summaryrefslogtreecommitdiff
path: root/src/arch/isa_parser.py
blob: bbdd95bb00fe76b5f3179cc4be6be38321290b7a (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
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
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
# Copyright (c) 2003-2005 The Regents of The University of Michigan
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are
# met: redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer;
# redistributions in binary form must reproduce the above copyright
# notice, this list of conditions and the following disclaimer in the
# documentation and/or other materials provided with the distribution;
# neither the name of the copyright holders nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#
# Authors: Steve Reinhardt
#          Korey Sewell

import os
import sys
import re
import string
import traceback
# get type names
from types import *

# Prepend the directory where the PLY lex & yacc modules are found
# to the search path.  Assumes we're compiling in a subdirectory
# of 'build' in the current tree.
sys.path[0:0] = [os.environ['M5_PLY']]

from ply import lex
from ply import yacc

#####################################################################
#
#                                Lexer
#
# The PLY lexer module takes two things as input:
# - A list of token names (the string list 'tokens')
# - A regular expression describing a match for each token.  The
#   regexp for token FOO can be provided in two ways:
#   - as a string variable named t_FOO
#   - as the doc string for a function named t_FOO.  In this case,
#     the function is also executed, allowing an action to be
#     associated with each token match.
#
#####################################################################

# Reserved words.  These are listed separately as they are matched
# using the same regexp as generic IDs, but distinguished in the
# t_ID() function.  The PLY documentation suggests this approach.
reserved = (
    'BITFIELD', 'DECODE', 'DECODER', 'DEFAULT', 'DEF', 'EXEC', 'FORMAT',
    'HEADER', 'LET', 'NAMESPACE', 'OPERAND_TYPES', 'OPERANDS',
    'OUTPUT', 'SIGNED', 'TEMPLATE'
    )

# List of tokens.  The lex module requires this.
tokens = reserved + (
    # identifier
    'ID',

    # integer literal
    'INTLIT',

    # string literal
    'STRLIT',

    # code literal
    'CODELIT',

    # ( ) [ ] { } < > , ; . : :: *
    'LPAREN', 'RPAREN',
    'LBRACKET', 'RBRACKET',
    'LBRACE', 'RBRACE',
    'LESS', 'GREATER', 'EQUALS',
    'COMMA', 'SEMI', 'DOT', 'COLON', 'DBLCOLON',
    'ASTERISK',

    # C preprocessor directives
    'CPPDIRECTIVE'

# The following are matched but never returned. commented out to
# suppress PLY warning
    # newfile directive
#    'NEWFILE',

    # endfile directive
#    'ENDFILE'
)

# Regular expressions for token matching
t_LPAREN           = r'\('
t_RPAREN           = r'\)'
t_LBRACKET         = r'\['
t_RBRACKET         = r'\]'
t_LBRACE           = r'\{'
t_RBRACE           = r'\}'
t_LESS             = r'\<'
t_GREATER          = r'\>'
t_EQUALS           = r'='
t_COMMA            = r','
t_SEMI             = r';'
t_DOT              = r'\.'
t_COLON            = r':'
t_DBLCOLON         = r'::'
t_ASTERISK	   = r'\*'

# Identifiers and reserved words
reserved_map = { }
for r in reserved:
    reserved_map[r.lower()] = r

def t_ID(t):
    r'[A-Za-z_]\w*'
    t.type = reserved_map.get(t.value,'ID')
    return t

# Integer literal
def t_INTLIT(t):
    r'(0x[\da-fA-F]+)|\d+'
    try:
        t.value = int(t.value,0)
    except ValueError:
        error(t.lexer.lineno, 'Integer value "%s" too large' % t.value)
        t.value = 0
    return t

# String literal.  Note that these use only single quotes, and
# can span multiple lines.
def t_STRLIT(t):
    r"(?m)'([^'])+'"
    # strip off quotes
    t.value = t.value[1:-1]
    t.lexer.lineno += t.value.count('\n')
    return t


# "Code literal"... like a string literal, but delimiters are
# '{{' and '}}' so they get formatted nicely under emacs c-mode
def t_CODELIT(t):
    r"(?m)\{\{([^\}]|}(?!\}))+\}\}"
    # strip off {{ & }}
    t.value = t.value[2:-2]
    t.lexer.lineno += t.value.count('\n')
    return t

def t_CPPDIRECTIVE(t):
    r'^\#[^\#].*\n'
    t.lexer.lineno += t.value.count('\n')
    return t

def t_NEWFILE(t):
    r'^\#\#newfile\s+"[\w/.-]*"'
    fileNameStack.push((t.value[11:-1], t.lexer.lineno))
    t.lexer.lineno = 0

def t_ENDFILE(t):
    r'^\#\#endfile'
    (old_filename, t.lexer.lineno) = fileNameStack.pop()

#
# The functions t_NEWLINE, t_ignore, and t_error are
# special for the lex module.
#

# Newlines
def t_NEWLINE(t):
    r'\n+'
    t.lexer.lineno += t.value.count('\n')

# Comments
def t_comment(t):
    r'//.*'

# Completely ignored characters
t_ignore           = ' \t\x0c'

# Error handler
def t_error(t):
    error(t.lexer.lineno, "illegal character '%s'" % t.value[0])
    t.skip(1)

# Build the lexer
lexer = lex.lex()

#####################################################################
#
#                                Parser
#
# Every function whose name starts with 'p_' defines a grammar rule.
# The rule is encoded in the function's doc string, while the
# function body provides the action taken when the rule is matched.
# The argument to each function is a list of the values of the
# rule's symbols: t[0] for the LHS, and t[1..n] for the symbols
# on the RHS.  For tokens, the value is copied from the t.value
# attribute provided by the lexer.  For non-terminals, the value
# is assigned by the producing rule; i.e., the job of the grammar
# rule function is to set the value for the non-terminal on the LHS
# (by assigning to t[0]).
#####################################################################

# The LHS of the first grammar rule is used as the start symbol
# (in this case, 'specification').  Note that this rule enforces
# that there will be exactly one namespace declaration, with 0 or more
# global defs/decls before and after it.  The defs & decls before
# the namespace decl will be outside the namespace; those after
# will be inside.  The decoder function is always inside the namespace.
def p_specification(t):
    'specification : opt_defs_and_outputs name_decl opt_defs_and_outputs decode_block'
    global_code = t[1]
    isa_name = t[2]
    namespace = isa_name + "Inst"
    # wrap the decode block as a function definition
    t[4].wrap_decode_block('''
StaticInstPtr
%(isa_name)s::decodeInst(%(isa_name)s::ExtMachInst machInst)
{
    using namespace %(namespace)s;
''' % vars(), '}')
    # both the latter output blocks and the decode block are in the namespace
    namespace_code = t[3] + t[4]
    # pass it all back to the caller of yacc.parse()
    t[0] = (isa_name, namespace, global_code, namespace_code)

# ISA name declaration looks like "namespace <foo>;"
def p_name_decl(t):
    'name_decl : NAMESPACE ID SEMI'
    t[0] = t[2]

# 'opt_defs_and_outputs' is a possibly empty sequence of
# def and/or output statements.
def p_opt_defs_and_outputs_0(t):
    'opt_defs_and_outputs : empty'
    t[0] = GenCode()

def p_opt_defs_and_outputs_1(t):
    'opt_defs_and_outputs : defs_and_outputs'
    t[0] = t[1]

def p_defs_and_outputs_0(t):
    'defs_and_outputs : def_or_output'
    t[0] = t[1]

def p_defs_and_outputs_1(t):
    'defs_and_outputs : defs_and_outputs def_or_output'
    t[0] = t[1] + t[2]

# The list of possible definition/output statements.
def p_def_or_output(t):
    '''def_or_output : def_format
                     | def_bitfield
                     | def_bitfield_struct
                     | def_template
                     | def_operand_types
                     | def_operands
                     | output_header
                     | output_decoder
                     | output_exec
                     | global_let'''
    t[0] = t[1]

# Output blocks 'output <foo> {{...}}' (C++ code blocks) are copied
# directly to the appropriate output section.


# Protect any non-dict-substitution '%'s in a format string
# (i.e. those not followed by '(')
def protect_non_subst_percents(s):
    return re.sub(r'%(?!\()', '%%', s)

# Massage output block by substituting in template definitions and bit
# operators.  We handle '%'s embedded in the string that don't
# indicate template substitutions (or CPU-specific symbols, which get
# handled in GenCode) by doubling them first so that the format
# operation will reduce them back to single '%'s.
def process_output(s):
    s = protect_non_subst_percents(s)
    # protects cpu-specific symbols too
    s = protect_cpu_symbols(s)
    return substBitOps(s % templateMap)

def p_output_header(t):
    'output_header : OUTPUT HEADER CODELIT SEMI'
    t[0] = GenCode(header_output = process_output(t[3]))

def p_output_decoder(t):
    'output_decoder : OUTPUT DECODER CODELIT SEMI'
    t[0] = GenCode(decoder_output = process_output(t[3]))

def p_output_exec(t):
    'output_exec : OUTPUT EXEC CODELIT SEMI'
    t[0] = GenCode(exec_output = process_output(t[3]))

# global let blocks 'let {{...}}' (Python code blocks) are executed
# directly when seen.  Note that these execute in a special variable
# context 'exportContext' to prevent the code from polluting this
# script's namespace.
def p_global_let(t):
    'global_let : LET CODELIT SEMI'
    updateExportContext()
    exportContext["header_output"] = ''
    exportContext["decoder_output"] = ''
    exportContext["exec_output"] = ''
    exportContext["decode_block"] = ''
    try:
        exec fixPythonIndentation(t[2]) in exportContext
    except Exception, exc:
        error(t.lexer.lineno,
              'error: %s in global let block "%s".' % (exc, t[2]))
    t[0] = GenCode(header_output = exportContext["header_output"],
                   decoder_output = exportContext["decoder_output"],
                   exec_output = exportContext["exec_output"],
                   decode_block = exportContext["decode_block"])

# Define the mapping from operand type extensions to C++ types and bit
# widths (stored in operandTypeMap).
def p_def_operand_types(t):
    'def_operand_types : DEF OPERAND_TYPES CODELIT SEMI'
    try:
        userDict = eval('{' + t[3] + '}')
    except Exception, exc:
        error(t.lexer.lineno,
              'error: %s in def operand_types block "%s".' % (exc, t[3]))
    buildOperandTypeMap(userDict, t.lexer.lineno)
    t[0] = GenCode() # contributes nothing to the output C++ file

# Define the mapping from operand names to operand classes and other
# traits.  Stored in operandNameMap.
def p_def_operands(t):
    'def_operands : DEF OPERANDS CODELIT SEMI'
    if not globals().has_key('operandTypeMap'):
        error(t.lexer.lineno,
              'error: operand types must be defined before operands')
    try:
        userDict = eval('{' + t[3] + '}')
    except Exception, exc:
        error(t.lexer.lineno,
              'error: %s in def operands block "%s".' % (exc, t[3]))
    buildOperandNameMap(userDict, t.lexer.lineno)
    t[0] = GenCode() # contributes nothing to the output C++ file

# A bitfield definition looks like:
# 'def [signed] bitfield <ID> [<first>:<last>]'
# This generates a preprocessor macro in the output file.
def p_def_bitfield_0(t):
    'def_bitfield : DEF opt_signed BITFIELD ID LESS INTLIT COLON INTLIT GREATER SEMI'
    expr = 'bits(machInst, %2d, %2d)' % (t[6], t[8])
    if (t[2] == 'signed'):
        expr = 'sext<%d>(%s)' % (t[6] - t[8] + 1, expr)
    hash_define = '#undef %s\n#define %s\t%s\n' % (t[4], t[4], expr)
    t[0] = GenCode(header_output = hash_define)

# alternate form for single bit: 'def [signed] bitfield <ID> [<bit>]'
def p_def_bitfield_1(t):
    'def_bitfield : DEF opt_signed BITFIELD ID LESS INTLIT GREATER SEMI'
    expr = 'bits(machInst, %2d, %2d)' % (t[6], t[6])
    if (t[2] == 'signed'):
        expr = 'sext<%d>(%s)' % (1, expr)
    hash_define = '#undef %s\n#define %s\t%s\n' % (t[4], t[4], expr)
    t[0] = GenCode(header_output = hash_define)

# alternate form for structure member: 'def bitfield <ID> <ID>'
def p_def_bitfield_struct(t):
    'def_bitfield_struct : DEF opt_signed BITFIELD ID id_with_dot SEMI'
    if (t[2] != ''):
        error(t.lexer.lineno, 'error: structure bitfields are always unsigned.')
    expr = 'machInst.%s' % t[5]
    hash_define = '#undef %s\n#define %s\t%s\n' % (t[4], t[4], expr)
    t[0] = GenCode(header_output = hash_define)

def p_id_with_dot_0(t):
    'id_with_dot : ID'
    t[0] = t[1]

def p_id_with_dot_1(t):
    'id_with_dot : ID DOT id_with_dot'
    t[0] = t[1] + t[2] + t[3]

def p_opt_signed_0(t):
    'opt_signed : SIGNED'
    t[0] = t[1]

def p_opt_signed_1(t):
    'opt_signed : empty'
    t[0] = ''

# Global map variable to hold templates
templateMap = {}

def p_def_template(t):
    'def_template : DEF TEMPLATE ID CODELIT SEMI'
    templateMap[t[3]] = Template(t[4])
    t[0] = GenCode()

# An instruction format definition looks like
# "def format <fmt>(<params>) {{...}};"
def p_def_format(t):
    'def_format : DEF FORMAT ID LPAREN param_list RPAREN CODELIT SEMI'
    (id, params, code) = (t[3], t[5], t[7])
    defFormat(id, params, code, t.lexer.lineno)
    t[0] = GenCode()

# The formal parameter list for an instruction format is a possibly
# empty list of comma-separated parameters.  Positional (standard,
# non-keyword) parameters must come first, followed by keyword
# parameters, followed by a '*foo' parameter that gets excess
# positional arguments (as in Python).  Each of these three parameter
# categories is optional.
#
# Note that we do not support the '**foo' parameter for collecting
# otherwise undefined keyword args.  Otherwise the parameter list is
# (I believe) identical to what is supported in Python.
#
# The param list generates a tuple, where the first element is a list of
# the positional params and the second element is a dict containing the
# keyword params.
def p_param_list_0(t):
    'param_list : positional_param_list COMMA nonpositional_param_list'
    t[0] = t[1] + t[3]

def p_param_list_1(t):
    '''param_list : positional_param_list
                  | nonpositional_param_list'''
    t[0] = t[1]

def p_positional_param_list_0(t):
    'positional_param_list : empty'
    t[0] = []

def p_positional_param_list_1(t):
    'positional_param_list : ID'
    t[0] = [t[1]]

def p_positional_param_list_2(t):
    'positional_param_list : positional_param_list COMMA ID'
    t[0] = t[1] + [t[3]]

def p_nonpositional_param_list_0(t):
    'nonpositional_param_list : keyword_param_list COMMA excess_args_param'
    t[0] = t[1] + t[3]

def p_nonpositional_param_list_1(t):
    '''nonpositional_param_list : keyword_param_list
                                | excess_args_param'''
    t[0] = t[1]

def p_keyword_param_list_0(t):
    'keyword_param_list : keyword_param'
    t[0] = [t[1]]

def p_keyword_param_list_1(t):
    'keyword_param_list : keyword_param_list COMMA keyword_param'
    t[0] = t[1] + [t[3]]

def p_keyword_param(t):
    'keyword_param : ID EQUALS expr'
    t[0] = t[1] + ' = ' + t[3].__repr__()

def p_excess_args_param(t):
    'excess_args_param : ASTERISK ID'
    # Just concatenate them: '*ID'.  Wrap in list to be consistent
    # with positional_param_list and keyword_param_list.
    t[0] = [t[1] + t[2]]

# End of format definition-related rules.
##############

#
# A decode block looks like:
#	decode <field1> [, <field2>]* [default <inst>] { ... }
#
def p_decode_block(t):
    'decode_block : DECODE ID opt_default LBRACE decode_stmt_list RBRACE'
    default_defaults = defaultStack.pop()
    codeObj = t[5]
    # use the "default defaults" only if there was no explicit
    # default statement in decode_stmt_list
    if not codeObj.has_decode_default:
        codeObj += default_defaults
    codeObj.wrap_decode_block('switch (%s) {\n' % t[2], '}\n')
    t[0] = codeObj

# The opt_default statement serves only to push the "default defaults"
# onto defaultStack.  This value will be used by nested decode blocks,
# and used and popped off when the current decode_block is processed
# (in p_decode_block() above).
def p_opt_default_0(t):
    'opt_default : empty'
    # no default specified: reuse the one currently at the top of the stack
    defaultStack.push(defaultStack.top())
    # no meaningful value returned
    t[0] = None

def p_opt_default_1(t):
    'opt_default : DEFAULT inst'
    # push the new default
    codeObj = t[2]
    codeObj.wrap_decode_block('\ndefault:\n', 'break;\n')
    defaultStack.push(codeObj)
    # no meaningful value returned
    t[0] = None

def p_decode_stmt_list_0(t):
    'decode_stmt_list : decode_stmt'
    t[0] = t[1]

def p_decode_stmt_list_1(t):
    'decode_stmt_list : decode_stmt decode_stmt_list'
    if (t[1].has_decode_default and t[2].has_decode_default):
        error(t.lexer.lineno, 'Two default cases in decode block')
    t[0] = t[1] + t[2]

#
# Decode statement rules
#
# There are four types of statements allowed in a decode block:
# 1. Format blocks 'format <foo> { ... }'
# 2. Nested decode blocks
# 3. Instruction definitions.
# 4. C preprocessor directives.


# Preprocessor directives found in a decode statement list are passed
# through to the output, replicated to all of the output code
# streams.  This works well for ifdefs, so we can ifdef out both the
# declarations and the decode cases generated by an instruction
# definition.  Handling them as part of the grammar makes it easy to
# keep them in the right place with respect to the code generated by
# the other statements.
def p_decode_stmt_cpp(t):
    'decode_stmt : CPPDIRECTIVE'
    t[0] = GenCode(t[1], t[1], t[1], t[1])

# A format block 'format <foo> { ... }' sets the default instruction
# format used to handle instruction definitions inside the block.
# This format can be overridden by using an explicit format on the
# instruction definition or with a nested format block.
def p_decode_stmt_format(t):
    'decode_stmt : FORMAT push_format_id LBRACE decode_stmt_list RBRACE'
    # The format will be pushed on the stack when 'push_format_id' is
    # processed (see below).  Once the parser has recognized the full
    # production (though the right brace), we're done with the format,
    # so now we can pop it.
    formatStack.pop()
    t[0] = t[4]

# This rule exists so we can set the current format (& push the stack)
# when we recognize the format name part of the format block.
def p_push_format_id(t):
    'push_format_id : ID'
    try:
        formatStack.push(formatMap[t[1]])
        t[0] = ('', '// format %s' % t[1])
    except KeyError:
        error(t.lexer.lineno, 'instruction format "%s" not defined.' % t[1])

# Nested decode block: if the value of the current field matches the
# specified constant, do a nested decode on some other field.
def p_decode_stmt_decode(t):
    'decode_stmt : case_label COLON decode_block'
    label = t[1]
    codeObj = t[3]
    # just wrap the decoding code from the block as a case in the
    # outer switch statement.
    codeObj.wrap_decode_block('\n%s:\n' % label)
    codeObj.has_decode_default = (label == 'default')
    t[0] = codeObj

# Instruction definition (finally!).
def p_decode_stmt_inst(t):
    'decode_stmt : case_label COLON inst SEMI'
    label = t[1]
    codeObj = t[3]
    codeObj.wrap_decode_block('\n%s:' % label, 'break;\n')
    codeObj.has_decode_default = (label == 'default')
    t[0] = codeObj

# The case label is either a list of one or more constants or 'default'
def p_case_label_0(t):
    'case_label : intlit_list'
    t[0] = ': '.join(map(lambda a: 'case %#x' % a, t[1]))

def p_case_label_1(t):
    'case_label : DEFAULT'
    t[0] = 'default'

#
# The constant list for a decode case label must be non-empty, but may have
# one or more comma-separated integer literals in it.
#
def p_intlit_list_0(t):
    'intlit_list : INTLIT'
    t[0] = [t[1]]

def p_intlit_list_1(t):
    'intlit_list : intlit_list COMMA INTLIT'
    t[0] = t[1]
    t[0].append(t[3])

# Define an instruction using the current instruction format (specified
# by an enclosing format block).
# "<mnemonic>(<args>)"
def p_inst_0(t):
    'inst : ID LPAREN arg_list RPAREN'
    # Pass the ID and arg list to the current format class to deal with.
    currentFormat = formatStack.top()
    codeObj = currentFormat.defineInst(t[1], t[3], t.lexer.lineno)
    args = ','.join(map(str, t[3]))
    args = re.sub('(?m)^', '//', args)
    args = re.sub('^//', '', args)
    comment = '\n// %s::%s(%s)\n' % (currentFormat.id, t[1], args)
    codeObj.prepend_all(comment)
    t[0] = codeObj

# Define an instruction using an explicitly specified format:
# "<fmt>::<mnemonic>(<args>)"
def p_inst_1(t):
    'inst : ID DBLCOLON ID LPAREN arg_list RPAREN'
    try:
        format = formatMap[t[1]]
    except KeyError:
        error(t.lexer.lineno, 'instruction format "%s" not defined.' % t[1])
    codeObj = format.defineInst(t[3], t[5], t.lexer.lineno)
    comment = '\n// %s::%s(%s)\n' % (t[1], t[3], t[5])
    codeObj.prepend_all(comment)
    t[0] = codeObj

# The arg list generates a tuple, where the first element is a list of
# the positional args and the second element is a dict containing the
# keyword args.
def p_arg_list_0(t):
    'arg_list : positional_arg_list COMMA keyword_arg_list'
    t[0] = ( t[1], t[3] )

def p_arg_list_1(t):
    'arg_list : positional_arg_list'
    t[0] = ( t[1], {} )

def p_arg_list_2(t):
    'arg_list : keyword_arg_list'
    t[0] = ( [], t[1] )

def p_positional_arg_list_0(t):
    'positional_arg_list : empty'
    t[0] = []

def p_positional_arg_list_1(t):
    'positional_arg_list : expr'
    t[0] = [t[1]]

def p_positional_arg_list_2(t):
    'positional_arg_list : positional_arg_list COMMA expr'
    t[0] = t[1] + [t[3]]

def p_keyword_arg_list_0(t):
    'keyword_arg_list : keyword_arg'
    t[0] = t[1]

def p_keyword_arg_list_1(t):
    'keyword_arg_list : keyword_arg_list COMMA keyword_arg'
    t[0] = t[1]
    t[0].update(t[3])

def p_keyword_arg(t):
    'keyword_arg : ID EQUALS expr'
    t[0] = { t[1] : t[3] }

#
# Basic expressions.  These constitute the argument values of
# "function calls" (i.e. instruction definitions in the decode block)
# and default values for formal parameters of format functions.
#
# Right now, these are either strings, integers, or (recursively)
# lists of exprs (using Python square-bracket list syntax).  Note that
# bare identifiers are trated as string constants here (since there
# isn't really a variable namespace to refer to).
#
def p_expr_0(t):
    '''expr : ID
            | INTLIT
            | STRLIT
            | CODELIT'''
    t[0] = t[1]

def p_expr_1(t):
    '''expr : LBRACKET list_expr RBRACKET'''
    t[0] = t[2]

def p_list_expr_0(t):
    'list_expr : expr'
    t[0] = [t[1]]

def p_list_expr_1(t):
    'list_expr : list_expr COMMA expr'
    t[0] = t[1] + [t[3]]

def p_list_expr_2(t):
    'list_expr : empty'
    t[0] = []

#
# Empty production... use in other rules for readability.
#
def p_empty(t):
    'empty :'
    pass

# Parse error handler.  Note that the argument here is the offending
# *token*, not a grammar symbol (hence the need to use t.value)
def p_error(t):
    if t:
        error(t.lexer.lineno, "syntax error at '%s'" % t.value)
    else:
        error(0, "unknown syntax error", True)

# END OF GRAMMAR RULES
#
# Now build the parser.
parser = yacc.yacc()


#####################################################################
#
#                           Support Classes
#
#####################################################################

# Expand template with CPU-specific references into a dictionary with
# an entry for each CPU model name.  The entry key is the model name
# and the corresponding value is the template with the CPU-specific
# refs substituted for that model.
def expand_cpu_symbols_to_dict(template):
    # Protect '%'s that don't go with CPU-specific terms
    t = re.sub(r'%(?!\(CPU_)', '%%', template)
    result = {}
    for cpu in cpu_models:
        result[cpu.name] = t % cpu.strings
    return result

# *If* the template has CPU-specific references, return a single
# string containing a copy of the template for each CPU model with the
# corresponding values substituted in.  If the template has no
# CPU-specific references, it is returned unmodified.
def expand_cpu_symbols_to_string(template):
    if template.find('%(CPU_') != -1:
        return reduce(lambda x,y: x+y,
                      expand_cpu_symbols_to_dict(template).values())
    else:
        return template

# Protect CPU-specific references by doubling the corresponding '%'s
# (in preparation for substituting a different set of references into
# the template).
def protect_cpu_symbols(template):
    return re.sub(r'%(?=\(CPU_)', '%%', template)

###############
# GenCode class
#
# The GenCode class encapsulates generated code destined for various
# output files.  The header_output and decoder_output attributes are
# strings containing code destined for decoder.hh and decoder.cc
# respectively.  The decode_block attribute contains code to be
# incorporated in the decode function itself (that will also end up in
# decoder.cc).  The exec_output attribute is a dictionary with a key
# for each CPU model name; the value associated with a particular key
# is the string of code for that CPU model's exec.cc file.  The
# has_decode_default attribute is used in the decode block to allow
# explicit default clauses to override default default clauses.

class GenCode:
    # Constructor.  At this point we substitute out all CPU-specific
    # symbols.  For the exec output, these go into the per-model
    # dictionary.  For all other output types they get collapsed into
    # a single string.
    def __init__(self,
                 header_output = '', decoder_output = '', exec_output = '',
                 decode_block = '', has_decode_default = False):
        self.header_output = expand_cpu_symbols_to_string(header_output)
        self.decoder_output = expand_cpu_symbols_to_string(decoder_output)
        if isinstance(exec_output, dict):
            self.exec_output = exec_output
        elif isinstance(exec_output, str):
            # If the exec_output arg is a single string, we replicate
            # it for each of the CPU models, substituting and
            # %(CPU_foo)s params appropriately.
            self.exec_output = expand_cpu_symbols_to_dict(exec_output)
        self.decode_block = expand_cpu_symbols_to_string(decode_block)
        self.has_decode_default = has_decode_default

    # Override '+' operator: generate a new GenCode object that
    # concatenates all the individual strings in the operands.
    def __add__(self, other):
        exec_output = {}
        for cpu in cpu_models:
            n = cpu.name
            exec_output[n] = self.exec_output[n] + other.exec_output[n]
        return GenCode(self.header_output + other.header_output,
                       self.decoder_output + other.decoder_output,
                       exec_output,
                       self.decode_block + other.decode_block,
                       self.has_decode_default or other.has_decode_default)

    # Prepend a string (typically a comment) to all the strings.
    def prepend_all(self, pre):
        self.header_output = pre + self.header_output
        self.decoder_output  = pre + self.decoder_output
        self.decode_block = pre + self.decode_block
        for cpu in cpu_models:
            self.exec_output[cpu.name] = pre + self.exec_output[cpu.name]

    # Wrap the decode block in a pair of strings (e.g., 'case foo:'
    # and 'break;').  Used to build the big nested switch statement.
    def wrap_decode_block(self, pre, post = ''):
        self.decode_block = pre + indent(self.decode_block) + post

################
# Format object.
#
# A format object encapsulates an instruction format.  It must provide
# a defineInst() method that generates the code for an instruction
# definition.

exportContextSymbols = ('InstObjParams', 'makeList', 're', 'string')

exportContext = {}

def updateExportContext():
    exportContext.update(exportDict(*exportContextSymbols))
    exportContext.update(templateMap)

def exportDict(*symNames):
    return dict([(s, eval(s)) for s in symNames])


class Format:
    def __init__(self, id, params, code):
        # constructor: just save away arguments
        self.id = id
        self.params = params
        label = 'def format ' + id
        self.user_code = compile(fixPythonIndentation(code), label, 'exec')
        param_list = string.join(params, ", ")
        f = '''def defInst(_code, _context, %s):
                my_locals = vars().copy()
                exec _code in _context, my_locals
                return my_locals\n''' % param_list
        c = compile(f, label + ' wrapper', 'exec')
        exec c
        self.func = defInst

    def defineInst(self, name, args, lineno):
        context = {}
        updateExportContext()
        context.update(exportContext)
        if len(name):
            Name = name[0].upper()
            if len(name) > 1:
                Name += name[1:]
        context.update({ 'name': name, 'Name': Name })
        try:
            vars = self.func(self.user_code, context, *args[0], **args[1])
        except Exception, exc:
            error(lineno, 'error defining "%s": %s.' % (name, exc))
        for k in vars.keys():
            if k not in ('header_output', 'decoder_output',
                         'exec_output', 'decode_block'):
                del vars[k]
        return GenCode(**vars)

# Special null format to catch an implicit-format instruction
# definition outside of any format block.
class NoFormat:
    def __init__(self):
        self.defaultInst = ''

    def defineInst(self, name, args, lineno):
        error(lineno,
              'instruction definition "%s" with no active format!' % name)

# This dictionary maps format name strings to Format objects.
formatMap = {}

# Define a new format
def defFormat(id, params, code, lineno):
    # make sure we haven't already defined this one
    if formatMap.get(id, None) != None:
        error(lineno, 'format %s redefined.' % id)
    # create new object and store in global map
    formatMap[id] = Format(id, params, code)


##############
# Stack: a simple stack object.  Used for both formats (formatStack)
# and default cases (defaultStack).  Simply wraps a list to give more
# stack-like syntax and enable initialization with an argument list
# (as opposed to an argument that's a list).

class Stack(list):
    def __init__(self, *items):
        list.__init__(self, items)

    def push(self, item):
        self.append(item);

    def top(self):
        return self[-1]

# The global format stack.
formatStack = Stack(NoFormat())

# The global default case stack.
defaultStack = Stack( None )

# Global stack that tracks current file and line number.
# Each element is a tuple (filename, lineno) that records the
# *current* filename and the line number in the *previous* file where
# it was included.
fileNameStack = Stack()

###################
# Utility functions

#
# Indent every line in string 's' by two spaces
# (except preprocessor directives).
# Used to make nested code blocks look pretty.
#
def indent(s):
    return re.sub(r'(?m)^(?!#)', '  ', s)

#
# Munge a somewhat arbitrarily formatted piece of Python code
# (e.g. from a format 'let' block) into something whose indentation
# will get by the Python parser.
#
# The two keys here are that Python will give a syntax error if
# there's any whitespace at the beginning of the first line, and that
# all lines at the same lexical nesting level must have identical
# indentation.  Unfortunately the way code literals work, an entire
# let block tends to have some initial indentation.  Rather than
# trying to figure out what that is and strip it off, we prepend 'if
# 1:' to make the let code the nested block inside the if (and have
# the parser automatically deal with the indentation for us).
#
# We don't want to do this if (1) the code block is empty or (2) the
# first line of the block doesn't have any whitespace at the front.

def fixPythonIndentation(s):
    # get rid of blank lines first
    s = re.sub(r'(?m)^\s*\n', '', s);
    if (s != '' and re.match(r'[ \t]', s[0])):
        s = 'if 1:\n' + s
    return s

# Error handler.  Just call exit.  Output formatted to work under
# Emacs compile-mode.  Optional 'print_traceback' arg, if set to True,
# prints a Python stack backtrace too (can be handy when trying to
# debug the parser itself).
def error(lineno, string, print_traceback = False):
    spaces = ""
    for (filename, line) in fileNameStack[0:-1]:
        print spaces + "In file included from " + filename + ":"
        spaces += "  "
    # Print a Python stack backtrace if requested.
    if (print_traceback):
        traceback.print_exc()
    if lineno != 0:
        line_str = "%d:" % lineno
    else:
        line_str = ""
    sys.exit(spaces + "%s:%s %s" % (fileNameStack[-1][0], line_str, string))


#####################################################################
#
#                      Bitfield Operator Support
#
#####################################################################

bitOp1ArgRE = re.compile(r'<\s*(\w+)\s*:\s*>')

bitOpWordRE = re.compile(r'(?<![\w\.])([\w\.]+)<\s*(\w+)\s*:\s*(\w+)\s*>')
bitOpExprRE = re.compile(r'\)<\s*(\w+)\s*:\s*(\w+)\s*>')

def substBitOps(code):
    # first convert single-bit selectors to two-index form
    # i.e., <n> --> <n:n>
    code = bitOp1ArgRE.sub(r'<\1:\1>', code)
    # simple case: selector applied to ID (name)
    # i.e., foo<a:b> --> bits(foo, a, b)
    code = bitOpWordRE.sub(r'bits(\1, \2, \3)', code)
    # if selector is applied to expression (ending in ')'),
    # we need to search backward for matching '('
    match = bitOpExprRE.search(code)
    while match:
        exprEnd = match.start()
        here = exprEnd - 1
        nestLevel = 1
        while nestLevel > 0:
            if code[here] == '(':
                nestLevel -= 1
            elif code[here] == ')':
                nestLevel += 1
            here -= 1
            if here < 0:
                sys.exit("Didn't find '('!")
        exprStart = here+1
        newExpr = r'bits(%s, %s, %s)' % (code[exprStart:exprEnd+1],
                                         match.group(1), match.group(2))
        code = code[:exprStart] + newExpr + code[match.end():]
        match = bitOpExprRE.search(code)
    return code


####################
# Template objects.
#
# Template objects are format strings that allow substitution from
# the attribute spaces of other objects (e.g. InstObjParams instances).

labelRE = re.compile(r'(?<!%)%\(([^\)]+)\)[sd]')

class Template:
    def __init__(self, t):
        self.template = t

    def subst(self, d):
        myDict = None

        # Protect non-Python-dict substitutions (e.g. if there's a printf
        # in the templated C++ code)
        template = protect_non_subst_percents(self.template)
        # CPU-model-specific substitutions are handled later (in GenCode).
        template = protect_cpu_symbols(template)

        # Build a dict ('myDict') to use for the template substitution.
        # Start with the template namespace.  Make a copy since we're
        # going to modify it.
        myDict = templateMap.copy()

        if isinstance(d, InstObjParams):
            # If we're dealing with an InstObjParams object, we need
            # to be a little more sophisticated.  The instruction-wide
            # parameters are already formed, but the parameters which
            # are only function wide still need to be generated.
            compositeCode = ''

            myDict.update(d.__dict__)
            # The "operands" and "snippets" attributes of the InstObjParams
            # objects are for internal use and not substitution.
            del myDict['operands']
            del myDict['snippets']

            snippetLabels = [l for l in labelRE.findall(template)
                             if d.snippets.has_key(l)]

            snippets = dict([(s, mungeSnippet(d.snippets[s]))
                             for s in snippetLabels])

            myDict.update(snippets)

            compositeCode = ' '.join(map(str, snippets.values()))

            # Add in template itself in case it references any
            # operands explicitly (like Mem)
            compositeCode += ' ' + template

            operands = SubOperandList(compositeCode, d.operands)

            myDict['op_decl'] = operands.concatAttrStrings('op_decl')

            is_src = lambda op: op.is_src
            is_dest = lambda op: op.is_dest

            myDict['op_src_decl'] = \
                      operands.concatSomeAttrStrings(is_src, 'op_src_decl')
            myDict['op_dest_decl'] = \
                      operands.concatSomeAttrStrings(is_dest, 'op_dest_decl')

            myDict['op_rd'] = operands.concatAttrStrings('op_rd')
            myDict['op_wb'] = operands.concatAttrStrings('op_wb')

            if d.operands.memOperand:
                myDict['mem_acc_size'] = d.operands.memOperand.mem_acc_size
                myDict['mem_acc_type'] = d.operands.memOperand.mem_acc_type

        elif isinstance(d, dict):
            # if the argument is a dictionary, we just use it.
            myDict.update(d)
        elif hasattr(d, '__dict__'):
            # if the argument is an object, we use its attribute map.
            myDict.update(d.__dict__)
        else:
            raise TypeError, "Template.subst() arg must be or have dictionary"
        return template % myDict

    # Convert to string.  This handles the case when a template with a
    # CPU-specific term gets interpolated into another template or into
    # an output block.
    def __str__(self):
        return expand_cpu_symbols_to_string(self.template)

#####################################################################
#
#                             Code Parser
#
# The remaining code is the support for automatically extracting
# instruction characteristics from pseudocode.
#
#####################################################################

# Force the argument to be a list.  Useful for flags, where a caller
# can specify a singleton flag or a list of flags.  Also usful for
# converting tuples to lists so they can be modified.
def makeList(arg):
    if isinstance(arg, list):
        return arg
    elif isinstance(arg, tuple):
        return list(arg)
    elif not arg:
        return []
    else:
        return [ arg ]

# Generate operandTypeMap from the user's 'def operand_types'
# statement.
def buildOperandTypeMap(userDict, lineno):
    global operandTypeMap
    operandTypeMap = {}
    for (ext, (desc, size)) in userDict.iteritems():
        if desc == 'signed int':
            ctype = 'int%d_t' % size
            is_signed = 1
        elif desc == 'unsigned int':
            ctype = 'uint%d_t' % size
            is_signed = 0
        elif desc == 'float':
            is_signed = 1	# shouldn't really matter
            if size == 32:
                ctype = 'float'
            elif size == 64:
                ctype = 'double'
        elif desc == 'twin64 int':
            is_signed = 0
            ctype = 'Twin64_t'
        elif desc == 'twin32 int':
            is_signed = 0
            ctype = 'Twin32_t'
        if ctype == '':
            error(lineno, 'Unrecognized type description "%s" in userDict')
        operandTypeMap[ext] = (size, ctype, is_signed)

#
#
#
# Base class for operand descriptors.  An instance of this class (or
# actually a class derived from this one) represents a specific
# operand for a code block (e.g, "Rc.sq" as a dest). Intermediate
# derived classes encapsulates the traits of a particular operand type
# (e.g., "32-bit integer register").
#
class Operand(object):
    def __init__(self, full_name, ext, is_src, is_dest):
        self.full_name = full_name
        self.ext = ext
        self.is_src = is_src
        self.is_dest = is_dest
        # The 'effective extension' (eff_ext) is either the actual
        # extension, if one was explicitly provided, or the default.
        if ext:
            self.eff_ext = ext
        else:
            self.eff_ext = self.dflt_ext

        (self.size, self.ctype, self.is_signed) = operandTypeMap[self.eff_ext]

        # note that mem_acc_size is undefined for non-mem operands...
        # template must be careful not to use it if it doesn't apply.
        if self.isMem():
            self.mem_acc_size = self.makeAccSize()
            if self.ctype in ['Twin32_t', 'Twin64_t']:
                self.mem_acc_type = 'Twin'
            else:
                self.mem_acc_type = 'uint'

    # Finalize additional fields (primarily code fields).  This step
    # is done separately since some of these fields may depend on the
    # register index enumeration that hasn't been performed yet at the
    # time of __init__().
    def finalize(self):
        self.flags = self.getFlags()
        self.constructor = self.makeConstructor()
        self.op_decl = self.makeDecl()

        if self.is_src:
            self.op_rd = self.makeRead()
            self.op_src_decl = self.makeDecl()
        else:
            self.op_rd = ''
            self.op_src_decl = ''

        if self.is_dest:
            self.op_wb = self.makeWrite()
            self.op_dest_decl = self.makeDecl()
        else:
            self.op_wb = ''
            self.op_dest_decl = ''

    def isMem(self):
        return 0

    def isReg(self):
        return 0

    def isFloatReg(self):
        return 0

    def isIntReg(self):
        return 0

    def isControlReg(self):
        return 0

    def isIControlReg(self):
        return 0

    def getFlags(self):
        # note the empty slice '[:]' gives us a copy of self.flags[0]
        # instead of a reference to it
        my_flags = self.flags[0][:]
        if self.is_src:
            my_flags += self.flags[1]
        if self.is_dest:
            my_flags += self.flags[2]
        return my_flags

    def makeDecl(self):
        # Note that initializations in the declarations are solely
        # to avoid 'uninitialized variable' errors from the compiler.
        return self.ctype + ' ' + self.base_name + ' = 0;\n';

class IntRegOperand(Operand):
    def isReg(self):
        return 1

    def isIntReg(self):
        return 1

    def makeConstructor(self):
        c = ''
        if self.is_src:
            c += '\n\t_srcRegIdx[%d] = %s;' % \
                 (self.src_reg_idx, self.reg_spec)
        if self.is_dest:
            c += '\n\t_destRegIdx[%d] = %s;' % \
                 (self.dest_reg_idx, self.reg_spec)
        return c

    def makeRead(self):
        if (self.ctype == 'float' or self.ctype == 'double'):
            error(0, 'Attempt to read integer register as FP')
        if (self.size == self.dflt_size):
            return '%s = xc->readIntRegOperand(this, %d);\n' % \
                   (self.base_name, self.src_reg_idx)
        elif (self.size > self.dflt_size):
            int_reg_val = 'xc->readIntRegOperand(this, %d)' % \
                          (self.src_reg_idx)
            if (self.is_signed):
                int_reg_val = 'sext<%d>(%s)' % (self.dflt_size, int_reg_val)
            return '%s = %s;\n' % (self.base_name, int_reg_val)
        else:
            return '%s = bits(xc->readIntRegOperand(this, %d), %d, 0);\n' % \
                   (self.base_name, self.src_reg_idx, self.size-1)

    def makeWrite(self):
        if (self.ctype == 'float' or self.ctype == 'double'):
            error(0, 'Attempt to write integer register as FP')
        if (self.size != self.dflt_size and self.is_signed):
            final_val = 'sext<%d>(%s)' % (self.size, self.base_name)
        else:
            final_val = self.base_name
        wb = '''
        {
            %s final_val = %s;
            xc->setIntRegOperand(this, %d, final_val);\n
            if (traceData) { traceData->setData(final_val); }
        }''' % (self.dflt_ctype, final_val, self.dest_reg_idx)
        return wb

class FloatRegOperand(Operand):
    def isReg(self):
        return 1

    def isFloatReg(self):
        return 1

    def makeConstructor(self):
        c = ''
        if self.is_src:
            c += '\n\t_srcRegIdx[%d] = %s + FP_Base_DepTag;' % \
                 (self.src_reg_idx, self.reg_spec)
        if self.is_dest:
            c += '\n\t_destRegIdx[%d] = %s + FP_Base_DepTag;' % \
                 (self.dest_reg_idx, self.reg_spec)
        return c

    def makeRead(self):
        bit_select = 0
        width = 0;
        if (self.ctype == 'float'):
            func = 'readFloatRegOperand'
            width = 32;
        elif (self.ctype == 'double'):
            func = 'readFloatRegOperand'
            width = 64;
        else:
            func = 'readFloatRegOperandBits'
            if (self.ctype == 'uint32_t'):
                width = 32;
            elif (self.ctype == 'uint64_t'):
                width = 64;
            if (self.size != self.dflt_size):
                bit_select = 1
        if width:
            base = 'xc->%s(this, %d, %d)' % \
                   (func, self.src_reg_idx, width)
        else:
            base = 'xc->%s(this, %d)' % \
                   (func, self.src_reg_idx)
        if bit_select:
            return '%s = bits(%s, %d, 0);\n' % \
                   (self.base_name, base, self.size-1)
        else:
            return '%s = %s;\n' % (self.base_name, base)

    def makeWrite(self):
        final_val = self.base_name
        final_ctype = self.ctype
        widthSpecifier = ''
        width = 0
        if (self.ctype == 'float'):
            width = 32
            func = 'setFloatRegOperand'
        elif (self.ctype == 'double'):
            width = 64
            func = 'setFloatRegOperand'
        elif (self.ctype == 'uint32_t'):
            func = 'setFloatRegOperandBits'
            width = 32
        elif (self.ctype == 'uint64_t'):
            func = 'setFloatRegOperandBits'
            width = 64
        else:
            func = 'setFloatRegOperandBits'
            final_ctype = 'uint%d_t' % self.dflt_size
            if (self.size != self.dflt_size and self.is_signed):
                final_val = 'sext<%d>(%s)' % (self.size, self.base_name)
        if width:
            widthSpecifier = ', %d' % width
        wb = '''
        {
            %s final_val = %s;
            xc->%s(this, %d, final_val%s);\n
            if (traceData) { traceData->setData(final_val); }
        }''' % (final_ctype, final_val, func, self.dest_reg_idx,
                widthSpecifier)
        return wb

class ControlRegOperand(Operand):
    def isReg(self):
        return 1

    def isControlReg(self):
        return 1

    def makeConstructor(self):
        c = ''
        if self.is_src:
            c += '\n\t_srcRegIdx[%d] = %s + Ctrl_Base_DepTag;' % \
                 (self.src_reg_idx, self.reg_spec)
        if self.is_dest:
            c += '\n\t_destRegIdx[%d] = %s + Ctrl_Base_DepTag;' % \
                 (self.dest_reg_idx, self.reg_spec)
        return c

    def makeRead(self):
        bit_select = 0
        if (self.ctype == 'float' or self.ctype == 'double'):
            error(0, 'Attempt to read control register as FP')
        base = 'xc->readMiscRegOperand(this, %s)' % self.src_reg_idx
        if self.size == self.dflt_size:
            return '%s = %s;\n' % (self.base_name, base)
        else:
            return '%s = bits(%s, %d, 0);\n' % \
                   (self.base_name, base, self.size-1)

    def makeWrite(self):
        if (self.ctype == 'float' or self.ctype == 'double'):
            error(0, 'Attempt to write control register as FP')
        wb = 'xc->setMiscRegOperand(this, %s, %s);\n' % \
             (self.dest_reg_idx, self.base_name)
        wb += 'if (traceData) { traceData->setData(%s); }' % \
              self.base_name
        return wb

class IControlRegOperand(Operand):
    def isReg(self):
        return 1

    def isIControlReg(self):
        return 1

    def makeConstructor(self):
        c = ''
        if self.is_src:
            c += '\n\t_srcRegIdx[%d] = %s + Ctrl_Base_DepTag;' % \
                 (self.src_reg_idx, self.reg_spec)
        if self.is_dest:
            c += '\n\t_destRegIdx[%d] = %s + Ctrl_Base_DepTag;' % \
                 (self.dest_reg_idx, self.reg_spec)
        return c

    def makeRead(self):
        bit_select = 0
        if (self.ctype == 'float' or self.ctype == 'double'):
            error(0, 'Attempt to read control register as FP')
        base = 'xc->readMiscReg(%s)' % self.reg_spec
        if self.size == self.dflt_size:
            return '%s = %s;\n' % (self.base_name, base)
        else:
            return '%s = bits(%s, %d, 0);\n' % \
                   (self.base_name, base, self.size-1)

    def makeWrite(self):
        if (self.ctype == 'float' or self.ctype == 'double'):
            error(0, 'Attempt to write control register as FP')
        wb = 'xc->setMiscReg(%s, %s);\n' % \
             (self.reg_spec, self.base_name)
        wb += 'if (traceData) { traceData->setData(%s); }' % \
              self.base_name
        return wb

class ControlBitfieldOperand(ControlRegOperand):
    def makeRead(self):
        bit_select = 0
        if (self.ctype == 'float' or self.ctype == 'double'):
            error(0, 'Attempt to read control register as FP')
        base = 'xc->readMiscReg(%s)' % self.reg_spec
        name = self.base_name
        return '%s = bits(%s, %s_HI, %s_LO);' % \
               (name, base, name, name)

    def makeWrite(self):
        if (self.ctype == 'float' or self.ctype == 'double'):
            error(0, 'Attempt to write control register as FP')
        base = 'xc->readMiscReg(%s)' % self.reg_spec
        name = self.base_name
        wb_val = 'insertBits(%s, %s_HI, %s_LO, %s)' % \
                    (base, name, name, self.base_name)
        wb = 'xc->setMiscRegOperand(this, %s, %s );\n' % (self.dest_reg_idx, wb_val)
        wb += 'if (traceData) { traceData->setData(%s); }' % \
              self.base_name
        return wb

class MemOperand(Operand):
    def isMem(self):
        return 1

    def makeConstructor(self):
        return ''

    def makeDecl(self):
        # Note that initializations in the declarations are solely
        # to avoid 'uninitialized variable' errors from the compiler.
        # Declare memory data variable.
        if self.ctype in ['Twin32_t','Twin64_t']:
            return "%s %s; %s.a = 0; %s.b = 0;\n" % (self.ctype, self.base_name,
                    self.base_name, self.base_name)
        c = '%s %s = 0;\n' % (self.ctype, self.base_name)
        return c

    def makeRead(self):
        return ''

    def makeWrite(self):
        return ''

    # Return the memory access size *in bits*, suitable for
    # forming a type via "uint%d_t".  Divide by 8 if you want bytes.
    def makeAccSize(self):
        return self.size

class UPCOperand(Operand):
    def makeConstructor(self):
        return ''

    def makeRead(self):
        return '%s = xc->readMicroPC();\n' % self.base_name

    def makeWrite(self):
        return 'xc->setMicroPC(%s);\n' % self.base_name

class NUPCOperand(Operand):
    def makeConstructor(self):
        return ''

    def makeRead(self):
        return '%s = xc->readNextMicroPC();\n' % self.base_name

    def makeWrite(self):
        return 'xc->setNextMicroPC(%s);\n' % self.base_name

class NPCOperand(Operand):
    def makeConstructor(self):
        return ''

    def makeRead(self):
        return '%s = xc->readNextPC();\n' % self.base_name

    def makeWrite(self):
        return 'xc->setNextPC(%s);\n' % self.base_name

class NNPCOperand(Operand):
    def makeConstructor(self):
        return ''

    def makeRead(self):
        return '%s = xc->readNextNPC();\n' % self.base_name

    def makeWrite(self):
        return 'xc->setNextNPC(%s);\n' % self.base_name

def buildOperandNameMap(userDict, lineno):
    global operandNameMap
    operandNameMap = {}
    for (op_name, val) in userDict.iteritems():
        (base_cls_name, dflt_ext, reg_spec, flags, sort_pri) = val
        (dflt_size, dflt_ctype, dflt_is_signed) = operandTypeMap[dflt_ext]
        # Canonical flag structure is a triple of lists, where each list
        # indicates the set of flags implied by this operand always, when
        # used as a source, and when used as a dest, respectively.
        # For simplicity this can be initialized using a variety of fairly
        # obvious shortcuts; we convert these to canonical form here.
        if not flags:
            # no flags specified (e.g., 'None')
            flags = ( [], [], [] )
        elif isinstance(flags, str):
            # a single flag: assumed to be unconditional
            flags = ( [ flags ], [], [] )
        elif isinstance(flags, list):
            # a list of flags: also assumed to be unconditional
            flags = ( flags, [], [] )
        elif isinstance(flags, tuple):
            # it's a tuple: it should be a triple,
            # but each item could be a single string or a list
            (uncond_flags, src_flags, dest_flags) = flags
            flags = (makeList(uncond_flags),
                     makeList(src_flags), makeList(dest_flags))
        # Accumulate attributes of new operand class in tmp_dict
        tmp_dict = {}
        for attr in ('dflt_ext', 'reg_spec', 'flags', 'sort_pri',
                     'dflt_size', 'dflt_ctype', 'dflt_is_signed'):
            tmp_dict[attr] = eval(attr)
        tmp_dict['base_name'] = op_name
        # New class name will be e.g. "IntReg_Ra"
        cls_name = base_cls_name + '_' + op_name
        # Evaluate string arg to get class object.  Note that the
        # actual base class for "IntReg" is "IntRegOperand", i.e. we
        # have to append "Operand".
        try:
            base_cls = eval(base_cls_name + 'Operand')
        except NameError:
            error(lineno,
                  'error: unknown operand base class "%s"' % base_cls_name)
        # The following statement creates a new class called
        # <cls_name> as a subclass of <base_cls> with the attributes
        # in tmp_dict, just as if we evaluated a class declaration.
        operandNameMap[op_name] = type(cls_name, (base_cls,), tmp_dict)

    # Define operand variables.
    operands = userDict.keys()

    operandsREString = (r'''
    (?<![\w\.])	     # neg. lookbehind assertion: prevent partial matches
    ((%s)(?:\.(\w+))?)   # match: operand with optional '.' then suffix
    (?![\w\.])	     # neg. lookahead assertion: prevent partial matches
    '''
                        % string.join(operands, '|'))

    global operandsRE
    operandsRE = re.compile(operandsREString, re.MULTILINE|re.VERBOSE)

    # Same as operandsREString, but extension is mandatory, and only two
    # groups are returned (base and ext, not full name as above).
    # Used for subtituting '_' for '.' to make C++ identifiers.
    operandsWithExtREString = (r'(?<![\w\.])(%s)\.(\w+)(?![\w\.])'
                               % string.join(operands, '|'))

    global operandsWithExtRE
    operandsWithExtRE = re.compile(operandsWithExtREString, re.MULTILINE)

maxInstSrcRegs = 0
maxInstDestRegs = 0

class OperandList:

    # Find all the operands in the given code block.  Returns an operand
    # descriptor list (instance of class OperandList).
    def __init__(self, code):
        self.items = []
        self.bases = {}
        # delete comments so we don't match on reg specifiers inside
        code = commentRE.sub('', code)
        # search for operands
        next_pos = 0
        while 1:
            match = operandsRE.search(code, next_pos)
            if not match:
                # no more matches: we're done
                break
            op = match.groups()
            # regexp groups are operand full name, base, and extension
            (op_full, op_base, op_ext) = op
            # if the token following the operand is an assignment, this is
            # a destination (LHS), else it's a source (RHS)
            is_dest = (assignRE.match(code, match.end()) != None)
            is_src = not is_dest
            # see if we've already seen this one
            op_desc = self.find_base(op_base)
            if op_desc:
                if op_desc.ext != op_ext:
                    error(0, 'Inconsistent extensions for operand %s' % \
                          op_base)
                op_desc.is_src = op_desc.is_src or is_src
                op_desc.is_dest = op_desc.is_dest or is_dest
            else:
                # new operand: create new descriptor
                op_desc = operandNameMap[op_base](op_full, op_ext,
                                                  is_src, is_dest)
                self.append(op_desc)
            # start next search after end of current match
            next_pos = match.end()
        self.sort()
        # enumerate source & dest register operands... used in building
        # constructor later
        self.numSrcRegs = 0
        self.numDestRegs = 0
        self.numFPDestRegs = 0
        self.numIntDestRegs = 0
        self.memOperand = None
        for op_desc in self.items:
            if op_desc.isReg():
                if op_desc.is_src:
                    op_desc.src_reg_idx = self.numSrcRegs
                    self.numSrcRegs += 1
                if op_desc.is_dest:
                    op_desc.dest_reg_idx = self.numDestRegs
                    self.numDestRegs += 1
                    if op_desc.isFloatReg():
                        self.numFPDestRegs += 1
                    elif op_desc.isIntReg():
                        self.numIntDestRegs += 1
            elif op_desc.isMem():
                if self.memOperand:
                    error(0, "Code block has more than one memory operand.")
                self.memOperand = op_desc
        global maxInstSrcRegs
        global maxInstDestRegs
        if maxInstSrcRegs < self.numSrcRegs:
            maxInstSrcRegs = self.numSrcRegs
        if maxInstDestRegs < self.numDestRegs:
            maxInstDestRegs = self.numDestRegs
        # now make a final pass to finalize op_desc fields that may depend
        # on the register enumeration
        for op_desc in self.items:
            op_desc.finalize()

    def __len__(self):
        return len(self.items)

    def __getitem__(self, index):
        return self.items[index]

    def append(self, op_desc):
        self.items.append(op_desc)
        self.bases[op_desc.base_name] = op_desc

    def find_base(self, base_name):
        # like self.bases[base_name], but returns None if not found
        # (rather than raising exception)
        return self.bases.get(base_name)

    # internal helper function for concat[Some]Attr{Strings|Lists}
    def __internalConcatAttrs(self, attr_name, filter, result):
        for op_desc in self.items:
            if filter(op_desc):
                result += getattr(op_desc, attr_name)
        return result

    # return a single string that is the concatenation of the (string)
    # values of the specified attribute for all operands
    def concatAttrStrings(self, attr_name):
        return self.__internalConcatAttrs(attr_name, lambda x: 1, '')

    # like concatAttrStrings, but only include the values for the operands
    # for which the provided filter function returns true
    def concatSomeAttrStrings(self, filter, attr_name):
        return self.__internalConcatAttrs(attr_name, filter, '')

    # return a single list that is the concatenation of the (list)
    # values of the specified attribute for all operands
    def concatAttrLists(self, attr_name):
        return self.__internalConcatAttrs(attr_name, lambda x: 1, [])

    # like concatAttrLists, but only include the values for the operands
    # for which the provided filter function returns true
    def concatSomeAttrLists(self, filter, attr_name):
        return self.__internalConcatAttrs(attr_name, filter, [])

    def sort(self):
        self.items.sort(lambda a, b: a.sort_pri - b.sort_pri)

class SubOperandList(OperandList):

    # Find all the operands in the given code block.  Returns an operand
    # descriptor list (instance of class OperandList).
    def __init__(self, code, master_list):
        self.items = []
        self.bases = {}
        # delete comments so we don't match on reg specifiers inside
        code = commentRE.sub('', code)
        # search for operands
        next_pos = 0
        while 1:
            match = operandsRE.search(code, next_pos)
            if not match:
                # no more matches: we're done
                break
            op = match.groups()
            # regexp groups are operand full name, base, and extension
            (op_full, op_base, op_ext) = op
            # find this op in the master list
            op_desc = master_list.find_base(op_base)
            if not op_desc:
                error(0, 'Found operand %s which is not in the master list!' \
                        ' This is an internal error' % \
                          op_base)
            else:
                # See if we've already found this operand
                op_desc = self.find_base(op_base)
                if not op_desc:
                    # if not, add a reference to it to this sub list
                    self.append(master_list.bases[op_base])

            # start next search after end of current match
            next_pos = match.end()
        self.sort()
        self.memOperand = None
        for op_desc in self.items:
            if op_desc.isMem():
                if self.memOperand:
                    error(0, "Code block has more than one memory operand.")
                self.memOperand = op_desc

# Regular expression object to match C++ comments
# (used in findOperands())
commentRE = re.compile(r'//.*\n')

# Regular expression object to match assignment statements
# (used in findOperands())
assignRE = re.compile(r'\s*=(?!=)', re.MULTILINE)

# Munge operand names in code string to make legal C++ variable names.
# This means getting rid of the type extension if any.
# (Will match base_name attribute of Operand object.)
def substMungedOpNames(code):
    return operandsWithExtRE.sub(r'\1', code)

# Fix up code snippets for final substitution in templates.
def mungeSnippet(s):
    if isinstance(s, str):
        return substMungedOpNames(substBitOps(s))
    else:
        return s

def makeFlagConstructor(flag_list):
    if len(flag_list) == 0:
        return ''
    # filter out repeated flags
    flag_list.sort()
    i = 1
    while i < len(flag_list):
        if flag_list[i] == flag_list[i-1]:
            del flag_list[i]
        else:
            i += 1
    pre = '\n\tflags['
    post = '] = true;'
    code = pre + string.join(flag_list, post + pre) + post
    return code

# Assume all instruction flags are of the form 'IsFoo'
instFlagRE = re.compile(r'Is.*')

# OpClass constants end in 'Op' except No_OpClass
opClassRE = re.compile(r'.*Op|No_OpClass')

class InstObjParams:
    def __init__(self, mnem, class_name, base_class = '',
                 snippets = {}, opt_args = []):
        self.mnemonic = mnem
        self.class_name = class_name
        self.base_class = base_class
        if not isinstance(snippets, dict):
            snippets = {'code' : snippets}
        compositeCode = ' '.join(map(str, snippets.values()))
        self.snippets = snippets

        self.operands = OperandList(compositeCode)
        self.constructor = self.operands.concatAttrStrings('constructor')
        self.constructor += \
                 '\n\t_numSrcRegs = %d;' % self.operands.numSrcRegs
        self.constructor += \
                 '\n\t_numDestRegs = %d;' % self.operands.numDestRegs
        self.constructor += \
                 '\n\t_numFPDestRegs = %d;' % self.operands.numFPDestRegs
        self.constructor += \
                 '\n\t_numIntDestRegs = %d;' % self.operands.numIntDestRegs
        self.flags = self.operands.concatAttrLists('flags')

        # Make a basic guess on the operand class (function unit type).
        # These are good enough for most cases, and can be overridden
        # later otherwise.
        if 'IsStore' in self.flags:
            self.op_class = 'MemWriteOp'
        elif 'IsLoad' in self.flags or 'IsPrefetch' in self.flags:
            self.op_class = 'MemReadOp'
        elif 'IsFloating' in self.flags:
            self.op_class = 'FloatAddOp'
        else:
            self.op_class = 'IntAluOp'

        # Optional arguments are assumed to be either StaticInst flags
        # or an OpClass value.  To avoid having to import a complete
        # list of these values to match against, we do it ad-hoc
        # with regexps.
        for oa in opt_args:
            if instFlagRE.match(oa):
                self.flags.append(oa)
            elif opClassRE.match(oa):
                self.op_class = oa
            else:
                error(0, 'InstObjParams: optional arg "%s" not recognized '
                      'as StaticInst::Flag or OpClass.' % oa)

        # add flag initialization to contructor here to include
        # any flags added via opt_args
        self.constructor += makeFlagConstructor(self.flags)

        # if 'IsFloating' is set, add call to the FP enable check
        # function (which should be provided by isa_desc via a declare)
        if 'IsFloating' in self.flags:
            self.fp_enable_check = 'fault = checkFpEnableFault(xc);'
        else:
            self.fp_enable_check = ''

#######################
#
# Output file template
#

file_template = '''
/*
 * DO NOT EDIT THIS FILE!!!
 *
 * It was automatically generated from the ISA description in %(filename)s
 */

%(includes)s

%(global_output)s

namespace %(namespace)s {

%(namespace_output)s

} // namespace %(namespace)s

%(decode_function)s
'''

max_inst_regs_template = '''
/*
 * DO NOT EDIT THIS FILE!!!
 *
 * It was automatically generated from the ISA description in %(filename)s
 */

namespace %(namespace)s {

    const int MaxInstSrcRegs = %(MaxInstSrcRegs)d;
    const int MaxInstDestRegs = %(MaxInstDestRegs)d;

} // namespace %(namespace)s

'''


# Update the output file only if the new contents are different from
# the current contents.  Minimizes the files that need to be rebuilt
# after minor changes.
def update_if_needed(file, contents):
    update = False
    if os.access(file, os.R_OK):
        f = open(file, 'r')
        old_contents = f.read()
        f.close()
        if contents != old_contents:
            print 'Updating', file
            os.remove(file) # in case it's write-protected
            update = True
        else:
            print 'File', file, 'is unchanged'
    else:
        print 'Generating', file
        update = True
    if update:
        f = open(file, 'w')
        f.write(contents)
        f.close()

# This regular expression matches '##include' directives
includeRE = re.compile(r'^\s*##include\s+"(?P<filename>[\w/.-]*)".*$',
                       re.MULTILINE)

# Function to replace a matched '##include' directive with the
# contents of the specified file (with nested ##includes replaced
# recursively).  'matchobj' is an re match object (from a match of
# includeRE) and 'dirname' is the directory relative to which the file
# path should be resolved.
def replace_include(matchobj, dirname):
    fname = matchobj.group('filename')
    full_fname = os.path.normpath(os.path.join(dirname, fname))
    contents = '##newfile "%s"\n%s\n##endfile\n' % \
               (full_fname, read_and_flatten(full_fname))
    return contents

# Read a file and recursively flatten nested '##include' files.
def read_and_flatten(filename):
    current_dir = os.path.dirname(filename)
    try:
        contents = open(filename).read()
    except IOError:
        error(0, 'Error including file "%s"' % filename)
    fileNameStack.push((filename, 0))
    # Find any includes and include them
    contents = includeRE.sub(lambda m: replace_include(m, current_dir),
                             contents)
    fileNameStack.pop()
    return contents

#
# Read in and parse the ISA description.
#
def parse_isa_desc(isa_desc_file, output_dir):
    # Read file and (recursively) all included files into a string.
    # PLY requires that the input be in a single string so we have to
    # do this up front.
    isa_desc = read_and_flatten(isa_desc_file)

    # Initialize filename stack with outer file.
    fileNameStack.push((isa_desc_file, 0))

    # Parse it.
    (isa_name, namespace, global_code, namespace_code) = \
        parser.parse(isa_desc, lexer=lexer)

    # grab the last three path components of isa_desc_file to put in
    # the output
    filename = '/'.join(isa_desc_file.split('/')[-3:])

    # generate decoder.hh
    includes = '#include "base/bitfield.hh" // for bitfield support'
    global_output = global_code.header_output
    namespace_output = namespace_code.header_output
    decode_function = ''
    update_if_needed(output_dir + '/decoder.hh', file_template % vars())

    # generate decoder.cc
    includes = '#include "decoder.hh"'
    global_output = global_code.decoder_output
    namespace_output = namespace_code.decoder_output
    # namespace_output += namespace_code.decode_block
    decode_function = namespace_code.decode_block
    update_if_needed(output_dir + '/decoder.cc', file_template % vars())

    # generate per-cpu exec files
    for cpu in cpu_models:
        includes = '#include "decoder.hh"\n'
        includes += cpu.includes
        global_output = global_code.exec_output[cpu.name]
        namespace_output = namespace_code.exec_output[cpu.name]
        decode_function = ''
        update_if_needed(output_dir + '/' + cpu.filename,
                          file_template % vars())

    # The variable names here are hacky, but this will creat local variables
    # which will be referenced in vars() which have the value of the globals.
    global maxInstSrcRegs
    MaxInstSrcRegs = maxInstSrcRegs
    global maxInstDestRegs
    MaxInstDestRegs = maxInstDestRegs
    # max_inst_regs.hh
    update_if_needed(output_dir + '/max_inst_regs.hh', \
            max_inst_regs_template % vars())

# global list of CpuModel objects (see cpu_models.py)
cpu_models = []

# Called as script: get args from command line.
# Args are: <path to cpu_models.py> <isa desc file> <output dir> <cpu models>
if __name__ == '__main__':
    execfile(sys.argv[1])  # read in CpuModel definitions
    cpu_models = [CpuModel.dict[cpu] for cpu in sys.argv[4:]]
    parse_isa_desc(sys.argv[2], sys.argv[3])