summaryrefslogtreecommitdiff
path: root/sim/common/genmloop.sh
blob: 6720799a3ef8f8ec575b8841e16564a68e7ae123 (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
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
# Generate the main loop of the simulator.
# Copyright (C) 1996, 1997, 1998, 1999 Free Software Foundation, Inc.
# Contributed by Cygnus Support.
#
# This file is part of the GNU simulators.
#
# 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, 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.
#
# This file creates two files: eng.hin and mloop.cin.
# eng.hin defines a few macros that specify what kind of engine was selected
# based on the arguments to this script.
# mloop.cin contains the engine.
#
# ??? Rename mloop.c to eng.c?
# ??? Rename mainloop.in to engine.in?
# ??? Add options to specify output file names?
# ??? Rename this file to genengine.sh?
#
# Syntax: genmloop.sh [options]
#
# Options:
#
# -mono | -multi
#    - specify single cpu or multiple cpus (number specifyable at runtime),
#      maximum number is a configuration parameter
#    - -multi wip
#
# -fast: include support for fast execution in addition to full featured mode
#
#	Full featured mode is for tracing, profiling, etc. and is always
#	provided.  Fast mode contains no frills, except speed.
#	A target need only provide a "full" version of one of
#	simple,scache,pbb.  If the target wants it can also provide a fast
#	version of same.  It can't provide more than this.
#	??? Later add ability to have another set of full/fast semantics
#	for use in with-devices/with-smp situations (pbb can be inappropriate
#	here).
#
# -full-switch: same as -fast but for full featured version of -switch
#	Only needed if -fast present.
#
# -simple: simple execution engine (the default)
#
#	This engine fetches and executes one instruction at a time.
#	Field extraction is done in the semantic routines.
#
#	??? There are two possible flavours of -simple.  One that extracts
#	fields in the semantic routine (which is what is implemented here),
#	and one that stores the extracted fields in ARGBUF before calling the
#	semantic routine.  The latter is essentially the -scache case with a
#	cache size of one (and the scache lookup code removed).  There are no
#	current uses of this and it's not clear when doing this would be a win.
#	More complicated ISA's that want to use -simple may find this a win.
#	Should this ever be desirable, implement a new engine style here and
#	call it -extract (or some such).  It's believed that the CGEN-generated
#	code for the -scache case would be usable here, so no new code
#	generation option would be needed for CGEN.
#
# -scache: use the scache to speed things up (not always a win)
#
#	This engine caches the extracted instruction before executing it.
#	When executing instructions they are first looked up in the scache.
#
# -pbb: same as -scache but extract a (pseudo-) basic block at a time
#
#	This engine is basically identical to the scache version except that
#	extraction is done a pseudo-basic-block at a time and the address of
#	the scache entry of a branch target is recorded as well.
#	Additional speedups are then possible by defering Ctrl-C checking
#	to the end of basic blocks and by threading the insns together.
#	We call them pseudo-basic-block's instead of just basic-blocks because
#	they're not necessarily basic-blocks, though normally are.
#
# -parallel-read: support parallel execution with read-before-exec support.
# -parallel-write: support parallel execution with write-after-exec support.
#
#	One of these options is specified in addition to -simple, -scache,
#	-pbb.  Note that while the code can determine if the cpu supports
#	parallel execution with HAVE_PARALLEL_INSNS [and thus this option is
#	technically unnecessary], having this option cuts down on the clutter
#	in the result.
#
# -switch file: specify file containing semantics implemented as a switch()
#
# -cpu <cpu-family>
#
#	Specify the cpu family name.
#
# -infile <input-file>
#
#	Specify the mainloop.in input file.
#
# Only one of -scache/-pbb may be selected.
# -simple is the default.
#
####
#
# TODO
# - build mainloop.in from .cpu file

type=mono
#scache=
#fast=
#full_switch=
#pbb=
parallel=no
switch=
cpu="unknown"
infile=""

while test $# -gt 0
do
	case $1 in
	-mono) type=mono ;;
	-multi) type=multi ;;
	-no-fast) ;;
	-fast) fast=yes ;;
	-full-switch) full_switch=yes ;;
	-simple) ;;
	-scache) scache=yes ;;
	-pbb) pbb=yes ;;
	-no-parallel) ;;
	-parallel-read) parallel=read ;;
	-parallel-write) parallel=write ;;
	-switch) shift ; switch=$1 ;;
	-cpu) shift ; cpu=$1 ;;
	-infile) shift ; infile=$1 ;;
	*) echo "unknown option: $1" >&2 ; exit 1 ;;
	esac
	shift
done

# Argument validation.

if [ x$scache = xyes -a x$pbb = xyes ] ; then
    echo "only one of -scache and -pbb may be selected" >&2
    exit 1
fi

if [ "x$cpu" = xunknown ] ; then
    echo "cpu family not specified" >&2
    exit 1
fi

if [ "x$infile" = x ] ; then
    echo "mainloop.in not specified" >&2
    exit 1
fi

lowercase='abcdefghijklmnopqrstuvwxyz'
uppercase='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
CPU=`echo ${cpu} | tr "${lowercase}" "${uppercase}"`

##########################################################################

rm -f eng.hin
exec 1>eng.hin

echo "/* engine configuration for ${cpu} */"
echo ""

echo "/* WITH_FAST: non-zero if a fast version of the engine is available"
echo "   in addition to the full-featured version.  */"
if [ x$fast = xyes ] ; then
	echo "#define WITH_FAST 1"
else
	echo "#define WITH_FAST 0"
fi

echo ""
echo "/* WITH_SCACHE_PBB_${CPU}: non-zero if the pbb engine was selected.  */"
if [ x$pbb = xyes ] ; then
	echo "#define WITH_SCACHE_PBB_${CPU} 1"
else
	echo "#define WITH_SCACHE_PBB_${CPU} 0"
fi

echo ""
echo "/* HAVE_PARALLEL_INSNS: non-zero if cpu can parallelly execute > 1 insn.  */"
if [ x$parallel != xno ] ; then
	echo "#define HAVE_PARALLEL_INSNS 1"
	if [ x$parallel = xread ] ; then
	    echo "/* Parallel execution is supported by read-before-exec.  */"
	    echo "#define WITH_PARALLEL_READ 1"
	    echo "#define WITH_PARALLEL_WRITE 0"
	else
	    echo "/* Parallel execution is supported by write-after-exec.  */"
	    echo "#define WITH_PARALLEL_READ 0"
	    echo "#define WITH_PARALLEL_WRITE 1"
	fi
else
	echo "#define HAVE_PARALLEL_INSNS 0"
	echo "#define WITH_PARALLEL_READ 0"
	echo "#define WITH_PARALLEL_WRITE 0"
fi

if [ "x$switch" != x ] ; then
	echo ""
	echo "/* WITH_SEM_SWITCH_FULL: non-zero if full-featured engine is"
	echo "   implemented as a switch().  */"
	if [ x$fast != xyes -o x$full_switch = xyes ] ; then
		echo "#define WITH_SEM_SWITCH_FULL 1"
	else
		echo "#define WITH_SEM_SWITCH_FULL 0"
	fi
	echo ""
	echo "/* WITH_SEM_SWITCH_FAST: non-zero if fast engine is"
	echo "   implemented as a switch().  */"
	if [ x$fast = xyes ] ; then
		echo "#define WITH_SEM_SWITCH_FAST 1"
	else
		echo "#define WITH_SEM_SWITCH_FAST 0"
	fi
fi

# Decls of functions we define.

echo ""
echo "/* Functions defined in the generated mainloop.c file"
echo "   (which doesn't necessarily have that file name).  */"
echo ""
echo "extern ENGINE_FN ${cpu}_engine_run_full;"
echo "extern ENGINE_FN ${cpu}_engine_run_fast;"

if [ x$pbb = xyes ] ; then
	echo ""
	echo "extern SEM_PC ${cpu}_pbb_begin (SIM_CPU *, int);"
	echo "extern SEM_PC ${cpu}_pbb_chain (SIM_CPU *, SEM_ARG);"
	echo "extern SEM_PC ${cpu}_pbb_cti_chain (SIM_CPU *, SEM_ARG, SEM_PC *, PCADDR);"
	echo "extern void ${cpu}_pbb_before (SIM_CPU *, SCACHE *);"
	echo "extern void ${cpu}_pbb_after (SIM_CPU *, SCACHE *);"
fi

##########################################################################

rm -f tmp-mloop.cin mloop.cin
exec 1>tmp-mloop.cin

# We use @cpu@ instead of ${cpu} because we still need to run sed to handle
# transformation of @cpu@ for mainloop.in, so there's no need to use ${cpu}
# here.

cat << EOF
/* This file is generated by the genmloop script.  DO NOT EDIT! */

/* Enable switch() support in cgen headers.  */
#define SEM_IN_SWITCH

#define WANT_CPU @cpu@
#define WANT_CPU_@CPU@

#include "sim-main.h"
#include "bfd.h"
#include "cgen-mem.h"
#include "cgen-ops.h"
#include "sim-assert.h"

/* Fill in the administrative ARGBUF fields required by all insns,
   virtual and real.  */

static INLINE void
@cpu@_fill_argbuf (const SIM_CPU *cpu, ARGBUF *abuf, const IDESC *idesc,
		    PCADDR pc, int fast_p)
{
#if WITH_SCACHE
  SEM_SET_CODE (abuf, idesc, fast_p);
  ARGBUF_ADDR (abuf) = pc;
#endif
  ARGBUF_IDESC (abuf) = idesc;
}

/* Fill in tracing/profiling fields of an ARGBUF.  */

static INLINE void
@cpu@_fill_argbuf_tp (const SIM_CPU *cpu, ARGBUF *abuf,
		       int trace_p, int profile_p)
{
  ARGBUF_TRACE_P (abuf) = trace_p;
  ARGBUF_PROFILE_P (abuf) = profile_p;
}

#if WITH_SCACHE_PBB

/* Emit the "x-before" handler.
   x-before is emitted before each insn (serial or parallel).
   This is as opposed to x-after which is only emitted at the end of a group
   of parallel insns.  */

static INLINE void
@cpu@_emit_before (SIM_CPU *current_cpu, SCACHE *sc, PCADDR pc, int first_p)
{
  ARGBUF *abuf = &sc[0].argbuf;
  const IDESC *id = & CPU_IDESC (current_cpu) [@CPU@_INSN_X_BEFORE];

  abuf->fields.before.first_p = first_p;
  @cpu@_fill_argbuf (current_cpu, abuf, id, pc, 0);
  /* no need to set trace_p,profile_p */
}

/* Emit the "x-after" handler.
   x-after is emitted after a serial insn or at the end of a group of
   parallel insns.  */

static INLINE void
@cpu@_emit_after (SIM_CPU *current_cpu, SCACHE *sc, PCADDR pc)
{
  ARGBUF *abuf = &sc[0].argbuf;
  const IDESC *id = & CPU_IDESC (current_cpu) [@CPU@_INSN_X_AFTER];

  @cpu@_fill_argbuf (current_cpu, abuf, id, pc, 0);
  /* no need to set trace_p,profile_p */
}

#endif /* WITH_SCACHE_PBB */

EOF

${SHELL} $infile support

##########################################################################

# Simple engine: fetch an instruction, execute the instruction.
#
# Instruction fields are not extracted into ARGBUF, they are extracted in
# the semantic routines themselves.  However, there is still a need to pass
# and return misc. information to the semantic routines so we still use ARGBUF.
# [One could certainly implement things differently and remove ARGBUF.
# It's not clear this is necessarily always a win.]
# ??? The use of the SCACHE struct is for consistency with the with-scache
# case though it might be a source of confusion.

if [ x$scache != xyes -a x$pbb != xyes ] ; then

    cat << EOF

#define FAST_P 0

void
@cpu@_engine_run_full (SIM_CPU *current_cpu)
{
#define FAST_P 0
  SIM_DESC current_state = CPU_STATE (current_cpu);
  /* ??? Use of SCACHE is a bit of a hack as we don't actually use the scache.
     We do however use ARGBUF so for consistency with the other engine flavours
     the SCACHE type is used.  */
  SCACHE cache[MAX_LIW_INSNS];
  SCACHE *sc = &cache[0];

EOF

if [ x$parallel != xno ] ; then
  cat << EOF
  PAREXEC pbufs[MAX_PARALLEL_INSNS];
  PAREXEC *par_exec;

EOF
fi

# Any initialization code before looping starts.
# Note that this code may declare some locals.
${SHELL} $infile init

if [ x$parallel != xno ] ; then
  cat << EOF

#if defined (HAVE_PARALLEL_EXEC) && defined (__GNUC__)
  {
    if (! CPU_IDESC_READ_INIT_P (current_cpu))
      {
/* ??? Later maybe paste read.c in when building mainloop.c.  */
#define DEFINE_LABELS
#include "readx.c"
	CPU_IDESC_READ_INIT_P (current_cpu) = 1;
      }
  }
#endif

EOF
fi

cat << EOF

#if WITH_SEM_SWITCH_FULL && defined (__GNUC__)
  {
    if (! CPU_IDESC_SEM_INIT_P (current_cpu))
      {
/* ??? Later maybe paste sem-switch.c in when building mainloop.c.  */
#define DEFINE_LABELS
#include "$switch"
	CPU_IDESC_SEM_INIT_P (current_cpu) = 1;
      }
  }
#endif

  do
    {
/* begin full-exec-simple */
EOF

${SHELL} $infile full-exec-simple

cat << EOF
/* end full-exec-simple */

      ++ CPU_INSN_COUNT (current_cpu);
    }
  while (0 /*CPU_RUNNING_P (current_cpu)*/);
}

#undef FAST_P

EOF

####################################

# Simple engine: fast version.
# ??? A somewhat dubious effort, but for completeness' sake.

if [ x$fast = xyes ] ; then

    cat << EOF

#define FAST_P 1

FIXME: "fast simple version unimplemented, delete -fast arg to genmloop.sh."

#undef FAST_P

EOF

fi # -fast

fi # simple engine

##########################################################################

# Scache engine: lookup insn in scache, fetch if missing, then execute it.

if [ x$scache = xyes ] ; then

    cat << EOF

static INLINE SCACHE *
@cpu@_scache_lookup (SIM_CPU *current_cpu, PCADDR vpc, SCACHE *scache,
                     unsigned int hash_mask, int FAST_P)
{
  /* First step: look up current insn in hash table.  */
  SCACHE *sc = scache + SCACHE_HASH_PC (vpc, hash_mask);

  /* If the entry isn't the one we want (cache miss),
     fetch and decode the instruction.  */
  if (sc->argbuf.addr != vpc)
    {
      if (FAST_P)
	PROFILE_COUNT_SCACHE_MISS (current_cpu);

/* begin extract-scache */
EOF

${SHELL} $infile extract-scache

cat << EOF
/* end extract-scache */
    }
  else if (FAST_P)
    {
      PROFILE_COUNT_SCACHE_HIT (current_cpu);
      /* Make core access statistics come out right.
	 The size is a guess, but it's currently not used either.  */
      PROFILE_COUNT_CORE (current_cpu, vpc, 2, exec_map);
    }

  return sc;
}

#define FAST_P 0

void
@cpu@_engine_run_full (SIM_CPU *current_cpu)
{
  SIM_DESC current_state = CPU_STATE (current_cpu);
  SCACHE *scache = CPU_SCACHE_CACHE (current_cpu);
  unsigned int hash_mask = CPU_SCACHE_HASH_MASK (current_cpu);
  SEM_PC vpc;

EOF

if [ x$parallel != xno ] ; then
  cat << EOF
  PAREXEC pbufs[MAX_PARALLEL_INSNS];
  PAREXEC *par_exec;

EOF
fi

# Any initialization code before looping starts.
# Note that this code may declare some locals.
${SHELL} $infile init

if [ x$parallel != xno ] ; then
  cat << EOF

#if defined (HAVE_PARALLEL_EXEC) && defined (__GNUC__)
  {
    if (! CPU_IDESC_READ_INIT_P (current_cpu))
      {
/* ??? Later maybe paste read.c in when building mainloop.c.  */
#define DEFINE_LABELS
#include "readx.c"
	CPU_IDESC_READ_INIT_P (current_cpu) = 1;
      }
  }
#endif

EOF
fi

cat << EOF

  vpc = GET_H_PC ();

  do
    {
      SCACHE *sc;

      sc = @cpu@_scache_lookup (current_cpu, vpc, scache, hash_mask, FAST_P);

/* begin full-exec-scache */
EOF

${SHELL} $infile full-exec-scache

cat << EOF
/* end full-exec-scache */

      SET_H_PC (vpc);

      ++ CPU_INSN_COUNT (current_cpu);
    }
  while (0 /*CPU_RUNNING_P (current_cpu)*/);
}

#undef FAST_P

EOF

####################################

# Scache engine: fast version.

if [ x$fast = xyes ] ; then

    cat << EOF

#define FAST_P 1

void
@cpu@_engine_run_fast (SIM_CPU *current_cpu)
{
  SIM_DESC current_state = CPU_STATE (current_cpu);
  SCACHE *scache = CPU_SCACHE_CACHE (current_cpu);
  unsigned int hash_mask = CPU_SCACHE_HASH_MASK (current_cpu);
  SEM_PC vpc;

EOF

if [ x$parallel != xno ] ; then
  cat << EOF
  PAREXEC pbufs[MAX_PARALLEL_INSNS];
  PAREXEC *par_exec;

EOF
fi

# Any initialization code before looping starts.
# Note that this code may declare some locals.
${SHELL} $infile init

if [ x$parallel != xno ] ; then
  cat << EOF

#if defined (HAVE_PARALLEL_EXEC) && defined (__GNUC__)
  {
    if (! CPU_IDESC_READ_INIT_P (current_cpu))
      {
/* ??? Later maybe paste read.c in when building mainloop.c.  */
#define DEFINE_LABELS
#include "readx.c"
	CPU_IDESC_READ_INIT_P (current_cpu) = 1;
      }
  }
#endif

EOF
fi # parallel != no

cat << EOF

#if WITH_SEM_SWITCH_FAST && defined (__GNUC__)
  {
    if (! CPU_IDESC_SEM_INIT_P (current_cpu))
      {
/* ??? Later maybe paste sem-switch.c in when building mainloop.c.  */
#define DEFINE_LABELS
#include "$switch"
	CPU_IDESC_SEM_INIT_P (current_cpu) = 1;
      }
  }
#endif

  vpc = GET_H_PC ();

  do
    {
      SCACHE *sc;

      sc = @cpu@_scache_lookup (current_cpu, vpc, scache, hash_mask, FAST_P);

/* begin fast-exec-scache */
EOF

${SHELL} $infile fast-exec-scache

cat << EOF
/* end fast-exec-scache */

      SET_H_PC (vpc);

      ++ CPU_INSN_COUNT (current_cpu);
    }
  while (0 /*CPU_RUNNING_P (current_cpu)*/);
}

#undef FAST_P

EOF

fi # -fast

fi # -scache

##########################################################################

# Compilation engine: lookup insn in scache, extract a pbb
# (pseudo-basic-block) if missing, then execute the pbb.
# A "pbb" is a sequence of insns up to the next cti insn or until
# some prespecified maximum.
# CTI: control transfer instruction.

if [ x$pbb = xyes ] ; then

    cat << EOF

/* Record address of cti terminating a pbb.  */
#define SET_CTI_VPC(sc) do { _cti_sc = (sc); } while (0)
/* Record number of [real] insns in pbb.  */
#define SET_INSN_COUNT(n) do { _insn_count = (n); } while (0)

/* Fetch and extract a pseudo-basic-block.
   FAST_P is non-zero if no tracing/profiling/etc. is wanted.  */

INLINE SEM_PC
@cpu@_pbb_begin (SIM_CPU *current_cpu, int FAST_P)
{
  SEM_PC new_vpc;
  PCADDR pc;
  SCACHE *sc;
  int max_insns = CPU_SCACHE_MAX_CHAIN_LENGTH (current_cpu);

  pc = GET_H_PC ();

  new_vpc = scache_lookup_or_alloc (current_cpu, pc, max_insns, &sc);
  if (! new_vpc)
    {
      /* Leading '_' to avoid collision with mainloop.in.  */
      int _insn_count = 0;
      SCACHE *orig_sc = sc;
      SCACHE *_cti_sc = NULL;
      int slice_insns = CPU_MAX_SLICE_INSNS (current_cpu);

      /* First figure out how many instructions to compile.
	 MAX_INSNS is the size of the allocated buffer, which includes space
	 for before/after handlers if they're being used.
	 SLICE_INSNS is the maxinum number of real insns that can be
	 executed.  Zero means "as many as we want".  */
      /* ??? max_insns is serving two incompatible roles.
	 1) Number of slots available in scache buffer.
	 2) Number of real insns to execute.
	 They're incompatible because there are virtual insns emitted too
	 (chain,cti-chain,before,after handlers).  */

      if (slice_insns == 1)
	{
	  /* No need to worry about extra slots required for virtual insns
	     and parallel exec support because MAX_CHAIN_LENGTH is
	     guaranteed to be big enough to execute at least 1 insn!  */
	  max_insns = 1;
	}
      else
	{
	  /* Allow enough slop so that while compiling insns, if max_insns > 0
	     then there's guaranteed to be enough space to emit one real insn.
	     MAX_CHAIN_LENGTH is typically much longer than
	     the normal number of insns between cti's anyway.  */
	  max_insns -= (1 /* one for the trailing chain insn */
			+ (FAST_P
			   ? 0
			   : (1 + MAX_PARALLEL_INSNS) /* before+after */)
			+ (MAX_PARALLEL_INSNS > 1
			   ? (MAX_PARALLEL_INSNS * 2)
			   : 0));

	  /* Account for before/after handlers.  */
	  if (! FAST_P)
	    slice_insns *= 3;

	  if (slice_insns > 0
	      && slice_insns < max_insns)
	    max_insns = slice_insns;
	}

      new_vpc = sc;

      /* SC,PC must be updated to point passed the last entry used.
	 SET_CTI_VPC must be called if pbb is terminated by a cti.
	 SET_INSN_COUNT must be called to record number of real insns in
	 pbb [could be computed by us of course, extra cpu but perhaps
	 negligible enough].  */

/* begin extract-pbb */
EOF

${SHELL} $infile extract-pbb

cat << EOF
/* end extract-pbb */

      /* The last one is a pseudo-insn to link to the next chain.
	 It is also used to record the insn count for this chain.  */
      {
	const IDESC *id;

	/* Was pbb terminated by a cti?  */
	if (_cti_sc)
	  {
	    id = & CPU_IDESC (current_cpu) [@CPU@_INSN_X_CTI_CHAIN];
	  }
	else
	  {
	    id = & CPU_IDESC (current_cpu) [@CPU@_INSN_X_CHAIN];
	  }
	SEM_SET_CODE (&sc->argbuf, id, FAST_P);
	sc->argbuf.idesc = id;
	sc->argbuf.addr = pc;
	sc->argbuf.fields.chain.insn_count = _insn_count;
	sc->argbuf.fields.chain.next = 0;
	++sc;
      }

      /* Update the pointer to the next free entry, may not have used as
	 many entries as was asked for.  */
      CPU_SCACHE_NEXT_FREE (current_cpu) = sc;
      /* Record length of chain if profiling.
	 This includes virtual insns since they count against
	 max_insns too.  */
      if (! FAST_P)
	PROFILE_COUNT_SCACHE_CHAIN_LENGTH (current_cpu, sc - orig_sc);
    }

  return new_vpc;
}

/* Chain to the next block from a non-cti terminated previous block.  */

INLINE SEM_PC
@cpu@_pbb_chain (SIM_CPU *current_cpu, SEM_ARG sem_arg)
{
  ARGBUF *abuf = SEM_ARGBUF (sem_arg);

  PBB_UPDATE_INSN_COUNT (current_cpu, sem_arg);

  SET_H_PC (abuf->addr);

  /* If not running forever, exit back to main loop.  */
  if (CPU_MAX_SLICE_INSNS (current_cpu) != 0
      /* Also exit back to main loop if there's an event.
         Note that if CPU_MAX_SLICE_INSNS != 1, events won't get processed
	 at the "right" time, but then that was what was asked for.
	 There is no silver bullet for simulator engines.
         ??? Clearly this needs a cleaner interface.
	 At present it's just so Ctrl-C works.  */
      || STATE_EVENTS (CPU_STATE (current_cpu))->work_pending)
    CPU_RUNNING_P (current_cpu) = 0;

  /* If chained to next block, go straight to it.  */
  if (abuf->fields.chain.next)
    return abuf->fields.chain.next;
  /* See if next block has already been compiled.  */
  abuf->fields.chain.next = scache_lookup (current_cpu, abuf->addr);
  if (abuf->fields.chain.next)
    return abuf->fields.chain.next;
  /* Nope, so next insn is a virtual insn to invoke the compiler
     (begin a pbb).  */
  return CPU_SCACHE_PBB_BEGIN (current_cpu);
}

/* Chain to the next block from a cti terminated previous block.
   NEW_VPC_PTR is one of SEM_BRANCH_UNTAKEN, SEM_BRANCH_UNCACHEABLE, or
   a pointer to a location containing the SEM_PC of the branch's address.
   NEW_PC is the target's branch address, and is only valid if
   NEW_VPC_PTR != SEM_BRANCH_UNTAKEN.  */

INLINE SEM_PC
@cpu@_pbb_cti_chain (SIM_CPU *current_cpu, SEM_ARG sem_arg,
		     SEM_PC *new_vpc_ptr, PCADDR new_pc)
{
  PBB_UPDATE_INSN_COUNT (current_cpu, sem_arg);

  /* If not running forever, exit back to main loop.  */
  if (CPU_MAX_SLICE_INSNS (current_cpu) != 0
      /* Also exit back to main loop if there's an event.
         Note that if CPU_MAX_SLICE_INSNS != 1, events won't get processed
	 at the "right" time, but then that was what was asked for.
	 There is no silver bullet for simulator engines.
         ??? Clearly this needs a cleaner interface.
	 At present it's just so Ctrl-C works.  */
      || STATE_EVENTS (CPU_STATE (current_cpu))->work_pending)
    CPU_RUNNING_P (current_cpu) = 0;

  /* Restart compiler if we branched to an uncacheable address
     (e.g. "j reg").  */
  if (new_vpc_ptr == SEM_BRANCH_UNCACHEABLE)
    {
      SET_H_PC (new_pc);
      return CPU_SCACHE_PBB_BEGIN (current_cpu);
    }

  /* If branch wasn't taken, update the pc and set BR_ADDR_PTR to our
     next chain ptr.  */
  if (new_vpc_ptr == SEM_BRANCH_UNTAKEN)
    {
      ARGBUF *abuf = SEM_ARGBUF (sem_arg);
      SET_H_PC (abuf->addr);
      new_vpc_ptr = &abuf->fields.chain.next;
    }
  else
    {
      SET_H_PC (new_pc);
    }

  /* If chained to next block, go straight to it.  */
  if (*new_vpc_ptr)
    return *new_vpc_ptr;
  /* See if next block has already been compiled.  */
  *new_vpc_ptr = scache_lookup (current_cpu, GET_H_PC ());
  if (*new_vpc_ptr)
    return *new_vpc_ptr;
  /* Nope, so next insn is a virtual insn to invoke the compiler
     (begin a pbb).  */
  return CPU_SCACHE_PBB_BEGIN (current_cpu);
}

/* x-before handler.
   This is called before each insn.  */

void
@cpu@_pbb_before (SIM_CPU *current_cpu, SCACHE *sc)
{
  SEM_ARG sem_arg = sc;
  const ARGBUF *abuf = SEM_ARGBUF (sem_arg);
  int first_p = abuf->fields.before.first_p;
  const ARGBUF *cur_abuf = SEM_ARGBUF (sc + 1);
  const IDESC *cur_idesc = cur_abuf->idesc;
  PCADDR pc = cur_abuf->addr;

  if (ARGBUF_PROFILE_P (cur_abuf))
    PROFILE_COUNT_INSN (current_cpu, pc, cur_idesc->num);

  /* If this isn't the first insn, finish up the previous one.  */

  if (! first_p)
    {
      if (PROFILE_MODEL_P (current_cpu))
	{
	  const SEM_ARG prev_sem_arg = sc - 1;
	  const ARGBUF *prev_abuf = SEM_ARGBUF (prev_sem_arg);
	  const IDESC *prev_idesc = prev_abuf->idesc;
	  int cycles;

	  /* ??? May want to measure all insns if doing insn tracing.  */
	  if (ARGBUF_PROFILE_P (prev_abuf))
	    {
	      cycles = (*prev_idesc->timing->model_fn) (current_cpu, prev_sem_arg);
	      @cpu@_model_insn_after (current_cpu, 0 /*last_p*/, cycles);
	    }
	}

      TRACE_INSN_FINI (current_cpu, cur_abuf, 0 /*last_p*/);
    }

  /* FIXME: Later make cover macros: PROFILE_INSN_{INIT,FINI}.  */
  if (PROFILE_MODEL_P (current_cpu)
      && ARGBUF_PROFILE_P (cur_abuf))
    @cpu@_model_insn_before (current_cpu, first_p);

  TRACE_INSN_INIT (current_cpu, cur_abuf, first_p);
  TRACE_INSN (current_cpu, cur_idesc->idata, cur_abuf, pc);
}

/* x-after handler.
   This is called after a serial insn or at the end of a group of parallel
   insns.  */

void
@cpu@_pbb_after (SIM_CPU *current_cpu, SCACHE *sc)
{
  SEM_ARG sem_arg = sc;
  const ARGBUF *abuf = SEM_ARGBUF (sem_arg);
  const SEM_ARG prev_sem_arg = sc - 1;
  const ARGBUF *prev_abuf = SEM_ARGBUF (prev_sem_arg);

  /* ??? May want to measure all insns if doing insn tracing.  */
  if (PROFILE_MODEL_P (current_cpu)
      && ARGBUF_PROFILE_P (prev_abuf))
    {
      const IDESC *prev_idesc = prev_abuf->idesc;
      int cycles;

      cycles = (*prev_idesc->timing->model_fn) (current_cpu, prev_sem_arg);
      @cpu@_model_insn_after (current_cpu, 1 /*last_p*/, cycles);
    }
  TRACE_INSN_FINI (current_cpu, prev_abuf, 1 /*last_p*/);
}

#define FAST_P 0

void
@cpu@_engine_run_full (SIM_CPU *current_cpu)
{
  SIM_DESC current_state = CPU_STATE (current_cpu);
  SCACHE *scache = CPU_SCACHE_CACHE (current_cpu);
  /* virtual program counter */
  SEM_PC vpc;
#if WITH_SEM_SWITCH_FULL
  /* For communication between cti's and cti-chain.  */
  PCADDR pbb_br_npc;
  SEM_PC *pbb_br_npc_ptr;
#endif

EOF

if [ x$parallel != xno ] ; then
  cat << EOF
  PAREXEC pbufs[MAX_PARALLEL_INSNS];
  PAREXEC *par_exec = &pbufs[0];

EOF
fi

# Any initialization code before looping starts.
# Note that this code may declare some locals.
${SHELL} $infile init

cat << EOF

  if (! CPU_IDESC_SEM_INIT_P (current_cpu))
    {
      /* ??? 'twould be nice to move this up a level and only call it once.
	 On the other hand, in the "let's go fast" case the test is only done
	 once per pbb (since we only return to the main loop at the end of
	 a pbb).  And in the "let's run until we're done" case we don't return
	 until the program exits.  */

#if WITH_SEM_SWITCH_FULL && defined (__GNUC__)
/* ??? Later maybe paste sem-switch.c in when building mainloop.c.  */
#define DEFINE_LABELS
#include "$switch"
#endif

      /* Initialize the "begin (compile) a pbb" virtual insn.  */
      vpc = CPU_SCACHE_PBB_BEGIN (current_cpu);
      SEM_SET_FULL_CODE (SEM_ARGBUF (vpc),
			 & CPU_IDESC (current_cpu) [@CPU@_INSN_X_BEGIN]);
      vpc->argbuf.idesc = & CPU_IDESC (current_cpu) [@CPU@_INSN_X_BEGIN];

      CPU_IDESC_SEM_INIT_P (current_cpu) = 1;
    }

  CPU_RUNNING_P (current_cpu) = 1;
  /* ??? In the case where we're returning to the main loop after every
     pbb we don't want to call pbb_begin each time (which hashes on the pc
     and does a table lookup).  A way to speed this up is to save vpc
     between calls.  */
  vpc = @cpu@_pbb_begin (current_cpu, FAST_P);

  do
    {
/* begin full-exec-pbb */
EOF

${SHELL} $infile full-exec-pbb

cat << EOF
/* end full-exec-pbb */
    }
  while (CPU_RUNNING_P (current_cpu));
}

#undef FAST_P

EOF

####################################

# Compile engine: fast version.

if [ x$fast = xyes ] ; then

    cat << EOF

#define FAST_P 1

void
@cpu@_engine_run_fast (SIM_CPU *current_cpu)
{
  SIM_DESC current_state = CPU_STATE (current_cpu);
  SCACHE *scache = CPU_SCACHE_CACHE (current_cpu);
  /* virtual program counter */
  SEM_PC vpc;
#if WITH_SEM_SWITCH_FAST
  /* For communication between cti's and cti-chain.  */
  PCADDR pbb_br_npc;
  SEM_PC *pbb_br_npc_ptr;
#endif

EOF

if [ x$parallel != xno ] ; then
  cat << EOF
  PAREXEC pbufs[MAX_PARALLEL_INSNS];
  PAREXEC *par_exec = &pbufs[0];

EOF
fi

# Any initialization code before looping starts.
# Note that this code may declare some locals.
${SHELL} $infile init

cat << EOF

  if (! CPU_IDESC_SEM_INIT_P (current_cpu))
    {
      /* ??? 'twould be nice to move this up a level and only call it once.
	 On the other hand, in the "let's go fast" case the test is only done
	 once per pbb (since we only return to the main loop at the end of
	 a pbb).  And in the "let's run until we're done" case we don't return
	 until the program exits.  */

#if WITH_SEM_SWITCH_FAST && defined (__GNUC__)
/* ??? Later maybe paste sem-switch.c in when building mainloop.c.  */
#define DEFINE_LABELS
#include "$switch"
#endif

      /* Initialize the "begin (compile) a pbb" virtual insn.  */
      vpc = CPU_SCACHE_PBB_BEGIN (current_cpu);
      SEM_SET_FAST_CODE (SEM_ARGBUF (vpc),
			 & CPU_IDESC (current_cpu) [@CPU@_INSN_X_BEGIN]);
      vpc->argbuf.idesc = & CPU_IDESC (current_cpu) [@CPU@_INSN_X_BEGIN];

      CPU_IDESC_SEM_INIT_P (current_cpu) = 1;
    }

  CPU_RUNNING_P (current_cpu) = 1;
  /* ??? In the case where we're returning to the main loop after every
     pbb we don't want to call pbb_begin each time (which hashes on the pc
     and does a table lookup).  A way to speed this up is to save vpc
     between calls.  */
  vpc = @cpu@_pbb_begin (current_cpu, FAST_P);

  do
    {
/* begin fast-exec-pbb */
EOF

${SHELL} $infile fast-exec-pbb

cat << EOF
/* end fast-exec-pbb */
    }
  while (CPU_RUNNING_P (current_cpu));
}

#undef FAST_P

EOF
fi # -fast

fi # -pbb

# Process @cpu@,@CPU@ appearing in mainloop.in.
sed -e "s/@cpu@/$cpu/g" -e "s/@CPU@/$CPU/g" < tmp-mloop.cin > mloop.cin
rc=$?
rm -f tmp-mloop.cin

exit $rc