * Makefile.am (EXTRA_DIST): Bring up to date.
470         * Makefile.in: Regenerate.
471         * src/frv/eabi.S: Remove RCS keyword.
472
473 2007-04-06  Richard Henderson  <rth@redhat.com>
474
475         * configure.ac: Tidy target case.
476         (HAVE_LONG_DOUBLE): Allow the target to override.
477         * configure: Regenerate.
478         * include/ffi.h.in: Don't define ffi_type_foo if
479         LIBFFI_HIDE_BASIC_TYPES is defined.
480         (ffi_type_longdouble): If not HAVE_LONG_DOUBLE, define
481         to ffi_type_double.
482         * types.c (LIBFFI_HIDE_BASIC_TYPES): Define.
483         (FFI_TYPEDEF, ffi_type_void): Mark the data const.
484         (ffi_type_longdouble): Special case for Alpha.  Don't define
485         if long double == double.
486
487         * src/alpha/ffi.c (FFI_TYPE_LONGDOUBLE): Assert unique value.
488         (ffi_prep_cif_machdep): Handle it as the 128-bit type.
489         (ffi_call, ffi_closure_osf_inner): Likewise.
490         (ffi_closure_osf_inner): Likewise.  Mark hidden.
491         (ffi_call_osf, ffi_closure_osf): Mark hidden.
492         * src/alpha/ffitarget.h (FFI_LAST_ABI): Tidy definition.
493         * src/alpha/osf.S (ffi_call_osf, ffi_closure_osf): Mark hidden.
494         (load_table): Handle 128-bit long double.
495
496         * testsuite/libffi.call/float4.c: Add -mieee for alpha.
497
498 2007-04-06  Tom Tromey  <tromey@redhat.com>
499
500         PR libffi/31491:
501         * README: Fixed bug in example.
502
503 2007-04-03  Jakub Jelinek  <jakub@redhat.com>
504
505         * src/closures.c: Include sys/statfs.h.
506         (_GNU_SOURCE): Define on Linux.
507         (FFI_MMAP_EXEC_SELINUX): Define.
508         (selinux_enabled): New variable.
509         (selinux_enabled_check): New function.
510         (is_selinux_enabled): Define.
511         (dlmmap): Use it.
512
513 2007-03-24  Uros Bizjak  <ubizjak@gmail.com>
514
515         * testsuite/libffi.call/return_fl2.c (return_fl): Mark as static.
516         Use 'volatile float sum' to create sum of floats to avoid false
517         negative due to excess precision on ix86 targets.
518         (main): Ditto. 
519
520 2007-03-08  Alexandre Oliva  <aoliva@redhat.com>
521
522         * src/powerpc/ffi.c (flush_icache): Fix left-over from previous
523         patch.
524         (ffi_prep_closure_loc): Remove unneeded casts.  Add needed ones.
525
526 2007-03-07  Alexandre Oliva  <aoliva@redhat.com>
527
528         * include/ffi.h.in (ffi_closure_alloc, ffi_closure_free): New.
529         (ffi_prep_closure_loc): New.
530         (ffi_prep_raw_closure_loc): New.
531         (ffi_prep_java_raw_closure_loc): New.
532         * src/closures.c: New file.
533         * src/dlmalloc.c [FFI_MMAP_EXEC_WRIT] (struct malloc_segment):
534         Replace sflags with exec_offset.
535         [FFI_MMAP_EXEC_WRIT] (mmap_exec_offset, add_segment_exec_offset,
536         sub_segment_exec_offset): New macros.
537         (get_segment_flags, set_segment_flags, check_segment_merge): New
538         macros.
539         (is_mmapped_segment, is_extern_segment): Use get_segment_flags.
540         (add_segment, sys_alloc, create_mspace, create_mspace_with_base,
541         destroy_mspace): Use new macros.
542         (sys_alloc): Silence warning.
543         * Makefile.am (libffi_la_SOURCES): Add src/closures.c.
544         * Makefile.in: Rebuilt.
545         * src/prep_cif [FFI_CLOSURES] (ffi_prep_closure): Implement in
546         terms of ffi_prep_closure_loc.
547         * src/raw_api.c (ffi_prep_raw_closure_loc): Renamed and adjusted
548         from...
549         (ffi_prep_raw_closure): ... this.  Re-implement in terms of the
550         renamed version.
551         * src/java_raw_api (ffi_prep_java_raw_closure_loc): Renamed and
552         adjusted from...
553         (ffi_prep_java_raw_closure): ... this.  Re-implement in terms of
554         the renamed version.
555         * src/alpha/ffi.c (ffi_prep_closure_loc): Renamed from
556         (ffi_prep_closure): ... this.
557         * src/pa/ffi.c: Likewise.
558         * src/cris/ffi.c: Likewise.  Adjust.
559         * src/frv/ffi.c: Likewise.
560         * src/ia64/ffi.c: Likewise.
561         * src/mips/ffi.c: Likewise.
562         * src/powerpc/ffi_darwin.c: Likewise.
563         * src/s390/ffi.c: Likewise.
564         * src/sh/ffi.c: Likewise.
565         * src/sh64/ffi.c: Likewise.
566         * src/sparc/ffi.c: Likewise.
567         * src/x86/ffi64.c: Likewise.
568         * src/x86/ffi.c: Likewise.
569         (FFI_INIT_TRAMPOLINE): Adjust.
570         (ffi_prep_raw_closure_loc): Renamed and adjusted from...
571         (ffi_prep_raw_closure): ... this.
572         * src/powerpc/ffi.c (ffi_prep_closure_loc): Renamed from
573         (ffi_prep_closure): ... this.
574         (flush_icache): Adjust.
575
576 2007-03-07  Alexandre Oliva  <aoliva@redhat.com>
577
578         * src/dlmalloc.c: New file, imported version 2.8.3 of Doug
579         Lea's malloc.
580
581 2007-03-01  Brooks Moses  <brooks.moses@codesourcery.com>
582
583         * Makefile.am: Add dummy install-pdf target.
584         * Makefile.in: Regenerate
585
586 2007-02-13  Andreas Krebbel  <krebbel1@de.ibm.com>
587
588         * src/s390/ffi.c (ffi_prep_args, ffi_prep_cif_machdep,
589         ffi_closure_helper_SYSV): Add long double handling.
590
591 2007-02-02  Jakub Jelinek  <jakub@redhat.com>
592
593         * src/powerpc/linux64.S (ffi_call_LINUX64): Move restore of r2
594         immediately after bctrl instruction.
595
596 2007-01-18  Alexandre Oliva  <aoliva@redhat.com>
597
598         * Makefile.am (all-recursive, install-recursive,
599         mostlyclean-recursive, clean-recursive, distclean-recursive,
600         maintainer-clean-recursive): Add missing targets.
601         * Makefile.in: Rebuilt.
602
603 2006-12-14  Andreas Tobler  <a.tobler@schweiz.org>
604
605         * configure.ac: Add TARGET for x86_64-*-darwin*.
606         * Makefile.am (nodist_libffi_la_SOURCES): Add rules for 64-bit sources
607         for X86_DARWIN.
608         * src/x86/ffitarget.h: Set trampoline size for x86_64-*-darwin*.
609         * src/x86/darwin64.S: New file for x86_64-*-darwin* support.
610         * configure: Regenerate.
611         * Makefile.in: Regenerate.
612         * include/Makefile.in: Regenerate.
613         * testsuite/Makefile.in: Regenerate.
614         * testsuite/libffi.special/unwindtest_ffi_call.cc: New test case for
615         ffi_call only.
616
617 2006-12-13  Andreas Tobler <a.tobler@schweiz.org>
618
619         * aclocal.m4: Regenerate with aclocal -I .. as written in the
620         Makefile.am.
621
622 2006-10-31  Geoffrey Keating  <geoffk@apple.com>
623
624         * src/powerpc/ffi_darwin.c (darwin_adjust_aggregate_sizes): New.
625         (ffi_prep_cif_machdep): Call darwin_adjust_aggregate_sizes for
626         Darwin.
627         * testsuite/libffi.call/nested_struct4.c: Remove Darwin XFAIL.
628         * testsuite/libffi.call/nested_struct6.c: Remove Darwin XFAIL.
629
630 2006-10-10  Paolo Bonzini  <bonzini@gnu.org>
631             Sandro Tolaini  <tolaini@libero.it>
632
633         * configure.ac [i*86-*-darwin*]: Set X86_DARWIN symbol and 
634         conditional.
635         * configure: Regenerated.
636         * Makefile.am (nodist_libffi_la_SOURCES) [X86_DARWIN]: New case.
637         (EXTRA_DIST): Add src/x86/darwin.S.
638         * Makefile.in: Regenerated.
639         * include/Makefile.in: Regenerated.
640         * testsuite/Makefile.in: Regenerated.
641
642         * src/x86/ffi.c (ffi_prep_cif_machdep) [X86_DARWIN]: Treat like
643         X86_WIN32, and additionally align stack to 16 bytes.
644         * src/x86/darwin.S: New, based on sysv.S.
645         * src/prep_cif.c (ffi_prep_cif) [X86_DARWIN]: Align > 8-byte structs.
646
647 2006-09-12  David Daney  <ddaney@avtrex.com>
648
649         PR libffi/23935
650         * include/Makefile.am: Install both ffi.h and ffitarget.h in
651         $(libdir)/gcc/$(target_alias)/$(gcc_version)/include.
652         * aclocal.m4: Regenerated for automake 1.9.6.
653         * Makefile.in: Regenerated.
654         * include/Makefile.in: Regenerated.
655         * testsuite/Makefile.in: Regenerated.
656
657 2006-08-17  Andreas Tobler  <a.tobler@schweiz.ch>
658
659         * include/ffi_common.h (struct): Revert accidental commit.
660
661 2006-08-15  Andreas Tobler  <a.tobler@schweiz.ch>
662
663         * include/ffi_common.h: Remove lint directives.
664         * include/ffi.h.in: Likewise.
665
666 2006-07-25  Torsten Schoenfeld  <kaffeetisch@gmx.de>
667
668         * include/ffi.h.in (ffi_type_ulong, ffi_type_slong): Define correctly
669         for 32-bit architectures.
670         * testsuite/libffi.call/return_ul.c: New test case.
671
672 2006-07-19  David Daney  <ddaney@avtrex.com>
673
674         * testsuite/libffi.call/closure_fn6.c: Remove xfail for mips,
675         xfail remains for mips64.
676
677 2006-05-23  Carlos O'Donell  <carlos@codesourcery.com>
678
679         * Makefile.am: Add install-html target. Add install-html to .PHONY
680         * Makefile.in: Regenerate.
681         * aclocal.m4: Regenerate.
682         * include/Makefile.in: Regenerate.
683         * testsuite/Makefile.in: Regenerate.
684
685 2006-05-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
686
687         * pa/ffi.c (ffi_prep_args_pa32): Load floating point arguments from
688         stack slot.
689
690 2006-04-22  Andreas Tobler  <a.tobler@schweiz.ch>
691
692         * README: Remove notice about 'Crazy Comments'.
693         * src/debug.c: Remove lint directives. Cleanup white spaces.
694         * src/java_raw_api.c: Likewise.
695         * src/prep_cif.c: Likewise.
696         * src/raw_api.c: Likewise.
697         * src/ffitest.c: Delete. No longer needed, all test cases migrated
698         to the testsuite.
699         * src/arm/ffi.c: Remove lint directives.
700         * src/m32r/ffi.c: Likewise.
701         * src/pa/ffi.c: Likewise.
702         * src/powerpc/ffi.c: Likewise.
703         * src/powerpc/ffi_darwin.c: Likewise.
704         * src/sh/ffi.c: Likewise.
705         * src/sh64/ffi.c: Likewise.
706         * src/x86/ffi.c: Likewise.
707         * testsuite/libffi.call/float2.c: Likewise.
708         * testsuite/libffi.call/promotion.c: Likewise.
709         * testsuite/libffi.call/struct1.c: Likewise.
710
711 2006-04-13  Andreas Tobler  <a.tobler@schweiz.ch>
712
713         * src/pa/hpux32.S: Correct unwind offset calculation for
714         ffi_closure_pa32.
715         * src/pa/linux.S: Likewise.
716
717 2006-04-12  James E Wilson  <wilson@specifix.com>
718
719         PR libgcj/26483
720         * src/ia64/ffi.c (stf_spill, ldf_fill): Rewrite as macros.
721         (hfa_type_load): Call stf_spill.
722         (hfa_type_store): Call ldf_fill.
723         (ffi_call): Adjust calls to above routines.  Add local temps for
724         macro result.
725
726 2006-04-10  Matthias Klose  <doko@debian.org>
727
728         * testsuite/lib/libffi-dg.exp (libffi-init): Recognize multilib
729         directory names containing underscores.
730
731 2006-04-07  James E Wilson  <wilson@specifix.com>
732
733         * testsuite/libffi.call/float4.c: New testcase.
734
735 2006-04-05  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
736             Andreas Tobler  <a.tobler@schweiz.ch>
737
738         * Makefile.am: Add PA_HPUX port.
739         * Makefile.in: Regenerate.
740         * include/Makefile.in: Likewise.
741         * testsuite/Makefile.in: Likewise.
742         * configure.ac: Add PA_HPUX rules.
743         * configure: Regenerate.
744         * src/pa/ffitarget.h: Rename linux target to PA_LINUX.
745         Add PA_HPUX and PA64_HPUX.
746         Rename FFI_LINUX ABI to FFI_PA32 ABI.
747         (FFI_TRAMPOLINE_SIZE): Define for 32-bit HP-UX targets.
748         (FFI_TYPE_SMALL_STRUCT2): Define.
749         (FFI_TYPE_SMALL_STRUCT4): Likewise.
750         (FFI_TYPE_SMALL_STRUCT8): Likewise.
751         (FFI_TYPE_SMALL_STRUCT3): Redefine.
752         (FFI_TYPE_SMALL_STRUCT5): Likewise.
753         (FFI_TYPE_SMALL_STRUCT6): Likewise.
754         (FFI_TYPE_SMALL_STRUCT7): Likewise.
755         * src/pa/ffi.c (ROUND_DOWN): Delete.
756         (fldw, fstw, fldd, fstd): Use '__asm__'.
757         (ffi_struct_type): Add support for FFI_TYPE_SMALL_STRUCT2,
758         FFI_TYPE_SMALL_STRUCT4 and FFI_TYPE_SMALL_STRUCT8.
759         (ffi_prep_args_LINUX): Rename to ffi_prep_args_pa32. Update comment.
760         Simplify incrementing of stack slot variable. Change type of local
761         'n' to unsigned int.
762         (ffi_size_stack_LINUX): Rename to ffi_size_stack_pa32. Handle long
763         double on PA_HPUX.
764         (ffi_prep_cif_machdep): Likewise.
765         (ffi_call): Likewise.
766         (ffi_closure_inner_LINUX): Rename to ffi_closure_inner_pa32. Change
767         return type to ffi_status. Simplify incrementing of stack slot
768         variable. Only copy floating point argument registers when PA_LINUX
769         is true. Reformat debug statement.
770         Add support for FFI_TYPE_SMALL_STRUCT2, FFI_TYPE_SMALL_STRUCT4 and
771         FFI_TYPE_SMALL_STRUCT8.
772         (ffi_closure_LINUX): Rename to ffi_closure_pa32. Add 'extern' to
773         declaration.
774         (ffi_prep_closure): Make linux trampoline conditional on PA_LINUX.
775         Add nops to cache flush.  Add trampoline for PA_HPUX.
776         * src/pa/hpux32.S: New file.
777         * src/pa/linux.S (ffi_call_LINUX): Rename to ffi_call_pa32. Rename
778         ffi_prep_args_LINUX to ffi_prep_args_pa32.
779         Localize labels. Add support for 2, 4 and 8-byte small structs. Handle
780         unaligned destinations in 3, 5, 6 and 7-byte small structs. Order
781         argument type checks so that common argument types appear first.
782         (ffi_closure_LINUX): Rename to ffi_closure_pa32. Rename
783         ffi_closure_inner_LINUX to ffi_closure_inner_pa32.
784
785 2006-03-24  Alan Modra  <amodra@bigpond.net.au>
786
787         * src/powerpc/ffitarget.h (enum ffi_abi): Add FFI_LINUX.  Default
788         for 32-bit using IBM extended double format.  Fix FFI_LAST_ABI.
789         * src/powerpc/ffi.c (ffi_prep_args_SYSV): Handle linux variant of
790         FFI_TYPE_LONGDOUBLE.
791         (ffi_prep_args64): Assert using IBM extended double.
792         (ffi_prep_cif_machdep): Don't munge FFI_TYPE_LONGDOUBLE type.
793         Handle FFI_LINUX FFI_TYPE_LONGDOUBLE return and args.
794         (ffi_call): Handle FFI_LINUX.
795         (ffi_closure_helper_SYSV): Non FFI_LINUX long double return needs
796         gpr3 return pointer as for struct return.  Handle FFI_LINUX
797         FFI_TYPE_LONGDOUBLE return and args.  Don't increment "nf"
798         unnecessarily.
799         * src/powerpc/ppc_closure.S (ffi_closure_SYSV): Load both f1 and f2
800         for FFI_TYPE_LONGDOUBLE.  Move epilogue insns into case table.
801         Don't use r6 as pointer to results, instead use sp offset.  Don't
802         make a special call to load lr with case table address, instead
803         use offset from previous call.
804         * src/powerpc/sysv.S (ffi_call_SYSV): Save long double return.
805         * src/powerpc/linux64.S (ffi_call_LINUX64): Simplify long double
806         return.
807
808 2006-03-15  Kaz Kojima  <kkojima@gcc.gnu.org>
809
810         * src/sh64/ffi.c (ffi_prep_cif_machdep): Handle float arguments
811         passed with FP registers correctly.
812         (ffi_closure_helper_SYSV): Likewise.
813         * src/sh64/sysv.S: Likewise.
814
815 2006-03-01  Andreas Tobler  <a.tobler@schweiz.ch>
816
817         * testsuite/libffi.special/unwindtest.cc (closure_test_fn): Mark cif,
818         args and userdata unused.
819         (closure_test_fn1): Mark cif and userdata unused.
820         (main): Remove unused res.
821
822 2006-02-28  Andreas Tobler  <a.tobler@schweiz.ch>
823
824         * testsuite/libffi.call/call.exp: Adjust FSF address. Add test runs for
825         -O2, -O3, -Os and the warning flags -W -Wall.
826         * testsuite/libffi.special/special.exp: Likewise.
827         * testsuite/libffi.call/ffitest.h: Add an __UNUSED__ macro to mark
828         unused parameter unused for gcc or else do nothing.
829         * testsuite/libffi.special/ffitestcxx.h: Likewise.
830         * testsuite/libffi.call/cls_12byte.c (cls_struct_12byte_gn): Mark cif
831         and userdata unused.
832         * testsuite/libffi.call/cls_16byte.c (cls_struct_16byte_gn): Likewise.
833         * testsuite/libffi.call/cls_18byte.c (cls_struct_18byte_gn): Likewise.
834         * testsuite/libffi.call/cls_19byte.c (cls_struct_19byte_gn): Likewise.
835         * testsuite/libffi.call/cls_1_1byte.c (cls_struct_1_1byte_gn): Likewise.
836         * testsuite/libffi.call/cls_20byte.c (cls_struct_20byte_gn): Likewise.
837         * testsuite/libffi.call/cls_20byte1.c (cls_struct_20byte_gn): Likewise.
838         * testsuite/libffi.call/cls_24byte.c (cls_struct_24byte_gn): Likewise.
839         * testsuite/libffi.call/cls_2byte.c (cls_struct_2byte_gn): Likewise.
840         * testsuite/libffi.call/cls_3_1byte.c (cls_struct_3_1byte_gn): Likewise.
841         * testsuite/libffi.call/cls_3byte1.c (cls_struct_3byte_gn): Likewise.
842         * testsuite/libffi.call/cls_3byte2.c (cls_struct_3byte_gn1): Likewise.
843         * testsuite/libffi.call/cls_4_1byte.c (cls_struct_4_1byte_gn): Likewise.
844         * testsuite/libffi.call/cls_4byte.c (cls_struct_4byte_gn): Likewise.
845         * testsuite/libffi.call/cls_5_1_byte.c (cls_struct_5byte_gn): Likewise.
846         * testsuite/libffi.call/cls_5byte.c (cls_struct_5byte_gn): Likewise.
847         * testsuite/libffi.call/cls_64byte.c (cls_struct_64byte_gn): Likewise.
848         * testsuite/libffi.call/cls_6_1_byte.c (cls_struct_6byte_gn): Likewise.
849         * testsuite/libffi.call/cls_6byte.c (cls_struct_6byte_gn): Likewise.
850         * testsuite/libffi.call/cls_7_1_byte.c (cls_struct_7byte_gn): Likewise.
851         * testsuite/libffi.call/cls_7byte.c (cls_struct_7byte_gn): Likewise.
852         * testsuite/libffi.call/cls_8byte.c (cls_struct_8byte_gn): Likewise.
853         * testsuite/libffi.call/cls_9byte1.c (cls_struct_9byte_gn): Likewise.
854         * testsuite/libffi.call/cls_9byte2.c (cls_struct_9byte_gn): Likewise.
855         * testsuite/libffi.call/cls_align_double.c (cls_struct_align_gn):
856         Likewise.
857         * testsuite/libffi.call/cls_align_float.c (cls_struct_align_gn):
858         Likewise.
859         * testsuite/libffi.call/cls_align_longdouble.c (cls_struct_align_gn):
860         Likewise.
861         * testsuite/libffi.call/cls_align_pointer.c (cls_struct_align_fn): Cast
862         void* to avoid compiler warning.
863         (main): Likewise.
864         (cls_struct_align_gn): Mark cif and userdata unused.
865         * testsuite/libffi.call/cls_align_sint16.c (cls_struct_align_gn):
866         Likewise.
867         * testsuite/libffi.call/cls_align_sint32.c (cls_struct_align_gn):
868         Likewise.
869         * testsuite/libffi.call/cls_align_sint64.c (cls_struct_align_gn):
870         Likewise.
871         * testsuite/libffi.call/cls_align_uint16.c (cls_struct_align_gn):
872         Likewise.
873         * testsuite/libffi.call/cls_align_uint32.c (cls_struct_align_gn):
874         Likewise.
875         * testsuite/libffi.call/cls_double.c (cls_ret_double_fn): Likewise.
876         * testsuite/libffi.call/cls_float.c (cls_ret_float_fn): Likewise.
877         * testsuite/libffi.call/cls_multi_schar.c (test_func_gn): Mark cif and
878         data unused.
879         (main): Cast res_call to silence gcc.
880         * testsuite/libffi.call/cls_multi_sshort.c (test_func_gn): Mark cif and
881         data unused.
882         (main): Cast res_call to silence gcc.
883         * testsuite/libffi.call/cls_multi_sshortchar.c (test_func_gn): Mark cif
884         and data unused.
885         (main): Cast res_call to silence gcc.
886         * testsuite/libffi.call/cls_multi_uchar.c (test_func_gn): Mark cif and
887         data unused.
888         (main): Cast res_call to silence gcc.
889         * testsuite/libffi.call/cls_multi_ushort.c (test_func_gn): Mark cif and
890         data unused.
891         (main): Cast res_call to silence gcc.
892         * testsuite/libffi.call/cls_multi_ushortchar.c (test_func_gn): Mark cif
893         and data unused.
894         (main): Cast res_call to silence gcc.
895         * testsuite/libffi.call/cls_schar.c (cls_ret_schar_fn): Mark cif and
896         userdata unused.
897         (cls_ret_schar_fn): Cast printf parameter to silence gcc.
898         * testsuite/libffi.call/cls_sint.c (cls_ret_sint_fn): Mark cif and
899         userdata unused.
900         (cls_ret_sint_fn): Cast printf parameter to silence gcc.
901         * testsuite/libffi.call/cls_sshort.c (cls_ret_sshort_fn): Mark cif and
902         userdata unused.
903         (cls_ret_sshort_fn): Cast printf parameter to silence gcc.
904         * testsuite/libffi.call/cls_uchar.c (cls_ret_uchar_fn):  Mark cif and
905         userdata unused.
906         (cls_ret_uchar_fn): Cast printf parameter to silence gcc.
907         * testsuite/libffi.call/cls_uint.c (cls_ret_uint_fn): Mark cif and
908         userdata unused.
909         (cls_ret_uint_fn): Cast printf parameter to silence gcc.
910         * testsuite/libffi.call/cls_ulonglong.c (cls_ret_ulonglong_fn): Mark cif
911         and userdata unused.
912         * testsuite/libffi.call/cls_ushort.c (cls_ret_ushort_fn): Mark cif and
913         userdata unused.
914         (cls_ret_ushort_fn): Cast printf parameter to silence gcc.
915         * testsuite/libffi.call/float.c (floating): Remove unused parameter e.
916         * testsuite/libffi.call/float1.c (main): Remove unused variable i.
917         Cleanup white spaces.
918         * testsuite/libffi.call/negint.c (checking): Remove unused variable i.
919         * testsuite/libffi.call/nested_struct.c (cls_struct_combined_gn): Mark
920         cif and userdata unused.
921         * testsuite/libffi.call/nested_struct1.c (cls_struct_combined_gn):
922         Likewise.
923         * testsuite/libffi.call/nested_struct10.c (B_gn): Likewise.
924         * testsuite/libffi.call/nested_struct2.c (B_fn): Adjust printf
925         formatters to silence gcc.
926         (B_gn): Mark cif and userdata unused.
927         * testsuite/libffi.call/nested_struct3.c (B_gn): Mark cif and userdata
928         unused.
929         * testsuite/libffi.call/nested_struct4.c: Mention related PR.
930         (B_gn): Mark cif and userdata unused.
931         * testsuite/libffi.call/nested_struct5.c (B_gn): Mark cif and userdata
932         unused.
933         * testsuite/libffi.call/nested_struct6.c: Mention related PR.
934         (B_gn): Mark cif and userdata unused.
935         * testsuite/libffi.call/nested_struct7.c (B_gn): Mark cif and userdata
936         unused.
937         * testsuite/libffi.call/nested_struct8.c (B_gn): Likewise.
938         * testsuite/libffi.call/nested_struct9.c (B_gn): Likewise.
939         * testsuite/libffi.call/problem1.c (stub): Likewise.
940         * testsuite/libffi.call/pyobjc-tc.c (main): Cast the result to silence
941         gcc.
942         * testsuite/libffi.call/return_fl2.c (return_fl): Add the note mentioned
943         in the last commit for this test case in the test case itself.
944         * testsuite/libffi.call/closure_fn0.c (closure_test_fn0): Mark cif as
945         unused.
946         * testsuite/libffi.call/closure_fn1.c (closure_test_fn1): Likewise.
947         * testsuite/libffi.call/closure_fn2.c (closure_test_fn2): Likewise.
948         * testsuite/libffi.call/closure_fn3.c (closure_test_fn3): Likewise.
949         * testsuite/libffi.call/closure_fn4.c (closure_test_fn0): Likewise.
950         * testsuite/libffi.call/closure_fn5.c (closure_test_fn5): Likewise.
951         * testsuite/libffi.call/closure_fn6.c (closure_test_fn0): Likewise.
952
953 2006-02-22  Kaz Kojima  <kkojima@gcc.gnu.org>
954
955         * src/sh/sysv.S: Fix register numbers in the FDE for
956         ffi_closure_SYSV.
957
958 2006-02-20  Andreas Tobler  <a.tobler@schweiz.ch>
959
960         * testsuite/libffi.call/return_fl2.c (return_fl): Remove static
961         declaration to avoid a false negative on ix86. See PR323.
962
963 2006-02-18  Kaz Kojima  <kkojima@gcc.gnu.org>
964
965         * src/sh/ffi.c (ffi_closure_helper_SYSV): Remove unused variable
966         and cast integer to void * if needed.  Update the pointer to
967         the FP register saved area correctly.
968         
969 2006-02-17  Andreas Tobler  <a.tobler@schweiz.ch>
970
971         * testsuite/libffi.call/nested_struct6.c: XFAIL this test until PR25630
972         is fixed.
973         * testsuite/libffi.call/nested_struct4.c: Likewise.
974
975 2006-02-16  Andreas Tobler  <a.tobler@schweiz.ch>
976
977         * testsuite/libffi.call/return_dbl.c: New test case.
978         * testsuite/libffi.call/return_dbl1.c: Likewise.
979         * testsuite/libffi.call/return_dbl2.c: Likewise.
980         * testsuite/libffi.call/return_fl.c: Likewise.
981         * testsuite/libffi.call/return_fl1.c: Likewise.
982         * testsuite/libffi.call/return_fl2.c: Likewise.
983         * testsuite/libffi.call/return_fl3.c: Likewise.
984         * testsuite/libffi.call/closure_fn6.c: Likewise.
985
986         * testsuite/libffi.call/nested_struct2.c: Remove ffi_type_mylong
987         definition.
988         * testsuite/libffi.call/ffitest.h: Add ffi_type_mylong definition
989         here to be used by other test cases too.
990
991         * testsuite/libffi.call/nested_struct10.c: New test case.
992         * testsuite/libffi.call/nested_struct9.c: Likewise.
993         * testsuite/libffi.call/nested_struct8.c: Likewise.
994         * testsuite/libffi.call/nested_struct7.c: Likewise.
995         * testsuite/libffi.call/nested_struct6.c: Likewise.
996         * testsuite/libffi.call/nested_struct5.c: Likewise.
997         * testsuite/libffi.call/nested_struct4.c: Likewise.
998
999 2006-01-21  Andreas Tobler  <a.tobler@schweiz.ch>
1000
1001         * configure.ac: Enable libffi for sparc64-*-freebsd*.
1002         * configure: Rebuilt.
1003
1004 2006-01-18  Jakub Jelinek  <jakub@redhat.com>
1005
1006         * src/powerpc/sysv.S (smst_two_register): Don't call __ashldi3,
1007         instead do the shifting inline.
1008         * src/powerpc/ppc_closure.S (ffi_closure_SYSV): Don't compute %r5
1009         shift count unconditionally.  Simplify load sequences for 1, 2, 3, 4
1010         and 8 byte structs, for the remaining struct sizes don't call
1011         __lshrdi3, instead do the shifting inline.
1012
1013 2005-12-07  Thiemo Seufer  <ths@networkno.de>
1014
1015         * src/mips/ffitarget.h: Remove obsolete sgidefs.h include. Add
1016         missing parentheses.
1017         * src/mips/o32.S (ffi_call_O32): Code formatting. Define
1018         and use A3_OFF, FP_OFF, RA_OFF. Micro-optimizations.
1019         (ffi_closure_O32): Likewise, but with newly defined A3_OFF2,
1020         A2_OFF2, A1_OFF2, A0_OFF2, RA_OFF2, FP_OFF2, S0_OFF2, GP_OFF2,
1021         V1_OFF2, V0_OFF2, FA_1_1_OFF2, FA_1_0_OFF2, FA_0_1_OFF2,
1022         FA_0_0_OFF2.
1023         * src/mips/ffi.c (ffi_prep_args): Code formatting. Fix
1024         endianness bugs.
1025         (ffi_prep_closure): Improve trampoline instruction scheduling.
1026         (ffi_closure_mips_inner_O32): Fix endianness bugs.
1027
1028 2005-12-03  Alan Modra  <amodra@bigpond.net.au>
1029
1030         * src/powerpc/ffi.c: Formatting.
1031         (ffi_prep_args_SYSV): Avoid possible aliasing problems by using unions.
1032         (ffi_prep_args64): Likewise.
1033
1034 2005-09-30  Geoffrey Keating  <geoffk@apple.com>
1035
1036         * testsuite/lib/libffi-dg.exp (libffi_target_compile): For
1037         darwin, use -shared-libgcc not -lgcc_s, and explain why.
1038
1039 2005-09-26  Tom Tromey  <tromey@redhat.com>
1040
1041         * testsuite/libffi.call/float1.c (value_type): New typedef.
1042         (CANARY): New define.
1043         (main): Check for result buffer overflow.
1044         * src/powerpc/linux64.S: Handle linux64 long double returns.
1045         * src/powerpc/ffi.c (FLAG_RETURNS_128BITS): New constant.
1046         (ffi_prep_cif_machdep): Handle linux64 long double returns.
1047
1048 2005-08-25  Alan Modra  <amodra@bigpond.net.au>
1049
1050         PR target/23404
1051         * src/powerpc/ffi.c (ffi_prep_args_SYSV): Correct placement of stack
1052         homed fp args.
1053         (ffi_status ffi_prep_cif_machdep): Correct stack sizing for same.
1054
1055 2005-08-11  Jakub Jelinek  <jakub@redhat.com>
1056
1057         * configure.ac (HAVE_HIDDEN_VISIBILITY_ATTRIBUTE): New test.
1058         (AH_BOTTOM): Add FFI_HIDDEN definition.
1059         * configure: Rebuilt.
1060         * fficonfig.h.in: Rebuilt.
1061         * src/powerpc/ffi.c (hidden): Remove.
1062         (ffi_closure_LINUX64, ffi_prep_args64, ffi_call_LINUX64,
1063         ffi_closure_helper_LINUX64): Use FFI_HIDDEN instead of hidden.
1064         * src/powerpc/linux64_closure.S (ffi_closure_LINUX64,
1065         .ffi_closure_LINUX64): Use FFI_HIDDEN instead of .hidden.
1066         * src/x86/ffi.c (ffi_closure_SYSV, ffi_closure_raw_SYSV): Remove,
1067         add FFI_HIDDEN to its prototype.
1068         (ffi_closure_SYSV_inner): New.
1069         * src/x86/sysv.S (ffi_closure_SYSV, ffi_closure_raw_SYSV): New.
1070         * src/x86/win32.S (ffi_closure_SYSV, ffi_closure_raw_SYSV): New.
1071
1072 2005-08-10  Alfred M. Szmidt  <ams@gnu.org>
1073
1074         PR libffi/21819:
1075         * configure: Rebuilt.
1076         * configure.ac: Handle i*86-*-gnu*.
1077
1078 2005-08-09  Jakub Jelinek  <jakub@redhat.com>
1079
1080         * src/powerpc/ppc_closure.S (ffi_closure_SYSV): Use
1081         DW_CFA_offset_extended_sf rather than
1082         DW_CFA_GNU_negative_offset_extended.
1083         * src/powerpc/sysv.S (ffi_call_SYSV): Likewise.
1084
1085 2005-07-22  SUGIOKA Toshinobu  <sugioka@itonet.co.jp>
1086
1087         * src/sh/sysv.S (ffi_call_SYSV): Stop argument popping correctly
1088         on sh3.
1089         (ffi_closure_SYSV): Change the stack layout for sh3 struct argument.
1090         * src/sh/ffi.c (ffi_prep_args): Fix sh3 argument copy, when it is
1091         partially on register.
1092         (ffi_closure_helper_SYSV): Likewise.
1093         (ffi_prep_cif_machdep): Don't set too many cif->flags.
1094
1095 2005-07-20  Kaz Kojima  <kkojima@gcc.gnu.org>
1096
1097         * src/sh/ffi.c (ffi_call): Handle small structures correctly.
1098         Remove empty line.
1099         * src/sh64/ffi.c (simple_type): Remove.
1100         (return_type): Handle small structures correctly.
1101         (ffi_prep_args): Likewise.
1102         (ffi_call): Likewise.
1103         (ffi_closure_helper_SYSV): Likewise.
1104         * src/sh64/sysv.S (ffi_call_SYSV): Handle 1, 2 and 4-byte return.
1105         Emit position independent code if PIC and remove wrong datalabel
1106         prefixes from EH data.
1107
1108 2005-07-19  Andreas Tobler  <a.tobler@schweiz.ch>
1109
1110         * Makefile.am (nodist_libffi_la_SOURCES): Add POWERPC_FREEBSD.
1111         * Makefile.in: Regenerate.
1112         * include/Makefile.in: Likewise.
1113         * testsuite/Makefile.in: Likewise.
1114         * configure.ac: Add POWERPC_FREEBSD rules.
1115         * configure: Regenerate.
1116         * src/powerpc/ffitarget.h: Add POWERPC_FREEBSD rules.
1117         (FFI_SYSV_TYPE_SMALL_STRUCT): Define.
1118         * src/powerpc/ffi.c: Add flags to handle small structure returns
1119         in ffi_call_SYSV.
1120         (ffi_prep_cif_machdep): Handle small structures for SYSV 4 ABI.
1121         Aka FFI_SYSV.
1122         (ffi_closure_helper_SYSV): Likewise.
1123         * src/powerpc/ppc_closure.S: Add return types for small structures.
1124         * src/powerpc/sysv.S: Add bits to handle small structures for
1125         final SYSV 4 ABI.
1126
1127 2005-07-10  Andreas Tobler  <a.tobler@schweiz.ch>
1128
1129         * testsuite/libffi.call/cls_5_1_byte.c: New test file.
1130         * testsuite/libffi.call/cls_6_1_byte.c: Likewise.
1131         * testsuite/libffi.call/cls_7_1_byte.c: Likewise.
1132
1133 2005-07-05  Randolph Chung  <tausq@debian.org>
1134
1135         * src/pa/ffi.c (ffi_struct_type): Rename FFI_TYPE_SMALL_STRUCT1
1136         as FFI_TYPE_SMALL_STRUCT3.  Break out handling for 5-7 byte
1137         structures.  Kill compilation warnings.
1138         (ffi_closure_inner_LINUX): Print return values as hex in debug
1139         message.  Rename FFI_TYPE_SMALL_STRUCT1 as FFI_TYPE_SMALL_STRUCT3.
1140         Properly handle 5-7 byte structure returns.
1141         * src/pa/ffitarget.h (FFI_TYPE_SMALL_STRUCT1)
1142         (FFI_TYPE_SMALL_STRUCT2): Remove.
1143         (FFI_TYPE_SMALL_STRUCT3, FFI_TYPE_SMALL_STRUCT5)
1144         (FFI_TYPE_SMALL_STRUCT6, FFI_TYPE_SMALL_STRUCT7): Define.
1145         * src/pa/linux.S: Mark source file as using PA1.1 assembly.
1146         (checksmst1, checksmst2): Remove.
1147         (checksmst3): Optimize handling of 3-byte struct returns.
1148         (checksmst567): Properly handle 5-7 byte struct returns.
1149
1150 2005-06-15  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
1151
1152         PR libgcj/21943
1153         * src/mips/n32.S: Enforce PIC code.
1154         * src/mips/o32.S: Likewise.
1155
1156 2005-06-15  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
1157
1158         * configure.ac: Treat i*86-*-solaris2.10 and up as X86_64.
1159         * configure: Regenerate.
1160         
1161 2005-06-01  Alan Modra  <amodra@bigpond.net.au>
1162
1163         * src/powerpc/ppc_closure.S (ffi_closure_SYSV): Don't use JUMPTARGET
1164         to call ffi_closure_helper_SYSV.  Append @local instead.
1165         * src/powerpc/sysv.S (ffi_call_SYSV): Likewise for ffi_prep_args_SYSV.
1166
1167 2005-05-17  Kelley Cook  <kcook@gcc.gnu.org>
1168
1169         * configure.ac: Use AC_C_BIGENDIAN instead of AC_C_BIGENDIAN_CROSS.
1170         Use AC_CHECK_SIZEOF instead of AC_COMPILE_CHECK_SIZEOF.
1171         * Makefile.am (ACLOCAL_AMFLAGS): Remove -I ../config.
1172         * aclocal.m4, configure, fficonfig.h.in, Makefile.in,
1173         include/Makefile.in, testsuite/Makefile.in: Regenerate.
1174         
1175 2005-05-09  Mike Stump  <mrs@apple.com>
1176
1177         * configure: Regenerate.
1178
1179 2005-05-08  Richard Henderson  <rth@redhat.com>
1180
1181         PR libffi/21285
1182         * src/alpha/osf.S: Update unwind into to match code.
1183
1184 2005-05-04  Andreas Degert <ad@papyrus-gmbh.de>
1185             Richard Henderson  <rth@redhat.com>
1186
1187         * src/x86/ffi64.c (ffi_prep_cif_machdep): Save sse-used flag in
1188         bit 11 of flags.
1189         (ffi_call): Mask return type field.  Pass ssecount to ffi_call_unix64.
1190         (ffi_prep_closure): Set carry bit if sse-used flag set.
1191         * src/x86/unix64.S (ffi_call_unix64): Add ssecount argument.
1192         Only load sse registers if ssecount non-zero.
1193         (ffi_closure_unix64): Only save sse registers if carry set on entry.
1194
1195 2005-04-29  Ralf Corsepius  <ralf.corsepius@rtems.org>
1196
1197         * configure.ac: Add i*86-*-rtems*, sparc*-*-rtems*,
1198         powerpc-*rtems*, arm*-*-rtems*, sh-*-rtems*.
1199         * configure: Regenerate.
1200
1201 2005-04-20  Hans-Peter Nilsson  <hp@axis.com>
1202
1203         * testsuite/lib/libffi-dg.exp (libffi-dg-test-1): In regsub use,
1204         have Tcl8.3-compatible intermediate variable.
1205
1206 2005-04-18  Simon Posnjak <simon.posnjak@siol.net> 
1207             Hans-Peter Nilsson  <hp@axis.com>
1208
1209         * Makefile.am: Add CRIS support.
1210         * configure.ac: Likewise.
1211         * Makefile.in, configure, testsuite/Makefile.in,
1212         include/Makefile.in: Regenerate.
1213         * src/cris: New directory.
1214         * src/cris/ffi.c, src/cris/sysv.S, src/cris/ffitarget.h: New files.
1215         * src/prep_cif.c (ffi_prep_cif): Wrap in #ifndef __CRIS__.
1216
1217         * testsuite/lib/libffi-dg.exp (libffi-dg-test-1): Replace \n with
1218         \r?\n in output tests.
1219
1220 2005-04-12  Mike Stump  <mrs@apple.com>
1221
1222         * configure: Regenerate.
1223
1224 2005-03-30  Hans Boehm  <Hans.Boehm@hp.com>
1225
1226         * src/ia64/ffitarget.h (ffi_arg): Use long long instead of DI.
1227         
1228 2005-03-30  Steve Ellcey  <sje@cup.hp.com>
1229
1230         * src/ia64/ffitarget.h (ffi_arg) ADD DI attribute.
1231         (ffi_sarg) Ditto.
1232         * src/ia64/unix.S (ffi_closure_unix): Extend gp
1233         to 64 bits in ILP32 mode.
1234         Load 64 bits even for short data.
1235
1236 2005-03-23  Mike Stump  <mrs@apple.com>
1237
1238         * src/powerpc/darwin.S: Update for -m64 multilib.
1239         * src/powerpc/darwin_closure.S: Likewise.
1240
1241 2005-03-21  Zack Weinberg  <zack@codesourcery.com>
1242
1243         * configure.ac: Do not invoke TL_AC_GCC_VERSION.
1244         Do not set tool_include_dir.
1245         * aclocal.m4, configure, Makefile.in, testsuite/Makefile.in:
1246         Regenerate.
1247         * include/Makefile.am: Set gcc_version and toollibffidir.
1248         * include/Makefile.in: Regenerate.
1249
1250 2005-02-22  Andrew Haley  <aph@redhat.com>
1251
1252         * src/powerpc/ffi.c (ffi_prep_cif_machdep): Bump alignment to
1253         odd-numbered register pairs for 64-bit integer types.
1254
1255 2005-02-23  Andreas Tobler  <a.tobler@schweiz.ch>
1256
1257         PR libffi/20104
1258         * testsuite/libffi.call/return_ll1.c: New test case.
1259
1260 2005-02-11  Janis Johnson  <janis187@us.ibm.com>
1261
1262         * testsuite/libffi.call/cls_align_longdouble.c: Remove dg-options.
1263         * testsuite/libffi.call/float.c: Ditto.
1264         * testsuite/libffi.call/float2.c: Ditto.
1265         * testsuite/libffi.call/float3.c: Ditto.
1266
1267 2005-02-08  Andreas Tobler  <a.tobler@schweiz.ch>
1268
1269         * src/frv/ffitarget.h: Remove PPC stuff which does not belong to frv.
1270
1271 2005-01-12  Eric Botcazou  <ebotcazou@libertysurf.fr>
1272
1273         * testsuite/libffi.special/special.exp (cxx_options): Add
1274         -shared-libgcc.
1275
1276 2004-12-31  Richard Henderson  <rth@redhat.com>
1277
1278         * src/types.c (FFI_AGGREGATE_TYPEDEF): Remove.
1279         (FFI_TYPEDEF): Rename from FFI_INTEGRAL_TYPEDEF.  Replace size and
1280         offset parameters with a type parameter; deduce size and structure
1281         alignment.  Update all users.
1282
1283 2004-12-31  Richard Henderson  <rth@redhat.com>
1284
1285         * src/types.c (FFI_TYPE_POINTER): Define with sizeof.
1286         (FFI_TYPE_LONGDOUBLE): Fix for ia64.
1287         * src/ia64/ffitarget.h (struct ffi_ia64_trampoline_struct): Move
1288         into ffi_prep_closure.
1289         * src/ia64/ia64_flags.h, src/ia64/ffi.c, src/ia64/unix.S: Rewrite
1290         from scratch.
1291
1292 2004-12-27  Richard Henderson  <rth@redhat.com>
1293
1294         * src/x86/unix64.S: Fix typo in unwind info.
1295
1296 2004-12-25  Richard Henderson  <rth@redhat.com>
1297
1298         * src/x86/ffi64.c (struct register_args): Rename from stackLayout.
1299         (enum x86_64_reg_class): Add X86_64_COMPLEX_X87_CLASS.
1300         (merge_classes): Check for it.
1301         (SSE_CLASS_P): New.
1302         (classify_argument): Pass byte_offset by value; perform all updates
1303         inside struct case.
1304         (examine_argument): Add classes argument; handle
1305         X86_64_COMPLEX_X87_CLASS.
1306         (ffi_prep_args): Merge into ...
1307         (ffi_call): ... here.  Share stack frame with ffi_call_unix64.
1308         (ffi_prep_cif_machdep): Setup cif->flags for proper structure return.
1309         (ffi_fill_return_value): Remove.
1310         (ffi_prep_closure): Remove dead assert.
1311         (ffi_closure_unix64_inner): Rename from ffi_closure_UNIX64_inner.
1312         Rewrite to use struct register_args instead of va_list.  Create
1313         flags for handling structure returns.
1314         * src/x86/unix64.S: Remove dead strings.
1315         (ffi_call_unix64): Rename from ffi_call_UNIX64.  Rewrite to share
1316         stack frame with ffi_call.  Handle structure returns properly.
1317         (float2sse, floatfloat2sse, double2sse): Remove.
1318         (sse2float, sse2double, sse2floatfloat): Remove.
1319         (ffi_closure_unix64): Rename from ffi_closure_UNIX64.  Rewrite
1320         to handle structure returns properly.
1321
1322 2004-12-08  David Edelsohn  <edelsohn@gnu.org>
1323
1324         * Makefile.am (AM_MAKEFLAGS): Remove duplicate LIBCFLAGS and
1325         PICFLAG.
1326         * Makefile.in: Regenerated.
1327
1328 2004-12-02  Richard Sandiford  <rsandifo@redhat.com>
1329
1330         * configure.ac: Use TL_AC_GCC_VERSION to set gcc_version.
1331         * configure, aclocal.m4, Makefile.in: Regenerate.
1332         * include/Makefile.in, testsuite/Makefile.in: Regenerate.
1333
1334 2004-11-29  Kelley Cook  <kcook@gcc.gnu.org>
1335
1336         * configure: Regenerate for libtool change.
1337
1338 2004-11-25  Kelley Cook  <kcook@gcc.gnu.org>
1339
1340         * configure: Regenerate for libtool reversion.
1341
1342 2004-11-24  Kelley Cook  <kcook@gcc.gnu.org>
1343
1344         * configure: Regenerate for libtool change.
1345
1346 2004-11-23  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
1347
1348         * testsuite/lib/libffi-dg.exp: Use new procs in target-libpath.exp.
1349
1350 2004-11-23  Richard Sandiford  <rsandifo@redhat.com>
1351
1352         * src/mips/o32.S (ffi_call_O32, ffi_closure_O32): Use jalr instead
1353         of jal.  Use an absolute encoding for the frame information.
1354
1355 2004-11-23  Kelley Cook  <kcook@gcc.gnu.org>
1356
1357         * Makefile.am: Remove no-dependencies.  Add ACLOCAL_AMFLAGS.
1358         * acinclude.m4: Delete logic for sincludes.
1359         * aclocal.m4, Makefile.in, configure: Regenerate.
1360         * include/Makefile: Likewise.
1361         * testsuite/Makefile: Likewise.
1362
1363 2004-11-22  Eric Botcazou  <ebotcazou@libertysurf.fr>
1364
1365         * src/sparc/ffi.c (ffi_prep_closure): Align doubles and 64-bit integers
1366         on a 8-byte boundary.
1367         * src/sparc/v8.S (ffi_closure_v8): Reserve frame space for arguments.
1368
1369 2004-10-27  Richard Earnshaw  <rearnsha@arm.com>
1370
1371         * src/arm/ffi.c (ffi_prep_cif_machdep): Handle functions that return
1372         long long values.  Round stack allocation to a multiple of 8 bytes
1373         for ATPCS compatibility.
1374         * src/arm/sysv.S (ffi_call_SYSV): Rework to avoid use of APCS register
1375         names.  Handle returning long long types.  Add Thumb and interworking
1376         support.  Improve soft-float code.
1377
1378 2004-10-27  Richard Earnshaw  <rearnsha@arm.com>
1379
1380         * testsuite/lib/libffi-db.exp (load_gcc_lib): New function.
1381         (libffi_exit): New function.
1382         (libffi_init): Build the testglue wrapper if needed.
1383
1384 2004-10-25  Eric Botcazou  <ebotcazou@libertysurf.fr>
1385
1386         PR other/18138
1387         * testsuite/lib/libffi-dg.exp: Accept more than one multilib libgcc.
1388
1389 2004-10-25  Kazuhiro Inaoka <inaoka.kazuhiro@renesas.com>
1390
1391         * src/m32r/libffitarget.h (FFI_CLOSURES): Set to 0.
1392
1393 2004-10-20  Kaz Kojima  <kkojima@gcc.gnu.org>
1394
1395         * src/sh/sysv.S (ffi_call_SYSV): Don't align for double data.
1396         * testsuite/libffi.call/float3.c: New test case.
1397
1398 2004-10-18  Kaz Kojima  <kkojima@gcc.gnu.org>
1399
1400         * src/sh/ffi.c (ffi_prep_closure): Set T bit in trampoline for
1401         the function returning a structure pointed with R2.
1402         * src/sh/sysv.S (ffi_closure_SYSV): Use R2 as the pointer to
1403         the structure return value if T bit set.  Emit position
1404         independent code and EH data if PIC.
1405
1406 2004-10-13  Kazuhiro Inaoka  <inaoka.kazuhiro@renesas.com>
1407
1408         * Makefile.am: Add m32r support.
1409         * configure.ac: Likewise.
1410         * Makefile.in: Regenerate.
1411         * confiugre: Regenerate.
1412         * src/types.c: Add m32r port to FFI_INTERNAL_TYPEDEF
1413         (uint64, sint64, double, longdouble)
1414         * src/m32r: New directory.
1415         * src/m32r/ffi.c: New file.
1416         * src/m32r/sysv.S: Likewise.
1417         * src/m32r/ffitarget.h: Likewise.
1418
1419 2004-10-02  Kaz Kojima  <kkojima@gcc.gnu.org>
1420
1421         * testsuite/libffi.call/negint.c: New test case.
1422
1423 2004-09-14  H.J. Lu  <hongjiu.lu@intel.com>
1424
1425         PR libgcj/17465
1426         * testsuite/lib/libffi-dg.exp: Don't use global ld_library_path.
1427         Set up LD_LIBRARY_PATH, SHLIB_PATH, LD_LIBRARYN32_PATH,
1428         LD_LIBRARY64_PATH, LD_LIBRARY_PATH_32, LD_LIBRARY_PATH_64 and
1429         DYLD_LIBRARY_PATH.
1430
1431 2004-09-05  Andreas Tobler  <a.tobler@schweiz.ch>
1432
1433         * testsuite/libffi.call/many_win32.c: Remove whitespaces.
1434         * testsuite/libffi.call/promotion.c: Likewise.
1435         * testsuite/libffi.call/return_ll.c: Remove unused var. Cleanup
1436         whitespaces.
1437         * testsuite/libffi.call/return_sc.c: Likewise.
1438         * testsuite/libffi.call/return_uc.c: Likewise.
1439
1440 2004-09-05  Andreas Tobler  <a.tobler@schweiz.ch>
1441
1442         * src/powerpc/darwin.S: Fix comments and identation.
1443         * src/powerpc/darwin_closure.S: Likewise.
1444
1445 2004-09-02  Andreas Tobler  <a.tobler@schweiz.ch>
1446
1447         * src/powerpc/ffi_darwin.c: Add flag for longdouble return values.
1448         (ffi_prep_args): Handle longdouble arguments.
1449         (ffi_prep_cif_machdep): Set flags for longdouble. Calculate space for
1450         longdouble.
1451         (ffi_closure_helper_DARWIN): Add closure handling for longdouble.
1452         * src/powerpc/darwin.S (_ffi_call_DARWIN): Add handling of longdouble
1453         values.
1454         * src/powerpc/darwin_closure.S (_ffi_closure_ASM): Likewise.
1455         * src/types.c: Defined longdouble size and alignment for darwin.
1456
1457 2004-09-02  Andreas Tobler  <a.tobler@schweiz.ch>
1458
1459         * src/powerpc/aix.S: Remove whitespaces.
1460         * src/powerpc/aix_closure.S: Likewise.
1461         * src/powerpc/asm.h: Likewise.
1462         * src/powerpc/ffi.c: Likewise.
1463         * src/powerpc/ffitarget.h: Likewise.
1464         * src/powerpc/linux64.S: Likewise.
1465         * src/powerpc/linux64_closure.S: Likewise.
1466         * src/powerpc/ppc_closure.S: Likewise.
1467         * src/powerpc/sysv.S: Likewise.
1468
1469 2004-08-30  Anthony Green  <green@redhat.com>
1470
1471         * Makefile.am: Add frv support.
1472         * Makefile.in, testsuite/Makefile.in: Rebuilt.
1473         * configure.ac: Read configure.host.
1474         * configure.in: Read configure.host.
1475         * configure.host: New file.  frv-elf needs libgloss.
1476         * include/ffi.h.in: Force ffi_closure to have a nice big (8)
1477         alignment.  This is needed to frv and shouldn't harm the others.
1478         * include/ffi_common.h (ALIGN_DOWN): New macro.
1479         * src/frv/ffi.c, src/frv/ffitarget.h, src/frv/eabi.S: New files.
1480
1481 2004-08-24  David Daney  <daney@avtrex.com>
1482
1483         * testsuite/libffi.call/closure_fn0.c: Xfail mips64* instead of mips*.
1484         * testsuite/libffi.call/closure_fn1.c: Likewise.
1485         * testsuite/libffi.call/closure_fn2.c  Likewise.
1486         * testsuite/libffi.call/closure_fn3.c: Likewise.
1487         * testsuite/libffi.call/closure_fn4.c: Likewise.
1488         * testsuite/libffi.call/closure_fn5.c: Likewise.
1489         * testsuite/libffi.call/cls_18byte.c: Likewise.
1490         * testsuite/libffi.call/cls_19byte.c: Likewise.
1491         * testsuite/libffi.call/cls_1_1byte.c: Likewise.
1492         * testsuite/libffi.call/cls_20byte.c: Likewise.
1493         * testsuite/libffi.call/cls_20byte1.c: Likewise.
1494         * testsuite/libffi.call/cls_24byte.c: Likewise.
1495         * testsuite/libffi.call/cls_2byte.c: Likewise.
1496         * testsuite/libffi.call/cls_3_1byte.c: Likewise.
1497         * testsuite/libffi.call/cls_3byte1.c: Likewise.
1498         * testsuite/libffi.call/cls_3byte2.c: Likewise.
1499         * testsuite/libffi.call/cls_4_1byte.c: Likewise.
1500         * testsuite/libffi.call/cls_4byte.c: Likewise.
1501         * testsuite/libffi.call/cls_64byte.c: Likewise.
1502         * testsuite/libffi.call/cls_6byte.c: Likewise.
1503         * testsuite/libffi.call/cls_7byte.c: Likewise.
1504         * testsuite/libffi.call/cls_8byte.c: Likewise.
1505         * testsuite/libffi.call/cls_9byte1.c: Likewise.
1506         * testsuite/libffi.call/cls_9byte2.c: Likewise.
1507         * testsuite/libffi.call/cls_align_double.c: Likewise.
1508         * testsuite/libffi.call/cls_align_float.c: Likewise.
1509         * testsuite/libffi.call/cls_align_longdouble.c: Likewise.
1510         * testsuite/libffi.call/cls_align_pointer.c: Likewise.
1511         * testsuite/libffi.call/cls_align_sint16.c: Likewise.
1512         * testsuite/libffi.call/cls_align_sint32.c: Likewise.
1513         * testsuite/libffi.call/cls_align_sint64.c: Likewise.
1514         * testsuite/libffi.call/cls_align_uint16.c: Likewise.
1515         * testsuite/libffi.call/cls_align_uint32.c: Likewise.
1516         * testsuite/libffi.call/cls_align_uint64.c: Likewise.
1517         * testsuite/libffi.call/cls_double.c: Likewise.
1518         * testsuite/libffi.call/cls_float.c: Likewise.
1519         * testsuite/libffi.call/cls_multi_schar.c: Likewise.
1520         * testsuite/libffi.call/cls_multi_sshort.c: Likewise.
1521         * testsuite/libffi.call/cls_multi_sshortchar.c: Likewise.
1522         * testsuite/libffi.call/cls_multi_uchar.c: Likewise.
1523         * testsuite/libffi.call/cls_multi_ushort.c: Likewise.
1524         * testsuite/libffi.call/cls_multi_ushortchar.c: Likewise.
1525         * testsuite/libffi.call/cls_schar.c: Likewise.
1526         * testsuite/libffi.call/cls_sint.c: Likewise.
1527         * testsuite/libffi.call/cls_sshort.c: Likewise.
1528         * testsuite/libffi.call/cls_uchar.c: Likewise.
1529         * testsuite/libffi.call/cls_uint.c: Likewise.
1530         * testsuite/libffi.call/cls_ulonglong.c: Likewise.
1531         * testsuite/libffi.call/cls_ushort.c: Likewise.
1532         * testsuite/libffi.call/nested_struct.c: Likewise.
1533         * testsuite/libffi.call/nested_struct1.c: Likewise.
1534         * testsuite/libffi.call/nested_struct2.c: Likewise.
1535         * testsuite/libffi.call/nested_struct3.c: Likewise.
1536         * testsuite/libffi.call/problem1.c: Likewise.
1537         * testsuite/libffi.special/unwindtest.cc: Likewise.
1538         * testsuite/libffi.call/cls_12byte.c: Likewise and set return value
1539         to zero.
1540         * testsuite/libffi.call/cls_16byte.c: Likewise.
1541         * testsuite/libffi.call/cls_5byte.c: Likewise.
1542
1543 2004-08-23  David Daney <daney@avtrex.com>
1544
1545         PR libgcj/13141
1546         * src/mips/ffitarget.h (FFI_O32_SOFT_FLOAT): New ABI.
1547         * src/mips/ffi.c (ffi_prep_args): Fix alignment calculation.
1548         (ffi_prep_cif_machdep): Handle FFI_O32_SOFT_FLOAT floating point
1549         parameters and return types.
1550         (ffi_call): Handle FFI_O32_SOFT_FLOAT ABI.
1551         (ffi_prep_closure): Ditto.
1552         (ffi_closure_mips_inner_O32): Handle FFI_O32_SOFT_FLOAT ABI, fix
1553         alignment calculations.
1554         * src/mips/o32.S (ffi_closure_O32): Don't use floating point
1555         instructions if FFI_O32_SOFT_FLOAT, make stack frame ABI compliant.
1556
1557 2004-08-14  Casey Marshall <csm@gnu.org>
1558
1559         * src/mips/ffi.c (ffi_pref_cif_machdep): set `cif->flags' to
1560         contain `FFI_TYPE_UINT64' as return type for any 64-bit
1561         integer (O32 ABI only).
1562         (ffi_prep_closure): new function.
1563         (ffi_closure_mips_inner_O32): new function.
1564         * src/mips/ffitarget.h: Define `FFI_CLOSURES' and
1565         `FFI_TRAMPOLINE_SIZE' appropriately if the ABI is o32.
1566         * src/mips/o32.S (ffi_call_O32): add labels for .eh_frame. Return
1567         64 bit integers correctly.
1568         (ffi_closure_O32): new function.
1569         Added DWARF-2 unwind info for both functions.
1570
1571 2004-08-10  Andrew Haley  <aph@redhat.com>
1572
1573         * src/x86/ffi64.c (ffi_prep_args ): 8-align all stack arguments.
1574
1575 2004-08-01  Robert Millan  <robertmh@gnu.org>
1576
1577         * configure.ac: Detect knetbsd-gnu and kfreebsd-gnu.
1578         * configure: Regenerate.
1579
1580 2004-07-30  Maciej W. Rozycki  <macro@linux-mips.org>
1581
1582         * acinclude.m4 (AC_FUNC_MMAP_BLACKLIST): Check for <sys/mman.h>
1583         and mmap() explicitly instead of relying on preset autoconf cache
1584         variables.
1585         * aclocal.m4: Regenerate.
1586         * configure: Regenerate.
1587
1588 2004-07-11  Ulrich Weigand  <uweigand@de.ibm.com>
1589
1590         * src/s390/ffi.c (ffi_prep_args): Fix C aliasing violation.
1591         (ffi_check_float_struct): Remove unused prototype.
1592
1593 2004-06-30  Geoffrey Keating  <geoffk@apple.com>
1594
1595         * src/powerpc/ffi_darwin.c (flush_icache): ';' is a comment
1596         character on Darwin, use '\n\t' instead.
1597
1598 2004-06-26  Matthias Klose  <doko@debian.org>
1599
1600         * libtool-version: Fix typo in revision/age.
1601
1602 2004-06-17  Matthias Klose  <doko@debian.org>
1603
1604         * libtool-version: New.
1605         * Makefile.am (libffi_la_LDFLAGS): Use -version-info for soname.
1606         * Makefile.in: Regenerate.
1607
1608 2004-06-15  Paolo Bonzini  <bonzini@gnu.org>
1609
1610         * Makefile.am: Remove useless multilib rules.
1611         * Makefile.in: Regenerate.
1612         * aclocal.m4: Regenerate with automake 1.8.5.
1613         * configure.ac: Remove useless multilib configury.
1614         * configure: Regenerate.
1615
1616 2004-06-15  Paolo Bonzini  <bonzini@gnu.org>
1617
1618         * .cvsignore: New file.
1619
1620 2004-06-10  Jakub Jelinek  <jakub@redhat.com>
1621
1622         * src/ia64/unix.S (ffi_call_unix): Insert group barrier break
1623         fp_done.
1624         (ffi_closure_UNIX): Fix f14/f15 adjustment if FLOAT_SZ is ever
1625         changed from 8.
1626
1627 2004-06-06  Sean McNeil  <sean@mcneil.com>
1628
1629         * configure.ac: Add x86_64-*-freebsd* support.
1630         * configure: Regenerate.
1631
1632 2004-04-26  Joe Buck <jbuck@welsh-buck.org>
1633
1634         Bug 15093
1635         * configure.ac: Test for existence of mmap and sys/mman.h before
1636         checking blacklist.  Fix suggested by Jim Wilson.
1637         * configure: Regenerate.
1638
1639 2004-04-26  Matt Austern  <austern@apple.com>
1640
1641         * src/powerpc/darwin.S: Go through a non-lazy pointer for initial
1642         FDE location.
1643         * src/powerpc/darwin_closure.S: Likewise.
1644
1645 2004-04-24  Andreas Tobler  <a.tobler@schweiz.ch>
1646
1647         * testsuite/libffi.call/cls_multi_schar.c (main): Fix initialization
1648         error. Reported by Thomas Heller <theller@python.net>.
1649         * testsuite/libffi.call/cls_multi_sshort.c (main): Likewise.
1650         * testsuite/libffi.call/cls_multi_ushort.c (main): Likewise.
1651
1652 2004-03-20  Matthias Klose  <doko@debian.org>
1653
1654         * src/pa/linux.S: Fix typo.
1655
1656 2004-03-19  Matthias Klose  <doko@debian.org>
1657
1658         * Makefile.am: Update.
1659         * Makefile.in: Regenerate.
1660         * src/pa/ffi.h.in: Remove.
1661         * src/pa/ffitarget.h: New file.
1662
1663 2004-02-10  Randolph Chung  <tausq@debian.org>
1664
1665         * Makefile.am: Add PA support.
1666         * Makefile.in: Regenerate.
1667         * include/Makefile.in: Regenerate.
1668         * configure.ac: Add PA target.
1669         * configure: Regenerate.
1670         * src/pa/ffi.c: New file.
1671         * src/pa/ffi.h.in: Add PA support.
1672         * src/pa/linux.S: New file.
1673         * prep_cif.c: Add PA support.
1674
1675 2004-03-16  Hosaka Yuji  <hos@tamanegi.org>
1676
1677         * src/types.c: Fix alignment size of X86_WIN32 case int64 and
1678         double.
1679         * src/x86/ffi.c (ffi_prep_args): Replace ecif->cif->rtype->type
1680         with ecif->cif->flags.
1681         (ffi_call, ffi_prep_incoming_args_SYSV): Replace cif->rtype->type
1682         with cif->flags.
1683         (ffi_prep_cif_machdep): Add X86_WIN32 struct case.
1684         (ffi_closure_SYSV): Add 1 or 2-bytes struct case for X86_WIN32.
1685         * src/x86/win32.S (retstruct1b, retstruct2b, sc_retstruct1b,
1686         sc_retstruct2b): Add for 1 or 2-bytes struct case.
1687
1688 2004-03-15 Kelley Cook <kcook@gcc.gnu.org>
1689
1690         * configure.in: Rename file to ...
1691         * configure.ac: ... this.
1692         * fficonfig.h.in: Regenerate.
1693         * Makefile.in: Regenerate.
1694         * include/Makefile.in: Regenerate.
1695         * testsuite/Makefile.in: Regenerate.
1696
1697 2004-03-12  Matt Austern  <austern@apple.com>
1698
1699         * src/powerpc/darwin.S: Fix EH information so it corresponds to
1700         changes in EH format resulting from addition of linkonce support.
1701         * src/powerpc/darwin_closure.S: Likewise.
1702
1703 2004-03-11  Andreas Tobler  <a.tobler@schweiz.ch>
1704             Paolo Bonzini  <bonzini@gnu.org>
1705
1706         * Makefile.am (AUTOMAKE_OPTIONS): Set them.
1707         Remove VPATH. Remove rules for object files. Remove multilib support.
1708         (AM_CCASFLAGS): Add.
1709         * configure.in (AC_CONFIG_HEADERS): Relace AM_CONFIG_HEADER.
1710         (AC_PREREQ): Bump version to 2.59.
1711         (AC_INIT): Fill with version info and bug address.
1712         (ORIGINAL_LD_FOR_MULTILIBS): Remove.
1713         (AM_ENABLE_MULTILIB): Use this instead of AC_ARG_ENABLE.
1714         De-precious CC so that the right flags are passed down to multilibs.
1715         (AC_MSG_ERROR): Replace obsolete macro AC_ERROR.
1716         (AC_CONFIG_FILES): Replace obsolete macro AC_LINK_FILES.
1717         (AC_OUTPUT): Reorganize the output with AC_CONFIG_COMMANDS.
1718         * configure: Rebuilt.
1719         * aclocal.m4: Likewise.
1720         * Makefile.in, include/Makefile.in, testsuite/Makefile.in: Likewise.
1721         * fficonfig.h.in: Likewise.
1722
1723 2004-03-11  Andreas Schwab  <schwab@suse.de>
1724
1725         * src/ia64/ffi.c (ffi_prep_incoming_args_UNIX): Get floating point
1726         arguments from fp registers only for the first 8 parameter slots.
1727         Don't convert a float parameter when passed in memory.
1728
1729 2004-03-09  Hans-Peter Nilsson  <hp@axis.com>
1730
1731         * configure: Regenerate for config/accross.m4 correction.
1732
1733 2004-02-25  Matt Kraai  <kraai@alumni.cmu.edu>
1734
1735         * src/powerpc/ffi.c (ffi_prep_args_SYSV): Change
1736         ecif->cif->bytes to bytes.
1737         (ffi_prep_cif_machdep): Add braces around nested if statement.
1738
1739 2004-02-09  Alan Modra  <amodra@bigpond.net.au>
1740
1741         * src/types.c (pointer): POWERPC64 has 8 byte pointers.
1742
1743         * src/powerpc/ffi.c (ffi_prep_args64): Correct long double handling.
1744         (ffi_closure_helper_LINUX64): Fix typo.
1745         * testsuite/libffi.call/cls_align_longdouble.c: Pass -mlong-double-128
1746         for powerpc64-*-*.
1747         * testsuite/libffi.call/float.c: Likewise.
1748         * testsuite/libffi.call/float2.c: Likewise.
1749
1750 2004-02-08  Alan Modra  <amodra@bigpond.net.au>
1751
1752         * src/powerpc/ffi.c (ffi_prep_cif_machdep <FFI_LINUX64>): Correct
1753         long double function return and long double arg handling.
1754         (ffi_closure_helper_LINUX64): Formatting.  Delete unused "ng" var.
1755         Use "end_pfr" instead of "nf".  Correct long double handling.
1756         Localise "temp".
1757         * src/powerpc/linux64.S (ffi_call_LINUX64): Save f2 long double
1758         return value.
1759         * src/powerpc/linux64_closure.S (ffi_closure_LINUX64): Allocate
1760         space for long double return value.  Adjust stack frame and offsets.
1761         Load f2 long double return.
1762
1763 2004-02-07  Alan Modra  <amodra@bigpond.net.au>
1764
1765         * src/types.c: Use 16 byte long double for POWERPC64.
1766
1767 2004-01-25  Eric Botcazou  <ebotcazou@libertysurf.fr>
1768
1769         * src/sparc/ffi.c (ffi_prep_args_v9): Shift the parameter array
1770         when the structure return address is passed in %o0.
1771         (ffi_V9_return_struct): Rename into ffi_v9_layout_struct.
1772         (ffi_v9_layout_struct): Align the field following a nested structure
1773         on a word boundary.  Use memmove instead of memcpy.
1774         (ffi_call): Update call to ffi_V9_return_struct.
1775         (ffi_prep_closure): Define 'ctx' only for V8.
1776         (ffi_closure_sparc_inner): Clone into ffi_closure_sparc_inner_v8
1777         and ffi_closure_sparc_inner_v9.
1778         (ffi_closure_sparc_inner_v8): Return long doubles by reference.
1779         Always skip the structure return address.  For structures and long
1780         doubles, copy the argument directly.
1781         (ffi_closure_sparc_inner_v9): Skip the structure return address only
1782         if required.  Shift the maximum floating-point slot accordingly.  For
1783         big structures, copy the argument directly; otherwise, left-justify the
1784         argument and call ffi_v9_layout_struct to lay out the structure on
1785         the stack.
1786         * src/sparc/v8.S: Undef STACKFRAME before defining it.
1787         (ffi_closure_v8): Pass the structure return address.  Update call to
1788         ffi_closure_sparc_inner_v8.  Short-circuit FFI_TYPE_INT handling.
1789         Skip the 'unimp' insn when returning long doubles and structures.
1790         * src/sparc/v9.S: Undef STACKFRAME before defining it.
1791         (ffi_closure_v9): Increase the frame size by 2 words.  Short-circuit
1792         FFI_TYPE_INT handling.  Load structures both in integers and
1793         floating-point registers on return.
1794         * README: Update status of the SPARC port.
1795
1796 2004-01-24  Andreas Tobler  <a.tobler@schweiz.ch>
1797
1798         * testsuite/libffi.call/pyobjc-tc.c (main): Treat result value
1799         as of type ffi_arg.
1800         * testsuite/libffi.call/struct3.c (main): Fix CHECK.
1801
1802 2004-01-22  Ulrich Weigand  <uweigand@de.ibm.com>
1803
1804         * testsuite/libffi.call/cls_uint.c (cls_ret_uint_fn): Treat result
1805         value as of type ffi_arg, not unsigned int.
1806
1807 2004-01-21  Michael Ritzert  <ritzert@t-online.de>
1808
1809         * ffi64.c (ffi_prep_args): Cast the RHS of an assignment instead
1810         of the LHS.
1811
1812 2004-01-12  Andreas Tobler  <a.tobler@schweiz.ch>
1813
1814         * testsuite/lib/libffi-dg.exp: Set LD_LIBRARY_PATH_32 for
1815         Solaris.
1816
1817 2004-01-08  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
1818
1819         * testsuite/libffi.call/ffitest.h (allocate_mmap): Cast MAP_FAILED
1820         to void *.
1821
1822 2003-12-10  Richard Henderson  <rth@redhat.com>
1823
1824         * testsuite/libffi.call/cls_align_pointer.c: Cast pointers to
1825         size_t instead of int.
1826
1827 2003-12-04  Hosaka Yuji  <hos@tamanegi.org>
1828
1829         * testsuite/libffi.call/many_win32.c: Include <float.h>.
1830         * testsuite/libffi.call/many_win32.c (main): Replace variable
1831         int i with unsigned long ul.
1832
1833         * testsuite/libffi.call/cls_align_uint64.c: New test case.
1834         * testsuite/libffi.call/cls_align_sint64.c: Likewise.
1835         * testsuite/libffi.call/cls_align_uint32.c: Likewise.
1836         * testsuite/libffi.call/cls_align_sint32.c: Likewise.
1837         * testsuite/libffi.call/cls_align_uint16.c: Likewise.
1838         * testsuite/libffi.call/cls_align_sint16.c: Likewise.
1839         * testsuite/libffi.call/cls_align_float.c: Likewise.
1840         * testsuite/libffi.call/cls_align_double.c: Likewise.
1841         * testsuite/libffi.call/cls_align_longdouble.c: Likewise.
1842         * testsuite/libffi.call/cls_align_pointer.c: Likewise.
1843
1844 2003-12-02  Hosaka Yuji  <hos@tamanegi.org>
1845
1846         PR other/13221
1847         * src/x86/ffi.c (ffi_prep_args, ffi_prep_incoming_args_SYSV):
1848         Align arguments to 32 bits.
1849
1850 2003-12-01  Andreas Tobler  <a.tobler@schweiz.ch>
1851
1852         PR other/13221
1853         * testsuite/libffi.call/cls_multi_sshort.c: New test case.
1854         * testsuite/libffi.call/cls_multi_sshortchar.c: Likewise.
1855         * testsuite/libffi.call/cls_multi_uchar.c: Likewise.
1856         * testsuite/libffi.call/cls_multi_schar.c: Likewise.
1857         * testsuite/libffi.call/cls_multi_ushortchar.c: Likewise.
1858         * testsuite/libffi.call/cls_multi_ushort.c: Likewise.
1859
1860         * testsuite/libffi.special/unwindtest.cc: Cosmetics.
1861
1862 2003-11-26  Kaveh R. Ghazi  <ghazi@caip.rutgers.edu>
1863
1864         * testsuite/libffi.call/ffitest.h: Include <fcntl.h>.
1865         * testsuite/libffi.special/ffitestcxx.h: Likewise.
1866
1867 2003-11-22  Andreas Tobler  <a.tobler@schweiz.ch>
1868
1869         * Makefile.in: Rebuilt.
1870         * configure: Likewise.
1871         * testsuite/libffi.special/unwindtest.cc: Convert the mmap to
1872         the right type.
1873
1874 2003-11-21  Andreas Jaeger  <aj@suse.de>
1875             Andreas Tobler  <a.tobler@schweiz.ch>
1876
1877         * acinclude.m4: Add AC_FUNC_MMAP_BLACKLIST.
1878         * configure.in: Call AC_FUNC_MMAP_BLACKLIST.
1879         * Makefile.in: Rebuilt.
1880         * aclocal.m4: Likewise.
1881         * configure: Likewise.
1882         * fficonfig.h.in: Likewise.
1883         * testsuite/lib/libffi-dg.exp: Add include dir.
1884         * testsuite/libffi.call/ffitest.h: Add MMAP definitions.
1885         * testsuite/libffi.special/ffitestcxx.h: Likewise.
1886         * testsuite/libffi.call/closure_fn0.c: Use MMAP functionality
1887         for ffi_closure if available.
1888         * testsuite/libffi.call/closure_fn1.c: Likewise.
1889         * testsuite/libffi.call/closure_fn2.c: Likewise.
1890         * testsuite/libffi.call/closure_fn3.c: Likewise.
1891         * testsuite/libffi.call/closure_fn4.c: Likewise.
1892         * testsuite/libffi.call/closure_fn5.c: Likewise.
1893         * testsuite/libffi.call/cls_12byte.c: Likewise.
1894         * testsuite/libffi.call/cls_16byte.c: Likewise.
1895         * testsuite/libffi.call/cls_18byte.c: Likewise.
1896         * testsuite/libffi.call/cls_19byte.c: Likewise.
1897         * testsuite/libffi.call/cls_1_1byte.c: Likewise.
1898         * testsuite/libffi.call/cls_20byte.c: Likewise.
1899         * testsuite/libffi.call/cls_20byte1.c: Likewise.
1900         * testsuite/libffi.call/cls_24byte.c: Likewise.
1901         * testsuite/libffi.call/cls_2byte.c: Likewise.
1902         * testsuite/libffi.call/cls_3_1byte.c: Likewise.
1903         * testsuite/libffi.call/cls_3byte1.c: Likewise.
1904         * testsuite/libffi.call/cls_3byte2.c: Likewise.
1905         * testsuite/libffi.call/cls_4_1byte.c: Likewise.
1906         * testsuite/libffi.call/cls_4byte.c: Likewise.
1907         * testsuite/libffi.call/cls_5byte.c: Likewise.
1908         * testsuite/libffi.call/cls_64byte.c: Likewise.
1909         * testsuite/libffi.call/cls_6byte.c: Likewise.
1910         * testsuite/libffi.call/cls_7byte.c: Likewise.
1911         * testsuite/libffi.call/cls_8byte.c: Likewise.
1912         * testsuite/libffi.call/cls_9byte1.c: Likewise.
1913         * testsuite/libffi.call/cls_9byte2.c: Likewise.
1914         * testsuite/libffi.call/cls_double.c: Likewise.
1915         * testsuite/libffi.call/cls_float.c: Likewise.
1916         * testsuite/libffi.call/cls_schar.c: Likewise.
1917         * testsuite/libffi.call/cls_sint.c: Likewise.
1918         * testsuite/libffi.call/cls_sshort.c: Likewise.
1919         * testsuite/libffi.call/cls_uchar.c: Likewise.
1920         * testsuite/libffi.call/cls_uint.c: Likewise.
1921         * testsuite/libffi.call/cls_ulonglong.c: Likewise.
1922         * testsuite/libffi.call/cls_ushort.c: Likewise.
1923         * testsuite/libffi.call/nested_struct.c: Likewise.
1924         * testsuite/libffi.call/nested_struct1.c: Likewise.
1925         * testsuite/libffi.call/nested_struct2.c: Likewise.
1926         * testsuite/libffi.call/nested_struct3.c: Likewise.
1927         * testsuite/libffi.call/problem1.c: Likewise.
1928         * testsuite/libffi.special/unwindtest.cc: Likewise.
1929
1930 2003-11-20  Andreas Tobler  <a.tobler@schweiz.ch>
1931
1932         * testsuite/lib/libffi-dg.exp: Make the -lgcc_s conditional.
1933
1934 2003-11-19  Andreas Tobler  <a.tobler@schweiz.ch>
1935
1936         * testsuite/lib/libffi-dg.exp: Add DYLD_LIBRARY_PATH for darwin.
1937         Add -lgcc_s to additional flags.
1938
1939 2003-11-12  Andreas Tobler  <a.tobler@schweiz.ch>
1940
1941         * configure.in, include/Makefile.am: PR libgcj/11147, install
1942         the ffitarget.h header file in a gcc versioned and target
1943         dependent place.
1944         * configure: Regenerated.
1945         * Makefile.in, include/Makefile.in: Likewise.
1946         * testsuite/Makefile.in: Likewise.
1947
1948 2003-11-09  Andreas Tobler  <a.tobler@schweiz.ch>
1949
1950         * testsuite/libffi.call/closure_fn0.c: Print result and check
1951         with dg-output to make debugging easier.
1952         * testsuite/libffi.call/closure_fn1.c: Likewise.
1953         * testsuite/libffi.call/closure_fn2.c: Likewise.
1954         * testsuite/libffi.call/closure_fn3.c: Likewise.
1955         * testsuite/libffi.call/closure_fn4.c: Likewise.
1956         * testsuite/libffi.call/closure_fn5.c: Likewise.
1957         * testsuite/libffi.call/cls_12byte.c: Likewise.
1958         * testsuite/libffi.call/cls_16byte.c: Likewise.
1959         * testsuite/libffi.call/cls_18byte.c: Likewise.
1960         * testsuite/libffi.call/cls_19byte.c: Likewise.
1961         * testsuite/libffi.call/cls_1_1byte.c: Likewise.
1962         * testsuite/libffi.call/cls_20byte.c: Likewise.
1963         * testsuite/libffi.call/cls_20byte1.c: Likewise.
1964         * testsuite/libffi.call/cls_24byte.c: Likewise.
1965         * testsuite/libffi.call/cls_2byte.c: Likewise.
1966         * testsuite/libffi.call/cls_3_1byte.c: Likewise.
1967         * testsuite/libffi.call/cls_3byte1.c: Likewise.
1968         * testsuite/libffi.call/cls_3byte2.c: Likewise.
1969         * testsuite/libffi.call/cls_4_1byte.c: Likewise.
1970         * testsuite/libffi.call/cls_4byte.c: Likewise.
1971         * testsuite/libffi.call/cls_5byte.c: Likewise.
1972         * testsuite/libffi.call/cls_64byte.c: Likewise.
1973         * testsuite/libffi.call/cls_6byte.c: Likewise.
1974         * testsuite/libffi.call/cls_7byte.c: Likewise.
1975         * testsuite/libffi.call/cls_8byte.c: Likewise.
1976         * testsuite/libffi.call/cls_9byte1.c: Likewise.
1977         * testsuite/libffi.call/cls_9byte2.c: Likewise.
1978         * testsuite/libffi.call/cls_double.c: Likewise.
1979         * testsuite/libffi.call/cls_float.c: Likewise.
1980         * testsuite/libffi.call/cls_schar.c: Likewise.
1981         * testsuite/libffi.call/cls_sint.c: Likewise.
1982         * testsuite/libffi.call/cls_sshort.c: Likewise.
1983         * testsuite/libffi.call/cls_uchar.c: Likewise.
1984         * testsuite/libffi.call/cls_uint.c: Likewise.
1985         * testsuite/libffi.call/cls_ulonglong.c: Likewise.
1986         * testsuite/libffi.call/cls_ushort.c: Likewise.
1987         * testsuite/libffi.call/problem1.c: Likewise.
1988
1989         * testsuite/libffi.special/unwindtest.cc: Make ffi_closure
1990         static.
1991
1992 2003-11-08  Andreas Tobler  <a.tobler@schweiz.ch>
1993
1994         * testsuite/libffi.call/cls_9byte2.c: New test case.
1995         * testsuite/libffi.call/cls_9byte1.c: Likewise.
1996         * testsuite/libffi.call/cls_64byte.c: Likewise.
1997         * testsuite/libffi.call/cls_20byte1.c: Likewise.
1998         * testsuite/libffi.call/cls_19byte.c: Likewise.
1999         * testsuite/libffi.call/cls_18byte.c: Likewise.
2000         * testsuite/libffi.call/closure_fn4.c: Likewise.
2001         * testsuite/libffi.call/closure_fn5.c: Likewise.
2002         * testsuite/libffi.call/cls_schar.c: Likewise.
2003         * testsuite/libffi.call/cls_sint.c: Likewise.
2004         * testsuite/libffi.call/cls_sshort.c: Likewise.
2005         * testsuite/libffi.call/nested_struct2.c: Likewise.
2006         * testsuite/libffi.call/nested_struct3.c: Likewise.
2007
2008 2003-11-08  Andreas Tobler  <a.tobler@schweiz.ch>
2009
2010         * testsuite/libffi.call/cls_double.c: Do a check on the result.
2011         * testsuite/libffi.call/cls_uchar.c: Likewise.
2012         * testsuite/libffi.call/cls_uint.c: Likewise.
2013         * testsuite/libffi.call/cls_ulonglong.c: Likewise.
2014         * testsuite/libffi.call/cls_ushort.c: Likewise.
2015         * testsuite/libffi.call/return_sc.c: Cleanup whitespaces.
2016
2017 2003-11-06  Andreas Tobler  <a.tobler@schweiz.ch>
2018
2019         * src/prep_cif.c (ffi_prep_cif): Move the validity check after
2020         the initialization.
2021
2022 2003-10-23  Andreas Tobler  <a.tobler@schweiz.ch>
2023
2024         * src/java_raw_api.c (ffi_java_ptrarray_to_raw): Replace
2025         FFI_ASSERT(FALSE) with FFI_ASSERT(0).
2026
2027 2003-10-22  David Daney  <ddaney@avtrex.com>
2028
2029         * src/mips/ffitarget.h: Replace undefined UINT32 and friends with
2030         __attribute__((__mode__(__SI__))) and friends.
2031
2032 2003-10-22  Andreas Schwab  <schwab@suse.de>
2033
2034         * src/ia64/ffi.c: Replace FALSE/TRUE with false/true.
2035
2036 2003-10-21  Andreas Tobler  <a.tobler@schweiz.ch>
2037
2038         * configure.in: AC_LINK_FILES(ffitarget.h).
2039         * configure: Regenerate.
2040         * Makefile.in: Likewise.
2041         * include/Makefile.in: Likewise.
2042         * testsuite/Makefile.in: Likewise.
2043         * fficonfig.h.in: Likewise.
2044
2045 2003-10-21  Paolo Bonzini  <bonzini@gnu.org>
2046             Richard Henderson  <rth@redhat.com>
2047
2048         Avoid that ffi.h includes fficonfig.h.
2049
2050         * Makefile.am (EXTRA_DIST): Include ffitarget.h files
2051         (TARGET_SRC_MIPS_GCC): Renamed to TARGET_SRC_MIPS_IRIX.
2052         (TARGET_SRC_MIPS_SGI): Removed.
2053         (MIPS_GCC): Renamed to TARGET_SRC_MIPS_IRIX.
2054         (MIPS_SGI): Removed.
2055         (CLEANFILES): Removed.
2056         (mostlyclean-am, clean-am, mostlyclean-sub, clean-sub): New
2057         targets.
2058         * acconfig.h: Removed.
2059         * configure.in: Compute sizeofs only for double and long double.
2060         Use them to define and subst HAVE_LONG_DOUBLE.  Include comments
2061         into AC_DEFINE instead of using acconfig.h.  Create
2062         include/ffitarget.h instead of include/fficonfig.h.  Rename
2063         MIPS_GCC to MIPS_IRIX, drop MIPS_SGI since we are in gcc's tree.
2064         AC_DEFINE EH_FRAME_FLAGS.
2065         * include/Makefile.am (DISTCLEANFILES): New automake macro.
2066         (hack_DATA): Add ffitarget.h.
2067         * include/ffi.h.in: Remove all system specific definitions.
2068         Declare raw API even if it is not installed, why bother?
2069         Use limits.h instead of SIZEOF_* to define ffi_type_*.  Do
2070         not define EH_FRAME_FLAGS, it is in fficonfig.h now.  Include
2071         ffitarget.h instead of fficonfig.h.  Remove ALIGN macro.
2072         (UINT_ARG, INT_ARG): Removed, use ffi_arg and ffi_sarg instead.
2073         * include/ffi_common.h (bool): Do not define.
2074         (ffi_assert): Accept failed assertion.
2075         (ffi_type_test): Return void and accept file/line.
2076         (FFI_ASSERT): Pass stringized failed assertion.
2077         (FFI_ASSERT_AT): New macro.
2078         (FFI_ASSERT_VALID_TYPE): New macro.
2079         (UINT8, SINT8, UINT16, SINT16, UINT32, SINT32,
2080         UINT64, SINT64): Define here with gcc's __attribute__ macro
2081         instead of in ffi.h
2082         (FLOAT32, ALIGN): Define here instead of in ffi.h
2083         * include/ffi-mips.h: Removed.  Its content moved to
2084         src/mips/ffitarget.h after separating assembly and C sections.
2085         * src/alpha/ffi.c, src/alpha/ffi.c, src/java_raw_api.c
2086         src/prep_cif.c, src/raw_api.c, src/ia64/ffi.c,
2087         src/mips/ffi.c, src/mips/n32.S, src/mips/o32.S,
2088         src/mips/ffitarget.h, src/sparc/ffi.c, src/x86/ffi64.c:
2089         SIZEOF_ARG -> FFI_SIZEOF_ARG.
2090         * src/ia64/ffi.c: Include stdbool.h (provided by GCC 2.95+).
2091         * src/debug.c (ffi_assert): Accept stringized failed assertion.
2092         (ffi_type_test): Rewritten.
2093         * src/prep-cif.c (initialize_aggregate, ffi_prep_cif): Call
2094         FFI_ASSERT_VALID_TYPE.
2095         * src/alpha/ffitarget.h, src/arm/ffitarget.h,
2096         src/ia64/ffitarget.h, src/m68k/ffitarget.h,
2097         src/mips/ffitarget.h, src/powerpc/ffitarget.h,
2098         src/s390/ffitarget.h, src/sh/ffitarget.h,
2099         src/sh64/ffitarget.h, src/sparc/ffitarget.h,
2100         src/x86/ffitarget.h: New files.
2101         * src/alpha/osf.S, src/arm/sysv.S, src/ia64/unix.S,
2102         src/m68k/sysv.S, src/mips/n32.S, src/mips/o32.S,
2103         src/powerpc/aix.S, src/powerpc/darwin.S,
2104         src/powerpc/ffi_darwin.c, src/powerpc/linux64.S,
2105         src/powerpc/linux64_closure.S, src/powerpc/ppc_closure.S,
2106         src/powerpc/sysv.S, src/s390/sysv.S, src/sh/sysv.S,
2107         src/sh64/sysv.S, src/sparc/v8.S, src/sparc/v9.S,
2108         src/x86/sysv.S, src/x86/unix64.S, src/x86/win32.S:
2109         include fficonfig.h
2110
2111 2003-10-20  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
2112
2113         * src/mips/ffi.c: Use _ABIN32, _ABIO32 instead of external
2114         _MIPS_SIM_NABI32, _MIPS_SIM_ABI32.
2115
2116 2003-10-19  Andreas Tobler  <a.tobler@schweiz.ch>
2117
2118         * src/powerpc/ffi_darwin.c (ffi_prep_args): Declare bytes again.
2119         Used when FFI_DEBUG = 1.
2120
2121 2003-10-14  Alan Modra  <amodra@bigpond.net.au>
2122
2123         * src/types.c (double, longdouble): Default POWERPC64 to 8 byte size
2124         and align.
2125
2126 2003-10-06  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
2127
2128         * include/ffi_mips.h: Define FFI_MIPS_N32 for N32/N64 ABIs,
2129         FFI_MIPS_O32 for O32 ABI.
2130
2131 2003-10-01  Andreas Tobler  <a.tobler@schweiz.ch>
2132
2133         * testsuite/lib/libffi-dg.exp: Set LD_LIBRARY_PATH_64 for
2134         SPARC64. Cleanup whitespaces.
2135
2136 2003-09-19  Andreas Tobler  <a.tobler@schweiz.ch>
2137
2138         * testsuite/libffi.call/closure_fn0.c: Xfail mips, arm,
2139         strongarm, xscale. Cleanup whitespaces.
2140         * testsuite/libffi.call/closure_fn1.c: Likewise.
2141         * testsuite/libffi.call/closure_fn2.c: Likewise.
2142         * testsuite/libffi.call/closure_fn3.c: Likewise.
2143         * testsuite/libffi.call/cls_12byte.c: Likewise.
2144         * testsuite/libffi.call/cls_16byte.c: Likewise.
2145         * testsuite/libffi.call/cls_1_1byte.c: Likewise.
2146         * testsuite/libffi.call/cls_20byte.c: Likewise.
2147         * testsuite/libffi.call/cls_24byte.c: Likewise.
2148         * testsuite/libffi.call/cls_2byte.c: Likewise.
2149         * testsuite/libffi.call/cls_3_1byte.c: Likewise.
2150         * testsuite/libffi.call/cls_3byte1.c: Likewise.
2151         * testsuite/libffi.call/cls_3byte2.c: Likewise.
2152         * testsuite/libffi.call/cls_4_1byte.c: Likewise.
2153         * testsuite/libffi.call/cls_4byte.c: Likewise.
2154         * testsuite/libffi.call/cls_5byte.c: Likewise.
2155         * testsuite/libffi.call/cls_6byte.c: Likewise.
2156         * testsuite/libffi.call/cls_7byte.c: Likewise.
2157         * testsuite/libffi.call/cls_8byte.c: Likewise.
2158         * testsuite/libffi.call/cls_double.c: Likewise.
2159         * testsuite/libffi.call/cls_float.c: Likewise.
2160         * testsuite/libffi.call/cls_uchar.c: Likewise.
2161         * testsuite/libffi.call/cls_uint.c: Likewise.
2162         * testsuite/libffi.call/cls_ulonglong.c: Likewise.
2163         * testsuite/libffi.call/cls_ushort.c: Likewise.
2164         * testsuite/libffi.call/nested_struct.c: Likewise.
2165         * testsuite/libffi.call/nested_struct1.c: Likewise.
2166         * testsuite/libffi.call/problem1.c: Likewise.
2167         * testsuite/libffi.special/unwindtest.cc: Likewise.
2168         * testsuite/libffi.call/pyobjc-tc.c: Cleanup whitespaces.
2169
2170 2003-09-18  David Edelsohn  <edelsohn@gnu.org>
2171
2172         * src/powerpc/aix.S: Cleanup whitespaces.
2173         * src/powerpc/aix_closure.S: Likewise.
2174
2175 2003-09-18  Andreas Tobler  <a.tobler@schweiz.ch>
2176
2177         * src/powerpc/darwin.S: Cleanup whitespaces, comment formatting.
2178         * src/powerpc/darwin_closure.S: Likewise.
2179         * src/powerpc/ffi_darwin.c: Likewise.
2180
2181 2003-09-18  Andreas Tobler  <a.tobler@schweiz.ch>
2182             David Edelsohn  <edelsohn@gnu.org>
2183
2184         * src/types.c (double): Add AIX and Darwin to the right TYPEDEF.
2185         * src/powerpc/aix_closure.S: Remove the pointer to the outgoing
2186         parameter stack.
2187         * src/powerpc/darwin_closure.S: Likewise.
2188         * src/powerpc/ffi_darwin.c (ffi_prep_args): Handle structures
2189         according to the Darwin/AIX ABI.
2190         (ffi_prep_cif_machdep): Likewise.
2191         (ffi_closure_helper_DARWIN): Likewise.
2192         Remove the outgoing parameter stack logic. Simplify the evaluation
2193         of the different CASE types.
2194         (ffi_prep_clousure): Avoid the casts on lvalues. Change the branch
2195         statement in the trampoline code.
2196
2197 2003-09-18  Kaz Kojima  <kkojima@gcc.gnu.org>
2198
2199         * src/sh/ffi.c (ffi_prep_args): Take account into the alignement
2200         for the register size.
2201         (ffi_closure_helper_SYSV): Handle the structure return value
2202         address correctly.
2203         (ffi_closure_helper_SYSV): Return the appropriate type when
2204         the registers are used for the structure return value.
2205         * src/sh/sysv.S (ffi_closure_SYSV): Fix the stack layout for
2206         the 64-bit return value.  Update copyright years.
2207
2208 2003-09-17  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
2209
2210         * testsuite/lib/libffi-dg.exp (libffi_target_compile): Search in
2211         srcdir for ffi_mips.h.
2212
2213 2003-09-12  Alan Modra  <amodra@bigpond.net.au>
2214
2215         * src/prep_cif.c (initialize_aggregate): Include tail padding in
2216         structure size.
2217         * src/powerpc/linux64_closure.S (ffi_closure_LINUX64): Correct
2218         placement of float result.
2219         * testsuite/libffi.special/unwindtest.cc (closure_test_fn1): Correct
2220         cast of "resp" for big-endian 64 bit machines.
2221
2222 2003-09-11  Alan Modra  <amodra@bigpond.net.au>
2223
2224         * src/types.c (double, longdouble): Merge identical SH and ARM
2225         typedefs, and add POWERPC64.
2226         * src/powerpc/ffi.c (ffi_prep_args64): Correct next_arg calc for
2227         struct split over gpr and rest.
2228         (ffi_prep_cif_machdep): Correct intarg_count for structures.
2229         * src/powerpc/linux64.S (ffi_call_LINUX64): Fix gpr offsets.
2230
2231 2003-09-09  Andreas Tobler  <a.tobler@schweiz.ch>
2232
2233         * src/powerpc/ffi.c (ffi_closure_helper_SYSV) Handle struct
2234         passing correctly.
2235
2236 2003-09-09  Alan Modra  <amodra@bigpond.net.au>
2237
2238         * configure: Regenerate.
2239
2240 2003-09-04  Andreas Tobler  <a.tobler@schweiz.ch>
2241
2242         * Makefile.am: Remove build rules for ffitest.
2243         * Makefile.in: Rebuilt.
2244
2245 2003-09-04  Andreas Tobler  <a.tobler@schweiz.ch>
2246
2247         * src/java_raw_api.c: Include <stdlib.h> to fix compiler warning
2248         about implicit declaration of abort().
2249
2250 2003-09-04  Andreas Tobler  <a.tobler@schweiz.ch>
2251
2252         * Makefile.am: Add dejagnu test framework. Fixes PR other/11411.
2253         * Makefile.in: Rebuilt.
2254         * configure.in: Add dejagnu test framework.
2255         * configure: Rebuilt.
2256
2257         * testsuite/Makefile.am: New file.
2258         * testsuite/Makefile.in: Built
2259         * testsuite/lib/libffi-dg.exp: New file.
2260         * testsuite/config/default.exp: Likewise.
2261         * testsuite/libffi.call/call.exp: Likewise.
2262         * testsuite/libffi.call/ffitest.h: Likewise.
2263         * testsuite/libffi.call/closure_fn0.c: Likewise.
2264         * testsuite/libffi.call/closure_fn1.c: Likewise.
2265         * testsuite/libffi.call/closure_fn2.c: Likewise.
2266         * testsuite/libffi.call/closure_fn3.c: Likewise.
2267         * testsuite/libffi.call/cls_1_1byte.c: Likewise.
2268         * testsuite/libffi.call/cls_3_1byte.c: Likewise.
2269         * testsuite/libffi.call/cls_4_1byte.c: Likewise.
2270         * testsuite/libffi.call/cls_2byte.c: Likewise.
2271         * testsuite/libffi.call/cls_3byte1.c: Likewise.
2272         * testsuite/libffi.call/cls_3byte2.c: Likewise.
2273         * testsuite/libffi.call/cls_4byte.c: Likewise.
2274         * testsuite/libffi.call/cls_5byte.c: Likewise.
2275         * testsuite/libffi.call/cls_6byte.c: Likewise.
2276         * testsuite/libffi.call/cls_7byte.c: Likewise.
2277         * testsuite/libffi.call/cls_8byte.c: Likewise.
2278         * testsuite/libffi.call/cls_12byte.c: Likewise.
2279         * testsuite/libffi.call/cls_16byte.c: Likewise.
2280         * testsuite/libffi.call/cls_20byte.c: Likewise.
2281         * testsuite/libffi.call/cls_24byte.c: Likewise.
2282         * testsuite/libffi.call/cls_double.c: Likewise.
2283         * testsuite/libffi.call/cls_float.c: Likewise.
2284         * testsuite/libffi.call/cls_uchar.c: Likewise.
2285         * testsuite/libffi.call/cls_uint.c: Likewise.
2286         * testsuite/libffi.call/cls_ulonglong.c: Likewise.
2287         * testsuite/libffi.call/cls_ushort.c: Likewise.
2288         * testsuite/libffi.call/float.c: Likewise.
2289         * testsuite/libffi.call/float1.c: Likewise.
2290         * testsuite/libffi.call/float2.c: Likewise.
2291         * testsuite/libffi.call/many.c: Likewise.
2292         * testsuite/libffi.call/many_win32.c: Likewise.
2293         * testsuite/libffi.call/nested_struct.c: Likewise.
2294         * testsuite/libffi.call/nested_struct1.c: Likewise.
2295         * testsuite/libffi.call/pyobjc-tc.c: Likewise.
2296         * testsuite/libffi.call/problem1.c: Likewise.
2297         * testsuite/libffi.call/promotion.c: Likewise.
2298         * testsuite/libffi.call/return_ll.c: Likewise.
2299         * testsuite/libffi.call/return_sc.c: Likewise.
2300         * testsuite/libffi.call/return_uc.c: Likewise.
2301         * testsuite/libffi.call/strlen.c: Likewise.
2302         * testsuite/libffi.call/strlen_win32.c: Likewise.
2303         * testsuite/libffi.call/struct1.c: Likewise.
2304         * testsuite/libffi.call/struct2.c: Likewise.
2305         * testsuite/libffi.call/struct3.c: Likewise.
2306         * testsuite/libffi.call/struct4.c: Likewise.
2307         * testsuite/libffi.call/struct5.c: Likewise.
2308         * testsuite/libffi.call/struct6.c: Likewise.
2309         * testsuite/libffi.call/struct7.c: Likewise.
2310         * testsuite/libffi.call/struct8.c: Likewise.
2311         * testsuite/libffi.call/struct9.c: Likewise.
2312         * testsuite/libffi.special/special.exp: New file.
2313         * testsuite/libffi.special/ffitestcxx.h: Likewise.
2314         * testsuite/libffi.special/unwindtest.cc: Likewise.
2315
2316
2317 2003-08-13  Kaz Kojima  <kkojima@gcc.gnu.org>
2318
2319         * src/sh/ffi.c (OFS_INT16): Set 0 for little endian case.  Update
2320         copyright years.
2321
2322 2003-08-02  Alan Modra  <amodra@bigpond.net.au>
2323
2324         * src/powerpc/ffi.c (ffi_prep_args64): Modify for changed gcc
2325         structure passing.
2326         (ffi_closure_helper_LINUX64): Likewise.
2327         * src/powerpc/linux64.S: Remove code writing to parm save area.
2328         * src/powerpc/linux64_closure.S (ffi_closure_LINUX64): Use return
2329         address in lr from ffi_closure_helper_LINUX64 call to calculate
2330         table address.  Optimize function tail.
2331
2332 2003-07-28  Andreas Tobler  <a.tobler@schweiz.ch>
2333
2334         * src/sparc/ffi.c: Handle all floating point registers.
2335         * src/sparc/v9.S: Likewise. Fixes second part of PR target/11410.
2336
2337 2003-07-11  Gerald Pfeifer  <pfeifer@dbai.tuwien.ac.at>
2338
2339         * README: Note that libffi is not part of GCC.  Update the project
2340         URL and status.
2341
2342 2003-06-19  Franz Sirl  <Franz.Sirl-kernel@lauterbach.com>
2343
2344         * src/powerpc/ppc_closure.S: Include ffi.h.
2345
2346 2003-06-13  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
2347
2348         * src/x86/sysv.S: Avoid gas-only .uleb128/.sleb128 directives.
2349         Use C style comments.
2350
2351 2003-06-13  Kaz Kojima  <kkojima@rr.iij4u.or.jp>
2352
2353         * Makefile.am: Add SHmedia support.  Fix a typo of SH support.
2354         * Makefile.in: Regenerate.
2355         * configure.in (sh64-*-linux*, sh5*-*-linux*): Add target.
2356         * configure: Regenerate.
2357         * include/ffi.h.in: Add SHmedia support.
2358         * src/sh64/ffi.c: New file.
2359         * src/sh64/sysv.S: New file.
2360
2361 2003-05-16  Jakub Jelinek  <jakub@redhat.com>
2362
2363         * configure.in (HAVE_RO_EH_FRAME): Check whether .eh_frame section
2364         should be read-only.
2365         * configure: Rebuilt.
2366         * fficonfig.h.in: Rebuilt.
2367         * include/ffi.h.in (EH_FRAME_FLAGS): Define.
2368         * src/alpha/osf.S: Use EH_FRAME_FLAGS.
2369         * src/powerpc/linux64.S: Likewise.
2370         * src/powerpc/linux64_closure.S: Likewise.  Include ffi.h.
2371         * src/powerpc/sysv.S: Use EH_FRAME_FLAGS.  Use pcrel encoding
2372         if -fpic/-fPIC/-mrelocatable.
2373         * src/powerpc/powerpc_closure.S: Likewise.
2374         * src/sparc/v8.S: If HAVE_RO_EH_FRAME is defined, don't include
2375         #write in .eh_frame flags.
2376         * src/sparc/v9.S: Likewise.
2377         * src/x86/unix64.S: Use EH_FRAME_FLAGS.
2378         * src/x86/sysv.S: Likewise.  Use pcrel encoding if -fpic/-fPIC.
2379         * src/s390/sysv.S: Use EH_FRAME_FLAGS.  Include ffi.h.
2380
2381 2003-05-07  Jeff Sturm  <jsturm@one-point.com>
2382
2383         Fixes PR bootstrap/10656
2384         * configure.in (HAVE_AS_REGISTER_PSEUDO_OP): Test assembler
2385         support for .register pseudo-op.
2386         * src/sparc/v8.S: Use it.
2387         * fficonfig.h.in: Rebuilt.
2388         * configure: Rebuilt.
2389
2390 2003-04-18  Jakub Jelinek  <jakub@redhat.com>
2391
2392         * include/ffi.h.in (POWERPC64): Define if 64-bit.
2393         (enum ffi_abi): Add FFI_LINUX64 on POWERPC.
2394         Make it the default on POWERPC64.
2395         (FFI_TRAMPOLINE_SIZE): Define to 24 on POWERPC64.
2396         * configure.in: Change powerpc-*-linux* into powerpc*-*-linux*.
2397         * configure: Rebuilt.
2398         * src/powerpc/ffi.c (hidden): Define.
2399         (ffi_prep_args_SYSV): Renamed from
2400         ffi_prep_args.  Cast pointers to unsigned long to shut up warnings.
2401         (NUM_GPR_ARG_REGISTERS64, NUM_FPR_ARG_REGISTERS64,
2402         ASM_NEEDS_REGISTERS64): New.
2403         (ffi_prep_args64): New function.
2404         (ffi_prep_cif_machdep): Handle FFI_LINUX64 ABI.
2405         (ffi_call): Likewise.
2406         (ffi_prep_closure): Likewise.
2407         (flush_icache): Surround by #ifndef POWERPC64.
2408         (ffi_dblfl): New union type.
2409         (ffi_closure_helper_SYSV): Use it to avoid aliasing problems.
2410         (ffi_closure_helper_LINUX64): New function.
2411         * src/powerpc/ppc_closure.S: Surround whole file by #ifndef
2412         __powerpc64__.
2413         * src/powerpc/sysv.S: Likewise.
2414         (ffi_call_SYSV): Rename ffi_prep_args to ffi_prep_args_SYSV.
2415         * src/powerpc/linux64.S: New file.
2416         * src/powerpc/linux64_closure.S: New file.
2417         * Makefile.am (EXTRA_DIST): Add src/powerpc/linux64.S and
2418         src/powerpc/linux64_closure.S.
2419         (TARGET_SRC_POWERPC): Likewise.
2420
2421         * src/ffitest.c (closure_test_fn, closure_test_fn1, closure_test_fn2,
2422         closure_test_fn3): Fix result printing on big-endian 64-bit
2423         machines.
2424         (main): Print tst2_arg instead of uninitialized tst2_result.
2425
2426         * src/ffitest.c (main): Hide what closure pointer really points to
2427         from the compiler.
2428
2429 2003-04-16  Richard Earnshaw  <rearnsha@arm.com>
2430
2431         * configure.in (arm-*-netbsdelf*): Add configuration.
2432         (configure): Regenerated.
2433
2434 2003-04-04  Loren J. Rittle  <ljrittle@acm.org>
2435
2436         * include/Makefile.in: Regenerate.
2437
2438 2003-03-21  Zdenek Dvorak  <rakdver@atrey.karlin.mff.cuni.cz>
2439
2440         * libffi/include/ffi.h.in: Define X86 instead of X86_64 in 32
2441         bit mode.
2442         * libffi/src/x86/ffi.c (ffi_closure_SYSV, ffi_closure_raw_SYSV):
2443         Receive closure pointer through parameter, read args using
2444         __builtin_dwarf_cfa.
2445         (FFI_INIT_TRAMPOLINE): Send closure reference through eax.
2446
2447 2003-03-12  Andreas Schwab  <schwab@suse.de>
2448
2449         * configure.in: Avoid trailing /. in toolexeclibdir.
2450         * configure: Rebuilt.
2451
2452 2003-03-03  Andreas Tobler <a.tobler@schweiz.ch>
2453
2454         * src/powerpc/darwin_closure.S: Recode to fit dynamic libraries.
2455
2456 2003-02-06  Andreas Tobler <a.tobler@schweiz.ch>
2457
2458         * libffi/src/powerpc/darwin_closure.S:
2459         Fix alignement bug, allocate 8 bytes for the result.
2460         * libffi/src/powerpc/aix_closure.S:
2461         Likewise.
2462         * libffi/src/powerpc/ffi_darwin.c:
2463         Update stackframe description for aix/darwin_closure.S.
2464
2465 2003-02-06  Jakub Jelinek  <jakub@redhat.com>
2466
2467         * src/s390/ffi.c (ffi_closure_helper_SYSV): Add hidden visibility
2468         attribute.
2469
2470 2003-01-31  Christian Cornelssen  <ccorn@cs.tu-berlin.de>,
2471             Andreas Schwab  <schwab@suse.de>
2472
2473         * configure.in: Adjust command to source config-ml.in to account
2474         for changes to the libffi_basedir definition.
2475         (libffi_basedir): Remove ${srcdir} from value and include trailing
2476         slash if nonempty.
2477
2478         * configure: Regenerate.
2479
2480 2003-01-29  Franz Sirl  <Franz.Sirl-kernel@lauterbach.com>
2481
2482         * src/powerpc/ppc_closure.S: Recode to fit shared libs.
2483
2484 2003-01-28  Andrew Haley  <aph@redhat.com>
2485
2486         * include/ffi.h.in: Enable FFI_CLOSURES for x86_64.
2487         * src/x86/ffi64.c (ffi_prep_closure): New.
2488         (ffi_closure_UNIX64_inner): New.
2489         * src/x86/unix64.S (ffi_closure_UNIX64): New.
2490
2491 2003-01-27  Alexandre Oliva  <aoliva@redhat.com>
2492
2493         * configure.in (toolexecdir, toolexeclibdir): Set and AC_SUBST.
2494         Remove USE_LIBDIR conditional.
2495         * Makefile.am (toolexecdir, toolexeclibdir): Don't override.
2496         * Makefile.in, configure: Rebuilt.
2497
2498 2003-01027  David Edelsohn  <edelsohn@gnu.org>
2499
2500         * Makefile.am (TARGET_SRC_POWERPC_AIX): Fix typo.
2501         * Makefile.in: Regenerate.
2502
2503 2003-01-22  Andrew Haley  <aph@redhat.com>
2504
2505         * src/powerpc/darwin.S (_ffi_call_AIX): Add Augmentation size to
2506         unwind info.
2507
2508 2003-01-21  Andreas Tobler  <a.tobler@schweiz.ch>
2509
2510         * src/powerpc/darwin.S: Add unwind info.
2511         * src/powerpc/darwin_closure.S: Likewise.
2512
2513 2003-01-14  Andrew Haley  <aph@redhat.com>
2514
2515         * src/x86/ffi64.c (ffi_prep_args): Check for void retval.
2516         (ffi_prep_cif_machdep): Likewise.
2517         * src/x86/unix64.S: Add unwind info.
2518
2519 2003-01-14  Andreas Jaeger  <aj@suse.de>
2520
2521         * src/ffitest.c (main): Only use ffi_closures if those are
2522         supported.
2523
2524 2003-01-13 Andreas Tobler <a.tobler@schweiz.ch>
2525
2526         * libffi/src/ffitest.c
2527          add closure testcases
2528
2529 2003-01-13 Kevin B. Hendricks <khendricks@ivey.uwo.ca>
2530
2531         * libffi/src/powerpc/ffi.c
2532          fix alignment bug for float (4 byte aligned iso 8 byte)
2533
2534 2003-01-09  Geoffrey Keating  <geoffk@apple.com>
2535
2536         * src/powerpc/ffi_darwin.c: Remove RCS version string.
2537         * src/powerpc/darwin.S: Remove RCS version string.
2538
2539 2003-01-03  Jeff Sturm  <jsturm@one-point.com>
2540
2541         * include/ffi.h.in: Add closure defines for SPARC, SPARC64.
2542         * src/ffitest.c (main): Use static storage for closure.
2543         * src/sparc/ffi.c (ffi_prep_closure, ffi_closure_sparc_inner): New.
2544         * src/sparc/v8.S (ffi_closure_v8): New.
2545         * src/sparc/v9.S (ffi_closure_v9): New.
2546
2547 2002-11-10  Ranjit Mathew <rmathew@hotmail.com>
2548
2549         * include/ffi.h.in: Added FFI_STDCALL ffi_type
2550           enumeration for X86_WIN32.
2551         * src/x86/win32.S: Added ffi_call_STDCALL function
2552           definition.
2553         * src/x86/ffi.c (ffi_call/ffi_raw_call): Added
2554           switch cases for recognising FFI_STDCALL and
2555           calling ffi_call_STDCALL if target is X86_WIN32.
2556         * src/ffitest.c (my_stdcall_strlen/stdcall_many):
2557           stdcall versions of the "my_strlen" and "many"
2558           test functions (for X86_WIN32).
2559           Added test cases to test stdcall invocation using
2560           these functions.
2561
2562 2002-12-02  Kaz Kojima  <kkojima@gcc.gnu.org>
2563
2564         * src/sh/sysv.S: Add DWARF2 unwind info.
2565
2566 2002-11-27  Ulrich Weigand  <uweigand@de.ibm.com>
2567
2568         * src/s390/sysv.S (.eh_frame section): Make section read-only.
2569
2570 2002-11-26  Jim Wilson  <wilson@redhat.com>
2571
2572         * src/types.c (FFI_TYPE_POINTER): Has size 8 on IA64.
2573
2574 2002-11-23  H.J. Lu <hjl@gnu.org>
2575
2576         * acinclude.m4: Add dummy AM_PROG_LIBTOOL.
2577         Include ../config/accross.m4.
2578         * aclocal.m4; Rebuild.
2579         * configure: Likewise.
2580
2581 2002-11-15  Ulrich Weigand  <uweigand@de.ibm.com>
2582
2583         * src/s390/sysv.S (.eh_frame section): Adapt to pcrel FDE encoding.
2584
2585 2002-11-11  DJ Delorie  <dj@redhat.com>
2586
2587         * configure.in: Look for common files in the right place.
2588
2589 2002-10-08  Ulrich Weigand  <uweigand@de.ibm.com>
2590
2591         * src/java_raw_api.c (ffi_java_raw_to_ptrarray): Interpret
2592         raw data as _Jv_word values, not ffi_raw.
2593         (ffi_java_ptrarray_to_raw): Likewise.
2594         (ffi_java_rvalue_to_raw): New function.
2595         (ffi_java_raw_call): Call it.
2596         (ffi_java_raw_to_rvalue): New function.
2597         (ffi_java_translate_args): Call it.
2598         * src/ffitest.c (closure_test_fn): Interpret return value
2599         as ffi_arg, not int.
2600         * src/s390/ffi.c (ffi_prep_cif_machdep): Add missing
2601         FFI_TYPE_POINTER case.
2602         (ffi_closure_helper_SYSV): Likewise.  Also, assume return
2603         values extended to word size.
2604
2605 2002-10-02  Andreas Jaeger  <aj@suse.de>
2606
2607         * src/x86/ffi64.c (ffi_prep_cif_machdep): Remove debug output.
2608
2609 2002-10-01  Bo Thorsen  <bo@smetana.suse.de>
2610
2611         * include/ffi.h.in: Fix i386 win32 compilation.
2612
2613 2002-09-30  Ulrich Weigand  <uweigand@de.ibm.com>
2614
2615         * configure.in: Add s390x-*-linux-* target.
2616         * configure: Regenerate.
2617         * include/ffi.h.in: Define S390X for s390x targets.
2618         (FFI_CLOSURES): Define for s390/s390x.
2619         (FFI_TRAMPOLINE_SIZE): Likewise.
2620         (FFI_NATIVE_RAW_API): Likewise.
2621         * src/prep_cif.c (ffi_prep_cif): Do not compute stack space for s390.
2622         * src/types.c (FFI_TYPE_POINTER): Use 8-byte pointers on s390x.
2623         * src/s390/ffi.c: Major rework of existing code.  Add support for
2624         s390x targets.  Add closure support.
2625         * src/s390/sysv.S: Likewise.
2626
2627 2002-09-29  Richard Earnshaw  <rearnsha@arm.com>
2628
2629         * src/arm/sysv.S: Fix typo.
2630
2631 2002-09-28  Richard Earnshaw  <rearnsha@arm.com>
2632
2633         * src/arm/sysv.S: If we don't have machine/asm.h and the pre-processor
2634         has defined __USER_LABEL_PREFIX__, then use it in CNAME.
2635         (ffi_call_SYSV): Handle soft-float.
2636
2637 2002-09-27  Bo Thorsen  <bo@suse.de>
2638
2639         * include/ffi.h.in: Fix multilib x86-64 support.
2640
2641 2002-09-22  Kaveh R. Ghazi  <ghazi@caip.rutgers.edu>
2642
2643         * Makefile.am (all-multi): Fix multilib parallel build.
2644
2645 2002-07-19  Kaz Kojima  <kkojima@gcc.gnu.org>
2646
2647         * configure.in (sh[34]*-*-linux*): Add brackets.
2648         * configure: Regenerate.
2649
2650 2002-07-18  Kaz Kojima  <kkojima@gcc.gnu.org>
2651
2652         * Makefile.am: Add SH support.
2653         * Makefile.in: Regenerate.
2654         * configure.in (sh-*-linux*, sh[34]*-*-linux*): Add target.
2655         * configure: Regenerate.
2656         * include/ffi.h.in: Add SH support.
2657         * src/sh/ffi.c: New file.
2658         * src/sh/sysv.S: New file.
2659         * src/types.c: Add SH support.
2660
2661 2002-07-16  Bo Thorsen  <bo@suse.de>
2662
2663         * src/x86/ffi64.c: New file that adds x86-64 support.
2664         * src/x86/unix64.S: New file that handles argument setup for
2665         x86-64.
2666         * src/x86/sysv.S: Don't use this on x86-64.
2667         * src/x86/ffi.c: Don't use this on x86-64.
2668         Remove unused vars.
2669         * src/prep_cif.c (ffi_prep_cif): Don't do stack size calculation
2670         for x86-64.
2671         * src/ffitest.c (struct6): New test that tests a special case in
2672         the x86-64 ABI.
2673         (struct7): Likewise.
2674         (struct8): Likewise.
2675         (struct9): Likewise.
2676         (closure_test_fn): Silence warning about this when it's not used.
2677         (main): Add the new tests.
2678         (main): Fix a couple of wrong casts and silence some compiler warnings.
2679         * include/ffi.h.in: Add x86-64 ABI definition.
2680         * fficonfig.h.in: Regenerate.
2681         * Makefile.am: Add x86-64 support.
2682         * configure.in: Likewise.
2683         * Makefile.in: Regenerate.
2684         * configure: Likewise.
2685
2686 2002-06-24  Bo Thorsen  <bo@suse.de>
2687
2688         * src/types.c: Merge settings for similar architectures.
2689         Add x86-64 sizes and alignments.
2690
2691 2002-06-23  Bo Thorsen  <bo@suse.de>
2692
2693         * src/arm/ffi.c (ffi_prep_args): Remove unused vars.
2694         * src/sparc/ffi.c (ffi_prep_args_v8): Likewise.
2695         * src/mips/ffi.c (ffi_prep_args): Likewise.
2696         * src/m68k/ffi.c (ffi_prep_args): Likewise.
2697
2698 2002-07-18  H.J. Lu  (hjl@gnu.org)
2699
2700         * Makefile.am (TARGET_SRC_MIPS_LINUX): New.
2701         (libffi_la_SOURCES): Support MIPS_LINUX.
2702         (libffi_convenience_la_SOURCES): Likewise.
2703         * Makefile.in: Regenerated.
2704
2705         * configure.in (mips64*-*): Skip.
2706         (mips*-*-linux*): New.
2707         * configure: Regenerated.
2708
2709         * src/mips/ffi.c: Include <sgidefs.h>.
2710
2711 2002-06-06  Ulrich Weigand  <uweigand@de.ibm.com>
2712
2713         * src/s390/sysv.S: Save/restore %r6.  Add DWARF-2 unwind info.
2714
2715 2002-05-27  Roger Sayle  <roger@eyesopen.com>
2716
2717         * src/x86/ffi.c (ffi_prep_args): Remove reference to avn.
2718
2719 2002-05-27  Bo Thorsen  <bo@suse.de>
2720
2721         * src/x86/ffi.c (ffi_prep_args): Remove unused variable and
2722         fix formatting.
2723
2724 2002-05-13  Andreas Tobler  <a.tobler@schweiz.ch>
2725
2726         * src/powerpc/ffi_darwin.c (ffi_prep_closure): Declare fd at
2727         beginning of function (for older apple cc).
2728
2729 2002-05-08  Alexandre Oliva  <aoliva@redhat.com>
2730
2731         * configure.in (ORIGINAL_LD_FOR_MULTILIBS): Preserve LD at
2732         script entry, and set LD to it when configuring multilibs.
2733         * configure: Rebuilt.
2734
2735 2002-05-05  Jason Thorpe  <thorpej@wasabisystems.com>
2736
2737         * configure.in (sparc64-*-netbsd*): Add target.
2738         (sparc-*-netbsdelf*): Likewise.
2739         * configure: Regenerate.
2740
2741 2002-04-28  David S. Miller  <davem@redhat.com>
2742
2743         * configure.in, configure: Fix SPARC test in previous change.
2744
2745 2002-04-29  Gerhard Tonn  <GerhardTonn@swol.de>
2746
2747         * Makefile.am: Add Linux for S/390 support.
2748         * Makefile.in: Regenerate.
2749         * configure.in: Add Linux for S/390 support.
2750         * configure: Regenerate.
2751         * include/ffi.h.in: Add Linux for S/390 support.
2752         * src/s390/ffi.c: New file from libffi CVS tree.
2753         * src/s390/sysv.S: New file from libffi CVS tree.
2754
2755 2002-04-28  Jakub Jelinek  <jakub@redhat.com>
2756
2757         * configure.in (HAVE_AS_SPARC_UA_PCREL): Check for working
2758         %r_disp32().
2759         * src/sparc/v8.S: Use it.
2760         * src/sparc/v9.S: Likewise.
2761         * fficonfig.h.in: Rebuilt.
2762         * configure: Rebuilt.
2763
2764 2002-04-08  Hans Boehm  <Hans_Boehm@hp.com>
2765
2766         * src/java_raw_api.c (ffi_java_raw_size): Handle FFI_TYPE_DOUBLE
2767         correctly.
2768         * src/ia64/unix.S: Add unwind information. Fix comments.
2769         Save sp in a way that's compatible with unwind info.
2770         (ffi_call_unix): Correctly restore sp in all cases.
2771         * src/ia64/ffi.c: Add, fix comments.
2772
2773 2002-04-08  Jakub Jelinek  <jakub@redhat.com>
2774
2775         * src/sparc/v8.S: Make .eh_frame dependent on target word size.
2776
2777 2002-04-06  Jason Thorpe  <thorpej@wasabisystems.com>
2778
2779         * configure.in (alpha*-*-netbsd*): Add target.
2780         * configure: Regenerate.
2781
2782 2002-04-04  Jeff Sturm  <jsturm@one-point.com>
2783
2784         * src/sparc/v8.S: Add unwind info.
2785         * src/sparc/v9.S: Likewise.
2786
2787 2002-03-30  Krister Walfridsson  <cato@df.lth.se>
2788
2789         * configure.in: Enable i*86-*-netbsdelf*.
2790         * configure: Rebuilt.
2791
2792 2002-03-29  David Billinghurst <David.Billinghurst@riotinto.com>
2793
2794         PR other/2620
2795         * src/mips/n32.s: Delete
2796         * src/mips/o32.s: Delete
2797
2798 2002-03-21  Loren J. Rittle  <ljrittle@acm.org>
2799
2800         * configure.in: Enable alpha*-*-freebsd*.
2801         * configure: Rebuilt.
2802
2803 2002-03-17  Bryce McKinlay  <bryce@waitaki.otago.ac.nz>
2804
2805         * Makefile.am: libfficonvenience -> libffi_convenience.
2806         * Makefile.in: Rebuilt.
2807
2808         * Makefile.am: Define ffitest_OBJECTS.
2809         * Makefile.in: Rebuilt.
2810
2811 2002-03-07  Andreas Tobler  <toa@pop.agri.ch>
2812             David Edelsohn  <edelsohn@gnu.org>
2813
2814         * Makefile.am (EXTRA_DIST): Add Darwin and AIX closure files.
2815         (TARGET_SRC_POWERPC_AIX): Add aix_closure.S.
2816         (TARGET_SRC_POWERPC_DARWIN): Add darwin_closure.S.
2817         * Makefile.in: Regenerate.
2818         * include/ffi.h.in: Add AIX and Darwin closure definitions.
2819         * src/powerpc/ffi_darwin.c (ffi_prep_closure): New function.
2820         (flush_icache, flush_range): New functions.
2821         (ffi_closure_helper_DARWIN): New function.
2822         * src/powerpc/aix_closure.S: New file.
2823         * src/powerpc/darwin_closure.S: New file.
2824
2825 2002-02-24  Jeff Sturm  <jsturm@one-point.com>
2826
2827         * include/ffi.h.in: Add typedef for ffi_arg.
2828         * src/ffitest.c (main): Declare rint with ffi_arg.
2829
2830 2002-02-21  Andreas Tobler  <toa@pop.agri.ch>
2831
2832         * src/powerpc/ffi_darwin.c (ffi_prep_args): Skip appropriate
2833         number of GPRs for floating-point arguments.
2834
2835 2002-01-31  Anthony Green  <green@redhat.com>
2836
2837         * configure: Rebuilt.
2838         * configure.in: Replace CHECK_SIZEOF and endian tests with
2839         cross-compiler friendly macros.
2840         * aclocal.m4 (AC_COMPILE_CHECK_SIZEOF, AC_C_BIGENDIAN_CROSS): New
2841         macros.
2842
2843 2002-01-18  David Edelsohn  <edelsohn@gnu.org>
2844
2845         * src/powerpc/darwin.S (_ffi_call_AIX): New.
2846         * src/powerpc/aix.S (ffi_call_DARWIN): New.
2847
2848 2002-01-17  David Edelsohn  <edelsohn@gnu.org>
2849
2850         * Makefile.am (EXTRA_DIST): Add Darwin and AIX files.
2851         (TARGET_SRC_POWERPC_AIX): New.
2852         (POWERPC_AIX): New stanza.
2853         * Makefile.in: Regenerate.
2854         * configure.in: Add AIX case.
2855         * configure: Regenerate.
2856         * include/ffi.h.in (ffi_abi): Add FFI_AIX.
2857         * src/powerpc/ffi_darwin.c (ffi_status): Use "long" to scale frame
2858         size.  Fix "long double" support.
2859         (ffi_call): Add FFI_AIX case.
2860         * src/powerpc/aix.S: New.
2861
2862 2001-10-09  John Hornkvist  <john@toastedmarshmallow.com>
2863
2864         Implement Darwin PowerPC ABI.
2865         * configure.in: Handle powerpc-*-darwin*.
2866         * Makefile.am: Set source files for POWERPC_DARWIN.
2867         * configure: Rebuilt.
2868         * Makefile.in: Rebuilt.
2869         * include/ffi.h.in: Define FFI_DARWIN and FFI_DEFAULT_ABI for
2870         POWERPC_DARWIN.
2871         * src/powerpc/darwin.S: New file.
2872         * src/powerpc/ffi_darwin.c: New file.
2873
2874 2001-10-07  Joseph S. Myers  <jsm28@cam.ac.uk>
2875
2876         * src/x86/ffi.c: Fix spelling error of "separate" as "seperate".
2877
2878 2001-07-16  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
2879
2880         * src/x86/sysv.S: Avoid gas-only .balign directive.
2881         Use C style comments.
2882
2883 2001-07-16  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
2884
2885         * src/alpha/ffi.c (ffi_prep_closure): Avoid gas-only mnemonic.
2886         Fixes PR bootstrap/3563.
2887
2888 2001-06-26  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
2889
2890         * src/alpha/osf.S (ffi_closure_osf): Use .rdata for ECOFF.
2891
2892 2001-06-25  Rainer Orth  <ro@TechFak.Uni-Bielefeld.DE>
2893
2894         * configure.in: Recognize sparc*-sun-* host.
2895         * configure: Regenerate.
2896
2897 2001-06-06  Andrew Haley  <aph@redhat.com>
2898
2899         * src/alpha/osf.S (__FRAME_BEGIN__): Conditionalize for ELF.
2900
2901 2001-06-03  Andrew Haley  <aph@redhat.com>
2902
2903         * src/alpha/osf.S: Add unwind info.
2904         * src/powerpc/sysv.S: Add unwind info.
2905         * src/powerpc/ppc_closure.S: Likewise.
2906
2907 2000-05-31  Jeff Sturm  <jsturm@one-point.com>
2908
2909         * configure.in: Fix AC_ARG_ENABLE usage.
2910         * configure: Rebuilt.
2911
2912 2001-05-06  Bryce McKinlay  <bryce@waitaki.otago.ac.nz>
2913
2914         * configure.in: Remove warning about beta code.
2915         * configure: Rebuilt.
2916
2917 2001-04-25  Hans Boehm <Hans_Boehm@hp.com>
2918
2919         * src/ia64/unix.S: Restore stack pointer when returning from
2920         ffi_closure_UNIX.
2921         * src/ia64/ffi.c: Fix typo in comment.
2922
2923 2001-04-18  Jim Wilson  <wilson@redhat.com>
2924
2925         * src/ia64/unix.S: Delete unnecessary increment and decrement of loc2
2926         to eliminate RAW DV.
2927
2928 2001-04-12  Bryce McKinlay  <bryce@albatross.co.nz>
2929
2930         * Makefile.am: Make a libtool convenience library.
2931         * Makefile.in: Rebuilt.
2932
2933 2001-03-29  Bryce McKinlay  <bryce@albatross.co.nz>
2934
2935         * configure.in: Use different syntax for subdirectory creation.
2936         * configure: Rebuilt.
2937
2938 2001-03-27  Jon Beniston  <jon@beniston.com>
2939
2940         * configure.in: Added X86_WIN32 target (Win32, CygWin, MingW).
2941         * configure: Rebuilt.
2942         * Makefile.am: Added X86_WIN32 target support.
2943         * Makefile.in: Rebuilt.
2944
2945         * include/ffi.h.in: Added X86_WIN32 target support.
2946
2947         * src/ffitest.c: Doesn't run structure tests for X86_WIN32 targets.
2948         * src/types.c: Added X86_WIN32 target support.
2949
2950         * src/x86/win32.S: New file. Based on sysv.S, but with EH
2951         stuff removed and made to work with CygWin's gas.
2952
2953 2001-03-26  Bryce McKinlay  <bryce@albatross.co.nz>
2954
2955         * configure.in: Make target subdirectory in build dir.
2956         * Makefile.am: Override suffix based rules to specify correct output
2957         subdirectory.
2958         * Makefile.in: Rebuilt.
2959         * configure: Rebuilt.
2960
2961 2001-03-23  Kevin B Hendricks  <khendricks@ivey.uwo.ca>
2962
2963         * src/powerpc/ppc_closure.S: New file.
2964         * src/powerpc/ffi.c (ffi_prep_args): Fixed ABI compatibility bug
2965         involving long long and register pairs.
2966         (ffi_prep_closure): New function.
2967         (flush_icache): Likewise.
2968         (ffi_closure_helper_SYSV): Likewise.
2969         * include/ffi.h.in (FFI_CLOSURES): Define on PPC.
2970         (FFI_TRAMPOLINE_SIZE): Likewise.
2971         (FFI_NATIVE_RAW_API): Likewise.
2972         * Makefile.in: Rebuilt.
2973         * Makefile.am (EXTRA_DIST): Added src/powerpc/ppc_closure.S.
2974         (TARGET_SRC_POWERPC): Likewise.
2975
2976 2001-03-19  Tom Tromey  <tromey@redhat.com>
2977
2978         * Makefile.in: Rebuilt.
2979         * Makefile.am (ffitest_LDFLAGS): New macro.
2980
2981 2001-03-02  Nick Clifton  <nickc@redhat.com>
2982
2983         * include/ffi.h.in: Remove RCS ident string.
2984         * include/ffi_mips.h: Remove RCS ident string.
2985         * src/debug.c: Remove RCS ident string.
2986         * src/ffitest.c: Remove RCS ident string.
2987         * src/prep_cif.c: Remove RCS ident string.
2988         * src/types.c: Remove RCS ident string.
2989         * src/alpha/ffi.c: Remove RCS ident string.
2990         * src/alpha/osf.S: Remove RCS ident string.
2991         * src/arm/ffi.c: Remove RCS ident string.
2992         * src/arm/sysv.S: Remove RCS ident string.
2993         * src/mips/ffi.c: Remove RCS ident string.
2994         * src/mips/n32.S: Remove RCS ident string.
2995         * src/mips/o32.S: Remove RCS ident string.
2996         * src/sparc/ffi.c: Remove RCS ident string.
2997         * src/sparc/v8.S: Remove RCS ident string.
2998         * src/sparc/v9.S: Remove RCS ident string.
2999         * src/x86/ffi.c: Remove RCS ident string.
3000         * src/x86/sysv.S: Remove RCS ident string.
3001
3002 2001-02-08  Joseph S. Myers  <jsm28@cam.ac.uk>
3003
3004         * include/ffi.h.in: Change sourceware.cygnus.com references to
3005         gcc.gnu.org.
3006
3007 2000-12-09  Richard Henderson  <rth@redhat.com>
3008
3009         * src/alpha/ffi.c (ffi_call): Simplify struct return test.
3010         (ffi_closure_osf_inner): Index rather than increment avalue
3011         and arg_types.  Give ffi_closure_osf the raw return value type.
3012         * src/alpha/osf.S (ffi_closure_osf): Handle return value type
3013         promotion.
3014
3015 2000-12-07  Richard Henderson  <rth@redhat.com>
3016
3017         * src/raw_api.c (ffi_translate_args): Fix typo.
3018         (ffi_prep_closure): Likewise.
3019
3020         * include/ffi.h.in [ALPHA]: Define FFI_CLOSURES and
3021         FFI_TRAMPOLINE_SIZE.
3022         * src/alpha/ffi.c (ffi_prep_cif_machdep): Adjust minimal
3023         cif->bytes for new ffi_call_osf implementation.
3024         (ffi_prep_args): Absorb into ...
3025         (ffi_call): ... here.  Do all stack allocation here and
3026         avoid a callback function.
3027         (ffi_prep_closure, ffi_closure_osf_inner): New.
3028         * src/alpha/osf.S (ffi_call_osf): Reimplement with no callback.
3029         (ffi_closure_osf): New.
3030
3031 2000-09-10  Alexandre Oliva  <aoliva@redhat.com>
3032
3033         * config.guess, config.sub, install-sh: Removed.
3034         * ltconfig, ltmain.sh, missing, mkinstalldirs: Likewise.
3035         * Makefile.in: Rebuilt.
3036
3037         * acinclude.m4: Include libtool macros from the top level.
3038         * aclocal.m4, configure: Rebuilt.
3039
3040 2000-08-22  Alexandre Oliva  <aoliva@redhat.com>
3041
3042         * configure.in [i*86-*-freebsd*] (TARGET, TARGETDIR): Set.
3043         * configure: Rebuilt.
3044
3045 2000-05-11  Scott Bambrough  <scottb@netwinder.org>
3046
3047         * libffi/src/arm/sysv.S (ffi_call_SYSV): Doubles are not saved to
3048         memory correctly.  Use conditional instructions, not branches where
3049         possible.
3050
3051 2000-05-04  Tom Tromey  <tromey@cygnus.com>
3052
3053         * configure: Rebuilt.
3054         * configure.in: Match `arm*-*-linux-*'.
3055         From Chris Dornan <cdornan@arm.com>.
3056
3057 2000-04-28  Jakub Jelinek  <jakub@redhat.com>
3058
3059         * Makefile.am (SUBDIRS): Define.
3060         (AM_MAKEFLAGS): Likewise.
3061         (Multilib support.): Add section.
3062         * Makefile.in: Rebuilt.
3063         * ltconfig (extra_compiler_flags, extra_compiler_flags_value):
3064         New variables. Set for gcc using -print-multi-lib. Export them
3065         to libtool.
3066         (sparc64-*-linux-gnu*): Use libsuff 64 for search paths.
3067         * ltmain.sh (B|b|V): Don't throw away gcc's -B, -b and -V options
3068         for -shared links.
3069         (extra_compiler_flags_value, extra_compiler_flags): Check these
3070         for extra compiler options which need to be passed down in
3071         compiler_flags.
3072
3073 2000-04-16  Anthony Green  <green@redhat.com>
3074
3075         * configure: Rebuilt.
3076         * configure.in: Change i*86-pc-linux* to i*86-*-linux*.
3077
3078 2000-04-14  Jakub Jelinek  <jakub@redhat.com>
3079
3080         * include/ffi.h.in (SPARC64): Define for 64bit SPARC builds.
3081         Set SPARC FFI_DEFAULT_ABI based on SPARC64 define.
3082         * src/sparc/ffi.c (ffi_prep_args_v8): Renamed from ffi_prep_args.
3083         Replace all void * sizeofs with sizeof(int).
3084         Only compare type with FFI_TYPE_LONGDOUBLE if LONGDOUBLE is
3085         different than DOUBLE.
3086         Remove FFI_TYPE_SINT32 and FFI_TYPE_UINT32 cases (handled elsewhere).
3087         (ffi_prep_args_v9): New function.
3088         (ffi_prep_cif_machdep): Handle V9 ABI and long long on V8.
3089         (ffi_V9_return_struct): New function.
3090         (ffi_call): Handle FFI_V9 ABI from 64bit code and FFI_V8 ABI from
3091         32bit code (not yet cross-arch calls).
3092         * src/sparc/v8.S: Add struct return delay nop.
3093         Handle long long.
3094         * src/sparc/v9.S: New file.
3095         * src/prep_cif.c (ffi_prep_cif): Return structure pointer
3096         is used on sparc64 only for structures larger than 32 bytes.
3097         Pass by reference for structures is done for structure arguments
3098         larger than 16 bytes.
3099         * src/ffitest.c (main): Use 64bit rint on sparc64.
3100         Run long long tests on sparc.
3101         * src/types.c (FFI_TYPE_POINTER): Pointer is 64bit on alpha and
3102         sparc64.
3103         (FFI_TYPE_LONGDOUBLE): long double is 128 bit aligned to 128 bits
3104         on sparc64.
3105         * configure.in (sparc-*-linux*): New supported target.
3106         (sparc64-*-linux*): Likewise.
3107         * configure: Rebuilt.
3108         * Makefile.am: Add v9.S to SPARC files.
3109         * Makefile.in: Likewise.
3110         (LINK): Surround $(CCLD) into double quotes, so that multilib
3111         compiles work correctly.
3112
3113 2000-04-04  Alexandre Petit-Bianco  <apbianco@cygnus.com>
3114
3115         * configure: Rebuilt.
3116         * configure.in: (i*86-*-solaris*): New libffi target. Patch
3117         proposed by Bryce McKinlay.
3118
3119 2000-03-20  Tom Tromey  <tromey@cygnus.com>
3120
3121         * Makefile.in: Hand edit for java_raw_api.lo.
3122
3123 2000-03-08  Bryce McKinlay  <bryce@albatross.co.nz>
3124
3125         * config.guess, config.sub: Update from the gcc tree.
3126         Fix for PR libgcj/168.
3127
3128 2000-03-03  Tom Tromey  <tromey@cygnus.com>
3129
3130         * Makefile.in: Fixed ia64 by hand.
3131
3132         * configure: Rebuilt.
3133         * configure.in (--enable-multilib): New option.
3134         (libffi_basedir): New subst.
3135         (AC_OUTPUT): Added multilib code.
3136
3137 2000-03-02  Tom Tromey  <tromey@cygnus.com>
3138
3139         * Makefile.in: Rebuilt.
3140         * Makefile.am (TARGET_SRC_IA64): Use `ia64', not `alpha', as
3141         directory name.
3142
3143 2000-02-25  Hans Boehm <boehm@acm.org>
3144
3145         * src/ia64/ffi.c, src/ia64/ia64_flags.h, src/ia64/unix.S: New
3146         files.
3147         * src/raw_api.c (ffi_translate_args): Fixed typo in argument
3148         list.
3149         (ffi_prep_raw_closure): Use ffi_translate_args, not
3150         ffi_closure_translate.
3151         * src/java_raw_api.c: New file.
3152         * src/ffitest.c (closure_test_fn): New function.
3153         (main): Define `rint' as long long on IA64.  Added new test when
3154         FFI_CLOSURES is defined.
3155         * include/ffi.h.in (ALIGN): Use size_t, not unsigned.
3156         (ffi_abi): Recognize IA64.
3157         (ffi_raw): Added `flt' field.
3158         Added "Java raw API" code.
3159         * configure.in: Recognize ia64.
3160         * Makefile.am (TARGET_SRC_IA64): New macro.
3161         (libffi_la_common_SOURCES): Added java_raw_api.c.
3162         (libffi_la_SOURCES): Define in IA64 case.
3163
3164 2000-01-04  Tom Tromey  <tromey@cygnus.com>
3165
3166         * Makefile.in: Rebuilt with newer automake.
3167
3168 1999-12-31  Tom Tromey  <tromey@cygnus.com>
3169
3170         * Makefile.am (INCLUDES): Added -I$(top_srcdir)/src.
3171
3172 1999-09-01  Tom Tromey  <tromey@cygnus.com>
3173
3174         * include/ffi.h.in: Removed PACKAGE and VERSION defines and
3175         undefs.
3176         * fficonfig.h.in: Rebuilt.
3177         * configure: Rebuilt.
3178         * configure.in: Pass 3rd argument to AM_INIT_AUTOMAKE.
3179         Use AM_PROG_LIBTOOL (automake 1.4 compatibility).
3180         * acconfig.h: Don't #undef PACKAGE or VERSION.
3181
3182 1999-08-09  Anthony Green  <green@cygnus.com>
3183
3184         * include/ffi.h.in: Try to work around messy header problem
3185         with PACKAGE and VERSION.
3186
3187         * configure: Rebuilt.
3188         * configure.in: Change version to 2.00-beta.
3189
3190         * fficonfig.h.in: Rebuilt.
3191         * acconfig.h (FFI_NO_STRUCTS, FFI_NO_RAW_API): Define.
3192
3193         * src/x86/ffi.c (ffi_raw_call): Rename.
3194
3195 1999-08-02  Kresten Krab Thorup  <krab@dominiq.is.s.u-tokyo.ac.jp>
3196
3197         * src/x86/ffi.c (ffi_closure_SYSV): New function.
3198         (ffi_prep_incoming_args_SYSV): Ditto.
3199         (ffi_prep_closure): Ditto.
3200         (ffi_closure_raw_SYSV): Ditto.
3201         (ffi_prep_raw_closure): More ditto.
3202         (ffi_call_raw): Final ditto.
3203
3204         * include/ffi.h.in: Add definitions for closure and raw API.
3205
3206         * src/x86/ffi.c (ffi_prep_cif_machdep): Added case for
3207         FFI_TYPE_UINT64.
3208
3209         * Makefile.am (libffi_la_common_SOURCES): Added raw_api.c
3210
3211         * src/raw_api.c: New file.
3212
3213         * include/ffi.h.in (ffi_raw): New type.
3214         (UINT_ARG, SINT_ARG): New defines.
3215         (ffi_closure, ffi_raw_closure): New types.
3216         (ffi_prep_closure, ffi_prep_raw_closure): New declarations.
3217
3218         * configure.in: Add check for endianness and sizeof void*.
3219
3220         * src/x86/sysv.S (ffi_call_SYSV): Call fixup routine via argument,
3221         instead of directly.
3222
3223         * configure: Rebuilt.
3224
3225 Thu Jul  8 14:28:42 1999  Anthony Green  <green@cygnus.com>
3226
3227         * configure.in: Add x86 and powerpc BeOS configurations.
3228         From Makoto Kato <m_kato@ga2.so-net.ne.jp>.
3229
3230 1999-05-09  Anthony Green  <green@cygnus.com>
3231
3232         * configure.in: Add warning about this being beta code.
3233         Remove src/Makefile.am from the picture.
3234         * configure: Rebuilt.
3235
3236         * Makefile.am: Move logic from src/Makefile.am.  Add changes
3237         to support libffi as a target library.
3238         * Makefile.in: Rebuilt.
3239
3240         * aclocal.m4, config.guess, config.sub, ltconfig, ltmain.sh:
3241         Upgraded to new autoconf, automake, libtool.
3242
3243         * README: Tweaks.
3244
3245         * LICENSE: Update copyright date.
3246
3247         * src/Makefile.am, src/Makefile.in: Removed.
3248
3249 1998-11-29  Anthony Green  <green@cygnus.com>
3250
3251         * include/ChangeLog: Removed.
3252         * src/ChangeLog: Removed.
3253         * src/mips/ChangeLog: Removed.
3254         * src/sparc/ChangeLog: Remboved.
3255         * src/x86/ChangeLog: Removed.
3256
3257         * ChangeLog.v1: Created.