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
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
|
//===- llvm/CodeGen/GlobalISel/IRTranslator.cpp - IRTranslator ---*- C++ -*-==//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
/// \file
/// This file implements the IRTranslator class.
//===----------------------------------------------------------------------===//
#include "llvm/CodeGen/GlobalISel/IRTranslator.h"
#include "llvm/ADT/PostOrderIterator.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/ScopeExit.h"
#include "llvm/ADT/SmallSet.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/Analysis/AliasAnalysis.h"
#include "llvm/Analysis/AssumptionCache.h"
#include "llvm/Analysis/BranchProbabilityInfo.h"
#include "llvm/Analysis/Loads.h"
#include "llvm/Analysis/OptimizationRemarkEmitter.h"
#include "llvm/Analysis/ValueTracking.h"
#include "llvm/CodeGen/Analysis.h"
#include "llvm/CodeGen/GlobalISel/CSEInfo.h"
#include "llvm/CodeGen/GlobalISel/CSEMIRBuilder.h"
#include "llvm/CodeGen/GlobalISel/CallLowering.h"
#include "llvm/CodeGen/GlobalISel/GISelChangeObserver.h"
#include "llvm/CodeGen/GlobalISel/InlineAsmLowering.h"
#include "llvm/CodeGen/GlobalISel/MachineIRBuilder.h"
#include "llvm/CodeGen/LowLevelTypeUtils.h"
#include "llvm/CodeGen/MachineBasicBlock.h"
#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineMemOperand.h"
#include "llvm/CodeGen/MachineModuleInfo.h"
#include "llvm/CodeGen/MachineOperand.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
#include "llvm/CodeGen/RuntimeLibcalls.h"
#include "llvm/CodeGen/StackProtector.h"
#include "llvm/CodeGen/SwitchLoweringUtils.h"
#include "llvm/CodeGen/TargetFrameLowering.h"
#include "llvm/CodeGen/TargetInstrInfo.h"
#include "llvm/CodeGen/TargetLowering.h"
#include "llvm/CodeGen/TargetOpcodes.h"
#include "llvm/CodeGen/TargetPassConfig.h"
#include "llvm/CodeGen/TargetRegisterInfo.h"
#include "llvm/CodeGen/TargetSubtargetInfo.h"
#include "llvm/CodeGenTypes/LowLevelType.h"
#include "llvm/IR/BasicBlock.h"
#include "llvm/IR/CFG.h"
#include "llvm/IR/Constant.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/DiagnosticInfo.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/GetElementPtrTypeIterator.h"
#include "llvm/IR/InlineAsm.h"
#include "llvm/IR/InstrTypes.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/Intrinsics.h"
#include "llvm/IR/IntrinsicsAMDGPU.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Metadata.h"
#include "llvm/IR/PatternMatch.h"
#include "llvm/IR/Statepoint.h"
#include "llvm/IR/Type.h"
#include "llvm/IR/User.h"
#include "llvm/IR/Value.h"
#include "llvm/InitializePasses.h"
#include "llvm/MC/MCContext.h"
#include "llvm/Pass.h"
#include "llvm/Support/Casting.h"
#include "llvm/Support/CodeGen.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/MathExtras.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Target/TargetIntrinsicInfo.h"
#include "llvm/Target/TargetMachine.h"
#include "llvm/Transforms/Utils/Local.h"
#include "llvm/Transforms/Utils/MemoryOpRemark.h"
#include <algorithm>
#include <cassert>
#include <cstdint>
#include <iterator>
#include <optional>
#include <string>
#include <utility>
#include <vector>
#define DEBUG_TYPE "irtranslator"
using namespace llvm;
static cl::opt<bool>
EnableCSEInIRTranslator("enable-cse-in-irtranslator",
cl::desc("Should enable CSE in irtranslator"),
cl::Optional, cl::init(false));
char IRTranslator::ID = 0;
INITIALIZE_PASS_BEGIN(IRTranslator, DEBUG_TYPE, "IRTranslator LLVM IR -> MI",
false, false)
INITIALIZE_PASS_DEPENDENCY(TargetPassConfig)
INITIALIZE_PASS_DEPENDENCY(GISelCSEAnalysisWrapperPass)
INITIALIZE_PASS_DEPENDENCY(BlockFrequencyInfoWrapperPass)
INITIALIZE_PASS_DEPENDENCY(StackProtector)
INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)
INITIALIZE_PASS_END(IRTranslator, DEBUG_TYPE, "IRTranslator LLVM IR -> MI",
false, false)
static void reportTranslationError(MachineFunction &MF,
const TargetPassConfig &TPC,
OptimizationRemarkEmitter &ORE,
OptimizationRemarkMissed &R) {
MF.getProperties().set(MachineFunctionProperties::Property::FailedISel);
// Print the function name explicitly if we don't have a debug location (which
// makes the diagnostic less useful) or if we're going to emit a raw error.
if (!R.getLocation().isValid() || TPC.isGlobalISelAbortEnabled())
R << (" (in function: " + MF.getName() + ")").str();
if (TPC.isGlobalISelAbortEnabled())
report_fatal_error(Twine(R.getMsg()));
else
ORE.emit(R);
}
IRTranslator::IRTranslator(CodeGenOptLevel optlevel)
: MachineFunctionPass(ID), OptLevel(optlevel) {}
#ifndef NDEBUG
namespace {
/// Verify that every instruction created has the same DILocation as the
/// instruction being translated.
class DILocationVerifier : public GISelChangeObserver {
const Instruction *CurrInst = nullptr;
public:
DILocationVerifier() = default;
~DILocationVerifier() = default;
const Instruction *getCurrentInst() const { return CurrInst; }
void setCurrentInst(const Instruction *Inst) { CurrInst = Inst; }
void erasingInstr(MachineInstr &MI) override {}
void changingInstr(MachineInstr &MI) override {}
void changedInstr(MachineInstr &MI) override {}
void createdInstr(MachineInstr &MI) override {
assert(getCurrentInst() && "Inserted instruction without a current MI");
// Only print the check message if we're actually checking it.
#ifndef NDEBUG
LLVM_DEBUG(dbgs() << "Checking DILocation from " << *CurrInst
<< " was copied to " << MI);
#endif
// We allow insts in the entry block to have no debug loc because
// they could have originated from constants, and we don't want a jumpy
// debug experience.
assert((CurrInst->getDebugLoc() == MI.getDebugLoc() ||
(MI.getParent()->isEntryBlock() && !MI.getDebugLoc()) ||
(MI.isDebugInstr())) &&
"Line info was not transferred to all instructions");
}
};
} // namespace
#endif // ifndef NDEBUG
void IRTranslator::getAnalysisUsage(AnalysisUsage &AU) const {
AU.addRequired<StackProtector>();
AU.addRequired<TargetPassConfig>();
AU.addRequired<GISelCSEAnalysisWrapperPass>();
AU.addRequired<AssumptionCacheTracker>();
if (OptLevel != CodeGenOptLevel::None) {
AU.addRequired<BranchProbabilityInfoWrapperPass>();
AU.addRequired<AAResultsWrapperPass>();
}
AU.addRequired<TargetLibraryInfoWrapperPass>();
AU.addPreserved<TargetLibraryInfoWrapperPass>();
getSelectionDAGFallbackAnalysisUsage(AU);
MachineFunctionPass::getAnalysisUsage(AU);
}
IRTranslator::ValueToVRegInfo::VRegListT &
IRTranslator::allocateVRegs(const Value &Val) {
auto VRegsIt = VMap.findVRegs(Val);
if (VRegsIt != VMap.vregs_end())
return *VRegsIt->second;
auto *Regs = VMap.getVRegs(Val);
auto *Offsets = VMap.getOffsets(Val);
SmallVector<LLT, 4> SplitTys;
computeValueLLTs(*DL, *Val.getType(), SplitTys,
Offsets->empty() ? Offsets : nullptr);
for (unsigned i = 0; i < SplitTys.size(); ++i)
Regs->push_back(0);
return *Regs;
}
ArrayRef<Register> IRTranslator::getOrCreateVRegs(const Value &Val) {
auto VRegsIt = VMap.findVRegs(Val);
if (VRegsIt != VMap.vregs_end())
return *VRegsIt->second;
if (Val.getType()->isVoidTy())
return *VMap.getVRegs(Val);
// Create entry for this type.
auto *VRegs = VMap.getVRegs(Val);
auto *Offsets = VMap.getOffsets(Val);
assert(Val.getType()->isSized() &&
"Don't know how to create an empty vreg");
SmallVector<LLT, 4> SplitTys;
computeValueLLTs(*DL, *Val.getType(), SplitTys,
Offsets->empty() ? Offsets : nullptr);
if (!isa<Constant>(Val)) {
for (auto Ty : SplitTys)
VRegs->push_back(MRI->createGenericVirtualRegister(Ty));
return *VRegs;
}
if (Val.getType()->isAggregateType()) {
// UndefValue, ConstantAggregateZero
auto &C = cast<Constant>(Val);
unsigned Idx = 0;
while (auto Elt = C.getAggregateElement(Idx++)) {
auto EltRegs = getOrCreateVRegs(*Elt);
llvm::copy(EltRegs, std::back_inserter(*VRegs));
}
} else {
assert(SplitTys.size() == 1 && "unexpectedly split LLT");
VRegs->push_back(MRI->createGenericVirtualRegister(SplitTys[0]));
bool Success = translate(cast<Constant>(Val), VRegs->front());
if (!Success) {
OptimizationRemarkMissed R("gisel-irtranslator", "GISelFailure",
MF->getFunction().getSubprogram(),
&MF->getFunction().getEntryBlock());
R << "unable to translate constant: " << ore::NV("Type", Val.getType());
reportTranslationError(*MF, *TPC, *ORE, R);
return *VRegs;
}
}
return *VRegs;
}
int IRTranslator::getOrCreateFrameIndex(const AllocaInst &AI) {
auto MapEntry = FrameIndices.find(&AI);
if (MapEntry != FrameIndices.end())
return MapEntry->second;
uint64_t ElementSize = DL->getTypeAllocSize(AI.getAllocatedType());
uint64_t Size =
ElementSize * cast<ConstantInt>(AI.getArraySize())->getZExtValue();
// Always allocate at least one byte.
Size = std::max<uint64_t>(Size, 1u);
int &FI = FrameIndices[&AI];
FI = MF->getFrameInfo().CreateStackObject(Size, AI.getAlign(), false, &AI);
return FI;
}
Align IRTranslator::getMemOpAlign(const Instruction &I) {
if (const StoreInst *SI = dyn_cast<StoreInst>(&I))
return SI->getAlign();
if (const LoadInst *LI = dyn_cast<LoadInst>(&I))
return LI->getAlign();
if (const AtomicCmpXchgInst *AI = dyn_cast<AtomicCmpXchgInst>(&I))
return AI->getAlign();
if (const AtomicRMWInst *AI = dyn_cast<AtomicRMWInst>(&I))
return AI->getAlign();
OptimizationRemarkMissed R("gisel-irtranslator", "", &I);
R << "unable to translate memop: " << ore::NV("Opcode", &I);
reportTranslationError(*MF, *TPC, *ORE, R);
return Align(1);
}
MachineBasicBlock &IRTranslator::getMBB(const BasicBlock &BB) {
MachineBasicBlock *&MBB = BBToMBB[&BB];
assert(MBB && "BasicBlock was not encountered before");
return *MBB;
}
void IRTranslator::addMachineCFGPred(CFGEdge Edge, MachineBasicBlock *NewPred) {
assert(NewPred && "new predecessor must be a real MachineBasicBlock");
MachinePreds[Edge].push_back(NewPred);
}
bool IRTranslator::translateBinaryOp(unsigned Opcode, const User &U,
MachineIRBuilder &MIRBuilder) {
// Get or create a virtual register for each value.
// Unless the value is a Constant => loadimm cst?
// or inline constant each time?
// Creation of a virtual register needs to have a size.
Register Op0 = getOrCreateVReg(*U.getOperand(0));
Register Op1 = getOrCreateVReg(*U.getOperand(1));
Register Res = getOrCreateVReg(U);
uint32_t Flags = 0;
if (isa<Instruction>(U)) {
const Instruction &I = cast<Instruction>(U);
Flags = MachineInstr::copyFlagsFromInstruction(I);
}
MIRBuilder.buildInstr(Opcode, {Res}, {Op0, Op1}, Flags);
return true;
}
bool IRTranslator::translateUnaryOp(unsigned Opcode, const User &U,
MachineIRBuilder &MIRBuilder) {
Register Op0 = getOrCreateVReg(*U.getOperand(0));
Register Res = getOrCreateVReg(U);
uint32_t Flags = 0;
if (isa<Instruction>(U)) {
const Instruction &I = cast<Instruction>(U);
Flags = MachineInstr::copyFlagsFromInstruction(I);
}
MIRBuilder.buildInstr(Opcode, {Res}, {Op0}, Flags);
return true;
}
bool IRTranslator::translateFNeg(const User &U, MachineIRBuilder &MIRBuilder) {
return translateUnaryOp(TargetOpcode::G_FNEG, U, MIRBuilder);
}
bool IRTranslator::translateCompare(const User &U,
MachineIRBuilder &MIRBuilder) {
auto *CI = dyn_cast<CmpInst>(&U);
Register Op0 = getOrCreateVReg(*U.getOperand(0));
Register Op1 = getOrCreateVReg(*U.getOperand(1));
Register Res = getOrCreateVReg(U);
CmpInst::Predicate Pred =
CI ? CI->getPredicate() : static_cast<CmpInst::Predicate>(
cast<ConstantExpr>(U).getPredicate());
if (CmpInst::isIntPredicate(Pred))
MIRBuilder.buildICmp(Pred, Res, Op0, Op1);
else if (Pred == CmpInst::FCMP_FALSE)
MIRBuilder.buildCopy(
Res, getOrCreateVReg(*Constant::getNullValue(U.getType())));
else if (Pred == CmpInst::FCMP_TRUE)
MIRBuilder.buildCopy(
Res, getOrCreateVReg(*Constant::getAllOnesValue(U.getType())));
else {
uint32_t Flags = 0;
if (CI)
Flags = MachineInstr::copyFlagsFromInstruction(*CI);
MIRBuilder.buildFCmp(Pred, Res, Op0, Op1, Flags);
}
return true;
}
bool IRTranslator::translateRet(const User &U, MachineIRBuilder &MIRBuilder) {
const ReturnInst &RI = cast<ReturnInst>(U);
const Value *Ret = RI.getReturnValue();
if (Ret && DL->getTypeStoreSize(Ret->getType()).isZero())
Ret = nullptr;
ArrayRef<Register> VRegs;
if (Ret)
VRegs = getOrCreateVRegs(*Ret);
Register SwiftErrorVReg = 0;
if (CLI->supportSwiftError() && SwiftError.getFunctionArg()) {
SwiftErrorVReg = SwiftError.getOrCreateVRegUseAt(
&RI, &MIRBuilder.getMBB(), SwiftError.getFunctionArg());
}
// The target may mess up with the insertion point, but
// this is not important as a return is the last instruction
// of the block anyway.
return CLI->lowerReturn(MIRBuilder, Ret, VRegs, FuncInfo, SwiftErrorVReg);
}
void IRTranslator::emitBranchForMergedCondition(
const Value *Cond, MachineBasicBlock *TBB, MachineBasicBlock *FBB,
MachineBasicBlock *CurBB, MachineBasicBlock *SwitchBB,
BranchProbability TProb, BranchProbability FProb, bool InvertCond) {
// If the leaf of the tree is a comparison, merge the condition into
// the caseblock.
if (const CmpInst *BOp = dyn_cast<CmpInst>(Cond)) {
CmpInst::Predicate Condition;
if (const ICmpInst *IC = dyn_cast<ICmpInst>(Cond)) {
Condition = InvertCond ? IC->getInversePredicate() : IC->getPredicate();
} else {
const FCmpInst *FC = cast<FCmpInst>(Cond);
Condition = InvertCond ? FC->getInversePredicate() : FC->getPredicate();
}
SwitchCG::CaseBlock CB(Condition, false, BOp->getOperand(0),
BOp->getOperand(1), nullptr, TBB, FBB, CurBB,
CurBuilder->getDebugLoc(), TProb, FProb);
SL->SwitchCases.push_back(CB);
return;
}
// Create a CaseBlock record representing this branch.
CmpInst::Predicate Pred = InvertCond ? CmpInst::ICMP_NE : CmpInst::ICMP_EQ;
SwitchCG::CaseBlock CB(
Pred, false, Cond, ConstantInt::getTrue(MF->getFunction().getContext()),
nullptr, TBB, FBB, CurBB, CurBuilder->getDebugLoc(), TProb, FProb);
SL->SwitchCases.push_back(CB);
}
static bool isValInBlock(const Value *V, const BasicBlock *BB) {
if (const Instruction *I = dyn_cast<Instruction>(V))
return I->getParent() == BB;
return true;
}
void IRTranslator::findMergedConditions(
const Value *Cond, MachineBasicBlock *TBB, MachineBasicBlock *FBB,
MachineBasicBlock *CurBB, MachineBasicBlock *SwitchBB,
Instruction::BinaryOps Opc, BranchProbability TProb,
BranchProbability FProb, bool InvertCond) {
using namespace PatternMatch;
assert((Opc == Instruction::And || Opc == Instruction::Or) &&
"Expected Opc to be AND/OR");
// Skip over not part of the tree and remember to invert op and operands at
// next level.
Value *NotCond;
if (match(Cond, m_OneUse(m_Not(m_Value(NotCond)))) &&
isValInBlock(NotCond, CurBB->getBasicBlock())) {
findMergedConditions(NotCond, TBB, FBB, CurBB, SwitchBB, Opc, TProb, FProb,
!InvertCond);
return;
}
const Instruction *BOp = dyn_cast<Instruction>(Cond);
const Value *BOpOp0, *BOpOp1;
// Compute the effective opcode for Cond, taking into account whether it needs
// to be inverted, e.g.
// and (not (or A, B)), C
// gets lowered as
// and (and (not A, not B), C)
Instruction::BinaryOps BOpc = (Instruction::BinaryOps)0;
if (BOp) {
BOpc = match(BOp, m_LogicalAnd(m_Value(BOpOp0), m_Value(BOpOp1)))
? Instruction::And
: (match(BOp, m_LogicalOr(m_Value(BOpOp0), m_Value(BOpOp1)))
? Instruction::Or
: (Instruction::BinaryOps)0);
if (InvertCond) {
if (BOpc == Instruction::And)
BOpc = Instruction::Or;
else if (BOpc == Instruction::Or)
BOpc = Instruction::And;
}
}
// If this node is not part of the or/and tree, emit it as a branch.
// Note that all nodes in the tree should have same opcode.
bool BOpIsInOrAndTree = BOpc && BOpc == Opc && BOp->hasOneUse();
if (!BOpIsInOrAndTree || BOp->getParent() != CurBB->getBasicBlock() ||
!isValInBlock(BOpOp0, CurBB->getBasicBlock()) ||
!isValInBlock(BOpOp1, CurBB->getBasicBlock())) {
emitBranchForMergedCondition(Cond, TBB, FBB, CurBB, SwitchBB, TProb, FProb,
InvertCond);
return;
}
// Create TmpBB after CurBB.
MachineFunction::iterator BBI(CurBB);
MachineBasicBlock *TmpBB =
MF->CreateMachineBasicBlock(CurBB->getBasicBlock());
CurBB->getParent()->insert(++BBI, TmpBB);
if (Opc == Instruction::Or) {
// Codegen X | Y as:
// BB1:
// jmp_if_X TBB
// jmp TmpBB
// TmpBB:
// jmp_if_Y TBB
// jmp FBB
//
// We have flexibility in setting Prob for BB1 and Prob for TmpBB.
// The requirement is that
// TrueProb for BB1 + (FalseProb for BB1 * TrueProb for TmpBB)
// = TrueProb for original BB.
// Assuming the original probabilities are A and B, one choice is to set
// BB1's probabilities to A/2 and A/2+B, and set TmpBB's probabilities to
// A/(1+B) and 2B/(1+B). This choice assumes that
// TrueProb for BB1 == FalseProb for BB1 * TrueProb for TmpBB.
// Another choice is to assume TrueProb for BB1 equals to TrueProb for
// TmpBB, but the math is more complicated.
auto NewTrueProb = TProb / 2;
auto NewFalseProb = TProb / 2 + FProb;
// Emit the LHS condition.
findMergedConditions(BOpOp0, TBB, TmpBB, CurBB, SwitchBB, Opc, NewTrueProb,
NewFalseProb, InvertCond);
// Normalize A/2 and B to get A/(1+B) and 2B/(1+B).
SmallVector<BranchProbability, 2> Probs{TProb / 2, FProb};
BranchProbability::normalizeProbabilities(Probs.begin(), Probs.end());
// Emit the RHS condition into TmpBB.
findMergedConditions(BOpOp1, TBB, FBB, TmpBB, SwitchBB, Opc, Probs[0],
Probs[1], InvertCond);
} else {
assert(Opc == Instruction::And && "Unknown merge op!");
// Codegen X & Y as:
// BB1:
// jmp_if_X TmpBB
// jmp FBB
// TmpBB:
// jmp_if_Y TBB
// jmp FBB
//
// This requires creation of TmpBB after CurBB.
// We have flexibility in setting Prob for BB1 and Prob for TmpBB.
// The requirement is that
// FalseProb for BB1 + (TrueProb for BB1 * FalseProb for TmpBB)
// = FalseProb for original BB.
// Assuming the original probabilities are A and B, one choice is to set
// BB1's probabilities to A+B/2 and B/2, and set TmpBB's probabilities to
// 2A/(1+A) and B/(1+A). This choice assumes that FalseProb for BB1 ==
// TrueProb for BB1 * FalseProb for TmpBB.
auto NewTrueProb = TProb + FProb / 2;
auto NewFalseProb = FProb / 2;
// Emit the LHS condition.
findMergedConditions(BOpOp0, TmpBB, FBB, CurBB, SwitchBB, Opc, NewTrueProb,
NewFalseProb, InvertCond);
// Normalize A and B/2 to get 2A/(1+A) and B/(1+A).
SmallVector<BranchProbability, 2> Probs{TProb, FProb / 2};
BranchProbability::normalizeProbabilities(Probs.begin(), Probs.end());
// Emit the RHS condition into TmpBB.
findMergedConditions(BOpOp1, TBB, FBB, TmpBB, SwitchBB, Opc, Probs[0],
Probs[1], InvertCond);
}
}
bool IRTranslator::shouldEmitAsBranches(
const std::vector<SwitchCG::CaseBlock> &Cases) {
// For multiple cases, it's better to emit as branches.
if (Cases.size() != 2)
return true;
// If this is two comparisons of the same values or'd or and'd together, they
// will get folded into a single comparison, so don't emit two blocks.
if ((Cases[0].CmpLHS == Cases[1].CmpLHS &&
Cases[0].CmpRHS == Cases[1].CmpRHS) ||
(Cases[0].CmpRHS == Cases[1].CmpLHS &&
Cases[0].CmpLHS == Cases[1].CmpRHS)) {
return false;
}
// Handle: (X != null) | (Y != null) --> (X|Y) != 0
// Handle: (X == null) & (Y == null) --> (X|Y) == 0
if (Cases[0].CmpRHS == Cases[1].CmpRHS &&
Cases[0].PredInfo.Pred == Cases[1].PredInfo.Pred &&
isa<Constant>(Cases[0].CmpRHS) &&
cast<Constant>(Cases[0].CmpRHS)->isNullValue()) {
if (Cases[0].PredInfo.Pred == CmpInst::ICMP_EQ &&
Cases[0].TrueBB == Cases[1].ThisBB)
return false;
if (Cases[0].PredInfo.Pred == CmpInst::ICMP_NE &&
Cases[0].FalseBB == Cases[1].ThisBB)
return false;
}
return true;
}
bool IRTranslator::translateBr(const User &U, MachineIRBuilder &MIRBuilder) {
const BranchInst &BrInst = cast<BranchInst>(U);
auto &CurMBB = MIRBuilder.getMBB();
auto *Succ0MBB = &getMBB(*BrInst.getSuccessor(0));
if (BrInst.isUnconditional()) {
// If the unconditional target is the layout successor, fallthrough.
if (OptLevel == CodeGenOptLevel::None ||
!CurMBB.isLayoutSuccessor(Succ0MBB))
MIRBuilder.buildBr(*Succ0MBB);
// Link successors.
for (const BasicBlock *Succ : successors(&BrInst))
CurMBB.addSuccessor(&getMBB(*Succ));
return true;
}
// If this condition is one of the special cases we handle, do special stuff
// now.
const Value *CondVal = BrInst.getCondition();
MachineBasicBlock *Succ1MBB = &getMBB(*BrInst.getSuccessor(1));
const auto &TLI = *MF->getSubtarget().getTargetLowering();
// If this is a series of conditions that are or'd or and'd together, emit
// this as a sequence of branches instead of setcc's with and/or operations.
// As long as jumps are not expensive (exceptions for multi-use logic ops,
// unpredictable branches, and vector extracts because those jumps are likely
// expensive for any target), this should improve performance.
// For example, instead of something like:
// cmp A, B
// C = seteq
// cmp D, E
// F = setle
// or C, F
// jnz foo
// Emit:
// cmp A, B
// je foo
// cmp D, E
// jle foo
using namespace PatternMatch;
const Instruction *CondI = dyn_cast<Instruction>(CondVal);
if (!TLI.isJumpExpensive() && CondI && CondI->hasOneUse() &&
!BrInst.hasMetadata(LLVMContext::MD_unpredictable)) {
Instruction::BinaryOps Opcode = (Instruction::BinaryOps)0;
Value *Vec;
const Value *BOp0, *BOp1;
if (match(CondI, m_LogicalAnd(m_Value(BOp0), m_Value(BOp1))))
Opcode = Instruction::And;
else if (match(CondI, m_LogicalOr(m_Value(BOp0), m_Value(BOp1))))
Opcode = Instruction::Or;
if (Opcode && !(match(BOp0, m_ExtractElt(m_Value(Vec), m_Value())) &&
match(BOp1, m_ExtractElt(m_Specific(Vec), m_Value())))) {
findMergedConditions(CondI, Succ0MBB, Succ1MBB, &CurMBB, &CurMBB, Opcode,
getEdgeProbability(&CurMBB, Succ0MBB),
getEdgeProbability(&CurMBB, Succ1MBB),
/*InvertCond=*/false);
assert(SL->SwitchCases[0].ThisBB == &CurMBB && "Unexpected lowering!");
// Allow some cases to be rejected.
if (shouldEmitAsBranches(SL->SwitchCases)) {
// Emit the branch for this block.
emitSwitchCase(SL->SwitchCases[0], &CurMBB, *CurBuilder);
SL->SwitchCases.erase(SL->SwitchCases.begin());
return true;
}
// Okay, we decided not to do this, remove any inserted MBB's and clear
// SwitchCases.
for (unsigned I = 1, E = SL->SwitchCases.size(); I != E; ++I)
MF->erase(SL->SwitchCases[I].ThisBB);
SL->SwitchCases.clear();
}
}
// Create a CaseBlock record representing this branch.
SwitchCG::CaseBlock CB(CmpInst::ICMP_EQ, false, CondVal,
ConstantInt::getTrue(MF->getFunction().getContext()),
nullptr, Succ0MBB, Succ1MBB, &CurMBB,
CurBuilder->getDebugLoc());
// Use emitSwitchCase to actually insert the fast branch sequence for this
// cond branch.
emitSwitchCase(CB, &CurMBB, *CurBuilder);
return true;
}
void IRTranslator::addSuccessorWithProb(MachineBasicBlock *Src,
MachineBasicBlock *Dst,
BranchProbability Prob) {
if (!FuncInfo.BPI) {
Src->addSuccessorWithoutProb(Dst);
return;
}
if (Prob.isUnknown())
Prob = getEdgeProbability(Src, Dst);
Src->addSuccessor(Dst, Prob);
}
BranchProbability
IRTranslator::getEdgeProbability(const MachineBasicBlock *Src,
const MachineBasicBlock *Dst) const {
const BasicBlock *SrcBB = Src->getBasicBlock();
const BasicBlock *DstBB = Dst->getBasicBlock();
if (!FuncInfo.BPI) {
// If BPI is not available, set the default probability as 1 / N, where N is
// the number of successors.
auto SuccSize = std::max<uint32_t>(succ_size(SrcBB), 1);
return BranchProbability(1, SuccSize);
}
return FuncInfo.BPI->getEdgeProbability(SrcBB, DstBB);
}
bool IRTranslator::translateSwitch(const User &U, MachineIRBuilder &MIB) {
using namespace SwitchCG;
// Extract cases from the switch.
const SwitchInst &SI = cast<SwitchInst>(U);
BranchProbabilityInfo *BPI = FuncInfo.BPI;
CaseClusterVector Clusters;
Clusters.reserve(SI.getNumCases());
for (const auto &I : SI.cases()) {
MachineBasicBlock *Succ = &getMBB(*I.getCaseSuccessor());
assert(Succ && "Could not find successor mbb in mapping");
const ConstantInt *CaseVal = I.getCaseValue();
BranchProbability Prob =
BPI ? BPI->getEdgeProbability(SI.getParent(), I.getSuccessorIndex())
: BranchProbability(1, SI.getNumCases() + 1);
Clusters.push_back(CaseCluster::range(CaseVal, CaseVal, Succ, Prob));
}
MachineBasicBlock *DefaultMBB = &getMBB(*SI.getDefaultDest());
// Cluster adjacent cases with the same destination. We do this at all
// optimization levels because it's cheap to do and will make codegen faster
// if there are many clusters.
sortAndRangeify(Clusters);
MachineBasicBlock *SwitchMBB = &getMBB(*SI.getParent());
// If there is only the default destination, jump there directly.
if (Clusters.empty()) {
SwitchMBB->addSuccessor(DefaultMBB);
if (DefaultMBB != SwitchMBB->getNextNode())
MIB.buildBr(*DefaultMBB);
return true;
}
SL->findJumpTables(Clusters, &SI, std::nullopt, DefaultMBB, nullptr, nullptr);
SL->findBitTestClusters(Clusters, &SI);
LLVM_DEBUG({
dbgs() << "Case clusters: ";
for (const CaseCluster &C : Clusters) {
if (C.Kind == CC_JumpTable)
dbgs() << "JT:";
if (C.Kind == CC_BitTests)
dbgs() << "BT:";
C.Low->getValue().print(dbgs(), true);
if (C.Low != C.High) {
dbgs() << '-';
C.High->getValue().print(dbgs(), true);
}
dbgs() << ' ';
}
dbgs() << '\n';
});
assert(!Clusters.empty());
SwitchWorkList WorkList;
CaseClusterIt First = Clusters.begin();
CaseClusterIt Last = Clusters.end() - 1;
auto DefaultProb = getEdgeProbability(SwitchMBB, DefaultMBB);
WorkList.push_back({SwitchMBB, First, Last, nullptr, nullptr, DefaultProb});
while (!WorkList.empty()) {
SwitchWorkListItem W = WorkList.pop_back_val();
unsigned NumClusters = W.LastCluster - W.FirstCluster + 1;
// For optimized builds, lower large range as a balanced binary tree.
if (NumClusters > 3 &&
MF->getTarget().getOptLevel() != CodeGenOptLevel::None &&
!DefaultMBB->getParent()->getFunction().hasMinSize()) {
splitWorkItem(WorkList, W, SI.getCondition(), SwitchMBB, MIB);
continue;
}
if (!lowerSwitchWorkItem(W, SI.getCondition(), SwitchMBB, DefaultMBB, MIB))
return false;
}
return true;
}
void IRTranslator::splitWorkItem(SwitchCG::SwitchWorkList &WorkList,
const SwitchCG::SwitchWorkListItem &W,
Value *Cond, MachineBasicBlock *SwitchMBB,
MachineIRBuilder &MIB) {
using namespace SwitchCG;
assert(W.FirstCluster->Low->getValue().slt(W.LastCluster->Low->getValue()) &&
"Clusters not sorted?");
assert(W.LastCluster - W.FirstCluster + 1 >= 2 && "Too small to split!");
auto [LastLeft, FirstRight, LeftProb, RightProb] =
SL->computeSplitWorkItemInfo(W);
// Use the first element on the right as pivot since we will make less-than
// comparisons against it.
CaseClusterIt PivotCluster = FirstRight;
assert(PivotCluster > W.FirstCluster);
assert(PivotCluster <= W.LastCluster);
CaseClusterIt FirstLeft = W.FirstCluster;
CaseClusterIt LastRight = W.LastCluster;
const ConstantInt *Pivot = PivotCluster->Low;
// New blocks will be inserted immediately after the current one.
MachineFunction::iterator BBI(W.MBB);
++BBI;
// We will branch to the LHS if Value < Pivot. If LHS is a single cluster,
// we can branch to its destination directly if it's squeezed exactly in
// between the known lower bound and Pivot - 1.
MachineBasicBlock *LeftMBB;
if (FirstLeft == LastLeft && FirstLeft->Kind == CC_Range &&
FirstLeft->Low == W.GE &&
(FirstLeft->High->getValue() + 1LL) == Pivot->getValue()) {
LeftMBB = FirstLeft->MBB;
} else {
LeftMBB = FuncInfo.MF->CreateMachineBasicBlock(W.MBB->getBasicBlock());
FuncInfo.MF->insert(BBI, LeftMBB);
WorkList.push_back(
{LeftMBB, FirstLeft, LastLeft, W.GE, Pivot, W.DefaultProb / 2});
}
// Similarly, we will branch to the RHS if Value >= Pivot. If RHS is a
// single cluster, RHS.Low == Pivot, and we can branch to its destination
// directly if RHS.High equals the current upper bound.
MachineBasicBlock *RightMBB;
if (FirstRight == LastRight && FirstRight->Kind == CC_Range && W.LT &&
(FirstRight->High->getValue() + 1ULL) == W.LT->getValue()) {
RightMBB = FirstRight->MBB;
} else {
RightMBB = FuncInfo.MF->CreateMachineBasicBlock(W.MBB->getBasicBlock());
FuncInfo.MF->insert(BBI, RightMBB);
WorkList.push_back(
{RightMBB, FirstRight, LastRight, Pivot, W.LT, W.DefaultProb / 2});
}
// Create the CaseBlock record that will be used to lower the branch.
CaseBlock CB(ICmpInst::Predicate::ICMP_SLT, false, Cond, Pivot, nullptr,
LeftMBB, RightMBB, W.MBB, MIB.getDebugLoc(), LeftProb,
RightProb);
if (W.MBB == SwitchMBB)
emitSwitchCase(CB, SwitchMBB, MIB);
else
SL->SwitchCases.push_back(CB);
}
void IRTranslator::emitJumpTable(SwitchCG::JumpTable &JT,
MachineBasicBlock *MBB) {
// Emit the code for the jump table
assert(JT.Reg != -1U && "Should lower JT Header first!");
MachineIRBuilder MIB(*MBB->getParent());
MIB.setMBB(*MBB);
MIB.setDebugLoc(CurBuilder->getDebugLoc());
Type *PtrIRTy = PointerType::getUnqual(MF->getFunction().getContext());
const LLT PtrTy = getLLTForType(*PtrIRTy, *DL);
auto Table = MIB.buildJumpTable(PtrTy, JT.JTI);
MIB.buildBrJT(Table.getReg(0), JT.JTI, JT.Reg);
}
bool IRTranslator::emitJumpTableHeader(SwitchCG::JumpTable &JT,
SwitchCG::JumpTableHeader &JTH,
MachineBasicBlock *HeaderBB) {
MachineIRBuilder MIB(*HeaderBB->getParent());
MIB.setMBB(*HeaderBB);
MIB.setDebugLoc(CurBuilder->getDebugLoc());
const Value &SValue = *JTH.SValue;
// Subtract the lowest switch case value from the value being switched on.
const LLT SwitchTy = getLLTForType(*SValue.getType(), *DL);
Register SwitchOpReg = getOrCreateVReg(SValue);
auto FirstCst = MIB.buildConstant(SwitchTy, JTH.First);
auto Sub = MIB.buildSub({SwitchTy}, SwitchOpReg, FirstCst);
// This value may be smaller or larger than the target's pointer type, and
// therefore require extension or truncating.
auto *PtrIRTy = PointerType::getUnqual(SValue.getContext());
const LLT PtrScalarTy = LLT::scalar(DL->getTypeSizeInBits(PtrIRTy));
Sub = MIB.buildZExtOrTrunc(PtrScalarTy, Sub);
JT.Reg = Sub.getReg(0);
if (JTH.FallthroughUnreachable) {
if (JT.MBB != HeaderBB->getNextNode())
MIB.buildBr(*JT.MBB);
return true;
}
// Emit the range check for the jump table, and branch to the default block
// for the switch statement if the value being switched on exceeds the
// largest case in the switch.
auto Cst = getOrCreateVReg(
*ConstantInt::get(SValue.getType(), JTH.Last - JTH.First));
Cst = MIB.buildZExtOrTrunc(PtrScalarTy, Cst).getReg(0);
auto Cmp = MIB.buildICmp(CmpInst::ICMP_UGT, LLT::scalar(1), Sub, Cst);
auto BrCond = MIB.buildBrCond(Cmp.getReg(0), *JT.Default);
// Avoid emitting unnecessary branches to the next block.
if (JT.MBB != HeaderBB->getNextNode())
BrCond = MIB.buildBr(*JT.MBB);
return true;
}
void IRTranslator::emitSwitchCase(SwitchCG::CaseBlock &CB,
MachineBasicBlock *SwitchBB,
MachineIRBuilder &MIB) {
Register CondLHS = getOrCreateVReg(*CB.CmpLHS);
Register Cond;
DebugLoc OldDbgLoc = MIB.getDebugLoc();
MIB.setDebugLoc(CB.DbgLoc);
MIB.setMBB(*CB.ThisBB);
if (CB.PredInfo.NoCmp) {
// Branch or fall through to TrueBB.
addSuccessorWithProb(CB.ThisBB, CB.TrueBB, CB.TrueProb);
addMachineCFGPred({SwitchBB->getBasicBlock(), CB.TrueBB->getBasicBlock()},
CB.ThisBB);
CB.ThisBB->normalizeSuccProbs();
if (CB.TrueBB != CB.ThisBB->getNextNode())
MIB.buildBr(*CB.TrueBB);
MIB.setDebugLoc(OldDbgLoc);
return;
}
const LLT i1Ty = LLT::scalar(1);
// Build the compare.
if (!CB.CmpMHS) {
const auto *CI = dyn_cast<ConstantInt>(CB.CmpRHS);
// For conditional branch lowering, we might try to do something silly like
// emit an G_ICMP to compare an existing G_ICMP i1 result with true. If so,
// just re-use the existing condition vreg.
if (MRI->getType(CondLHS).getSizeInBits() == 1 && CI && CI->isOne() &&
CB.PredInfo.Pred == CmpInst::ICMP_EQ) {
Cond = CondLHS;
} else {
Register CondRHS = getOrCreateVReg(*CB.CmpRHS);
if (CmpInst::isFPPredicate(CB.PredInfo.Pred))
Cond =
MIB.buildFCmp(CB.PredInfo.Pred, i1Ty, CondLHS, CondRHS).getReg(0);
else
Cond =
MIB.buildICmp(CB.PredInfo.Pred, i1Ty, CondLHS, CondRHS).getReg(0);
}
} else {
assert(CB.PredInfo.Pred == CmpInst::ICMP_SLE &&
"Can only handle SLE ranges");
const APInt& Low = cast<ConstantInt>(CB.CmpLHS)->getValue();
const APInt& High = cast<ConstantInt>(CB.CmpRHS)->getValue();
Register CmpOpReg = getOrCreateVReg(*CB.CmpMHS);
if (cast<ConstantInt>(CB.CmpLHS)->isMinValue(true)) {
Register CondRHS = getOrCreateVReg(*CB.CmpRHS);
Cond =
MIB.buildICmp(CmpInst::ICMP_SLE, i1Ty, CmpOpReg, CondRHS).getReg(0);
} else {
const LLT CmpTy = MRI->getType(CmpOpReg);
auto Sub = MIB.buildSub({CmpTy}, CmpOpReg, CondLHS);
auto Diff = MIB.buildConstant(CmpTy, High - Low);
Cond = MIB.buildICmp(CmpInst::ICMP_ULE, i1Ty, Sub, Diff).getReg(0);
}
}
// Update successor info
addSuccessorWithProb(CB.ThisBB, CB.TrueBB, CB.TrueProb);
addMachineCFGPred({SwitchBB->getBasicBlock(), CB.TrueBB->getBasicBlock()},
CB.ThisBB);
// TrueBB and FalseBB are always different unless the incoming IR is
// degenerate. This only happens when running llc on weird IR.
if (CB.TrueBB != CB.FalseBB)
addSuccessorWithProb(CB.ThisBB, CB.FalseBB, CB.FalseProb);
CB.ThisBB->normalizeSuccProbs();
addMachineCFGPred({SwitchBB->getBasicBlock(), CB.FalseBB->getBasicBlock()},
CB.ThisBB);
MIB.buildBrCond(Cond, *CB.TrueBB);
MIB.buildBr(*CB.FalseBB);
MIB.setDebugLoc(OldDbgLoc);
}
bool IRTranslator::lowerJumpTableWorkItem(SwitchCG::SwitchWorkListItem W,
MachineBasicBlock *SwitchMBB,
MachineBasicBlock *CurMBB,
MachineBasicBlock *DefaultMBB,
MachineIRBuilder &MIB,
MachineFunction::iterator BBI,
BranchProbability UnhandledProbs,
SwitchCG::CaseClusterIt I,
MachineBasicBlock *Fallthrough,
bool FallthroughUnreachable) {
using namespace SwitchCG;
MachineFunction *CurMF = SwitchMBB->getParent();
// FIXME: Optimize away range check based on pivot comparisons.
JumpTableHeader *JTH = &SL->JTCases[I->JTCasesIndex].first;
SwitchCG::JumpTable *JT = &SL->JTCases[I->JTCasesIndex].second;
BranchProbability DefaultProb = W.DefaultProb;
// The jump block hasn't been inserted yet; insert it here.
MachineBasicBlock *JumpMBB = JT->MBB;
CurMF->insert(BBI, JumpMBB);
// Since the jump table block is separate from the switch block, we need
// to keep track of it as a machine predecessor to the default block,
// otherwise we lose the phi edges.
addMachineCFGPred({SwitchMBB->getBasicBlock(), DefaultMBB->getBasicBlock()},
CurMBB);
addMachineCFGPred({SwitchMBB->getBasicBlock(), DefaultMBB->getBasicBlock()},
JumpMBB);
auto JumpProb = I->Prob;
auto FallthroughProb = UnhandledProbs;
// If the default statement is a target of the jump table, we evenly
// distribute the default probability to successors of CurMBB. Also
// update the probability on the edge from JumpMBB to Fallthrough.
for (MachineBasicBlock::succ_iterator SI = JumpMBB->succ_begin(),
SE = JumpMBB->succ_end();
SI != SE; ++SI) {
if (*SI == DefaultMBB) {
JumpProb += DefaultProb / 2;
FallthroughProb -= DefaultProb / 2;
JumpMBB->setSuccProbability(SI, DefaultProb / 2);
JumpMBB->normalizeSuccProbs();
} else {
// Also record edges from the jump table block to it's successors.
addMachineCFGPred({SwitchMBB->getBasicBlock(), (*SI)->getBasicBlock()},
JumpMBB);
}
}
if (FallthroughUnreachable)
JTH->FallthroughUnreachable = true;
if (!JTH->FallthroughUnreachable)
addSuccessorWithProb(CurMBB, Fallthrough, FallthroughProb);
addSuccessorWithProb(CurMBB, JumpMBB, JumpProb);
CurMBB->normalizeSuccProbs();
// The jump table header will be inserted in our current block, do the
// range check, and fall through to our fallthrough block.
JTH->HeaderBB = CurMBB;
JT->Default = Fallthrough; // FIXME: Move Default to JumpTableHeader.
// If we're in the right place, emit the jump table header right now.
if (CurMBB == SwitchMBB) {
if (!emitJumpTableHeader(*JT, *JTH, CurMBB))
return false;
JTH->Emitted = true;
}
return true;
}
bool IRTranslator::lowerSwitchRangeWorkItem(SwitchCG::CaseClusterIt I,
Value *Cond,
MachineBasicBlock *Fallthrough,
bool FallthroughUnreachable,
BranchProbability UnhandledProbs,
MachineBasicBlock *CurMBB,
MachineIRBuilder &MIB,
MachineBasicBlock *SwitchMBB) {
using namespace SwitchCG;
const Value *RHS, *LHS, *MHS;
CmpInst::Predicate Pred;
if (I->Low == I->High) {
// Check Cond == I->Low.
Pred = CmpInst::ICMP_EQ;
LHS = Cond;
RHS = I->Low;
MHS = nullptr;
} else {
// Check I->Low <= Cond <= I->High.
Pred = CmpInst::ICMP_SLE;
LHS = I->Low;
MHS = Cond;
RHS = I->High;
}
// If Fallthrough is unreachable, fold away the comparison.
// The false probability is the sum of all unhandled cases.
CaseBlock CB(Pred, FallthroughUnreachable, LHS, RHS, MHS, I->MBB, Fallthrough,
CurMBB, MIB.getDebugLoc(), I->Prob, UnhandledProbs);
emitSwitchCase(CB, SwitchMBB, MIB);
return true;
}
void IRTranslator::emitBitTestHeader(SwitchCG::BitTestBlock &B,
MachineBasicBlock *SwitchBB) {
MachineIRBuilder &MIB = *CurBuilder;
MIB.setMBB(*SwitchBB);
// Subtract the minimum value.
Register SwitchOpReg = getOrCreateVReg(*B.SValue);
LLT SwitchOpTy = MRI->getType(SwitchOpReg);
Register MinValReg = MIB.buildConstant(SwitchOpTy, B.First).getReg(0);
auto RangeSub = MIB.buildSub(SwitchOpTy, SwitchOpReg, MinValReg);
Type *PtrIRTy = PointerType::getUnqual(MF->getFunction().getContext());
const LLT PtrTy = getLLTForType(*PtrIRTy, *DL);
LLT MaskTy = SwitchOpTy;
if (MaskTy.getSizeInBits() > PtrTy.getSizeInBits() ||
!llvm::has_single_bit<uint32_t>(MaskTy.getSizeInBits()))
MaskTy = LLT::scalar(PtrTy.getSizeInBits());
else {
// Ensure that the type will fit the mask value.
for (unsigned I = 0, E = B.Cases.size(); I != E; ++I) {
if (!isUIntN(SwitchOpTy.getSizeInBits(), B.Cases[I].Mask)) {
// Switch table case range are encoded into series of masks.
// Just use pointer type, it's guaranteed to fit.
MaskTy = LLT::scalar(PtrTy.getSizeInBits());
break;
}
}
}
Register SubReg = RangeSub.getReg(0);
if (SwitchOpTy != MaskTy)
SubReg = MIB.buildZExtOrTrunc(MaskTy, SubReg).getReg(0);
B.RegVT = getMVTForLLT(MaskTy);
B.Reg = SubReg;
MachineBasicBlock *MBB = B.Cases[0].ThisBB;
if (!B.FallthroughUnreachable)
addSuccessorWithProb(SwitchBB, B.Default, B.DefaultProb);
addSuccessorWithProb(SwitchBB, MBB, B.Prob);
SwitchBB->normalizeSuccProbs();
if (!B.FallthroughUnreachable) {
// Conditional branch to the default block.
auto RangeCst = MIB.buildConstant(SwitchOpTy, B.Range);
auto RangeCmp = MIB.buildICmp(CmpInst::Predicate::ICMP_UGT, LLT::scalar(1),
RangeSub, RangeCst);
MIB.buildBrCond(RangeCmp, *B.Default);
}
// Avoid emitting unnecessary branches to the next block.
if (MBB != SwitchBB->getNextNode())
MIB.buildBr(*MBB);
}
void IRTranslator::emitBitTestCase(SwitchCG::BitTestBlock &BB,
MachineBasicBlock *NextMBB,
BranchProbability BranchProbToNext,
Register Reg, SwitchCG::BitTestCase &B,
MachineBasicBlock *SwitchBB) {
MachineIRBuilder &MIB = *CurBuilder;
MIB.setMBB(*SwitchBB);
LLT SwitchTy = getLLTForMVT(BB.RegVT);
Register Cmp;
unsigned PopCount = llvm::popcount(B.Mask);
if (PopCount == 1) {
// Testing for a single bit; just compare the shift count with what it
// would need to be to shift a 1 bit in that position.
auto MaskTrailingZeros =
MIB.buildConstant(SwitchTy, llvm::countr_zero(B.Mask));
Cmp =
MIB.buildICmp(ICmpInst::ICMP_EQ, LLT::scalar(1), Reg, MaskTrailingZeros)
.getReg(0);
} else if (PopCount == BB.Range) {
// There is only one zero bit in the range, test for it directly.
auto MaskTrailingOnes =
MIB.buildConstant(SwitchTy, llvm::countr_one(B.Mask));
Cmp = MIB.buildICmp(CmpInst::ICMP_NE, LLT::scalar(1), Reg, MaskTrailingOnes)
.getReg(0);
} else {
// Make desired shift.
auto CstOne = MIB.buildConstant(SwitchTy, 1);
auto SwitchVal = MIB.buildShl(SwitchTy, CstOne, Reg);
// Emit bit tests and jumps.
auto CstMask = MIB.buildConstant(SwitchTy, B.Mask);
auto AndOp = MIB.buildAnd(SwitchTy, SwitchVal, CstMask);
auto CstZero = MIB.buildConstant(SwitchTy, 0);
Cmp = MIB.buildICmp(CmpInst::ICMP_NE, LLT::scalar(1), AndOp, CstZero)
.getReg(0);
}
// The branch probability from SwitchBB to B.TargetBB is B.ExtraProb.
addSuccessorWithProb(SwitchBB, B.TargetBB, B.ExtraProb);
// The branch probability from SwitchBB to NextMBB is BranchProbToNext.
addSuccessorWithProb(SwitchBB, NextMBB, BranchProbToNext);
// It is not guaranteed that the sum of B.ExtraProb and BranchProbToNext is
// one as they are relative probabilities (and thus work more like weights),
// and hence we need to normalize them to let the sum of them become one.
SwitchBB->normalizeSuccProbs();
// Record the fact that the IR edge from the header to the bit test target
// will go through our new block. Neeeded for PHIs to have nodes added.
addMachineCFGPred({BB.Parent->getBasicBlock(), B.TargetBB->getBasicBlock()},
SwitchBB);
MIB.buildBrCond(Cmp, *B.TargetBB);
// Avoid emitting unnecessary branches to the next block.
if (NextMBB != SwitchBB->getNextNode())
MIB.buildBr(*NextMBB);
}
bool IRTranslator::lowerBitTestWorkItem(
SwitchCG::SwitchWorkListItem W, MachineBasicBlock *SwitchMBB,
MachineBasicBlock *CurMBB, MachineBasicBlock *DefaultMBB,
MachineIRBuilder &MIB, MachineFunction::iterator BBI,
BranchProbability DefaultProb, BranchProbability UnhandledProbs,
SwitchCG::CaseClusterIt I, MachineBasicBlock *Fallthrough,
bool FallthroughUnreachable) {
using namespace SwitchCG;
MachineFunction *CurMF = SwitchMBB->getParent();
// FIXME: Optimize away range check based on pivot comparisons.
BitTestBlock *BTB = &SL->BitTestCases[I->BTCasesIndex];
// The bit test blocks haven't been inserted yet; insert them here.
for (BitTestCase &BTC : BTB->Cases)
CurMF->insert(BBI, BTC.ThisBB);
// Fill in fields of the BitTestBlock.
BTB->Parent = CurMBB;
BTB->Default = Fallthrough;
BTB->DefaultProb = UnhandledProbs;
// If the cases in bit test don't form a contiguous range, we evenly
// distribute the probability on the edge to Fallthrough to two
// successors of CurMBB.
if (!BTB->ContiguousRange) {
BTB->Prob += DefaultProb / 2;
BTB->DefaultProb -= DefaultProb / 2;
}
if (FallthroughUnreachable)
BTB->FallthroughUnreachable = true;
// If we're in the right place, emit the bit test header right now.
if (CurMBB == SwitchMBB) {
emitBitTestHeader(*BTB, SwitchMBB);
BTB->Emitted = true;
}
return true;
}
bool IRTranslator::lowerSwitchWorkItem(SwitchCG::SwitchWorkListItem W,
Value *Cond,
MachineBasicBlock *SwitchMBB,
MachineBasicBlock *DefaultMBB,
MachineIRBuilder &MIB) {
using namespace SwitchCG;
MachineFunction *CurMF = FuncInfo.MF;
MachineBasicBlock *NextMBB = nullptr;
MachineFunction::iterator BBI(W.MBB);
if (++BBI != FuncInfo.MF->end())
NextMBB = &*BBI;
if (EnableOpts) {
// Here, we order cases by probability so the most likely case will be
// checked first. However, two clusters can have the same probability in
// which case their relative ordering is non-deterministic. So we use Low
// as a tie-breaker as clusters are guaranteed to never overlap.
llvm::sort(W.FirstCluster, W.LastCluster + 1,
[](const CaseCluster &a, const CaseCluster &b) {
return a.Prob != b.Prob
? a.Prob > b.Prob
: a.Low->getValue().slt(b.Low->getValue());
});
// Rearrange the case blocks so that the last one falls through if possible
// without changing the order of probabilities.
for (CaseClusterIt I = W.LastCluster; I > W.FirstCluster;) {
--I;
if (I->Prob > W.LastCluster->Prob)
break;
if (I->Kind == CC_Range && I->MBB == NextMBB) {
std::swap(*I, *W.LastCluster);
break;
}
}
}
// Compute total probability.
BranchProbability DefaultProb = W.DefaultProb;
BranchProbability UnhandledProbs = DefaultProb;
for (CaseClusterIt I = W.FirstCluster; I <= W.LastCluster; ++I)
UnhandledProbs += I->Prob;
MachineBasicBlock *CurMBB = W.MBB;
for (CaseClusterIt I = W.FirstCluster, E = W.LastCluster; I <= E; ++I) {
bool FallthroughUnreachable = false;
MachineBasicBlock *Fallthrough;
if (I == W.LastCluster) {
// For the last cluster, fall through to the default destination.
Fallthrough = DefaultMBB;
FallthroughUnreachable = isa<UnreachableInst>(
DefaultMBB->getBasicBlock()->getFirstNonPHIOrDbg());
} else {
Fallthrough = CurMF->CreateMachineBasicBlock(CurMBB->getBasicBlock());
CurMF->insert(BBI, Fallthrough);
}
UnhandledProbs -= I->Prob;
switch (I->Kind) {
case CC_BitTests: {
if (!lowerBitTestWorkItem(W, SwitchMBB, CurMBB, DefaultMBB, MIB, BBI,
DefaultProb, UnhandledProbs, I, Fallthrough,
FallthroughUnreachable)) {
LLVM_DEBUG(dbgs() << "Failed to lower bit test for switch");
return false;
}
break;
}
case CC_JumpTable: {
if (!lowerJumpTableWorkItem(W, SwitchMBB, CurMBB, DefaultMBB, MIB, BBI,
UnhandledProbs, I, Fallthrough,
FallthroughUnreachable)) {
LLVM_DEBUG(dbgs() << "Failed to lower jump table");
return false;
}
break;
}
case CC_Range: {
if (!lowerSwitchRangeWorkItem(I, Cond, Fallthrough,
FallthroughUnreachable, UnhandledProbs,
CurMBB, MIB, SwitchMBB)) {
LLVM_DEBUG(dbgs() << "Failed to lower switch range");
return false;
}
break;
}
}
CurMBB = Fallthrough;
}
return true;
}
bool IRTranslator::translateIndirectBr(const User &U,
MachineIRBuilder &MIRBuilder) {
const IndirectBrInst &BrInst = cast<IndirectBrInst>(U);
const Register Tgt = getOrCreateVReg(*BrInst.getAddress());
MIRBuilder.buildBrIndirect(Tgt);
// Link successors.
SmallPtrSet<const BasicBlock *, 32> AddedSuccessors;
MachineBasicBlock &CurBB = MIRBuilder.getMBB();
for (const BasicBlock *Succ : successors(&BrInst)) {
// It's legal for indirectbr instructions to have duplicate blocks in the
// destination list. We don't allow this in MIR. Skip anything that's
// already a successor.
if (!AddedSuccessors.insert(Succ).second)
continue;
CurBB.addSuccessor(&getMBB(*Succ));
}
return true;
}
static bool isSwiftError(const Value *V) {
if (auto Arg = dyn_cast<Argument>(V))
return Arg->hasSwiftErrorAttr();
if (auto AI = dyn_cast<AllocaInst>(V))
return AI->isSwiftError();
return false;
}
bool IRTranslator::translateLoad(const User &U, MachineIRBuilder &MIRBuilder) {
const LoadInst &LI = cast<LoadInst>(U);
unsigned StoreSize = DL->getTypeStoreSize(LI.getType());
if (StoreSize == 0)
return true;
ArrayRef<Register> Regs = getOrCreateVRegs(LI);
ArrayRef<uint64_t> Offsets = *VMap.getOffsets(LI);
Register Base = getOrCreateVReg(*LI.getPointerOperand());
AAMDNodes AAInfo = LI.getAAMetadata();
const Value *Ptr = LI.getPointerOperand();
Type *OffsetIRTy = DL->getIndexType(Ptr->getType());
LLT OffsetTy = getLLTForType(*OffsetIRTy, *DL);
if (CLI->supportSwiftError() && isSwiftError(Ptr)) {
assert(Regs.size() == 1 && "swifterror should be single pointer");
Register VReg =
SwiftError.getOrCreateVRegUseAt(&LI, &MIRBuilder.getMBB(), Ptr);
MIRBuilder.buildCopy(Regs[0], VReg);
return true;
}
auto &TLI = *MF->getSubtarget().getTargetLowering();
MachineMemOperand::Flags Flags =
TLI.getLoadMemOperandFlags(LI, *DL, AC, LibInfo);
if (AA && !(Flags & MachineMemOperand::MOInvariant)) {
if (AA->pointsToConstantMemory(
MemoryLocation(Ptr, LocationSize::precise(StoreSize), AAInfo))) {
Flags |= MachineMemOperand::MOInvariant;
}
}
const MDNode *Ranges =
Regs.size() == 1 ? LI.getMetadata(LLVMContext::MD_range) : nullptr;
for (unsigned i = 0; i < Regs.size(); ++i) {
Register Addr;
MIRBuilder.materializePtrAdd(Addr, Base, OffsetTy, Offsets[i] / 8);
MachinePointerInfo Ptr(LI.getPointerOperand(), Offsets[i] / 8);
Align BaseAlign = getMemOpAlign(LI);
auto MMO = MF->getMachineMemOperand(
Ptr, Flags, MRI->getType(Regs[i]),
commonAlignment(BaseAlign, Offsets[i] / 8), AAInfo, Ranges,
LI.getSyncScopeID(), LI.getOrdering());
MIRBuilder.buildLoad(Regs[i], Addr, *MMO);
}
return true;
}
bool IRTranslator::translateStore(const User &U, MachineIRBuilder &MIRBuilder) {
const StoreInst &SI = cast<StoreInst>(U);
if (DL->getTypeStoreSize(SI.getValueOperand()->getType()) == 0)
return true;
ArrayRef<Register> Vals = getOrCreateVRegs(*SI.getValueOperand());
ArrayRef<uint64_t> Offsets = *VMap.getOffsets(*SI.getValueOperand());
Register Base = getOrCreateVReg(*SI.getPointerOperand());
Type *OffsetIRTy = DL->getIndexType(SI.getPointerOperandType());
LLT OffsetTy = getLLTForType(*OffsetIRTy, *DL);
if (CLI->supportSwiftError() && isSwiftError(SI.getPointerOperand())) {
assert(Vals.size() == 1 && "swifterror should be single pointer");
Register VReg = SwiftError.getOrCreateVRegDefAt(&SI, &MIRBuilder.getMBB(),
SI.getPointerOperand());
MIRBuilder.buildCopy(VReg, Vals[0]);
return true;
}
auto &TLI = *MF->getSubtarget().getTargetLowering();
MachineMemOperand::Flags Flags = TLI.getStoreMemOperandFlags(SI, *DL);
for (unsigned i = 0; i < Vals.size(); ++i) {
Register Addr;
MIRBuilder.materializePtrAdd(Addr, Base, OffsetTy, Offsets[i] / 8);
MachinePointerInfo Ptr(SI.getPointerOperand(), Offsets[i] / 8);
Align BaseAlign = getMemOpAlign(SI);
auto MMO = MF->getMachineMemOperand(
Ptr, Flags, MRI->getType(Vals[i]),
commonAlignment(BaseAlign, Offsets[i] / 8), SI.getAAMetadata(), nullptr,
SI.getSyncScopeID(), SI.getOrdering());
MIRBuilder.buildStore(Vals[i], Addr, *MMO);
}
return true;
}
static uint64_t getOffsetFromIndices(const User &U, const DataLayout &DL) {
const Value *Src = U.getOperand(0);
Type *Int32Ty = Type::getInt32Ty(U.getContext());
// getIndexedOffsetInType is designed for GEPs, so the first index is the
// usual array element rather than looking into the actual aggregate.
SmallVector<Value *, 1> Indices;
Indices.push_back(ConstantInt::get(Int32Ty, 0));
if (const ExtractValueInst *EVI = dyn_cast<ExtractValueInst>(&U)) {
for (auto Idx : EVI->indices())
Indices.push_back(ConstantInt::get(Int32Ty, Idx));
} else if (const InsertValueInst *IVI = dyn_cast<InsertValueInst>(&U)) {
for (auto Idx : IVI->indices())
Indices.push_back(ConstantInt::get(Int32Ty, Idx));
} else {
for (unsigned i = 1; i < U.getNumOperands(); ++i)
Indices.push_back(U.getOperand(i));
}
return 8 * static_cast<uint64_t>(
DL.getIndexedOffsetInType(Src->getType(), Indices));
}
bool IRTranslator::translateExtractValue(const User &U,
MachineIRBuilder &MIRBuilder) {
const Value *Src = U.getOperand(0);
uint64_t Offset = getOffsetFromIndices(U, *DL);
ArrayRef<Register> SrcRegs = getOrCreateVRegs(*Src);
ArrayRef<uint64_t> Offsets = *VMap.getOffsets(*Src);
unsigned Idx = llvm::lower_bound(Offsets, Offset) - Offsets.begin();
auto &DstRegs = allocateVRegs(U);
for (unsigned i = 0; i < DstRegs.size(); ++i)
DstRegs[i] = SrcRegs[Idx++];
return true;
}
bool IRTranslator::translateInsertValue(const User &U,
MachineIRBuilder &MIRBuilder) {
const Value *Src = U.getOperand(0);
uint64_t Offset = getOffsetFromIndices(U, *DL);
auto &DstRegs = allocateVRegs(U);
ArrayRef<uint64_t> DstOffsets = *VMap.getOffsets(U);
ArrayRef<Register> SrcRegs = getOrCreateVRegs(*Src);
ArrayRef<Register> InsertedRegs = getOrCreateVRegs(*U.getOperand(1));
auto *InsertedIt = InsertedRegs.begin();
for (unsigned i = 0; i < DstRegs.size(); ++i) {
if (DstOffsets[i] >= Offset && InsertedIt != InsertedRegs.end())
DstRegs[i] = *InsertedIt++;
else
DstRegs[i] = SrcRegs[i];
}
return true;
}
bool IRTranslator::translateSelect(const User &U,
MachineIRBuilder &MIRBuilder) {
Register Tst = getOrCreateVReg(*U.getOperand(0));
ArrayRef<Register> ResRegs = getOrCreateVRegs(U);
ArrayRef<Register> Op0Regs = getOrCreateVRegs(*U.getOperand(1));
ArrayRef<Register> Op1Regs = getOrCreateVRegs(*U.getOperand(2));
uint32_t Flags = 0;
if (const SelectInst *SI = dyn_cast<SelectInst>(&U))
Flags = MachineInstr::copyFlagsFromInstruction(*SI);
for (unsigned i = 0; i < ResRegs.size(); ++i) {
MIRBuilder.buildSelect(ResRegs[i], Tst, Op0Regs[i], Op1Regs[i], Flags);
}
return true;
}
bool IRTranslator::translateCopy(const User &U, const Value &V,
MachineIRBuilder &MIRBuilder) {
Register Src = getOrCreateVReg(V);
auto &Regs = *VMap.getVRegs(U);
if (Regs.empty()) {
Regs.push_back(Src);
VMap.getOffsets(U)->push_back(0);
} else {
// If we already assigned a vreg for this instruction, we can't change that.
// Emit a copy to satisfy the users we already emitted.
MIRBuilder.buildCopy(Regs[0], Src);
}
return true;
}
bool IRTranslator::translateBitCast(const User &U,
MachineIRBuilder &MIRBuilder) {
// If we're bitcasting to the source type, we can reuse the source vreg.
if (getLLTForType(*U.getOperand(0)->getType(), *DL) ==
getLLTForType(*U.getType(), *DL)) {
// If the source is a ConstantInt then it was probably created by
// ConstantHoisting and we should leave it alone.
if (isa<ConstantInt>(U.getOperand(0)))
return translateCast(TargetOpcode::G_CONSTANT_FOLD_BARRIER, U,
MIRBuilder);
return translateCopy(U, *U.getOperand(0), MIRBuilder);
}
return translateCast(TargetOpcode::G_BITCAST, U, MIRBuilder);
}
bool IRTranslator::translateCast(unsigned Opcode, const User &U,
MachineIRBuilder &MIRBuilder) {
if (U.getType()->getScalarType()->isBFloatTy() ||
U.getOperand(0)->getType()->getScalarType()->isBFloatTy())
return false;
Register Op = getOrCreateVReg(*U.getOperand(0));
Register Res = getOrCreateVReg(U);
MIRBuilder.buildInstr(Opcode, {Res}, {Op});
return true;
}
bool IRTranslator::translateGetElementPtr(const User &U,
MachineIRBuilder &MIRBuilder) {
Value &Op0 = *U.getOperand(0);
Register BaseReg = getOrCreateVReg(Op0);
Type *PtrIRTy = Op0.getType();
LLT PtrTy = getLLTForType(*PtrIRTy, *DL);
Type *OffsetIRTy = DL->getIndexType(PtrIRTy);
LLT OffsetTy = getLLTForType(*OffsetIRTy, *DL);
uint32_t Flags = 0;
if (isa<Instruction>(U)) {
const Instruction &I = cast<Instruction>(U);
Flags = MachineInstr::copyFlagsFromInstruction(I);
}
// Normalize Vector GEP - all scalar operands should be converted to the
// splat vector.
unsigned VectorWidth = 0;
// True if we should use a splat vector; using VectorWidth alone is not
// sufficient.
bool WantSplatVector = false;
if (auto *VT = dyn_cast<VectorType>(U.getType())) {
VectorWidth = cast<FixedVectorType>(VT)->getNumElements();
// We don't produce 1 x N vectors; those are treated as scalars.
WantSplatVector = VectorWidth > 1;
}
// We might need to splat the base pointer into a vector if the offsets
// are vectors.
if (WantSplatVector && !PtrTy.isVector()) {
BaseReg =
MIRBuilder
.buildSplatVector(LLT::fixed_vector(VectorWidth, PtrTy), BaseReg)
.getReg(0);
PtrIRTy = FixedVectorType::get(PtrIRTy, VectorWidth);
PtrTy = getLLTForType(*PtrIRTy, *DL);
OffsetIRTy = DL->getIndexType(PtrIRTy);
OffsetTy = getLLTForType(*OffsetIRTy, *DL);
}
int64_t Offset = 0;
for (gep_type_iterator GTI = gep_type_begin(&U), E = gep_type_end(&U);
GTI != E; ++GTI) {
const Value *Idx = GTI.getOperand();
if (StructType *StTy = GTI.getStructTypeOrNull()) {
unsigned Field = cast<Constant>(Idx)->getUniqueInteger().getZExtValue();
Offset += DL->getStructLayout(StTy)->getElementOffset(Field);
continue;
} else {
uint64_t ElementSize = GTI.getSequentialElementStride(*DL);
// If this is a scalar constant or a splat vector of constants,
// handle it quickly.
if (const auto *CI = dyn_cast<ConstantInt>(Idx)) {
if (std::optional<int64_t> Val = CI->getValue().trySExtValue()) {
Offset += ElementSize * *Val;
continue;
}
}
if (Offset != 0) {
auto OffsetMIB = MIRBuilder.buildConstant({OffsetTy}, Offset);
BaseReg = MIRBuilder.buildPtrAdd(PtrTy, BaseReg, OffsetMIB.getReg(0))
.getReg(0);
Offset = 0;
}
Register IdxReg = getOrCreateVReg(*Idx);
LLT IdxTy = MRI->getType(IdxReg);
if (IdxTy != OffsetTy) {
if (!IdxTy.isVector() && WantSplatVector) {
IdxReg = MIRBuilder.buildSplatVector(
OffsetTy.changeElementType(IdxTy), IdxReg).getReg(0);
}
IdxReg = MIRBuilder.buildSExtOrTrunc(OffsetTy, IdxReg).getReg(0);
}
// N = N + Idx * ElementSize;
// Avoid doing it for ElementSize of 1.
Register GepOffsetReg;
if (ElementSize != 1) {
auto ElementSizeMIB = MIRBuilder.buildConstant(
getLLTForType(*OffsetIRTy, *DL), ElementSize);
GepOffsetReg =
MIRBuilder.buildMul(OffsetTy, IdxReg, ElementSizeMIB).getReg(0);
} else
GepOffsetReg = IdxReg;
BaseReg = MIRBuilder.buildPtrAdd(PtrTy, BaseReg, GepOffsetReg).getReg(0);
}
}
if (Offset != 0) {
auto OffsetMIB =
MIRBuilder.buildConstant(OffsetTy, Offset);
if (int64_t(Offset) >= 0 && cast<GEPOperator>(U).isInBounds())
Flags |= MachineInstr::MIFlag::NoUWrap;
MIRBuilder.buildPtrAdd(getOrCreateVReg(U), BaseReg, OffsetMIB.getReg(0),
Flags);
return true;
}
MIRBuilder.buildCopy(getOrCreateVReg(U), BaseReg);
return true;
}
bool IRTranslator::translateMemFunc(const CallInst &CI,
MachineIRBuilder &MIRBuilder,
unsigned Opcode) {
const Value *SrcPtr = CI.getArgOperand(1);
// If the source is undef, then just emit a nop.
if (isa<UndefValue>(SrcPtr))
return true;
SmallVector<Register, 3> SrcRegs;
unsigned MinPtrSize = UINT_MAX;
for (auto AI = CI.arg_begin(), AE = CI.arg_end(); std::next(AI) != AE; ++AI) {
Register SrcReg = getOrCreateVReg(**AI);
LLT SrcTy = MRI->getType(SrcReg);
if (SrcTy.isPointer())
MinPtrSize = std::min<unsigned>(SrcTy.getSizeInBits(), MinPtrSize);
SrcRegs.push_back(SrcReg);
}
LLT SizeTy = LLT::scalar(MinPtrSize);
// The size operand should be the minimum of the pointer sizes.
Register &SizeOpReg = SrcRegs[SrcRegs.size() - 1];
if (MRI->getType(SizeOpReg) != SizeTy)
SizeOpReg = MIRBuilder.buildZExtOrTrunc(SizeTy, SizeOpReg).getReg(0);
auto ICall = MIRBuilder.buildInstr(Opcode);
for (Register SrcReg : SrcRegs)
ICall.addUse(SrcReg);
Align DstAlign;
Align SrcAlign;
unsigned IsVol =
cast<ConstantInt>(CI.getArgOperand(CI.arg_size() - 1))->getZExtValue();
ConstantInt *CopySize = nullptr;
if (auto *MCI = dyn_cast<MemCpyInst>(&CI)) {
DstAlign = MCI->getDestAlign().valueOrOne();
SrcAlign = MCI->getSourceAlign().valueOrOne();
CopySize = dyn_cast<ConstantInt>(MCI->getArgOperand(2));
} else if (auto *MCI = dyn_cast<MemCpyInlineInst>(&CI)) {
DstAlign = MCI->getDestAlign().valueOrOne();
SrcAlign = MCI->getSourceAlign().valueOrOne();
CopySize = dyn_cast<ConstantInt>(MCI->getArgOperand(2));
} else if (auto *MMI = dyn_cast<MemMoveInst>(&CI)) {
DstAlign = MMI->getDestAlign().valueOrOne();
SrcAlign = MMI->getSourceAlign().valueOrOne();
CopySize = dyn_cast<ConstantInt>(MMI->getArgOperand(2));
} else {
auto *MSI = cast<MemSetInst>(&CI);
DstAlign = MSI->getDestAlign().valueOrOne();
}
if (Opcode != TargetOpcode::G_MEMCPY_INLINE) {
// We need to propagate the tail call flag from the IR inst as an argument.
// Otherwise, we have to pessimize and assume later that we cannot tail call
// any memory intrinsics.
ICall.addImm(CI.isTailCall() ? 1 : 0);
}
// Create mem operands to store the alignment and volatile info.
MachineMemOperand::Flags LoadFlags = MachineMemOperand::MOLoad;
MachineMemOperand::Flags StoreFlags = MachineMemOperand::MOStore;
if (IsVol) {
LoadFlags |= MachineMemOperand::MOVolatile;
StoreFlags |= MachineMemOperand::MOVolatile;
}
AAMDNodes AAInfo = CI.getAAMetadata();
if (AA && CopySize &&
AA->pointsToConstantMemory(MemoryLocation(
SrcPtr, LocationSize::precise(CopySize->getZExtValue()), AAInfo))) {
LoadFlags |= MachineMemOperand::MOInvariant;
// FIXME: pointsToConstantMemory probably does not imply dereferenceable,
// but the previous usage implied it did. Probably should check
// isDereferenceableAndAlignedPointer.
LoadFlags |= MachineMemOperand::MODereferenceable;
}
ICall.addMemOperand(
MF->getMachineMemOperand(MachinePointerInfo(CI.getArgOperand(0)),
StoreFlags, 1, DstAlign, AAInfo));
if (Opcode != TargetOpcode::G_MEMSET)
ICall.addMemOperand(MF->getMachineMemOperand(
MachinePointerInfo(SrcPtr), LoadFlags, 1, SrcAlign, AAInfo));
return true;
}
void IRTranslator::getStackGuard(Register DstReg,
MachineIRBuilder &MIRBuilder) {
const TargetRegisterInfo *TRI = MF->getSubtarget().getRegisterInfo();
MRI->setRegClass(DstReg, TRI->getPointerRegClass(*MF));
auto MIB =
MIRBuilder.buildInstr(TargetOpcode::LOAD_STACK_GUARD, {DstReg}, {});
auto &TLI = *MF->getSubtarget().getTargetLowering();
Value *Global = TLI.getSDagStackGuard(*MF->getFunction().getParent());
if (!Global)
return;
unsigned AddrSpace = Global->getType()->getPointerAddressSpace();
LLT PtrTy = LLT::pointer(AddrSpace, DL->getPointerSizeInBits(AddrSpace));
MachinePointerInfo MPInfo(Global);
auto Flags = MachineMemOperand::MOLoad | MachineMemOperand::MOInvariant |
MachineMemOperand::MODereferenceable;
MachineMemOperand *MemRef = MF->getMachineMemOperand(
MPInfo, Flags, PtrTy, DL->getPointerABIAlignment(AddrSpace));
MIB.setMemRefs({MemRef});
}
bool IRTranslator::translateOverflowIntrinsic(const CallInst &CI, unsigned Op,
MachineIRBuilder &MIRBuilder) {
ArrayRef<Register> ResRegs = getOrCreateVRegs(CI);
MIRBuilder.buildInstr(
Op, {ResRegs[0], ResRegs[1]},
{getOrCreateVReg(*CI.getOperand(0)), getOrCreateVReg(*CI.getOperand(1))});
return true;
}
bool IRTranslator::translateFixedPointIntrinsic(unsigned Op, const CallInst &CI,
MachineIRBuilder &MIRBuilder) {
Register Dst = getOrCreateVReg(CI);
Register Src0 = getOrCreateVReg(*CI.getOperand(0));
Register Src1 = getOrCreateVReg(*CI.getOperand(1));
uint64_t Scale = cast<ConstantInt>(CI.getOperand(2))->getZExtValue();
MIRBuilder.buildInstr(Op, {Dst}, { Src0, Src1, Scale });
return true;
}
unsigned IRTranslator::getSimpleIntrinsicOpcode(Intrinsic::ID ID) {
switch (ID) {
default:
break;
case Intrinsic::bswap:
return TargetOpcode::G_BSWAP;
case Intrinsic::bitreverse:
return TargetOpcode::G_BITREVERSE;
case Intrinsic::fshl:
return TargetOpcode::G_FSHL;
case Intrinsic::fshr:
return TargetOpcode::G_FSHR;
case Intrinsic::ceil:
return TargetOpcode::G_FCEIL;
case Intrinsic::cos:
return TargetOpcode::G_FCOS;
case Intrinsic::ctpop:
return TargetOpcode::G_CTPOP;
case Intrinsic::exp:
return TargetOpcode::G_FEXP;
case Intrinsic::exp2:
return TargetOpcode::G_FEXP2;
case Intrinsic::exp10:
return TargetOpcode::G_FEXP10;
case Intrinsic::fabs:
return TargetOpcode::G_FABS;
case Intrinsic::copysign:
return TargetOpcode::G_FCOPYSIGN;
case Intrinsic::minnum:
return TargetOpcode::G_FMINNUM;
case Intrinsic::maxnum:
return TargetOpcode::G_FMAXNUM;
case Intrinsic::minimum:
return TargetOpcode::G_FMINIMUM;
case Intrinsic::maximum:
return TargetOpcode::G_FMAXIMUM;
case Intrinsic::canonicalize:
return TargetOpcode::G_FCANONICALIZE;
case Intrinsic::floor:
return TargetOpcode::G_FFLOOR;
case Intrinsic::fma:
return TargetOpcode::G_FMA;
case Intrinsic::log:
return TargetOpcode::G_FLOG;
case Intrinsic::log2:
return TargetOpcode::G_FLOG2;
case Intrinsic::log10:
return TargetOpcode::G_FLOG10;
case Intrinsic::ldexp:
return TargetOpcode::G_FLDEXP;
case Intrinsic::nearbyint:
return TargetOpcode::G_FNEARBYINT;
case Intrinsic::pow:
return TargetOpcode::G_FPOW;
case Intrinsic::powi:
return TargetOpcode::G_FPOWI;
case Intrinsic::rint:
return TargetOpcode::G_FRINT;
case Intrinsic::round:
return TargetOpcode::G_INTRINSIC_ROUND;
case Intrinsic::roundeven:
return TargetOpcode::G_INTRINSIC_ROUNDEVEN;
case Intrinsic::sin:
return TargetOpcode::G_FSIN;
case Intrinsic::sqrt:
return TargetOpcode::G_FSQRT;
case Intrinsic::trunc:
return TargetOpcode::G_INTRINSIC_TRUNC;
case Intrinsic::readcyclecounter:
return TargetOpcode::G_READCYCLECOUNTER;
case Intrinsic::readsteadycounter:
return TargetOpcode::G_READSTEADYCOUNTER;
case Intrinsic::ptrmask:
return TargetOpcode::G_PTRMASK;
case Intrinsic::lrint:
return TargetOpcode::G_INTRINSIC_LRINT;
// FADD/FMUL require checking the FMF, so are handled elsewhere.
case Intrinsic::vector_reduce_fmin:
return TargetOpcode::G_VECREDUCE_FMIN;
case Intrinsic::vector_reduce_fmax:
return TargetOpcode::G_VECREDUCE_FMAX;
case Intrinsic::vector_reduce_fminimum:
return TargetOpcode::G_VECREDUCE_FMINIMUM;
case Intrinsic::vector_reduce_fmaximum:
return TargetOpcode::G_VECREDUCE_FMAXIMUM;
case Intrinsic::vector_reduce_add:
return TargetOpcode::G_VECREDUCE_ADD;
case Intrinsic::vector_reduce_mul:
return TargetOpcode::G_VECREDUCE_MUL;
case Intrinsic::vector_reduce_and:
return TargetOpcode::G_VECREDUCE_AND;
case Intrinsic::vector_reduce_or:
return TargetOpcode::G_VECREDUCE_OR;
case Intrinsic::vector_reduce_xor:
return TargetOpcode::G_VECREDUCE_XOR;
case Intrinsic::vector_reduce_smax:
return TargetOpcode::G_VECREDUCE_SMAX;
case Intrinsic::vector_reduce_smin:
return TargetOpcode::G_VECREDUCE_SMIN;
case Intrinsic::vector_reduce_umax:
return TargetOpcode::G_VECREDUCE_UMAX;
case Intrinsic::vector_reduce_umin:
return TargetOpcode::G_VECREDUCE_UMIN;
case Intrinsic::lround:
return TargetOpcode::G_LROUND;
case Intrinsic::llround:
return TargetOpcode::G_LLROUND;
case Intrinsic::get_fpenv:
return TargetOpcode::G_GET_FPENV;
case Intrinsic::get_fpmode:
return TargetOpcode::G_GET_FPMODE;
}
return Intrinsic::not_intrinsic;
}
bool IRTranslator::translateSimpleIntrinsic(const CallInst &CI,
Intrinsic::ID ID,
MachineIRBuilder &MIRBuilder) {
unsigned Op = getSimpleIntrinsicOpcode(ID);
// Is this a simple intrinsic?
if (Op == Intrinsic::not_intrinsic)
return false;
// Yes. Let's translate it.
SmallVector<llvm::SrcOp, 4> VRegs;
for (const auto &Arg : CI.args())
VRegs.push_back(getOrCreateVReg(*Arg));
MIRBuilder.buildInstr(Op, {getOrCreateVReg(CI)}, VRegs,
MachineInstr::copyFlagsFromInstruction(CI));
return true;
}
// TODO: Include ConstainedOps.def when all strict instructions are defined.
static unsigned getConstrainedOpcode(Intrinsic::ID ID) {
switch (ID) {
case Intrinsic::experimental_constrained_fadd:
return TargetOpcode::G_STRICT_FADD;
case Intrinsic::experimental_constrained_fsub:
return TargetOpcode::G_STRICT_FSUB;
case Intrinsic::experimental_constrained_fmul:
return TargetOpcode::G_STRICT_FMUL;
case Intrinsic::experimental_constrained_fdiv:
return TargetOpcode::G_STRICT_FDIV;
case Intrinsic::experimental_constrained_frem:
return TargetOpcode::G_STRICT_FREM;
case Intrinsic::experimental_constrained_fma:
return TargetOpcode::G_STRICT_FMA;
case Intrinsic::experimental_constrained_sqrt:
return TargetOpcode::G_STRICT_FSQRT;
case Intrinsic::experimental_constrained_ldexp:
return TargetOpcode::G_STRICT_FLDEXP;
default:
return 0;
}
}
bool IRTranslator::translateConstrainedFPIntrinsic(
const ConstrainedFPIntrinsic &FPI, MachineIRBuilder &MIRBuilder) {
fp::ExceptionBehavior EB = *FPI.getExceptionBehavior();
unsigned Opcode = getConstrainedOpcode(FPI.getIntrinsicID());
if (!Opcode)
return false;
uint32_t Flags = MachineInstr::copyFlagsFromInstruction(FPI);
if (EB == fp::ExceptionBehavior::ebIgnore)
Flags |= MachineInstr::NoFPExcept;
SmallVector<llvm::SrcOp, 4> VRegs;
VRegs.push_back(getOrCreateVReg(*FPI.getArgOperand(0)));
if (!FPI.isUnaryOp())
VRegs.push_back(getOrCreateVReg(*FPI.getArgOperand(1)));
if (FPI.isTernaryOp())
VRegs.push_back(getOrCreateVReg(*FPI.getArgOperand(2)));
MIRBuilder.buildInstr(Opcode, {getOrCreateVReg(FPI)}, VRegs, Flags);
return true;
}
std::optional<MCRegister> IRTranslator::getArgPhysReg(Argument &Arg) {
auto VRegs = getOrCreateVRegs(Arg);
if (VRegs.size() != 1)
return std::nullopt;
// Arguments are lowered as a copy of a livein physical register.
auto *VRegDef = MF->getRegInfo().getVRegDef(VRegs[0]);
if (!VRegDef || !VRegDef->isCopy())
return std::nullopt;
return VRegDef->getOperand(1).getReg().asMCReg();
}
bool IRTranslator::translateIfEntryValueArgument(bool isDeclare, Value *Val,
const DILocalVariable *Var,
const DIExpression *Expr,
const DebugLoc &DL,
MachineIRBuilder &MIRBuilder) {
auto *Arg = dyn_cast<Argument>(Val);
if (!Arg)
return false;
if (!Expr->isEntryValue())
return false;
std::optional<MCRegister> PhysReg = getArgPhysReg(*Arg);
if (!PhysReg) {
LLVM_DEBUG(dbgs() << "Dropping dbg." << (isDeclare ? "declare" : "value")
<< ": expression is entry_value but "
<< "couldn't find a physical register\n");
LLVM_DEBUG(dbgs() << *Var << "\n");
return true;
}
if (isDeclare) {
// Append an op deref to account for the fact that this is a dbg_declare.
Expr = DIExpression::append(Expr, dwarf::DW_OP_deref);
MF->setVariableDbgInfo(Var, Expr, *PhysReg, DL);
} else {
MIRBuilder.buildDirectDbgValue(*PhysReg, Var, Expr);
}
return true;
}
bool IRTranslator::translateKnownIntrinsic(const CallInst &CI, Intrinsic::ID ID,
MachineIRBuilder &MIRBuilder) {
if (auto *MI = dyn_cast<AnyMemIntrinsic>(&CI)) {
if (ORE->enabled()) {
if (MemoryOpRemark::canHandle(MI, *LibInfo)) {
MemoryOpRemark R(*ORE, "gisel-irtranslator-memsize", *DL, *LibInfo);
R.visit(MI);
}
}
}
// If this is a simple intrinsic (that is, we just need to add a def of
// a vreg, and uses for each arg operand, then translate it.
if (translateSimpleIntrinsic(CI, ID, MIRBuilder))
return true;
switch (ID) {
default:
break;
case Intrinsic::lifetime_start:
case Intrinsic::lifetime_end: {
// No stack colouring in O0, discard region information.
if (MF->getTarget().getOptLevel() == CodeGenOptLevel::None)
return true;
unsigned Op = ID == Intrinsic::lifetime_start ? TargetOpcode::LIFETIME_START
: TargetOpcode::LIFETIME_END;
// Get the underlying objects for the location passed on the lifetime
// marker.
SmallVector<const Value *, 4> Allocas;
getUnderlyingObjects(CI.getArgOperand(1), Allocas);
// Iterate over each underlying object, creating lifetime markers for each
// static alloca. Quit if we find a non-static alloca.
for (const Value *V : Allocas) {
const AllocaInst *AI = dyn_cast<AllocaInst>(V);
if (!AI)
continue;
if (!AI->isStaticAlloca())
return true;
MIRBuilder.buildInstr(Op).addFrameIndex(getOrCreateFrameIndex(*AI));
}
return true;
}
case Intrinsic::dbg_declare: {
const DbgDeclareInst &DI = cast<DbgDeclareInst>(CI);
assert(DI.getVariable() && "Missing variable");
translateDbgDeclareRecord(DI.getAddress(), DI.hasArgList(), DI.getVariable(),
DI.getExpression(), DI.getDebugLoc(), MIRBuilder);
return true;
}
case Intrinsic::dbg_label: {
const DbgLabelInst &DI = cast<DbgLabelInst>(CI);
assert(DI.getLabel() && "Missing label");
assert(DI.getLabel()->isValidLocationForIntrinsic(
MIRBuilder.getDebugLoc()) &&
"Expected inlined-at fields to agree");
MIRBuilder.buildDbgLabel(DI.getLabel());
return true;
}
case Intrinsic::vaend:
// No target I know of cares about va_end. Certainly no in-tree target
// does. Simplest intrinsic ever!
return true;
case Intrinsic::vastart: {
auto &TLI = *MF->getSubtarget().getTargetLowering();
Value *Ptr = CI.getArgOperand(0);
unsigned ListSize = TLI.getVaListSizeInBits(*DL) / 8;
Align Alignment = getKnownAlignment(Ptr, *DL);
MIRBuilder.buildInstr(TargetOpcode::G_VASTART, {}, {getOrCreateVReg(*Ptr)})
.addMemOperand(MF->getMachineMemOperand(MachinePointerInfo(Ptr),
MachineMemOperand::MOStore,
ListSize, Alignment));
return true;
}
case Intrinsic::dbg_assign:
// A dbg.assign is a dbg.value with more information about stack locations,
// typically produced during optimisation of variables with leaked
// addresses. We can treat it like a normal dbg_value intrinsic here; to
// benefit from the full analysis of stack/SSA locations, GlobalISel would
// need to register for and use the AssignmentTrackingAnalysis pass.
LLVM_FALLTHROUGH;
case Intrinsic::dbg_value: {
// This form of DBG_VALUE is target-independent.
const DbgValueInst &DI = cast<DbgValueInst>(CI);
translateDbgValueRecord(DI.getValue(), DI.hasArgList(), DI.getVariable(),
DI.getExpression(), DI.getDebugLoc(), MIRBuilder);
return true;
}
case Intrinsic::uadd_with_overflow:
return translateOverflowIntrinsic(CI, TargetOpcode::G_UADDO, MIRBuilder);
case Intrinsic::sadd_with_overflow:
return translateOverflowIntrinsic(CI, TargetOpcode::G_SADDO, MIRBuilder);
case Intrinsic::usub_with_overflow:
return translateOverflowIntrinsic(CI, TargetOpcode::G_USUBO, MIRBuilder);
case Intrinsic::ssub_with_overflow:
return translateOverflowIntrinsic(CI, TargetOpcode::G_SSUBO, MIRBuilder);
case Intrinsic::umul_with_overflow:
return translateOverflowIntrinsic(CI, TargetOpcode::G_UMULO, MIRBuilder);
case Intrinsic::smul_with_overflow:
return translateOverflowIntrinsic(CI, TargetOpcode::G_SMULO, MIRBuilder);
case Intrinsic::uadd_sat:
return translateBinaryOp(TargetOpcode::G_UADDSAT, CI, MIRBuilder);
case Intrinsic::sadd_sat:
return translateBinaryOp(TargetOpcode::G_SADDSAT, CI, MIRBuilder);
case Intrinsic::usub_sat:
return translateBinaryOp(TargetOpcode::G_USUBSAT, CI, MIRBuilder);
case Intrinsic::ssub_sat:
return translateBinaryOp(TargetOpcode::G_SSUBSAT, CI, MIRBuilder);
case Intrinsic::ushl_sat:
return translateBinaryOp(TargetOpcode::G_USHLSAT, CI, MIRBuilder);
case Intrinsic::sshl_sat:
return translateBinaryOp(TargetOpcode::G_SSHLSAT, CI, MIRBuilder);
case Intrinsic::umin:
return translateBinaryOp(TargetOpcode::G_UMIN, CI, MIRBuilder);
case Intrinsic::umax:
return translateBinaryOp(TargetOpcode::G_UMAX, CI, MIRBuilder);
case Intrinsic::smin:
return translateBinaryOp(TargetOpcode::G_SMIN, CI, MIRBuilder);
case Intrinsic::smax:
return translateBinaryOp(TargetOpcode::G_SMAX, CI, MIRBuilder);
case Intrinsic::abs:
// TODO: Preserve "int min is poison" arg in GMIR?
return translateUnaryOp(TargetOpcode::G_ABS, CI, MIRBuilder);
case Intrinsic::smul_fix:
return translateFixedPointIntrinsic(TargetOpcode::G_SMULFIX, CI, MIRBuilder);
case Intrinsic::umul_fix:
return translateFixedPointIntrinsic(TargetOpcode::G_UMULFIX, CI, MIRBuilder);
case Intrinsic::smul_fix_sat:
return translateFixedPointIntrinsic(TargetOpcode::G_SMULFIXSAT, CI, MIRBuilder);
case Intrinsic::umul_fix_sat:
return translateFixedPointIntrinsic(TargetOpcode::G_UMULFIXSAT, CI, MIRBuilder);
case Intrinsic::sdiv_fix:
return translateFixedPointIntrinsic(TargetOpcode::G_SDIVFIX, CI, MIRBuilder);
case Intrinsic::udiv_fix:
return translateFixedPointIntrinsic(TargetOpcode::G_UDIVFIX, CI, MIRBuilder);
case Intrinsic::sdiv_fix_sat:
return translateFixedPointIntrinsic(TargetOpcode::G_SDIVFIXSAT, CI, MIRBuilder);
case Intrinsic::udiv_fix_sat:
return translateFixedPointIntrinsic(TargetOpcode::G_UDIVFIXSAT, CI, MIRBuilder);
case Intrinsic::fmuladd: {
const TargetMachine &TM = MF->getTarget();
const TargetLowering &TLI = *MF->getSubtarget().getTargetLowering();
Register Dst = getOrCreateVReg(CI);
Register Op0 = getOrCreateVReg(*CI.getArgOperand(0));
Register Op1 = getOrCreateVReg(*CI.getArgOperand(1));
Register Op2 = getOrCreateVReg(*CI.getArgOperand(2));
if (TM.Options.AllowFPOpFusion != FPOpFusion::Strict &&
TLI.isFMAFasterThanFMulAndFAdd(*MF,
TLI.getValueType(*DL, CI.getType()))) {
// TODO: Revisit this to see if we should move this part of the
// lowering to the combiner.
MIRBuilder.buildFMA(Dst, Op0, Op1, Op2,
MachineInstr::copyFlagsFromInstruction(CI));
} else {
LLT Ty = getLLTForType(*CI.getType(), *DL);
auto FMul = MIRBuilder.buildFMul(
Ty, Op0, Op1, MachineInstr::copyFlagsFromInstruction(CI));
MIRBuilder.buildFAdd(Dst, FMul, Op2,
MachineInstr::copyFlagsFromInstruction(CI));
}
return true;
}
case Intrinsic::convert_from_fp16:
// FIXME: This intrinsic should probably be removed from the IR.
MIRBuilder.buildFPExt(getOrCreateVReg(CI),
getOrCreateVReg(*CI.getArgOperand(0)),
MachineInstr::copyFlagsFromInstruction(CI));
return true;
case Intrinsic::convert_to_fp16:
// FIXME: This intrinsic should probably be removed from the IR.
MIRBuilder.buildFPTrunc(getOrCreateVReg(CI),
getOrCreateVReg(*CI.getArgOperand(0)),
MachineInstr::copyFlagsFromInstruction(CI));
return true;
case Intrinsic::frexp: {
ArrayRef<Register> VRegs = getOrCreateVRegs(CI);
MIRBuilder.buildFFrexp(VRegs[0], VRegs[1],
getOrCreateVReg(*CI.getArgOperand(0)),
MachineInstr::copyFlagsFromInstruction(CI));
return true;
}
case Intrinsic::memcpy_inline:
return translateMemFunc(CI, MIRBuilder, TargetOpcode::G_MEMCPY_INLINE);
case Intrinsic::memcpy:
return translateMemFunc(CI, MIRBuilder, TargetOpcode::G_MEMCPY);
case Intrinsic::memmove:
return translateMemFunc(CI, MIRBuilder, TargetOpcode::G_MEMMOVE);
case Intrinsic::memset:
return translateMemFunc(CI, MIRBuilder, TargetOpcode::G_MEMSET);
case Intrinsic::eh_typeid_for: {
GlobalValue *GV = ExtractTypeInfo(CI.getArgOperand(0));
Register Reg = getOrCreateVReg(CI);
unsigned TypeID = MF->getTypeIDFor(GV);
MIRBuilder.buildConstant(Reg, TypeID);
return true;
}
case Intrinsic::objectsize:
llvm_unreachable("llvm.objectsize.* should have been lowered already");
case Intrinsic::is_constant:
llvm_unreachable("llvm.is.constant.* should have been lowered already");
case Intrinsic::stackguard:
getStackGuard(getOrCreateVReg(CI), MIRBuilder);
return true;
case Intrinsic::stackprotector: {
const TargetLowering &TLI = *MF->getSubtarget().getTargetLowering();
LLT PtrTy = getLLTForType(*CI.getArgOperand(0)->getType(), *DL);
Register GuardVal;
if (TLI.useLoadStackGuardNode()) {
GuardVal = MRI->createGenericVirtualRegister(PtrTy);
getStackGuard(GuardVal, MIRBuilder);
} else
GuardVal = getOrCreateVReg(*CI.getArgOperand(0)); // The guard's value.
AllocaInst *Slot = cast<AllocaInst>(CI.getArgOperand(1));
int FI = getOrCreateFrameIndex(*Slot);
MF->getFrameInfo().setStackProtectorIndex(FI);
MIRBuilder.buildStore(
GuardVal, getOrCreateVReg(*Slot),
*MF->getMachineMemOperand(MachinePointerInfo::getFixedStack(*MF, FI),
MachineMemOperand::MOStore |
MachineMemOperand::MOVolatile,
PtrTy, Align(8)));
return true;
}
case Intrinsic::stacksave: {
MIRBuilder.buildInstr(TargetOpcode::G_STACKSAVE, {getOrCreateVReg(CI)}, {});
return true;
}
case Intrinsic::stackrestore: {
MIRBuilder.buildInstr(TargetOpcode::G_STACKRESTORE, {},
{getOrCreateVReg(*CI.getArgOperand(0))});
return true;
}
case Intrinsic::cttz:
case Intrinsic::ctlz: {
ConstantInt *Cst = cast<ConstantInt>(CI.getArgOperand(1));
bool isTrailing = ID == Intrinsic::cttz;
unsigned Opcode = isTrailing
? Cst->isZero() ? TargetOpcode::G_CTTZ
: TargetOpcode::G_CTTZ_ZERO_UNDEF
: Cst->isZero() ? TargetOpcode::G_CTLZ
: TargetOpcode::G_CTLZ_ZERO_UNDEF;
MIRBuilder.buildInstr(Opcode, {getOrCreateVReg(CI)},
{getOrCreateVReg(*CI.getArgOperand(0))});
return true;
}
case Intrinsic::invariant_start: {
LLT PtrTy = getLLTForType(*CI.getArgOperand(0)->getType(), *DL);
Register Undef = MRI->createGenericVirtualRegister(PtrTy);
MIRBuilder.buildUndef(Undef);
return true;
}
case Intrinsic::invariant_end:
return true;
case Intrinsic::expect:
case Intrinsic::annotation:
case Intrinsic::ptr_annotation:
case Intrinsic::launder_invariant_group:
case Intrinsic::strip_invariant_group: {
// Drop the intrinsic, but forward the value.
MIRBuilder.buildCopy(getOrCreateVReg(CI),
getOrCreateVReg(*CI.getArgOperand(0)));
return true;
}
case Intrinsic::assume:
case Intrinsic::experimental_noalias_scope_decl:
case Intrinsic::var_annotation:
case Intrinsic::sideeffect:
// Discard annotate attributes, assumptions, and artificial side-effects.
return true;
case Intrinsic::read_volatile_register:
case Intrinsic::read_register: {
Value *Arg = CI.getArgOperand(0);
MIRBuilder
.buildInstr(TargetOpcode::G_READ_REGISTER, {getOrCreateVReg(CI)}, {})
.addMetadata(cast<MDNode>(cast<MetadataAsValue>(Arg)->getMetadata()));
return true;
}
case Intrinsic::write_register: {
Value *Arg = CI.getArgOperand(0);
MIRBuilder.buildInstr(TargetOpcode::G_WRITE_REGISTER)
.addMetadata(cast<MDNode>(cast<MetadataAsValue>(Arg)->getMetadata()))
.addUse(getOrCreateVReg(*CI.getArgOperand(1)));
return true;
}
case Intrinsic::localescape: {
MachineBasicBlock &EntryMBB = MF->front();
StringRef EscapedName = GlobalValue::dropLLVMManglingEscape(MF->getName());
// Directly emit some LOCAL_ESCAPE machine instrs. Label assignment emission
// is the same on all targets.
for (unsigned Idx = 0, E = CI.arg_size(); Idx < E; ++Idx) {
Value *Arg = CI.getArgOperand(Idx)->stripPointerCasts();
if (isa<ConstantPointerNull>(Arg))
continue; // Skip null pointers. They represent a hole in index space.
int FI = getOrCreateFrameIndex(*cast<AllocaInst>(Arg));
MCSymbol *FrameAllocSym =
MF->getMMI().getContext().getOrCreateFrameAllocSymbol(EscapedName,
Idx);
// This should be inserted at the start of the entry block.
auto LocalEscape =
MIRBuilder.buildInstrNoInsert(TargetOpcode::LOCAL_ESCAPE)
.addSym(FrameAllocSym)
.addFrameIndex(FI);
EntryMBB.insert(EntryMBB.begin(), LocalEscape);
}
return true;
}
case Intrinsic::vector_reduce_fadd:
case Intrinsic::vector_reduce_fmul: {
// Need to check for the reassoc flag to decide whether we want a
// sequential reduction opcode or not.
Register Dst = getOrCreateVReg(CI);
Register ScalarSrc = getOrCreateVReg(*CI.getArgOperand(0));
Register VecSrc = getOrCreateVReg(*CI.getArgOperand(1));
unsigned Opc = 0;
if (!CI.hasAllowReassoc()) {
// The sequential ordering case.
Opc = ID == Intrinsic::vector_reduce_fadd
? TargetOpcode::G_VECREDUCE_SEQ_FADD
: TargetOpcode::G_VECREDUCE_SEQ_FMUL;
MIRBuilder.buildInstr(Opc, {Dst}, {ScalarSrc, VecSrc},
MachineInstr::copyFlagsFromInstruction(CI));
return true;
}
// We split the operation into a separate G_FADD/G_FMUL + the reduce,
// since the associativity doesn't matter.
unsigned ScalarOpc;
if (ID == Intrinsic::vector_reduce_fadd) {
Opc = TargetOpcode::G_VECREDUCE_FADD;
ScalarOpc = TargetOpcode::G_FADD;
} else {
Opc = TargetOpcode::G_VECREDUCE_FMUL;
ScalarOpc = TargetOpcode::G_FMUL;
}
LLT DstTy = MRI->getType(Dst);
auto Rdx = MIRBuilder.buildInstr(
Opc, {DstTy}, {VecSrc}, MachineInstr::copyFlagsFromInstruction(CI));
MIRBuilder.buildInstr(ScalarOpc, {Dst}, {ScalarSrc, Rdx},
MachineInstr::copyFlagsFromInstruction(CI));
return true;
}
case Intrinsic::trap:
case Intrinsic::debugtrap:
case Intrinsic::ubsantrap: {
StringRef TrapFuncName =
CI.getAttributes().getFnAttr("trap-func-name").getValueAsString();
if (TrapFuncName.empty())
break; // Use the default handling.
CallLowering::CallLoweringInfo Info;
if (ID == Intrinsic::ubsantrap) {
Info.OrigArgs.push_back({getOrCreateVRegs(*CI.getArgOperand(0)),
CI.getArgOperand(0)->getType(), 0});
}
Info.Callee = MachineOperand::CreateES(TrapFuncName.data());
Info.CB = &CI;
Info.OrigRet = {Register(), Type::getVoidTy(CI.getContext()), 0};
return CLI->lowerCall(MIRBuilder, Info);
}
case Intrinsic::amdgcn_cs_chain:
return translateCallBase(CI, MIRBuilder);
case Intrinsic::fptrunc_round: {
uint32_t Flags = MachineInstr::copyFlagsFromInstruction(CI);
// Convert the metadata argument to a constant integer
Metadata *MD = cast<MetadataAsValue>(CI.getArgOperand(1))->getMetadata();
std::optional<RoundingMode> RoundMode =
convertStrToRoundingMode(cast<MDString>(MD)->getString());
// Add the Rounding mode as an integer
MIRBuilder
.buildInstr(TargetOpcode::G_INTRINSIC_FPTRUNC_ROUND,
{getOrCreateVReg(CI)},
{getOrCreateVReg(*CI.getArgOperand(0))}, Flags)
.addImm((int)*RoundMode);
return true;
}
case Intrinsic::is_fpclass: {
Value *FpValue = CI.getOperand(0);
ConstantInt *TestMaskValue = cast<ConstantInt>(CI.getOperand(1));
MIRBuilder
.buildInstr(TargetOpcode::G_IS_FPCLASS, {getOrCreateVReg(CI)},
{getOrCreateVReg(*FpValue)})
.addImm(TestMaskValue->getZExtValue());
return true;
}
case Intrinsic::set_fpenv: {
Value *FPEnv = CI.getOperand(0);
MIRBuilder.buildInstr(TargetOpcode::G_SET_FPENV, {},
{getOrCreateVReg(*FPEnv)});
return true;
}
case Intrinsic::reset_fpenv: {
MIRBuilder.buildInstr(TargetOpcode::G_RESET_FPENV, {}, {});
return true;
}
case Intrinsic::set_fpmode: {
Value *FPState = CI.getOperand(0);
MIRBuilder.buildInstr(TargetOpcode::G_SET_FPMODE, {},
{ getOrCreateVReg(*FPState) });
return true;
}
case Intrinsic::reset_fpmode: {
MIRBuilder.buildInstr(TargetOpcode::G_RESET_FPMODE, {}, {});
return true;
}
case Intrinsic::prefetch: {
Value *Addr = CI.getOperand(0);
unsigned RW = cast<ConstantInt>(CI.getOperand(1))->getZExtValue();
unsigned Locality = cast<ConstantInt>(CI.getOperand(2))->getZExtValue();
unsigned CacheType = cast<ConstantInt>(CI.getOperand(3))->getZExtValue();
auto Flags = RW ? MachineMemOperand::MOStore : MachineMemOperand::MOLoad;
auto &MMO = *MF->getMachineMemOperand(MachinePointerInfo(Addr), Flags,
LLT(), Align());
MIRBuilder.buildPrefetch(getOrCreateVReg(*Addr), RW, Locality, CacheType,
MMO);
return true;
}
#define INSTRUCTION(NAME, NARG, ROUND_MODE, INTRINSIC) \
case Intrinsic::INTRINSIC:
#include "llvm/IR/ConstrainedOps.def"
return translateConstrainedFPIntrinsic(cast<ConstrainedFPIntrinsic>(CI),
MIRBuilder);
}
return false;
}
bool IRTranslator::translateInlineAsm(const CallBase &CB,
MachineIRBuilder &MIRBuilder) {
const InlineAsmLowering *ALI = MF->getSubtarget().getInlineAsmLowering();
if (!ALI) {
LLVM_DEBUG(
dbgs() << "Inline asm lowering is not supported for this target yet\n");
return false;
}
return ALI->lowerInlineAsm(
MIRBuilder, CB, [&](const Value &Val) { return getOrCreateVRegs(Val); });
}
bool IRTranslator::translateCallBase(const CallBase &CB,
MachineIRBuilder &MIRBuilder) {
ArrayRef<Register> Res = getOrCreateVRegs(CB);
SmallVector<ArrayRef<Register>, 8> Args;
Register SwiftInVReg = 0;
Register SwiftErrorVReg = 0;
for (const auto &Arg : CB.args()) {
if (CLI->supportSwiftError() && isSwiftError(Arg)) {
assert(SwiftInVReg == 0 && "Expected only one swift error argument");
LLT Ty = getLLTForType(*Arg->getType(), *DL);
SwiftInVReg = MRI->createGenericVirtualRegister(Ty);
MIRBuilder.buildCopy(SwiftInVReg, SwiftError.getOrCreateVRegUseAt(
&CB, &MIRBuilder.getMBB(), Arg));
Args.emplace_back(ArrayRef(SwiftInVReg));
SwiftErrorVReg =
SwiftError.getOrCreateVRegDefAt(&CB, &MIRBuilder.getMBB(), Arg);
continue;
}
Args.push_back(getOrCreateVRegs(*Arg));
}
if (auto *CI = dyn_cast<CallInst>(&CB)) {
if (ORE->enabled()) {
if (MemoryOpRemark::canHandle(CI, *LibInfo)) {
MemoryOpRemark R(*ORE, "gisel-irtranslator-memsize", *DL, *LibInfo);
R.visit(CI);
}
}
}
// We don't set HasCalls on MFI here yet because call lowering may decide to
// optimize into tail calls. Instead, we defer that to selection where a final
// scan is done to check if any instructions are calls.
bool Success =
CLI->lowerCall(MIRBuilder, CB, Res, Args, SwiftErrorVReg,
[&]() { return getOrCreateVReg(*CB.getCalledOperand()); });
// Check if we just inserted a tail call.
if (Success) {
assert(!HasTailCall && "Can't tail call return twice from block?");
const TargetInstrInfo *TII = MF->getSubtarget().getInstrInfo();
HasTailCall = TII->isTailCall(*std::prev(MIRBuilder.getInsertPt()));
}
return Success;
}
bool IRTranslator::translateCall(const User &U, MachineIRBuilder &MIRBuilder) {
const CallInst &CI = cast<CallInst>(U);
auto TII = MF->getTarget().getIntrinsicInfo();
const Function *F = CI.getCalledFunction();
// FIXME: support Windows dllimport function calls and calls through
// weak symbols.
if (F && (F->hasDLLImportStorageClass() ||
(MF->getTarget().getTargetTriple().isOSWindows() &&
F->hasExternalWeakLinkage())))
return false;
// FIXME: support control flow guard targets.
if (CI.countOperandBundlesOfType(LLVMContext::OB_cfguardtarget))
return false;
// FIXME: support statepoints and related.
if (isa<GCStatepointInst, GCRelocateInst, GCResultInst>(U))
return false;
if (CI.isInlineAsm())
return translateInlineAsm(CI, MIRBuilder);
diagnoseDontCall(CI);
Intrinsic::ID ID = Intrinsic::not_intrinsic;
if (F && F->isIntrinsic()) {
ID = F->getIntrinsicID();
if (TII && ID == Intrinsic::not_intrinsic)
ID = static_cast<Intrinsic::ID>(TII->getIntrinsicID(F));
}
if (!F || !F->isIntrinsic() || ID == Intrinsic::not_intrinsic)
return translateCallBase(CI, MIRBuilder);
assert(ID != Intrinsic::not_intrinsic && "unknown intrinsic");
if (translateKnownIntrinsic(CI, ID, MIRBuilder))
return true;
ArrayRef<Register> ResultRegs;
if (!CI.getType()->isVoidTy())
ResultRegs = getOrCreateVRegs(CI);
// Ignore the callsite attributes. Backend code is most likely not expecting
// an intrinsic to sometimes have side effects and sometimes not.
MachineInstrBuilder MIB = MIRBuilder.buildIntrinsic(ID, ResultRegs);
if (isa<FPMathOperator>(CI))
MIB->copyIRFlags(CI);
for (const auto &Arg : enumerate(CI.args())) {
// If this is required to be an immediate, don't materialize it in a
// register.
if (CI.paramHasAttr(Arg.index(), Attribute::ImmArg)) {
if (ConstantInt *CI = dyn_cast<ConstantInt>(Arg.value())) {
// imm arguments are more convenient than cimm (and realistically
// probably sufficient), so use them.
assert(CI->getBitWidth() <= 64 &&
"large intrinsic immediates not handled");
MIB.addImm(CI->getSExtValue());
} else {
MIB.addFPImm(cast<ConstantFP>(Arg.value()));
}
} else if (auto *MDVal = dyn_cast<MetadataAsValue>(Arg.value())) {
auto *MD = MDVal->getMetadata();
auto *MDN = dyn_cast<MDNode>(MD);
if (!MDN) {
if (auto *ConstMD = dyn_cast<ConstantAsMetadata>(MD))
MDN = MDNode::get(MF->getFunction().getContext(), ConstMD);
else // This was probably an MDString.
return false;
}
MIB.addMetadata(MDN);
} else {
ArrayRef<Register> VRegs = getOrCreateVRegs(*Arg.value());
if (VRegs.size() > 1)
return false;
MIB.addUse(VRegs[0]);
}
}
// Add a MachineMemOperand if it is a target mem intrinsic.
const TargetLowering &TLI = *MF->getSubtarget().getTargetLowering();
TargetLowering::IntrinsicInfo Info;
// TODO: Add a GlobalISel version of getTgtMemIntrinsic.
if (TLI.getTgtMemIntrinsic(Info, CI, *MF, ID)) {
Align Alignment = Info.align.value_or(
DL->getABITypeAlign(Info.memVT.getTypeForEVT(F->getContext())));
LLT MemTy = Info.memVT.isSimple()
? getLLTForMVT(Info.memVT.getSimpleVT())
: LLT::scalar(Info.memVT.getStoreSizeInBits());
// TODO: We currently just fallback to address space 0 if getTgtMemIntrinsic
// didn't yield anything useful.
MachinePointerInfo MPI;
if (Info.ptrVal)
MPI = MachinePointerInfo(Info.ptrVal, Info.offset);
else if (Info.fallbackAddressSpace)
MPI = MachinePointerInfo(*Info.fallbackAddressSpace);
MIB.addMemOperand(
MF->getMachineMemOperand(MPI, Info.flags, MemTy, Alignment, CI.getAAMetadata()));
}
return true;
}
bool IRTranslator::findUnwindDestinations(
const BasicBlock *EHPadBB,
BranchProbability Prob,
SmallVectorImpl<std::pair<MachineBasicBlock *, BranchProbability>>
&UnwindDests) {
EHPersonality Personality = classifyEHPersonality(
EHPadBB->getParent()->getFunction().getPersonalityFn());
bool IsMSVCCXX = Personality == EHPersonality::MSVC_CXX;
bool IsCoreCLR = Personality == EHPersonality::CoreCLR;
bool IsWasmCXX = Personality == EHPersonality::Wasm_CXX;
bool IsSEH = isAsynchronousEHPersonality(Personality);
if (IsWasmCXX) {
// Ignore this for now.
return false;
}
while (EHPadBB) {
const Instruction *Pad = EHPadBB->getFirstNonPHI();
BasicBlock *NewEHPadBB = nullptr;
if (isa<LandingPadInst>(Pad)) {
// Stop on landingpads. They are not funclets.
UnwindDests.emplace_back(&getMBB(*EHPadBB), Prob);
break;
}
if (isa<CleanupPadInst>(Pad)) {
// Stop on cleanup pads. Cleanups are always funclet entries for all known
// personalities.
UnwindDests.emplace_back(&getMBB(*EHPadBB), Prob);
UnwindDests.back().first->setIsEHScopeEntry();
UnwindDests.back().first->setIsEHFuncletEntry();
break;
}
if (auto *CatchSwitch = dyn_cast<CatchSwitchInst>(Pad)) {
// Add the catchpad handlers to the possible destinations.
for (const BasicBlock *CatchPadBB : CatchSwitch->handlers()) {
UnwindDests.emplace_back(&getMBB(*CatchPadBB), Prob);
// For MSVC++ and the CLR, catchblocks are funclets and need prologues.
if (IsMSVCCXX || IsCoreCLR)
UnwindDests.back().first->setIsEHFuncletEntry();
if (!IsSEH)
UnwindDests.back().first->setIsEHScopeEntry();
}
NewEHPadBB = CatchSwitch->getUnwindDest();
} else {
continue;
}
BranchProbabilityInfo *BPI = FuncInfo.BPI;
if (BPI && NewEHPadBB)
Prob *= BPI->getEdgeProbability(EHPadBB, NewEHPadBB);
EHPadBB = NewEHPadBB;
}
return true;
}
bool IRTranslator::translateInvoke(const User &U,
MachineIRBuilder &MIRBuilder) {
const InvokeInst &I = cast<InvokeInst>(U);
MCContext &Context = MF->getContext();
const BasicBlock *ReturnBB = I.getSuccessor(0);
const BasicBlock *EHPadBB = I.getSuccessor(1);
const Function *Fn = I.getCalledFunction();
// FIXME: support invoking patchpoint and statepoint intrinsics.
if (Fn && Fn->isIntrinsic())
return false;
// FIXME: support whatever these are.
if (I.countOperandBundlesOfType(LLVMContext::OB_deopt))
return false;
// FIXME: support control flow guard targets.
if (I.countOperandBundlesOfType(LLVMContext::OB_cfguardtarget))
return false;
// FIXME: support Windows exception handling.
if (!isa<LandingPadInst>(EHPadBB->getFirstNonPHI()))
return false;
// FIXME: support Windows dllimport function calls and calls through
// weak symbols.
if (Fn && (Fn->hasDLLImportStorageClass() ||
(MF->getTarget().getTargetTriple().isOSWindows() &&
Fn->hasExternalWeakLinkage())))
return false;
bool LowerInlineAsm = I.isInlineAsm();
bool NeedEHLabel = true;
// Emit the actual call, bracketed by EH_LABELs so that the MF knows about
// the region covered by the try.
MCSymbol *BeginSymbol = nullptr;
if (NeedEHLabel) {
MIRBuilder.buildInstr(TargetOpcode::G_INVOKE_REGION_START);
BeginSymbol = Context.createTempSymbol();
MIRBuilder.buildInstr(TargetOpcode::EH_LABEL).addSym(BeginSymbol);
}
if (LowerInlineAsm) {
if (!translateInlineAsm(I, MIRBuilder))
return false;
} else if (!translateCallBase(I, MIRBuilder))
return false;
MCSymbol *EndSymbol = nullptr;
if (NeedEHLabel) {
EndSymbol = Context.createTempSymbol();
MIRBuilder.buildInstr(TargetOpcode::EH_LABEL).addSym(EndSymbol);
}
SmallVector<std::pair<MachineBasicBlock *, BranchProbability>, 1> UnwindDests;
BranchProbabilityInfo *BPI = FuncInfo.BPI;
MachineBasicBlock *InvokeMBB = &MIRBuilder.getMBB();
BranchProbability EHPadBBProb =
BPI ? BPI->getEdgeProbability(InvokeMBB->getBasicBlock(), EHPadBB)
: BranchProbability::getZero();
if (!findUnwindDestinations(EHPadBB, EHPadBBProb, UnwindDests))
return false;
MachineBasicBlock &EHPadMBB = getMBB(*EHPadBB),
&ReturnMBB = getMBB(*ReturnBB);
// Update successor info.
addSuccessorWithProb(InvokeMBB, &ReturnMBB);
for (auto &UnwindDest : UnwindDests) {
UnwindDest.first->setIsEHPad();
addSuccessorWithProb(InvokeMBB, UnwindDest.first, UnwindDest.second);
}
InvokeMBB->normalizeSuccProbs();
if (NeedEHLabel) {
assert(BeginSymbol && "Expected a begin symbol!");
assert(EndSymbol && "Expected an end symbol!");
MF->addInvoke(&EHPadMBB, BeginSymbol, EndSymbol);
}
MIRBuilder.buildBr(ReturnMBB);
return true;
}
bool IRTranslator::translateCallBr(const User &U,
MachineIRBuilder &MIRBuilder) {
// FIXME: Implement this.
return false;
}
bool IRTranslator::translateLandingPad(const User &U,
MachineIRBuilder &MIRBuilder) {
const LandingPadInst &LP = cast<LandingPadInst>(U);
MachineBasicBlock &MBB = MIRBuilder.getMBB();
MBB.setIsEHPad();
// If there aren't registers to copy the values into (e.g., during SjLj
// exceptions), then don't bother.
auto &TLI = *MF->getSubtarget().getTargetLowering();
const Constant *PersonalityFn = MF->getFunction().getPersonalityFn();
if (TLI.getExceptionPointerRegister(PersonalityFn) == 0 &&
TLI.getExceptionSelectorRegister(PersonalityFn) == 0)
return true;
// If landingpad's return type is token type, we don't create DAG nodes
// for its exception pointer and selector value. The extraction of exception
// pointer or selector value from token type landingpads is not currently
// supported.
if (LP.getType()->isTokenTy())
return true;
// Add a label to mark the beginning of the landing pad. Deletion of the
// landing pad can thus be detected via the MachineModuleInfo.
MIRBuilder.buildInstr(TargetOpcode::EH_LABEL)
.addSym(MF->addLandingPad(&MBB));
// If the unwinder does not preserve all registers, ensure that the
// function marks the clobbered registers as used.
const TargetRegisterInfo &TRI = *MF->getSubtarget().getRegisterInfo();
if (auto *RegMask = TRI.getCustomEHPadPreservedMask(*MF))
MF->getRegInfo().addPhysRegsUsedFromRegMask(RegMask);
LLT Ty = getLLTForType(*LP.getType(), *DL);
Register Undef = MRI->createGenericVirtualRegister(Ty);
MIRBuilder.buildUndef(Undef);
SmallVector<LLT, 2> Tys;
for (Type *Ty : cast<StructType>(LP.getType())->elements())
Tys.push_back(getLLTForType(*Ty, *DL));
assert(Tys.size() == 2 && "Only two-valued landingpads are supported");
// Mark exception register as live in.
Register ExceptionReg = TLI.getExceptionPointerRegister(PersonalityFn);
if (!ExceptionReg)
return false;
MBB.addLiveIn(ExceptionReg);
ArrayRef<Register> ResRegs = getOrCreateVRegs(LP);
MIRBuilder.buildCopy(ResRegs[0], ExceptionReg);
Register SelectorReg = TLI.getExceptionSelectorRegister(PersonalityFn);
if (!SelectorReg)
return false;
MBB.addLiveIn(SelectorReg);
Register PtrVReg = MRI->createGenericVirtualRegister(Tys[0]);
MIRBuilder.buildCopy(PtrVReg, SelectorReg);
MIRBuilder.buildCast(ResRegs[1], PtrVReg);
return true;
}
bool IRTranslator::translateAlloca(const User &U,
MachineIRBuilder &MIRBuilder) {
auto &AI = cast<AllocaInst>(U);
if (AI.isSwiftError())
return true;
if (AI.isStaticAlloca()) {
Register Res = getOrCreateVReg(AI);
int FI = getOrCreateFrameIndex(AI);
MIRBuilder.buildFrameIndex(Res, FI);
return true;
}
// FIXME: support stack probing for Windows.
if (MF->getTarget().getTargetTriple().isOSWindows())
return false;
// Now we're in the harder dynamic case.
Register NumElts = getOrCreateVReg(*AI.getArraySize());
Type *IntPtrIRTy = DL->getIntPtrType(AI.getType());
LLT IntPtrTy = getLLTForType(*IntPtrIRTy, *DL);
if (MRI->getType(NumElts) != IntPtrTy) {
Register ExtElts = MRI->createGenericVirtualRegister(IntPtrTy);
MIRBuilder.buildZExtOrTrunc(ExtElts, NumElts);
NumElts = ExtElts;
}
Type *Ty = AI.getAllocatedType();
Register AllocSize = MRI->createGenericVirtualRegister(IntPtrTy);
Register TySize =
getOrCreateVReg(*ConstantInt::get(IntPtrIRTy, DL->getTypeAllocSize(Ty)));
MIRBuilder.buildMul(AllocSize, NumElts, TySize);
// Round the size of the allocation up to the stack alignment size
// by add SA-1 to the size. This doesn't overflow because we're computing
// an address inside an alloca.
Align StackAlign = MF->getSubtarget().getFrameLowering()->getStackAlign();
auto SAMinusOne = MIRBuilder.buildConstant(IntPtrTy, StackAlign.value() - 1);
auto AllocAdd = MIRBuilder.buildAdd(IntPtrTy, AllocSize, SAMinusOne,
MachineInstr::NoUWrap);
auto AlignCst =
MIRBuilder.buildConstant(IntPtrTy, ~(uint64_t)(StackAlign.value() - 1));
auto AlignedAlloc = MIRBuilder.buildAnd(IntPtrTy, AllocAdd, AlignCst);
Align Alignment = std::max(AI.getAlign(), DL->getPrefTypeAlign(Ty));
if (Alignment <= StackAlign)
Alignment = Align(1);
MIRBuilder.buildDynStackAlloc(getOrCreateVReg(AI), AlignedAlloc, Alignment);
MF->getFrameInfo().CreateVariableSizedObject(Alignment, &AI);
assert(MF->getFrameInfo().hasVarSizedObjects());
return true;
}
bool IRTranslator::translateVAArg(const User &U, MachineIRBuilder &MIRBuilder) {
// FIXME: We may need more info about the type. Because of how LLT works,
// we're completely discarding the i64/double distinction here (amongst
// others). Fortunately the ABIs I know of where that matters don't use va_arg
// anyway but that's not guaranteed.
MIRBuilder.buildInstr(TargetOpcode::G_VAARG, {getOrCreateVReg(U)},
{getOrCreateVReg(*U.getOperand(0)),
DL->getABITypeAlign(U.getType()).value()});
return true;
}
bool IRTranslator::translateUnreachable(const User &U, MachineIRBuilder &MIRBuilder) {
if (!MF->getTarget().Options.TrapUnreachable)
return true;
auto &UI = cast<UnreachableInst>(U);
// We may be able to ignore unreachable behind a noreturn call.
if (MF->getTarget().Options.NoTrapAfterNoreturn) {
const BasicBlock &BB = *UI.getParent();
if (&UI != &BB.front()) {
BasicBlock::const_iterator PredI =
std::prev(BasicBlock::const_iterator(UI));
if (const CallInst *Call = dyn_cast<CallInst>(&*PredI)) {
if (Call->doesNotReturn())
return true;
}
}
}
MIRBuilder.buildIntrinsic(Intrinsic::trap, ArrayRef<Register>());
return true;
}
bool IRTranslator::translateInsertElement(const User &U,
MachineIRBuilder &MIRBuilder) {
// If it is a <1 x Ty> vector, use the scalar as it is
// not a legal vector type in LLT.
if (auto *FVT = dyn_cast<FixedVectorType>(U.getType());
FVT && FVT->getNumElements() == 1)
return translateCopy(U, *U.getOperand(1), MIRBuilder);
Register Res = getOrCreateVReg(U);
Register Val = getOrCreateVReg(*U.getOperand(0));
Register Elt = getOrCreateVReg(*U.getOperand(1));
Register Idx = getOrCreateVReg(*U.getOperand(2));
MIRBuilder.buildInsertVectorElement(Res, Val, Elt, Idx);
return true;
}
bool IRTranslator::translateExtractElement(const User &U,
MachineIRBuilder &MIRBuilder) {
// If it is a <1 x Ty> vector, use the scalar as it is
// not a legal vector type in LLT.
if (cast<FixedVectorType>(U.getOperand(0)->getType())->getNumElements() == 1)
return translateCopy(U, *U.getOperand(0), MIRBuilder);
Register Res = getOrCreateVReg(U);
Register Val = getOrCreateVReg(*U.getOperand(0));
const auto &TLI = *MF->getSubtarget().getTargetLowering();
unsigned PreferredVecIdxWidth = TLI.getVectorIdxTy(*DL).getSizeInBits();
Register Idx;
if (auto *CI = dyn_cast<ConstantInt>(U.getOperand(1))) {
if (CI->getBitWidth() != PreferredVecIdxWidth) {
APInt NewIdx = CI->getValue().zextOrTrunc(PreferredVecIdxWidth);
auto *NewIdxCI = ConstantInt::get(CI->getContext(), NewIdx);
Idx = getOrCreateVReg(*NewIdxCI);
}
}
if (!Idx)
Idx = getOrCreateVReg(*U.getOperand(1));
if (MRI->getType(Idx).getSizeInBits() != PreferredVecIdxWidth) {
const LLT VecIdxTy = LLT::scalar(PreferredVecIdxWidth);
Idx = MIRBuilder.buildZExtOrTrunc(VecIdxTy, Idx).getReg(0);
}
MIRBuilder.buildExtractVectorElement(Res, Val, Idx);
return true;
}
bool IRTranslator::translateShuffleVector(const User &U,
MachineIRBuilder &MIRBuilder) {
ArrayRef<int> Mask;
if (auto *SVI = dyn_cast<ShuffleVectorInst>(&U))
Mask = SVI->getShuffleMask();
else
Mask = cast<ConstantExpr>(U).getShuffleMask();
ArrayRef<int> MaskAlloc = MF->allocateShuffleMask(Mask);
MIRBuilder
.buildInstr(TargetOpcode::G_SHUFFLE_VECTOR, {getOrCreateVReg(U)},
{getOrCreateVReg(*U.getOperand(0)),
getOrCreateVReg(*U.getOperand(1))})
.addShuffleMask(MaskAlloc);
return true;
}
bool IRTranslator::translatePHI(const User &U, MachineIRBuilder &MIRBuilder) {
const PHINode &PI = cast<PHINode>(U);
SmallVector<MachineInstr *, 4> Insts;
for (auto Reg : getOrCreateVRegs(PI)) {
auto MIB = MIRBuilder.buildInstr(TargetOpcode::G_PHI, {Reg}, {});
Insts.push_back(MIB.getInstr());
}
PendingPHIs.emplace_back(&PI, std::move(Insts));
return true;
}
bool IRTranslator::translateAtomicCmpXchg(const User &U,
MachineIRBuilder &MIRBuilder) {
const AtomicCmpXchgInst &I = cast<AtomicCmpXchgInst>(U);
auto &TLI = *MF->getSubtarget().getTargetLowering();
auto Flags = TLI.getAtomicMemOperandFlags(I, *DL);
auto Res = getOrCreateVRegs(I);
Register OldValRes = Res[0];
Register SuccessRes = Res[1];
Register Addr = getOrCreateVReg(*I.getPointerOperand());
Register Cmp = getOrCreateVReg(*I.getCompareOperand());
Register NewVal = getOrCreateVReg(*I.getNewValOperand());
MIRBuilder.buildAtomicCmpXchgWithSuccess(
OldValRes, SuccessRes, Addr, Cmp, NewVal,
*MF->getMachineMemOperand(
MachinePointerInfo(I.getPointerOperand()), Flags, MRI->getType(Cmp),
getMemOpAlign(I), I.getAAMetadata(), nullptr, I.getSyncScopeID(),
I.getSuccessOrdering(), I.getFailureOrdering()));
return true;
}
bool IRTranslator::translateAtomicRMW(const User &U,
MachineIRBuilder &MIRBuilder) {
const AtomicRMWInst &I = cast<AtomicRMWInst>(U);
auto &TLI = *MF->getSubtarget().getTargetLowering();
auto Flags = TLI.getAtomicMemOperandFlags(I, *DL);
Register Res = getOrCreateVReg(I);
Register Addr = getOrCreateVReg(*I.getPointerOperand());
Register Val = getOrCreateVReg(*I.getValOperand());
unsigned Opcode = 0;
switch (I.getOperation()) {
default:
return false;
case AtomicRMWInst::Xchg:
Opcode = TargetOpcode::G_ATOMICRMW_XCHG;
break;
case AtomicRMWInst::Add:
Opcode = TargetOpcode::G_ATOMICRMW_ADD;
break;
case AtomicRMWInst::Sub:
Opcode = TargetOpcode::G_ATOMICRMW_SUB;
break;
case AtomicRMWInst::And:
Opcode = TargetOpcode::G_ATOMICRMW_AND;
break;
case AtomicRMWInst::Nand:
Opcode = TargetOpcode::G_ATOMICRMW_NAND;
break;
case AtomicRMWInst::Or:
Opcode = TargetOpcode::G_ATOMICRMW_OR;
break;
case AtomicRMWInst::Xor:
Opcode = TargetOpcode::G_ATOMICRMW_XOR;
break;
case AtomicRMWInst::Max:
Opcode = TargetOpcode::G_ATOMICRMW_MAX;
break;
case AtomicRMWInst::Min:
Opcode = TargetOpcode::G_ATOMICRMW_MIN;
break;
case AtomicRMWInst::UMax:
Opcode = TargetOpcode::G_ATOMICRMW_UMAX;
break;
case AtomicRMWInst::UMin:
Opcode = TargetOpcode::G_ATOMICRMW_UMIN;
break;
case AtomicRMWInst::FAdd:
Opcode = TargetOpcode::G_ATOMICRMW_FADD;
break;
case AtomicRMWInst::FSub:
Opcode = TargetOpcode::G_ATOMICRMW_FSUB;
break;
case AtomicRMWInst::FMax:
Opcode = TargetOpcode::G_ATOMICRMW_FMAX;
break;
case AtomicRMWInst::FMin:
Opcode = TargetOpcode::G_ATOMICRMW_FMIN;
break;
case AtomicRMWInst::UIncWrap:
Opcode = TargetOpcode::G_ATOMICRMW_UINC_WRAP;
break;
case AtomicRMWInst::UDecWrap:
Opcode = TargetOpcode::G_ATOMICRMW_UDEC_WRAP;
break;
}
MIRBuilder.buildAtomicRMW(
Opcode, Res, Addr, Val,
*MF->getMachineMemOperand(MachinePointerInfo(I.getPointerOperand()),
Flags, MRI->getType(Val), getMemOpAlign(I),
I.getAAMetadata(), nullptr, I.getSyncScopeID(),
I.getOrdering()));
return true;
}
bool IRTranslator::translateFence(const User &U,
MachineIRBuilder &MIRBuilder) {
const FenceInst &Fence = cast<FenceInst>(U);
MIRBuilder.buildFence(static_cast<unsigned>(Fence.getOrdering()),
Fence.getSyncScopeID());
return true;
}
bool IRTranslator::translateFreeze(const User &U,
MachineIRBuilder &MIRBuilder) {
const ArrayRef<Register> DstRegs = getOrCreateVRegs(U);
const ArrayRef<Register> SrcRegs = getOrCreateVRegs(*U.getOperand(0));
assert(DstRegs.size() == SrcRegs.size() &&
"Freeze with different source and destination type?");
for (unsigned I = 0; I < DstRegs.size(); ++I) {
MIRBuilder.buildFreeze(DstRegs[I], SrcRegs[I]);
}
return true;
}
void IRTranslator::finishPendingPhis() {
#ifndef NDEBUG
DILocationVerifier Verifier;
GISelObserverWrapper WrapperObserver(&Verifier);
RAIIDelegateInstaller DelInstall(*MF, &WrapperObserver);
#endif // ifndef NDEBUG
for (auto &Phi : PendingPHIs) {
const PHINode *PI = Phi.first;
if (PI->getType()->isEmptyTy())
continue;
ArrayRef<MachineInstr *> ComponentPHIs = Phi.second;
MachineBasicBlock *PhiMBB = ComponentPHIs[0]->getParent();
EntryBuilder->setDebugLoc(PI->getDebugLoc());
#ifndef NDEBUG
Verifier.setCurrentInst(PI);
#endif // ifndef NDEBUG
SmallSet<const MachineBasicBlock *, 16> SeenPreds;
for (unsigned i = 0; i < PI->getNumIncomingValues(); ++i) {
auto IRPred = PI->getIncomingBlock(i);
ArrayRef<Register> ValRegs = getOrCreateVRegs(*PI->getIncomingValue(i));
for (auto *Pred : getMachinePredBBs({IRPred, PI->getParent()})) {
if (SeenPreds.count(Pred) || !PhiMBB->isPredecessor(Pred))
continue;
SeenPreds.insert(Pred);
for (unsigned j = 0; j < ValRegs.size(); ++j) {
MachineInstrBuilder MIB(*MF, ComponentPHIs[j]);
MIB.addUse(ValRegs[j]);
MIB.addMBB(Pred);
}
}
}
}
}
void IRTranslator::translateDbgValueRecord(Value *V, bool HasArgList,
const DILocalVariable *Variable,
const DIExpression *Expression,
const DebugLoc &DL,
MachineIRBuilder &MIRBuilder) {
assert(Variable->isValidLocationForIntrinsic(DL) &&
"Expected inlined-at fields to agree");
// Act as if we're handling a debug intrinsic.
MIRBuilder.setDebugLoc(DL);
if (!V || HasArgList) {
// DI cannot produce a valid DBG_VALUE, so produce an undef DBG_VALUE to
// terminate any prior location.
MIRBuilder.buildIndirectDbgValue(0, Variable, Expression);
return;
}
if (const auto *CI = dyn_cast<Constant>(V)) {
MIRBuilder.buildConstDbgValue(*CI, Variable, Expression);
return;
}
if (auto *AI = dyn_cast<AllocaInst>(V);
AI && AI->isStaticAlloca() && Expression->startsWithDeref()) {
// If the value is an alloca and the expression starts with a
// dereference, track a stack slot instead of a register, as registers
// may be clobbered.
auto ExprOperands = Expression->getElements();
auto *ExprDerefRemoved =
DIExpression::get(AI->getContext(), ExprOperands.drop_front());
MIRBuilder.buildFIDbgValue(getOrCreateFrameIndex(*AI), Variable,
ExprDerefRemoved);
return;
}
if (translateIfEntryValueArgument(false, V, Variable, Expression, DL,
MIRBuilder))
return;
for (Register Reg : getOrCreateVRegs(*V)) {
// FIXME: This does not handle register-indirect values at offset 0. The
// direct/indirect thing shouldn't really be handled by something as
// implicit as reg+noreg vs reg+imm in the first place, but it seems
// pretty baked in right now.
MIRBuilder.buildDirectDbgValue(Reg, Variable, Expression);
}
return;
}
void IRTranslator::translateDbgDeclareRecord(Value *Address, bool HasArgList,
const DILocalVariable *Variable,
const DIExpression *Expression,
const DebugLoc &DL,
MachineIRBuilder &MIRBuilder) {
if (!Address || isa<UndefValue>(Address)) {
LLVM_DEBUG(dbgs() << "Dropping debug info for " << *Variable << "\n");
return;
}
assert(Variable->isValidLocationForIntrinsic(DL) &&
"Expected inlined-at fields to agree");
auto AI = dyn_cast<AllocaInst>(Address);
if (AI && AI->isStaticAlloca()) {
// Static allocas are tracked at the MF level, no need for DBG_VALUE
// instructions (in fact, they get ignored if they *do* exist).
MF->setVariableDbgInfo(Variable, Expression,
getOrCreateFrameIndex(*AI), DL);
return;
}
if (translateIfEntryValueArgument(true, Address, Variable,
Expression, DL,
MIRBuilder))
return;
// A dbg.declare describes the address of a source variable, so lower it
// into an indirect DBG_VALUE.
MIRBuilder.setDebugLoc(DL);
MIRBuilder.buildIndirectDbgValue(getOrCreateVReg(*Address),
Variable, Expression);
return;
}
void IRTranslator::translateDbgInfo(const Instruction &Inst,
MachineIRBuilder &MIRBuilder) {
for (DPValue &DPV : DPValue::filter(Inst.getDbgValueRange())) {
const DILocalVariable *Variable = DPV.getVariable();
const DIExpression *Expression = DPV.getExpression();
Value *V = DPV.getVariableLocationOp(0);
if (DPV.isDbgDeclare())
translateDbgDeclareRecord(V, DPV.hasArgList(), Variable,
Expression, DPV.getDebugLoc(), MIRBuilder);
else
translateDbgValueRecord(V, DPV.hasArgList(), Variable,
Expression, DPV.getDebugLoc(), MIRBuilder);
}
}
bool IRTranslator::translate(const Instruction &Inst) {
CurBuilder->setDebugLoc(Inst.getDebugLoc());
CurBuilder->setPCSections(Inst.getMetadata(LLVMContext::MD_pcsections));
auto &TLI = *MF->getSubtarget().getTargetLowering();
if (TLI.fallBackToDAGISel(Inst))
return false;
switch (Inst.getOpcode()) {
#define HANDLE_INST(NUM, OPCODE, CLASS) \
case Instruction::OPCODE: \
return translate##OPCODE(Inst, *CurBuilder.get());
#include "llvm/IR/Instruction.def"
default:
return false;
}
}
bool IRTranslator::translate(const Constant &C, Register Reg) {
// We only emit constants into the entry block from here. To prevent jumpy
// debug behaviour remove debug line.
if (auto CurrInstDL = CurBuilder->getDL())
EntryBuilder->setDebugLoc(DebugLoc());
if (auto CI = dyn_cast<ConstantInt>(&C))
EntryBuilder->buildConstant(Reg, *CI);
else if (auto CF = dyn_cast<ConstantFP>(&C))
EntryBuilder->buildFConstant(Reg, *CF);
else if (isa<UndefValue>(C))
EntryBuilder->buildUndef(Reg);
else if (isa<ConstantPointerNull>(C))
EntryBuilder->buildConstant(Reg, 0);
else if (auto GV = dyn_cast<GlobalValue>(&C))
EntryBuilder->buildGlobalValue(Reg, GV);
else if (auto CAZ = dyn_cast<ConstantAggregateZero>(&C)) {
if (!isa<FixedVectorType>(CAZ->getType()))
return false;
// Return the scalar if it is a <1 x Ty> vector.
unsigned NumElts = CAZ->getElementCount().getFixedValue();
if (NumElts == 1)
return translateCopy(C, *CAZ->getElementValue(0u), *EntryBuilder);
SmallVector<Register, 4> Ops;
for (unsigned I = 0; I < NumElts; ++I) {
Constant &Elt = *CAZ->getElementValue(I);
Ops.push_back(getOrCreateVReg(Elt));
}
EntryBuilder->buildBuildVector(Reg, Ops);
} else if (auto CV = dyn_cast<ConstantDataVector>(&C)) {
// Return the scalar if it is a <1 x Ty> vector.
if (CV->getNumElements() == 1)
return translateCopy(C, *CV->getElementAsConstant(0), *EntryBuilder);
SmallVector<Register, 4> Ops;
for (unsigned i = 0; i < CV->getNumElements(); ++i) {
Constant &Elt = *CV->getElementAsConstant(i);
Ops.push_back(getOrCreateVReg(Elt));
}
EntryBuilder->buildBuildVector(Reg, Ops);
} else if (auto CE = dyn_cast<ConstantExpr>(&C)) {
switch(CE->getOpcode()) {
#define HANDLE_INST(NUM, OPCODE, CLASS) \
case Instruction::OPCODE: \
return translate##OPCODE(*CE, *EntryBuilder.get());
#include "llvm/IR/Instruction.def"
default:
return false;
}
} else if (auto CV = dyn_cast<ConstantVector>(&C)) {
if (CV->getNumOperands() == 1)
return translateCopy(C, *CV->getOperand(0), *EntryBuilder);
SmallVector<Register, 4> Ops;
for (unsigned i = 0; i < CV->getNumOperands(); ++i) {
Ops.push_back(getOrCreateVReg(*CV->getOperand(i)));
}
EntryBuilder->buildBuildVector(Reg, Ops);
} else if (auto *BA = dyn_cast<BlockAddress>(&C)) {
EntryBuilder->buildBlockAddress(Reg, BA);
} else
return false;
return true;
}
bool IRTranslator::finalizeBasicBlock(const BasicBlock &BB,
MachineBasicBlock &MBB) {
for (auto &BTB : SL->BitTestCases) {
// Emit header first, if it wasn't already emitted.
if (!BTB.Emitted)
emitBitTestHeader(BTB, BTB.Parent);
BranchProbability UnhandledProb = BTB.Prob;
for (unsigned j = 0, ej = BTB.Cases.size(); j != ej; ++j) {
UnhandledProb -= BTB.Cases[j].ExtraProb;
// Set the current basic block to the mbb we wish to insert the code into
MachineBasicBlock *MBB = BTB.Cases[j].ThisBB;
// If all cases cover a contiguous range, it is not necessary to jump to
// the default block after the last bit test fails. This is because the
// range check during bit test header creation has guaranteed that every
// case here doesn't go outside the range. In this case, there is no need
// to perform the last bit test, as it will always be true. Instead, make
// the second-to-last bit-test fall through to the target of the last bit
// test, and delete the last bit test.
MachineBasicBlock *NextMBB;
if ((BTB.ContiguousRange || BTB.FallthroughUnreachable) && j + 2 == ej) {
// Second-to-last bit-test with contiguous range: fall through to the
// target of the final bit test.
NextMBB = BTB.Cases[j + 1].TargetBB;
} else if (j + 1 == ej) {
// For the last bit test, fall through to Default.
NextMBB = BTB.Default;
} else {
// Otherwise, fall through to the next bit test.
NextMBB = BTB.Cases[j + 1].ThisBB;
}
emitBitTestCase(BTB, NextMBB, UnhandledProb, BTB.Reg, BTB.Cases[j], MBB);
if ((BTB.ContiguousRange || BTB.FallthroughUnreachable) && j + 2 == ej) {
// We need to record the replacement phi edge here that normally
// happens in emitBitTestCase before we delete the case, otherwise the
// phi edge will be lost.
addMachineCFGPred({BTB.Parent->getBasicBlock(),
BTB.Cases[ej - 1].TargetBB->getBasicBlock()},
MBB);
// Since we're not going to use the final bit test, remove it.
BTB.Cases.pop_back();
break;
}
}
// This is "default" BB. We have two jumps to it. From "header" BB and from
// last "case" BB, unless the latter was skipped.
CFGEdge HeaderToDefaultEdge = {BTB.Parent->getBasicBlock(),
BTB.Default->getBasicBlock()};
addMachineCFGPred(HeaderToDefaultEdge, BTB.Parent);
if (!BTB.ContiguousRange) {
addMachineCFGPred(HeaderToDefaultEdge, BTB.Cases.back().ThisBB);
}
}
SL->BitTestCases.clear();
for (auto &JTCase : SL->JTCases) {
// Emit header first, if it wasn't already emitted.
if (!JTCase.first.Emitted)
emitJumpTableHeader(JTCase.second, JTCase.first, JTCase.first.HeaderBB);
emitJumpTable(JTCase.second, JTCase.second.MBB);
}
SL->JTCases.clear();
for (auto &SwCase : SL->SwitchCases)
emitSwitchCase(SwCase, &CurBuilder->getMBB(), *CurBuilder);
SL->SwitchCases.clear();
// Check if we need to generate stack-protector guard checks.
StackProtector &SP = getAnalysis<StackProtector>();
if (SP.shouldEmitSDCheck(BB)) {
const TargetLowering &TLI = *MF->getSubtarget().getTargetLowering();
bool FunctionBasedInstrumentation =
TLI.getSSPStackGuardCheck(*MF->getFunction().getParent());
SPDescriptor.initialize(&BB, &MBB, FunctionBasedInstrumentation);
}
// Handle stack protector.
if (SPDescriptor.shouldEmitFunctionBasedCheckStackProtector()) {
LLVM_DEBUG(dbgs() << "Unimplemented stack protector case\n");
return false;
} else if (SPDescriptor.shouldEmitStackProtector()) {
MachineBasicBlock *ParentMBB = SPDescriptor.getParentMBB();
MachineBasicBlock *SuccessMBB = SPDescriptor.getSuccessMBB();
// Find the split point to split the parent mbb. At the same time copy all
// physical registers used in the tail of parent mbb into virtual registers
// before the split point and back into physical registers after the split
// point. This prevents us needing to deal with Live-ins and many other
// register allocation issues caused by us splitting the parent mbb. The
// register allocator will clean up said virtual copies later on.
MachineBasicBlock::iterator SplitPoint = findSplitPointForStackProtector(
ParentMBB, *MF->getSubtarget().getInstrInfo());
// Splice the terminator of ParentMBB into SuccessMBB.
SuccessMBB->splice(SuccessMBB->end(), ParentMBB, SplitPoint,
ParentMBB->end());
// Add compare/jump on neq/jump to the parent BB.
if (!emitSPDescriptorParent(SPDescriptor, ParentMBB))
return false;
// CodeGen Failure MBB if we have not codegened it yet.
MachineBasicBlock *FailureMBB = SPDescriptor.getFailureMBB();
if (FailureMBB->empty()) {
if (!emitSPDescriptorFailure(SPDescriptor, FailureMBB))
return false;
}
// Clear the Per-BB State.
SPDescriptor.resetPerBBState();
}
return true;
}
bool IRTranslator::emitSPDescriptorParent(StackProtectorDescriptor &SPD,
MachineBasicBlock *ParentBB) {
CurBuilder->setInsertPt(*ParentBB, ParentBB->end());
// First create the loads to the guard/stack slot for the comparison.
const TargetLowering &TLI = *MF->getSubtarget().getTargetLowering();
Type *PtrIRTy = PointerType::getUnqual(MF->getFunction().getContext());
const LLT PtrTy = getLLTForType(*PtrIRTy, *DL);
LLT PtrMemTy = getLLTForMVT(TLI.getPointerMemTy(*DL));
MachineFrameInfo &MFI = ParentBB->getParent()->getFrameInfo();
int FI = MFI.getStackProtectorIndex();
Register Guard;
Register StackSlotPtr = CurBuilder->buildFrameIndex(PtrTy, FI).getReg(0);
const Module &M = *ParentBB->getParent()->getFunction().getParent();
Align Align = DL->getPrefTypeAlign(PointerType::getUnqual(M.getContext()));
// Generate code to load the content of the guard slot.
Register GuardVal =
CurBuilder
->buildLoad(PtrMemTy, StackSlotPtr,
MachinePointerInfo::getFixedStack(*MF, FI), Align,
MachineMemOperand::MOLoad | MachineMemOperand::MOVolatile)
.getReg(0);
if (TLI.useStackGuardXorFP()) {
LLVM_DEBUG(dbgs() << "Stack protector xor'ing with FP not yet implemented");
return false;
}
// Retrieve guard check function, nullptr if instrumentation is inlined.
if (const Function *GuardCheckFn = TLI.getSSPStackGuardCheck(M)) {
// This path is currently untestable on GlobalISel, since the only platform
// that needs this seems to be Windows, and we fall back on that currently.
// The code still lives here in case that changes.
// Silence warning about unused variable until the code below that uses
// 'GuardCheckFn' is enabled.
(void)GuardCheckFn;
return false;
#if 0
// The target provides a guard check function to validate the guard value.
// Generate a call to that function with the content of the guard slot as
// argument.
FunctionType *FnTy = GuardCheckFn->getFunctionType();
assert(FnTy->getNumParams() == 1 && "Invalid function signature");
ISD::ArgFlagsTy Flags;
if (GuardCheckFn->hasAttribute(1, Attribute::AttrKind::InReg))
Flags.setInReg();
CallLowering::ArgInfo GuardArgInfo(
{GuardVal, FnTy->getParamType(0), {Flags}});
CallLowering::CallLoweringInfo Info;
Info.OrigArgs.push_back(GuardArgInfo);
Info.CallConv = GuardCheckFn->getCallingConv();
Info.Callee = MachineOperand::CreateGA(GuardCheckFn, 0);
Info.OrigRet = {Register(), FnTy->getReturnType()};
if (!CLI->lowerCall(MIRBuilder, Info)) {
LLVM_DEBUG(dbgs() << "Failed to lower call to stack protector check\n");
return false;
}
return true;
#endif
}
// If useLoadStackGuardNode returns true, generate LOAD_STACK_GUARD.
// Otherwise, emit a volatile load to retrieve the stack guard value.
if (TLI.useLoadStackGuardNode()) {
Guard =
MRI->createGenericVirtualRegister(LLT::scalar(PtrTy.getSizeInBits()));
getStackGuard(Guard, *CurBuilder);
} else {
// TODO: test using android subtarget when we support @llvm.thread.pointer.
const Value *IRGuard = TLI.getSDagStackGuard(M);
Register GuardPtr = getOrCreateVReg(*IRGuard);
Guard = CurBuilder
->buildLoad(PtrMemTy, GuardPtr,
MachinePointerInfo::getFixedStack(*MF, FI), Align,
MachineMemOperand::MOLoad |
MachineMemOperand::MOVolatile)
.getReg(0);
}
// Perform the comparison.
auto Cmp =
CurBuilder->buildICmp(CmpInst::ICMP_NE, LLT::scalar(1), Guard, GuardVal);
// If the guard/stackslot do not equal, branch to failure MBB.
CurBuilder->buildBrCond(Cmp, *SPD.getFailureMBB());
// Otherwise branch to success MBB.
CurBuilder->buildBr(*SPD.getSuccessMBB());
return true;
}
bool IRTranslator::emitSPDescriptorFailure(StackProtectorDescriptor &SPD,
MachineBasicBlock *FailureBB) {
CurBuilder->setInsertPt(*FailureBB, FailureBB->end());
const TargetLowering &TLI = *MF->getSubtarget().getTargetLowering();
const RTLIB::Libcall Libcall = RTLIB::STACKPROTECTOR_CHECK_FAIL;
const char *Name = TLI.getLibcallName(Libcall);
CallLowering::CallLoweringInfo Info;
Info.CallConv = TLI.getLibcallCallingConv(Libcall);
Info.Callee = MachineOperand::CreateES(Name);
Info.OrigRet = {Register(), Type::getVoidTy(MF->getFunction().getContext()),
0};
if (!CLI->lowerCall(*CurBuilder, Info)) {
LLVM_DEBUG(dbgs() << "Failed to lower call to stack protector fail\n");
return false;
}
// On PS4/PS5, the "return address" must still be within the calling
// function, even if it's at the very end, so emit an explicit TRAP here.
// WebAssembly needs an unreachable instruction after a non-returning call,
// because the function return type can be different from __stack_chk_fail's
// return type (void).
const TargetMachine &TM = MF->getTarget();
if (TM.getTargetTriple().isPS() || TM.getTargetTriple().isWasm()) {
LLVM_DEBUG(dbgs() << "Unhandled trap emission for stack protector fail\n");
return false;
}
return true;
}
void IRTranslator::finalizeFunction() {
// Release the memory used by the different maps we
// needed during the translation.
PendingPHIs.clear();
VMap.reset();
FrameIndices.clear();
MachinePreds.clear();
// MachineIRBuilder::DebugLoc can outlive the DILocation it holds. Clear it
// to avoid accessing free’d memory (in runOnMachineFunction) and to avoid
// destroying it twice (in ~IRTranslator() and ~LLVMContext())
EntryBuilder.reset();
CurBuilder.reset();
FuncInfo.clear();
SPDescriptor.resetPerFunctionState();
}
/// Returns true if a BasicBlock \p BB within a variadic function contains a
/// variadic musttail call.
static bool checkForMustTailInVarArgFn(bool IsVarArg, const BasicBlock &BB) {
if (!IsVarArg)
return false;
// Walk the block backwards, because tail calls usually only appear at the end
// of a block.
return llvm::any_of(llvm::reverse(BB), [](const Instruction &I) {
const auto *CI = dyn_cast<CallInst>(&I);
return CI && CI->isMustTailCall();
});
}
bool IRTranslator::runOnMachineFunction(MachineFunction &CurMF) {
MF = &CurMF;
const Function &F = MF->getFunction();
GISelCSEAnalysisWrapper &Wrapper =
getAnalysis<GISelCSEAnalysisWrapperPass>().getCSEWrapper();
// Set the CSEConfig and run the analysis.
GISelCSEInfo *CSEInfo = nullptr;
TPC = &getAnalysis<TargetPassConfig>();
bool EnableCSE = EnableCSEInIRTranslator.getNumOccurrences()
? EnableCSEInIRTranslator
: TPC->isGISelCSEEnabled();
if (EnableCSE) {
EntryBuilder = std::make_unique<CSEMIRBuilder>(CurMF);
CSEInfo = &Wrapper.get(TPC->getCSEConfig());
EntryBuilder->setCSEInfo(CSEInfo);
CurBuilder = std::make_unique<CSEMIRBuilder>(CurMF);
CurBuilder->setCSEInfo(CSEInfo);
} else {
EntryBuilder = std::make_unique<MachineIRBuilder>();
CurBuilder = std::make_unique<MachineIRBuilder>();
}
CLI = MF->getSubtarget().getCallLowering();
CurBuilder->setMF(*MF);
EntryBuilder->setMF(*MF);
MRI = &MF->getRegInfo();
DL = &F.getParent()->getDataLayout();
ORE = std::make_unique<OptimizationRemarkEmitter>(&F);
const TargetMachine &TM = MF->getTarget();
TM.resetTargetOptions(F);
EnableOpts = OptLevel != CodeGenOptLevel::None && !skipFunction(F);
FuncInfo.MF = MF;
if (EnableOpts) {
AA = &getAnalysis<AAResultsWrapperPass>().getAAResults();
FuncInfo.BPI = &getAnalysis<BranchProbabilityInfoWrapperPass>().getBPI();
} else {
AA = nullptr;
FuncInfo.BPI = nullptr;
}
AC = &getAnalysis<AssumptionCacheTracker>().getAssumptionCache(
MF->getFunction());
LibInfo = &getAnalysis<TargetLibraryInfoWrapperPass>().getTLI(F);
FuncInfo.CanLowerReturn = CLI->checkReturnTypeForCallConv(*MF);
const auto &TLI = *MF->getSubtarget().getTargetLowering();
SL = std::make_unique<GISelSwitchLowering>(this, FuncInfo);
SL->init(TLI, TM, *DL);
assert(PendingPHIs.empty() && "stale PHIs");
// Targets which want to use big endian can enable it using
// enableBigEndian()
if (!DL->isLittleEndian() && !CLI->enableBigEndian()) {
// Currently we don't properly handle big endian code.
OptimizationRemarkMissed R("gisel-irtranslator", "GISelFailure",
F.getSubprogram(), &F.getEntryBlock());
R << "unable to translate in big endian mode";
reportTranslationError(*MF, *TPC, *ORE, R);
}
// Release the per-function state when we return, whether we succeeded or not.
auto FinalizeOnReturn = make_scope_exit([this]() { finalizeFunction(); });
// Setup a separate basic-block for the arguments and constants
MachineBasicBlock *EntryBB = MF->CreateMachineBasicBlock();
MF->push_back(EntryBB);
EntryBuilder->setMBB(*EntryBB);
DebugLoc DbgLoc = F.getEntryBlock().getFirstNonPHI()->getDebugLoc();
SwiftError.setFunction(CurMF);
SwiftError.createEntriesInEntryBlock(DbgLoc);
bool IsVarArg = F.isVarArg();
bool HasMustTailInVarArgFn = false;
// Create all blocks, in IR order, to preserve the layout.
for (const BasicBlock &BB: F) {
auto *&MBB = BBToMBB[&BB];
MBB = MF->CreateMachineBasicBlock(&BB);
MF->push_back(MBB);
if (BB.hasAddressTaken())
MBB->setAddressTakenIRBlock(const_cast<BasicBlock *>(&BB));
if (!HasMustTailInVarArgFn)
HasMustTailInVarArgFn = checkForMustTailInVarArgFn(IsVarArg, BB);
}
MF->getFrameInfo().setHasMustTailInVarArgFunc(HasMustTailInVarArgFn);
// Make our arguments/constants entry block fallthrough to the IR entry block.
EntryBB->addSuccessor(&getMBB(F.front()));
if (CLI->fallBackToDAGISel(*MF)) {
OptimizationRemarkMissed R("gisel-irtranslator", "GISelFailure",
F.getSubprogram(), &F.getEntryBlock());
R << "unable to lower function: " << ore::NV("Prototype", F.getType());
reportTranslationError(*MF, *TPC, *ORE, R);
return false;
}
// Lower the actual args into this basic block.
SmallVector<ArrayRef<Register>, 8> VRegArgs;
for (const Argument &Arg: F.args()) {
if (DL->getTypeStoreSize(Arg.getType()).isZero())
continue; // Don't handle zero sized types.
ArrayRef<Register> VRegs = getOrCreateVRegs(Arg);
VRegArgs.push_back(VRegs);
if (Arg.hasSwiftErrorAttr()) {
assert(VRegs.size() == 1 && "Too many vregs for Swift error");
SwiftError.setCurrentVReg(EntryBB, SwiftError.getFunctionArg(), VRegs[0]);
}
}
if (!CLI->lowerFormalArguments(*EntryBuilder, F, VRegArgs, FuncInfo)) {
OptimizationRemarkMissed R("gisel-irtranslator", "GISelFailure",
F.getSubprogram(), &F.getEntryBlock());
R << "unable to lower arguments: " << ore::NV("Prototype", F.getType());
reportTranslationError(*MF, *TPC, *ORE, R);
return false;
}
// Need to visit defs before uses when translating instructions.
GISelObserverWrapper WrapperObserver;
if (EnableCSE && CSEInfo)
WrapperObserver.addObserver(CSEInfo);
{
ReversePostOrderTraversal<const Function *> RPOT(&F);
#ifndef NDEBUG
DILocationVerifier Verifier;
WrapperObserver.addObserver(&Verifier);
#endif // ifndef NDEBUG
RAIIDelegateInstaller DelInstall(*MF, &WrapperObserver);
RAIIMFObserverInstaller ObsInstall(*MF, WrapperObserver);
for (const BasicBlock *BB : RPOT) {
MachineBasicBlock &MBB = getMBB(*BB);
// Set the insertion point of all the following translations to
// the end of this basic block.
CurBuilder->setMBB(MBB);
HasTailCall = false;
for (const Instruction &Inst : *BB) {
// If we translated a tail call in the last step, then we know
// everything after the call is either a return, or something that is
// handled by the call itself. (E.g. a lifetime marker or assume
// intrinsic.) In this case, we should stop translating the block and
// move on.
if (HasTailCall)
break;
#ifndef NDEBUG
Verifier.setCurrentInst(&Inst);
#endif // ifndef NDEBUG
// Translate any debug-info attached to the instruction.
translateDbgInfo(Inst, *CurBuilder.get());
if (translate(Inst))
continue;
OptimizationRemarkMissed R("gisel-irtranslator", "GISelFailure",
Inst.getDebugLoc(), BB);
R << "unable to translate instruction: " << ore::NV("Opcode", &Inst);
if (ORE->allowExtraAnalysis("gisel-irtranslator")) {
std::string InstStrStorage;
raw_string_ostream InstStr(InstStrStorage);
InstStr << Inst;
R << ": '" << InstStr.str() << "'";
}
reportTranslationError(*MF, *TPC, *ORE, R);
return false;
}
if (!finalizeBasicBlock(*BB, MBB)) {
OptimizationRemarkMissed R("gisel-irtranslator", "GISelFailure",
BB->getTerminator()->getDebugLoc(), BB);
R << "unable to translate basic block";
reportTranslationError(*MF, *TPC, *ORE, R);
return false;
}
}
#ifndef NDEBUG
WrapperObserver.removeObserver(&Verifier);
#endif
}
finishPendingPhis();
SwiftError.propagateVRegs();
// Merge the argument lowering and constants block with its single
// successor, the LLVM-IR entry block. We want the basic block to
// be maximal.
assert(EntryBB->succ_size() == 1 &&
"Custom BB used for lowering should have only one successor");
// Get the successor of the current entry block.
MachineBasicBlock &NewEntryBB = **EntryBB->succ_begin();
assert(NewEntryBB.pred_size() == 1 &&
"LLVM-IR entry block has a predecessor!?");
// Move all the instruction from the current entry block to the
// new entry block.
NewEntryBB.splice(NewEntryBB.begin(), EntryBB, EntryBB->begin(),
EntryBB->end());
// Update the live-in information for the new entry block.
for (const MachineBasicBlock::RegisterMaskPair &LiveIn : EntryBB->liveins())
NewEntryBB.addLiveIn(LiveIn);
NewEntryBB.sortUniqueLiveIns();
// Get rid of the now empty basic block.
EntryBB->removeSuccessor(&NewEntryBB);
MF->remove(EntryBB);
MF->deleteMachineBasicBlock(EntryBB);
assert(&MF->front() == &NewEntryBB &&
"New entry wasn't next in the list of basic block!");
// Initialize stack protector information.
StackProtector &SP = getAnalysis<StackProtector>();
SP.copyToMachineFrameInfo(MF->getFrameInfo());
return false;
}
|