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
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
|
//===------- LegalizeVectorTypes.cpp - Legalization of vector types -------===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file performs vector type splitting and scalarization for LegalizeTypes.
// Scalarization is the act of changing a computation in an illegal one-element
// vector type to be a computation in its scalar element type. For example,
// implementing <1 x f32> arithmetic in a scalar f32 register. This is needed
// as a base case when scalarizing vector arithmetic like <4 x f32>, which
// eventually decomposes to scalars if the target doesn't support v4f32 or v2f32
// types.
// Splitting is the act of changing a computation in an invalid vector type to
// be a computation in two vectors of half the size. For example, implementing
// <128 x f32> operations in terms of two <64 x f32> operations.
//
//===----------------------------------------------------------------------===//
#include "LegalizeTypes.h"
#include "llvm/Target/TargetData.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/raw_ostream.h"
using namespace llvm;
//===----------------------------------------------------------------------===//
// Result Vector Scalarization: <1 x ty> -> ty.
//===----------------------------------------------------------------------===//
void DAGTypeLegalizer::ScalarizeVectorResult(SDNode *N, unsigned ResNo) {
DEBUG(dbgs() << "Scalarize node result " << ResNo << ": ";
N->dump(&DAG);
dbgs() << "\n");
SDValue R = SDValue();
switch (N->getOpcode()) {
default:
#ifndef NDEBUG
dbgs() << "ScalarizeVectorResult #" << ResNo << ": ";
N->dump(&DAG);
dbgs() << "\n";
#endif
report_fatal_error("Do not know how to scalarize the result of this "
"operator!\n");
case ISD::MERGE_VALUES: R = ScalarizeVecRes_MERGE_VALUES(N, ResNo);break;
case ISD::BITCAST: R = ScalarizeVecRes_BITCAST(N); break;
case ISD::BUILD_VECTOR: R = N->getOperand(0); break;
case ISD::CONVERT_RNDSAT: R = ScalarizeVecRes_CONVERT_RNDSAT(N); break;
case ISD::EXTRACT_SUBVECTOR: R = ScalarizeVecRes_EXTRACT_SUBVECTOR(N); break;
case ISD::FP_ROUND: R = ScalarizeVecRes_FP_ROUND(N); break;
case ISD::FP_ROUND_INREG: R = ScalarizeVecRes_InregOp(N); break;
case ISD::FPOWI: R = ScalarizeVecRes_FPOWI(N); break;
case ISD::INSERT_VECTOR_ELT: R = ScalarizeVecRes_INSERT_VECTOR_ELT(N); break;
case ISD::LOAD: R = ScalarizeVecRes_LOAD(cast<LoadSDNode>(N));break;
case ISD::SCALAR_TO_VECTOR: R = ScalarizeVecRes_SCALAR_TO_VECTOR(N); break;
case ISD::SIGN_EXTEND_INREG: R = ScalarizeVecRes_InregOp(N); break;
case ISD::SELECT: R = ScalarizeVecRes_SELECT(N); break;
case ISD::SELECT_CC: R = ScalarizeVecRes_SELECT_CC(N); break;
case ISD::SETCC: R = ScalarizeVecRes_SETCC(N); break;
case ISD::UNDEF: R = ScalarizeVecRes_UNDEF(N); break;
case ISD::VECTOR_SHUFFLE: R = ScalarizeVecRes_VECTOR_SHUFFLE(N); break;
case ISD::ANY_EXTEND:
case ISD::CTLZ:
case ISD::CTPOP:
case ISD::CTTZ:
case ISD::FABS:
case ISD::FCEIL:
case ISD::FCOS:
case ISD::FEXP:
case ISD::FEXP2:
case ISD::FFLOOR:
case ISD::FLOG:
case ISD::FLOG10:
case ISD::FLOG2:
case ISD::FNEARBYINT:
case ISD::FNEG:
case ISD::FP_EXTEND:
case ISD::FP_TO_SINT:
case ISD::FP_TO_UINT:
case ISD::FRINT:
case ISD::FSIN:
case ISD::FSQRT:
case ISD::FTRUNC:
case ISD::SIGN_EXTEND:
case ISD::SINT_TO_FP:
case ISD::TRUNCATE:
case ISD::UINT_TO_FP:
case ISD::ZERO_EXTEND:
R = ScalarizeVecRes_UnaryOp(N);
break;
case ISD::ADD:
case ISD::AND:
case ISD::FADD:
case ISD::FDIV:
case ISD::FMUL:
case ISD::FPOW:
case ISD::FREM:
case ISD::FSUB:
case ISD::MUL:
case ISD::OR:
case ISD::SDIV:
case ISD::SREM:
case ISD::SUB:
case ISD::UDIV:
case ISD::UREM:
case ISD::XOR:
case ISD::SHL:
case ISD::SRA:
case ISD::SRL:
R = ScalarizeVecRes_BinOp(N);
break;
}
// If R is null, the sub-method took care of registering the result.
if (R.getNode())
SetScalarizedVector(SDValue(N, ResNo), R);
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_BinOp(SDNode *N) {
SDValue LHS = GetScalarizedVector(N->getOperand(0));
SDValue RHS = GetScalarizedVector(N->getOperand(1));
return DAG.getNode(N->getOpcode(), N->getDebugLoc(),
LHS.getValueType(), LHS, RHS);
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_MERGE_VALUES(SDNode *N,
unsigned ResNo) {
SDValue Op = DisintegrateMERGE_VALUES(N, ResNo);
return GetScalarizedVector(Op);
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_BITCAST(SDNode *N) {
EVT NewVT = N->getValueType(0).getVectorElementType();
return DAG.getNode(ISD::BITCAST, N->getDebugLoc(),
NewVT, N->getOperand(0));
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_CONVERT_RNDSAT(SDNode *N) {
EVT NewVT = N->getValueType(0).getVectorElementType();
SDValue Op0 = GetScalarizedVector(N->getOperand(0));
return DAG.getConvertRndSat(NewVT, N->getDebugLoc(),
Op0, DAG.getValueType(NewVT),
DAG.getValueType(Op0.getValueType()),
N->getOperand(3),
N->getOperand(4),
cast<CvtRndSatSDNode>(N)->getCvtCode());
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_EXTRACT_SUBVECTOR(SDNode *N) {
return DAG.getNode(ISD::EXTRACT_VECTOR_ELT, N->getDebugLoc(),
N->getValueType(0).getVectorElementType(),
N->getOperand(0), N->getOperand(1));
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_FP_ROUND(SDNode *N) {
EVT NewVT = N->getValueType(0).getVectorElementType();
SDValue Op = GetScalarizedVector(N->getOperand(0));
return DAG.getNode(ISD::FP_ROUND, N->getDebugLoc(),
NewVT, Op, N->getOperand(1));
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_FPOWI(SDNode *N) {
SDValue Op = GetScalarizedVector(N->getOperand(0));
return DAG.getNode(ISD::FPOWI, N->getDebugLoc(),
Op.getValueType(), Op, N->getOperand(1));
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_INSERT_VECTOR_ELT(SDNode *N) {
// The value to insert may have a wider type than the vector element type,
// so be sure to truncate it to the element type if necessary.
SDValue Op = N->getOperand(1);
EVT EltVT = N->getValueType(0).getVectorElementType();
if (Op.getValueType() != EltVT)
// FIXME: Can this happen for floating point types?
Op = DAG.getNode(ISD::TRUNCATE, N->getDebugLoc(), EltVT, Op);
return Op;
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_LOAD(LoadSDNode *N) {
assert(N->isUnindexed() && "Indexed vector load?");
SDValue Result = DAG.getLoad(ISD::UNINDEXED,
N->getExtensionType(),
N->getValueType(0).getVectorElementType(),
N->getDebugLoc(),
N->getChain(), N->getBasePtr(),
DAG.getUNDEF(N->getBasePtr().getValueType()),
N->getPointerInfo(),
N->getMemoryVT().getVectorElementType(),
N->isVolatile(), N->isNonTemporal(),
N->isInvariant(), N->getOriginalAlignment());
// Legalized the chain result - switch anything that used the old chain to
// use the new one.
ReplaceValueWith(SDValue(N, 1), Result.getValue(1));
return Result;
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_UnaryOp(SDNode *N) {
// Get the dest type - it doesn't always match the input type, e.g. int_to_fp.
EVT DestVT = N->getValueType(0).getVectorElementType();
SDValue Op = GetScalarizedVector(N->getOperand(0));
return DAG.getNode(N->getOpcode(), N->getDebugLoc(), DestVT, Op);
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_InregOp(SDNode *N) {
EVT EltVT = N->getValueType(0).getVectorElementType();
EVT ExtVT = cast<VTSDNode>(N->getOperand(1))->getVT().getVectorElementType();
SDValue LHS = GetScalarizedVector(N->getOperand(0));
return DAG.getNode(N->getOpcode(), N->getDebugLoc(), EltVT,
LHS, DAG.getValueType(ExtVT));
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_SCALAR_TO_VECTOR(SDNode *N) {
// If the operand is wider than the vector element type then it is implicitly
// truncated. Make that explicit here.
EVT EltVT = N->getValueType(0).getVectorElementType();
SDValue InOp = N->getOperand(0);
if (InOp.getValueType() != EltVT)
return DAG.getNode(ISD::TRUNCATE, N->getDebugLoc(), EltVT, InOp);
return InOp;
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_SELECT(SDNode *N) {
SDValue LHS = GetScalarizedVector(N->getOperand(1));
return DAG.getNode(ISD::SELECT, N->getDebugLoc(),
LHS.getValueType(), N->getOperand(0), LHS,
GetScalarizedVector(N->getOperand(2)));
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_SELECT_CC(SDNode *N) {
SDValue LHS = GetScalarizedVector(N->getOperand(2));
return DAG.getNode(ISD::SELECT_CC, N->getDebugLoc(), LHS.getValueType(),
N->getOperand(0), N->getOperand(1),
LHS, GetScalarizedVector(N->getOperand(3)),
N->getOperand(4));
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_SETCC(SDNode *N) {
assert(N->getValueType(0).isVector() ==
N->getOperand(0).getValueType().isVector() &&
"Scalar/Vector type mismatch");
if (N->getValueType(0).isVector()) return ScalarizeVecRes_VSETCC(N);
SDValue LHS = GetScalarizedVector(N->getOperand(0));
SDValue RHS = GetScalarizedVector(N->getOperand(1));
DebugLoc DL = N->getDebugLoc();
// Turn it into a scalar SETCC.
return DAG.getNode(ISD::SETCC, DL, MVT::i1, LHS, RHS, N->getOperand(2));
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_UNDEF(SDNode *N) {
return DAG.getUNDEF(N->getValueType(0).getVectorElementType());
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_VECTOR_SHUFFLE(SDNode *N) {
// Figure out if the scalar is the LHS or RHS and return it.
SDValue Arg = N->getOperand(2).getOperand(0);
if (Arg.getOpcode() == ISD::UNDEF)
return DAG.getUNDEF(N->getValueType(0).getVectorElementType());
unsigned Op = !cast<ConstantSDNode>(Arg)->isNullValue();
return GetScalarizedVector(N->getOperand(Op));
}
SDValue DAGTypeLegalizer::ScalarizeVecRes_VSETCC(SDNode *N) {
assert(N->getValueType(0).isVector() &&
N->getOperand(0).getValueType().isVector() &&
"Operand types must be vectors");
SDValue LHS = GetScalarizedVector(N->getOperand(0));
SDValue RHS = GetScalarizedVector(N->getOperand(1));
EVT NVT = N->getValueType(0).getVectorElementType();
DebugLoc DL = N->getDebugLoc();
// Turn it into a scalar SETCC.
SDValue Res = DAG.getNode(ISD::SETCC, DL, MVT::i1, LHS, RHS,
N->getOperand(2));
// Vectors may have a different boolean contents to scalars. Promote the
// value appropriately.
ISD::NodeType ExtendCode =
TargetLowering::getExtendForContent(TLI.getBooleanContents(true));
return DAG.getNode(ExtendCode, DL, NVT, Res);
}
//===----------------------------------------------------------------------===//
// Operand Vector Scalarization <1 x ty> -> ty.
//===----------------------------------------------------------------------===//
bool DAGTypeLegalizer::ScalarizeVectorOperand(SDNode *N, unsigned OpNo) {
DEBUG(dbgs() << "Scalarize node operand " << OpNo << ": ";
N->dump(&DAG);
dbgs() << "\n");
SDValue Res = SDValue();
if (Res.getNode() == 0) {
switch (N->getOpcode()) {
default:
#ifndef NDEBUG
dbgs() << "ScalarizeVectorOperand Op #" << OpNo << ": ";
N->dump(&DAG);
dbgs() << "\n";
#endif
llvm_unreachable("Do not know how to scalarize this operator's operand!");
case ISD::BITCAST:
Res = ScalarizeVecOp_BITCAST(N);
break;
case ISD::CONCAT_VECTORS:
Res = ScalarizeVecOp_CONCAT_VECTORS(N);
break;
case ISD::EXTRACT_VECTOR_ELT:
Res = ScalarizeVecOp_EXTRACT_VECTOR_ELT(N);
break;
case ISD::STORE:
Res = ScalarizeVecOp_STORE(cast<StoreSDNode>(N), OpNo);
break;
}
}
// If the result is null, the sub-method took care of registering results etc.
if (!Res.getNode()) return false;
// If the result is N, the sub-method updated N in place. Tell the legalizer
// core about this.
if (Res.getNode() == N)
return true;
assert(Res.getValueType() == N->getValueType(0) && N->getNumValues() == 1 &&
"Invalid operand expansion");
ReplaceValueWith(SDValue(N, 0), Res);
return false;
}
/// ScalarizeVecOp_BITCAST - If the value to convert is a vector that needs
/// to be scalarized, it must be <1 x ty>. Convert the element instead.
SDValue DAGTypeLegalizer::ScalarizeVecOp_BITCAST(SDNode *N) {
SDValue Elt = GetScalarizedVector(N->getOperand(0));
return DAG.getNode(ISD::BITCAST, N->getDebugLoc(),
N->getValueType(0), Elt);
}
/// ScalarizeVecOp_CONCAT_VECTORS - The vectors to concatenate have length one -
/// use a BUILD_VECTOR instead.
SDValue DAGTypeLegalizer::ScalarizeVecOp_CONCAT_VECTORS(SDNode *N) {
SmallVector<SDValue, 8> Ops(N->getNumOperands());
for (unsigned i = 0, e = N->getNumOperands(); i < e; ++i)
Ops[i] = GetScalarizedVector(N->getOperand(i));
return DAG.getNode(ISD::BUILD_VECTOR, N->getDebugLoc(), N->getValueType(0),
&Ops[0], Ops.size());
}
/// ScalarizeVecOp_EXTRACT_VECTOR_ELT - If the input is a vector that needs to
/// be scalarized, it must be <1 x ty>, so just return the element, ignoring the
/// index.
SDValue DAGTypeLegalizer::ScalarizeVecOp_EXTRACT_VECTOR_ELT(SDNode *N) {
SDValue Res = GetScalarizedVector(N->getOperand(0));
if (Res.getValueType() != N->getValueType(0))
Res = DAG.getNode(ISD::ANY_EXTEND, N->getDebugLoc(), N->getValueType(0),
Res);
return Res;
}
/// ScalarizeVecOp_STORE - If the value to store is a vector that needs to be
/// scalarized, it must be <1 x ty>. Just store the element.
SDValue DAGTypeLegalizer::ScalarizeVecOp_STORE(StoreSDNode *N, unsigned OpNo){
assert(N->isUnindexed() && "Indexed store of one-element vector?");
assert(OpNo == 1 && "Do not know how to scalarize this operand!");
DebugLoc dl = N->getDebugLoc();
if (N->isTruncatingStore())
return DAG.getTruncStore(N->getChain(), dl,
GetScalarizedVector(N->getOperand(1)),
N->getBasePtr(), N->getPointerInfo(),
N->getMemoryVT().getVectorElementType(),
N->isVolatile(), N->isNonTemporal(),
N->getAlignment());
return DAG.getStore(N->getChain(), dl, GetScalarizedVector(N->getOperand(1)),
N->getBasePtr(), N->getPointerInfo(),
N->isVolatile(), N->isNonTemporal(),
N->getOriginalAlignment());
}
//===----------------------------------------------------------------------===//
// Result Vector Splitting
//===----------------------------------------------------------------------===//
/// SplitVectorResult - This method is called when the specified result of the
/// specified node is found to need vector splitting. At this point, the node
/// may also have invalid operands or may have other results that need
/// legalization, we just know that (at least) one result needs vector
/// splitting.
void DAGTypeLegalizer::SplitVectorResult(SDNode *N, unsigned ResNo) {
DEBUG(dbgs() << "Split node result: ";
N->dump(&DAG);
dbgs() << "\n");
SDValue Lo, Hi;
switch (N->getOpcode()) {
default:
#ifndef NDEBUG
dbgs() << "SplitVectorResult #" << ResNo << ": ";
N->dump(&DAG);
dbgs() << "\n";
#endif
llvm_unreachable("Do not know how to split the result of this operator!");
case ISD::MERGE_VALUES: SplitRes_MERGE_VALUES(N, ResNo, Lo, Hi); break;
case ISD::VSELECT:
case ISD::SELECT: SplitRes_SELECT(N, Lo, Hi); break;
case ISD::SELECT_CC: SplitRes_SELECT_CC(N, Lo, Hi); break;
case ISD::UNDEF: SplitRes_UNDEF(N, Lo, Hi); break;
case ISD::BITCAST: SplitVecRes_BITCAST(N, Lo, Hi); break;
case ISD::BUILD_VECTOR: SplitVecRes_BUILD_VECTOR(N, Lo, Hi); break;
case ISD::CONCAT_VECTORS: SplitVecRes_CONCAT_VECTORS(N, Lo, Hi); break;
case ISD::EXTRACT_SUBVECTOR: SplitVecRes_EXTRACT_SUBVECTOR(N, Lo, Hi); break;
case ISD::FP_ROUND_INREG: SplitVecRes_InregOp(N, Lo, Hi); break;
case ISD::FPOWI: SplitVecRes_FPOWI(N, Lo, Hi); break;
case ISD::INSERT_VECTOR_ELT: SplitVecRes_INSERT_VECTOR_ELT(N, Lo, Hi); break;
case ISD::SCALAR_TO_VECTOR: SplitVecRes_SCALAR_TO_VECTOR(N, Lo, Hi); break;
case ISD::SIGN_EXTEND_INREG: SplitVecRes_InregOp(N, Lo, Hi); break;
case ISD::LOAD:
SplitVecRes_LOAD(cast<LoadSDNode>(N), Lo, Hi);
break;
case ISD::SETCC:
SplitVecRes_SETCC(N, Lo, Hi);
break;
case ISD::VECTOR_SHUFFLE:
SplitVecRes_VECTOR_SHUFFLE(cast<ShuffleVectorSDNode>(N), Lo, Hi);
break;
case ISD::ANY_EXTEND:
case ISD::CONVERT_RNDSAT:
case ISD::CTLZ:
case ISD::CTPOP:
case ISD::CTTZ:
case ISD::FABS:
case ISD::FCEIL:
case ISD::FCOS:
case ISD::FEXP:
case ISD::FEXP2:
case ISD::FFLOOR:
case ISD::FLOG:
case ISD::FLOG10:
case ISD::FLOG2:
case ISD::FNEARBYINT:
case ISD::FNEG:
case ISD::FP_EXTEND:
case ISD::FP_ROUND:
case ISD::FP_TO_SINT:
case ISD::FP_TO_UINT:
case ISD::FRINT:
case ISD::FSIN:
case ISD::FSQRT:
case ISD::FTRUNC:
case ISD::SIGN_EXTEND:
case ISD::SINT_TO_FP:
case ISD::TRUNCATE:
case ISD::UINT_TO_FP:
case ISD::ZERO_EXTEND:
SplitVecRes_UnaryOp(N, Lo, Hi);
break;
case ISD::ADD:
case ISD::SUB:
case ISD::MUL:
case ISD::FADD:
case ISD::FSUB:
case ISD::FMUL:
case ISD::SDIV:
case ISD::UDIV:
case ISD::FDIV:
case ISD::FPOW:
case ISD::AND:
case ISD::OR:
case ISD::XOR:
case ISD::SHL:
case ISD::SRA:
case ISD::SRL:
case ISD::UREM:
case ISD::SREM:
case ISD::FREM:
SplitVecRes_BinOp(N, Lo, Hi);
break;
}
// If Lo/Hi is null, the sub-method took care of registering results etc.
if (Lo.getNode())
SetSplitVector(SDValue(N, ResNo), Lo, Hi);
}
void DAGTypeLegalizer::SplitVecRes_BinOp(SDNode *N, SDValue &Lo,
SDValue &Hi) {
SDValue LHSLo, LHSHi;
GetSplitVector(N->getOperand(0), LHSLo, LHSHi);
SDValue RHSLo, RHSHi;
GetSplitVector(N->getOperand(1), RHSLo, RHSHi);
DebugLoc dl = N->getDebugLoc();
Lo = DAG.getNode(N->getOpcode(), dl, LHSLo.getValueType(), LHSLo, RHSLo);
Hi = DAG.getNode(N->getOpcode(), dl, LHSHi.getValueType(), LHSHi, RHSHi);
}
void DAGTypeLegalizer::SplitVecRes_BITCAST(SDNode *N, SDValue &Lo,
SDValue &Hi) {
// We know the result is a vector. The input may be either a vector or a
// scalar value.
EVT LoVT, HiVT;
GetSplitDestVTs(N->getValueType(0), LoVT, HiVT);
DebugLoc dl = N->getDebugLoc();
SDValue InOp = N->getOperand(0);
EVT InVT = InOp.getValueType();
// Handle some special cases efficiently.
switch (getTypeAction(InVT)) {
case TargetLowering::TypeLegal:
case TargetLowering::TypePromoteInteger:
case TargetLowering::TypeSoftenFloat:
case TargetLowering::TypeScalarizeVector:
case TargetLowering::TypeWidenVector:
break;
case TargetLowering::TypeExpandInteger:
case TargetLowering::TypeExpandFloat:
// A scalar to vector conversion, where the scalar needs expansion.
// If the vector is being split in two then we can just convert the
// expanded pieces.
if (LoVT == HiVT) {
GetExpandedOp(InOp, Lo, Hi);
if (TLI.isBigEndian())
std::swap(Lo, Hi);
Lo = DAG.getNode(ISD::BITCAST, dl, LoVT, Lo);
Hi = DAG.getNode(ISD::BITCAST, dl, HiVT, Hi);
return;
}
break;
case TargetLowering::TypeSplitVector:
// If the input is a vector that needs to be split, convert each split
// piece of the input now.
GetSplitVector(InOp, Lo, Hi);
Lo = DAG.getNode(ISD::BITCAST, dl, LoVT, Lo);
Hi = DAG.getNode(ISD::BITCAST, dl, HiVT, Hi);
return;
}
// In the general case, convert the input to an integer and split it by hand.
EVT LoIntVT = EVT::getIntegerVT(*DAG.getContext(), LoVT.getSizeInBits());
EVT HiIntVT = EVT::getIntegerVT(*DAG.getContext(), HiVT.getSizeInBits());
if (TLI.isBigEndian())
std::swap(LoIntVT, HiIntVT);
SplitInteger(BitConvertToInteger(InOp), LoIntVT, HiIntVT, Lo, Hi);
if (TLI.isBigEndian())
std::swap(Lo, Hi);
Lo = DAG.getNode(ISD::BITCAST, dl, LoVT, Lo);
Hi = DAG.getNode(ISD::BITCAST, dl, HiVT, Hi);
}
void DAGTypeLegalizer::SplitVecRes_BUILD_VECTOR(SDNode *N, SDValue &Lo,
SDValue &Hi) {
EVT LoVT, HiVT;
DebugLoc dl = N->getDebugLoc();
GetSplitDestVTs(N->getValueType(0), LoVT, HiVT);
unsigned LoNumElts = LoVT.getVectorNumElements();
SmallVector<SDValue, 8> LoOps(N->op_begin(), N->op_begin()+LoNumElts);
Lo = DAG.getNode(ISD::BUILD_VECTOR, dl, LoVT, &LoOps[0], LoOps.size());
SmallVector<SDValue, 8> HiOps(N->op_begin()+LoNumElts, N->op_end());
Hi = DAG.getNode(ISD::BUILD_VECTOR, dl, HiVT, &HiOps[0], HiOps.size());
}
void DAGTypeLegalizer::SplitVecRes_CONCAT_VECTORS(SDNode *N, SDValue &Lo,
SDValue &Hi) {
assert(!(N->getNumOperands() & 1) && "Unsupported CONCAT_VECTORS");
DebugLoc dl = N->getDebugLoc();
unsigned NumSubvectors = N->getNumOperands() / 2;
if (NumSubvectors == 1) {
Lo = N->getOperand(0);
Hi = N->getOperand(1);
return;
}
EVT LoVT, HiVT;
GetSplitDestVTs(N->getValueType(0), LoVT, HiVT);
SmallVector<SDValue, 8> LoOps(N->op_begin(), N->op_begin()+NumSubvectors);
Lo = DAG.getNode(ISD::CONCAT_VECTORS, dl, LoVT, &LoOps[0], LoOps.size());
SmallVector<SDValue, 8> HiOps(N->op_begin()+NumSubvectors, N->op_end());
Hi = DAG.getNode(ISD::CONCAT_VECTORS, dl, HiVT, &HiOps[0], HiOps.size());
}
void DAGTypeLegalizer::SplitVecRes_EXTRACT_SUBVECTOR(SDNode *N, SDValue &Lo,
SDValue &Hi) {
SDValue Vec = N->getOperand(0);
SDValue Idx = N->getOperand(1);
DebugLoc dl = N->getDebugLoc();
EVT LoVT, HiVT;
GetSplitDestVTs(N->getValueType(0), LoVT, HiVT);
Lo = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, LoVT, Vec, Idx);
uint64_t IdxVal = cast<ConstantSDNode>(Idx)->getZExtValue();
Hi = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, HiVT, Vec,
DAG.getIntPtrConstant(IdxVal + LoVT.getVectorNumElements()));
}
void DAGTypeLegalizer::SplitVecRes_FPOWI(SDNode *N, SDValue &Lo,
SDValue &Hi) {
DebugLoc dl = N->getDebugLoc();
GetSplitVector(N->getOperand(0), Lo, Hi);
Lo = DAG.getNode(ISD::FPOWI, dl, Lo.getValueType(), Lo, N->getOperand(1));
Hi = DAG.getNode(ISD::FPOWI, dl, Hi.getValueType(), Hi, N->getOperand(1));
}
void DAGTypeLegalizer::SplitVecRes_InregOp(SDNode *N, SDValue &Lo,
SDValue &Hi) {
SDValue LHSLo, LHSHi;
GetSplitVector(N->getOperand(0), LHSLo, LHSHi);
DebugLoc dl = N->getDebugLoc();
EVT LoVT, HiVT;
GetSplitDestVTs(cast<VTSDNode>(N->getOperand(1))->getVT(), LoVT, HiVT);
Lo = DAG.getNode(N->getOpcode(), dl, LHSLo.getValueType(), LHSLo,
DAG.getValueType(LoVT));
Hi = DAG.getNode(N->getOpcode(), dl, LHSHi.getValueType(), LHSHi,
DAG.getValueType(HiVT));
}
void DAGTypeLegalizer::SplitVecRes_INSERT_VECTOR_ELT(SDNode *N, SDValue &Lo,
SDValue &Hi) {
SDValue Vec = N->getOperand(0);
SDValue Elt = N->getOperand(1);
SDValue Idx = N->getOperand(2);
DebugLoc dl = N->getDebugLoc();
GetSplitVector(Vec, Lo, Hi);
if (ConstantSDNode *CIdx = dyn_cast<ConstantSDNode>(Idx)) {
unsigned IdxVal = CIdx->getZExtValue();
unsigned LoNumElts = Lo.getValueType().getVectorNumElements();
if (IdxVal < LoNumElts)
Lo = DAG.getNode(ISD::INSERT_VECTOR_ELT, dl,
Lo.getValueType(), Lo, Elt, Idx);
else
Hi = DAG.getNode(ISD::INSERT_VECTOR_ELT, dl, Hi.getValueType(), Hi, Elt,
DAG.getIntPtrConstant(IdxVal - LoNumElts));
return;
}
// Spill the vector to the stack.
EVT VecVT = Vec.getValueType();
EVT EltVT = VecVT.getVectorElementType();
SDValue StackPtr = DAG.CreateStackTemporary(VecVT);
SDValue Store = DAG.getStore(DAG.getEntryNode(), dl, Vec, StackPtr,
MachinePointerInfo(), false, false, 0);
// Store the new element. This may be larger than the vector element type,
// so use a truncating store.
SDValue EltPtr = GetVectorElementPointer(StackPtr, EltVT, Idx);
Type *VecType = VecVT.getTypeForEVT(*DAG.getContext());
unsigned Alignment =
TLI.getTargetData()->getPrefTypeAlignment(VecType);
Store = DAG.getTruncStore(Store, dl, Elt, EltPtr, MachinePointerInfo(), EltVT,
false, false, 0);
// Load the Lo part from the stack slot.
Lo = DAG.getLoad(Lo.getValueType(), dl, Store, StackPtr, MachinePointerInfo(),
false, false, false, 0);
// Increment the pointer to the other part.
unsigned IncrementSize = Lo.getValueType().getSizeInBits() / 8;
StackPtr = DAG.getNode(ISD::ADD, dl, StackPtr.getValueType(), StackPtr,
DAG.getIntPtrConstant(IncrementSize));
// Load the Hi part from the stack slot.
Hi = DAG.getLoad(Hi.getValueType(), dl, Store, StackPtr, MachinePointerInfo(),
false, false, false, MinAlign(Alignment, IncrementSize));
}
void DAGTypeLegalizer::SplitVecRes_SCALAR_TO_VECTOR(SDNode *N, SDValue &Lo,
SDValue &Hi) {
EVT LoVT, HiVT;
DebugLoc dl = N->getDebugLoc();
GetSplitDestVTs(N->getValueType(0), LoVT, HiVT);
Lo = DAG.getNode(ISD::SCALAR_TO_VECTOR, dl, LoVT, N->getOperand(0));
Hi = DAG.getUNDEF(HiVT);
}
void DAGTypeLegalizer::SplitVecRes_LOAD(LoadSDNode *LD, SDValue &Lo,
SDValue &Hi) {
assert(ISD::isUNINDEXEDLoad(LD) && "Indexed load during type legalization!");
EVT LoVT, HiVT;
DebugLoc dl = LD->getDebugLoc();
GetSplitDestVTs(LD->getValueType(0), LoVT, HiVT);
ISD::LoadExtType ExtType = LD->getExtensionType();
SDValue Ch = LD->getChain();
SDValue Ptr = LD->getBasePtr();
SDValue Offset = DAG.getUNDEF(Ptr.getValueType());
EVT MemoryVT = LD->getMemoryVT();
unsigned Alignment = LD->getOriginalAlignment();
bool isVolatile = LD->isVolatile();
bool isNonTemporal = LD->isNonTemporal();
bool isInvariant = LD->isInvariant();
EVT LoMemVT, HiMemVT;
GetSplitDestVTs(MemoryVT, LoMemVT, HiMemVT);
Lo = DAG.getLoad(ISD::UNINDEXED, ExtType, LoVT, dl, Ch, Ptr, Offset,
LD->getPointerInfo(), LoMemVT, isVolatile, isNonTemporal,
isInvariant, Alignment);
unsigned IncrementSize = LoMemVT.getSizeInBits()/8;
Ptr = DAG.getNode(ISD::ADD, dl, Ptr.getValueType(), Ptr,
DAG.getIntPtrConstant(IncrementSize));
Hi = DAG.getLoad(ISD::UNINDEXED, ExtType, HiVT, dl, Ch, Ptr, Offset,
LD->getPointerInfo().getWithOffset(IncrementSize),
HiMemVT, isVolatile, isNonTemporal, isInvariant, Alignment);
// Build a factor node to remember that this load is independent of the
// other one.
Ch = DAG.getNode(ISD::TokenFactor, dl, MVT::Other, Lo.getValue(1),
Hi.getValue(1));
// Legalized the chain result - switch anything that used the old chain to
// use the new one.
ReplaceValueWith(SDValue(LD, 1), Ch);
}
void DAGTypeLegalizer::SplitVecRes_SETCC(SDNode *N, SDValue &Lo, SDValue &Hi) {
assert(N->getValueType(0).isVector() &&
N->getOperand(0).getValueType().isVector() &&
"Operand types must be vectors");
EVT LoVT, HiVT;
DebugLoc DL = N->getDebugLoc();
GetSplitDestVTs(N->getValueType(0), LoVT, HiVT);
// Split the input.
EVT InVT = N->getOperand(0).getValueType();
SDValue LL, LH, RL, RH;
EVT InNVT = EVT::getVectorVT(*DAG.getContext(), InVT.getVectorElementType(),
LoVT.getVectorNumElements());
LL = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, InNVT, N->getOperand(0),
DAG.getIntPtrConstant(0));
LH = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, InNVT, N->getOperand(0),
DAG.getIntPtrConstant(InNVT.getVectorNumElements()));
RL = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, InNVT, N->getOperand(1),
DAG.getIntPtrConstant(0));
RH = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, InNVT, N->getOperand(1),
DAG.getIntPtrConstant(InNVT.getVectorNumElements()));
Lo = DAG.getNode(N->getOpcode(), DL, LoVT, LL, RL, N->getOperand(2));
Hi = DAG.getNode(N->getOpcode(), DL, HiVT, LH, RH, N->getOperand(2));
}
void DAGTypeLegalizer::SplitVecRes_UnaryOp(SDNode *N, SDValue &Lo,
SDValue &Hi) {
// Get the dest types - they may not match the input types, e.g. int_to_fp.
EVT LoVT, HiVT;
DebugLoc dl = N->getDebugLoc();
GetSplitDestVTs(N->getValueType(0), LoVT, HiVT);
// If the input also splits, handle it directly for a compile time speedup.
// Otherwise split it by hand.
EVT InVT = N->getOperand(0).getValueType();
if (getTypeAction(InVT) == TargetLowering::TypeSplitVector) {
GetSplitVector(N->getOperand(0), Lo, Hi);
} else {
EVT InNVT = EVT::getVectorVT(*DAG.getContext(), InVT.getVectorElementType(),
LoVT.getVectorNumElements());
Lo = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, InNVT, N->getOperand(0),
DAG.getIntPtrConstant(0));
Hi = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, InNVT, N->getOperand(0),
DAG.getIntPtrConstant(InNVT.getVectorNumElements()));
}
if (N->getOpcode() == ISD::FP_ROUND) {
Lo = DAG.getNode(N->getOpcode(), dl, LoVT, Lo, N->getOperand(1));
Hi = DAG.getNode(N->getOpcode(), dl, HiVT, Hi, N->getOperand(1));
} else if (N->getOpcode() == ISD::CONVERT_RNDSAT) {
SDValue DTyOpLo = DAG.getValueType(LoVT);
SDValue DTyOpHi = DAG.getValueType(HiVT);
SDValue STyOpLo = DAG.getValueType(Lo.getValueType());
SDValue STyOpHi = DAG.getValueType(Hi.getValueType());
SDValue RndOp = N->getOperand(3);
SDValue SatOp = N->getOperand(4);
ISD::CvtCode CvtCode = cast<CvtRndSatSDNode>(N)->getCvtCode();
Lo = DAG.getConvertRndSat(LoVT, dl, Lo, DTyOpLo, STyOpLo, RndOp, SatOp,
CvtCode);
Hi = DAG.getConvertRndSat(HiVT, dl, Hi, DTyOpHi, STyOpHi, RndOp, SatOp,
CvtCode);
} else {
Lo = DAG.getNode(N->getOpcode(), dl, LoVT, Lo);
Hi = DAG.getNode(N->getOpcode(), dl, HiVT, Hi);
}
}
void DAGTypeLegalizer::SplitVecRes_VECTOR_SHUFFLE(ShuffleVectorSDNode *N,
SDValue &Lo, SDValue &Hi) {
// The low and high parts of the original input give four input vectors.
SDValue Inputs[4];
DebugLoc dl = N->getDebugLoc();
GetSplitVector(N->getOperand(0), Inputs[0], Inputs[1]);
GetSplitVector(N->getOperand(1), Inputs[2], Inputs[3]);
EVT NewVT = Inputs[0].getValueType();
unsigned NewElts = NewVT.getVectorNumElements();
// If Lo or Hi uses elements from at most two of the four input vectors, then
// express it as a vector shuffle of those two inputs. Otherwise extract the
// input elements by hand and construct the Lo/Hi output using a BUILD_VECTOR.
SmallVector<int, 16> Ops;
for (unsigned High = 0; High < 2; ++High) {
SDValue &Output = High ? Hi : Lo;
// Build a shuffle mask for the output, discovering on the fly which
// input vectors to use as shuffle operands (recorded in InputUsed).
// If building a suitable shuffle vector proves too hard, then bail
// out with useBuildVector set.
unsigned InputUsed[2] = { -1U, -1U }; // Not yet discovered.
unsigned FirstMaskIdx = High * NewElts;
bool useBuildVector = false;
for (unsigned MaskOffset = 0; MaskOffset < NewElts; ++MaskOffset) {
// The mask element. This indexes into the input.
int Idx = N->getMaskElt(FirstMaskIdx + MaskOffset);
// The input vector this mask element indexes into.
unsigned Input = (unsigned)Idx / NewElts;
if (Input >= array_lengthof(Inputs)) {
// The mask element does not index into any input vector.
Ops.push_back(-1);
continue;
}
// Turn the index into an offset from the start of the input vector.
Idx -= Input * NewElts;
// Find or create a shuffle vector operand to hold this input.
unsigned OpNo;
for (OpNo = 0; OpNo < array_lengthof(InputUsed); ++OpNo) {
if (InputUsed[OpNo] == Input) {
// This input vector is already an operand.
break;
} else if (InputUsed[OpNo] == -1U) {
// Create a new operand for this input vector.
InputUsed[OpNo] = Input;
break;
}
}
if (OpNo >= array_lengthof(InputUsed)) {
// More than two input vectors used! Give up on trying to create a
// shuffle vector. Insert all elements into a BUILD_VECTOR instead.
useBuildVector = true;
break;
}
// Add the mask index for the new shuffle vector.
Ops.push_back(Idx + OpNo * NewElts);
}
if (useBuildVector) {
EVT EltVT = NewVT.getVectorElementType();
SmallVector<SDValue, 16> SVOps;
// Extract the input elements by hand.
for (unsigned MaskOffset = 0; MaskOffset < NewElts; ++MaskOffset) {
// The mask element. This indexes into the input.
int Idx = N->getMaskElt(FirstMaskIdx + MaskOffset);
// The input vector this mask element indexes into.
unsigned Input = (unsigned)Idx / NewElts;
if (Input >= array_lengthof(Inputs)) {
// The mask element is "undef" or indexes off the end of the input.
SVOps.push_back(DAG.getUNDEF(EltVT));
continue;
}
// Turn the index into an offset from the start of the input vector.
Idx -= Input * NewElts;
// Extract the vector element by hand.
SVOps.push_back(DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, EltVT,
Inputs[Input], DAG.getIntPtrConstant(Idx)));
}
// Construct the Lo/Hi output using a BUILD_VECTOR.
Output = DAG.getNode(ISD::BUILD_VECTOR,dl,NewVT, &SVOps[0], SVOps.size());
} else if (InputUsed[0] == -1U) {
// No input vectors were used! The result is undefined.
Output = DAG.getUNDEF(NewVT);
} else {
SDValue Op0 = Inputs[InputUsed[0]];
// If only one input was used, use an undefined vector for the other.
SDValue Op1 = InputUsed[1] == -1U ?
DAG.getUNDEF(NewVT) : Inputs[InputUsed[1]];
// At least one input vector was used. Create a new shuffle vector.
Output = DAG.getVectorShuffle(NewVT, dl, Op0, Op1, &Ops[0]);
}
Ops.clear();
}
}
//===----------------------------------------------------------------------===//
// Operand Vector Splitting
//===----------------------------------------------------------------------===//
/// SplitVectorOperand - This method is called when the specified operand of the
/// specified node is found to need vector splitting. At this point, all of the
/// result types of the node are known to be legal, but other operands of the
/// node may need legalization as well as the specified one.
bool DAGTypeLegalizer::SplitVectorOperand(SDNode *N, unsigned OpNo) {
DEBUG(dbgs() << "Split node operand: ";
N->dump(&DAG);
dbgs() << "\n");
SDValue Res = SDValue();
if (Res.getNode() == 0) {
switch (N->getOpcode()) {
default:
#ifndef NDEBUG
dbgs() << "SplitVectorOperand Op #" << OpNo << ": ";
N->dump(&DAG);
dbgs() << "\n";
#endif
llvm_unreachable("Do not know how to split this operator's operand!");
case ISD::SETCC: Res = SplitVecOp_VSETCC(N); break;
case ISD::BITCAST: Res = SplitVecOp_BITCAST(N); break;
case ISD::EXTRACT_SUBVECTOR: Res = SplitVecOp_EXTRACT_SUBVECTOR(N); break;
case ISD::EXTRACT_VECTOR_ELT:Res = SplitVecOp_EXTRACT_VECTOR_ELT(N); break;
case ISD::CONCAT_VECTORS: Res = SplitVecOp_CONCAT_VECTORS(N); break;
case ISD::FP_ROUND: Res = SplitVecOp_FP_ROUND(N); break;
case ISD::STORE:
Res = SplitVecOp_STORE(cast<StoreSDNode>(N), OpNo);
break;
case ISD::CTTZ:
case ISD::CTLZ:
case ISD::CTPOP:
case ISD::FP_EXTEND:
case ISD::FP_TO_SINT:
case ISD::FP_TO_UINT:
case ISD::SINT_TO_FP:
case ISD::UINT_TO_FP:
case ISD::FTRUNC:
case ISD::TRUNCATE:
case ISD::SIGN_EXTEND:
case ISD::ZERO_EXTEND:
case ISD::ANY_EXTEND:
Res = SplitVecOp_UnaryOp(N);
break;
}
}
// If the result is null, the sub-method took care of registering results etc.
if (!Res.getNode()) return false;
// If the result is N, the sub-method updated N in place. Tell the legalizer
// core about this.
if (Res.getNode() == N)
return true;
assert(Res.getValueType() == N->getValueType(0) && N->getNumValues() == 1 &&
"Invalid operand expansion");
ReplaceValueWith(SDValue(N, 0), Res);
return false;
}
SDValue DAGTypeLegalizer::SplitVecOp_UnaryOp(SDNode *N) {
// The result has a legal vector type, but the input needs splitting.
EVT ResVT = N->getValueType(0);
SDValue Lo, Hi;
DebugLoc dl = N->getDebugLoc();
GetSplitVector(N->getOperand(0), Lo, Hi);
EVT InVT = Lo.getValueType();
EVT OutVT = EVT::getVectorVT(*DAG.getContext(), ResVT.getVectorElementType(),
InVT.getVectorNumElements());
Lo = DAG.getNode(N->getOpcode(), dl, OutVT, Lo);
Hi = DAG.getNode(N->getOpcode(), dl, OutVT, Hi);
return DAG.getNode(ISD::CONCAT_VECTORS, dl, ResVT, Lo, Hi);
}
SDValue DAGTypeLegalizer::SplitVecOp_BITCAST(SDNode *N) {
// For example, i64 = BITCAST v4i16 on alpha. Typically the vector will
// end up being split all the way down to individual components. Convert the
// split pieces into integers and reassemble.
SDValue Lo, Hi;
GetSplitVector(N->getOperand(0), Lo, Hi);
Lo = BitConvertToInteger(Lo);
Hi = BitConvertToInteger(Hi);
if (TLI.isBigEndian())
std::swap(Lo, Hi);
return DAG.getNode(ISD::BITCAST, N->getDebugLoc(), N->getValueType(0),
JoinIntegers(Lo, Hi));
}
SDValue DAGTypeLegalizer::SplitVecOp_EXTRACT_SUBVECTOR(SDNode *N) {
// We know that the extracted result type is legal.
EVT SubVT = N->getValueType(0);
SDValue Idx = N->getOperand(1);
DebugLoc dl = N->getDebugLoc();
SDValue Lo, Hi;
GetSplitVector(N->getOperand(0), Lo, Hi);
uint64_t LoElts = Lo.getValueType().getVectorNumElements();
uint64_t IdxVal = cast<ConstantSDNode>(Idx)->getZExtValue();
if (IdxVal < LoElts) {
assert(IdxVal + SubVT.getVectorNumElements() <= LoElts &&
"Extracted subvector crosses vector split!");
return DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, SubVT, Lo, Idx);
} else {
return DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, SubVT, Hi,
DAG.getConstant(IdxVal - LoElts, Idx.getValueType()));
}
}
SDValue DAGTypeLegalizer::SplitVecOp_EXTRACT_VECTOR_ELT(SDNode *N) {
SDValue Vec = N->getOperand(0);
SDValue Idx = N->getOperand(1);
EVT VecVT = Vec.getValueType();
if (isa<ConstantSDNode>(Idx)) {
uint64_t IdxVal = cast<ConstantSDNode>(Idx)->getZExtValue();
assert(IdxVal < VecVT.getVectorNumElements() && "Invalid vector index!");
SDValue Lo, Hi;
GetSplitVector(Vec, Lo, Hi);
uint64_t LoElts = Lo.getValueType().getVectorNumElements();
if (IdxVal < LoElts)
return SDValue(DAG.UpdateNodeOperands(N, Lo, Idx), 0);
return SDValue(DAG.UpdateNodeOperands(N, Hi,
DAG.getConstant(IdxVal - LoElts,
Idx.getValueType())), 0);
}
// Store the vector to the stack.
EVT EltVT = VecVT.getVectorElementType();
DebugLoc dl = N->getDebugLoc();
SDValue StackPtr = DAG.CreateStackTemporary(VecVT);
SDValue Store = DAG.getStore(DAG.getEntryNode(), dl, Vec, StackPtr,
MachinePointerInfo(), false, false, 0);
// Load back the required element.
StackPtr = GetVectorElementPointer(StackPtr, EltVT, Idx);
return DAG.getExtLoad(ISD::EXTLOAD, dl, N->getValueType(0), Store, StackPtr,
MachinePointerInfo(), EltVT, false, false, 0);
}
SDValue DAGTypeLegalizer::SplitVecOp_STORE(StoreSDNode *N, unsigned OpNo) {
assert(N->isUnindexed() && "Indexed store of vector?");
assert(OpNo == 1 && "Can only split the stored value");
DebugLoc DL = N->getDebugLoc();
bool isTruncating = N->isTruncatingStore();
SDValue Ch = N->getChain();
SDValue Ptr = N->getBasePtr();
EVT MemoryVT = N->getMemoryVT();
unsigned Alignment = N->getOriginalAlignment();
bool isVol = N->isVolatile();
bool isNT = N->isNonTemporal();
SDValue Lo, Hi;
GetSplitVector(N->getOperand(1), Lo, Hi);
EVT LoMemVT, HiMemVT;
GetSplitDestVTs(MemoryVT, LoMemVT, HiMemVT);
unsigned IncrementSize = LoMemVT.getSizeInBits()/8;
if (isTruncating)
Lo = DAG.getTruncStore(Ch, DL, Lo, Ptr, N->getPointerInfo(),
LoMemVT, isVol, isNT, Alignment);
else
Lo = DAG.getStore(Ch, DL, Lo, Ptr, N->getPointerInfo(),
isVol, isNT, Alignment);
// Increment the pointer to the other half.
Ptr = DAG.getNode(ISD::ADD, DL, Ptr.getValueType(), Ptr,
DAG.getIntPtrConstant(IncrementSize));
if (isTruncating)
Hi = DAG.getTruncStore(Ch, DL, Hi, Ptr,
N->getPointerInfo().getWithOffset(IncrementSize),
HiMemVT, isVol, isNT, Alignment);
else
Hi = DAG.getStore(Ch, DL, Hi, Ptr,
N->getPointerInfo().getWithOffset(IncrementSize),
isVol, isNT, Alignment);
return DAG.getNode(ISD::TokenFactor, DL, MVT::Other, Lo, Hi);
}
SDValue DAGTypeLegalizer::SplitVecOp_CONCAT_VECTORS(SDNode *N) {
DebugLoc DL = N->getDebugLoc();
// The input operands all must have the same type, and we know the result the
// result type is valid. Convert this to a buildvector which extracts all the
// input elements.
// TODO: If the input elements are power-two vectors, we could convert this to
// a new CONCAT_VECTORS node with elements that are half-wide.
SmallVector<SDValue, 32> Elts;
EVT EltVT = N->getValueType(0).getVectorElementType();
for (unsigned op = 0, e = N->getNumOperands(); op != e; ++op) {
SDValue Op = N->getOperand(op);
for (unsigned i = 0, e = Op.getValueType().getVectorNumElements();
i != e; ++i) {
Elts.push_back(DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, EltVT,
Op, DAG.getIntPtrConstant(i)));
}
}
return DAG.getNode(ISD::BUILD_VECTOR, DL, N->getValueType(0),
&Elts[0], Elts.size());
}
SDValue DAGTypeLegalizer::SplitVecOp_VSETCC(SDNode *N) {
assert(N->getValueType(0).isVector() &&
N->getOperand(0).getValueType().isVector() &&
"Operand types must be vectors");
// The result has a legal vector type, but the input needs splitting.
SDValue Lo0, Hi0, Lo1, Hi1, LoRes, HiRes;
DebugLoc DL = N->getDebugLoc();
GetSplitVector(N->getOperand(0), Lo0, Hi0);
GetSplitVector(N->getOperand(1), Lo1, Hi1);
unsigned PartElements = Lo0.getValueType().getVectorNumElements();
EVT PartResVT = EVT::getVectorVT(*DAG.getContext(), MVT::i1, PartElements);
EVT WideResVT = EVT::getVectorVT(*DAG.getContext(), MVT::i1, 2*PartElements);
LoRes = DAG.getNode(ISD::SETCC, DL, PartResVT, Lo0, Lo1, N->getOperand(2));
HiRes = DAG.getNode(ISD::SETCC, DL, PartResVT, Hi0, Hi1, N->getOperand(2));
SDValue Con = DAG.getNode(ISD::CONCAT_VECTORS, DL, WideResVT, LoRes, HiRes);
return PromoteTargetBoolean(Con, N->getValueType(0));
}
SDValue DAGTypeLegalizer::SplitVecOp_FP_ROUND(SDNode *N) {
// The result has a legal vector type, but the input needs splitting.
EVT ResVT = N->getValueType(0);
SDValue Lo, Hi;
DebugLoc DL = N->getDebugLoc();
GetSplitVector(N->getOperand(0), Lo, Hi);
EVT InVT = Lo.getValueType();
EVT OutVT = EVT::getVectorVT(*DAG.getContext(), ResVT.getVectorElementType(),
InVT.getVectorNumElements());
Lo = DAG.getNode(ISD::FP_ROUND, DL, OutVT, Lo, N->getOperand(1));
Hi = DAG.getNode(ISD::FP_ROUND, DL, OutVT, Hi, N->getOperand(1));
return DAG.getNode(ISD::CONCAT_VECTORS, DL, ResVT, Lo, Hi);
}
//===----------------------------------------------------------------------===//
// Result Vector Widening
//===----------------------------------------------------------------------===//
void DAGTypeLegalizer::WidenVectorResult(SDNode *N, unsigned ResNo) {
DEBUG(dbgs() << "Widen node result " << ResNo << ": ";
N->dump(&DAG);
dbgs() << "\n");
// See if the target wants to custom widen this node.
if (CustomWidenLowerNode(N, N->getValueType(ResNo)))
return;
SDValue Res = SDValue();
switch (N->getOpcode()) {
default:
#ifndef NDEBUG
dbgs() << "WidenVectorResult #" << ResNo << ": ";
N->dump(&DAG);
dbgs() << "\n";
#endif
llvm_unreachable("Do not know how to widen the result of this operator!");
case ISD::MERGE_VALUES: Res = WidenVecRes_MERGE_VALUES(N, ResNo); break;
case ISD::BITCAST: Res = WidenVecRes_BITCAST(N); break;
case ISD::BUILD_VECTOR: Res = WidenVecRes_BUILD_VECTOR(N); break;
case ISD::CONCAT_VECTORS: Res = WidenVecRes_CONCAT_VECTORS(N); break;
case ISD::CONVERT_RNDSAT: Res = WidenVecRes_CONVERT_RNDSAT(N); break;
case ISD::EXTRACT_SUBVECTOR: Res = WidenVecRes_EXTRACT_SUBVECTOR(N); break;
case ISD::FP_ROUND_INREG: Res = WidenVecRes_InregOp(N); break;
case ISD::INSERT_VECTOR_ELT: Res = WidenVecRes_INSERT_VECTOR_ELT(N); break;
case ISD::LOAD: Res = WidenVecRes_LOAD(N); break;
case ISD::SCALAR_TO_VECTOR: Res = WidenVecRes_SCALAR_TO_VECTOR(N); break;
case ISD::SIGN_EXTEND_INREG: Res = WidenVecRes_InregOp(N); break;
case ISD::VSELECT:
case ISD::SELECT: Res = WidenVecRes_SELECT(N); break;
case ISD::SELECT_CC: Res = WidenVecRes_SELECT_CC(N); break;
case ISD::SETCC: Res = WidenVecRes_SETCC(N); break;
case ISD::UNDEF: Res = WidenVecRes_UNDEF(N); break;
case ISD::VECTOR_SHUFFLE:
Res = WidenVecRes_VECTOR_SHUFFLE(cast<ShuffleVectorSDNode>(N));
break;
case ISD::ADD:
case ISD::AND:
case ISD::BSWAP:
case ISD::FADD:
case ISD::FCOPYSIGN:
case ISD::FDIV:
case ISD::FMUL:
case ISD::FPOW:
case ISD::FREM:
case ISD::FSUB:
case ISD::MUL:
case ISD::MULHS:
case ISD::MULHU:
case ISD::OR:
case ISD::SDIV:
case ISD::SREM:
case ISD::UDIV:
case ISD::UREM:
case ISD::SUB:
case ISD::XOR:
Res = WidenVecRes_Binary(N);
break;
case ISD::FPOWI:
Res = WidenVecRes_POWI(N);
break;
case ISD::SHL:
case ISD::SRA:
case ISD::SRL:
Res = WidenVecRes_Shift(N);
break;
case ISD::ANY_EXTEND:
case ISD::FP_EXTEND:
case ISD::FP_ROUND:
case ISD::FP_TO_SINT:
case ISD::FP_TO_UINT:
case ISD::SIGN_EXTEND:
case ISD::SINT_TO_FP:
case ISD::TRUNCATE:
case ISD::UINT_TO_FP:
case ISD::ZERO_EXTEND:
Res = WidenVecRes_Convert(N);
break;
case ISD::CTLZ:
case ISD::CTPOP:
case ISD::CTTZ:
case ISD::FABS:
case ISD::FCEIL:
case ISD::FCOS:
case ISD::FEXP:
case ISD::FEXP2:
case ISD::FFLOOR:
case ISD::FLOG:
case ISD::FLOG10:
case ISD::FLOG2:
case ISD::FNEARBYINT:
case ISD::FNEG:
case ISD::FRINT:
case ISD::FSIN:
case ISD::FSQRT:
case ISD::FTRUNC:
Res = WidenVecRes_Unary(N);
break;
}
// If Res is null, the sub-method took care of registering the result.
if (Res.getNode())
SetWidenedVector(SDValue(N, ResNo), Res);
}
SDValue DAGTypeLegalizer::WidenVecRes_Binary(SDNode *N) {
// Binary op widening.
unsigned Opcode = N->getOpcode();
DebugLoc dl = N->getDebugLoc();
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
EVT WidenEltVT = WidenVT.getVectorElementType();
EVT VT = WidenVT;
unsigned NumElts = VT.getVectorNumElements();
while (!TLI.isTypeLegal(VT) && NumElts != 1) {
NumElts = NumElts / 2;
VT = EVT::getVectorVT(*DAG.getContext(), WidenEltVT, NumElts);
}
if (NumElts != 1 && !TLI.canOpTrap(N->getOpcode(), VT)) {
// Operation doesn't trap so just widen as normal.
SDValue InOp1 = GetWidenedVector(N->getOperand(0));
SDValue InOp2 = GetWidenedVector(N->getOperand(1));
return DAG.getNode(N->getOpcode(), dl, WidenVT, InOp1, InOp2);
}
// No legal vector version so unroll the vector operation and then widen.
if (NumElts == 1)
return DAG.UnrollVectorOp(N, WidenVT.getVectorNumElements());
// Since the operation can trap, apply operation on the original vector.
EVT MaxVT = VT;
SDValue InOp1 = GetWidenedVector(N->getOperand(0));
SDValue InOp2 = GetWidenedVector(N->getOperand(1));
unsigned CurNumElts = N->getValueType(0).getVectorNumElements();
SmallVector<SDValue, 16> ConcatOps(CurNumElts);
unsigned ConcatEnd = 0; // Current ConcatOps index.
int Idx = 0; // Current Idx into input vectors.
// NumElts := greatest legal vector size (at most WidenVT)
// while (orig. vector has unhandled elements) {
// take munches of size NumElts from the beginning and add to ConcatOps
// NumElts := next smaller supported vector size or 1
// }
while (CurNumElts != 0) {
while (CurNumElts >= NumElts) {
SDValue EOp1 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, VT, InOp1,
DAG.getIntPtrConstant(Idx));
SDValue EOp2 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, VT, InOp2,
DAG.getIntPtrConstant(Idx));
ConcatOps[ConcatEnd++] = DAG.getNode(Opcode, dl, VT, EOp1, EOp2);
Idx += NumElts;
CurNumElts -= NumElts;
}
do {
NumElts = NumElts / 2;
VT = EVT::getVectorVT(*DAG.getContext(), WidenEltVT, NumElts);
} while (!TLI.isTypeLegal(VT) && NumElts != 1);
if (NumElts == 1) {
for (unsigned i = 0; i != CurNumElts; ++i, ++Idx) {
SDValue EOp1 = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, WidenEltVT,
InOp1, DAG.getIntPtrConstant(Idx));
SDValue EOp2 = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, WidenEltVT,
InOp2, DAG.getIntPtrConstant(Idx));
ConcatOps[ConcatEnd++] = DAG.getNode(Opcode, dl, WidenEltVT,
EOp1, EOp2);
}
CurNumElts = 0;
}
}
// Check to see if we have a single operation with the widen type.
if (ConcatEnd == 1) {
VT = ConcatOps[0].getValueType();
if (VT == WidenVT)
return ConcatOps[0];
}
// while (Some element of ConcatOps is not of type MaxVT) {
// From the end of ConcatOps, collect elements of the same type and put
// them into an op of the next larger supported type
// }
while (ConcatOps[ConcatEnd-1].getValueType() != MaxVT) {
Idx = ConcatEnd - 1;
VT = ConcatOps[Idx--].getValueType();
while (Idx >= 0 && ConcatOps[Idx].getValueType() == VT)
Idx--;
int NextSize = VT.isVector() ? VT.getVectorNumElements() : 1;
EVT NextVT;
do {
NextSize *= 2;
NextVT = EVT::getVectorVT(*DAG.getContext(), WidenEltVT, NextSize);
} while (!TLI.isTypeLegal(NextVT));
if (!VT.isVector()) {
// Scalar type, create an INSERT_VECTOR_ELEMENT of type NextVT
SDValue VecOp = DAG.getUNDEF(NextVT);
unsigned NumToInsert = ConcatEnd - Idx - 1;
for (unsigned i = 0, OpIdx = Idx+1; i < NumToInsert; i++, OpIdx++) {
VecOp = DAG.getNode(ISD::INSERT_VECTOR_ELT, dl, NextVT, VecOp,
ConcatOps[OpIdx], DAG.getIntPtrConstant(i));
}
ConcatOps[Idx+1] = VecOp;
ConcatEnd = Idx + 2;
} else {
// Vector type, create a CONCAT_VECTORS of type NextVT
SDValue undefVec = DAG.getUNDEF(VT);
unsigned OpsToConcat = NextSize/VT.getVectorNumElements();
SmallVector<SDValue, 16> SubConcatOps(OpsToConcat);
unsigned RealVals = ConcatEnd - Idx - 1;
unsigned SubConcatEnd = 0;
unsigned SubConcatIdx = Idx + 1;
while (SubConcatEnd < RealVals)
SubConcatOps[SubConcatEnd++] = ConcatOps[++Idx];
while (SubConcatEnd < OpsToConcat)
SubConcatOps[SubConcatEnd++] = undefVec;
ConcatOps[SubConcatIdx] = DAG.getNode(ISD::CONCAT_VECTORS, dl,
NextVT, &SubConcatOps[0],
OpsToConcat);
ConcatEnd = SubConcatIdx + 1;
}
}
// Check to see if we have a single operation with the widen type.
if (ConcatEnd == 1) {
VT = ConcatOps[0].getValueType();
if (VT == WidenVT)
return ConcatOps[0];
}
// add undefs of size MaxVT until ConcatOps grows to length of WidenVT
unsigned NumOps = WidenVT.getVectorNumElements()/MaxVT.getVectorNumElements();
if (NumOps != ConcatEnd ) {
SDValue UndefVal = DAG.getUNDEF(MaxVT);
for (unsigned j = ConcatEnd; j < NumOps; ++j)
ConcatOps[j] = UndefVal;
}
return DAG.getNode(ISD::CONCAT_VECTORS, dl, WidenVT, &ConcatOps[0], NumOps);
}
SDValue DAGTypeLegalizer::WidenVecRes_Convert(SDNode *N) {
SDValue InOp = N->getOperand(0);
DebugLoc DL = N->getDebugLoc();
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
unsigned WidenNumElts = WidenVT.getVectorNumElements();
EVT InVT = InOp.getValueType();
EVT InEltVT = InVT.getVectorElementType();
EVT InWidenVT = EVT::getVectorVT(*DAG.getContext(), InEltVT, WidenNumElts);
unsigned Opcode = N->getOpcode();
unsigned InVTNumElts = InVT.getVectorNumElements();
if (getTypeAction(InVT) == TargetLowering::TypeWidenVector) {
InOp = GetWidenedVector(N->getOperand(0));
InVT = InOp.getValueType();
InVTNumElts = InVT.getVectorNumElements();
if (InVTNumElts == WidenNumElts) {
if (N->getNumOperands() == 1)
return DAG.getNode(Opcode, DL, WidenVT, InOp);
return DAG.getNode(Opcode, DL, WidenVT, InOp, N->getOperand(1));
}
}
if (TLI.isTypeLegal(InWidenVT)) {
// Because the result and the input are different vector types, widening
// the result could create a legal type but widening the input might make
// it an illegal type that might lead to repeatedly splitting the input
// and then widening it. To avoid this, we widen the input only if
// it results in a legal type.
if (WidenNumElts % InVTNumElts == 0) {
// Widen the input and call convert on the widened input vector.
unsigned NumConcat = WidenNumElts/InVTNumElts;
SmallVector<SDValue, 16> Ops(NumConcat);
Ops[0] = InOp;
SDValue UndefVal = DAG.getUNDEF(InVT);
for (unsigned i = 1; i != NumConcat; ++i)
Ops[i] = UndefVal;
SDValue InVec = DAG.getNode(ISD::CONCAT_VECTORS, DL, InWidenVT,
&Ops[0], NumConcat);
if (N->getNumOperands() == 1)
return DAG.getNode(Opcode, DL, WidenVT, InVec);
return DAG.getNode(Opcode, DL, WidenVT, InVec, N->getOperand(1));
}
if (InVTNumElts % WidenNumElts == 0) {
SDValue InVal = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, InWidenVT,
InOp, DAG.getIntPtrConstant(0));
// Extract the input and convert the shorten input vector.
if (N->getNumOperands() == 1)
return DAG.getNode(Opcode, DL, WidenVT, InVal);
return DAG.getNode(Opcode, DL, WidenVT, InVal, N->getOperand(1));
}
}
// Otherwise unroll into some nasty scalar code and rebuild the vector.
SmallVector<SDValue, 16> Ops(WidenNumElts);
EVT EltVT = WidenVT.getVectorElementType();
unsigned MinElts = std::min(InVTNumElts, WidenNumElts);
unsigned i;
for (i=0; i < MinElts; ++i) {
SDValue Val = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, InEltVT, InOp,
DAG.getIntPtrConstant(i));
if (N->getNumOperands() == 1)
Ops[i] = DAG.getNode(Opcode, DL, EltVT, Val);
else
Ops[i] = DAG.getNode(Opcode, DL, EltVT, Val, N->getOperand(1));
}
SDValue UndefVal = DAG.getUNDEF(EltVT);
for (; i < WidenNumElts; ++i)
Ops[i] = UndefVal;
return DAG.getNode(ISD::BUILD_VECTOR, DL, WidenVT, &Ops[0], WidenNumElts);
}
SDValue DAGTypeLegalizer::WidenVecRes_POWI(SDNode *N) {
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
SDValue InOp = GetWidenedVector(N->getOperand(0));
SDValue ShOp = N->getOperand(1);
return DAG.getNode(N->getOpcode(), N->getDebugLoc(), WidenVT, InOp, ShOp);
}
SDValue DAGTypeLegalizer::WidenVecRes_Shift(SDNode *N) {
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
SDValue InOp = GetWidenedVector(N->getOperand(0));
SDValue ShOp = N->getOperand(1);
EVT ShVT = ShOp.getValueType();
if (getTypeAction(ShVT) == TargetLowering::TypeWidenVector) {
ShOp = GetWidenedVector(ShOp);
ShVT = ShOp.getValueType();
}
EVT ShWidenVT = EVT::getVectorVT(*DAG.getContext(),
ShVT.getVectorElementType(),
WidenVT.getVectorNumElements());
if (ShVT != ShWidenVT)
ShOp = ModifyToType(ShOp, ShWidenVT);
return DAG.getNode(N->getOpcode(), N->getDebugLoc(), WidenVT, InOp, ShOp);
}
SDValue DAGTypeLegalizer::WidenVecRes_Unary(SDNode *N) {
// Unary op widening.
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
SDValue InOp = GetWidenedVector(N->getOperand(0));
return DAG.getNode(N->getOpcode(), N->getDebugLoc(), WidenVT, InOp);
}
SDValue DAGTypeLegalizer::WidenVecRes_InregOp(SDNode *N) {
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
EVT ExtVT = EVT::getVectorVT(*DAG.getContext(),
cast<VTSDNode>(N->getOperand(1))->getVT()
.getVectorElementType(),
WidenVT.getVectorNumElements());
SDValue WidenLHS = GetWidenedVector(N->getOperand(0));
return DAG.getNode(N->getOpcode(), N->getDebugLoc(),
WidenVT, WidenLHS, DAG.getValueType(ExtVT));
}
SDValue DAGTypeLegalizer::WidenVecRes_MERGE_VALUES(SDNode *N, unsigned ResNo) {
SDValue WidenVec = DisintegrateMERGE_VALUES(N, ResNo);
return GetWidenedVector(WidenVec);
}
SDValue DAGTypeLegalizer::WidenVecRes_BITCAST(SDNode *N) {
SDValue InOp = N->getOperand(0);
EVT InVT = InOp.getValueType();
EVT VT = N->getValueType(0);
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), VT);
DebugLoc dl = N->getDebugLoc();
switch (getTypeAction(InVT)) {
default:
assert(false && "Unknown type action!");
break;
case TargetLowering::TypeLegal:
break;
case TargetLowering::TypePromoteInteger:
// If the InOp is promoted to the same size, convert it. Otherwise,
// fall out of the switch and widen the promoted input.
InOp = GetPromotedInteger(InOp);
InVT = InOp.getValueType();
if (WidenVT.bitsEq(InVT))
return DAG.getNode(ISD::BITCAST, dl, WidenVT, InOp);
break;
case TargetLowering::TypeSoftenFloat:
case TargetLowering::TypeExpandInteger:
case TargetLowering::TypeExpandFloat:
case TargetLowering::TypeScalarizeVector:
case TargetLowering::TypeSplitVector:
break;
case TargetLowering::TypeWidenVector:
// If the InOp is widened to the same size, convert it. Otherwise, fall
// out of the switch and widen the widened input.
InOp = GetWidenedVector(InOp);
InVT = InOp.getValueType();
if (WidenVT.bitsEq(InVT))
// The input widens to the same size. Convert to the widen value.
return DAG.getNode(ISD::BITCAST, dl, WidenVT, InOp);
break;
}
unsigned WidenSize = WidenVT.getSizeInBits();
unsigned InSize = InVT.getSizeInBits();
// x86mmx is not an acceptable vector element type, so don't try.
if (WidenSize % InSize == 0 && InVT != MVT::x86mmx) {
// Determine new input vector type. The new input vector type will use
// the same element type (if its a vector) or use the input type as a
// vector. It is the same size as the type to widen to.
EVT NewInVT;
unsigned NewNumElts = WidenSize / InSize;
if (InVT.isVector()) {
EVT InEltVT = InVT.getVectorElementType();
NewInVT = EVT::getVectorVT(*DAG.getContext(), InEltVT,
WidenSize / InEltVT.getSizeInBits());
} else {
NewInVT = EVT::getVectorVT(*DAG.getContext(), InVT, NewNumElts);
}
if (TLI.isTypeLegal(NewInVT)) {
// Because the result and the input are different vector types, widening
// the result could create a legal type but widening the input might make
// it an illegal type that might lead to repeatedly splitting the input
// and then widening it. To avoid this, we widen the input only if
// it results in a legal type.
SmallVector<SDValue, 16> Ops(NewNumElts);
SDValue UndefVal = DAG.getUNDEF(InVT);
Ops[0] = InOp;
for (unsigned i = 1; i < NewNumElts; ++i)
Ops[i] = UndefVal;
SDValue NewVec;
if (InVT.isVector())
NewVec = DAG.getNode(ISD::CONCAT_VECTORS, dl,
NewInVT, &Ops[0], NewNumElts);
else
NewVec = DAG.getNode(ISD::BUILD_VECTOR, dl,
NewInVT, &Ops[0], NewNumElts);
return DAG.getNode(ISD::BITCAST, dl, WidenVT, NewVec);
}
}
return CreateStackStoreLoad(InOp, WidenVT);
}
SDValue DAGTypeLegalizer::WidenVecRes_BUILD_VECTOR(SDNode *N) {
DebugLoc dl = N->getDebugLoc();
// Build a vector with undefined for the new nodes.
EVT VT = N->getValueType(0);
EVT EltVT = VT.getVectorElementType();
unsigned NumElts = VT.getVectorNumElements();
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), VT);
unsigned WidenNumElts = WidenVT.getVectorNumElements();
SmallVector<SDValue, 16> NewOps(N->op_begin(), N->op_end());
NewOps.reserve(WidenNumElts);
for (unsigned i = NumElts; i < WidenNumElts; ++i)
NewOps.push_back(DAG.getUNDEF(EltVT));
return DAG.getNode(ISD::BUILD_VECTOR, dl, WidenVT, &NewOps[0], NewOps.size());
}
SDValue DAGTypeLegalizer::WidenVecRes_CONCAT_VECTORS(SDNode *N) {
EVT InVT = N->getOperand(0).getValueType();
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
DebugLoc dl = N->getDebugLoc();
unsigned WidenNumElts = WidenVT.getVectorNumElements();
unsigned NumInElts = InVT.getVectorNumElements();
unsigned NumOperands = N->getNumOperands();
bool InputWidened = false; // Indicates we need to widen the input.
if (getTypeAction(InVT) != TargetLowering::TypeWidenVector) {
if (WidenVT.getVectorNumElements() % InVT.getVectorNumElements() == 0) {
// Add undef vectors to widen to correct length.
unsigned NumConcat = WidenVT.getVectorNumElements() /
InVT.getVectorNumElements();
SDValue UndefVal = DAG.getUNDEF(InVT);
SmallVector<SDValue, 16> Ops(NumConcat);
for (unsigned i=0; i < NumOperands; ++i)
Ops[i] = N->getOperand(i);
for (unsigned i = NumOperands; i != NumConcat; ++i)
Ops[i] = UndefVal;
return DAG.getNode(ISD::CONCAT_VECTORS, dl, WidenVT, &Ops[0], NumConcat);
}
} else {
InputWidened = true;
if (WidenVT == TLI.getTypeToTransformTo(*DAG.getContext(), InVT)) {
// The inputs and the result are widen to the same value.
unsigned i;
for (i=1; i < NumOperands; ++i)
if (N->getOperand(i).getOpcode() != ISD::UNDEF)
break;
if (i == NumOperands)
// Everything but the first operand is an UNDEF so just return the
// widened first operand.
return GetWidenedVector(N->getOperand(0));
if (NumOperands == 2) {
// Replace concat of two operands with a shuffle.
SmallVector<int, 16> MaskOps(WidenNumElts, -1);
for (unsigned i = 0; i < NumInElts; ++i) {
MaskOps[i] = i;
MaskOps[i + NumInElts] = i + WidenNumElts;
}
return DAG.getVectorShuffle(WidenVT, dl,
GetWidenedVector(N->getOperand(0)),
GetWidenedVector(N->getOperand(1)),
&MaskOps[0]);
}
}
}
// Fall back to use extracts and build vector.
EVT EltVT = WidenVT.getVectorElementType();
SmallVector<SDValue, 16> Ops(WidenNumElts);
unsigned Idx = 0;
for (unsigned i=0; i < NumOperands; ++i) {
SDValue InOp = N->getOperand(i);
if (InputWidened)
InOp = GetWidenedVector(InOp);
for (unsigned j=0; j < NumInElts; ++j)
Ops[Idx++] = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, EltVT, InOp,
DAG.getIntPtrConstant(j));
}
SDValue UndefVal = DAG.getUNDEF(EltVT);
for (; Idx < WidenNumElts; ++Idx)
Ops[Idx] = UndefVal;
return DAG.getNode(ISD::BUILD_VECTOR, dl, WidenVT, &Ops[0], WidenNumElts);
}
SDValue DAGTypeLegalizer::WidenVecRes_CONVERT_RNDSAT(SDNode *N) {
DebugLoc dl = N->getDebugLoc();
SDValue InOp = N->getOperand(0);
SDValue RndOp = N->getOperand(3);
SDValue SatOp = N->getOperand(4);
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
unsigned WidenNumElts = WidenVT.getVectorNumElements();
EVT InVT = InOp.getValueType();
EVT InEltVT = InVT.getVectorElementType();
EVT InWidenVT = EVT::getVectorVT(*DAG.getContext(), InEltVT, WidenNumElts);
SDValue DTyOp = DAG.getValueType(WidenVT);
SDValue STyOp = DAG.getValueType(InWidenVT);
ISD::CvtCode CvtCode = cast<CvtRndSatSDNode>(N)->getCvtCode();
unsigned InVTNumElts = InVT.getVectorNumElements();
if (getTypeAction(InVT) == TargetLowering::TypeWidenVector) {
InOp = GetWidenedVector(InOp);
InVT = InOp.getValueType();
InVTNumElts = InVT.getVectorNumElements();
if (InVTNumElts == WidenNumElts)
return DAG.getConvertRndSat(WidenVT, dl, InOp, DTyOp, STyOp, RndOp,
SatOp, CvtCode);
}
if (TLI.isTypeLegal(InWidenVT)) {
// Because the result and the input are different vector types, widening
// the result could create a legal type but widening the input might make
// it an illegal type that might lead to repeatedly splitting the input
// and then widening it. To avoid this, we widen the input only if
// it results in a legal type.
if (WidenNumElts % InVTNumElts == 0) {
// Widen the input and call convert on the widened input vector.
unsigned NumConcat = WidenNumElts/InVTNumElts;
SmallVector<SDValue, 16> Ops(NumConcat);
Ops[0] = InOp;
SDValue UndefVal = DAG.getUNDEF(InVT);
for (unsigned i = 1; i != NumConcat; ++i)
Ops[i] = UndefVal;
InOp = DAG.getNode(ISD::CONCAT_VECTORS, dl, InWidenVT, &Ops[0],NumConcat);
return DAG.getConvertRndSat(WidenVT, dl, InOp, DTyOp, STyOp, RndOp,
SatOp, CvtCode);
}
if (InVTNumElts % WidenNumElts == 0) {
// Extract the input and convert the shorten input vector.
InOp = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, InWidenVT, InOp,
DAG.getIntPtrConstant(0));
return DAG.getConvertRndSat(WidenVT, dl, InOp, DTyOp, STyOp, RndOp,
SatOp, CvtCode);
}
}
// Otherwise unroll into some nasty scalar code and rebuild the vector.
SmallVector<SDValue, 16> Ops(WidenNumElts);
EVT EltVT = WidenVT.getVectorElementType();
DTyOp = DAG.getValueType(EltVT);
STyOp = DAG.getValueType(InEltVT);
unsigned MinElts = std::min(InVTNumElts, WidenNumElts);
unsigned i;
for (i=0; i < MinElts; ++i) {
SDValue ExtVal = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, InEltVT, InOp,
DAG.getIntPtrConstant(i));
Ops[i] = DAG.getConvertRndSat(WidenVT, dl, ExtVal, DTyOp, STyOp, RndOp,
SatOp, CvtCode);
}
SDValue UndefVal = DAG.getUNDEF(EltVT);
for (; i < WidenNumElts; ++i)
Ops[i] = UndefVal;
return DAG.getNode(ISD::BUILD_VECTOR, dl, WidenVT, &Ops[0], WidenNumElts);
}
SDValue DAGTypeLegalizer::WidenVecRes_EXTRACT_SUBVECTOR(SDNode *N) {
EVT VT = N->getValueType(0);
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), VT);
unsigned WidenNumElts = WidenVT.getVectorNumElements();
SDValue InOp = N->getOperand(0);
SDValue Idx = N->getOperand(1);
DebugLoc dl = N->getDebugLoc();
if (getTypeAction(InOp.getValueType()) == TargetLowering::TypeWidenVector)
InOp = GetWidenedVector(InOp);
EVT InVT = InOp.getValueType();
// Check if we can just return the input vector after widening.
uint64_t IdxVal = cast<ConstantSDNode>(Idx)->getZExtValue();
if (IdxVal == 0 && InVT == WidenVT)
return InOp;
// Check if we can extract from the vector.
unsigned InNumElts = InVT.getVectorNumElements();
if (IdxVal % WidenNumElts == 0 && IdxVal + WidenNumElts < InNumElts)
return DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, WidenVT, InOp, Idx);
// We could try widening the input to the right length but for now, extract
// the original elements, fill the rest with undefs and build a vector.
SmallVector<SDValue, 16> Ops(WidenNumElts);
EVT EltVT = VT.getVectorElementType();
unsigned NumElts = VT.getVectorNumElements();
unsigned i;
for (i=0; i < NumElts; ++i)
Ops[i] = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, EltVT, InOp,
DAG.getIntPtrConstant(IdxVal+i));
SDValue UndefVal = DAG.getUNDEF(EltVT);
for (; i < WidenNumElts; ++i)
Ops[i] = UndefVal;
return DAG.getNode(ISD::BUILD_VECTOR, dl, WidenVT, &Ops[0], WidenNumElts);
}
SDValue DAGTypeLegalizer::WidenVecRes_INSERT_VECTOR_ELT(SDNode *N) {
SDValue InOp = GetWidenedVector(N->getOperand(0));
return DAG.getNode(ISD::INSERT_VECTOR_ELT, N->getDebugLoc(),
InOp.getValueType(), InOp,
N->getOperand(1), N->getOperand(2));
}
SDValue DAGTypeLegalizer::WidenVecRes_LOAD(SDNode *N) {
LoadSDNode *LD = cast<LoadSDNode>(N);
ISD::LoadExtType ExtType = LD->getExtensionType();
SDValue Result;
SmallVector<SDValue, 16> LdChain; // Chain for the series of load
if (ExtType != ISD::NON_EXTLOAD)
Result = GenWidenVectorExtLoads(LdChain, LD, ExtType);
else
Result = GenWidenVectorLoads(LdChain, LD);
// If we generate a single load, we can use that for the chain. Otherwise,
// build a factor node to remember the multiple loads are independent and
// chain to that.
SDValue NewChain;
if (LdChain.size() == 1)
NewChain = LdChain[0];
else
NewChain = DAG.getNode(ISD::TokenFactor, LD->getDebugLoc(), MVT::Other,
&LdChain[0], LdChain.size());
// Modified the chain - switch anything that used the old chain to use
// the new one.
ReplaceValueWith(SDValue(N, 1), NewChain);
return Result;
}
SDValue DAGTypeLegalizer::WidenVecRes_SCALAR_TO_VECTOR(SDNode *N) {
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
return DAG.getNode(ISD::SCALAR_TO_VECTOR, N->getDebugLoc(),
WidenVT, N->getOperand(0));
}
SDValue DAGTypeLegalizer::WidenVecRes_SELECT(SDNode *N) {
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
unsigned WidenNumElts = WidenVT.getVectorNumElements();
SDValue Cond1 = N->getOperand(0);
EVT CondVT = Cond1.getValueType();
if (CondVT.isVector()) {
EVT CondEltVT = CondVT.getVectorElementType();
EVT CondWidenVT = EVT::getVectorVT(*DAG.getContext(),
CondEltVT, WidenNumElts);
if (getTypeAction(CondVT) == TargetLowering::TypeWidenVector)
Cond1 = GetWidenedVector(Cond1);
if (Cond1.getValueType() != CondWidenVT)
Cond1 = ModifyToType(Cond1, CondWidenVT);
}
SDValue InOp1 = GetWidenedVector(N->getOperand(1));
SDValue InOp2 = GetWidenedVector(N->getOperand(2));
assert(InOp1.getValueType() == WidenVT && InOp2.getValueType() == WidenVT);
return DAG.getNode(N->getOpcode(), N->getDebugLoc(),
WidenVT, Cond1, InOp1, InOp2);
}
SDValue DAGTypeLegalizer::WidenVecRes_SELECT_CC(SDNode *N) {
SDValue InOp1 = GetWidenedVector(N->getOperand(2));
SDValue InOp2 = GetWidenedVector(N->getOperand(3));
return DAG.getNode(ISD::SELECT_CC, N->getDebugLoc(),
InOp1.getValueType(), N->getOperand(0),
N->getOperand(1), InOp1, InOp2, N->getOperand(4));
}
SDValue DAGTypeLegalizer::WidenVecRes_SETCC(SDNode *N) {
assert(N->getValueType(0).isVector() ==
N->getOperand(0).getValueType().isVector() &&
"Scalar/Vector type mismatch");
if (N->getValueType(0).isVector()) return WidenVecRes_VSETCC(N);
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
SDValue InOp1 = GetWidenedVector(N->getOperand(0));
SDValue InOp2 = GetWidenedVector(N->getOperand(1));
return DAG.getNode(ISD::SETCC, N->getDebugLoc(), WidenVT,
InOp1, InOp2, N->getOperand(2));
}
SDValue DAGTypeLegalizer::WidenVecRes_UNDEF(SDNode *N) {
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
return DAG.getUNDEF(WidenVT);
}
SDValue DAGTypeLegalizer::WidenVecRes_VECTOR_SHUFFLE(ShuffleVectorSDNode *N) {
EVT VT = N->getValueType(0);
DebugLoc dl = N->getDebugLoc();
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), VT);
unsigned NumElts = VT.getVectorNumElements();
unsigned WidenNumElts = WidenVT.getVectorNumElements();
SDValue InOp1 = GetWidenedVector(N->getOperand(0));
SDValue InOp2 = GetWidenedVector(N->getOperand(1));
// Adjust mask based on new input vector length.
SmallVector<int, 16> NewMask;
for (unsigned i = 0; i != NumElts; ++i) {
int Idx = N->getMaskElt(i);
if (Idx < (int)NumElts)
NewMask.push_back(Idx);
else
NewMask.push_back(Idx - NumElts + WidenNumElts);
}
for (unsigned i = NumElts; i != WidenNumElts; ++i)
NewMask.push_back(-1);
return DAG.getVectorShuffle(WidenVT, dl, InOp1, InOp2, &NewMask[0]);
}
SDValue DAGTypeLegalizer::WidenVecRes_VSETCC(SDNode *N) {
assert(N->getValueType(0).isVector() &&
N->getOperand(0).getValueType().isVector() &&
"Operands must be vectors");
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(), N->getValueType(0));
unsigned WidenNumElts = WidenVT.getVectorNumElements();
SDValue InOp1 = N->getOperand(0);
EVT InVT = InOp1.getValueType();
assert(InVT.isVector() && "can not widen non vector type");
EVT WidenInVT = EVT::getVectorVT(*DAG.getContext(),
InVT.getVectorElementType(), WidenNumElts);
InOp1 = GetWidenedVector(InOp1);
SDValue InOp2 = GetWidenedVector(N->getOperand(1));
// Assume that the input and output will be widen appropriately. If not,
// we will have to unroll it at some point.
assert(InOp1.getValueType() == WidenInVT &&
InOp2.getValueType() == WidenInVT &&
"Input not widened to expected type!");
(void)WidenInVT;
return DAG.getNode(ISD::SETCC, N->getDebugLoc(),
WidenVT, InOp1, InOp2, N->getOperand(2));
}
//===----------------------------------------------------------------------===//
// Widen Vector Operand
//===----------------------------------------------------------------------===//
bool DAGTypeLegalizer::WidenVectorOperand(SDNode *N, unsigned ResNo) {
DEBUG(dbgs() << "Widen node operand " << ResNo << ": ";
N->dump(&DAG);
dbgs() << "\n");
SDValue Res = SDValue();
switch (N->getOpcode()) {
default:
#ifndef NDEBUG
dbgs() << "WidenVectorOperand op #" << ResNo << ": ";
N->dump(&DAG);
dbgs() << "\n";
#endif
llvm_unreachable("Do not know how to widen this operator's operand!");
case ISD::BITCAST: Res = WidenVecOp_BITCAST(N); break;
case ISD::CONCAT_VECTORS: Res = WidenVecOp_CONCAT_VECTORS(N); break;
case ISD::EXTRACT_SUBVECTOR: Res = WidenVecOp_EXTRACT_SUBVECTOR(N); break;
case ISD::EXTRACT_VECTOR_ELT: Res = WidenVecOp_EXTRACT_VECTOR_ELT(N); break;
case ISD::STORE: Res = WidenVecOp_STORE(N); break;
case ISD::SETCC: Res = WidenVecOp_SETCC(N); break;
case ISD::FP_EXTEND:
case ISD::FP_TO_SINT:
case ISD::FP_TO_UINT:
case ISD::SINT_TO_FP:
case ISD::UINT_TO_FP:
case ISD::TRUNCATE:
case ISD::SIGN_EXTEND:
case ISD::ZERO_EXTEND:
case ISD::ANY_EXTEND:
Res = WidenVecOp_Convert(N);
break;
}
// If Res is null, the sub-method took care of registering the result.
if (!Res.getNode()) return false;
// If the result is N, the sub-method updated N in place. Tell the legalizer
// core about this.
if (Res.getNode() == N)
return true;
assert(Res.getValueType() == N->getValueType(0) && N->getNumValues() == 1 &&
"Invalid operand expansion");
ReplaceValueWith(SDValue(N, 0), Res);
return false;
}
SDValue DAGTypeLegalizer::WidenVecOp_Convert(SDNode *N) {
// Since the result is legal and the input is illegal, it is unlikely
// that we can fix the input to a legal type so unroll the convert
// into some scalar code and create a nasty build vector.
EVT VT = N->getValueType(0);
EVT EltVT = VT.getVectorElementType();
DebugLoc dl = N->getDebugLoc();
unsigned NumElts = VT.getVectorNumElements();
SDValue InOp = N->getOperand(0);
if (getTypeAction(InOp.getValueType()) == TargetLowering::TypeWidenVector)
InOp = GetWidenedVector(InOp);
EVT InVT = InOp.getValueType();
EVT InEltVT = InVT.getVectorElementType();
unsigned Opcode = N->getOpcode();
SmallVector<SDValue, 16> Ops(NumElts);
for (unsigned i=0; i < NumElts; ++i)
Ops[i] = DAG.getNode(Opcode, dl, EltVT,
DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, InEltVT, InOp,
DAG.getIntPtrConstant(i)));
return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &Ops[0], NumElts);
}
SDValue DAGTypeLegalizer::WidenVecOp_BITCAST(SDNode *N) {
EVT VT = N->getValueType(0);
SDValue InOp = GetWidenedVector(N->getOperand(0));
EVT InWidenVT = InOp.getValueType();
DebugLoc dl = N->getDebugLoc();
// Check if we can convert between two legal vector types and extract.
unsigned InWidenSize = InWidenVT.getSizeInBits();
unsigned Size = VT.getSizeInBits();
// x86mmx is not an acceptable vector element type, so don't try.
if (InWidenSize % Size == 0 && !VT.isVector() && VT != MVT::x86mmx) {
unsigned NewNumElts = InWidenSize / Size;
EVT NewVT = EVT::getVectorVT(*DAG.getContext(), VT, NewNumElts);
if (TLI.isTypeLegal(NewVT)) {
SDValue BitOp = DAG.getNode(ISD::BITCAST, dl, NewVT, InOp);
return DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, VT, BitOp,
DAG.getIntPtrConstant(0));
}
}
return CreateStackStoreLoad(InOp, VT);
}
SDValue DAGTypeLegalizer::WidenVecOp_CONCAT_VECTORS(SDNode *N) {
// If the input vector is not legal, it is likely that we will not find a
// legal vector of the same size. Replace the concatenate vector with a
// nasty build vector.
EVT VT = N->getValueType(0);
EVT EltVT = VT.getVectorElementType();
DebugLoc dl = N->getDebugLoc();
unsigned NumElts = VT.getVectorNumElements();
SmallVector<SDValue, 16> Ops(NumElts);
EVT InVT = N->getOperand(0).getValueType();
unsigned NumInElts = InVT.getVectorNumElements();
unsigned Idx = 0;
unsigned NumOperands = N->getNumOperands();
for (unsigned i=0; i < NumOperands; ++i) {
SDValue InOp = N->getOperand(i);
if (getTypeAction(InOp.getValueType()) == TargetLowering::TypeWidenVector)
InOp = GetWidenedVector(InOp);
for (unsigned j=0; j < NumInElts; ++j)
Ops[Idx++] = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, EltVT, InOp,
DAG.getIntPtrConstant(j));
}
return DAG.getNode(ISD::BUILD_VECTOR, dl, VT, &Ops[0], NumElts);
}
SDValue DAGTypeLegalizer::WidenVecOp_EXTRACT_SUBVECTOR(SDNode *N) {
SDValue InOp = GetWidenedVector(N->getOperand(0));
return DAG.getNode(ISD::EXTRACT_SUBVECTOR, N->getDebugLoc(),
N->getValueType(0), InOp, N->getOperand(1));
}
SDValue DAGTypeLegalizer::WidenVecOp_EXTRACT_VECTOR_ELT(SDNode *N) {
SDValue InOp = GetWidenedVector(N->getOperand(0));
return DAG.getNode(ISD::EXTRACT_VECTOR_ELT, N->getDebugLoc(),
N->getValueType(0), InOp, N->getOperand(1));
}
SDValue DAGTypeLegalizer::WidenVecOp_STORE(SDNode *N) {
// We have to widen the value but we want only to store the original
// vector type.
StoreSDNode *ST = cast<StoreSDNode>(N);
SmallVector<SDValue, 16> StChain;
if (ST->isTruncatingStore())
GenWidenVectorTruncStores(StChain, ST);
else
GenWidenVectorStores(StChain, ST);
if (StChain.size() == 1)
return StChain[0];
else
return DAG.getNode(ISD::TokenFactor, ST->getDebugLoc(),
MVT::Other,&StChain[0],StChain.size());
}
SDValue DAGTypeLegalizer::WidenVecOp_SETCC(SDNode *N) {
SDValue InOp0 = GetWidenedVector(N->getOperand(0));
SDValue InOp1 = GetWidenedVector(N->getOperand(1));
DebugLoc dl = N->getDebugLoc();
// WARNING: In this code we widen the compare instruction with garbage.
// This garbage may contain denormal floats which may be slow. Is this a real
// concern ? Should we zero the unused lanes if this is a float compare ?
// Get a new SETCC node to compare the newly widened operands.
// Only some of the compared elements are legal.
EVT SVT = TLI.getSetCCResultType(InOp0.getValueType());
SDValue WideSETCC = DAG.getNode(ISD::SETCC, N->getDebugLoc(),
SVT, InOp0, InOp1, N->getOperand(2));
// Extract the needed results from the result vector.
EVT ResVT = EVT::getVectorVT(*DAG.getContext(),
SVT.getVectorElementType(),
N->getValueType(0).getVectorNumElements());
SDValue CC = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl,
ResVT, WideSETCC, DAG.getIntPtrConstant(0));
return PromoteTargetBoolean(CC, N->getValueType(0));
}
//===----------------------------------------------------------------------===//
// Vector Widening Utilities
//===----------------------------------------------------------------------===//
// Utility function to find the type to chop up a widen vector for load/store
// TLI: Target lowering used to determine legal types.
// Width: Width left need to load/store.
// WidenVT: The widen vector type to load to/store from
// Align: If 0, don't allow use of a wider type
// WidenEx: If Align is not 0, the amount additional we can load/store from.
static EVT FindMemType(SelectionDAG& DAG, const TargetLowering &TLI,
unsigned Width, EVT WidenVT,
unsigned Align = 0, unsigned WidenEx = 0) {
EVT WidenEltVT = WidenVT.getVectorElementType();
unsigned WidenWidth = WidenVT.getSizeInBits();
unsigned WidenEltWidth = WidenEltVT.getSizeInBits();
unsigned AlignInBits = Align*8;
// If we have one element to load/store, return it.
EVT RetVT = WidenEltVT;
if (Width == WidenEltWidth)
return RetVT;
// See if there is larger legal integer than the element type to load/store
unsigned VT;
for (VT = (unsigned)MVT::LAST_INTEGER_VALUETYPE;
VT >= (unsigned)MVT::FIRST_INTEGER_VALUETYPE; --VT) {
EVT MemVT((MVT::SimpleValueType) VT);
unsigned MemVTWidth = MemVT.getSizeInBits();
if (MemVT.getSizeInBits() <= WidenEltWidth)
break;
if (TLI.isTypeLegal(MemVT) && (WidenWidth % MemVTWidth) == 0 &&
isPowerOf2_32(WidenWidth / MemVTWidth) &&
(MemVTWidth <= Width ||
(Align!=0 && MemVTWidth<=AlignInBits && MemVTWidth<=Width+WidenEx))) {
RetVT = MemVT;
break;
}
}
// See if there is a larger vector type to load/store that has the same vector
// element type and is evenly divisible with the WidenVT.
for (VT = (unsigned)MVT::LAST_VECTOR_VALUETYPE;
VT >= (unsigned)MVT::FIRST_VECTOR_VALUETYPE; --VT) {
EVT MemVT = (MVT::SimpleValueType) VT;
unsigned MemVTWidth = MemVT.getSizeInBits();
if (TLI.isTypeLegal(MemVT) && WidenEltVT == MemVT.getVectorElementType() &&
(WidenWidth % MemVTWidth) == 0 &&
isPowerOf2_32(WidenWidth / MemVTWidth) &&
(MemVTWidth <= Width ||
(Align!=0 && MemVTWidth<=AlignInBits && MemVTWidth<=Width+WidenEx))) {
if (RetVT.getSizeInBits() < MemVTWidth || MemVT == WidenVT)
return MemVT;
}
}
return RetVT;
}
// Builds a vector type from scalar loads
// VecTy: Resulting Vector type
// LDOps: Load operators to build a vector type
// [Start,End) the list of loads to use.
static SDValue BuildVectorFromScalar(SelectionDAG& DAG, EVT VecTy,
SmallVector<SDValue, 16>& LdOps,
unsigned Start, unsigned End) {
DebugLoc dl = LdOps[Start].getDebugLoc();
EVT LdTy = LdOps[Start].getValueType();
unsigned Width = VecTy.getSizeInBits();
unsigned NumElts = Width / LdTy.getSizeInBits();
EVT NewVecVT = EVT::getVectorVT(*DAG.getContext(), LdTy, NumElts);
unsigned Idx = 1;
SDValue VecOp = DAG.getNode(ISD::SCALAR_TO_VECTOR, dl, NewVecVT,LdOps[Start]);
for (unsigned i = Start + 1; i != End; ++i) {
EVT NewLdTy = LdOps[i].getValueType();
if (NewLdTy != LdTy) {
NumElts = Width / NewLdTy.getSizeInBits();
NewVecVT = EVT::getVectorVT(*DAG.getContext(), NewLdTy, NumElts);
VecOp = DAG.getNode(ISD::BITCAST, dl, NewVecVT, VecOp);
// Readjust position and vector position based on new load type
Idx = Idx * LdTy.getSizeInBits() / NewLdTy.getSizeInBits();
LdTy = NewLdTy;
}
VecOp = DAG.getNode(ISD::INSERT_VECTOR_ELT, dl, NewVecVT, VecOp, LdOps[i],
DAG.getIntPtrConstant(Idx++));
}
return DAG.getNode(ISD::BITCAST, dl, VecTy, VecOp);
}
SDValue DAGTypeLegalizer::GenWidenVectorLoads(SmallVector<SDValue, 16> &LdChain,
LoadSDNode *LD) {
// The strategy assumes that we can efficiently load powers of two widths.
// The routines chops the vector into the largest vector loads with the same
// element type or scalar loads and then recombines it to the widen vector
// type.
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(),LD->getValueType(0));
unsigned WidenWidth = WidenVT.getSizeInBits();
EVT LdVT = LD->getMemoryVT();
DebugLoc dl = LD->getDebugLoc();
assert(LdVT.isVector() && WidenVT.isVector());
assert(LdVT.getVectorElementType() == WidenVT.getVectorElementType());
// Load information
SDValue Chain = LD->getChain();
SDValue BasePtr = LD->getBasePtr();
unsigned Align = LD->getAlignment();
bool isVolatile = LD->isVolatile();
bool isNonTemporal = LD->isNonTemporal();
bool isInvariant = LD->isInvariant();
int LdWidth = LdVT.getSizeInBits();
int WidthDiff = WidenWidth - LdWidth; // Difference
unsigned LdAlign = (isVolatile) ? 0 : Align; // Allow wider loads
// Find the vector type that can load from.
EVT NewVT = FindMemType(DAG, TLI, LdWidth, WidenVT, LdAlign, WidthDiff);
int NewVTWidth = NewVT.getSizeInBits();
SDValue LdOp = DAG.getLoad(NewVT, dl, Chain, BasePtr, LD->getPointerInfo(),
isVolatile, isNonTemporal, isInvariant, Align);
LdChain.push_back(LdOp.getValue(1));
// Check if we can load the element with one instruction
if (LdWidth <= NewVTWidth) {
if (!NewVT.isVector()) {
unsigned NumElts = WidenWidth / NewVTWidth;
EVT NewVecVT = EVT::getVectorVT(*DAG.getContext(), NewVT, NumElts);
SDValue VecOp = DAG.getNode(ISD::SCALAR_TO_VECTOR, dl, NewVecVT, LdOp);
return DAG.getNode(ISD::BITCAST, dl, WidenVT, VecOp);
}
if (NewVT == WidenVT)
return LdOp;
assert(WidenWidth % NewVTWidth == 0);
unsigned NumConcat = WidenWidth / NewVTWidth;
SmallVector<SDValue, 16> ConcatOps(NumConcat);
SDValue UndefVal = DAG.getUNDEF(NewVT);
ConcatOps[0] = LdOp;
for (unsigned i = 1; i != NumConcat; ++i)
ConcatOps[i] = UndefVal;
return DAG.getNode(ISD::CONCAT_VECTORS, dl, WidenVT, &ConcatOps[0],
NumConcat);
}
// Load vector by using multiple loads from largest vector to scalar
SmallVector<SDValue, 16> LdOps;
LdOps.push_back(LdOp);
LdWidth -= NewVTWidth;
unsigned Offset = 0;
while (LdWidth > 0) {
unsigned Increment = NewVTWidth / 8;
Offset += Increment;
BasePtr = DAG.getNode(ISD::ADD, dl, BasePtr.getValueType(), BasePtr,
DAG.getIntPtrConstant(Increment));
if (LdWidth < NewVTWidth) {
// Our current type we are using is too large, find a better size
NewVT = FindMemType(DAG, TLI, LdWidth, WidenVT, LdAlign, WidthDiff);
NewVTWidth = NewVT.getSizeInBits();
}
SDValue LdOp = DAG.getLoad(NewVT, dl, Chain, BasePtr,
LD->getPointerInfo().getWithOffset(Offset),
isVolatile,
isNonTemporal, isInvariant,
MinAlign(Align, Increment));
LdChain.push_back(LdOp.getValue(1));
LdOps.push_back(LdOp);
LdWidth -= NewVTWidth;
}
// Build the vector from the loads operations
unsigned End = LdOps.size();
if (!LdOps[0].getValueType().isVector())
// All the loads are scalar loads.
return BuildVectorFromScalar(DAG, WidenVT, LdOps, 0, End);
// If the load contains vectors, build the vector using concat vector.
// All of the vectors used to loads are power of 2 and the scalars load
// can be combined to make a power of 2 vector.
SmallVector<SDValue, 16> ConcatOps(End);
int i = End - 1;
int Idx = End;
EVT LdTy = LdOps[i].getValueType();
// First combine the scalar loads to a vector
if (!LdTy.isVector()) {
for (--i; i >= 0; --i) {
LdTy = LdOps[i].getValueType();
if (LdTy.isVector())
break;
}
ConcatOps[--Idx] = BuildVectorFromScalar(DAG, LdTy, LdOps, i+1, End);
}
ConcatOps[--Idx] = LdOps[i];
for (--i; i >= 0; --i) {
EVT NewLdTy = LdOps[i].getValueType();
if (NewLdTy != LdTy) {
// Create a larger vector
ConcatOps[End-1] = DAG.getNode(ISD::CONCAT_VECTORS, dl, NewLdTy,
&ConcatOps[Idx], End - Idx);
Idx = End - 1;
LdTy = NewLdTy;
}
ConcatOps[--Idx] = LdOps[i];
}
if (WidenWidth == LdTy.getSizeInBits()*(End - Idx))
return DAG.getNode(ISD::CONCAT_VECTORS, dl, WidenVT,
&ConcatOps[Idx], End - Idx);
// We need to fill the rest with undefs to build the vector
unsigned NumOps = WidenWidth / LdTy.getSizeInBits();
SmallVector<SDValue, 16> WidenOps(NumOps);
SDValue UndefVal = DAG.getUNDEF(LdTy);
{
unsigned i = 0;
for (; i != End-Idx; ++i)
WidenOps[i] = ConcatOps[Idx+i];
for (; i != NumOps; ++i)
WidenOps[i] = UndefVal;
}
return DAG.getNode(ISD::CONCAT_VECTORS, dl, WidenVT, &WidenOps[0],NumOps);
}
SDValue
DAGTypeLegalizer::GenWidenVectorExtLoads(SmallVector<SDValue, 16>& LdChain,
LoadSDNode * LD,
ISD::LoadExtType ExtType) {
// For extension loads, it may not be more efficient to chop up the vector
// and then extended it. Instead, we unroll the load and build a new vector.
EVT WidenVT = TLI.getTypeToTransformTo(*DAG.getContext(),LD->getValueType(0));
EVT LdVT = LD->getMemoryVT();
DebugLoc dl = LD->getDebugLoc();
assert(LdVT.isVector() && WidenVT.isVector());
// Load information
SDValue Chain = LD->getChain();
SDValue BasePtr = LD->getBasePtr();
unsigned Align = LD->getAlignment();
bool isVolatile = LD->isVolatile();
bool isNonTemporal = LD->isNonTemporal();
EVT EltVT = WidenVT.getVectorElementType();
EVT LdEltVT = LdVT.getVectorElementType();
unsigned NumElts = LdVT.getVectorNumElements();
// Load each element and widen
unsigned WidenNumElts = WidenVT.getVectorNumElements();
SmallVector<SDValue, 16> Ops(WidenNumElts);
unsigned Increment = LdEltVT.getSizeInBits() / 8;
Ops[0] = DAG.getExtLoad(ExtType, dl, EltVT, Chain, BasePtr,
LD->getPointerInfo(),
LdEltVT, isVolatile, isNonTemporal, Align);
LdChain.push_back(Ops[0].getValue(1));
unsigned i = 0, Offset = Increment;
for (i=1; i < NumElts; ++i, Offset += Increment) {
SDValue NewBasePtr = DAG.getNode(ISD::ADD, dl, BasePtr.getValueType(),
BasePtr, DAG.getIntPtrConstant(Offset));
Ops[i] = DAG.getExtLoad(ExtType, dl, EltVT, Chain, NewBasePtr,
LD->getPointerInfo().getWithOffset(Offset), LdEltVT,
isVolatile, isNonTemporal, Align);
LdChain.push_back(Ops[i].getValue(1));
}
// Fill the rest with undefs
SDValue UndefVal = DAG.getUNDEF(EltVT);
for (; i != WidenNumElts; ++i)
Ops[i] = UndefVal;
return DAG.getNode(ISD::BUILD_VECTOR, dl, WidenVT, &Ops[0], Ops.size());
}
void DAGTypeLegalizer::GenWidenVectorStores(SmallVector<SDValue, 16>& StChain,
StoreSDNode *ST) {
// The strategy assumes that we can efficiently store powers of two widths.
// The routines chops the vector into the largest vector stores with the same
// element type or scalar stores.
SDValue Chain = ST->getChain();
SDValue BasePtr = ST->getBasePtr();
unsigned Align = ST->getAlignment();
bool isVolatile = ST->isVolatile();
bool isNonTemporal = ST->isNonTemporal();
SDValue ValOp = GetWidenedVector(ST->getValue());
DebugLoc dl = ST->getDebugLoc();
EVT StVT = ST->getMemoryVT();
unsigned StWidth = StVT.getSizeInBits();
EVT ValVT = ValOp.getValueType();
unsigned ValWidth = ValVT.getSizeInBits();
EVT ValEltVT = ValVT.getVectorElementType();
unsigned ValEltWidth = ValEltVT.getSizeInBits();
assert(StVT.getVectorElementType() == ValEltVT);
int Idx = 0; // current index to store
unsigned Offset = 0; // offset from base to store
while (StWidth != 0) {
// Find the largest vector type we can store with
EVT NewVT = FindMemType(DAG, TLI, StWidth, ValVT);
unsigned NewVTWidth = NewVT.getSizeInBits();
unsigned Increment = NewVTWidth / 8;
if (NewVT.isVector()) {
unsigned NumVTElts = NewVT.getVectorNumElements();
do {
SDValue EOp = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, NewVT, ValOp,
DAG.getIntPtrConstant(Idx));
StChain.push_back(DAG.getStore(Chain, dl, EOp, BasePtr,
ST->getPointerInfo().getWithOffset(Offset),
isVolatile, isNonTemporal,
MinAlign(Align, Offset)));
StWidth -= NewVTWidth;
Offset += Increment;
Idx += NumVTElts;
BasePtr = DAG.getNode(ISD::ADD, dl, BasePtr.getValueType(), BasePtr,
DAG.getIntPtrConstant(Increment));
} while (StWidth != 0 && StWidth >= NewVTWidth);
} else {
// Cast the vector to the scalar type we can store
unsigned NumElts = ValWidth / NewVTWidth;
EVT NewVecVT = EVT::getVectorVT(*DAG.getContext(), NewVT, NumElts);
SDValue VecOp = DAG.getNode(ISD::BITCAST, dl, NewVecVT, ValOp);
// Readjust index position based on new vector type
Idx = Idx * ValEltWidth / NewVTWidth;
do {
SDValue EOp = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, NewVT, VecOp,
DAG.getIntPtrConstant(Idx++));
StChain.push_back(DAG.getStore(Chain, dl, EOp, BasePtr,
ST->getPointerInfo().getWithOffset(Offset),
isVolatile, isNonTemporal,
MinAlign(Align, Offset)));
StWidth -= NewVTWidth;
Offset += Increment;
BasePtr = DAG.getNode(ISD::ADD, dl, BasePtr.getValueType(), BasePtr,
DAG.getIntPtrConstant(Increment));
} while (StWidth != 0 && StWidth >= NewVTWidth);
// Restore index back to be relative to the original widen element type
Idx = Idx * NewVTWidth / ValEltWidth;
}
}
}
void
DAGTypeLegalizer::GenWidenVectorTruncStores(SmallVector<SDValue, 16>& StChain,
StoreSDNode *ST) {
// For extension loads, it may not be more efficient to truncate the vector
// and then store it. Instead, we extract each element and then store it.
SDValue Chain = ST->getChain();
SDValue BasePtr = ST->getBasePtr();
unsigned Align = ST->getAlignment();
bool isVolatile = ST->isVolatile();
bool isNonTemporal = ST->isNonTemporal();
SDValue ValOp = GetWidenedVector(ST->getValue());
DebugLoc dl = ST->getDebugLoc();
EVT StVT = ST->getMemoryVT();
EVT ValVT = ValOp.getValueType();
// It must be true that we the widen vector type is bigger than where
// we need to store.
assert(StVT.isVector() && ValOp.getValueType().isVector());
assert(StVT.bitsLT(ValOp.getValueType()));
// For truncating stores, we can not play the tricks of chopping legal
// vector types and bit cast it to the right type. Instead, we unroll
// the store.
EVT StEltVT = StVT.getVectorElementType();
EVT ValEltVT = ValVT.getVectorElementType();
unsigned Increment = ValEltVT.getSizeInBits() / 8;
unsigned NumElts = StVT.getVectorNumElements();
SDValue EOp = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, ValEltVT, ValOp,
DAG.getIntPtrConstant(0));
StChain.push_back(DAG.getTruncStore(Chain, dl, EOp, BasePtr,
ST->getPointerInfo(), StEltVT,
isVolatile, isNonTemporal, Align));
unsigned Offset = Increment;
for (unsigned i=1; i < NumElts; ++i, Offset += Increment) {
SDValue NewBasePtr = DAG.getNode(ISD::ADD, dl, BasePtr.getValueType(),
BasePtr, DAG.getIntPtrConstant(Offset));
SDValue EOp = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, ValEltVT, ValOp,
DAG.getIntPtrConstant(0));
StChain.push_back(DAG.getTruncStore(Chain, dl, EOp, NewBasePtr,
ST->getPointerInfo().getWithOffset(Offset),
StEltVT, isVolatile, isNonTemporal,
MinAlign(Align, Offset)));
}
}
/// Modifies a vector input (widen or narrows) to a vector of NVT. The
/// input vector must have the same element type as NVT.
SDValue DAGTypeLegalizer::ModifyToType(SDValue InOp, EVT NVT) {
// Note that InOp might have been widened so it might already have
// the right width or it might need be narrowed.
EVT InVT = InOp.getValueType();
assert(InVT.getVectorElementType() == NVT.getVectorElementType() &&
"input and widen element type must match");
DebugLoc dl = InOp.getDebugLoc();
// Check if InOp already has the right width.
if (InVT == NVT)
return InOp;
unsigned InNumElts = InVT.getVectorNumElements();
unsigned WidenNumElts = NVT.getVectorNumElements();
if (WidenNumElts > InNumElts && WidenNumElts % InNumElts == 0) {
unsigned NumConcat = WidenNumElts / InNumElts;
SmallVector<SDValue, 16> Ops(NumConcat);
SDValue UndefVal = DAG.getUNDEF(InVT);
Ops[0] = InOp;
for (unsigned i = 1; i != NumConcat; ++i)
Ops[i] = UndefVal;
return DAG.getNode(ISD::CONCAT_VECTORS, dl, NVT, &Ops[0], NumConcat);
}
if (WidenNumElts < InNumElts && InNumElts % WidenNumElts)
return DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, NVT, InOp,
DAG.getIntPtrConstant(0));
// Fall back to extract and build.
SmallVector<SDValue, 16> Ops(WidenNumElts);
EVT EltVT = NVT.getVectorElementType();
unsigned MinNumElts = std::min(WidenNumElts, InNumElts);
unsigned Idx;
for (Idx = 0; Idx < MinNumElts; ++Idx)
Ops[Idx] = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, EltVT, InOp,
DAG.getIntPtrConstant(Idx));
SDValue UndefVal = DAG.getUNDEF(EltVT);
for ( ; Idx < WidenNumElts; ++Idx)
Ops[Idx] = UndefVal;
return DAG.getNode(ISD::BUILD_VECTOR, dl, NVT, &Ops[0], WidenNumElts);
}
|