summaryrefslogtreecommitdiff
path: root/gdb/i960-tdep.c
blob: 35e47f1a6261f6ef439cf5500b3ed915f37c1a11 (plain)
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
/* Target-machine dependent code for the Intel 960
   Copyright 1991, 1992, 1993, 1994, 1995, 2001 Free Software
   Foundation, Inc.
   Contributed by Intel Corporation.
   examine_prologue and other parts contributed by Wind River Systems.

   This file is part of GDB.

   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 2 of the License, or
   (at your option) any later version.

   This program is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.

   You should have received a copy of the GNU General Public License
   along with this program; if not, write to the Free Software
   Foundation, Inc., 59 Temple Place - Suite 330,
   Boston, MA 02111-1307, USA.  */

#include "defs.h"
#include "symtab.h"
#include "value.h"
#include "frame.h"
#include "floatformat.h"
#include "target.h"
#include "gdbcore.h"
#include "inferior.h"
#include "regcache.h"

static CORE_ADDR next_insn (CORE_ADDR memaddr,
			    unsigned int *pword1, unsigned int *pword2);

/* Does the specified function use the "struct returning" convention
   or the "value returning" convention?  The "value returning" convention
   almost invariably returns the entire value in registers.  The
   "struct returning" convention often returns the entire value in
   memory, and passes a pointer (out of or into the function) saying
   where the value (is or should go).

   Since this sometimes depends on whether it was compiled with GCC,
   this is also an argument.  This is used in call_function to build a
   stack, and in value_being_returned to print return values.

   On i960, a structure is returned in registers g0-g3, if it will fit.
   If it's more than 16 bytes long, g13 pointed to it on entry.  */

int
i960_use_struct_convention (int gcc_p, struct type *type)
{
  return (TYPE_LENGTH (type) > 16);
}

/* gdb960 is always running on a non-960 host.  Check its characteristics.
   This routine must be called as part of gdb initialization.  */

static void
check_host (void)
{
  int i;

  static struct typestruct
    {
      int hostsize;		/* Size of type on host         */
      int i960size;		/* Size of type on i960         */
      char *typename;		/* Name of type, for error msg  */
    }
  types[] =
  {
    {
      sizeof (short), 2, "short"
    }
     ,
    {
      sizeof (int), 4, "int"
    }
     ,
    {
      sizeof (long), 4, "long"
    }
     ,
    {
      sizeof (float), 4, "float"
    }
     ,
    {
      sizeof (double), 8, "double"
    }
     ,
    {
      sizeof (char *), 4, "pointer"
    }
     ,
  };
#define TYPELEN	(sizeof(types) / sizeof(struct typestruct))

  /* Make sure that host type sizes are same as i960
   */
  for (i = 0; i < TYPELEN; i++)
    {
      if (types[i].hostsize != types[i].i960size)
	{
	  printf_unfiltered ("sizeof(%s) != %d:  PROCEED AT YOUR OWN RISK!\n",
			     types[i].typename, types[i].i960size);
	}

    }
}

/* Examine an i960 function prologue, recording the addresses at which
   registers are saved explicitly by the prologue code, and returning
   the address of the first instruction after the prologue (but not
   after the instruction at address LIMIT, as explained below).

   LIMIT places an upper bound on addresses of the instructions to be
   examined.  If the prologue code scan reaches LIMIT, the scan is
   aborted and LIMIT is returned.  This is used, when examining the
   prologue for the current frame, to keep examine_prologue () from
   claiming that a given register has been saved when in fact the
   instruction that saves it has not yet been executed.  LIMIT is used
   at other times to stop the scan when we hit code after the true
   function prologue (e.g. for the first source line) which might
   otherwise be mistaken for function prologue.

   The format of the function prologue matched by this routine is
   derived from examination of the source to gcc960 1.21, particularly
   the routine i960_function_prologue ().  A "regular expression" for
   the function prologue is given below:

   (lda LRn, g14
   mov g14, g[0-7]
   (mov 0, g14) | (lda 0, g14))?

   (mov[qtl]? g[0-15], r[4-15])*
   ((addo [1-31], sp, sp) | (lda n(sp), sp))?
   (st[qtl]? g[0-15], n(fp))*

   (cmpobne 0, g14, LFn
   mov sp, g14
   lda 0x30(sp), sp
   LFn: stq g0, (g14)
   stq g4, 0x10(g14)
   stq g8, 0x20(g14))?

   (st g14, n(fp))?
   (mov g13,r[4-15])?
 */

/* Macros for extracting fields from i960 instructions.  */

#define BITMASK(pos, width) (((0x1 << (width)) - 1) << (pos))
#define EXTRACT_FIELD(val, pos, width) ((val) >> (pos) & BITMASK (0, width))

#define REG_SRC1(insn)    EXTRACT_FIELD (insn, 0, 5)
#define REG_SRC2(insn)    EXTRACT_FIELD (insn, 14, 5)
#define REG_SRCDST(insn)  EXTRACT_FIELD (insn, 19, 5)
#define MEM_SRCDST(insn)  EXTRACT_FIELD (insn, 19, 5)
#define MEMA_OFFSET(insn) EXTRACT_FIELD (insn, 0, 12)

/* Fetch the instruction at ADDR, returning 0 if ADDR is beyond LIM or
   is not the address of a valid instruction, the address of the next
   instruction beyond ADDR otherwise.  *PWORD1 receives the first word
   of the instruction, and (for two-word instructions), *PWORD2 receives
   the second.  */

#define NEXT_PROLOGUE_INSN(addr, lim, pword1, pword2) \
  (((addr) < (lim)) ? next_insn (addr, pword1, pword2) : 0)

static CORE_ADDR
examine_prologue (register CORE_ADDR ip, register CORE_ADDR limit,
		  CORE_ADDR frame_addr, struct frame_saved_regs *fsr)
{
  register CORE_ADDR next_ip;
  register int src, dst;
  register unsigned int *pcode;
  unsigned int insn1, insn2;
  int size;
  int within_leaf_prologue;
  CORE_ADDR save_addr;
  static unsigned int varargs_prologue_code[] =
  {
    0x3507a00c,			/* cmpobne 0x0, g14, LFn */
    0x5cf01601,			/* mov sp, g14           */
    0x8c086030,			/* lda 0x30(sp), sp      */
    0xb2879000,			/* LFn: stq  g0, (g14)   */
    0xb2a7a010,			/* stq g4, 0x10(g14)     */
    0xb2c7a020			/* stq g8, 0x20(g14)     */
  };

  /* Accept a leaf procedure prologue code fragment if present.
     Note that ip might point to either the leaf or non-leaf
     entry point; we look for the non-leaf entry point first:  */

  within_leaf_prologue = 0;
  if ((next_ip = NEXT_PROLOGUE_INSN (ip, limit, &insn1, &insn2))
      && ((insn1 & 0xfffff000) == 0x8cf00000	/* lda LRx, g14 (MEMA) */
	  || (insn1 & 0xfffffc60) == 0x8cf03000))	/* lda LRx, g14 (MEMB) */
    {
      within_leaf_prologue = 1;
      next_ip = NEXT_PROLOGUE_INSN (next_ip, limit, &insn1, &insn2);
    }

  /* Now look for the prologue code at a leaf entry point:  */

  if (next_ip
      && (insn1 & 0xff87ffff) == 0x5c80161e	/* mov g14, gx */
      && REG_SRCDST (insn1) <= G0_REGNUM + 7)
    {
      within_leaf_prologue = 1;
      if ((next_ip = NEXT_PROLOGUE_INSN (next_ip, limit, &insn1, &insn2))
	  && (insn1 == 0x8cf00000	/* lda 0, g14 */
	      || insn1 == 0x5cf01e00))	/* mov 0, g14 */
	{
	  ip = next_ip;
	  next_ip = NEXT_PROLOGUE_INSN (ip, limit, &insn1, &insn2);
	  within_leaf_prologue = 0;
	}
    }

  /* If something that looks like the beginning of a leaf prologue
     has been seen, but the remainder of the prologue is missing, bail.
     We don't know what we've got.  */

  if (within_leaf_prologue)
    return (ip);

  /* Accept zero or more instances of "mov[qtl]? gx, ry", where y >= 4.
     This may cause us to mistake the moving of a register
     parameter to a local register for the saving of a callee-saved
     register, but that can't be helped, since with the
     "-fcall-saved" flag, any register can be made callee-saved.  */

  while (next_ip
	 && (insn1 & 0xfc802fb0) == 0x5c000610
	 && (dst = REG_SRCDST (insn1)) >= (R0_REGNUM + 4))
    {
      src = REG_SRC1 (insn1);
      size = EXTRACT_FIELD (insn1, 24, 2) + 1;
      save_addr = frame_addr + ((dst - R0_REGNUM) * 4);
      while (size--)
	{
	  fsr->regs[src++] = save_addr;
	  save_addr += 4;
	}
      ip = next_ip;
      next_ip = NEXT_PROLOGUE_INSN (ip, limit, &insn1, &insn2);
    }

  /* Accept an optional "addo n, sp, sp" or "lda n(sp), sp".  */

  if (next_ip &&
      ((insn1 & 0xffffffe0) == 0x59084800	/* addo n, sp, sp */
       || (insn1 & 0xfffff000) == 0x8c086000	/* lda n(sp), sp (MEMA) */
       || (insn1 & 0xfffffc60) == 0x8c087400))	/* lda n(sp), sp (MEMB) */
    {
      ip = next_ip;
      next_ip = NEXT_PROLOGUE_INSN (ip, limit, &insn1, &insn2);
    }

  /* Accept zero or more instances of "st[qtl]? gx, n(fp)".  
     This may cause us to mistake the copying of a register
     parameter to the frame for the saving of a callee-saved
     register, but that can't be helped, since with the
     "-fcall-saved" flag, any register can be made callee-saved.
     We can, however, refuse to accept a save of register g14,
     since that is matched explicitly below.  */

  while (next_ip &&
	 ((insn1 & 0xf787f000) == 0x9287e000	/* stl? gx, n(fp) (MEMA) */
	  || (insn1 & 0xf787fc60) == 0x9287f400		/* stl? gx, n(fp) (MEMB) */
	  || (insn1 & 0xef87f000) == 0xa287e000		/* st[tq] gx, n(fp) (MEMA) */
	  || (insn1 & 0xef87fc60) == 0xa287f400)	/* st[tq] gx, n(fp) (MEMB) */
	 && ((src = MEM_SRCDST (insn1)) != G14_REGNUM))
    {
      save_addr = frame_addr + ((insn1 & BITMASK (12, 1))
				? insn2 : MEMA_OFFSET (insn1));
      size = (insn1 & BITMASK (29, 1)) ? ((insn1 & BITMASK (28, 1)) ? 4 : 3)
	: ((insn1 & BITMASK (27, 1)) ? 2 : 1);
      while (size--)
	{
	  fsr->regs[src++] = save_addr;
	  save_addr += 4;
	}
      ip = next_ip;
      next_ip = NEXT_PROLOGUE_INSN (ip, limit, &insn1, &insn2);
    }

  /* Accept the varargs prologue code if present.  */

  size = sizeof (varargs_prologue_code) / sizeof (int);
  pcode = varargs_prologue_code;
  while (size-- && next_ip && *pcode++ == insn1)
    {
      ip = next_ip;
      next_ip = NEXT_PROLOGUE_INSN (ip, limit, &insn1, &insn2);
    }

  /* Accept an optional "st g14, n(fp)".  */

  if (next_ip &&
      ((insn1 & 0xfffff000) == 0x92f7e000	/* st g14, n(fp) (MEMA) */
       || (insn1 & 0xfffffc60) == 0x92f7f400))	/* st g14, n(fp) (MEMB) */
    {
      fsr->regs[G14_REGNUM] = frame_addr + ((insn1 & BITMASK (12, 1))
					    ? insn2 : MEMA_OFFSET (insn1));
      ip = next_ip;
      next_ip = NEXT_PROLOGUE_INSN (ip, limit, &insn1, &insn2);
    }

  /* Accept zero or one instance of "mov g13, ry", where y >= 4.
     This is saving the address where a struct should be returned.  */

  if (next_ip
      && (insn1 & 0xff802fbf) == 0x5c00061d
      && (dst = REG_SRCDST (insn1)) >= (R0_REGNUM + 4))
    {
      save_addr = frame_addr + ((dst - R0_REGNUM) * 4);
      fsr->regs[G0_REGNUM + 13] = save_addr;
      ip = next_ip;
#if 0				/* We'll need this once there is a subsequent instruction examined. */
      next_ip = NEXT_PROLOGUE_INSN (ip, limit, &insn1, &insn2);
#endif
    }

  return (ip);
}

/* Given an ip value corresponding to the start of a function,
   return the ip of the first instruction after the function 
   prologue.  */

CORE_ADDR
i960_skip_prologue (CORE_ADDR ip)
{
  struct frame_saved_regs saved_regs_dummy;
  struct symtab_and_line sal;
  CORE_ADDR limit;

  sal = find_pc_line (ip, 0);
  limit = (sal.end) ? sal.end : 0xffffffff;

  return (examine_prologue (ip, limit, (CORE_ADDR) 0, &saved_regs_dummy));
}

/* Put here the code to store, into a struct frame_saved_regs,
   the addresses of the saved registers of frame described by FRAME_INFO.
   This includes special registers such as pc and fp saved in special
   ways in the stack frame.  sp is even more special:
   the address we return for it IS the sp for the next frame.

   We cache the result of doing this in the frame_obstack, since it is
   fairly expensive.  */

void
frame_find_saved_regs (struct frame_info *fi, struct frame_saved_regs *fsr)
{
  register CORE_ADDR next_addr;
  register CORE_ADDR *saved_regs;
  register int regnum;
  register struct frame_saved_regs *cache_fsr;
  CORE_ADDR ip;
  struct symtab_and_line sal;
  CORE_ADDR limit;

  if (!fi->fsr)
    {
      cache_fsr = (struct frame_saved_regs *)
	frame_obstack_alloc (sizeof (struct frame_saved_regs));
      memset (cache_fsr, '\0', sizeof (struct frame_saved_regs));
      fi->fsr = cache_fsr;

      /* Find the start and end of the function prologue.  If the PC
         is in the function prologue, we only consider the part that
         has executed already.  */

      ip = get_pc_function_start (fi->pc);
      sal = find_pc_line (ip, 0);
      limit = (sal.end && sal.end < fi->pc) ? sal.end : fi->pc;

      examine_prologue (ip, limit, fi->frame, cache_fsr);

      /* Record the addresses at which the local registers are saved.
         Strictly speaking, we should only do this for non-leaf procedures,
         but no one will ever look at these values if it is a leaf procedure,
         since local registers are always caller-saved.  */

      next_addr = (CORE_ADDR) fi->frame;
      saved_regs = cache_fsr->regs;
      for (regnum = R0_REGNUM; regnum <= R15_REGNUM; regnum++)
	{
	  *saved_regs++ = next_addr;
	  next_addr += 4;
	}

      cache_fsr->regs[FP_REGNUM] = cache_fsr->regs[PFP_REGNUM];
    }

  *fsr = *fi->fsr;

  /* Fetch the value of the sp from memory every time, since it
     is conceivable that it has changed since the cache was flushed.  
     This unfortunately undoes much of the savings from caching the 
     saved register values.  I suggest adding an argument to 
     get_frame_saved_regs () specifying the register number we're
     interested in (or -1 for all registers).  This would be passed
     through to FRAME_FIND_SAVED_REGS (), permitting more efficient
     computation of saved register addresses (e.g., on the i960,
     we don't have to examine the prologue to find local registers). 
     -- markf@wrs.com 
     FIXME, we don't need to refetch this, since the cache is cleared
     every time the child process is restarted.  If GDB itself
     modifies SP, it has to clear the cache by hand (does it?).  -gnu */

  fsr->regs[SP_REGNUM] = read_memory_integer (fsr->regs[SP_REGNUM], 4);
}

/* Return the address of the argument block for the frame
   described by FI.  Returns 0 if the address is unknown.  */

CORE_ADDR
frame_args_address (struct frame_info *fi, int must_be_correct)
{
  struct frame_saved_regs fsr;
  CORE_ADDR ap;

  /* If g14 was saved in the frame by the function prologue code, return
     the saved value.  If the frame is current and we are being sloppy,
     return the value of g14.  Otherwise, return zero.  */

  get_frame_saved_regs (fi, &fsr);
  if (fsr.regs[G14_REGNUM])
    ap = read_memory_integer (fsr.regs[G14_REGNUM], 4);
  else
    {
      if (must_be_correct)
	return 0;		/* Don't cache this result */
      if (get_next_frame (fi))
	ap = 0;
      else
	ap = read_register (G14_REGNUM);
      if (ap == 0)
	ap = fi->frame;
    }
  fi->arg_pointer = ap;		/* Cache it for next time */
  return ap;
}

/* Return the address of the return struct for the frame
   described by FI.  Returns 0 if the address is unknown.  */

CORE_ADDR
frame_struct_result_address (struct frame_info *fi)
{
  struct frame_saved_regs fsr;
  CORE_ADDR ap;

  /* If the frame is non-current, check to see if g14 was saved in the
     frame by the function prologue code; return the saved value if so,
     zero otherwise.  If the frame is current, return the value of g14.

     FIXME, shouldn't this use the saved value as long as we are past
     the function prologue, and only use the current value if we have
     no saved value and are at TOS?   -- gnu@cygnus.com */

  if (get_next_frame (fi))
    {
      get_frame_saved_regs (fi, &fsr);
      if (fsr.regs[G13_REGNUM])
	ap = read_memory_integer (fsr.regs[G13_REGNUM], 4);
      else
	ap = 0;
    }
  else
    ap = read_register (G13_REGNUM);

  return ap;
}

/* Return address to which the currently executing leafproc will return,
   or 0 if IP, the value of the instruction pointer from the currently
   executing function, is not in a leafproc (or if we can't tell if it
   is).

   Do this by finding the starting address of the routine in which IP lies.
   If the instruction there is "mov g14, gx" (where x is in [0,7]), this
   is a leafproc and the return address is in register gx.  Well, this is
   true unless the return address points at a RET instruction in the current
   procedure, which indicates that we have a 'dual entry' routine that
   has been entered through the CALL entry point.  */

CORE_ADDR
leafproc_return (CORE_ADDR ip)
{
  register struct minimal_symbol *msymbol;
  char *p;
  int dst;
  unsigned int insn1, insn2;
  CORE_ADDR return_addr;

  if ((msymbol = lookup_minimal_symbol_by_pc (ip)) != NULL)
    {
      if ((p = strchr (SYMBOL_NAME (msymbol), '.')) && STREQ (p, ".lf"))
	{
	  if (next_insn (SYMBOL_VALUE_ADDRESS (msymbol), &insn1, &insn2)
	      && (insn1 & 0xff87ffff) == 0x5c80161e	/* mov g14, gx */
	      && (dst = REG_SRCDST (insn1)) <= G0_REGNUM + 7)
	    {
	      /* Get the return address.  If the "mov g14, gx" 
	         instruction hasn't been executed yet, read
	         the return address from g14; otherwise, read it
	         from the register into which g14 was moved.  */

	      return_addr =
		read_register ((ip == SYMBOL_VALUE_ADDRESS (msymbol))
			       ? G14_REGNUM : dst);

	      /* We know we are in a leaf procedure, but we don't know
	         whether the caller actually did a "bal" to the ".lf"
	         entry point, or a normal "call" to the non-leaf entry
	         point one instruction before.  In the latter case, the
	         return address will be the address of a "ret"
	         instruction within the procedure itself.  We test for
	         this below.  */

	      if (!next_insn (return_addr, &insn1, &insn2)
		  || (insn1 & 0xff000000) != 0xa000000	/* ret */
		  || lookup_minimal_symbol_by_pc (return_addr) != msymbol)
		return (return_addr);
	    }
	}
    }

  return (0);
}

/* Immediately after a function call, return the saved pc.
   Can't go through the frames for this because on some machines
   the new frame is not set up until the new function executes
   some instructions. 
   On the i960, the frame *is* set up immediately after the call,
   unless the function is a leaf procedure.  */

CORE_ADDR
saved_pc_after_call (struct frame_info *frame)
{
  CORE_ADDR saved_pc;

  saved_pc = leafproc_return (get_frame_pc (frame));
  if (!saved_pc)
    saved_pc = FRAME_SAVED_PC (frame);

  return saved_pc;
}

/* Discard from the stack the innermost frame,
   restoring all saved registers.  */

void
i960_pop_frame (void)
{
  register struct frame_info *current_fi, *prev_fi;
  register int i;
  CORE_ADDR save_addr;
  CORE_ADDR leaf_return_addr;
  struct frame_saved_regs fsr;
  char local_regs_buf[16 * 4];

  current_fi = get_current_frame ();

  /* First, undo what the hardware does when we return.
     If this is a non-leaf procedure, restore local registers from
     the save area in the calling frame.  Otherwise, load the return
     address obtained from leafproc_return () into the rip.  */

  leaf_return_addr = leafproc_return (current_fi->pc);
  if (!leaf_return_addr)
    {
      /* Non-leaf procedure.  Restore local registers, incl IP.  */
      prev_fi = get_prev_frame (current_fi);
      read_memory (prev_fi->frame, local_regs_buf, sizeof (local_regs_buf));
      write_register_bytes (REGISTER_BYTE (R0_REGNUM), local_regs_buf,
			    sizeof (local_regs_buf));

      /* Restore frame pointer.  */
      write_register (FP_REGNUM, prev_fi->frame);
    }
  else
    {
      /* Leaf procedure.  Just restore the return address into the IP.  */
      write_register (RIP_REGNUM, leaf_return_addr);
    }

  /* Now restore any global regs that the current function had saved. */
  get_frame_saved_regs (current_fi, &fsr);
  for (i = G0_REGNUM; i < G14_REGNUM; i++)
    {
      save_addr = fsr.regs[i];
      if (save_addr != 0)
	write_register (i, read_memory_integer (save_addr, 4));
    }

  /* Flush the frame cache, create a frame for the new innermost frame,
     and make it the current frame.  */

  flush_cached_frames ();
}

/* Given a 960 stop code (fault or trace), return the signal which
   corresponds.  */

enum target_signal
i960_fault_to_signal (int fault)
{
  switch (fault)
    {
    case 0:
      return TARGET_SIGNAL_BUS;	/* parallel fault */
    case 1:
      return TARGET_SIGNAL_UNKNOWN;
    case 2:
      return TARGET_SIGNAL_ILL;	/* operation fault */
    case 3:
      return TARGET_SIGNAL_FPE;	/* arithmetic fault */
    case 4:
      return TARGET_SIGNAL_FPE;	/* floating point fault */

      /* constraint fault.  This appears not to distinguish between
         a range constraint fault (which should be SIGFPE) and a privileged
         fault (which should be SIGILL).  */
    case 5:
      return TARGET_SIGNAL_ILL;

    case 6:
      return TARGET_SIGNAL_SEGV;	/* virtual memory fault */

      /* protection fault.  This is for an out-of-range argument to
         "calls".  I guess it also could be SIGILL. */
    case 7:
      return TARGET_SIGNAL_SEGV;

    case 8:
      return TARGET_SIGNAL_BUS;	/* machine fault */
    case 9:
      return TARGET_SIGNAL_BUS;	/* structural fault */
    case 0xa:
      return TARGET_SIGNAL_ILL;	/* type fault */
    case 0xb:
      return TARGET_SIGNAL_UNKNOWN;	/* reserved fault */
    case 0xc:
      return TARGET_SIGNAL_BUS;	/* process fault */
    case 0xd:
      return TARGET_SIGNAL_SEGV;	/* descriptor fault */
    case 0xe:
      return TARGET_SIGNAL_BUS;	/* event fault */
    case 0xf:
      return TARGET_SIGNAL_UNKNOWN;	/* reserved fault */
    case 0x10:
      return TARGET_SIGNAL_TRAP;	/* single-step trace */
    case 0x11:
      return TARGET_SIGNAL_TRAP;	/* branch trace */
    case 0x12:
      return TARGET_SIGNAL_TRAP;	/* call trace */
    case 0x13:
      return TARGET_SIGNAL_TRAP;	/* return trace */
    case 0x14:
      return TARGET_SIGNAL_TRAP;	/* pre-return trace */
    case 0x15:
      return TARGET_SIGNAL_TRAP;	/* supervisor call trace */
    case 0x16:
      return TARGET_SIGNAL_TRAP;	/* breakpoint trace */
    default:
      return TARGET_SIGNAL_UNKNOWN;
    }
}

/****************************************/
/* MEM format                           */
/****************************************/

struct tabent
{
  char *name;
  char numops;
};

/* Return instruction length, either 4 or 8.  When NOPRINT is non-zero
   (TRUE), don't output any text.  (Actually, as implemented, if NOPRINT
   is 0, abort() is called.) */

static int
mem (unsigned long memaddr, unsigned long word1, unsigned long word2,
     int noprint)
{
  int i, j;
  int len;
  int mode;
  int offset;
  const char *reg1, *reg2, *reg3;

  /* This lookup table is too sparse to make it worth typing in, but not
   * so large as to make a sparse array necessary.  We allocate the
   * table at runtime, initialize all entries to empty, and copy the
   * real ones in from an initialization table.
   *
   * NOTE: In this table, the meaning of 'numops' is:
   *       1: single operand
   *       2: 2 operands, load instruction
   *      -2: 2 operands, store instruction
   */
  static struct tabent *mem_tab = NULL;
/* Opcodes of 0x8X, 9X, aX, bX, and cX must be in the table.  */
#define MEM_MIN	0x80
#define MEM_MAX	0xcf
#define MEM_SIZ	((MEM_MAX-MEM_MIN+1) * sizeof(struct tabent))

  static struct
    {
      int opcode;
      char *name;
      char numops;
    }
  mem_init[] =
  {
    0x80, "ldob", 2,
      0x82, "stob", -2,
      0x84, "bx", 1,
      0x85, "balx", 2,
      0x86, "callx", 1,
      0x88, "ldos", 2,
      0x8a, "stos", -2,
      0x8c, "lda", 2,
      0x90, "ld", 2,
      0x92, "st", -2,
      0x98, "ldl", 2,
      0x9a, "stl", -2,
      0xa0, "ldt", 2,
      0xa2, "stt", -2,
      0xb0, "ldq", 2,
      0xb2, "stq", -2,
      0xc0, "ldib", 2,
      0xc2, "stib", -2,
      0xc8, "ldis", 2,
      0xca, "stis", -2,
      0, NULL, 0
  };

  if (mem_tab == NULL)
    {
      mem_tab = (struct tabent *) xmalloc (MEM_SIZ);
      memset (mem_tab, '\0', MEM_SIZ);
      for (i = 0; mem_init[i].opcode != 0; i++)
	{
	  j = mem_init[i].opcode - MEM_MIN;
	  mem_tab[j].name = mem_init[i].name;
	  mem_tab[j].numops = mem_init[i].numops;
	}
    }

  i = ((word1 >> 24) & 0xff) - MEM_MIN;
  mode = (word1 >> 10) & 0xf;

  if ((mem_tab[i].name != NULL)	/* Valid instruction */
      && ((mode == 5) || (mode >= 12)))
    {				/* With 32-bit displacement */
      len = 8;
    }
  else
    {
      len = 4;
    }

  if (noprint)
    {
      return len;
    }
  internal_error (__FILE__, __LINE__, "failed internal consistency check");
}

/* Read the i960 instruction at 'memaddr' and return the address of 
   the next instruction after that, or 0 if 'memaddr' is not the
   address of a valid instruction.  The first word of the instruction
   is stored at 'pword1', and the second word, if any, is stored at
   'pword2'.  */

static CORE_ADDR
next_insn (CORE_ADDR memaddr, unsigned int *pword1, unsigned int *pword2)
{
  int len;
  char buf[8];

  /* Read the two (potential) words of the instruction at once,
     to eliminate the overhead of two calls to read_memory ().
     FIXME: Loses if the first one is readable but the second is not
     (e.g. last word of the segment).  */

  read_memory (memaddr, buf, 8);
  *pword1 = extract_unsigned_integer (buf, 4);
  *pword2 = extract_unsigned_integer (buf + 4, 4);

  /* Divide instruction set into classes based on high 4 bits of opcode */

  switch ((*pword1 >> 28) & 0xf)
    {
    case 0x0:
    case 0x1:			/* ctrl */

    case 0x2:
    case 0x3:			/* cobr */

    case 0x5:
    case 0x6:
    case 0x7:			/* reg */
      len = 4;
      break;

    case 0x8:
    case 0x9:
    case 0xa:
    case 0xb:
    case 0xc:
      len = mem (memaddr, *pword1, *pword2, 1);
      break;

    default:			/* invalid instruction */
      len = 0;
      break;
    }

  if (len)
    return memaddr + len;
  else
    return 0;
}

/* 'start_frame' is a variable in the MON960 runtime startup routine
   that contains the frame pointer of the 'start' routine (the routine
   that calls 'main').  By reading its contents out of remote memory,
   we can tell where the frame chain ends:  backtraces should halt before
   they display this frame.  */

int
mon960_frame_chain_valid (CORE_ADDR chain, struct frame_info *curframe)
{
  struct symbol *sym;
  struct minimal_symbol *msymbol;

  /* crtmon960.o is an assembler module that is assumed to be linked
   * first in an i80960 executable.  It contains the true entry point;
   * it performs startup up initialization and then calls 'main'.
   *
   * 'sf' is the name of a variable in crtmon960.o that is set
   *      during startup to the address of the first frame.
   *
   * 'a' is the address of that variable in 80960 memory.
   */
  static char sf[] = "start_frame";
  CORE_ADDR a;


  chain &= ~0x3f;		/* Zero low 6 bits because previous frame pointers
				   contain return status info in them.  */
  if (chain == 0)
    {
      return 0;
    }

  sym = lookup_symbol (sf, 0, VAR_NAMESPACE, (int *) NULL,
		       (struct symtab **) NULL);
  if (sym != 0)
    {
      a = SYMBOL_VALUE (sym);
    }
  else
    {
      msymbol = lookup_minimal_symbol (sf, NULL, NULL);
      if (msymbol == NULL)
	return 0;
      a = SYMBOL_VALUE_ADDRESS (msymbol);
    }

  return (chain != read_memory_integer (a, 4));
}


void
_initialize_i960_tdep (void)
{
  check_host ();

  tm_print_insn = print_insn_i960;
}