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
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
|
// -*- mode:c++ -*-
//
// Alpha ISA description file.
//
let {{
global rcs_id
rcs_id = "$Id$"
}};
#include <sstream>
#include <iostream>
#include <iomanip>
#include <math.h>
#if defined(linux)
#include <fenv.h>
#endif
#include "base/cprintf.hh"
#include "base/misc.hh"
#include "cpu/exec_context.hh"
#include "cpu/exetrace.hh"
#include "cpu/full_cpu/dyn_inst.hh"
#include "cpu/simple_cpu/simple_cpu.hh"
#include "cpu/static_inst.hh"
#include "sim/annotation.hh"
#include "sim/sim_exit.hh"
#ifdef FULL_SYSTEM
#include "arch/alpha/ev5.hh"
#include "arch/alpha/pseudo_inst.hh"
#endif
namespace AlphaISA;
// Universal (format-independent) fields
def bitfield OPCODE <31:26>;
def bitfield RA <25:21>;
def bitfield RB <20:16>;
// Memory format
def signed bitfield MEMDISP <15: 0>; // displacement
def bitfield MEMFUNC <15: 0>; // function code (same field, unsigned)
// Memory-format jumps
def bitfield JMPFUNC <15:14>; // function code (disp<15:14>)
def bitfield JMPHINT <13: 0>; // tgt Icache idx hint (disp<13:0>)
// Branch format
def signed bitfield BRDISP <20: 0>; // displacement
// Integer operate format(s>;
def bitfield INTIMM <20:13>; // integer immediate (literal)
def bitfield IMM <12:12>; // immediate flag
def bitfield INTFUNC <11: 5>; // function code
def bitfield RC < 4: 0>; // dest reg
// Floating-point operate format
def bitfield FA <25:21>;
def bitfield FB <20:16>;
def bitfield FP_FULLFUNC <15: 5>; // complete function code
def bitfield FP_TRAPMODE <15:13>; // trapping mode
def bitfield FP_ROUNDMODE <12:11>; // rounding mode
def bitfield FP_TYPEFUNC <10: 5>; // type+func: handiest for decoding
def bitfield FP_SRCTYPE <10: 9>; // source reg type
def bitfield FP_SHORTFUNC < 8: 5>; // short function code
def bitfield FP_SHORTFUNC_TOP2 <8:7>; // top 2 bits of short func code
def bitfield FC < 4: 0>; // dest reg
// PALcode format
def bitfield PALFUNC <25: 0>; // function code
// EV5 PAL instructions:
// HW_LD/HW_ST
def bitfield HW_LDST_PHYS <15>; // address is physical
def bitfield HW_LDST_ALT <14>; // use ALT_MODE IPR
def bitfield HW_LDST_WRTCK <13>; // HW_LD only: fault if no write acc
def bitfield HW_LDST_QUAD <12>; // size: 0=32b, 1=64b
def bitfield HW_LDST_VPTE <11>; // HW_LD only: is PTE fetch
def bitfield HW_LDST_LOCK <10>; // HW_LD only: is load locked
def bitfield HW_LDST_COND <10>; // HW_ST only: is store conditional
def signed bitfield HW_LDST_DISP <9:0>; // signed displacement
// HW_REI
def bitfield HW_REI_TYP <15:14>; // type: stalling vs. non-stallingk
def bitfield HW_REI_MBZ <13: 0>; // must be zero
// HW_MTPR/MW_MFPR
def bitfield HW_IPR_IDX <15:0>; // IPR index
// M5 instructions
def bitfield M5FUNC <7:0>;
let {{
global operandTypeMap
operandTypeMap = {
'sb' : ('signed int', 8),
'ub' : ('unsigned int', 8),
'sw' : ('signed int', 16),
'uw' : ('unsigned int', 16),
'sl' : ('signed int', 32),
'ul' : ('unsigned int', 32),
'sq' : ('signed int', 64),
'uq' : ('unsigned int', 64),
'sf' : ('float', 32),
'df' : ('float', 64)
}
global operandTraitsMap
operandTraitsMap = {
# Int regs default to unsigned, but code should not count on this.
# For clarity, descriptions that depend on unsigned behavior should
# explicitly specify '.uq'.
'Ra': IntRegOperandTraits('uq', 'RA', 'IsInteger', 1),
'Rb': IntRegOperandTraits('uq', 'RB', 'IsInteger', 2),
'Rc': IntRegOperandTraits('uq', 'RC', 'IsInteger', 3),
'Fa': FloatRegOperandTraits('df', 'FA', 'IsFloating', 1),
'Fb': FloatRegOperandTraits('df', 'FB', 'IsFloating', 2),
'Fc': FloatRegOperandTraits('df', 'FC', 'IsFloating', 3),
'Mem': MemOperandTraits('uq', None,
('IsMemRef', 'IsLoad', 'IsStore'), 4),
'NPC': NPCOperandTraits('uq', None, ( None, None, 'IsControl' ), 4),
'Runiq': ControlRegOperandTraits('uq', 'Uniq', None, 1),
'FPCR': ControlRegOperandTraits('uq', 'Fpcr', None, 1),
# The next two are hacks for non-full-system call-pal emulation
'R0': IntRegOperandTraits('uq', '0', None, 1),
'R16': IntRegOperandTraits('uq', '16', None, 1),
}
defineDerivedOperandVars()
}};
declare {{
// just temporary, while comparing with old code for debugging
// #define SS_COMPATIBLE_DISASSEMBLY
/// Check "FP enabled" machine status bit. Called when executing any FP
/// instruction in full-system mode.
/// @retval Full-system mode: No_Fault if FP is enabled, Fen_Fault
/// if not. Non-full-system mode: always returns No_Fault.
#ifdef FULL_SYSTEM
template <class XC>
inline Fault checkFpEnableFault(XC *xc)
{
Fault fault = No_Fault; // dummy... this ipr access should not fault
if (!ICSR_FPE(xc->readIpr(AlphaISA::IPR_ICSR, fault))) {
fault = Fen_Fault;
}
return fault;
}
#else
template <class XC>
inline Fault checkFpEnableFault(XC *xc)
{
return No_Fault;
}
#endif
/**
* Base class for all Alpha static instructions.
*/
class AlphaStaticInst : public StaticInst<AlphaISA>
{
protected:
/// Make AlphaISA register dependence tags directly visible in
/// this class and derived classes. Maybe these should really
/// live here and not in the AlphaISA namespace.
enum DependenceTags {
FP_Base_DepTag = AlphaISA::FP_Base_DepTag,
Fpcr_DepTag = AlphaISA::Fpcr_DepTag,
Uniq_DepTag = AlphaISA::Uniq_DepTag,
IPR_Base_DepTag = AlphaISA::IPR_Base_DepTag
};
/// Constructor.
AlphaStaticInst(const char *mnem, MachInst _machInst,
OpClass __opClass)
: StaticInst<AlphaISA>(mnem, _machInst, __opClass)
{
}
/// Print a register name for disassembly given the unique
/// dependence tag number (FP or int).
void printReg(std::ostream &os, int reg)
{
if (reg < FP_Base_DepTag) {
ccprintf(os, "r%d", reg);
}
else {
ccprintf(os, "f%d", reg - FP_Base_DepTag);
}
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
std::stringstream ss;
ccprintf(ss, "%-10s ", mnemonic);
// just print the first two source regs... if there's
// a third one, it's a read-modify-write dest (Rc),
// e.g. for CMOVxx
if (_numSrcRegs > 0) {
printReg(ss, _srcRegIdx[0]);
}
if (_numSrcRegs > 1) {
ss << ",";
printReg(ss, _srcRegIdx[1]);
}
// just print the first dest... if there's a second one,
// it's generally implicit
if (_numDestRegs > 0) {
if (_numSrcRegs > 0)
ss << ",";
printReg(ss, _destRegIdx[0]);
}
return ss.str();
}
};
}};
def template BasicDeclare {{
/**
* Static instruction class for "%(mnemonic)s".
*/
class %(class_name)s : public %(base_class)s
{
public:
/// Constructor.
%(class_name)s(MachInst machInst)
: %(base_class)s("%(mnemonic)s", machInst, %(op_class)s)
{
%(constructor)s;
}
%(exec_func_declarations)s
};
}};
def template BasicExecute {{
Fault %(class_name)s::execute(%(cpu_model)s *xc,
Trace::InstRecord *traceData)
{
Fault fault = No_Fault;
%(fp_enable_check)s;
%(op_decl)s;
%(op_rd)s;
%(code)s;
if (fault == No_Fault) {
%(op_wb)s;
}
return fault;
}
}};
def template BasicDecode {{
return new %(class_name)s(machInst);
}};
def template BasicDecodeWithMnemonic {{
return new %(class_name)s("%(mnemonic)s", machInst);
}};
// The most basic instruction format... used only for a few misc. insts
def format BasicOperate(code, *flags) {{
iop = InstObjParams(name, Name, 'AlphaStaticInst', CodeBlock(code), flags)
return iop.subst('BasicDeclare', 'BasicDecode', 'BasicExecute')
}};
////////////////////////////////////////////////////////////////////
declare {{
/**
* Static instruction class for no-ops. This is a leaf class.
*/
class Nop : public AlphaStaticInst
{
/// Disassembly of original instruction.
const std::string originalDisassembly;
public:
/// Constructor
Nop(const std::string _originalDisassembly, MachInst _machInst)
: AlphaStaticInst("nop", _machInst, No_OpClass),
originalDisassembly(_originalDisassembly)
{
flags[IsNop] = true;
}
~Nop() { }
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
#ifdef SS_COMPATIBLE_DISASSEMBLY
return originalDisassembly;
#else
return csprintf("%-10s (%s)", "nop", originalDisassembly);
#endif
}
Fault execute(SimpleCPUExecContext *, Trace::InstRecord *)
{ return No_Fault; }
Fault execute(FullCPUExecContext *, Trace::InstRecord *)
{ return No_Fault; }
};
/// Helper function for decoding nops. Substitute Nop object
/// for original inst passed in as arg (and delete latter).
inline
AlphaStaticInst *
makeNop(AlphaStaticInst *inst)
{
AlphaStaticInst *nop = new Nop(inst->disassemble(0), inst->machInst);
delete inst;
return nop;
}
}};
def format Nop() {{
return ('', 'return new Nop("%s", machInst);\n' % name, 'return No_Fault;')
}};
// integer & FP operate instructions use Rc as dest, so check for
// Rc == 31 to detect nops
def template OperateNopCheckDecode {{
{
AlphaStaticInst *i = new %(class_name)s(machInst);
if (RC == 31) {
i = makeNop(i);
}
return i;
}
}};
// Like BasicOperate format, but generates NOP if RC/FC == 31
def format BasicOperateWithNopCheck(code, *opt_args) {{
iop = InstObjParams(name, Name, 'AlphaStaticInst', CodeBlock(code),
opt_args)
return iop.subst('BasicDeclare', 'OperateNopCheckDecode', 'BasicExecute')
}};
////////////////////////////////////////////////////////////////////
//
// Integer operate instructions
//
declare {{
/**
* Base class for integer immediate instructions.
*/
class IntegerImm : public AlphaStaticInst
{
protected:
/// Immediate operand value (unsigned 8-bit int).
uint8_t imm;
/// Constructor
IntegerImm(const char *mnem, MachInst _machInst, OpClass __opClass)
: AlphaStaticInst(mnem, _machInst, __opClass), imm(INTIMM)
{
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
std::stringstream ss;
ccprintf(ss, "%-10s ", mnemonic);
// just print the first source reg... if there's
// a second one, it's a read-modify-write dest (Rc),
// e.g. for CMOVxx
if (_numSrcRegs > 0) {
printReg(ss, _srcRegIdx[0]);
ss << ",";
}
ss << (int)imm;
if (_numDestRegs > 0) {
ss << ",";
printReg(ss, _destRegIdx[0]);
}
return ss.str();
}
};
}};
def template RegOrImmDecode {{
{
AlphaStaticInst *i =
(IMM) ? (AlphaStaticInst *)new %(class_name)sImm(machInst)
: (AlphaStaticInst *)new %(class_name)s(machInst);
if (RC == 31) {
i = makeNop(i);
}
return i;
}
}};
// Primary format for integer operate instructions:
// - Generates both reg-reg and reg-imm versions if Rb_or_imm is used.
// - Generates NOP if RC == 31.
def format IntegerOperate(code, *opt_flags) {{
# If the code block contains 'Rb_or_imm', we define two instructions,
# one using 'Rb' and one using 'imm', and have the decoder select
# the right one.
uses_imm = (code.find('Rb_or_imm') != -1)
if uses_imm:
orig_code = code
# base code is reg version:
# rewrite by substituting 'Rb' for 'Rb_or_imm'
code = re.sub(r'Rb_or_imm', 'Rb', orig_code)
# generate immediate version by substituting 'imm'
# note that imm takes no extenstion, so we extend
# the regexp to replace any extension as well
imm_code = re.sub(r'Rb_or_imm(\.\w+)?', 'imm', orig_code)
# generate declaration for register version
cblk = CodeBlock(code)
iop = InstObjParams(name, Name, 'AlphaStaticInst', cblk, opt_flags)
(decls, exec_code) = iop.subst('BasicDeclare', 'BasicExecute')
if uses_imm:
# append declaration for imm version
imm_cblk = CodeBlock(imm_code)
imm_iop = InstObjParams(name, Name + 'Imm', 'IntegerImm', imm_cblk,
opt_flags)
(imm_decls, imm_exec_code) = \
imm_iop.subst('BasicDeclare', 'BasicExecute')
decls += imm_decls
exec_code += imm_exec_code
# decode checks IMM bit to pick correct version
decode = iop.subst('RegOrImmDecode')
else:
# no imm version: just check for nop
decode = iop.subst('OperateNopCheckDecode')
return (decls, decode, exec_code)
}};
////////////////////////////////////////////////////////////////////
//
// Floating-point instructions
//
// Note that many FP-type instructions which do not support all the
// various rounding & trapping modes use the simpler format
// BasicOperateWithNopCheck.
//
declare {{
/**
* Base class for general floating-point instructions. Includes
* support for various Alpha rounding and trapping modes. Only FP
* instructions that require this support are derived from this
* class; the rest derive directly from AlphaStaticInst.
*/
class AlphaFP : public AlphaStaticInst
{
public:
/// Alpha FP rounding modes.
enum RoundingMode {
Chopped = 0, ///< round toward zero
Minus_Infinity = 1, ///< round toward minus infinity
Normal = 2, ///< round to nearest (default)
Dynamic = 3, ///< use FPCR setting (in instruction)
Plus_Infinity = 3 ///< round to plus inifinity (in FPCR)
};
/// Alpha FP trapping modes.
/// For instructions that produce integer results, the
/// "Underflow Enable" modes really mean "Overflow Enable", and
/// the assembly modifier is V rather than U.
enum TrappingMode {
/// default: nothing enabled
Imprecise = 0, ///< no modifier
/// underflow/overflow traps enabled, inexact disabled
Underflow_Imprecise = 1, ///< /U or /V
Underflow_Precise = 5, ///< /SU or /SV
/// underflow/overflow and inexact traps enabled
Underflow_Inexact_Precise = 7 ///< /SUI or /SVI
};
protected:
#if defined(linux)
static const int alphaToC99RoundingMode[];
#endif
/// Map enum RoundingMode values to disassembly suffixes.
static const char *roundingModeSuffix[];
/// Map enum TrappingMode values to FP disassembly suffixes.
static const char *fpTrappingModeSuffix[];
/// Map enum TrappingMode values to integer disassembly suffixes.
static const char *intTrappingModeSuffix[];
/// This instruction's rounding mode.
RoundingMode roundingMode;
/// This instruction's trapping mode.
TrappingMode trappingMode;
/// Constructor
AlphaFP(const char *mnem, MachInst _machInst, OpClass __opClass)
: AlphaStaticInst(mnem, _machInst, __opClass),
roundingMode((enum RoundingMode)FP_ROUNDMODE),
trappingMode((enum TrappingMode)FP_TRAPMODE)
{
if (trappingMode != Imprecise) {
warn("precise FP traps unimplemented\n");
}
}
#if defined(linux)
int
getC99RoundingMode(uint64_t fpcr_val)
{
if (roundingMode == Dynamic) {
return alphaToC99RoundingMode[bits(fpcr_val, 59, 58)];
}
else {
return alphaToC99RoundingMode[roundingMode];
}
}
#endif
// This differs from the AlphaStaticInst version only in
// printing suffixes for non-default rounding & trapping modes.
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
std::string mnem_str(mnemonic);
#ifndef SS_COMPATIBLE_DISASSEMBLY
std::string suffix("");
suffix += ((_destRegIdx[0] >= FP_Base_DepTag)
? fpTrappingModeSuffix[trappingMode]
: intTrappingModeSuffix[trappingMode]);
suffix += roundingModeSuffix[roundingMode];
if (suffix != "") {
mnem_str = csprintf("%s/%s", mnemonic, suffix);
}
#endif
std::stringstream ss;
ccprintf(ss, "%-10s ", mnem_str.c_str());
// just print the first two source regs... if there's
// a third one, it's a read-modify-write dest (Rc),
// e.g. for CMOVxx
if (_numSrcRegs > 0) {
printReg(ss, _srcRegIdx[0]);
}
if (_numSrcRegs > 1) {
ss << ",";
printReg(ss, _srcRegIdx[1]);
}
// just print the first dest... if there's a second one,
// it's generally implicit
if (_numDestRegs > 0) {
if (_numSrcRegs > 0)
ss << ",";
printReg(ss, _destRegIdx[0]);
}
return ss.str();
}
};
#if defined(linux)
const int AlphaFP::alphaToC99RoundingMode[] = {
FE_TOWARDZERO, // Chopped
FE_DOWNWARD, // Minus_Infinity
FE_TONEAREST, // Normal
FE_UPWARD // Dynamic in inst, Plus_Infinity in FPCR
};
#endif
const char *AlphaFP::roundingModeSuffix[] = { "c", "m", "", "d" };
// mark invalid trapping modes, but don't fail on them, because
// you could decode anything on a misspeculated path
const char *AlphaFP::fpTrappingModeSuffix[] =
{ "", "u", "INVTM2", "INVTM3", "INVTM4", "su", "INVTM6", "sui" };
const char *AlphaFP::intTrappingModeSuffix[] =
{ "", "v", "INVTM2", "INVTM3", "INVTM4", "sv", "INVTM6", "svi" };
}};
def template FloatingPointDecode {{
{
bool fast = (FP_TRAPMODE == AlphaFP::Imprecise
&& FP_ROUNDMODE == AlphaFP::Normal);
AlphaStaticInst *i =
fast ? (AlphaStaticInst *)new %(class_name)sFast(machInst) :
(AlphaStaticInst *)new %(class_name)sGeneral(machInst);
if (FC == 31) {
i = makeNop(i);
}
return i;
}
}};
// General format for floating-point operate instructions:
// - Checks trapping and rounding mode flags. Trapping modes
// currently unimplemented (will fail).
// - Generates NOP if FC == 31.
def format FloatingPointOperate(code, *opt_args) {{
iop = InstObjParams(name, Name, 'AlphaFP', CodeBlock(code), opt_args)
decode = iop.subst('FloatingPointDecode')
fast_iop = InstObjParams(name, Name + 'Fast', 'AlphaFP',
CodeBlock(code), opt_args)
(fast_declare, fast_exec) = fast_iop.subst('BasicDeclare', 'BasicExecute')
gen_code_prefix = r'''
#if defined(linux)
fesetround(getC99RoundingMode(xc->readFpcr()));
#endif
'''
gen_code_suffix = r'''
#if defined(linux)
fesetround(FE_TONEAREST);
#endif
'''
gen_iop = InstObjParams(name, Name + 'General', 'AlphaFP',
CodeBlock(gen_code_prefix + code + gen_code_suffix), opt_args)
(gen_declare, gen_exec) = gen_iop.subst('BasicDeclare', 'BasicExecute')
return (fast_declare + gen_declare, decode, fast_exec + gen_exec)
}};
////////////////////////////////////////////////////////////////////
//
// Memory-format instructions: LoadAddress, Load, Store
//
declare {{
/**
* Base class for general Alpha memory-format instructions.
*/
class Memory : public AlphaStaticInst
{
protected:
/// Displacement for EA calculation (signed).
int32_t disp;
/// Memory request flags. See mem_req_base.hh.
unsigned memAccessFlags;
/// Constructor
Memory(const char *mnem, MachInst _machInst, OpClass __opClass)
: AlphaStaticInst(mnem, _machInst, __opClass),
disp(MEMDISP), memAccessFlags(0)
{
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
return csprintf("%-10s %c%d,%d(r%d)", mnemonic,
flags[IsFloating] ? 'f' : 'r', RA, MEMDISP, RB);
}
};
/**
* Base class for a few miscellaneous memory-format insts
* that don't interpret the disp field: wh64, fetch, fetch_m, ecb.
* None of these instructions has a destination register either.
*/
class MemoryNoDisp : public AlphaStaticInst
{
protected:
/// Memory request flags. See mem_req_base.hh.
unsigned memAccessFlags;
/// Constructor
MemoryNoDisp(const char *mnem, MachInst _machInst, OpClass __opClass)
: AlphaStaticInst(mnem, _machInst, __opClass),
memAccessFlags(0)
{
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
return csprintf("%-10s (r%d)", mnemonic, RB);
}
};
/**
* Base class for "fake" effective-address computation
* instructions returnded by eaCompInst().
*/
class EACompBase : public AlphaStaticInst
{
public:
/// Constructor
EACompBase(MachInst machInst)
: AlphaStaticInst("(eacomp)", machInst, IntALU)
{
}
Fault execute(SimpleCPUExecContext *, Trace::InstRecord *)
{ panic("attempt to execute eacomp"); }
Fault execute(FullCPUExecContext *, Trace::InstRecord *)
{ panic("attempt to execute eacomp"); }
};
/**
* Base class for "fake" memory-access instructions returnded by
* memAccInst().
*/
class MemAccBase : public AlphaStaticInst
{
public:
/// Constructor
MemAccBase(MachInst machInst, OpClass __opClass)
: AlphaStaticInst("(memacc)", machInst, __opClass)
{
}
Fault execute(SimpleCPUExecContext *, Trace::InstRecord *)
{ panic("attempt to execute memacc"); }
Fault execute(FullCPUExecContext *, Trace::InstRecord *)
{ panic("attempt to execute memacc"); }
};
}};
def format LoadAddress(code) {{
iop = InstObjParams(name, Name, 'Memory', CodeBlock(code))
return iop.subst('BasicDeclare', 'BasicDecode', 'BasicExecute')
}};
def template LoadStoreDeclare {{
/**
* Static instruction class for "%(mnemonic)s".
*/
class %(class_name)s : public %(base_class)s
{
protected:
/**
* "Fake" effective address computation class for "%(mnemonic)s".
*/
class EAComp : public EACompBase
{
public:
/// Constructor
EAComp(MachInst machInst)
: EACompBase(machInst)
{
%(ea_constructor)s;
}
};
/**
* "Fake" memory access instruction class for "%(mnemonic)s".
*/
class MemAcc : public MemAccBase
{
public:
/// Constructor
MemAcc(MachInst machInst)
: MemAccBase(machInst, %(op_class)s)
{
%(memacc_constructor)s;
}
};
/// Pointer to EAComp object.
StaticInstPtr<AlphaISA> eaCompPtr;
/// Pointer to MemAcc object.
StaticInstPtr<AlphaISA> memAccPtr;
public:
StaticInstPtr<AlphaISA> eaCompInst() { return eaCompPtr; }
StaticInstPtr<AlphaISA> memAccInst() { return memAccPtr; }
/// Constructor.
%(class_name)s(MachInst machInst)
: %(base_class)s("%(mnemonic)s", machInst, %(op_class)s),
eaCompPtr(new EAComp(machInst)), memAccPtr(new MemAcc(machInst))
{
%(constructor)s;
}
%(exec_func_declarations)s
};
}};
def template LoadStoreExecute {{
Fault %(class_name)s::execute(%(cpu_model)s *xc,
Trace::InstRecord *traceData)
{
Addr EA;
Fault fault = No_Fault;
%(fp_enable_check)s;
%(op_decl)s;
%(op_nonmem_rd)s;
%(ea_code)s;
if (fault == No_Fault) {
%(op_mem_rd)s;
%(memacc_code)s;
}
if (fault == No_Fault) {
%(op_mem_wb)s;
}
if (fault == No_Fault) {
%(postacc_code)s;
}
if (fault == No_Fault) {
%(op_nonmem_wb)s;
}
return fault;
}
}};
def template PrefetchDeclare {{
/**
* Static instruction class for "%(mnemonic)s".
*/
class %(class_name)s : public %(base_class)s
{
protected:
/**
* "Fake" effective address computation class for "%(mnemonic)s".
*/
class EAComp : public EACompBase
{
public:
/// Constructor
EAComp(MachInst machInst)
: EACompBase(machInst)
{
%(ea_constructor)s;
}
};
/**
* "Fake" memory access instruction class for "%(mnemonic)s".
*/
class MemAcc : public MemAccBase
{
public:
/// Constructor
MemAcc(MachInst machInst)
: MemAccBase(machInst, %(op_class)s)
{
%(memacc_constructor)s;
}
};
/// Pointer to EAComp object.
StaticInstPtr<AlphaISA> eaCompPtr;
/// Pointer to MemAcc object.
StaticInstPtr<AlphaISA> memAccPtr;
public:
StaticInstPtr<AlphaISA> eaCompInst() { return eaCompPtr; }
StaticInstPtr<AlphaISA> memAccInst() { return memAccPtr; }
/// Constructor
%(class_name)s(MachInst machInst)
: %(base_class)s("%(mnemonic)s", machInst, %(op_class)s),
eaCompPtr(new EAComp(machInst)), memAccPtr(new MemAcc(machInst))
{
%(constructor)s;
}
%(exec_func_declarations)s
};
}};
def template PrefetchExecute {{
Fault %(class_name)s::execute(%(cpu_model)s *xc,
Trace::InstRecord *traceData)
{
Addr EA;
Fault fault = No_Fault;
%(fp_enable_check)s;
%(op_decl)s;
%(op_nonmem_rd)s;
%(ea_code)s;
if (fault == No_Fault) {
xc->prefetch(EA, memAccessFlags);
}
return No_Fault;
}
}};
// load instructions use Ra as dest, so check for
// Ra == 31 to detect nops
def template LoadNopCheckDecode {{
{
AlphaStaticInst *i = new %(class_name)s(machInst);
if (RA == 31) {
i = makeNop(i);
}
return i;
}
}};
// for some load instructions, Ra == 31 indicates a prefetch (not a nop)
def template LoadPrefetchCheckDecode {{
{
if (RA != 31) {
return new %(class_name)s(machInst);
}
else {
return new %(class_name)sPrefetch(machInst);
}
}
}};
let {{
global LoadStoreBase
def LoadStoreBase(name, Name, ea_code, memacc_code, postacc_code = '',
base_class = 'Memory', flags = [],
declare_template = 'LoadStoreDeclare',
decode_template = 'BasicDecode',
exec_template = 'LoadStoreExecute'):
# Segregate flags into instruction flags (handled by InstObjParams)
# and memory access flags (handled here).
# Would be nice to autogenerate this list, but oh well.
valid_mem_flags = ['LOCKED', 'EVICT_NEXT', 'PF_EXCLUSIVE']
inst_flags = []
mem_flags = []
for f in flags:
if f in valid_mem_flags:
mem_flags.append(f)
else:
inst_flags.append(f)
ea_cblk = CodeBlock(ea_code)
memacc_cblk = CodeBlock(memacc_code)
postacc_cblk = CodeBlock(postacc_code)
cblk = CodeBlock(ea_code + memacc_code + postacc_code)
iop = InstObjParams(name, Name, base_class, cblk, inst_flags)
iop.ea_constructor = ea_cblk.constructor
iop.ea_code = ea_cblk.code
iop.memacc_constructor = memacc_cblk.constructor
iop.memacc_code = memacc_cblk.code
iop.postacc_code = postacc_cblk.code
mem_flags = string.join(mem_flags, '|')
if mem_flags != '':
iop.constructor += '\n\tmemAccessFlags = ' + mem_flags + ';'
return iop.subst(declare_template, decode_template, exec_template)
}};
def format LoadOrNop(ea_code, memacc_code, *flags) {{
return LoadStoreBase(name, Name, ea_code, memacc_code,
flags = flags,
decode_template = 'LoadNopCheckDecode')
}};
// Note that the flags passed in apply only to the prefetch version
def format LoadOrPrefetch(ea_code, memacc_code, *pf_flags) {{
# declare the load instruction object and generate the decode block
(decls, decode, exec_code) = \
LoadStoreBase(name, Name, ea_code, memacc_code,
decode_template = 'LoadPrefetchCheckDecode')
# Declare the prefetch instruction object.
# convert flags from tuple to list to make them mutable
pf_flags = list(pf_flags) + ['IsMemRef', 'IsLoad', 'IsDataPrefetch', 'RdPort']
(pfdecls, pfdecode, pfexec) = \
LoadStoreBase(name, Name + 'Prefetch', ea_code, '',
flags = pf_flags,
declare_template = 'PrefetchDeclare',
exec_template = 'PrefetchExecute')
return (decls + pfdecls, decode, exec_code + pfexec)
}};
def format Store(ea_code, memacc_code, *flags) {{
return LoadStoreBase(name, Name, ea_code, memacc_code,
flags = flags)
}};
def format StoreCond(ea_code, memacc_code, postacc_code, *flags) {{
return LoadStoreBase(name, Name, ea_code, memacc_code, postacc_code,
flags = flags)
}};
// Use 'MemoryNoDisp' as base: for wh64, fetch, ecb
def format MiscPrefetch(ea_code, memacc_code, *flags) {{
return LoadStoreBase(name, Name, ea_code, memacc_code,
flags = flags, base_class = 'MemoryNoDisp')
}};
////////////////////////////////////////////////////////////////////
declare {{
/**
* Base class for instructions whose disassembly is not purely a
* function of the machine instruction (i.e., it depends on the
* PC). This class overrides the disassemble() method to check
* the PC and symbol table values before re-using a cached
* disassembly string. This is necessary for branches and jumps,
* where the disassembly string includes the target address (which
* may depend on the PC and/or symbol table).
*/
class PCDependentDisassembly : public AlphaStaticInst
{
protected:
/// Cached program counter from last disassembly
Addr cachedPC;
/// Cached symbol table pointer from last disassembly
const SymbolTable *cachedSymtab;
/// Constructor
PCDependentDisassembly(const char *mnem, MachInst _machInst,
OpClass __opClass)
: AlphaStaticInst(mnem, _machInst, __opClass),
cachedPC(0), cachedSymtab(0)
{
}
const std::string &disassemble(Addr pc, const SymbolTable *symtab)
{
if (!cachedDisassembly ||
pc != cachedPC || symtab != cachedSymtab)
{
if (cachedDisassembly)
delete cachedDisassembly;
cachedDisassembly =
new std::string(generateDisassembly(pc, symtab));
cachedPC = pc;
cachedSymtab = symtab;
}
return *cachedDisassembly;
}
};
/**
* Base class for branches (PC-relative control transfers),
* conditional or unconditional.
*/
class Branch : public PCDependentDisassembly
{
protected:
/// Displacement to target address (signed).
int32_t disp;
/// Constructor.
Branch(const char *mnem, MachInst _machInst, OpClass __opClass)
: PCDependentDisassembly(mnem, _machInst, __opClass),
disp(BRDISP << 2)
{
}
Addr branchTarget(Addr branchPC) const
{
return branchPC + 4 + disp;
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
std::stringstream ss;
ccprintf(ss, "%-10s ", mnemonic);
// There's only one register arg (RA), but it could be
// either a source (the condition for conditional
// branches) or a destination (the link reg for
// unconditional branches)
if (_numSrcRegs > 0) {
printReg(ss, _srcRegIdx[0]);
ss << ",";
}
else if (_numDestRegs > 0) {
printReg(ss, _destRegIdx[0]);
ss << ",";
}
#ifdef SS_COMPATIBLE_DISASSEMBLY
if (_numSrcRegs == 0 && _numDestRegs == 0) {
printReg(ss, 31);
ss << ",";
}
#endif
Addr target = pc + 4 + disp;
std::string str;
if (symtab && symtab->findSymbol(target, str))
ss << str;
else
ccprintf(ss, "0x%x", target);
return ss.str();
}
};
/**
* Base class for jumps (register-indirect control transfers). In
* the Alpha ISA, these are always unconditional.
*/
class Jump : public PCDependentDisassembly
{
protected:
/// Displacement to target address (signed).
int32_t disp;
public:
/// Constructor
Jump(const char *mnem, MachInst _machInst, OpClass __opClass)
: PCDependentDisassembly(mnem, _machInst, __opClass),
disp(BRDISP)
{
}
Addr branchTarget(ExecContext *xc) const
{
Addr NPC = xc->readPC() + 4;
uint64_t Rb = xc->readIntReg(_srcRegIdx[0]);
return (Rb & ~3) | (NPC & 1);
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
std::stringstream ss;
ccprintf(ss, "%-10s ", mnemonic);
#ifdef SS_COMPATIBLE_DISASSEMBLY
if (_numDestRegs == 0) {
printReg(ss, 31);
ss << ",";
}
#endif
if (_numDestRegs > 0) {
printReg(ss, _destRegIdx[0]);
ss << ",";
}
ccprintf(ss, "(r%d)", RB);
return ss.str();
}
};
}};
def template JumpOrBranchDecode {{
return (RA == 31)
? (StaticInst<AlphaISA> *)new %(class_name)s(machInst)
: (StaticInst<AlphaISA> *)new %(class_name)sAndLink(machInst);
}};
def format CondBranch(code) {{
code = 'bool cond;\n' + code + '\nif (cond) NPC = NPC + disp;\n';
iop = InstObjParams(name, Name, 'Branch', CodeBlock(code),
('IsDirectControl', 'IsCondControl'))
return iop.subst('BasicDeclare', 'BasicDecode', 'BasicExecute')
}};
let {{
global UncondCtrlBase
def UncondCtrlBase(name, Name, base_class, npc_expr, flags):
# Declare basic control transfer w/o link (i.e. link reg is R31)
nolink_code = 'NPC = %s;\n' % npc_expr
nolink_iop = InstObjParams(name, Name, base_class,
CodeBlock(nolink_code), flags)
(decls, exec_code) = nolink_iop.subst('BasicDeclare', 'BasicExecute')
# Generate declaration of '*AndLink' version, append to decls
link_code = 'Ra = NPC & ~3;\n' + nolink_code
link_iop = InstObjParams(name, Name + 'AndLink', base_class,
CodeBlock(link_code), flags)
(link_decls, link_exec_code) = \
link_iop.subst('BasicDeclare', 'BasicExecute')
decls += link_decls
exec_code += link_exec_code
# need to use link_iop for the decode template since it is expecting
# the shorter version of class_name (w/o "AndLink")
return (decls, nolink_iop.subst('JumpOrBranchDecode'), exec_code)
}};
def format UncondBranch(*flags) {{
flags += ('IsUncondControl', 'IsDirectControl')
return UncondCtrlBase(name, Name, 'Branch', 'NPC + disp', flags)
}};
def format Jump(*flags) {{
flags += ('IsUncondControl', 'IsIndirectControl')
return UncondCtrlBase(name, Name, 'Jump', '(Rb & ~3) | (NPC & 1)', flags)
}};
declare {{
/**
* Base class for emulated call_pal calls (used only in
* non-full-system mode).
*/
class EmulatedCallPal : public AlphaStaticInst
{
protected:
/// Constructor.
EmulatedCallPal(const char *mnem, MachInst _machInst,
OpClass __opClass)
: AlphaStaticInst(mnem, _machInst, __opClass)
{
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
#ifdef SS_COMPATIBLE_DISASSEMBLY
return csprintf("%s %s", "call_pal", mnemonic);
#else
return csprintf("%-10s %s", "call_pal", mnemonic);
#endif
}
};
}};
def format EmulatedCallPal(code) {{
iop = InstObjParams(name, Name, 'EmulatedCallPal', CodeBlock(code))
return iop.subst('BasicDeclare', 'BasicDecode', 'BasicExecute')
}};
declare {{
/**
* Base class for full-system-mode call_pal instructions.
* Probably could turn this into a leaf class and get rid of the
* parser template.
*/
class CallPalBase : public AlphaStaticInst
{
protected:
int palFunc; ///< Function code part of instruction
int palOffset; ///< Target PC, offset from IPR_PAL_BASE
bool palValid; ///< is the function code valid?
bool palPriv; ///< is this call privileged?
/// Constructor.
CallPalBase(const char *mnem, MachInst _machInst,
OpClass __opClass)
: AlphaStaticInst(mnem, _machInst, __opClass),
palFunc(PALFUNC)
{
// From the 21164 HRM (paraphrased):
// Bit 7 of the function code (mask 0x80) indicates
// whether the call is privileged (bit 7 == 0) or
// unprivileged (bit 7 == 1). The privileged call table
// starts at 0x2000, the unprivielged call table starts at
// 0x3000. Bits 5-0 (mask 0x3f) are used to calculate the
// offset.
const int palPrivMask = 0x80;
const int palOffsetMask = 0x3f;
// Pal call is invalid unless all other bits are 0
palValid = ((machInst & ~(palPrivMask | palOffsetMask)) == 0);
palPriv = ((machInst & palPrivMask) == 0);
int shortPalFunc = (machInst & palOffsetMask);
// Add 1 to base to set pal-mode bit
palOffset = (palPriv ? 0x2001 : 0x3001) + (shortPalFunc << 6);
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
return csprintf("%-10s %#x", "call_pal", palFunc);
}
};
}};
def format CallPal(code) {{
iop = InstObjParams(name, Name, 'CallPalBase', CodeBlock(code))
return iop.subst('BasicDeclare', 'BasicDecode', 'BasicExecute')
}};
//
// hw_ld, hw_st
//
declare {{
/**
* Base class for hw_ld and hw_st.
*/
class HwLoadStore : public AlphaStaticInst
{
protected:
/// Displacement for EA calculation (signed).
int16_t disp;
/// Memory request flags. See mem_req_base.hh.
unsigned memAccessFlags;
/// Constructor
HwLoadStore(const char *mnem, MachInst _machInst, OpClass __opClass)
: AlphaStaticInst(mnem, _machInst, __opClass), disp(HW_LDST_DISP)
{
memAccessFlags = 0;
if (HW_LDST_PHYS) memAccessFlags |= PHYSICAL;
if (HW_LDST_ALT) memAccessFlags |= ALTMODE;
if (HW_LDST_VPTE) memAccessFlags |= VPTE;
if (HW_LDST_LOCK) memAccessFlags |= LOCKED;
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
#ifdef SS_COMPATIBLE_DISASSEMBLY
return csprintf("%-10s r%d,%d(r%d)", mnemonic, RA, disp, RB);
#else
// HW_LDST_LOCK and HW_LDST_COND are the same bit.
const char *lock_str =
(HW_LDST_LOCK) ? (flags[IsLoad] ? ",LOCK" : ",COND") : "";
return csprintf("%-10s r%d,%d(r%d)%s%s%s%s%s",
mnemonic, RA, disp, RB,
HW_LDST_PHYS ? ",PHYS" : "",
HW_LDST_ALT ? ",ALT" : "",
HW_LDST_QUAD ? ",QUAD" : "",
HW_LDST_VPTE ? ",VPTE" : "",
lock_str);
#endif
}
};
}};
def format HwLoadStore(ea_code, memacc_code, class_ext, *flags) {{
return LoadStoreBase(name, Name + class_ext, ea_code, memacc_code,
flags = flags,
base_class = 'HwLoadStore')
}};
def format HwStoreCond(ea_code, memacc_code, postacc_code, class_ext, *flags) {{
return LoadStoreBase(name, Name + class_ext,
ea_code, memacc_code, postacc_code,
flags = flags,
base_class = 'HwLoadStore')
}};
declare {{
/**
* Base class for hw_mfpr and hw_mtpr.
*/
class HwMoveIPR : public AlphaStaticInst
{
protected:
/// Index of internal processor register.
int ipr_index;
/// Constructor
HwMoveIPR(const char *mnem, MachInst _machInst, OpClass __opClass)
: AlphaStaticInst(mnem, _machInst, __opClass),
ipr_index(HW_IPR_IDX)
{
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
if (_numSrcRegs > 0) {
// must be mtpr
return csprintf("%-10s r%d,IPR(%#x)",
mnemonic, RA, ipr_index);
}
else {
// must be mfpr
return csprintf("%-10s IPR(%#x),r%d",
mnemonic, ipr_index, RA);
}
}
};
}};
def format HwMoveIPR(code) {{
iop = InstObjParams(name, Name, 'HwMoveIPR', CodeBlock(code))
return iop.subst('BasicDeclare', 'BasicDecode', 'BasicExecute')
}};
declare {{
/**
* Static instruction class for unimplemented instructions that
* cause simulator termination. Note that these are recognized
* (legal) instructions that the simulator does not support; the
* 'Unknown' class is used for unrecognized/illegal instructions.
* This is a leaf class.
*/
class FailUnimplemented : public AlphaStaticInst
{
public:
/// Constructor
FailUnimplemented(const char *_mnemonic, MachInst _machInst)
: AlphaStaticInst(_mnemonic, _machInst, No_OpClass)
{
}
Fault execute(SimpleCPUExecContext *xc,
Trace::InstRecord *traceData)
{
panic("attempt to execute unimplemented instruction '%s' "
"(inst 0x%08x, opcode 0x%x)", mnemonic, machInst, OPCODE);
return Unimplemented_Opcode_Fault;
}
Fault execute(FullCPUExecContext *xc,
Trace::InstRecord *traceData)
{
// don't panic if this is a misspeculated instruction
if (!xc->misspeculating())
panic("attempt to execute unimplemented instruction '%s' "
"(inst 0x%08x, opcode 0x%x)",
mnemonic, machInst, OPCODE);
return Unimplemented_Opcode_Fault;
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
return csprintf("%-10s (unimplemented)", mnemonic);
}
};
/**
* Base class for unimplemented instructions that cause a warning
* to be printed (but do not terminate simulation). This
* implementation is a little screwy in that it will print a
* warning for each instance of a particular unimplemented machine
* instruction, not just for each unimplemented opcode. Should
* probably make the 'warned' flag a static member of the derived
* class.
*/
class WarnUnimplemented : public AlphaStaticInst
{
private:
/// Have we warned on this instruction yet?
bool warned;
public:
/// Constructor
WarnUnimplemented(const char *_mnemonic, MachInst _machInst)
: AlphaStaticInst(_mnemonic, _machInst, No_OpClass), warned(false)
{
}
Fault execute(SimpleCPUExecContext *xc,
Trace::InstRecord *traceData)
{
if (!warned) {
warn("instruction '%s' unimplemented\n", mnemonic);
warned = true;
}
return No_Fault;
}
Fault execute(FullCPUExecContext *xc,
Trace::InstRecord *traceData)
{
if (!xc->misspeculating() && !warned) {
warn("instruction '%s' unimplemented\n", mnemonic);
warned = true;
}
return No_Fault;
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
#ifdef SS_COMPATIBLE_DISASSEMBLY
return csprintf("%-10s", mnemonic);
#else
return csprintf("%-10s (unimplemented)", mnemonic);
#endif
}
};
}};
def template WarnUnimplDeclare {{
/**
* Static instruction class for "%(mnemonic)s".
*/
class %(class_name)s : public %(base_class)s
{
public:
/// Constructor
%(class_name)s(MachInst machInst)
: %(base_class)s("%(mnemonic)s", machInst)
{
}
};
}};
def format FailUnimpl() {{
iop = InstObjParams(name, 'FailUnimplemented')
return ('', iop.subst('BasicDecodeWithMnemonic'), '')
}};
def format WarnUnimpl() {{
iop = InstObjParams(name, Name, 'WarnUnimplemented')
return iop.subst('WarnUnimplDeclare', 'BasicDecode') + ['']
}};
declare {{
/**
* Static instruction class for unknown (illegal) instructions.
* These cause simulator termination if they are executed in a
* non-speculative mode. This is a leaf class.
*/
class Unknown : public AlphaStaticInst
{
public:
/// Constructor
Unknown(MachInst _machInst)
: AlphaStaticInst("unknown", _machInst, No_OpClass)
{
}
Fault execute(SimpleCPUExecContext *xc,
Trace::InstRecord *traceData)
{
panic("attempt to execute unknown instruction "
"(inst 0x%08x, opcode 0x%x)", machInst, OPCODE);
return Unimplemented_Opcode_Fault;
}
Fault execute(FullCPUExecContext *xc,
Trace::InstRecord *traceData)
{
// don't panic if this is a misspeculated instruction
if (!xc->misspeculating())
panic("attempt to execute unknown instruction "
"(inst 0x%08x, opcode 0x%x)", machInst, OPCODE);
return Unimplemented_Opcode_Fault;
}
std::string generateDisassembly(Addr pc, const SymbolTable *symtab)
{
return csprintf("%-10s (inst 0x%x, opcode 0x%x)",
"unknown", machInst, OPCODE);
}
};
}};
def format Unknown() {{
return ('', 'return new Unknown(machInst);\n', '')
}};
declare {{
/// Return opa + opb, summing carry into third arg.
inline uint64_t
addc(uint64_t opa, uint64_t opb, int &carry)
{
uint64_t res = opa + opb;
if (res < opa || res < opb)
++carry;
return res;
}
/// Multiply two 64-bit values (opa * opb), returning the 128-bit
/// product in res_hi and res_lo.
void
mul128(uint64_t opa, uint64_t opb, uint64_t &res_hi, uint64_t &res_lo)
{
// do a 64x64 --> 128 multiply using four 32x32 --> 64 multiplies
uint64_t opa_hi = opa<63:32>;
uint64_t opa_lo = opa<31:0>;
uint64_t opb_hi = opb<63:32>;
uint64_t opb_lo = opb<31:0>;
res_lo = opa_lo * opb_lo;
// The middle partial products logically belong in bit
// positions 95 to 32. Thus the lower 32 bits of each product
// sum into the upper 32 bits of the low result, while the
// upper 32 sum into the low 32 bits of the upper result.
uint64_t partial1 = opa_hi * opb_lo;
uint64_t partial2 = opa_lo * opb_hi;
uint64_t partial1_lo = partial1<31:0> << 32;
uint64_t partial1_hi = partial1<63:32>;
uint64_t partial2_lo = partial2<31:0> << 32;
uint64_t partial2_hi = partial2<63:32>;
// Add partial1_lo and partial2_lo to res_lo, keeping track
// of any carries out
int carry_out = 0;
res_lo = addc(partial1_lo, res_lo, carry_out);
res_lo = addc(partial2_lo, res_lo, carry_out);
// Now calculate the high 64 bits...
res_hi = (opa_hi * opb_hi) + partial1_hi + partial2_hi + carry_out;
}
/// Map 8-bit S-floating exponent to 11-bit T-floating exponent.
/// See Table 2-2 of Alpha AHB.
inline int
map_s(int old_exp)
{
int hibit = old_exp<7:>;
int lobits = old_exp<6:0>;
if (hibit == 1) {
return (lobits == 0x7f) ? 0x7ff : (0x400 | lobits);
}
else {
return (lobits == 0) ? 0 : (0x380 | lobits);
}
}
/// Convert a 32-bit S-floating value to the equivalent 64-bit
/// representation to be stored in an FP reg.
inline uint64_t
s_to_t(uint32_t s_val)
{
uint64_t tmp = s_val;
return (tmp<31:> << 63 // sign bit
| (uint64_t)map_s(tmp<30:23>) << 52 // exponent
| tmp<22:0> << 29); // fraction
}
/// Convert a 64-bit T-floating value to the equivalent 32-bit
/// S-floating representation to be stored in memory.
inline int32_t
t_to_s(uint64_t t_val)
{
return (t_val<63:62> << 30 // sign bit & hi exp bit
| t_val<58:29>); // rest of exp & fraction
}
}};
decode OPCODE default Unknown::unknown() {
format LoadAddress {
0x08: lda({{ Ra = Rb + disp; }});
0x09: ldah({{ Ra = Rb + (disp << 16); }});
}
format LoadOrNop {
0x0a: ldbu({{ EA = Rb + disp; }}, {{ Ra.uq = Mem.ub; }});
0x0c: ldwu({{ EA = Rb + disp; }}, {{ Ra.uq = Mem.uw; }});
0x0b: ldq_u({{ EA = (Rb + disp) & ~7; }}, {{ Ra = Mem.uq; }});
0x23: ldt({{ EA = Rb + disp; }}, {{ Fa = Mem.df; }});
0x2a: ldl_l({{ EA = Rb + disp; }}, {{ Ra.sl = Mem.sl; }}, LOCKED);
0x2b: ldq_l({{ EA = Rb + disp; }}, {{ Ra.uq = Mem.uq; }}, LOCKED);
0x20: copy_load({{EA = Ra;}},
{{ fault = xc->copySrcTranslate(EA);}},
IsMemRef, IsLoad, IsCopy);
}
format LoadOrPrefetch {
0x28: ldl({{ EA = Rb + disp; }}, {{ Ra.sl = Mem.sl; }});
0x29: ldq({{ EA = Rb + disp; }}, {{ Ra.uq = Mem.uq; }}, EVICT_NEXT);
// IsFloating flag on lds gets the prefetch to disassemble
// using f31 instead of r31... funcitonally it's unnecessary
0x22: lds({{ EA = Rb + disp; }}, {{ Fa.uq = s_to_t(Mem.ul); }},
PF_EXCLUSIVE, IsFloating);
}
format Store {
0x0e: stb({{ EA = Rb + disp; }}, {{ Mem.ub = Ra<7:0>; }});
0x0d: stw({{ EA = Rb + disp; }}, {{ Mem.uw = Ra<15:0>; }});
0x2c: stl({{ EA = Rb + disp; }}, {{ Mem.ul = Ra<31:0>; }});
0x2d: stq({{ EA = Rb + disp; }}, {{ Mem.uq = Ra.uq; }});
0x0f: stq_u({{ EA = (Rb + disp) & ~7; }}, {{ Mem.uq = Ra.uq; }});
0x26: sts({{ EA = Rb + disp; }}, {{ Mem.ul = t_to_s(Fa.uq); }});
0x27: stt({{ EA = Rb + disp; }}, {{ Mem.df = Fa; }});
0x24: copy_store({{EA = Rb;}},
{{ fault = xc->copy(EA);}},
IsMemRef, IsStore, IsCopy);
}
format StoreCond {
0x2e: stl_c({{ EA = Rb + disp; }}, {{ Mem.ul = Ra<31:0>; }},
{{
uint64_t tmp = Mem_write_result;
// see stq_c
Ra = (tmp == 0 || tmp == 1) ? tmp : Ra;
}}, LOCKED);
0x2f: stq_c({{ EA = Rb + disp; }}, {{ Mem.uq = Ra; }},
{{
uint64_t tmp = Mem_write_result;
// If the write operation returns 0 or 1, then
// this was a conventional store conditional,
// and the value indicates the success/failure
// of the operation. If another value is
// returned, then this was a Turbolaser
// mailbox access, and we don't update the
// result register at all.
Ra = (tmp == 0 || tmp == 1) ? tmp : Ra;
}}, LOCKED);
}
format IntegerOperate {
0x10: decode INTFUNC { // integer arithmetic operations
0x00: addl({{ Rc.sl = Ra.sl + Rb_or_imm.sl; }});
0x40: addlv({{
uint32_t tmp = Ra.sl + Rb_or_imm.sl;
// signed overflow occurs when operands have same sign
// and sign of result does not match.
if (Ra.sl<31:> == Rb_or_imm.sl<31:> && tmp<31:> != Ra.sl<31:>)
fault = Integer_Overflow_Fault;
Rc.sl = tmp;
}});
0x02: s4addl({{ Rc.sl = (Ra.sl << 2) + Rb_or_imm.sl; }});
0x12: s8addl({{ Rc.sl = (Ra.sl << 3) + Rb_or_imm.sl; }});
0x20: addq({{ Rc = Ra + Rb_or_imm; }});
0x60: addqv({{
uint64_t tmp = Ra + Rb_or_imm;
// signed overflow occurs when operands have same sign
// and sign of result does not match.
if (Ra<63:> == Rb_or_imm<63:> && tmp<63:> != Ra<63:>)
fault = Integer_Overflow_Fault;
Rc = tmp;
}});
0x22: s4addq({{ Rc = (Ra << 2) + Rb_or_imm; }});
0x32: s8addq({{ Rc = (Ra << 3) + Rb_or_imm; }});
0x09: subl({{ Rc.sl = Ra.sl - Rb_or_imm.sl; }});
0x49: sublv({{
uint32_t tmp = Ra.sl - Rb_or_imm.sl;
// signed overflow detection is same as for add,
// except we need to look at the *complemented*
// sign bit of the subtrahend (Rb), i.e., if the initial
// signs are the *same* then no overflow can occur
if (Ra.sl<31:> != Rb_or_imm.sl<31:> && tmp<31:> != Ra.sl<31:>)
fault = Integer_Overflow_Fault;
Rc.sl = tmp;
}});
0x0b: s4subl({{ Rc.sl = (Ra.sl << 2) - Rb_or_imm.sl; }});
0x1b: s8subl({{ Rc.sl = (Ra.sl << 3) - Rb_or_imm.sl; }});
0x29: subq({{ Rc = Ra - Rb_or_imm; }});
0x69: subqv({{
uint64_t tmp = Ra - Rb_or_imm;
// signed overflow detection is same as for add,
// except we need to look at the *complemented*
// sign bit of the subtrahend (Rb), i.e., if the initial
// signs are the *same* then no overflow can occur
if (Ra<63:> != Rb_or_imm<63:> && tmp<63:> != Ra<63:>)
fault = Integer_Overflow_Fault;
Rc = tmp;
}});
0x2b: s4subq({{ Rc = (Ra << 2) - Rb_or_imm; }});
0x3b: s8subq({{ Rc = (Ra << 3) - Rb_or_imm; }});
0x2d: cmpeq({{ Rc = (Ra == Rb_or_imm); }});
0x6d: cmple({{ Rc = (Ra.sq <= Rb_or_imm.sq); }});
0x4d: cmplt({{ Rc = (Ra.sq < Rb_or_imm.sq); }});
0x3d: cmpule({{ Rc = (Ra.uq <= Rb_or_imm.uq); }});
0x1d: cmpult({{ Rc = (Ra.uq < Rb_or_imm.uq); }});
0x0f: cmpbge({{
int hi = 7;
int lo = 0;
uint64_t tmp = 0;
for (int i = 0; i < 8; ++i) {
tmp |= (Ra.uq<hi:lo> >= Rb_or_imm.uq<hi:lo>) << i;
hi += 8;
lo += 8;
}
Rc = tmp;
}});
}
0x11: decode INTFUNC { // integer logical operations
0x00: and({{ Rc = Ra & Rb_or_imm; }});
0x08: bic({{ Rc = Ra & ~Rb_or_imm; }});
0x20: bis({{ Rc = Ra | Rb_or_imm; }});
0x28: ornot({{ Rc = Ra | ~Rb_or_imm; }});
0x40: xor({{ Rc = Ra ^ Rb_or_imm; }});
0x48: eqv({{ Rc = Ra ^ ~Rb_or_imm; }});
// conditional moves
0x14: cmovlbs({{ Rc = ((Ra & 1) == 1) ? Rb_or_imm : Rc; }});
0x16: cmovlbc({{ Rc = ((Ra & 1) == 0) ? Rb_or_imm : Rc; }});
0x24: cmoveq({{ Rc = (Ra == 0) ? Rb_or_imm : Rc; }});
0x26: cmovne({{ Rc = (Ra != 0) ? Rb_or_imm : Rc; }});
0x44: cmovlt({{ Rc = (Ra.sq < 0) ? Rb_or_imm : Rc; }});
0x46: cmovge({{ Rc = (Ra.sq >= 0) ? Rb_or_imm : Rc; }});
0x64: cmovle({{ Rc = (Ra.sq <= 0) ? Rb_or_imm : Rc; }});
0x66: cmovgt({{ Rc = (Ra.sq > 0) ? Rb_or_imm : Rc; }});
// For AMASK, RA must be R31.
0x61: decode RA {
31: amask({{ Rc = Rb_or_imm & ~ULL(0x17); }});
}
// For IMPLVER, RA must be R31 and the B operand
// must be the immediate value 1.
0x6c: decode RA {
31: decode IMM {
1: decode INTIMM {
// return EV5 for FULL_SYSTEM and EV6 otherwise
1: implver({{
#ifdef FULL_SYSTEM
Rc = 1;
#else
Rc = 2;
#endif
}});
}
}
}
#ifdef FULL_SYSTEM
// The mysterious 11.25...
0x25: WarnUnimpl::eleven25();
#endif
}
0x12: decode INTFUNC {
0x39: sll({{ Rc = Ra << Rb_or_imm<5:0>; }});
0x34: srl({{ Rc = Ra.uq >> Rb_or_imm<5:0>; }});
0x3c: sra({{ Rc = Ra.sq >> Rb_or_imm<5:0>; }});
0x02: mskbl({{ Rc = Ra & ~(mask( 8) << (Rb_or_imm<2:0> * 8)); }});
0x12: mskwl({{ Rc = Ra & ~(mask(16) << (Rb_or_imm<2:0> * 8)); }});
0x22: mskll({{ Rc = Ra & ~(mask(32) << (Rb_or_imm<2:0> * 8)); }});
0x32: mskql({{ Rc = Ra & ~(mask(64) << (Rb_or_imm<2:0> * 8)); }});
0x52: mskwh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra & ~(mask(16) >> (64 - 8 * bv))) : Ra;
}});
0x62: msklh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra & ~(mask(32) >> (64 - 8 * bv))) : Ra;
}});
0x72: mskqh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra & ~(mask(64) >> (64 - 8 * bv))) : Ra;
}});
0x06: extbl({{ Rc = (Ra.uq >> (Rb_or_imm<2:0> * 8))< 7:0>; }});
0x16: extwl({{ Rc = (Ra.uq >> (Rb_or_imm<2:0> * 8))<15:0>; }});
0x26: extll({{ Rc = (Ra.uq >> (Rb_or_imm<2:0> * 8))<31:0>; }});
0x36: extql({{ Rc = (Ra.uq >> (Rb_or_imm<2:0> * 8)); }});
0x5a: extwh({{
Rc = (Ra << (64 - (Rb_or_imm<2:0> * 8))<5:0>)<15:0>; }});
0x6a: extlh({{
Rc = (Ra << (64 - (Rb_or_imm<2:0> * 8))<5:0>)<31:0>; }});
0x7a: extqh({{
Rc = (Ra << (64 - (Rb_or_imm<2:0> * 8))<5:0>); }});
0x0b: insbl({{ Rc = Ra< 7:0> << (Rb_or_imm<2:0> * 8); }});
0x1b: inswl({{ Rc = Ra<15:0> << (Rb_or_imm<2:0> * 8); }});
0x2b: insll({{ Rc = Ra<31:0> << (Rb_or_imm<2:0> * 8); }});
0x3b: insql({{ Rc = Ra << (Rb_or_imm<2:0> * 8); }});
0x57: inswh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra.uq<15:0> >> (64 - 8 * bv)) : 0;
}});
0x67: inslh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra.uq<31:0> >> (64 - 8 * bv)) : 0;
}});
0x77: insqh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra.uq >> (64 - 8 * bv)) : 0;
}});
0x30: zap({{
uint64_t zapmask = 0;
for (int i = 0; i < 8; ++i) {
if (Rb_or_imm<i:>)
zapmask |= (mask(8) << (i * 8));
}
Rc = Ra & ~zapmask;
}});
0x31: zapnot({{
uint64_t zapmask = 0;
for (int i = 0; i < 8; ++i) {
if (!Rb_or_imm<i:>)
zapmask |= (mask(8) << (i * 8));
}
Rc = Ra & ~zapmask;
}});
}
0x13: decode INTFUNC { // integer multiplies
0x00: mull({{ Rc.sl = Ra.sl * Rb_or_imm.sl; }}, IntMULT);
0x20: mulq({{ Rc = Ra * Rb_or_imm; }}, IntMULT);
0x30: umulh({{
uint64_t hi, lo;
mul128(Ra, Rb_or_imm, hi, lo);
Rc = hi;
}}, IntMULT);
0x40: mullv({{
// 32-bit multiply with trap on overflow
int64_t Rax = Ra.sl; // sign extended version of Ra.sl
int64_t Rbx = Rb_or_imm.sl;
int64_t tmp = Rax * Rbx;
// To avoid overflow, all the upper 32 bits must match
// the sign bit of the lower 32. We code this as
// checking the upper 33 bits for all 0s or all 1s.
uint64_t sign_bits = tmp<63:31>;
if (sign_bits != 0 && sign_bits != mask(33))
fault = Integer_Overflow_Fault;
Rc.sl = tmp<31:0>;
}}, IntMULT);
0x60: mulqv({{
// 64-bit multiply with trap on overflow
uint64_t hi, lo;
mul128(Ra, Rb_or_imm, hi, lo);
// all the upper 64 bits must match the sign bit of
// the lower 64
if (!((hi == 0 && lo<63:> == 0) ||
(hi == mask(64) && lo<63:> == 1)))
fault = Integer_Overflow_Fault;
Rc = lo;
}}, IntMULT);
}
0x1c: decode INTFUNC {
0x00: decode RA { 31: sextb({{ Rc.sb = Rb_or_imm< 7:0>; }}); }
0x01: decode RA { 31: sextw({{ Rc.sw = Rb_or_imm<15:0>; }}); }
format FailUnimpl {
0x30: ctpop();
0x31: perr();
0x32: ctlz();
0x33: cttz();
0x34: unpkbw();
0x35: unpkbl();
0x36: pkwb();
0x37: pklb();
0x38: minsb8();
0x39: minsw4();
0x3a: minub8();
0x3b: minuw4();
0x3c: maxub8();
0x3d: maxuw4();
0x3e: maxsb8();
0x3f: maxsw4();
}
format BasicOperateWithNopCheck {
0x70: decode RB {
31: ftoit({{ Rc = Fa.uq; }}, FloatCVT);
}
0x78: decode RB {
31: ftois({{ Rc.sl = t_to_s(Fa.uq); }},
FloatCVT);
}
}
}
}
// Conditional branches.
format CondBranch {
0x39: beq({{ cond = (Ra == 0); }});
0x3d: bne({{ cond = (Ra != 0); }});
0x3e: bge({{ cond = (Ra.sq >= 0); }});
0x3f: bgt({{ cond = (Ra.sq > 0); }});
0x3b: ble({{ cond = (Ra.sq <= 0); }});
0x3a: blt({{ cond = (Ra.sq < 0); }});
0x38: blbc({{ cond = ((Ra & 1) == 0); }});
0x3c: blbs({{ cond = ((Ra & 1) == 1); }});
0x31: fbeq({{ cond = (Fa == 0); }});
0x35: fbne({{ cond = (Fa != 0); }});
0x36: fbge({{ cond = (Fa >= 0); }});
0x37: fbgt({{ cond = (Fa > 0); }});
0x33: fble({{ cond = (Fa <= 0); }});
0x32: fblt({{ cond = (Fa < 0); }});
}
// unconditional branches
format UncondBranch {
0x30: br();
0x34: bsr(IsCall);
}
// indirect branches
0x1a: decode JMPFUNC {
format Jump {
0: jmp();
1: jsr(IsCall);
2: ret(IsReturn);
3: jsr_coroutine(IsCall, IsReturn);
}
}
// IEEE floating point
0x14: decode FP_SHORTFUNC {
// Integer to FP register moves must have RB == 31
0x4: decode RB {
31: decode FP_FULLFUNC {
format BasicOperateWithNopCheck {
0x004: itofs({{ Fc.uq = s_to_t(Ra.ul); }}, FloatCVT);
0x024: itoft({{ Fc.uq = Ra.uq; }}, FloatCVT);
0x014: FailUnimpl::itoff(); // VAX-format conversion
}
}
}
// Square root instructions must have FA == 31
0xb: decode FA {
31: decode FP_TYPEFUNC {
format FloatingPointOperate {
#ifdef SS_COMPATIBLE_FP
0x0b: sqrts({{
if (Fb < 0.0)
fault = Arithmetic_Fault;
Fc = sqrt(Fb);
}}, FloatSQRT);
#else
0x0b: sqrts({{
if (Fb.sf < 0.0)
fault = Arithmetic_Fault;
Fc.sf = sqrt(Fb.sf);
}}, FloatSQRT);
#endif
0x2b: sqrtt({{
if (Fb < 0.0)
fault = Arithmetic_Fault;
Fc = sqrt(Fb);
}}, FloatSQRT);
}
}
}
// VAX-format sqrtf and sqrtg are not implemented
0xa: FailUnimpl::sqrtfg();
}
// IEEE floating point
0x16: decode FP_SHORTFUNC_TOP2 {
// The top two bits of the short function code break this space
// into four groups: binary ops, compares, reserved, and conversions.
// See Table 4-12 of AHB.
// Most of these instructions may have various trapping and
// rounding mode flags set; these are decoded in the
// FloatingPointDecode template used by the
// FloatingPointOperate format.
// add/sub/mul/div: just decode on the short function code
// and source type.
0: decode FP_TYPEFUNC {
format FloatingPointOperate {
#ifdef SS_COMPATIBLE_FP
0x00: adds({{ Fc = Fa + Fb; }});
0x01: subs({{ Fc = Fa - Fb; }});
0x02: muls({{ Fc = Fa * Fb; }}, FloatMULT);
0x03: divs({{ Fc = Fa / Fb; }}, FloatDIV);
#else
0x00: adds({{ Fc.sf = Fa.sf + Fb.sf; }});
0x01: subs({{ Fc.sf = Fa.sf - Fb.sf; }});
0x02: muls({{ Fc.sf = Fa.sf * Fb.sf; }}, FloatMULT);
0x03: divs({{ Fc.sf = Fa.sf / Fb.sf; }}, FloatDIV);
#endif
0x20: addt({{ Fc = Fa + Fb; }});
0x21: subt({{ Fc = Fa - Fb; }});
0x22: mult({{ Fc = Fa * Fb; }}, FloatMULT);
0x23: divt({{ Fc = Fa / Fb; }}, FloatDIV);
}
}
// Floating-point compare instructions must have the default
// rounding mode, and may use the default trapping mode or
// /SU. Both trapping modes are treated the same by M5; the
// only difference on the real hardware (as far a I can tell)
// is that without /SU you'd get an imprecise trap if you
// tried to compare a NaN with something else (instead of an
// "unordered" result).
1: decode FP_FULLFUNC {
format BasicOperateWithNopCheck {
0x0a5, 0x5a5: cmpteq({{ Fc = (Fa == Fb) ? 2.0 : 0.0; }},
FloatCMP);
0x0a7, 0x5a7: cmptle({{ Fc = (Fa <= Fb) ? 2.0 : 0.0; }},
FloatCMP);
0x0a6, 0x5a6: cmptlt({{ Fc = (Fa < Fb) ? 2.0 : 0.0; }},
FloatCMP);
0x0a4, 0x5a4: cmptun({{ // unordered
Fc = (!(Fa < Fb) && !(Fa == Fb) && !(Fa > Fb)) ? 2.0 : 0.0;
}}, FloatCMP);
}
}
// The FP-to-integer and integer-to-FP conversion insts
// require that FA be 31.
3: decode FA {
31: decode FP_TYPEFUNC {
format FloatingPointOperate {
0x2f: cvttq({{ Fc.sq = (int64_t)rint(Fb); }});
// The cvtts opcode is overloaded to be cvtst if the trap
// mode is 2 or 6 (which are not valid otherwise)
0x2c: decode FP_FULLFUNC {
format BasicOperateWithNopCheck {
// trap on denorm version "cvtst/s" is
// simulated same as cvtst
0x2ac, 0x6ac: cvtst({{ Fc = Fb.sf; }});
}
default: cvtts({{ Fc.sf = Fb; }});
}
// The trapping mode for integer-to-FP conversions
// must be /SUI or nothing; /U and /SU are not
// allowed. The full set of rounding modes are
// supported though.
0x3c: decode FP_TRAPMODE {
0,7: cvtqs({{ Fc.sf = Fb.sq; }});
}
0x3e: decode FP_TRAPMODE {
0,7: cvtqt({{ Fc = Fb.sq; }});
}
}
}
}
}
// misc FP operate
0x17: decode FP_FULLFUNC {
format BasicOperateWithNopCheck {
0x010: cvtlq({{
Fc.sl = (Fb.uq<63:62> << 30) | Fb.uq<58:29>;
}});
0x030: cvtql({{
Fc.uq = (Fb.uq<31:30> << 62) | (Fb.uq<29:0> << 29);
}});
// We treat the precise & imprecise trapping versions of
// cvtql identically.
0x130, 0x530: cvtqlv({{
// To avoid overflow, all the upper 32 bits must match
// the sign bit of the lower 32. We code this as
// checking the upper 33 bits for all 0s or all 1s.
uint64_t sign_bits = Fb.uq<63:31>;
if (sign_bits != 0 && sign_bits != mask(33))
fault = Integer_Overflow_Fault;
Fc.uq = (Fb.uq<31:30> << 62) | (Fb.uq<29:0> << 29);
}});
0x020: cpys({{ // copy sign
Fc.uq = (Fa.uq<63:> << 63) | Fb.uq<62:0>;
}});
0x021: cpysn({{ // copy sign negated
Fc.uq = (~Fa.uq<63:> << 63) | Fb.uq<62:0>;
}});
0x022: cpyse({{ // copy sign and exponent
Fc.uq = (Fa.uq<63:52> << 52) | Fb.uq<51:0>;
}});
0x02a: fcmoveq({{ Fc = (Fa == 0) ? Fb : Fc; }});
0x02b: fcmovne({{ Fc = (Fa != 0) ? Fb : Fc; }});
0x02c: fcmovlt({{ Fc = (Fa < 0) ? Fb : Fc; }});
0x02d: fcmovge({{ Fc = (Fa >= 0) ? Fb : Fc; }});
0x02e: fcmovle({{ Fc = (Fa <= 0) ? Fb : Fc; }});
0x02f: fcmovgt({{ Fc = (Fa > 0) ? Fb : Fc; }});
0x024: mt_fpcr({{ FPCR = Fa.uq; }});
0x025: mf_fpcr({{ Fa.uq = FPCR; }});
}
}
// miscellaneous mem-format ops
0x18: decode MEMFUNC {
format WarnUnimpl {
0x8000: fetch();
0xa000: fetch_m();
0xe800: ecb();
}
format MiscPrefetch {
0xf800: wh64({{ EA = Rb; }},
{{ xc->writeHint(EA, 64); }},
IsMemRef, IsStore, WrPort);
}
format BasicOperate {
0xc000: rpcc({{
#ifdef FULL_SYSTEM
Ra = xc->readIpr(AlphaISA::IPR_CC, fault);
#else
Ra = curTick;
#endif
}});
// All of the barrier instructions below do nothing in
// their execute() methods (hence the empty code blocks).
// All of their functionality is hard-coded in the
// pipeline based on the flags IsSerializing,
// IsMemBarrier, and IsWriteBarrier. In the current
// detailed CPU model, the execute() function only gets
// called at fetch, so there's no way to generate pipeline
// behavior at any other stage. Once we go to an
// exec-in-exec CPU model we should be able to get rid of
// these flags and implement this behavior via the
// execute() methods.
// trapb is just a barrier on integer traps, where excb is
// a barrier on integer and FP traps. "EXCB is thus a
// superset of TRAPB." (Alpha ARM, Sec 4.11.4) We treat
// them the same though.
0x0000: trapb({{ }}, IsSerializing, No_OpClass);
0x0400: excb({{ }}, IsSerializing, No_OpClass);
0x4000: mb({{ }}, IsMemBarrier, RdPort);
0x4400: wmb({{ }}, IsWriteBarrier, WrPort);
}
#ifdef FULL_SYSTEM
format BasicOperate {
0xe000: rc({{
Ra = xc->readIntrFlag();
if (!xc->misspeculating()) {
xc->setIntrFlag(0);
}
}});
0xf000: rs({{
Ra = xc->readIntrFlag();
if (!xc->misspeculating()) {
xc->setIntrFlag(1);
}
}});
}
#else
format FailUnimpl {
0xe000: rc();
0xf000: rs();
}
#endif
}
#ifdef FULL_SYSTEM
0x00: CallPal::call_pal({{
if (!palValid ||
(palPriv
&& xc->readIpr(AlphaISA::IPR_ICM, fault) != AlphaISA::mode_kernel)) {
// invalid pal function code, or attempt to do privileged
// PAL call in non-kernel mode
fault = Unimplemented_Opcode_Fault;
}
else {
bool dopal = true;
if (!xc->misspeculating()) {
// check to see if simulator wants to do something special
// on this PAL call (including maybe suppress it)
dopal = xc->simPalCheck(palFunc);
Annotate::Callpal(xc->xcBase(), palFunc);
if (dopal) {
AlphaISA::swap_palshadow(&xc->xcBase()->regs, true);
xc->setIpr(AlphaISA::IPR_EXC_ADDR, NPC);
}
}
// if we're misspeculating, it's still safe (if
// unrealistic) to set NPC, as the control-flow change
// won't get committed.
if (dopal) {
NPC = xc->readIpr(AlphaISA::IPR_PAL_BASE, fault) + palOffset;
}
}
}});
#else
0x00: decode PALFUNC {
format EmulatedCallPal {
0x00: halt ({{
if (!xc->misspeculating())
SimExit(curTick, "halt instruction encountered");
}});
0x83: callsys({{
if (!xc->misspeculating())
xc->syscall();
}});
// Read uniq reg into ABI return value register (r0)
0x9e: rduniq({{ R0 = Runiq; }});
// Write uniq reg with value from ABI arg register (r16)
0x9f: wruniq({{ Runiq = R16; }});
}
}
#endif
#ifdef FULL_SYSTEM
format HwLoadStore {
0x1b: decode HW_LDST_QUAD {
0: hw_ld({{ EA = (Rb + disp) & ~3; }}, {{ Ra = Mem.ul; }}, L);
1: hw_ld({{ EA = (Rb + disp) & ~7; }}, {{ Ra = Mem.uq; }}, Q);
}
0x1f: decode HW_LDST_COND {
0: decode HW_LDST_QUAD {
0: hw_st({{ EA = (Rb + disp) & ~3; }},
{{ Mem.ul = Ra<31:0>; }}, L);
1: hw_st({{ EA = (Rb + disp) & ~7; }},
{{ Mem.uq = Ra.uq; }}, Q);
}
1: FailUnimpl::hw_st_cond();
}
}
format BasicOperate {
0x1e: hw_rei({{ xc->hwrei(); }});
// M5 special opcodes use the reserved 0x01 opcode space
0x01: decode M5FUNC {
0x00: arm({{
if (!xc->misspeculating()) {
Annotate::ARM(xc->xcBase());
xc->xcBase()->kernelStats.arm();
}
}});
0x01: quiesce({{
if (!xc->misspeculating())
AlphaPseudo::quiesce(xc->xcBase());
}});
0x10: ivlb({{
if (!xc->misspeculating()) {
Annotate::BeginInterval(xc->xcBase());
xc->xcBase()->kernelStats.ivlb();
}
}}, No_OpClass);
0x11: ivle({{
if (!xc->misspeculating())
Annotate::EndInterval(xc->xcBase());
}}, No_OpClass);
0x20: m5exit_old({{
if (!xc->misspeculating())
AlphaPseudo::m5exit_old(xc->xcBase());
}}, No_OpClass);
0x21: m5exit({{
if (!xc->misspeculating())
AlphaPseudo::m5exit(xc->xcBase());
}}, No_OpClass);
0x30: initparam({{ Ra = xc->xcBase()->cpu->system->init_param; }});
0x40: resetstats({{
if (!xc->misspeculating())
AlphaPseudo::resetstats(xc->xcBase());
}});
0x41: dumpstats({{
if (!xc->misspeculating())
AlphaPseudo::dumpstats(xc->xcBase());
}});
0x42: dumpresetstats({{
if (!xc->misspeculating())
AlphaPseudo::dumpresetstats(xc->xcBase());
}});
0x43: m5checkpoint({{
if (!xc->misspeculating())
AlphaPseudo::m5checkpoint(xc->xcBase());
}});
}
}
format HwMoveIPR {
0x19: hw_mfpr({{
// this instruction is only valid in PAL mode
if (!xc->inPalMode()) {
fault = Unimplemented_Opcode_Fault;
}
else {
Ra = xc->readIpr(ipr_index, fault);
}
}});
0x1d: hw_mtpr({{
// this instruction is only valid in PAL mode
if (!xc->inPalMode()) {
fault = Unimplemented_Opcode_Fault;
}
else {
xc->setIpr(ipr_index, Ra);
if (traceData) { traceData->setData(Ra); }
}
}});
}
#endif
}
|