File: ffi.c

package info (click to toggle)
libffi 3.2.1-9
  • links: PTS
  • area: main
  • in suites: buster
  • size: 5,304 kB
  • sloc: ansic: 25,713; asm: 12,881; sh: 11,687; exp: 513; makefile: 427; python: 161; cpp: 128
file content (1179 lines) | stat: -rw-r--r-- 30,676 bytes parent folder | download | duplicates (2)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
/* Copyright (c) 2009, 2010, 2011, 2012 ARM Ltd.

Permission is hereby granted, free of charge, to any person obtaining
a copy of this software and associated documentation files (the
``Software''), to deal in the Software without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of the Software, and to
permit persons to whom the Software is furnished to do so, subject to
the following conditions:

The above copyright notice and this permission notice shall be
included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED ``AS IS'', WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.  */

#include <stdio.h>

#include <ffi.h>
#include <ffi_common.h>

#include <stdlib.h>

/* Stack alignment requirement in bytes */
#if defined (__APPLE__)
#define AARCH64_STACK_ALIGN 1
#else
#define AARCH64_STACK_ALIGN 16
#endif

#define N_X_ARG_REG 8
#define N_V_ARG_REG 8

#define AARCH64_FFI_WITH_V (1 << AARCH64_FFI_WITH_V_BIT)

union _d
{
  UINT64 d;
  UINT32 s[2];
};

struct call_context
{
  UINT64 x [AARCH64_N_XREG];
  struct
  {
    union _d d[2];
  } v [AARCH64_N_VREG];
};

#if defined (__clang__) && defined (__APPLE__)
extern void
sys_icache_invalidate (void *start, size_t len);
#endif

static inline void
ffi_clear_cache (void *start, void *end)
{
#if defined (__clang__) && defined (__APPLE__)
	sys_icache_invalidate (start, (char *)end - (char *)start);
#elif defined (__GNUC__)
	__builtin___clear_cache (start, end);
#else
#error "Missing builtin to flush instruction cache"
#endif
}

static void *
get_x_addr (struct call_context *context, unsigned n)
{
  return &context->x[n];
}

static void *
get_s_addr (struct call_context *context, unsigned n)
{
#if defined __AARCH64EB__
  return &context->v[n].d[1].s[1];
#else
  return &context->v[n].d[0].s[0];
#endif
}

static void *
get_d_addr (struct call_context *context, unsigned n)
{
#if defined __AARCH64EB__
  return &context->v[n].d[1];
#else
  return &context->v[n].d[0];
#endif
}

static void *
get_v_addr (struct call_context *context, unsigned n)
{
  return &context->v[n];
}

/* Return the memory location at which a basic type would reside
   were it to have been stored in register n.  */

static void *
get_basic_type_addr (unsigned short type, struct call_context *context,
		     unsigned n)
{
  switch (type)
    {
    case FFI_TYPE_FLOAT:
      return get_s_addr (context, n);
    case FFI_TYPE_DOUBLE:
      return get_d_addr (context, n);
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
    case FFI_TYPE_LONGDOUBLE:
      return get_v_addr (context, n);
#endif
    case FFI_TYPE_UINT8:
    case FFI_TYPE_SINT8:
    case FFI_TYPE_UINT16:
    case FFI_TYPE_SINT16:
    case FFI_TYPE_UINT32:
    case FFI_TYPE_SINT32:
    case FFI_TYPE_INT:
    case FFI_TYPE_POINTER:
    case FFI_TYPE_UINT64:
    case FFI_TYPE_SINT64:
      return get_x_addr (context, n);
    case FFI_TYPE_VOID:
      return NULL;
    default:
      FFI_ASSERT (0);
      return NULL;
    }
}

/* Return the alignment width for each of the basic types.  */

static size_t
get_basic_type_alignment (unsigned short type)
{
  switch (type)
    {
    case FFI_TYPE_FLOAT:
#if defined (__APPLE__)
      return sizeof (UINT32);
#endif
    case FFI_TYPE_DOUBLE:
      return sizeof (UINT64);
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
    case FFI_TYPE_LONGDOUBLE:
      return sizeof (long double);
#endif
    case FFI_TYPE_UINT8:
    case FFI_TYPE_SINT8:
#if defined (__APPLE__)
	  return sizeof (UINT8);
#endif
    case FFI_TYPE_UINT16:
    case FFI_TYPE_SINT16:
#if defined (__APPLE__)
	  return sizeof (UINT16);
#endif
    case FFI_TYPE_UINT32:
    case FFI_TYPE_INT:
    case FFI_TYPE_SINT32:
#if defined (__APPLE__)
	  return sizeof (UINT32);
#endif
    case FFI_TYPE_POINTER:
    case FFI_TYPE_UINT64:
    case FFI_TYPE_SINT64:
      return sizeof (UINT64);

    default:
      FFI_ASSERT (0);
      return 0;
    }
}

/* Return the size in bytes for each of the basic types.  */

static size_t
get_basic_type_size (unsigned short type)
{
  switch (type)
    {
    case FFI_TYPE_FLOAT:
      return sizeof (UINT32);
    case FFI_TYPE_DOUBLE:
      return sizeof (UINT64);
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
    case FFI_TYPE_LONGDOUBLE:
      return sizeof (long double);
#endif
    case FFI_TYPE_UINT8:
      return sizeof (UINT8);
    case FFI_TYPE_SINT8:
      return sizeof (SINT8);
    case FFI_TYPE_UINT16:
      return sizeof (UINT16);
    case FFI_TYPE_SINT16:
      return sizeof (SINT16);
    case FFI_TYPE_UINT32:
      return sizeof (UINT32);
    case FFI_TYPE_INT:
    case FFI_TYPE_SINT32:
      return sizeof (SINT32);
    case FFI_TYPE_POINTER:
    case FFI_TYPE_UINT64:
      return sizeof (UINT64);
    case FFI_TYPE_SINT64:
      return sizeof (SINT64);

    default:
      FFI_ASSERT (0);
      return 0;
    }
}

extern void
ffi_call_SYSV (unsigned (*)(struct call_context *context, unsigned char *,
			    extended_cif *),
               struct call_context *context,
               extended_cif *,
               size_t,
               void (*fn)(void));

extern void
ffi_closure_SYSV (ffi_closure *);

/* Test for an FFI floating point representation.  */

static unsigned
is_floating_type (unsigned short type)
{
  return (type == FFI_TYPE_FLOAT || type == FFI_TYPE_DOUBLE
	  || type == FFI_TYPE_LONGDOUBLE);
}

/* Test for a homogeneous structure.  */

static unsigned short
get_homogeneous_type (ffi_type *ty)
{
  if (ty->type == FFI_TYPE_STRUCT && ty->elements)
    {
      unsigned i;
      unsigned short candidate_type
	= get_homogeneous_type (ty->elements[0]);
      for (i =1; ty->elements[i]; i++)
	{
	  unsigned short iteration_type = 0;
	  /* If we have a nested struct, we must find its homogeneous type.
	     If that fits with our candidate type, we are still
	     homogeneous.  */
	  if (ty->elements[i]->type == FFI_TYPE_STRUCT
	      && ty->elements[i]->elements)
	    {
	      iteration_type = get_homogeneous_type (ty->elements[i]);
	    }
	  else
	    {
	      iteration_type = ty->elements[i]->type;
	    }

	  /* If we are not homogeneous, return FFI_TYPE_STRUCT.  */
	  if (candidate_type != iteration_type)
	    return FFI_TYPE_STRUCT;
	}
      return candidate_type;
    }

  /* Base case, we have no more levels of nesting, so we
     are a basic type, and so, trivially homogeneous in that type.  */
  return ty->type;
}

/* Determine the number of elements within a STRUCT.

   Note, we must handle nested structs.

   If ty is not a STRUCT this function will return 0.  */

static unsigned
element_count (ffi_type *ty)
{
  if (ty->type == FFI_TYPE_STRUCT && ty->elements)
    {
      unsigned n;
      unsigned elems = 0;
      for (n = 0; ty->elements[n]; n++)
	{
	  if (ty->elements[n]->type == FFI_TYPE_STRUCT
	      && ty->elements[n]->elements)
	    elems += element_count (ty->elements[n]);
	  else
	    elems++;
	}
      return elems;
    }
  return 0;
}

/* Test for a homogeneous floating point aggregate.

   A homogeneous floating point aggregate is a homogeneous aggregate of
   a half- single- or double- precision floating point type with one
   to four elements.  Note that this includes nested structs of the
   basic type.  */

static int
is_hfa (ffi_type *ty)
{
  if (ty->type == FFI_TYPE_STRUCT
      && ty->elements[0]
      && is_floating_type (get_homogeneous_type (ty)))
    {
      unsigned n = element_count (ty);
      return n >= 1 && n <= 4;
    }
  return 0;
}

/* Test if an ffi_type is a candidate for passing in a register.

   This test does not check that sufficient registers of the
   appropriate class are actually available, merely that IFF
   sufficient registers are available then the argument will be passed
   in register(s).

   Note that an ffi_type that is deemed to be a register candidate
   will always be returned in registers.

   Returns 1 if a register candidate else 0.  */

static int
is_register_candidate (ffi_type *ty)
{
  switch (ty->type)
    {
    case FFI_TYPE_VOID:
    case FFI_TYPE_FLOAT:
    case FFI_TYPE_DOUBLE:
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
    case FFI_TYPE_LONGDOUBLE:
#endif
    case FFI_TYPE_UINT8:
    case FFI_TYPE_UINT16:
    case FFI_TYPE_UINT32:
    case FFI_TYPE_UINT64:
    case FFI_TYPE_POINTER:
    case FFI_TYPE_SINT8:
    case FFI_TYPE_SINT16:
    case FFI_TYPE_SINT32:
    case FFI_TYPE_INT:
    case FFI_TYPE_SINT64:
      return 1;

    case FFI_TYPE_STRUCT:
      if (is_hfa (ty))
        {
          return 1;
        }
      else if (ty->size > 16)
        {
          /* Too large. Will be replaced with a pointer to memory. The
             pointer MAY be passed in a register, but the value will
             not. This test specifically fails since the argument will
             never be passed by value in registers. */
          return 0;
        }
      else
        {
          /* Might be passed in registers depending on the number of
             registers required. */
          return (ty->size + 7) / 8 < N_X_ARG_REG;
        }
      break;

    default:
      FFI_ASSERT (0);
      break;
    }

  return 0;
}

/* Test if an ffi_type argument or result is a candidate for a vector
   register.  */

static int
is_v_register_candidate (ffi_type *ty)
{
  return is_floating_type (ty->type)
	   || (ty->type == FFI_TYPE_STRUCT && is_hfa (ty));
}

/* Representation of the procedure call argument marshalling
   state.

   The terse state variable names match the names used in the AARCH64
   PCS. */

struct arg_state
{
  unsigned ngrn;                /* Next general-purpose register number. */
  unsigned nsrn;                /* Next vector register number. */
  size_t nsaa;                  /* Next stack offset. */

#if defined (__APPLE__)
  unsigned allocating_variadic;
#endif
};

/* Initialize a procedure call argument marshalling state.  */
static void
arg_init (struct arg_state *state, size_t call_frame_size)
{
  state->ngrn = 0;
  state->nsrn = 0;
  state->nsaa = 0;

#if defined (__APPLE__)
  state->allocating_variadic = 0;
#endif
}

/* Return the number of available consecutive core argument
   registers.  */

static unsigned
available_x (struct arg_state *state)
{
  return N_X_ARG_REG - state->ngrn;
}

/* Return the number of available consecutive vector argument
   registers.  */

static unsigned
available_v (struct arg_state *state)
{
  return N_V_ARG_REG - state->nsrn;
}

static void *
allocate_to_x (struct call_context *context, struct arg_state *state)
{
  FFI_ASSERT (state->ngrn < N_X_ARG_REG);
  return get_x_addr (context, (state->ngrn)++);
}

static void *
allocate_to_s (struct call_context *context, struct arg_state *state)
{
  FFI_ASSERT (state->nsrn < N_V_ARG_REG);
  return get_s_addr (context, (state->nsrn)++);
}

static void *
allocate_to_d (struct call_context *context, struct arg_state *state)
{
  FFI_ASSERT (state->nsrn < N_V_ARG_REG);
  return get_d_addr (context, (state->nsrn)++);
}

static void *
allocate_to_v (struct call_context *context, struct arg_state *state)
{
  FFI_ASSERT (state->nsrn < N_V_ARG_REG);
  return get_v_addr (context, (state->nsrn)++);
}

/* Allocate an aligned slot on the stack and return a pointer to it.  */
static void *
allocate_to_stack (struct arg_state *state, void *stack, size_t alignment,
		   size_t size)
{
  void *allocation;

  /* Round up the NSAA to the larger of 8 or the natural
     alignment of the argument's type.  */
  state->nsaa = ALIGN (state->nsaa, alignment);
  state->nsaa = ALIGN (state->nsaa, alignment);
#if defined (__APPLE__)
  if (state->allocating_variadic)
    state->nsaa = ALIGN (state->nsaa, 8);
#else
  state->nsaa = ALIGN (state->nsaa, 8);
#endif

  allocation = stack + state->nsaa;

  state->nsaa += size;
  return allocation;
}

static void
copy_basic_type (void *dest, void *source, unsigned short type)
{
  /* This is necessary to ensure that basic types are copied
     sign extended to 64-bits as libffi expects.  */
  switch (type)
    {
    case FFI_TYPE_FLOAT:
      *(float *) dest = *(float *) source;
      break;
    case FFI_TYPE_DOUBLE:
      *(double *) dest = *(double *) source;
      break;
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
    case FFI_TYPE_LONGDOUBLE:
      *(long double *) dest = *(long double *) source;
      break;
#endif
    case FFI_TYPE_UINT8:
      *(ffi_arg *) dest = *(UINT8 *) source;
      break;
    case FFI_TYPE_SINT8:
      *(ffi_sarg *) dest = *(SINT8 *) source;
      break;
    case FFI_TYPE_UINT16:
      *(ffi_arg *) dest = *(UINT16 *) source;
      break;
    case FFI_TYPE_SINT16:
      *(ffi_sarg *) dest = *(SINT16 *) source;
      break;
    case FFI_TYPE_UINT32:
      *(ffi_arg *) dest = *(UINT32 *) source;
      break;
    case FFI_TYPE_INT:
    case FFI_TYPE_SINT32:
      *(ffi_sarg *) dest = *(SINT32 *) source;
      break;
    case FFI_TYPE_POINTER:
    case FFI_TYPE_UINT64:
      *(ffi_arg *) dest = *(UINT64 *) source;
      break;
    case FFI_TYPE_SINT64:
      *(ffi_sarg *) dest = *(SINT64 *) source;
      break;
    case FFI_TYPE_VOID:
      break;

    default:
      FFI_ASSERT (0);
    }
}

static void
copy_hfa_to_reg_or_stack (void *memory,
			  ffi_type *ty,
			  struct call_context *context,
			  unsigned char *stack,
			  struct arg_state *state)
{
  unsigned elems = element_count (ty);
  if (available_v (state) < elems)
    {
      /* There are insufficient V registers. Further V register allocations
	 are prevented, the NSAA is adjusted (by allocate_to_stack ())
	 and the argument is copied to memory at the adjusted NSAA.  */
      state->nsrn = N_V_ARG_REG;
      memcpy (allocate_to_stack (state, stack, ty->alignment, ty->size),
	      memory,
	      ty->size);
    }
  else
    {
      int i;
      unsigned short type = get_homogeneous_type (ty);
      for (i = 0; i < elems; i++)
	{
	  void *reg = allocate_to_v (context, state);
	  copy_basic_type (reg, memory, type);
	  memory += get_basic_type_size (type);
	}
    }
}

/* Either allocate an appropriate register for the argument type, or if
   none are available, allocate a stack slot and return a pointer
   to the allocated space.  */

static void *
allocate_to_register_or_stack (struct call_context *context,
			       unsigned char *stack,
			       struct arg_state *state,
			       unsigned short type)
{
  size_t alignment = get_basic_type_alignment (type);
  size_t size = alignment;
  switch (type)
    {
    case FFI_TYPE_FLOAT:
      /* This is the only case for which the allocated stack size
	 should not match the alignment of the type.  */
      size = sizeof (UINT32);
      /* Fall through.  */
    case FFI_TYPE_DOUBLE:
      if (state->nsrn < N_V_ARG_REG)
	return allocate_to_d (context, state);
      state->nsrn = N_V_ARG_REG;
      break;
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
    case FFI_TYPE_LONGDOUBLE:
      if (state->nsrn < N_V_ARG_REG)
	return allocate_to_v (context, state);
      state->nsrn = N_V_ARG_REG;
      break;
#endif
    case FFI_TYPE_UINT8:
    case FFI_TYPE_SINT8:
    case FFI_TYPE_UINT16:
    case FFI_TYPE_SINT16:
    case FFI_TYPE_UINT32:
    case FFI_TYPE_SINT32:
    case FFI_TYPE_INT:
    case FFI_TYPE_POINTER:
    case FFI_TYPE_UINT64:
    case FFI_TYPE_SINT64:
      if (state->ngrn < N_X_ARG_REG)
	return allocate_to_x (context, state);
      state->ngrn = N_X_ARG_REG;
      break;
    default:
      FFI_ASSERT (0);
    }

    return allocate_to_stack (state, stack, alignment, size);
}

/* Copy a value to an appropriate register, or if none are
   available, to the stack.  */

static void
copy_to_register_or_stack (struct call_context *context,
			   unsigned char *stack,
			   struct arg_state *state,
			   void *value,
			   unsigned short type)
{
  copy_basic_type (
	  allocate_to_register_or_stack (context, stack, state, type),
	  value,
	  type);
}

/* Marshall the arguments from FFI representation to procedure call
   context and stack.  */

static unsigned
aarch64_prep_args (struct call_context *context, unsigned char *stack,
		   extended_cif *ecif)
{
  int i;
  struct arg_state state;

  arg_init (&state, ALIGN(ecif->cif->bytes, 16));

  for (i = 0; i < ecif->cif->nargs; i++)
    {
      ffi_type *ty = ecif->cif->arg_types[i];
      switch (ty->type)
	{
	case FFI_TYPE_VOID:
	  FFI_ASSERT (0);
	  break;

	/* If the argument is a basic type the argument is allocated to an
	   appropriate register, or if none are available, to the stack.  */
	case FFI_TYPE_FLOAT:
	case FFI_TYPE_DOUBLE:
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
	case FFI_TYPE_LONGDOUBLE:
#endif
	case FFI_TYPE_UINT8:
	case FFI_TYPE_SINT8:
	case FFI_TYPE_UINT16:
	case FFI_TYPE_SINT16:
	case FFI_TYPE_UINT32:
	case FFI_TYPE_INT:
	case FFI_TYPE_SINT32:
	case FFI_TYPE_POINTER:
	case FFI_TYPE_UINT64:
	case FFI_TYPE_SINT64:
	  copy_to_register_or_stack (context, stack, &state,
				     ecif->avalue[i], ty->type);
	  break;

	case FFI_TYPE_STRUCT:
	  if (is_hfa (ty))
	    {
	      copy_hfa_to_reg_or_stack (ecif->avalue[i], ty, context,
					stack, &state);
	    }
	  else if (ty->size > 16)
	    {
	      /* If the argument is a composite type that is larger than 16
		 bytes, then the argument has been copied to memory, and
		 the argument is replaced by a pointer to the copy.  */

	      copy_to_register_or_stack (context, stack, &state,
					 &(ecif->avalue[i]), FFI_TYPE_POINTER);
	    }
	  else if (available_x (&state) >= (ty->size + 7) / 8)
	    {
	      /* If the argument is a composite type and the size in
		 double-words is not more than the number of available
		 X registers, then the argument is copied into consecutive
		 X registers.  */
	      int j;
	      for (j = 0; j < (ty->size + 7) / 8; j++)
		{
		  memcpy (allocate_to_x (context, &state),
			  &(((UINT64 *) ecif->avalue[i])[j]),
			  sizeof (UINT64));
		}
	    }
	  else
	    {
	      /* Otherwise, there are insufficient X registers. Further X
		 register allocations are prevented, the NSAA is adjusted
		 (by allocate_to_stack ()) and the argument is copied to
		 memory at the adjusted NSAA.  */
	      state.ngrn = N_X_ARG_REG;

	      memcpy (allocate_to_stack (&state, stack, ty->alignment,
					 ty->size), ecif->avalue[i], ty->size);
	    }
	  break;

	default:
	  FFI_ASSERT (0);
	  break;
	}

#if defined (__APPLE__)
      if (i + 1 == ecif->cif->aarch64_nfixedargs)
	{
	  state.ngrn = N_X_ARG_REG;
	  state.nsrn = N_V_ARG_REG;

	  state.allocating_variadic = 1;
	}
#endif
    }

  return ecif->cif->aarch64_flags;
}

ffi_status
ffi_prep_cif_machdep (ffi_cif *cif)
{
  /* Round the stack up to a multiple of the stack alignment requirement. */
  cif->bytes =
    (cif->bytes + (AARCH64_STACK_ALIGN - 1)) & ~ (AARCH64_STACK_ALIGN - 1);

  /* Initialize our flags. We are interested if this CIF will touch a
     vector register, if so we will enable context save and load to
     those registers, otherwise not. This is intended to be friendly
     to lazy float context switching in the kernel.  */
  cif->aarch64_flags = 0;

  if (is_v_register_candidate (cif->rtype))
    {
      cif->aarch64_flags |= AARCH64_FFI_WITH_V;
    }
  else
    {
      int i;
      for (i = 0; i < cif->nargs; i++)
        if (is_v_register_candidate (cif->arg_types[i]))
          {
            cif->aarch64_flags |= AARCH64_FFI_WITH_V;
            break;
          }
    }

#if defined (__APPLE__)
  cif->aarch64_nfixedargs = 0;
#endif

  return FFI_OK;
}

#if defined (__APPLE__)

/* Perform Apple-specific cif processing for variadic calls */
ffi_status ffi_prep_cif_machdep_var(ffi_cif *cif,
				    unsigned int nfixedargs,
				    unsigned int ntotalargs)
{
  ffi_status status;

  status = ffi_prep_cif_machdep (cif);

  cif->aarch64_nfixedargs = nfixedargs;

  return status;
}

#endif

/* Call a function with the provided arguments and capture the return
   value.  */
void
ffi_call (ffi_cif *cif, void (*fn)(void), void *rvalue, void **avalue)
{
  extended_cif ecif;

  ecif.cif = cif;
  ecif.avalue = avalue;
  ecif.rvalue = rvalue;

  switch (cif->abi)
    {
    case FFI_SYSV:
      {
        struct call_context context;
	size_t stack_bytes;

	/* Figure out the total amount of stack space we need, the
	   above call frame space needs to be 16 bytes aligned to
	   ensure correct alignment of the first object inserted in
	   that space hence the ALIGN applied to cif->bytes.*/
	stack_bytes = ALIGN(cif->bytes, 16);

	memset (&context, 0, sizeof (context));
        if (is_register_candidate (cif->rtype))
          {
            ffi_call_SYSV (aarch64_prep_args, &context, &ecif, stack_bytes, fn);
            switch (cif->rtype->type)
              {
              case FFI_TYPE_VOID:
              case FFI_TYPE_FLOAT:
              case FFI_TYPE_DOUBLE:
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
              case FFI_TYPE_LONGDOUBLE:
#endif
              case FFI_TYPE_UINT8:
              case FFI_TYPE_SINT8:
              case FFI_TYPE_UINT16:
              case FFI_TYPE_SINT16:
              case FFI_TYPE_UINT32:
              case FFI_TYPE_SINT32:
              case FFI_TYPE_POINTER:
              case FFI_TYPE_UINT64:
              case FFI_TYPE_INT:
              case FFI_TYPE_SINT64:
		{
		  void *addr = get_basic_type_addr (cif->rtype->type,
						    &context, 0);
		  copy_basic_type (rvalue, addr, cif->rtype->type);
		  break;
		}

              case FFI_TYPE_STRUCT:
                if (is_hfa (cif->rtype))
		  {
		    int j;
		    unsigned short type = get_homogeneous_type (cif->rtype);
		    unsigned elems = element_count (cif->rtype);
		    for (j = 0; j < elems; j++)
		      {
			void *reg = get_basic_type_addr (type, &context, j);
			copy_basic_type (rvalue, reg, type);
			rvalue += get_basic_type_size (type);
		      }
		  }
                else if ((cif->rtype->size + 7) / 8 < N_X_ARG_REG)
                  {
                    size_t size = ALIGN (cif->rtype->size, sizeof (UINT64));
                    memcpy (rvalue, get_x_addr (&context, 0), size);
                  }
                else
                  {
                    FFI_ASSERT (0);
                  }
                break;

              default:
                FFI_ASSERT (0);
                break;
              }
          }
        else
          {
            memcpy (get_x_addr (&context, 8), &rvalue, sizeof (UINT64));
            ffi_call_SYSV (aarch64_prep_args, &context, &ecif,
			   stack_bytes, fn);
          }
        break;
      }

    default:
      FFI_ASSERT (0);
      break;
    }
}

static unsigned char trampoline [] =
{ 0x70, 0x00, 0x00, 0x58,	/* ldr	x16, 1f	*/
  0x91, 0x00, 0x00, 0x10,	/* adr	x17, 2f	*/
  0x00, 0x02, 0x1f, 0xd6	/* br	x16	*/
};

/* Build a trampoline.  */

#define FFI_INIT_TRAMPOLINE(TRAMP,FUN,CTX,FLAGS)			\
  ({unsigned char *__tramp = (unsigned char*)(TRAMP);			\
    UINT64  __fun = (UINT64)(FUN);					\
    UINT64  __ctx = (UINT64)(CTX);					\
    UINT64  __flags = (UINT64)(FLAGS);					\
    memcpy (__tramp, trampoline, sizeof (trampoline));			\
    memcpy (__tramp + 12, &__fun, sizeof (__fun));			\
    memcpy (__tramp + 20, &__ctx, sizeof (__ctx));			\
    memcpy (__tramp + 28, &__flags, sizeof (__flags));			\
    ffi_clear_cache(__tramp, __tramp + FFI_TRAMPOLINE_SIZE);		\
  })

ffi_status
ffi_prep_closure_loc (ffi_closure* closure,
                      ffi_cif* cif,
                      void (*fun)(ffi_cif*,void*,void**,void*),
                      void *user_data,
                      void *codeloc)
{
  if (cif->abi != FFI_SYSV)
    return FFI_BAD_ABI;

  FFI_INIT_TRAMPOLINE (&closure->tramp[0], &ffi_closure_SYSV, codeloc,
		       cif->aarch64_flags);

  closure->cif  = cif;
  closure->user_data = user_data;
  closure->fun  = fun;

  return FFI_OK;
}

/* Primary handler to setup and invoke a function within a closure.

   A closure when invoked enters via the assembler wrapper
   ffi_closure_SYSV(). The wrapper allocates a call context on the
   stack, saves the interesting registers (from the perspective of
   the calling convention) into the context then passes control to
   ffi_closure_SYSV_inner() passing the saved context and a pointer to
   the stack at the point ffi_closure_SYSV() was invoked.

   On the return path the assembler wrapper will reload call context
   registers.

   ffi_closure_SYSV_inner() marshalls the call context into ffi value
   descriptors, invokes the wrapped function, then marshalls the return
   value back into the call context.  */

void FFI_HIDDEN
ffi_closure_SYSV_inner (ffi_closure *closure, struct call_context *context,
			void *stack)
{
  ffi_cif *cif = closure->cif;
  void **avalue = (void**) alloca (cif->nargs * sizeof (void*));
  void *rvalue = NULL;
  int i;
  struct arg_state state;

  arg_init (&state, ALIGN(cif->bytes, 16));

  for (i = 0; i < cif->nargs; i++)
    {
      ffi_type *ty = cif->arg_types[i];

      switch (ty->type)
	{
	case FFI_TYPE_VOID:
	  FFI_ASSERT (0);
	  break;

	case FFI_TYPE_UINT8:
	case FFI_TYPE_SINT8:
	case FFI_TYPE_UINT16:
	case FFI_TYPE_SINT16:
	case FFI_TYPE_UINT32:
	case FFI_TYPE_SINT32:
	case FFI_TYPE_INT:
	case FFI_TYPE_POINTER:
	case FFI_TYPE_UINT64:
	case FFI_TYPE_SINT64:
	case  FFI_TYPE_FLOAT:
	case  FFI_TYPE_DOUBLE:
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
	case  FFI_TYPE_LONGDOUBLE:
	  avalue[i] = allocate_to_register_or_stack (context, stack,
						     &state, ty->type);
	  break;
#endif

	case FFI_TYPE_STRUCT:
	  if (is_hfa (ty))
	    {
	      unsigned n = element_count (ty);
	      if (available_v (&state) < n)
		{
		  state.nsrn = N_V_ARG_REG;
		  avalue[i] = allocate_to_stack (&state, stack, ty->alignment,
						 ty->size);
		}
	      else
		{
		  switch (get_homogeneous_type (ty))
		    {
		    case FFI_TYPE_FLOAT:
		      {
			/* Eeek! We need a pointer to the structure,
			   however the homogeneous float elements are
			   being passed in individual S registers,
			   therefore the structure is not represented as
			   a contiguous sequence of bytes in our saved
			   register context. We need to fake up a copy
			   of the structure laid out in memory
			   correctly. The fake can be tossed once the
			   closure function has returned hence alloca()
			   is sufficient. */
			int j;
			UINT32 *p = avalue[i] = alloca (ty->size);
			for (j = 0; j < element_count (ty); j++)
			  memcpy (&p[j],
				  allocate_to_s (context, &state),
				  sizeof (*p));
			break;
		      }

		    case FFI_TYPE_DOUBLE:
		      {
			/* Eeek! We need a pointer to the structure,
			   however the homogeneous float elements are
			   being passed in individual S registers,
			   therefore the structure is not represented as
			   a contiguous sequence of bytes in our saved
			   register context. We need to fake up a copy
			   of the structure laid out in memory
			   correctly. The fake can be tossed once the
			   closure function has returned hence alloca()
			   is sufficient. */
			int j;
			UINT64 *p = avalue[i] = alloca (ty->size);
			for (j = 0; j < element_count (ty); j++)
			  memcpy (&p[j],
				  allocate_to_d (context, &state),
				  sizeof (*p));
			break;
		      }

#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
		    case FFI_TYPE_LONGDOUBLE:
			  memcpy (&avalue[i],
				  allocate_to_v (context, &state),
				  sizeof (*avalue));
		      break;
#endif

		    default:
		      FFI_ASSERT (0);
		      break;
		    }
		}
	    }
	  else if (ty->size > 16)
	    {
	      /* Replace Composite type of size greater than 16 with a
		 pointer.  */
	      memcpy (&avalue[i],
		      allocate_to_register_or_stack (context, stack,
						     &state, FFI_TYPE_POINTER),
		      sizeof (avalue[i]));
	    }
	  else if (available_x (&state) >= (ty->size + 7) / 8)
	    {
	      avalue[i] = get_x_addr (context, state.ngrn);
	      state.ngrn += (ty->size + 7) / 8;
	    }
	  else
	    {
	      state.ngrn = N_X_ARG_REG;

	      avalue[i] = allocate_to_stack (&state, stack, ty->alignment,
					     ty->size);
	    }
	  break;

	default:
	  FFI_ASSERT (0);
	  break;
	}
    }

  /* Figure out where the return value will be passed, either in
     registers or in a memory block allocated by the caller and passed
     in x8.  */

  if (is_register_candidate (cif->rtype))
    {
      /* Register candidates are *always* returned in registers. */

      /* Allocate a scratchpad for the return value, we will let the
         callee scrible the result into the scratch pad then move the
         contents into the appropriate return value location for the
         call convention.  */
      rvalue = alloca (cif->rtype->size);
      (closure->fun) (cif, rvalue, avalue, closure->user_data);

      /* Copy the return value into the call context so that it is returned
         as expected to our caller.  */
      switch (cif->rtype->type)
        {
        case FFI_TYPE_VOID:
          break;

        case FFI_TYPE_UINT8:
        case FFI_TYPE_UINT16:
        case FFI_TYPE_UINT32:
        case FFI_TYPE_POINTER:
        case FFI_TYPE_UINT64:
        case FFI_TYPE_SINT8:
        case FFI_TYPE_SINT16:
        case FFI_TYPE_INT:
        case FFI_TYPE_SINT32:
        case FFI_TYPE_SINT64:
        case FFI_TYPE_FLOAT:
        case FFI_TYPE_DOUBLE:
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
        case FFI_TYPE_LONGDOUBLE:
#endif
	  {
	    void *addr = get_basic_type_addr (cif->rtype->type, context, 0);
	    copy_basic_type (addr, rvalue, cif->rtype->type);
            break;
	  }
        case FFI_TYPE_STRUCT:
          if (is_hfa (cif->rtype))
	    {
	      int j;
	      unsigned short type = get_homogeneous_type (cif->rtype);
	      unsigned elems = element_count (cif->rtype);
	      for (j = 0; j < elems; j++)
		{
		  void *reg = get_basic_type_addr (type, context, j);
		  copy_basic_type (reg, rvalue, type);
		  rvalue += get_basic_type_size (type);
		}
	    }
          else if ((cif->rtype->size + 7) / 8 < N_X_ARG_REG)
            {
              size_t size = ALIGN (cif->rtype->size, sizeof (UINT64)) ;
              memcpy (get_x_addr (context, 0), rvalue, size);
            }
          else
            {
              FFI_ASSERT (0);
            }
          break;
        default:
          FFI_ASSERT (0);
          break;
        }
    }
  else
    {
      memcpy (&rvalue, get_x_addr (context, 8), sizeof (UINT64));
      (closure->fun) (cif, rvalue, avalue, closure->user_data);
    }
}