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
|
#! /usr/bin/env python
# $Id$
# Copyright (c) 2003 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.
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_EXT'] + '/ply']
import lex
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', 'DECLARE', 'DECODE', 'DEFAULT', 'DEF', 'FORMAT',
'LET', 'NAMESPACE', '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',
# not used any more... commented out to suppress PLY warning
# 'LBRACKET', 'RBRACKET',
'LBRACE', 'RBRACE',
'LESS', 'GREATER',
'COMMA', 'SEMI', 'COLON', 'DBLCOLON',
'ASTERISK',
# C preprocessor directives
'CPPDIRECTIVE'
)
# Regular expressions for token matching
t_LPAREN = r'\('
t_RPAREN = r'\)'
# not used any more... commented out to suppress PLY warning
# t_LBRACKET = r'\['
# t_RBRACKET = r'\]'
t_LBRACE = r'\{'
t_RBRACE = r'\}'
t_LESS = r'\<'
t_GREATER = r'\>'
t_COMMA = r','
t_SEMI = 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.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.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.lineno += t.value.count('\n')
return t
def t_CPPDIRECTIVE(t):
r'^\#.*\n'
t.lineno += t.value.count('\n')
return t
#
# The functions t_NEWLINE, t_ignore, and t_error are
# special for the lex module.
#
# Newlines
def t_NEWLINE(t):
r'\n+'
t.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.lineno, "illegal character '%s'" % t.value[0])
t.skip(1)
# Build the 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]).
#####################################################################
# Not sure why, but we get a handful of shift/reduce conflicts on DECLARE.
# By default these get resolved as shifts, which is correct, but
# warnings are printed. Explicitly marking DECLARE as right-associative
# suppresses the warnings.
precedence = (
('right', 'DECLARE'),
)
# 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_declares name_decl opt_defs_and_declares decode_block'
global_decls1 = t[1]
isa_name = t[2]
namespace = isa_name + "Inst"
global_decls2 = t[3]
(inst_decls, decode_code, exec_code) = t[4]
decode_code = indent(decode_code)
# grab the last three path components of isa_desc_filename
filename = '/'.join(isa_desc_filename.split('/')[-3:])
# if the isa_desc file defines a 'rcs_id' string,
# echo that into the output too
try:
local_rcs_id = rcs_id
# strip $s out of ID so it doesn't get re-substituted
local_rcs_id = re.sub(r'\$', '', local_rcs_id)
except NameError:
local_rcs_id = 'Id: no RCS id found'
output = open(decoder_filename, 'w')
# split string to keep rcs from substituting this file's RCS id in
print >> output, '/* $Id' + '''$ */
/*
* Copyright (c) 2003
* The Regents of The University of Michigan
* All Rights Reserved
*
* This code is part of the M5 simulator, developed by Nathan Binkert,
* Erik Hallnor, Steve Raasch, and Steve Reinhardt, with contributions
* from Ron Dreslinski, Dave Greene, and Lisa Hsu.
*
* Permission is granted to use, copy, create derivative works and
* redistribute this software and such derivative works for any
* purpose, so long as the copyright notice above, this grant of
* permission, and the disclaimer below appear in all copies made; and
* so long as the name of The University of Michigan is not used in
* any advertising or publicity pertaining to the use or distribution
* of this software without specific, written prior authorization.
*
* THIS SOFTWARE IS PROVIDED AS IS, WITHOUT REPRESENTATION FROM THE
* UNIVERSITY OF MICHIGAN AS TO ITS FITNESS FOR ANY PURPOSE, AND
* WITHOUT WARRANTY BY THE UNIVERSITY OF MICHIGAN OF ANY KIND, EITHER
* EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE. THE REGENTS OF THE UNIVERSITY OF MICHIGAN SHALL NOT BE
* LIABLE FOR ANY DAMAGES, INCLUDING DIRECT, SPECIAL, INDIRECT,
* INCIDENTAL, OR CONSEQUENTIAL DAMAGES, WITH RESPECT TO ANY CLAIM
* ARISING OUT OF OR IN CONNECTION WITH THE USE OF THE SOFTWARE, EVEN
* IF IT HAS BEEN OR IS HEREAFTER ADVISED OF THE POSSIBILITY OF SUCH
* DAMAGES.
*/
/*
* DO NOT EDIT THIS FILE!!!
*
* It was automatically generated from this ISA description:
* Filename: %(filename)s
* RCS %(local_rcs_id)s
*/
#include "base/bitfield.hh" // required for bitfield support
/////////////////////////////////////
// Global defs (outside namespace) //
/////////////////////////////////////
%(global_decls1)s
/**
* Namespace for %(isa_name)s static instruction objects.
*/
namespace %(namespace)s
{
/////////////////////////////////////
// Global defs (within namespace) //
/////////////////////////////////////
%(global_decls2)s
////////////////////////////////////
// Declares from inst definitions //
////////////////////////////////////
%(inst_decls)s
%(exec_code)s
} // namespace %(namespace)s
//////////////////////
// Decoder function //
//////////////////////
StaticInstPtr<%(isa_name)s>
%(isa_name)s::decodeInst(%(isa_name)s::MachInst machInst)
{
using namespace %(namespace)s;
%(decode_code)s
} // decodeInst
''' % vars()
output.close()
# ISA name declaration looks like "namespace <foo>;"
def p_name_decl(t):
'name_decl : NAMESPACE ID SEMI'
t[0] = t[2]
# 'opt_defs_and_declares' is a possibly empty sequence of
# defs and/or declares.
def p_opt_defs_and_declares_0(t):
'opt_defs_and_declares : empty'
t[0] = ''
def p_opt_defs_and_declares_1(t):
'opt_defs_and_declares : defs_and_declares'
t[0] = t[1]
def p_defs_and_declares_0(t):
'defs_and_declares : def_or_declare'
t[0] = t[1]
def p_defs_and_declares_1(t):
'defs_and_declares : defs_and_declares def_or_declare'
t[0] = t[1] + t[2]
# The list of possible definition/declaration statements.
def p_def_or_declare(t):
'''def_or_declare : def_format
| def_bitfield
| def_template
| global_declare
| global_let
| cpp_directive'''
t[0] = t[1]
# preprocessor directives are copied directly to the output.
def p_cpp_directive(t):
'''cpp_directive : CPPDIRECTIVE'''
t[0] = t[1]
# Global declares 'declare {{...}}' (C++ code blocks) are copied
# directly to the output.
def p_global_declare(t):
'global_declare : DECLARE CODELIT SEMI'
t[0] = substBitOps(t[2])
# global let blocks 'let {{...}}' (Python code blocks) are executed
# directly when seen. These are typically used to initialize global
# Python variables used in later format definitions.
def p_global_let(t):
'global_let : LET CODELIT SEMI'
try:
exec(fixPythonIndentation(t[2]))
except:
error_bt(t.lineno(1), 'error in global let block "%s".' % t[2])
t[0] = '' # 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)
t[0] = '#undef %s\n#define %s\t%s\n' % (t[4], t[4], expr)
# 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)
t[0] = '#undef %s\n#define %s\t%s\n' % (t[4], t[4], expr)
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]] = t[4]
t[0] = ''
# 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.lineno(1))
# insert a comment into the output to note that the def was processed
t[0] = '''
//
// parser: format %s defined
//
''' % id
# The formal parameter list for an instruction format is a possibly
# empty list of comma-separated parameters.
def p_param_list_0(t):
'param_list : empty'
t[0] = [ ]
def p_param_list_1(t):
'param_list : param'
t[0] = [t[1]]
def p_param_list_2(t):
'param_list : param_list COMMA param'
t[0] = t[1]
t[0].append(t[3])
# Each formal parameter is either an identifier or an identifier
# preceded by an asterisk. As in Python, the latter (if present) gets
# a tuple containing all the excess positional arguments, allowing
# varargs functions.
def p_param_0(t):
'param : ID'
t[0] = t[1]
def p_param_1(t):
'param : ASTERISK ID'
# just concatenate them: '*ID'
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()
(decls, decode_code, exec_code, has_default) = t[5]
# use the "default defaults" only if there was no explicit
# default statement in decode_stmt_list
if not has_default:
(default_decls, default_decode, default_exec) = default_defaults
decls += default_decls
decode_code += default_decode
exec_code += default_exec
t[0] = (decls, '''
switch (%s) {
%s
}
''' % (t[2], indent(decode_code)), exec_code)
# 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
(decls, decode_code, exec_code) = t[2]
defaultStack.push((decls, '\ndefault:\n%sbreak;' % decode_code, exec_code))
# 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'
(decls1, decode_code1, exec_code1, has_default1) = t[1]
(decls2, decode_code2, exec_code2, has_default2) = t[2]
if (has_default1 and has_default2):
error(t.lineno(1), 'Two default cases in decode block')
t[0] = (decls1 + '\n' + decls2, decode_code1 + '\n' + decode_code2,
exec_code1 + '\n' + exec_code2, has_default1 or has_default2)
#
# 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 both the declaration and decode
# 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] = (t[1], t[1], t[1], 0)
# 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.lineno(1), '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, is_default) = t[1]
(decls, decode_code, exec_code) = t[3]
# just wrap the decoding code from the block as a case in the
# outer switch statement.
t[0] = (decls, '\n%s:\n%s' % (label, indent(decode_code)),
exec_code, is_default)
# Instruction definition (finally!).
def p_decode_stmt_inst(t):
'decode_stmt : case_label COLON inst SEMI'
(label, is_default) = t[1]
(decls, decode_code, exec_code) = t[3]
t[0] = (decls, '\n%s:%sbreak;' % (label, indent(decode_code)),
exec_code, is_default)
# 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])), 0)
def p_case_label_1(t):
'case_label : DEFAULT'
t[0] = ('default', 1)
#
# 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()
(decls, decode_code, exec_code) = \
currentFormat.defineInst(t[1], t[3], t.lineno(1))
args = ','.join(map(str, t[3]))
args = re.sub('(?m)^', '//', args)
args = re.sub('^//', '', args)
comment = '// %s::%s(%s)\n' % (currentFormat.id, t[1], args)
t[0] = (comment + decls, comment + decode_code, comment + exec_code)
# 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.lineno(1), 'instruction format "%s" not defined.' % t[1])
(decls, decode_code, exec_code) = \
format.defineInst(t[3], t[5], t.lineno(1))
comment = '// %s::%s(%s)\n' % (t[1], t[3], t[5])
t[0] = (comment + decls, comment + decode_code, comment + exec_code)
def p_arg_list_0(t):
'arg_list : empty'
t[0] = [ ]
def p_arg_list_1(t):
'arg_list : arg'
t[0] = [t[1]]
def p_arg_list_2(t):
'arg_list : arg_list COMMA arg'
t[0] = t[1]
t[0].append(t[3])
def p_arg(t):
'''arg : ID
| INTLIT
| STRLIT
| CODELIT'''
t[0] = t[1]
#
# 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.lineno, "syntax error at '%s'" % t.value)
else:
error_bt(0, "unknown syntax error")
# END OF GRAMMAR RULES
#
# Now build the parser.
yacc.yacc()
################
# Format object.
#
# A format object encapsulates an instruction format. It must provide
# a defineInst() method that generates the code for an instruction
# definition.
class Format:
def __init__(self, id, params, code):
# constructor: just save away arguments
self.id = id
self.params = params
# strip blank lines from code (ones at the end are troublesome)
code = re.sub(r'(?m)^\s*$', '', code);
if code == '':
code = ' pass\n'
param_list = string.join(params, ", ")
f = 'def defInst(name, Name, ' + param_list + '):\n' + code
c = compile(f, 'def format ' + id, 'exec')
exec(c)
self.func = defInst
def defineInst(self, name, args, lineno):
# automatically provide a capitalized version of mnemonic
Name = string.capitalize(name)
try:
retval = self.func(name, Name, *args)
except:
error_bt(lineno, 'error defining "%s".' % name)
return retval
# 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).
class Stack:
def __init__(self, initItem):
self.stack = [ initItem ]
def push(self, item):
self.stack.append(item);
def pop(self):
return self.stack.pop()
def top(self):
return self.stack[-1]
# The global format stack.
formatStack = Stack(NoFormat())
# The global default case stack.
defaultStack = Stack( None )
###################
# 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.
def error(lineno, string):
sys.exit("%s:%d: %s" % (isa_desc_filename, lineno, string))
# Like error(), but include a Python stack backtrace (for processing
# Python exceptions).
def error_bt(lineno, string):
traceback.print_exc()
print >> sys.stderr, "%s:%d: %s" % (isa_desc_filename, lineno, string)
sys.exit(1)
#####################################################################
#
# 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
#####################################################################
#
# Code Parser
#
# The remaining code is the support for automatically extracting
# instruction characteristics from pseudocode.
#
#####################################################################
# Force the argument to be a list
def makeList(list_or_item):
if not list_or_item:
return []
elif type(list_or_item) == ListType:
return list_or_item
else:
return [ list_or_item ]
# generate operandSizeMap based on provided operandTypeMap:
# basically generate equiv. C++ type and make is_signed flag
def buildOperandSizeMap():
global operandSizeMap
operandSizeMap = {}
for ext in operandTypeMap.keys():
(desc, size) = operandTypeMap[ext]
if desc == 'signed int':
type = 'int%d_t' % size
is_signed = 1
elif desc == 'unsigned int':
type = 'uint%d_t' % size
is_signed = 0
elif desc == 'float':
is_signed = 1 # shouldn't really matter
if size == 32:
type = 'float'
elif size == 64:
type = 'double'
if type == '':
error(0, 'Unrecognized type description "%s" in operandTypeMap')
operandSizeMap[ext] = (size, type, is_signed)
#
# Base class for operand traits. An instance of this class (or actually
# a class derived from this one) encapsulates the traits of a particular
# operand type (e.g., "32-bit integer register").
#
class OperandTraits:
def __init__(self, dflt_ext, reg_spec, flags, sort_pri):
# Force construction of operandSizeMap from operandTypeMap
# if it hasn't happened yet
if not globals().has_key('operandSizeMap'):
buildOperandSizeMap()
self.dflt_ext = dflt_ext
(self.dflt_size, self.dflt_type, self.dflt_is_signed) = \
operandSizeMap[dflt_ext]
self.reg_spec = reg_spec
# 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')
self.flags = ( [], [], [] )
elif type(flags) == StringType:
# a single flag: assumed to be unconditional
self.flags = ( [ flags ], [], [] )
elif type(flags) == ListType:
# a list of flags: also assumed to be unconditional
self.flags = ( flags, [], [] )
elif type(flags) == TupleType:
# 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
self.flags = (makeList(uncond_flags),
makeList(src_flags), makeList(dest_flags))
self.sort_pri = sort_pri
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 getFlags(self, op_desc):
# note the empty slice '[:]' gives us a copy of self.flags[0]
# instead of a reference to it
my_flags = self.flags[0][:]
if op_desc.is_src:
my_flags += self.flags[1]
if op_desc.is_dest:
my_flags += self.flags[2]
return my_flags
def makeDecl(self, op_desc):
(size, type, is_signed) = operandSizeMap[op_desc.eff_ext]
# Note that initializations in the declarations are solely
# to avoid 'uninitialized variable' errors from the compiler.
return type + ' ' + op_desc.munged_name + ' = 0;\n';
class IntRegOperandTraits(OperandTraits):
def isReg(self):
return 1
def isIntReg(self):
return 1
def makeConstructor(self, op_desc):
c = ''
if op_desc.is_src:
c += '\n\t_srcRegIdx[%d] = %s;' % \
(op_desc.src_reg_idx, self.reg_spec)
if op_desc.is_dest:
c += '\n\t_destRegIdx[%d] = %s;' % \
(op_desc.dest_reg_idx, self.reg_spec)
return c
def makeRead(self, op_desc):
(size, type, is_signed) = operandSizeMap[op_desc.eff_ext]
if (type == 'float' or type == 'double'):
error(0, 'Attempt to read integer register as FP')
if (size == self.dflt_size):
return '%s = xc->readIntReg(_srcRegIdx[%d]);\n' % \
(op_desc.munged_name, op_desc.src_reg_idx)
else:
return '%s = bits(xc->readIntReg(_srcRegIdx[%d]), %d, 0);\n' % \
(op_desc.munged_name, op_desc.src_reg_idx, size-1)
def makeWrite(self, op_desc):
(size, type, is_signed) = operandSizeMap[op_desc.eff_ext]
if (type == 'float' or type == 'double'):
error(0, 'Attempt to write integer register as FP')
if (size != self.dflt_size and is_signed):
final_val = 'sext<%d>(%s)' % (size, op_desc.munged_name)
else:
final_val = op_desc.munged_name
wb = '''
{
%s final_val = %s;
xc->setIntReg(_destRegIdx[%d], final_val);\n
if (traceData) { traceData->setData(final_val); }
}''' % (self.dflt_type, final_val, op_desc.dest_reg_idx)
return wb
class FloatRegOperandTraits(OperandTraits):
def isReg(self):
return 1
def isFloatReg(self):
return 1
def makeConstructor(self, op_desc):
c = ''
if op_desc.is_src:
c += '\n\t_srcRegIdx[%d] = %s + FP_Base_DepTag;' % \
(op_desc.src_reg_idx, self.reg_spec)
if op_desc.is_dest:
c += '\n\t_destRegIdx[%d] = %s + FP_Base_DepTag;' % \
(op_desc.dest_reg_idx, self.reg_spec)
return c
def makeRead(self, op_desc):
(size, type, is_signed) = operandSizeMap[op_desc.eff_ext]
bit_select = 0
if (type == 'float'):
func = 'readFloatRegSingle'
elif (type == 'double'):
func = 'readFloatRegDouble'
else:
func = 'readFloatRegInt'
if (size != self.dflt_size):
bit_select = 1
base = 'xc->%s(_srcRegIdx[%d] - FP_Base_DepTag)' % \
(func, op_desc.src_reg_idx)
if bit_select:
return '%s = bits(%s, %d, 0);\n' % \
(op_desc.munged_name, base, size-1)
else:
return '%s = %s;\n' % (op_desc.munged_name, base)
def makeWrite(self, op_desc):
(size, type, is_signed) = operandSizeMap[op_desc.eff_ext]
final_val = op_desc.munged_name
if (type == 'float'):
func = 'setFloatRegSingle'
elif (type == 'double'):
func = 'setFloatRegDouble'
else:
func = 'setFloatRegInt'
type = 'uint%d_t' % self.dflt_size
if (size != self.dflt_size and is_signed):
final_val = 'sext<%d>(%s)' % (size, op_desc.munged_name)
wb = '''
{
%s final_val = %s;
xc->%s(_destRegIdx[%d] - FP_Base_DepTag, final_val);\n
if (traceData) { traceData->setData(final_val); }
}''' % (type, final_val, func, op_desc.dest_reg_idx)
return wb
class ControlRegOperandTraits(OperandTraits):
def isReg(self):
return 1
def isControlReg(self):
return 1
def makeConstructor(self, op_desc):
c = ''
if op_desc.is_src:
c += '\n\t_srcRegIdx[%d] = %s_DepTag;' % \
(op_desc.src_reg_idx, self.reg_spec)
if op_desc.is_dest:
c += '\n\t_destRegIdx[%d] = %s_DepTag;' % \
(op_desc.dest_reg_idx, self.reg_spec)
return c
def makeRead(self, op_desc):
(size, type, is_signed) = operandSizeMap[op_desc.eff_ext]
bit_select = 0
if (type == 'float' or type == 'double'):
error(0, 'Attempt to read control register as FP')
base = 'xc->read%s()' % self.reg_spec
if size == self.dflt_size:
return '%s = %s;\n' % (op_desc.munged_name, base)
else:
return '%s = bits(%s, %d, 0);\n' % \
(op_desc.munged_name, base, size-1)
def makeWrite(self, op_desc):
(size, type, is_signed) = operandSizeMap[op_desc.eff_ext]
if (type == 'float' or type == 'double'):
error(0, 'Attempt to write control register as FP')
wb = 'xc->set%s(%s);\n' % (self.reg_spec, op_desc.munged_name)
wb += 'if (traceData) { traceData->setData(%s); }' % \
op_desc.munged_name
return wb
class MemOperandTraits(OperandTraits):
def isMem(self):
return 1
def makeConstructor(self, op_desc):
return ''
def makeDecl(self, op_desc):
(size, type, is_signed) = operandSizeMap[op_desc.eff_ext]
# Note that initializations in the declarations are solely
# to avoid 'uninitialized variable' errors from the compiler.
# Declare memory data variable.
c = '%s %s = 0;\n' % (type, op_desc.munged_name)
# Declare var to hold memory access flags.
c += 'unsigned %s_flags = memAccessFlags;\n' % op_desc.base_name
# If this operand is a dest (i.e., it's a store operation),
# then we need to declare a variable for the write result code
# as well.
if op_desc.is_dest:
c += 'uint64_t %s_write_result = 0;\n' % op_desc.base_name
return c
def makeRead(self, op_desc):
(size, type, is_signed) = operandSizeMap[op_desc.eff_ext]
eff_type = 'uint%d_t' % size
return 'fault = xc->read(EA, (%s&)%s, %s_flags);\n' \
% (eff_type, op_desc.munged_name, op_desc.base_name)
def makeWrite(self, op_desc):
(size, type, is_signed) = operandSizeMap[op_desc.eff_ext]
eff_type = 'uint%d_t' % size
return 'fault = xc->write((%s&)%s, EA, %s_flags,' \
' &%s_write_result);\n' \
% (eff_type, op_desc.munged_name, op_desc.base_name,
op_desc.base_name)
class NPCOperandTraits(OperandTraits):
def makeConstructor(self, op_desc):
return ''
def makeRead(self, op_desc):
return '%s = xc->readPC() + 4;\n' % op_desc.munged_name
def makeWrite(self, op_desc):
return 'xc->setNextPC(%s);\n' % op_desc.munged_name
#
# Define operand variables that get derived from the basic declaration
# of ISA-specific operands in operandTraitsMap. This function must be
# called by the ISA description file explicitly after defining
# operandTraitsMap (in a 'let' block).
#
def defineDerivedOperandVars():
global operands
operands = operandTraitsMap.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)
#
# Operand descriptor class. An instance of this class represents
# a specific operand for a code block.
#
class OperandDescriptor:
def __init__(self, full_name, base_name, ext, is_src, is_dest):
self.full_name = full_name
self.base_name = base_name
self.ext = ext
self.is_src = is_src
self.is_dest = is_dest
self.traits = operandTraitsMap[base_name]
# The 'effective extension' (eff_ext) is either the actual
# extension, if one was explicitly provided, or the default.
# The 'munged name' replaces the '.' between the base and
# extension (if any) with a '_' to make a legal C++ variable name.
if ext:
self.eff_ext = ext
self.munged_name = base_name + '_' + ext
else:
self.eff_ext = self.traits.dflt_ext
self.munged_name = base_name
# 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.traits.getFlags(self)
self.constructor = self.traits.makeConstructor(self)
self.op_decl = self.traits.makeDecl(self)
if self.is_src:
self.op_rd = self.traits.makeRead(self)
else:
self.op_rd = ''
if self.is_dest:
self.op_wb = self.traits.makeWrite(self)
else:
self.op_wb = ''
class OperandDescriptorList:
def __init__(self):
self.items = []
self.bases = {}
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.traits.sort_pri - b.traits.sort_pri)
# 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)
#
# Find all the operands in the given code block. Returns an operand
# descriptor list (instance of class OperandDescriptorList).
#
def findOperands(code):
operands = OperandDescriptorList()
# delete comments so we don't accidentally 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 = operands.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 = OperandDescriptor(op_full, op_base, op_ext,
is_src, is_dest)
operands.append(op_desc)
# start next search after end of current match
next_pos = match.end()
operands.sort()
# enumerate source & dest register operands... used in building
# constructor later
srcRegs = 0
destRegs = 0
operands.numFPDestRegs = 0
operands.numIntDestRegs = 0
for op_desc in operands:
if op_desc.traits.isReg():
if op_desc.is_src:
op_desc.src_reg_idx = srcRegs
srcRegs += 1
if op_desc.is_dest:
op_desc.dest_reg_idx = destRegs
destRegs += 1
if op_desc.traits.isFloatReg():
operands.numFPDestRegs += 1
elif op_desc.traits.isIntReg():
operands.numIntDestRegs += 1
operands.numSrcRegs = srcRegs
operands.numDestRegs = destRegs
# now make a final pass to finalize op_desc fields that may depend
# on the register enumeration
for op_desc in operands:
op_desc.finalize()
return operands
# Munge operand names in code string to make legal C++ variable names.
# (Will match munged_name attribute of OperandDescriptor object.)
def substMungedOpNames(code):
return operandsWithExtRE.sub(r'\1_\2', code)
def joinLists(t):
return map(string.join, t)
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
class CodeBlock:
def __init__(self, code):
self.orig_code = code
self.operands = findOperands(code)
self.code = substMungedOpNames(substBitOps(code))
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.op_decl = self.operands.concatAttrStrings('op_decl')
is_mem = lambda op: op.traits.isMem()
not_mem = lambda op: not op.traits.isMem()
self.op_rd = self.operands.concatAttrStrings('op_rd')
self.op_wb = self.operands.concatAttrStrings('op_wb')
self.op_mem_rd = \
self.operands.concatSomeAttrStrings(is_mem, 'op_rd')
self.op_mem_wb = \
self.operands.concatSomeAttrStrings(is_mem, 'op_wb')
self.op_nonmem_rd = \
self.operands.concatSomeAttrStrings(not_mem, 'op_rd')
self.op_nonmem_wb = \
self.operands.concatSomeAttrStrings(not_mem, 'op_wb')
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 will be overridden
# later otherwise.
if 'IsStore' in self.flags:
self.op_class = 'WrPort'
elif 'IsLoad' in self.flags or 'IsPrefetch' in self.flags:
self.op_class = 'RdPort'
elif 'IsFloating' in self.flags:
self.op_class = 'FloatADD'
else:
self.op_class = 'IntALU'
# Assume all instruction flags are of the form 'IsFoo'
instFlagRE = re.compile(r'Is.*')
# OpClass constants are just a little more complicated
opClassRE = re.compile(r'Int.*|Float.*|.*Port|No_OpClass')
class InstObjParams:
def __init__(self, mnem, class_name, base_class = '',
code_block = None, opt_args = []):
self.mnemonic = mnem
self.class_name = class_name
self.base_class = base_class
self.exec_func_declarations = '''
Fault execute(SimpleCPUExecContext *, Trace::InstRecord *);
Fault execute(FullCPUExecContext *, Trace::InstRecord *);
'''
if code_block:
for code_attr in code_block.__dict__.keys():
setattr(self, code_attr, getattr(code_block, code_attr))
else:
self.constructor = ''
self.flags = []
# 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 = ''
def _subst(self, template):
try:
return template % self.__dict__
except KeyError, key:
raise KeyError, 'InstObjParams.subst: no definition for %s' % key
def subst(self, *args):
result = []
for t in args:
try: template = templateMap[t]
except KeyError:
error(0, 'InstObjParams::subst: undefined template "%s"' % t)
if template.find('%(cpu_model)') != -1:
tmp = ''
for cpu_model in ('SimpleCPUExecContext', 'FullCPUExecContext'):
self.cpu_model = cpu_model
tmp += self._subst(template)
result.append(tmp)
else:
result.append(self._subst(template))
if len(args) == 1:
result = result[0]
return result
#
# Read in and parse the ISA description.
#
def parse_isa_desc(isa_desc_file, decoder_file):
# Arguments are the name of the ISA description (input) file and
# the name of the C++ decoder (output) file.
global isa_desc_filename, decoder_filename
isa_desc_filename = isa_desc_file
decoder_filename = decoder_file
# Suck the ISA description file in.
input = open(isa_desc_filename)
isa_desc = input.read()
input.close()
# Parse it.
yacc.parse(isa_desc)
# Called as script: get args from command line.
if __name__ == '__main__':
parse_isa_desc(sys.argv[1], sys.argv[2])
|