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
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
|
//===---- TargetInfo.cpp - Encapsulate target details -----------*- C++ -*-===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// These classes wrap the information about a call or function
// definition used to handle ABI compliancy.
//
//===----------------------------------------------------------------------===//
#include "TargetInfo.h"
#include "ABIInfo.h"
#include "CGCXXABI.h"
#include "CodeGenFunction.h"
#include "clang/AST/RecordLayout.h"
#include "clang/Frontend/CodeGenOptions.h"
#include "llvm/ADT/Triple.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/Type.h"
#include "llvm/Support/raw_ostream.h"
using namespace clang;
using namespace CodeGen;
static void AssignToArrayRange(CodeGen::CGBuilderTy &Builder,
llvm::Value *Array,
llvm::Value *Value,
unsigned FirstIndex,
unsigned LastIndex) {
// Alternatively, we could emit this as a loop in the source.
for (unsigned I = FirstIndex; I <= LastIndex; ++I) {
llvm::Value *Cell = Builder.CreateConstInBoundsGEP1_32(Array, I);
Builder.CreateStore(Value, Cell);
}
}
static bool isAggregateTypeForABI(QualType T) {
return !CodeGenFunction::hasScalarEvaluationKind(T) ||
T->isMemberFunctionPointerType();
}
ABIInfo::~ABIInfo() {}
static bool isRecordReturnIndirect(const RecordType *RT, CodeGen::CodeGenTypes &CGT) {
const CXXRecordDecl *RD = dyn_cast<CXXRecordDecl>(RT->getDecl());
if (!RD)
return false;
return CGT.CGM.getCXXABI().isReturnTypeIndirect(RD);
}
static bool isRecordReturnIndirect(QualType T, CodeGen::CodeGenTypes &CGT) {
const RecordType *RT = T->getAs<RecordType>();
if (!RT)
return false;
return isRecordReturnIndirect(RT, CGT);
}
static CGCXXABI::RecordArgABI getRecordArgABI(const RecordType *RT,
CodeGen::CodeGenTypes &CGT) {
const CXXRecordDecl *RD = dyn_cast<CXXRecordDecl>(RT->getDecl());
if (!RD)
return CGCXXABI::RAA_Default;
return CGT.CGM.getCXXABI().getRecordArgABI(RD);
}
static CGCXXABI::RecordArgABI getRecordArgABI(QualType T,
CodeGen::CodeGenTypes &CGT) {
const RecordType *RT = T->getAs<RecordType>();
if (!RT)
return CGCXXABI::RAA_Default;
return getRecordArgABI(RT, CGT);
}
ASTContext &ABIInfo::getContext() const {
return CGT.getContext();
}
llvm::LLVMContext &ABIInfo::getVMContext() const {
return CGT.getLLVMContext();
}
const llvm::DataLayout &ABIInfo::getDataLayout() const {
return CGT.getDataLayout();
}
const TargetInfo &ABIInfo::getTarget() const {
return CGT.getTarget();
}
void ABIArgInfo::dump() const {
raw_ostream &OS = llvm::errs();
OS << "(ABIArgInfo Kind=";
switch (TheKind) {
case Direct:
OS << "Direct Type=";
if (llvm::Type *Ty = getCoerceToType())
Ty->print(OS);
else
OS << "null";
break;
case Extend:
OS << "Extend";
break;
case Ignore:
OS << "Ignore";
break;
case Indirect:
OS << "Indirect Align=" << getIndirectAlign()
<< " ByVal=" << getIndirectByVal()
<< " Realign=" << getIndirectRealign();
break;
case Expand:
OS << "Expand";
break;
}
OS << ")\n";
}
TargetCodeGenInfo::~TargetCodeGenInfo() { delete Info; }
// If someone can figure out a general rule for this, that would be great.
// It's probably just doomed to be platform-dependent, though.
unsigned TargetCodeGenInfo::getSizeOfUnwindException() const {
// Verified for:
// x86-64 FreeBSD, Linux, Darwin
// x86-32 FreeBSD, Linux, Darwin
// PowerPC Linux, Darwin
// ARM Darwin (*not* EABI)
// AArch64 Linux
return 32;
}
bool TargetCodeGenInfo::isNoProtoCallVariadic(const CallArgList &args,
const FunctionNoProtoType *fnType) const {
// The following conventions are known to require this to be false:
// x86_stdcall
// MIPS
// For everything else, we just prefer false unless we opt out.
return false;
}
static bool isEmptyRecord(ASTContext &Context, QualType T, bool AllowArrays);
/// isEmptyField - Return true iff a the field is "empty", that is it
/// is an unnamed bit-field or an (array of) empty record(s).
static bool isEmptyField(ASTContext &Context, const FieldDecl *FD,
bool AllowArrays) {
if (FD->isUnnamedBitfield())
return true;
QualType FT = FD->getType();
// Constant arrays of empty records count as empty, strip them off.
// Constant arrays of zero length always count as empty.
if (AllowArrays)
while (const ConstantArrayType *AT = Context.getAsConstantArrayType(FT)) {
if (AT->getSize() == 0)
return true;
FT = AT->getElementType();
}
const RecordType *RT = FT->getAs<RecordType>();
if (!RT)
return false;
// C++ record fields are never empty, at least in the Itanium ABI.
//
// FIXME: We should use a predicate for whether this behavior is true in the
// current ABI.
if (isa<CXXRecordDecl>(RT->getDecl()))
return false;
return isEmptyRecord(Context, FT, AllowArrays);
}
/// isEmptyRecord - Return true iff a structure contains only empty
/// fields. Note that a structure with a flexible array member is not
/// considered empty.
static bool isEmptyRecord(ASTContext &Context, QualType T, bool AllowArrays) {
const RecordType *RT = T->getAs<RecordType>();
if (!RT)
return 0;
const RecordDecl *RD = RT->getDecl();
if (RD->hasFlexibleArrayMember())
return false;
// If this is a C++ record, check the bases first.
if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD))
for (CXXRecordDecl::base_class_const_iterator i = CXXRD->bases_begin(),
e = CXXRD->bases_end(); i != e; ++i)
if (!isEmptyRecord(Context, i->getType(), true))
return false;
for (RecordDecl::field_iterator i = RD->field_begin(), e = RD->field_end();
i != e; ++i)
if (!isEmptyField(Context, *i, AllowArrays))
return false;
return true;
}
/// isSingleElementStruct - Determine if a structure is a "single
/// element struct", i.e. it has exactly one non-empty field or
/// exactly one field which is itself a single element
/// struct. Structures with flexible array members are never
/// considered single element structs.
///
/// \return The field declaration for the single non-empty field, if
/// it exists.
static const Type *isSingleElementStruct(QualType T, ASTContext &Context) {
const RecordType *RT = T->getAsStructureType();
if (!RT)
return 0;
const RecordDecl *RD = RT->getDecl();
if (RD->hasFlexibleArrayMember())
return 0;
const Type *Found = 0;
// If this is a C++ record, check the bases first.
if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD)) {
for (CXXRecordDecl::base_class_const_iterator i = CXXRD->bases_begin(),
e = CXXRD->bases_end(); i != e; ++i) {
// Ignore empty records.
if (isEmptyRecord(Context, i->getType(), true))
continue;
// If we already found an element then this isn't a single-element struct.
if (Found)
return 0;
// If this is non-empty and not a single element struct, the composite
// cannot be a single element struct.
Found = isSingleElementStruct(i->getType(), Context);
if (!Found)
return 0;
}
}
// Check for single element.
for (RecordDecl::field_iterator i = RD->field_begin(), e = RD->field_end();
i != e; ++i) {
const FieldDecl *FD = *i;
QualType FT = FD->getType();
// Ignore empty fields.
if (isEmptyField(Context, FD, true))
continue;
// If we already found an element then this isn't a single-element
// struct.
if (Found)
return 0;
// Treat single element arrays as the element.
while (const ConstantArrayType *AT = Context.getAsConstantArrayType(FT)) {
if (AT->getSize().getZExtValue() != 1)
break;
FT = AT->getElementType();
}
if (!isAggregateTypeForABI(FT)) {
Found = FT.getTypePtr();
} else {
Found = isSingleElementStruct(FT, Context);
if (!Found)
return 0;
}
}
// We don't consider a struct a single-element struct if it has
// padding beyond the element type.
if (Found && Context.getTypeSize(Found) != Context.getTypeSize(T))
return 0;
return Found;
}
static bool is32Or64BitBasicType(QualType Ty, ASTContext &Context) {
// Treat complex types as the element type.
if (const ComplexType *CTy = Ty->getAs<ComplexType>())
Ty = CTy->getElementType();
// Check for a type which we know has a simple scalar argument-passing
// convention without any padding. (We're specifically looking for 32
// and 64-bit integer and integer-equivalents, float, and double.)
if (!Ty->getAs<BuiltinType>() && !Ty->hasPointerRepresentation() &&
!Ty->isEnumeralType() && !Ty->isBlockPointerType())
return false;
uint64_t Size = Context.getTypeSize(Ty);
return Size == 32 || Size == 64;
}
/// canExpandIndirectArgument - Test whether an argument type which is to be
/// passed indirectly (on the stack) would have the equivalent layout if it was
/// expanded into separate arguments. If so, we prefer to do the latter to avoid
/// inhibiting optimizations.
///
// FIXME: This predicate is missing many cases, currently it just follows
// llvm-gcc (checks that all fields are 32-bit or 64-bit primitive types). We
// should probably make this smarter, or better yet make the LLVM backend
// capable of handling it.
static bool canExpandIndirectArgument(QualType Ty, ASTContext &Context) {
// We can only expand structure types.
const RecordType *RT = Ty->getAs<RecordType>();
if (!RT)
return false;
// We can only expand (C) structures.
//
// FIXME: This needs to be generalized to handle classes as well.
const RecordDecl *RD = RT->getDecl();
if (!RD->isStruct() || isa<CXXRecordDecl>(RD))
return false;
uint64_t Size = 0;
for (RecordDecl::field_iterator i = RD->field_begin(), e = RD->field_end();
i != e; ++i) {
const FieldDecl *FD = *i;
if (!is32Or64BitBasicType(FD->getType(), Context))
return false;
// FIXME: Reject bit-fields wholesale; there are two problems, we don't know
// how to expand them yet, and the predicate for telling if a bitfield still
// counts as "basic" is more complicated than what we were doing previously.
if (FD->isBitField())
return false;
Size += Context.getTypeSize(FD->getType());
}
// Make sure there are not any holes in the struct.
if (Size != Context.getTypeSize(Ty))
return false;
return true;
}
namespace {
/// DefaultABIInfo - The default implementation for ABI specific
/// details. This implementation provides information which results in
/// self-consistent and sensible LLVM IR generation, but does not
/// conform to any particular ABI.
class DefaultABIInfo : public ABIInfo {
public:
DefaultABIInfo(CodeGen::CodeGenTypes &CGT) : ABIInfo(CGT) {}
ABIArgInfo classifyReturnType(QualType RetTy) const;
ABIArgInfo classifyArgumentType(QualType RetTy) const;
virtual void computeInfo(CGFunctionInfo &FI) const {
FI.getReturnInfo() = classifyReturnType(FI.getReturnType());
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it)
it->info = classifyArgumentType(it->type);
}
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
};
class DefaultTargetCodeGenInfo : public TargetCodeGenInfo {
public:
DefaultTargetCodeGenInfo(CodeGen::CodeGenTypes &CGT)
: TargetCodeGenInfo(new DefaultABIInfo(CGT)) {}
};
llvm::Value *DefaultABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
return 0;
}
ABIArgInfo DefaultABIInfo::classifyArgumentType(QualType Ty) const {
if (isAggregateTypeForABI(Ty)) {
// Records with non trivial destructors/constructors should not be passed
// by value.
if (isRecordReturnIndirect(Ty, CGT))
return ABIArgInfo::getIndirect(0, /*ByVal=*/false);
return ABIArgInfo::getIndirect(0);
}
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
return (Ty->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
ABIArgInfo DefaultABIInfo::classifyReturnType(QualType RetTy) const {
if (RetTy->isVoidType())
return ABIArgInfo::getIgnore();
if (isAggregateTypeForABI(RetTy))
return ABIArgInfo::getIndirect(0);
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = RetTy->getAs<EnumType>())
RetTy = EnumTy->getDecl()->getIntegerType();
return (RetTy->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
//===----------------------------------------------------------------------===//
// le32/PNaCl bitcode ABI Implementation
//
// This is a simplified version of the x86_32 ABI. Arguments and return values
// are always passed on the stack.
//===----------------------------------------------------------------------===//
class PNaClABIInfo : public ABIInfo {
public:
PNaClABIInfo(CodeGen::CodeGenTypes &CGT) : ABIInfo(CGT) {}
ABIArgInfo classifyReturnType(QualType RetTy) const;
ABIArgInfo classifyArgumentType(QualType RetTy) const;
virtual void computeInfo(CGFunctionInfo &FI) const;
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
};
class PNaClTargetCodeGenInfo : public TargetCodeGenInfo {
public:
PNaClTargetCodeGenInfo(CodeGen::CodeGenTypes &CGT)
: TargetCodeGenInfo(new PNaClABIInfo(CGT)) {}
/// For PNaCl we don't want llvm.pow.* intrinsics to be emitted instead
/// of library function calls.
bool emitIntrinsicForPow() const { return false; }
bool addAsmMemoryAroundSyncSynchronize() const { return true; } // @LOCALMOD
bool asmMemoryIsFence() const { return true; } // @LOCALMOD
};
void PNaClABIInfo::computeInfo(CGFunctionInfo &FI) const {
FI.getReturnInfo() = classifyReturnType(FI.getReturnType());
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it)
it->info = classifyArgumentType(it->type);
}
llvm::Value *PNaClABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
return 0;
}
/// \brief Classify argument of given type \p Ty.
ABIArgInfo PNaClABIInfo::classifyArgumentType(QualType Ty) const {
if (isAggregateTypeForABI(Ty)) {
if (CGCXXABI::RecordArgABI RAA = getRecordArgABI(Ty, CGT))
return ABIArgInfo::getIndirect(0, RAA == CGCXXABI::RAA_DirectInMemory);
return ABIArgInfo::getIndirect(0);
} else if (const EnumType *EnumTy = Ty->getAs<EnumType>()) {
// Treat an enum type as its underlying type.
Ty = EnumTy->getDecl()->getIntegerType();
} else if (Ty->isFloatingType()) {
// Floating-point types don't go inreg.
return ABIArgInfo::getDirect();
}
return (Ty->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
ABIArgInfo PNaClABIInfo::classifyReturnType(QualType RetTy) const {
if (RetTy->isVoidType())
return ABIArgInfo::getIgnore();
// In the PNaCl ABI we always return records/structures on the stack.
if (isAggregateTypeForABI(RetTy))
return ABIArgInfo::getIndirect(0);
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = RetTy->getAs<EnumType>())
RetTy = EnumTy->getDecl()->getIntegerType();
return (RetTy->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
/// IsX86_MMXType - Return true if this is an MMX type.
bool IsX86_MMXType(llvm::Type *IRType) {
// Return true if the type is an MMX type <2 x i32>, <4 x i16>, or <8 x i8>.
return IRType->isVectorTy() && IRType->getPrimitiveSizeInBits() == 64 &&
cast<llvm::VectorType>(IRType)->getElementType()->isIntegerTy() &&
IRType->getScalarSizeInBits() != 64;
}
static llvm::Type* X86AdjustInlineAsmType(CodeGen::CodeGenFunction &CGF,
StringRef Constraint,
llvm::Type* Ty) {
if ((Constraint == "y" || Constraint == "&y") && Ty->isVectorTy())
return llvm::Type::getX86_MMXTy(CGF.getLLVMContext());
return Ty;
}
//===----------------------------------------------------------------------===//
// X86-32 ABI Implementation
//===----------------------------------------------------------------------===//
/// X86_32ABIInfo - The X86-32 ABI information.
class X86_32ABIInfo : public ABIInfo {
enum Class {
Integer,
Float
};
static const unsigned MinABIStackAlignInBytes = 4;
bool IsDarwinVectorABI;
bool IsSmallStructInRegABI;
bool IsWin32StructABI;
unsigned DefaultNumRegisterParameters;
static bool isRegisterSize(unsigned Size) {
return (Size == 8 || Size == 16 || Size == 32 || Size == 64);
}
static bool shouldReturnTypeInRegister(QualType Ty, ASTContext &Context,
unsigned callingConvention);
/// getIndirectResult - Give a source type \arg Ty, return a suitable result
/// such that the argument will be passed in memory.
ABIArgInfo getIndirectResult(QualType Ty, bool ByVal,
unsigned &FreeRegs) const;
/// \brief Return the alignment to use for the given type on the stack.
unsigned getTypeStackAlignInBytes(QualType Ty, unsigned Align) const;
Class classify(QualType Ty) const;
ABIArgInfo classifyReturnType(QualType RetTy,
unsigned callingConvention) const;
ABIArgInfo classifyArgumentType(QualType RetTy, unsigned &FreeRegs,
bool IsFastCall) const;
bool shouldUseInReg(QualType Ty, unsigned &FreeRegs,
bool IsFastCall, bool &NeedsPadding) const;
public:
virtual void computeInfo(CGFunctionInfo &FI) const;
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
X86_32ABIInfo(CodeGen::CodeGenTypes &CGT, bool d, bool p, bool w,
unsigned r)
: ABIInfo(CGT), IsDarwinVectorABI(d), IsSmallStructInRegABI(p),
IsWin32StructABI(w), DefaultNumRegisterParameters(r) {}
};
class X86_32TargetCodeGenInfo : public TargetCodeGenInfo {
public:
X86_32TargetCodeGenInfo(CodeGen::CodeGenTypes &CGT,
bool d, bool p, bool w, unsigned r)
:TargetCodeGenInfo(new X86_32ABIInfo(CGT, d, p, w, r)) {}
void SetTargetAttributes(const Decl *D, llvm::GlobalValue *GV,
CodeGen::CodeGenModule &CGM) const;
int getDwarfEHStackPointer(CodeGen::CodeGenModule &CGM) const {
// Darwin uses different dwarf register numbers for EH.
if (CGM.getTarget().getTriple().isOSDarwin()) return 5;
return 4;
}
bool initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const;
llvm::Type* adjustInlineAsmType(CodeGen::CodeGenFunction &CGF,
StringRef Constraint,
llvm::Type* Ty) const {
return X86AdjustInlineAsmType(CGF, Constraint, Ty);
}
};
}
/// shouldReturnTypeInRegister - Determine if the given type should be
/// passed in a register (for the Darwin ABI).
bool X86_32ABIInfo::shouldReturnTypeInRegister(QualType Ty,
ASTContext &Context,
unsigned callingConvention) {
uint64_t Size = Context.getTypeSize(Ty);
// Type must be register sized.
if (!isRegisterSize(Size))
return false;
if (Ty->isVectorType()) {
// 64- and 128- bit vectors inside structures are not returned in
// registers.
if (Size == 64 || Size == 128)
return false;
return true;
}
// If this is a builtin, pointer, enum, complex type, member pointer, or
// member function pointer it is ok.
if (Ty->getAs<BuiltinType>() || Ty->hasPointerRepresentation() ||
Ty->isAnyComplexType() || Ty->isEnumeralType() ||
Ty->isBlockPointerType() || Ty->isMemberPointerType())
return true;
// Arrays are treated like records.
if (const ConstantArrayType *AT = Context.getAsConstantArrayType(Ty))
return shouldReturnTypeInRegister(AT->getElementType(), Context,
callingConvention);
// Otherwise, it must be a record type.
const RecordType *RT = Ty->getAs<RecordType>();
if (!RT) return false;
// FIXME: Traverse bases here too.
// For thiscall conventions, structures will never be returned in
// a register. This is for compatibility with the MSVC ABI
if (callingConvention == llvm::CallingConv::X86_ThisCall &&
RT->isStructureType()) {
return false;
}
// Structure types are passed in register if all fields would be
// passed in a register.
for (RecordDecl::field_iterator i = RT->getDecl()->field_begin(),
e = RT->getDecl()->field_end(); i != e; ++i) {
const FieldDecl *FD = *i;
// Empty fields are ignored.
if (isEmptyField(Context, FD, true))
continue;
// Check fields recursively.
if (!shouldReturnTypeInRegister(FD->getType(), Context,
callingConvention))
return false;
}
return true;
}
ABIArgInfo X86_32ABIInfo::classifyReturnType(QualType RetTy,
unsigned callingConvention) const {
if (RetTy->isVoidType())
return ABIArgInfo::getIgnore();
if (const VectorType *VT = RetTy->getAs<VectorType>()) {
// On Darwin, some vectors are returned in registers.
if (IsDarwinVectorABI) {
uint64_t Size = getContext().getTypeSize(RetTy);
// 128-bit vectors are a special case; they are returned in
// registers and we need to make sure to pick a type the LLVM
// backend will like.
if (Size == 128)
return ABIArgInfo::getDirect(llvm::VectorType::get(
llvm::Type::getInt64Ty(getVMContext()), 2));
// Always return in register if it fits in a general purpose
// register, or if it is 64 bits and has a single element.
if ((Size == 8 || Size == 16 || Size == 32) ||
(Size == 64 && VT->getNumElements() == 1))
return ABIArgInfo::getDirect(llvm::IntegerType::get(getVMContext(),
Size));
return ABIArgInfo::getIndirect(0);
}
return ABIArgInfo::getDirect();
}
if (isAggregateTypeForABI(RetTy)) {
if (const RecordType *RT = RetTy->getAs<RecordType>()) {
if (isRecordReturnIndirect(RT, CGT))
return ABIArgInfo::getIndirect(0, /*ByVal=*/false);
// Structures with flexible arrays are always indirect.
if (RT->getDecl()->hasFlexibleArrayMember())
return ABIArgInfo::getIndirect(0);
}
// If specified, structs and unions are always indirect.
if (!IsSmallStructInRegABI && !RetTy->isAnyComplexType())
return ABIArgInfo::getIndirect(0);
// Small structures which are register sized are generally returned
// in a register.
if (X86_32ABIInfo::shouldReturnTypeInRegister(RetTy, getContext(),
callingConvention)) {
uint64_t Size = getContext().getTypeSize(RetTy);
// As a special-case, if the struct is a "single-element" struct, and
// the field is of type "float" or "double", return it in a
// floating-point register. (MSVC does not apply this special case.)
// We apply a similar transformation for pointer types to improve the
// quality of the generated IR.
if (const Type *SeltTy = isSingleElementStruct(RetTy, getContext()))
if ((!IsWin32StructABI && SeltTy->isRealFloatingType())
|| SeltTy->hasPointerRepresentation())
return ABIArgInfo::getDirect(CGT.ConvertType(QualType(SeltTy, 0)));
// FIXME: We should be able to narrow this integer in cases with dead
// padding.
return ABIArgInfo::getDirect(llvm::IntegerType::get(getVMContext(),Size));
}
return ABIArgInfo::getIndirect(0);
}
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = RetTy->getAs<EnumType>())
RetTy = EnumTy->getDecl()->getIntegerType();
return (RetTy->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
static bool isSSEVectorType(ASTContext &Context, QualType Ty) {
return Ty->getAs<VectorType>() && Context.getTypeSize(Ty) == 128;
}
static bool isRecordWithSSEVectorType(ASTContext &Context, QualType Ty) {
const RecordType *RT = Ty->getAs<RecordType>();
if (!RT)
return 0;
const RecordDecl *RD = RT->getDecl();
// If this is a C++ record, check the bases first.
if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD))
for (CXXRecordDecl::base_class_const_iterator i = CXXRD->bases_begin(),
e = CXXRD->bases_end(); i != e; ++i)
if (!isRecordWithSSEVectorType(Context, i->getType()))
return false;
for (RecordDecl::field_iterator i = RD->field_begin(), e = RD->field_end();
i != e; ++i) {
QualType FT = i->getType();
if (isSSEVectorType(Context, FT))
return true;
if (isRecordWithSSEVectorType(Context, FT))
return true;
}
return false;
}
unsigned X86_32ABIInfo::getTypeStackAlignInBytes(QualType Ty,
unsigned Align) const {
// Otherwise, if the alignment is less than or equal to the minimum ABI
// alignment, just use the default; the backend will handle this.
if (Align <= MinABIStackAlignInBytes)
return 0; // Use default alignment.
// On non-Darwin, the stack type alignment is always 4.
if (!IsDarwinVectorABI) {
// Set explicit alignment, since we may need to realign the top.
return MinABIStackAlignInBytes;
}
// Otherwise, if the type contains an SSE vector type, the alignment is 16.
if (Align >= 16 && (isSSEVectorType(getContext(), Ty) ||
isRecordWithSSEVectorType(getContext(), Ty)))
return 16;
return MinABIStackAlignInBytes;
}
ABIArgInfo X86_32ABIInfo::getIndirectResult(QualType Ty, bool ByVal,
unsigned &FreeRegs) const {
if (!ByVal) {
if (FreeRegs) {
--FreeRegs; // Non byval indirects just use one pointer.
return ABIArgInfo::getIndirectInReg(0, false);
}
return ABIArgInfo::getIndirect(0, false);
}
// Compute the byval alignment.
unsigned TypeAlign = getContext().getTypeAlign(Ty) / 8;
unsigned StackAlign = getTypeStackAlignInBytes(Ty, TypeAlign);
if (StackAlign == 0)
return ABIArgInfo::getIndirect(4);
// If the stack alignment is less than the type alignment, realign the
// argument.
if (StackAlign < TypeAlign)
return ABIArgInfo::getIndirect(StackAlign, /*ByVal=*/true,
/*Realign=*/true);
return ABIArgInfo::getIndirect(StackAlign);
}
X86_32ABIInfo::Class X86_32ABIInfo::classify(QualType Ty) const {
const Type *T = isSingleElementStruct(Ty, getContext());
if (!T)
T = Ty.getTypePtr();
if (const BuiltinType *BT = T->getAs<BuiltinType>()) {
BuiltinType::Kind K = BT->getKind();
if (K == BuiltinType::Float || K == BuiltinType::Double)
return Float;
}
return Integer;
}
bool X86_32ABIInfo::shouldUseInReg(QualType Ty, unsigned &FreeRegs,
bool IsFastCall, bool &NeedsPadding) const {
NeedsPadding = false;
Class C = classify(Ty);
if (C == Float)
return false;
unsigned Size = getContext().getTypeSize(Ty);
unsigned SizeInRegs = (Size + 31) / 32;
if (SizeInRegs == 0)
return false;
if (SizeInRegs > FreeRegs) {
FreeRegs = 0;
return false;
}
FreeRegs -= SizeInRegs;
if (IsFastCall) {
if (Size > 32)
return false;
if (Ty->isIntegralOrEnumerationType())
return true;
if (Ty->isPointerType())
return true;
if (Ty->isReferenceType())
return true;
if (FreeRegs)
NeedsPadding = true;
return false;
}
return true;
}
ABIArgInfo X86_32ABIInfo::classifyArgumentType(QualType Ty,
unsigned &FreeRegs,
bool IsFastCall) const {
// FIXME: Set alignment on indirect arguments.
if (isAggregateTypeForABI(Ty)) {
if (const RecordType *RT = Ty->getAs<RecordType>()) {
if (IsWin32StructABI)
return getIndirectResult(Ty, true, FreeRegs);
if (CGCXXABI::RecordArgABI RAA = getRecordArgABI(RT, CGT))
return getIndirectResult(Ty, RAA == CGCXXABI::RAA_DirectInMemory, FreeRegs);
// Structures with flexible arrays are always indirect.
if (RT->getDecl()->hasFlexibleArrayMember())
return getIndirectResult(Ty, true, FreeRegs);
}
// Ignore empty structs/unions.
if (isEmptyRecord(getContext(), Ty, true))
return ABIArgInfo::getIgnore();
llvm::LLVMContext &LLVMContext = getVMContext();
llvm::IntegerType *Int32 = llvm::Type::getInt32Ty(LLVMContext);
bool NeedsPadding;
if (shouldUseInReg(Ty, FreeRegs, IsFastCall, NeedsPadding)) {
unsigned SizeInRegs = (getContext().getTypeSize(Ty) + 31) / 32;
SmallVector<llvm::Type*, 3> Elements;
for (unsigned I = 0; I < SizeInRegs; ++I)
Elements.push_back(Int32);
llvm::Type *Result = llvm::StructType::get(LLVMContext, Elements);
return ABIArgInfo::getDirectInReg(Result);
}
llvm::IntegerType *PaddingType = NeedsPadding ? Int32 : 0;
// Expand small (<= 128-bit) record types when we know that the stack layout
// of those arguments will match the struct. This is important because the
// LLVM backend isn't smart enough to remove byval, which inhibits many
// optimizations.
if (getContext().getTypeSize(Ty) <= 4*32 &&
canExpandIndirectArgument(Ty, getContext()))
return ABIArgInfo::getExpandWithPadding(IsFastCall, PaddingType);
return getIndirectResult(Ty, true, FreeRegs);
}
if (const VectorType *VT = Ty->getAs<VectorType>()) {
// On Darwin, some vectors are passed in memory, we handle this by passing
// it as an i8/i16/i32/i64.
if (IsDarwinVectorABI) {
uint64_t Size = getContext().getTypeSize(Ty);
if ((Size == 8 || Size == 16 || Size == 32) ||
(Size == 64 && VT->getNumElements() == 1))
return ABIArgInfo::getDirect(llvm::IntegerType::get(getVMContext(),
Size));
}
if (IsX86_MMXType(CGT.ConvertType(Ty)))
return ABIArgInfo::getDirect(llvm::IntegerType::get(getVMContext(), 64));
return ABIArgInfo::getDirect();
}
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
bool NeedsPadding;
bool InReg = shouldUseInReg(Ty, FreeRegs, IsFastCall, NeedsPadding);
if (Ty->isPromotableIntegerType()) {
if (InReg)
return ABIArgInfo::getExtendInReg();
return ABIArgInfo::getExtend();
}
if (InReg)
return ABIArgInfo::getDirectInReg();
return ABIArgInfo::getDirect();
}
void X86_32ABIInfo::computeInfo(CGFunctionInfo &FI) const {
FI.getReturnInfo() = classifyReturnType(FI.getReturnType(),
FI.getCallingConvention());
unsigned CC = FI.getCallingConvention();
bool IsFastCall = CC == llvm::CallingConv::X86_FastCall;
unsigned FreeRegs;
if (IsFastCall)
FreeRegs = 2;
else if (FI.getHasRegParm())
FreeRegs = FI.getRegParm();
else
FreeRegs = DefaultNumRegisterParameters;
// If the return value is indirect, then the hidden argument is consuming one
// integer register.
if (FI.getReturnInfo().isIndirect() && FreeRegs) {
--FreeRegs;
ABIArgInfo &Old = FI.getReturnInfo();
Old = ABIArgInfo::getIndirectInReg(Old.getIndirectAlign(),
Old.getIndirectByVal(),
Old.getIndirectRealign());
}
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it)
it->info = classifyArgumentType(it->type, FreeRegs, IsFastCall);
}
llvm::Value *X86_32ABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
llvm::Type *BPP = CGF.Int8PtrPtrTy;
CGBuilderTy &Builder = CGF.Builder;
llvm::Value *VAListAddrAsBPP = Builder.CreateBitCast(VAListAddr, BPP,
"ap");
llvm::Value *Addr = Builder.CreateLoad(VAListAddrAsBPP, "ap.cur");
// Compute if the address needs to be aligned
unsigned Align = CGF.getContext().getTypeAlignInChars(Ty).getQuantity();
Align = getTypeStackAlignInBytes(Ty, Align);
Align = std::max(Align, 4U);
if (Align > 4) {
// addr = (addr + align - 1) & -align;
llvm::Value *Offset =
llvm::ConstantInt::get(CGF.Int32Ty, Align - 1);
Addr = CGF.Builder.CreateGEP(Addr, Offset);
llvm::Value *AsInt = CGF.Builder.CreatePtrToInt(Addr,
CGF.Int32Ty);
llvm::Value *Mask = llvm::ConstantInt::get(CGF.Int32Ty, -Align);
Addr = CGF.Builder.CreateIntToPtr(CGF.Builder.CreateAnd(AsInt, Mask),
Addr->getType(),
"ap.cur.aligned");
}
llvm::Type *PTy =
llvm::PointerType::getUnqual(CGF.ConvertType(Ty));
llvm::Value *AddrTyped = Builder.CreateBitCast(Addr, PTy);
uint64_t Offset =
llvm::RoundUpToAlignment(CGF.getContext().getTypeSize(Ty) / 8, Align);
llvm::Value *NextAddr =
Builder.CreateGEP(Addr, llvm::ConstantInt::get(CGF.Int32Ty, Offset),
"ap.next");
Builder.CreateStore(NextAddr, VAListAddrAsBPP);
return AddrTyped;
}
void X86_32TargetCodeGenInfo::SetTargetAttributes(const Decl *D,
llvm::GlobalValue *GV,
CodeGen::CodeGenModule &CGM) const {
if (const FunctionDecl *FD = dyn_cast<FunctionDecl>(D)) {
if (FD->hasAttr<X86ForceAlignArgPointerAttr>()) {
// Get the LLVM function.
llvm::Function *Fn = cast<llvm::Function>(GV);
// Now add the 'alignstack' attribute with a value of 16.
llvm::AttrBuilder B;
B.addStackAlignmentAttr(16);
Fn->addAttributes(llvm::AttributeSet::FunctionIndex,
llvm::AttributeSet::get(CGM.getLLVMContext(),
llvm::AttributeSet::FunctionIndex,
B));
}
}
}
bool X86_32TargetCodeGenInfo::initDwarfEHRegSizeTable(
CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const {
CodeGen::CGBuilderTy &Builder = CGF.Builder;
llvm::Value *Four8 = llvm::ConstantInt::get(CGF.Int8Ty, 4);
// 0-7 are the eight integer registers; the order is different
// on Darwin (for EH), but the range is the same.
// 8 is %eip.
AssignToArrayRange(Builder, Address, Four8, 0, 8);
if (CGF.CGM.getTarget().getTriple().isOSDarwin()) {
// 12-16 are st(0..4). Not sure why we stop at 4.
// These have size 16, which is sizeof(long double) on
// platforms with 8-byte alignment for that type.
llvm::Value *Sixteen8 = llvm::ConstantInt::get(CGF.Int8Ty, 16);
AssignToArrayRange(Builder, Address, Sixteen8, 12, 16);
} else {
// 9 is %eflags, which doesn't get a size on Darwin for some
// reason.
Builder.CreateStore(Four8, Builder.CreateConstInBoundsGEP1_32(Address, 9));
// 11-16 are st(0..5). Not sure why we stop at 5.
// These have size 12, which is sizeof(long double) on
// platforms with 4-byte alignment for that type.
llvm::Value *Twelve8 = llvm::ConstantInt::get(CGF.Int8Ty, 12);
AssignToArrayRange(Builder, Address, Twelve8, 11, 16);
}
return false;
}
//===----------------------------------------------------------------------===//
// X86-64 ABI Implementation
//===----------------------------------------------------------------------===//
namespace {
/// X86_64ABIInfo - The X86_64 ABI information.
class X86_64ABIInfo : public ABIInfo {
enum Class {
Integer = 0,
SSE,
SSEUp,
X87,
X87Up,
ComplexX87,
NoClass,
Memory
};
/// merge - Implement the X86_64 ABI merging algorithm.
///
/// Merge an accumulating classification \arg Accum with a field
/// classification \arg Field.
///
/// \param Accum - The accumulating classification. This should
/// always be either NoClass or the result of a previous merge
/// call. In addition, this should never be Memory (the caller
/// should just return Memory for the aggregate).
static Class merge(Class Accum, Class Field);
/// postMerge - Implement the X86_64 ABI post merging algorithm.
///
/// Post merger cleanup, reduces a malformed Hi and Lo pair to
/// final MEMORY or SSE classes when necessary.
///
/// \param AggregateSize - The size of the current aggregate in
/// the classification process.
///
/// \param Lo - The classification for the parts of the type
/// residing in the low word of the containing object.
///
/// \param Hi - The classification for the parts of the type
/// residing in the higher words of the containing object.
///
void postMerge(unsigned AggregateSize, Class &Lo, Class &Hi) const;
/// classify - Determine the x86_64 register classes in which the
/// given type T should be passed.
///
/// \param Lo - The classification for the parts of the type
/// residing in the low word of the containing object.
///
/// \param Hi - The classification for the parts of the type
/// residing in the high word of the containing object.
///
/// \param OffsetBase - The bit offset of this type in the
/// containing object. Some parameters are classified different
/// depending on whether they straddle an eightbyte boundary.
///
/// If a word is unused its result will be NoClass; if a type should
/// be passed in Memory then at least the classification of \arg Lo
/// will be Memory.
///
/// The \arg Lo class will be NoClass iff the argument is ignored.
///
/// If the \arg Lo class is ComplexX87, then the \arg Hi class will
/// also be ComplexX87.
void classify(QualType T, uint64_t OffsetBase, Class &Lo, Class &Hi) const;
llvm::Type *GetByteVectorType(QualType Ty) const;
llvm::Type *GetSSETypeAtOffset(llvm::Type *IRType,
unsigned IROffset, QualType SourceTy,
unsigned SourceOffset) const;
llvm::Type *GetINTEGERTypeAtOffset(llvm::Type *IRType,
unsigned IROffset, QualType SourceTy,
unsigned SourceOffset) const;
/// getIndirectResult - Give a source type \arg Ty, return a suitable result
/// such that the argument will be returned in memory.
ABIArgInfo getIndirectReturnResult(QualType Ty) const;
/// getIndirectResult - Give a source type \arg Ty, return a suitable result
/// such that the argument will be passed in memory.
///
/// \param freeIntRegs - The number of free integer registers remaining
/// available.
ABIArgInfo getIndirectResult(QualType Ty, unsigned freeIntRegs) const;
ABIArgInfo classifyReturnType(QualType RetTy) const;
ABIArgInfo classifyArgumentType(QualType Ty,
unsigned freeIntRegs,
unsigned &neededInt,
unsigned &neededSSE) const;
bool IsIllegalVectorType(QualType Ty) const;
/// The 0.98 ABI revision clarified a lot of ambiguities,
/// unfortunately in ways that were not always consistent with
/// certain previous compilers. In particular, platforms which
/// required strict binary compatibility with older versions of GCC
/// may need to exempt themselves.
bool honorsRevision0_98() const {
return !getTarget().getTriple().isOSDarwin();
}
bool HasAVX;
// Some ABIs (e.g. X32 ABI and Native Client OS) use 32 bit pointers on
// 64-bit hardware.
bool Has64BitPointers;
public:
X86_64ABIInfo(CodeGen::CodeGenTypes &CGT, bool hasavx) :
ABIInfo(CGT), HasAVX(hasavx),
Has64BitPointers(CGT.getDataLayout().getPointerSize(0) == 8) {
}
bool isPassedUsingAVXType(QualType type) const {
unsigned neededInt, neededSSE;
// The freeIntRegs argument doesn't matter here.
ABIArgInfo info = classifyArgumentType(type, 0, neededInt, neededSSE);
if (info.isDirect()) {
llvm::Type *ty = info.getCoerceToType();
if (llvm::VectorType *vectorTy = dyn_cast_or_null<llvm::VectorType>(ty))
return (vectorTy->getBitWidth() > 128);
}
return false;
}
virtual void computeInfo(CGFunctionInfo &FI) const;
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
};
/// WinX86_64ABIInfo - The Windows X86_64 ABI information.
class WinX86_64ABIInfo : public ABIInfo {
ABIArgInfo classify(QualType Ty, bool IsReturnType) const;
public:
WinX86_64ABIInfo(CodeGen::CodeGenTypes &CGT) : ABIInfo(CGT) {}
virtual void computeInfo(CGFunctionInfo &FI) const;
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
};
class X86_64TargetCodeGenInfo : public TargetCodeGenInfo {
public:
X86_64TargetCodeGenInfo(CodeGen::CodeGenTypes &CGT, bool HasAVX)
: TargetCodeGenInfo(new X86_64ABIInfo(CGT, HasAVX)) {}
const X86_64ABIInfo &getABIInfo() const {
return static_cast<const X86_64ABIInfo&>(TargetCodeGenInfo::getABIInfo());
}
int getDwarfEHStackPointer(CodeGen::CodeGenModule &CGM) const {
return 7;
}
bool initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const {
llvm::Value *Eight8 = llvm::ConstantInt::get(CGF.Int8Ty, 8);
// 0-15 are the 16 integer registers.
// 16 is %rip.
AssignToArrayRange(CGF.Builder, Address, Eight8, 0, 16);
return false;
}
llvm::Type* adjustInlineAsmType(CodeGen::CodeGenFunction &CGF,
StringRef Constraint,
llvm::Type* Ty) const {
return X86AdjustInlineAsmType(CGF, Constraint, Ty);
}
bool isNoProtoCallVariadic(const CallArgList &args,
const FunctionNoProtoType *fnType) const {
// The default CC on x86-64 sets %al to the number of SSA
// registers used, and GCC sets this when calling an unprototyped
// function, so we override the default behavior. However, don't do
// that when AVX types are involved: the ABI explicitly states it is
// undefined, and it doesn't work in practice because of how the ABI
// defines varargs anyway.
if (fnType->getCallConv() == CC_Default || fnType->getCallConv() == CC_C) {
bool HasAVXType = false;
for (CallArgList::const_iterator
it = args.begin(), ie = args.end(); it != ie; ++it) {
if (getABIInfo().isPassedUsingAVXType(it->Ty)) {
HasAVXType = true;
break;
}
}
if (!HasAVXType)
return true;
}
return TargetCodeGenInfo::isNoProtoCallVariadic(args, fnType);
}
};
class WinX86_64TargetCodeGenInfo : public TargetCodeGenInfo {
public:
WinX86_64TargetCodeGenInfo(CodeGen::CodeGenTypes &CGT)
: TargetCodeGenInfo(new WinX86_64ABIInfo(CGT)) {}
int getDwarfEHStackPointer(CodeGen::CodeGenModule &CGM) const {
return 7;
}
bool initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const {
llvm::Value *Eight8 = llvm::ConstantInt::get(CGF.Int8Ty, 8);
// 0-15 are the 16 integer registers.
// 16 is %rip.
AssignToArrayRange(CGF.Builder, Address, Eight8, 0, 16);
return false;
}
};
}
void X86_64ABIInfo::postMerge(unsigned AggregateSize, Class &Lo,
Class &Hi) const {
// AMD64-ABI 3.2.3p2: Rule 5. Then a post merger cleanup is done:
//
// (a) If one of the classes is Memory, the whole argument is passed in
// memory.
//
// (b) If X87UP is not preceded by X87, the whole argument is passed in
// memory.
//
// (c) If the size of the aggregate exceeds two eightbytes and the first
// eightbyte isn't SSE or any other eightbyte isn't SSEUP, the whole
// argument is passed in memory. NOTE: This is necessary to keep the
// ABI working for processors that don't support the __m256 type.
//
// (d) If SSEUP is not preceded by SSE or SSEUP, it is converted to SSE.
//
// Some of these are enforced by the merging logic. Others can arise
// only with unions; for example:
// union { _Complex double; unsigned; }
//
// Note that clauses (b) and (c) were added in 0.98.
//
if (Hi == Memory)
Lo = Memory;
if (Hi == X87Up && Lo != X87 && honorsRevision0_98())
Lo = Memory;
if (AggregateSize > 128 && (Lo != SSE || Hi != SSEUp))
Lo = Memory;
if (Hi == SSEUp && Lo != SSE)
Hi = SSE;
}
X86_64ABIInfo::Class X86_64ABIInfo::merge(Class Accum, Class Field) {
// AMD64-ABI 3.2.3p2: Rule 4. Each field of an object is
// classified recursively so that always two fields are
// considered. The resulting class is calculated according to
// the classes of the fields in the eightbyte:
//
// (a) If both classes are equal, this is the resulting class.
//
// (b) If one of the classes is NO_CLASS, the resulting class is
// the other class.
//
// (c) If one of the classes is MEMORY, the result is the MEMORY
// class.
//
// (d) If one of the classes is INTEGER, the result is the
// INTEGER.
//
// (e) If one of the classes is X87, X87UP, COMPLEX_X87 class,
// MEMORY is used as class.
//
// (f) Otherwise class SSE is used.
// Accum should never be memory (we should have returned) or
// ComplexX87 (because this cannot be passed in a structure).
assert((Accum != Memory && Accum != ComplexX87) &&
"Invalid accumulated classification during merge.");
if (Accum == Field || Field == NoClass)
return Accum;
if (Field == Memory)
return Memory;
if (Accum == NoClass)
return Field;
if (Accum == Integer || Field == Integer)
return Integer;
if (Field == X87 || Field == X87Up || Field == ComplexX87 ||
Accum == X87 || Accum == X87Up)
return Memory;
return SSE;
}
void X86_64ABIInfo::classify(QualType Ty, uint64_t OffsetBase,
Class &Lo, Class &Hi) const {
// FIXME: This code can be simplified by introducing a simple value class for
// Class pairs with appropriate constructor methods for the various
// situations.
// FIXME: Some of the split computations are wrong; unaligned vectors
// shouldn't be passed in registers for example, so there is no chance they
// can straddle an eightbyte. Verify & simplify.
Lo = Hi = NoClass;
Class &Current = OffsetBase < 64 ? Lo : Hi;
Current = Memory;
if (const BuiltinType *BT = Ty->getAs<BuiltinType>()) {
BuiltinType::Kind k = BT->getKind();
if (k == BuiltinType::Void) {
Current = NoClass;
} else if (k == BuiltinType::Int128 || k == BuiltinType::UInt128) {
Lo = Integer;
Hi = Integer;
} else if (k >= BuiltinType::Bool && k <= BuiltinType::LongLong) {
Current = Integer;
} else if ((k == BuiltinType::Float || k == BuiltinType::Double) ||
(k == BuiltinType::LongDouble &&
getTarget().getTriple().getOS() == llvm::Triple::NaCl)) {
Current = SSE;
} else if (k == BuiltinType::LongDouble) {
Lo = X87;
Hi = X87Up;
}
// FIXME: _Decimal32 and _Decimal64 are SSE.
// FIXME: _float128 and _Decimal128 are (SSE, SSEUp).
return;
}
if (const EnumType *ET = Ty->getAs<EnumType>()) {
// Classify the underlying integer type.
classify(ET->getDecl()->getIntegerType(), OffsetBase, Lo, Hi);
return;
}
if (Ty->hasPointerRepresentation()) {
Current = Integer;
return;
}
if (Ty->isMemberPointerType()) {
if (Ty->isMemberFunctionPointerType() && Has64BitPointers)
Lo = Hi = Integer;
else
Current = Integer;
return;
}
if (const VectorType *VT = Ty->getAs<VectorType>()) {
uint64_t Size = getContext().getTypeSize(VT);
if (Size == 32) {
// gcc passes all <4 x char>, <2 x short>, <1 x int>, <1 x
// float> as integer.
Current = Integer;
// If this type crosses an eightbyte boundary, it should be
// split.
uint64_t EB_Real = (OffsetBase) / 64;
uint64_t EB_Imag = (OffsetBase + Size - 1) / 64;
if (EB_Real != EB_Imag)
Hi = Lo;
} else if (Size == 64) {
// gcc passes <1 x double> in memory. :(
if (VT->getElementType()->isSpecificBuiltinType(BuiltinType::Double))
return;
// gcc passes <1 x long long> as INTEGER.
if (VT->getElementType()->isSpecificBuiltinType(BuiltinType::LongLong) ||
VT->getElementType()->isSpecificBuiltinType(BuiltinType::ULongLong) ||
VT->getElementType()->isSpecificBuiltinType(BuiltinType::Long) ||
VT->getElementType()->isSpecificBuiltinType(BuiltinType::ULong))
Current = Integer;
else
Current = SSE;
// If this type crosses an eightbyte boundary, it should be
// split.
if (OffsetBase && OffsetBase != 64)
Hi = Lo;
} else if (Size == 128 || (HasAVX && Size == 256)) {
// Arguments of 256-bits are split into four eightbyte chunks. The
// least significant one belongs to class SSE and all the others to class
// SSEUP. The original Lo and Hi design considers that types can't be
// greater than 128-bits, so a 64-bit split in Hi and Lo makes sense.
// This design isn't correct for 256-bits, but since there're no cases
// where the upper parts would need to be inspected, avoid adding
// complexity and just consider Hi to match the 64-256 part.
Lo = SSE;
Hi = SSEUp;
}
return;
}
if (const ComplexType *CT = Ty->getAs<ComplexType>()) {
QualType ET = getContext().getCanonicalType(CT->getElementType());
uint64_t Size = getContext().getTypeSize(Ty);
if (ET->isIntegralOrEnumerationType()) {
if (Size <= 64)
Current = Integer;
else if (Size <= 128)
Lo = Hi = Integer;
} else if (ET == getContext().FloatTy)
Current = SSE;
else if (ET == getContext().DoubleTy ||
(ET == getContext().LongDoubleTy &&
getTarget().getTriple().getOS() == llvm::Triple::NaCl))
Lo = Hi = SSE;
else if (ET == getContext().LongDoubleTy)
Current = ComplexX87;
// If this complex type crosses an eightbyte boundary then it
// should be split.
uint64_t EB_Real = (OffsetBase) / 64;
uint64_t EB_Imag = (OffsetBase + getContext().getTypeSize(ET)) / 64;
if (Hi == NoClass && EB_Real != EB_Imag)
Hi = Lo;
return;
}
if (const ConstantArrayType *AT = getContext().getAsConstantArrayType(Ty)) {
// Arrays are treated like structures.
uint64_t Size = getContext().getTypeSize(Ty);
// AMD64-ABI 3.2.3p2: Rule 1. If the size of an object is larger
// than four eightbytes, ..., it has class MEMORY.
if (Size > 256)
return;
// AMD64-ABI 3.2.3p2: Rule 1. If ..., or it contains unaligned
// fields, it has class MEMORY.
//
// Only need to check alignment of array base.
if (OffsetBase % getContext().getTypeAlign(AT->getElementType()))
return;
// Otherwise implement simplified merge. We could be smarter about
// this, but it isn't worth it and would be harder to verify.
Current = NoClass;
uint64_t EltSize = getContext().getTypeSize(AT->getElementType());
uint64_t ArraySize = AT->getSize().getZExtValue();
// The only case a 256-bit wide vector could be used is when the array
// contains a single 256-bit element. Since Lo and Hi logic isn't extended
// to work for sizes wider than 128, early check and fallback to memory.
if (Size > 128 && EltSize != 256)
return;
for (uint64_t i=0, Offset=OffsetBase; i<ArraySize; ++i, Offset += EltSize) {
Class FieldLo, FieldHi;
classify(AT->getElementType(), Offset, FieldLo, FieldHi);
Lo = merge(Lo, FieldLo);
Hi = merge(Hi, FieldHi);
if (Lo == Memory || Hi == Memory)
break;
}
postMerge(Size, Lo, Hi);
assert((Hi != SSEUp || Lo == SSE) && "Invalid SSEUp array classification.");
return;
}
if (const RecordType *RT = Ty->getAs<RecordType>()) {
uint64_t Size = getContext().getTypeSize(Ty);
// AMD64-ABI 3.2.3p2: Rule 1. If the size of an object is larger
// than four eightbytes, ..., it has class MEMORY.
if (Size > 256)
return;
// AMD64-ABI 3.2.3p2: Rule 2. If a C++ object has either a non-trivial
// copy constructor or a non-trivial destructor, it is passed by invisible
// reference.
if (getRecordArgABI(RT, CGT))
return;
const RecordDecl *RD = RT->getDecl();
// Assume variable sized types are passed in memory.
if (RD->hasFlexibleArrayMember())
return;
const ASTRecordLayout &Layout = getContext().getASTRecordLayout(RD);
// Reset Lo class, this will be recomputed.
Current = NoClass;
// If this is a C++ record, classify the bases first.
if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD)) {
for (CXXRecordDecl::base_class_const_iterator i = CXXRD->bases_begin(),
e = CXXRD->bases_end(); i != e; ++i) {
assert(!i->isVirtual() && !i->getType()->isDependentType() &&
"Unexpected base class!");
const CXXRecordDecl *Base =
cast<CXXRecordDecl>(i->getType()->getAs<RecordType>()->getDecl());
// Classify this field.
//
// AMD64-ABI 3.2.3p2: Rule 3. If the size of the aggregate exceeds a
// single eightbyte, each is classified separately. Each eightbyte gets
// initialized to class NO_CLASS.
Class FieldLo, FieldHi;
uint64_t Offset =
OffsetBase + getContext().toBits(Layout.getBaseClassOffset(Base));
classify(i->getType(), Offset, FieldLo, FieldHi);
Lo = merge(Lo, FieldLo);
Hi = merge(Hi, FieldHi);
if (Lo == Memory || Hi == Memory)
break;
}
}
// Classify the fields one at a time, merging the results.
unsigned idx = 0;
for (RecordDecl::field_iterator i = RD->field_begin(), e = RD->field_end();
i != e; ++i, ++idx) {
uint64_t Offset = OffsetBase + Layout.getFieldOffset(idx);
bool BitField = i->isBitField();
// AMD64-ABI 3.2.3p2: Rule 1. If the size of an object is larger than
// four eightbytes, or it contains unaligned fields, it has class MEMORY.
//
// The only case a 256-bit wide vector could be used is when the struct
// contains a single 256-bit element. Since Lo and Hi logic isn't extended
// to work for sizes wider than 128, early check and fallback to memory.
//
if (Size > 128 && getContext().getTypeSize(i->getType()) != 256) {
Lo = Memory;
return;
}
// Note, skip this test for bit-fields, see below.
if (!BitField && Offset % getContext().getTypeAlign(i->getType())) {
Lo = Memory;
return;
}
// Classify this field.
//
// AMD64-ABI 3.2.3p2: Rule 3. If the size of the aggregate
// exceeds a single eightbyte, each is classified
// separately. Each eightbyte gets initialized to class
// NO_CLASS.
Class FieldLo, FieldHi;
// Bit-fields require special handling, they do not force the
// structure to be passed in memory even if unaligned, and
// therefore they can straddle an eightbyte.
if (BitField) {
// Ignore padding bit-fields.
if (i->isUnnamedBitfield())
continue;
uint64_t Offset = OffsetBase + Layout.getFieldOffset(idx);
uint64_t Size = i->getBitWidthValue(getContext());
uint64_t EB_Lo = Offset / 64;
uint64_t EB_Hi = (Offset + Size - 1) / 64;
FieldLo = FieldHi = NoClass;
if (EB_Lo) {
assert(EB_Hi == EB_Lo && "Invalid classification, type > 16 bytes.");
FieldLo = NoClass;
FieldHi = Integer;
} else {
FieldLo = Integer;
FieldHi = EB_Hi ? Integer : NoClass;
}
} else
classify(i->getType(), Offset, FieldLo, FieldHi);
Lo = merge(Lo, FieldLo);
Hi = merge(Hi, FieldHi);
if (Lo == Memory || Hi == Memory)
break;
}
postMerge(Size, Lo, Hi);
}
}
ABIArgInfo X86_64ABIInfo::getIndirectReturnResult(QualType Ty) const {
// If this is a scalar LLVM value then assume LLVM will pass it in the right
// place naturally.
if (!isAggregateTypeForABI(Ty)) {
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
return (Ty->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
return ABIArgInfo::getIndirect(0);
}
bool X86_64ABIInfo::IsIllegalVectorType(QualType Ty) const {
if (const VectorType *VecTy = Ty->getAs<VectorType>()) {
uint64_t Size = getContext().getTypeSize(VecTy);
unsigned LargestVector = HasAVX ? 256 : 128;
if (Size <= 64 || Size > LargestVector)
return true;
}
return false;
}
ABIArgInfo X86_64ABIInfo::getIndirectResult(QualType Ty,
unsigned freeIntRegs) const {
// If this is a scalar LLVM value then assume LLVM will pass it in the right
// place naturally.
//
// This assumption is optimistic, as there could be free registers available
// when we need to pass this argument in memory, and LLVM could try to pass
// the argument in the free register. This does not seem to happen currently,
// but this code would be much safer if we could mark the argument with
// 'onstack'. See PR12193.
if (!isAggregateTypeForABI(Ty) && !IsIllegalVectorType(Ty)) {
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
return (Ty->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
if (CGCXXABI::RecordArgABI RAA = getRecordArgABI(Ty, CGT))
return ABIArgInfo::getIndirect(0, RAA == CGCXXABI::RAA_DirectInMemory);
// Compute the byval alignment. We specify the alignment of the byval in all
// cases so that the mid-level optimizer knows the alignment of the byval.
unsigned Align = std::max(getContext().getTypeAlign(Ty) / 8, 8U);
// Attempt to avoid passing indirect results using byval when possible. This
// is important for good codegen.
//
// We do this by coercing the value into a scalar type which the backend can
// handle naturally (i.e., without using byval).
//
// For simplicity, we currently only do this when we have exhausted all of the
// free integer registers. Doing this when there are free integer registers
// would require more care, as we would have to ensure that the coerced value
// did not claim the unused register. That would require either reording the
// arguments to the function (so that any subsequent inreg values came first),
// or only doing this optimization when there were no following arguments that
// might be inreg.
//
// We currently expect it to be rare (particularly in well written code) for
// arguments to be passed on the stack when there are still free integer
// registers available (this would typically imply large structs being passed
// by value), so this seems like a fair tradeoff for now.
//
// We can revisit this if the backend grows support for 'onstack' parameter
// attributes. See PR12193.
if (freeIntRegs == 0) {
uint64_t Size = getContext().getTypeSize(Ty);
// If this type fits in an eightbyte, coerce it into the matching integral
// type, which will end up on the stack (with alignment 8).
if (Align == 8 && Size <= 64)
return ABIArgInfo::getDirect(llvm::IntegerType::get(getVMContext(),
Size));
}
return ABIArgInfo::getIndirect(Align);
}
/// GetByteVectorType - The ABI specifies that a value should be passed in an
/// full vector XMM/YMM register. Pick an LLVM IR type that will be passed as a
/// vector register.
llvm::Type *X86_64ABIInfo::GetByteVectorType(QualType Ty) const {
llvm::Type *IRType = CGT.ConvertType(Ty);
// Wrapper structs that just contain vectors are passed just like vectors,
// strip them off if present.
llvm::StructType *STy = dyn_cast<llvm::StructType>(IRType);
while (STy && STy->getNumElements() == 1) {
IRType = STy->getElementType(0);
STy = dyn_cast<llvm::StructType>(IRType);
}
// If the preferred type is a 16-byte vector, prefer to pass it.
if (llvm::VectorType *VT = dyn_cast<llvm::VectorType>(IRType)){
llvm::Type *EltTy = VT->getElementType();
unsigned BitWidth = VT->getBitWidth();
if ((BitWidth >= 128 && BitWidth <= 256) &&
(EltTy->isFloatTy() || EltTy->isDoubleTy() ||
EltTy->isIntegerTy(8) || EltTy->isIntegerTy(16) ||
EltTy->isIntegerTy(32) || EltTy->isIntegerTy(64) ||
EltTy->isIntegerTy(128)))
return VT;
}
return llvm::VectorType::get(llvm::Type::getDoubleTy(getVMContext()), 2);
}
/// BitsContainNoUserData - Return true if the specified [start,end) bit range
/// is known to either be off the end of the specified type or being in
/// alignment padding. The user type specified is known to be at most 128 bits
/// in size, and have passed through X86_64ABIInfo::classify with a successful
/// classification that put one of the two halves in the INTEGER class.
///
/// It is conservatively correct to return false.
static bool BitsContainNoUserData(QualType Ty, unsigned StartBit,
unsigned EndBit, ASTContext &Context) {
// If the bytes being queried are off the end of the type, there is no user
// data hiding here. This handles analysis of builtins, vectors and other
// types that don't contain interesting padding.
unsigned TySize = (unsigned)Context.getTypeSize(Ty);
if (TySize <= StartBit)
return true;
if (const ConstantArrayType *AT = Context.getAsConstantArrayType(Ty)) {
unsigned EltSize = (unsigned)Context.getTypeSize(AT->getElementType());
unsigned NumElts = (unsigned)AT->getSize().getZExtValue();
// Check each element to see if the element overlaps with the queried range.
for (unsigned i = 0; i != NumElts; ++i) {
// If the element is after the span we care about, then we're done..
unsigned EltOffset = i*EltSize;
if (EltOffset >= EndBit) break;
unsigned EltStart = EltOffset < StartBit ? StartBit-EltOffset :0;
if (!BitsContainNoUserData(AT->getElementType(), EltStart,
EndBit-EltOffset, Context))
return false;
}
// If it overlaps no elements, then it is safe to process as padding.
return true;
}
if (const RecordType *RT = Ty->getAs<RecordType>()) {
const RecordDecl *RD = RT->getDecl();
const ASTRecordLayout &Layout = Context.getASTRecordLayout(RD);
// If this is a C++ record, check the bases first.
if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD)) {
for (CXXRecordDecl::base_class_const_iterator i = CXXRD->bases_begin(),
e = CXXRD->bases_end(); i != e; ++i) {
assert(!i->isVirtual() && !i->getType()->isDependentType() &&
"Unexpected base class!");
const CXXRecordDecl *Base =
cast<CXXRecordDecl>(i->getType()->getAs<RecordType>()->getDecl());
// If the base is after the span we care about, ignore it.
unsigned BaseOffset = Context.toBits(Layout.getBaseClassOffset(Base));
if (BaseOffset >= EndBit) continue;
unsigned BaseStart = BaseOffset < StartBit ? StartBit-BaseOffset :0;
if (!BitsContainNoUserData(i->getType(), BaseStart,
EndBit-BaseOffset, Context))
return false;
}
}
// Verify that no field has data that overlaps the region of interest. Yes
// this could be sped up a lot by being smarter about queried fields,
// however we're only looking at structs up to 16 bytes, so we don't care
// much.
unsigned idx = 0;
for (RecordDecl::field_iterator i = RD->field_begin(), e = RD->field_end();
i != e; ++i, ++idx) {
unsigned FieldOffset = (unsigned)Layout.getFieldOffset(idx);
// If we found a field after the region we care about, then we're done.
if (FieldOffset >= EndBit) break;
unsigned FieldStart = FieldOffset < StartBit ? StartBit-FieldOffset :0;
if (!BitsContainNoUserData(i->getType(), FieldStart, EndBit-FieldOffset,
Context))
return false;
}
// If nothing in this record overlapped the area of interest, then we're
// clean.
return true;
}
return false;
}
/// ContainsFloatAtOffset - Return true if the specified LLVM IR type has a
/// float member at the specified offset. For example, {int,{float}} has a
/// float at offset 4. It is conservatively correct for this routine to return
/// false.
static bool ContainsFloatAtOffset(llvm::Type *IRType, unsigned IROffset,
const llvm::DataLayout &TD) {
// Base case if we find a float.
if (IROffset == 0 && IRType->isFloatTy())
return true;
// If this is a struct, recurse into the field at the specified offset.
if (llvm::StructType *STy = dyn_cast<llvm::StructType>(IRType)) {
const llvm::StructLayout *SL = TD.getStructLayout(STy);
unsigned Elt = SL->getElementContainingOffset(IROffset);
IROffset -= SL->getElementOffset(Elt);
return ContainsFloatAtOffset(STy->getElementType(Elt), IROffset, TD);
}
// If this is an array, recurse into the field at the specified offset.
if (llvm::ArrayType *ATy = dyn_cast<llvm::ArrayType>(IRType)) {
llvm::Type *EltTy = ATy->getElementType();
unsigned EltSize = TD.getTypeAllocSize(EltTy);
IROffset -= IROffset/EltSize*EltSize;
return ContainsFloatAtOffset(EltTy, IROffset, TD);
}
return false;
}
/// GetSSETypeAtOffset - Return a type that will be passed by the backend in the
/// low 8 bytes of an XMM register, corresponding to the SSE class.
llvm::Type *X86_64ABIInfo::
GetSSETypeAtOffset(llvm::Type *IRType, unsigned IROffset,
QualType SourceTy, unsigned SourceOffset) const {
// The only three choices we have are either double, <2 x float>, or float. We
// pass as float if the last 4 bytes is just padding. This happens for
// structs that contain 3 floats.
if (BitsContainNoUserData(SourceTy, SourceOffset*8+32,
SourceOffset*8+64, getContext()))
return llvm::Type::getFloatTy(getVMContext());
// We want to pass as <2 x float> if the LLVM IR type contains a float at
// offset+0 and offset+4. Walk the LLVM IR type to find out if this is the
// case.
if (ContainsFloatAtOffset(IRType, IROffset, getDataLayout()) &&
ContainsFloatAtOffset(IRType, IROffset+4, getDataLayout()))
return llvm::VectorType::get(llvm::Type::getFloatTy(getVMContext()), 2);
return llvm::Type::getDoubleTy(getVMContext());
}
/// GetINTEGERTypeAtOffset - The ABI specifies that a value should be passed in
/// an 8-byte GPR. This means that we either have a scalar or we are talking
/// about the high or low part of an up-to-16-byte struct. This routine picks
/// the best LLVM IR type to represent this, which may be i64 or may be anything
/// else that the backend will pass in a GPR that works better (e.g. i8, %foo*,
/// etc).
///
/// PrefType is an LLVM IR type that corresponds to (part of) the IR type for
/// the source type. IROffset is an offset in bytes into the LLVM IR type that
/// the 8-byte value references. PrefType may be null.
///
/// SourceTy is the source level type for the entire argument. SourceOffset is
/// an offset into this that we're processing (which is always either 0 or 8).
///
llvm::Type *X86_64ABIInfo::
GetINTEGERTypeAtOffset(llvm::Type *IRType, unsigned IROffset,
QualType SourceTy, unsigned SourceOffset) const {
// If we're dealing with an un-offset LLVM IR type, then it means that we're
// returning an 8-byte unit starting with it. See if we can safely use it.
if (IROffset == 0) {
// Pointers and int64's always fill the 8-byte unit.
if ((isa<llvm::PointerType>(IRType) && Has64BitPointers) ||
IRType->isIntegerTy(64))
return IRType;
// If we have a 1/2/4-byte integer, we can use it only if the rest of the
// goodness in the source type is just tail padding. This is allowed to
// kick in for struct {double,int} on the int, but not on
// struct{double,int,int} because we wouldn't return the second int. We
// have to do this analysis on the source type because we can't depend on
// unions being lowered a specific way etc.
if (IRType->isIntegerTy(8) || IRType->isIntegerTy(16) ||
IRType->isIntegerTy(32) ||
(isa<llvm::PointerType>(IRType) && !Has64BitPointers)) {
unsigned BitWidth = isa<llvm::PointerType>(IRType) ? 32 :
cast<llvm::IntegerType>(IRType)->getBitWidth();
if (BitsContainNoUserData(SourceTy, SourceOffset*8+BitWidth,
SourceOffset*8+64, getContext()))
return IRType;
}
}
if (llvm::StructType *STy = dyn_cast<llvm::StructType>(IRType)) {
// If this is a struct, recurse into the field at the specified offset.
const llvm::StructLayout *SL = getDataLayout().getStructLayout(STy);
if (IROffset < SL->getSizeInBytes()) {
unsigned FieldIdx = SL->getElementContainingOffset(IROffset);
IROffset -= SL->getElementOffset(FieldIdx);
return GetINTEGERTypeAtOffset(STy->getElementType(FieldIdx), IROffset,
SourceTy, SourceOffset);
}
}
if (llvm::ArrayType *ATy = dyn_cast<llvm::ArrayType>(IRType)) {
llvm::Type *EltTy = ATy->getElementType();
unsigned EltSize = getDataLayout().getTypeAllocSize(EltTy);
unsigned EltOffset = IROffset/EltSize*EltSize;
return GetINTEGERTypeAtOffset(EltTy, IROffset-EltOffset, SourceTy,
SourceOffset);
}
// Okay, we don't have any better idea of what to pass, so we pass this in an
// integer register that isn't too big to fit the rest of the struct.
unsigned TySizeInBytes =
(unsigned)getContext().getTypeSizeInChars(SourceTy).getQuantity();
assert(TySizeInBytes != SourceOffset && "Empty field?");
// It is always safe to classify this as an integer type up to i64 that
// isn't larger than the structure.
return llvm::IntegerType::get(getVMContext(),
std::min(TySizeInBytes-SourceOffset, 8U)*8);
}
/// GetX86_64ByValArgumentPair - Given a high and low type that can ideally
/// be used as elements of a two register pair to pass or return, return a
/// first class aggregate to represent them. For example, if the low part of
/// a by-value argument should be passed as i32* and the high part as float,
/// return {i32*, float}.
static llvm::Type *
GetX86_64ByValArgumentPair(llvm::Type *Lo, llvm::Type *Hi,
const llvm::DataLayout &TD) {
// In order to correctly satisfy the ABI, we need to the high part to start
// at offset 8. If the high and low parts we inferred are both 4-byte types
// (e.g. i32 and i32) then the resultant struct type ({i32,i32}) won't have
// the second element at offset 8. Check for this:
unsigned LoSize = (unsigned)TD.getTypeAllocSize(Lo);
unsigned HiAlign = TD.getABITypeAlignment(Hi);
unsigned HiStart = llvm::DataLayout::RoundUpAlignment(LoSize, HiAlign);
assert(HiStart != 0 && HiStart <= 8 && "Invalid x86-64 argument pair!");
// To handle this, we have to increase the size of the low part so that the
// second element will start at an 8 byte offset. We can't increase the size
// of the second element because it might make us access off the end of the
// struct.
if (HiStart != 8) {
// There are only two sorts of types the ABI generation code can produce for
// the low part of a pair that aren't 8 bytes in size: float or i8/i16/i32.
// Promote these to a larger type.
if (Lo->isFloatTy())
Lo = llvm::Type::getDoubleTy(Lo->getContext());
else {
assert(Lo->isIntegerTy() && "Invalid/unknown lo type");
Lo = llvm::Type::getInt64Ty(Lo->getContext());
}
}
llvm::StructType *Result = llvm::StructType::get(Lo, Hi, NULL);
// Verify that the second element is at an 8-byte offset.
assert(TD.getStructLayout(Result)->getElementOffset(1) == 8 &&
"Invalid x86-64 argument pair!");
return Result;
}
ABIArgInfo X86_64ABIInfo::
classifyReturnType(QualType RetTy) const {
// AMD64-ABI 3.2.3p4: Rule 1. Classify the return type with the
// classification algorithm.
X86_64ABIInfo::Class Lo, Hi;
classify(RetTy, 0, Lo, Hi);
// Check some invariants.
assert((Hi != Memory || Lo == Memory) && "Invalid memory classification.");
assert((Hi != SSEUp || Lo == SSE) && "Invalid SSEUp classification.");
llvm::Type *ResType = 0;
switch (Lo) {
case NoClass:
if (Hi == NoClass)
return ABIArgInfo::getIgnore();
// If the low part is just padding, it takes no register, leave ResType
// null.
assert((Hi == SSE || Hi == Integer || Hi == X87Up) &&
"Unknown missing lo part");
break;
case SSEUp:
case X87Up:
llvm_unreachable("Invalid classification for lo word.");
// AMD64-ABI 3.2.3p4: Rule 2. Types of class memory are returned via
// hidden argument.
case Memory:
return getIndirectReturnResult(RetTy);
// AMD64-ABI 3.2.3p4: Rule 3. If the class is INTEGER, the next
// available register of the sequence %rax, %rdx is used.
case Integer:
ResType = GetINTEGERTypeAtOffset(CGT.ConvertType(RetTy), 0, RetTy, 0);
// If we have a sign or zero extended integer, make sure to return Extend
// so that the parameter gets the right LLVM IR attributes.
if (Hi == NoClass && isa<llvm::IntegerType>(ResType)) {
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = RetTy->getAs<EnumType>())
RetTy = EnumTy->getDecl()->getIntegerType();
if (RetTy->isIntegralOrEnumerationType() &&
RetTy->isPromotableIntegerType())
return ABIArgInfo::getExtend();
}
break;
// AMD64-ABI 3.2.3p4: Rule 4. If the class is SSE, the next
// available SSE register of the sequence %xmm0, %xmm1 is used.
case SSE:
ResType = GetSSETypeAtOffset(CGT.ConvertType(RetTy), 0, RetTy, 0);
break;
// AMD64-ABI 3.2.3p4: Rule 6. If the class is X87, the value is
// returned on the X87 stack in %st0 as 80-bit x87 number.
case X87:
ResType = llvm::Type::getX86_FP80Ty(getVMContext());
break;
// AMD64-ABI 3.2.3p4: Rule 8. If the class is COMPLEX_X87, the real
// part of the value is returned in %st0 and the imaginary part in
// %st1.
case ComplexX87:
assert(Hi == ComplexX87 && "Unexpected ComplexX87 classification.");
ResType = llvm::StructType::get(llvm::Type::getX86_FP80Ty(getVMContext()),
llvm::Type::getX86_FP80Ty(getVMContext()),
NULL);
break;
}
llvm::Type *HighPart = 0;
switch (Hi) {
// Memory was handled previously and X87 should
// never occur as a hi class.
case Memory:
case X87:
llvm_unreachable("Invalid classification for hi word.");
case ComplexX87: // Previously handled.
case NoClass:
break;
case Integer:
HighPart = GetINTEGERTypeAtOffset(CGT.ConvertType(RetTy), 8, RetTy, 8);
if (Lo == NoClass) // Return HighPart at offset 8 in memory.
return ABIArgInfo::getDirect(HighPart, 8);
break;
case SSE:
HighPart = GetSSETypeAtOffset(CGT.ConvertType(RetTy), 8, RetTy, 8);
if (Lo == NoClass) // Return HighPart at offset 8 in memory.
return ABIArgInfo::getDirect(HighPart, 8);
break;
// AMD64-ABI 3.2.3p4: Rule 5. If the class is SSEUP, the eightbyte
// is passed in the next available eightbyte chunk if the last used
// vector register.
//
// SSEUP should always be preceded by SSE, just widen.
case SSEUp:
assert(Lo == SSE && "Unexpected SSEUp classification.");
ResType = GetByteVectorType(RetTy);
break;
// AMD64-ABI 3.2.3p4: Rule 7. If the class is X87UP, the value is
// returned together with the previous X87 value in %st0.
case X87Up:
// If X87Up is preceded by X87, we don't need to do
// anything. However, in some cases with unions it may not be
// preceded by X87. In such situations we follow gcc and pass the
// extra bits in an SSE reg.
if (Lo != X87) {
HighPart = GetSSETypeAtOffset(CGT.ConvertType(RetTy), 8, RetTy, 8);
if (Lo == NoClass) // Return HighPart at offset 8 in memory.
return ABIArgInfo::getDirect(HighPart, 8);
}
break;
}
// If a high part was specified, merge it together with the low part. It is
// known to pass in the high eightbyte of the result. We do this by forming a
// first class struct aggregate with the high and low part: {low, high}
if (HighPart)
ResType = GetX86_64ByValArgumentPair(ResType, HighPart, getDataLayout());
return ABIArgInfo::getDirect(ResType);
}
ABIArgInfo X86_64ABIInfo::classifyArgumentType(
QualType Ty, unsigned freeIntRegs, unsigned &neededInt, unsigned &neededSSE)
const
{
X86_64ABIInfo::Class Lo, Hi;
classify(Ty, 0, Lo, Hi);
// Check some invariants.
// FIXME: Enforce these by construction.
assert((Hi != Memory || Lo == Memory) && "Invalid memory classification.");
assert((Hi != SSEUp || Lo == SSE) && "Invalid SSEUp classification.");
neededInt = 0;
neededSSE = 0;
llvm::Type *ResType = 0;
switch (Lo) {
case NoClass:
if (Hi == NoClass)
return ABIArgInfo::getIgnore();
// If the low part is just padding, it takes no register, leave ResType
// null.
assert((Hi == SSE || Hi == Integer || Hi == X87Up) &&
"Unknown missing lo part");
break;
// AMD64-ABI 3.2.3p3: Rule 1. If the class is MEMORY, pass the argument
// on the stack.
case Memory:
// AMD64-ABI 3.2.3p3: Rule 5. If the class is X87, X87UP or
// COMPLEX_X87, it is passed in memory.
case X87:
case ComplexX87:
if (getRecordArgABI(Ty, CGT) == CGCXXABI::RAA_Indirect)
++neededInt;
return getIndirectResult(Ty, freeIntRegs);
case SSEUp:
case X87Up:
llvm_unreachable("Invalid classification for lo word.");
// AMD64-ABI 3.2.3p3: Rule 2. If the class is INTEGER, the next
// available register of the sequence %rdi, %rsi, %rdx, %rcx, %r8
// and %r9 is used.
case Integer:
++neededInt;
// Pick an 8-byte type based on the preferred type.
ResType = GetINTEGERTypeAtOffset(CGT.ConvertType(Ty), 0, Ty, 0);
// If we have a sign or zero extended integer, make sure to return Extend
// so that the parameter gets the right LLVM IR attributes.
if (Hi == NoClass && isa<llvm::IntegerType>(ResType)) {
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
if (Ty->isIntegralOrEnumerationType() &&
Ty->isPromotableIntegerType())
return ABIArgInfo::getExtend();
}
break;
// AMD64-ABI 3.2.3p3: Rule 3. If the class is SSE, the next
// available SSE register is used, the registers are taken in the
// order from %xmm0 to %xmm7.
case SSE: {
llvm::Type *IRType = CGT.ConvertType(Ty);
ResType = GetSSETypeAtOffset(IRType, 0, Ty, 0);
++neededSSE;
break;
}
}
llvm::Type *HighPart = 0;
switch (Hi) {
// Memory was handled previously, ComplexX87 and X87 should
// never occur as hi classes, and X87Up must be preceded by X87,
// which is passed in memory.
case Memory:
case X87:
case ComplexX87:
llvm_unreachable("Invalid classification for hi word.");
case NoClass: break;
case Integer:
++neededInt;
// Pick an 8-byte type based on the preferred type.
HighPart = GetINTEGERTypeAtOffset(CGT.ConvertType(Ty), 8, Ty, 8);
if (Lo == NoClass) // Pass HighPart at offset 8 in memory.
return ABIArgInfo::getDirect(HighPart, 8);
break;
// X87Up generally doesn't occur here (long double is passed in
// memory), except in situations involving unions.
case X87Up:
case SSE:
HighPart = GetSSETypeAtOffset(CGT.ConvertType(Ty), 8, Ty, 8);
if (Lo == NoClass) // Pass HighPart at offset 8 in memory.
return ABIArgInfo::getDirect(HighPart, 8);
++neededSSE;
break;
// AMD64-ABI 3.2.3p3: Rule 4. If the class is SSEUP, the
// eightbyte is passed in the upper half of the last used SSE
// register. This only happens when 128-bit vectors are passed.
case SSEUp:
assert(Lo == SSE && "Unexpected SSEUp classification");
ResType = GetByteVectorType(Ty);
break;
}
// If a high part was specified, merge it together with the low part. It is
// known to pass in the high eightbyte of the result. We do this by forming a
// first class struct aggregate with the high and low part: {low, high}
if (HighPart)
ResType = GetX86_64ByValArgumentPair(ResType, HighPart, getDataLayout());
return ABIArgInfo::getDirect(ResType);
}
void X86_64ABIInfo::computeInfo(CGFunctionInfo &FI) const {
FI.getReturnInfo() = classifyReturnType(FI.getReturnType());
// Keep track of the number of assigned registers.
unsigned freeIntRegs = 6, freeSSERegs = 8;
// If the return value is indirect, then the hidden argument is consuming one
// integer register.
if (FI.getReturnInfo().isIndirect())
--freeIntRegs;
// AMD64-ABI 3.2.3p3: Once arguments are classified, the registers
// get assigned (in left-to-right order) for passing as follows...
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it) {
unsigned neededInt, neededSSE;
it->info = classifyArgumentType(it->type, freeIntRegs, neededInt,
neededSSE);
// AMD64-ABI 3.2.3p3: If there are no registers available for any
// eightbyte of an argument, the whole argument is passed on the
// stack. If registers have already been assigned for some
// eightbytes of such an argument, the assignments get reverted.
if (freeIntRegs >= neededInt && freeSSERegs >= neededSSE) {
freeIntRegs -= neededInt;
freeSSERegs -= neededSSE;
} else {
it->info = getIndirectResult(it->type, freeIntRegs);
}
}
}
static llvm::Value *EmitVAArgFromMemory(llvm::Value *VAListAddr,
QualType Ty,
CodeGenFunction &CGF) {
llvm::Value *overflow_arg_area_p =
CGF.Builder.CreateStructGEP(VAListAddr, 2, "overflow_arg_area_p");
llvm::Value *overflow_arg_area =
CGF.Builder.CreateLoad(overflow_arg_area_p, "overflow_arg_area");
// AMD64-ABI 3.5.7p5: Step 7. Align l->overflow_arg_area upwards to a 16
// byte boundary if alignment needed by type exceeds 8 byte boundary.
// It isn't stated explicitly in the standard, but in practice we use
// alignment greater than 16 where necessary.
uint64_t Align = CGF.getContext().getTypeAlign(Ty) / 8;
if (Align > 8) {
// overflow_arg_area = (overflow_arg_area + align - 1) & -align;
llvm::Value *Offset =
llvm::ConstantInt::get(CGF.Int64Ty, Align - 1);
overflow_arg_area = CGF.Builder.CreateGEP(overflow_arg_area, Offset);
llvm::Value *AsInt = CGF.Builder.CreatePtrToInt(overflow_arg_area,
CGF.Int64Ty);
llvm::Value *Mask = llvm::ConstantInt::get(CGF.Int64Ty, -(uint64_t)Align);
overflow_arg_area =
CGF.Builder.CreateIntToPtr(CGF.Builder.CreateAnd(AsInt, Mask),
overflow_arg_area->getType(),
"overflow_arg_area.align");
}
// AMD64-ABI 3.5.7p5: Step 8. Fetch type from l->overflow_arg_area.
llvm::Type *LTy = CGF.ConvertTypeForMem(Ty);
llvm::Value *Res =
CGF.Builder.CreateBitCast(overflow_arg_area,
llvm::PointerType::getUnqual(LTy));
// AMD64-ABI 3.5.7p5: Step 9. Set l->overflow_arg_area to:
// l->overflow_arg_area + sizeof(type).
// AMD64-ABI 3.5.7p5: Step 10. Align l->overflow_arg_area upwards to
// an 8 byte boundary.
uint64_t SizeInBytes = (CGF.getContext().getTypeSize(Ty) + 7) / 8;
llvm::Value *Offset =
llvm::ConstantInt::get(CGF.Int32Ty, (SizeInBytes + 7) & ~7);
overflow_arg_area = CGF.Builder.CreateGEP(overflow_arg_area, Offset,
"overflow_arg_area.next");
CGF.Builder.CreateStore(overflow_arg_area, overflow_arg_area_p);
// AMD64-ABI 3.5.7p5: Step 11. Return the fetched type.
return Res;
}
llvm::Value *X86_64ABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
// Assume that va_list type is correct; should be pointer to LLVM type:
// struct {
// i32 gp_offset;
// i32 fp_offset;
// i8* overflow_arg_area;
// i8* reg_save_area;
// };
unsigned neededInt, neededSSE;
Ty = CGF.getContext().getCanonicalType(Ty);
ABIArgInfo AI = classifyArgumentType(Ty, 0, neededInt, neededSSE);
// AMD64-ABI 3.5.7p5: Step 1. Determine whether type may be passed
// in the registers. If not go to step 7.
if (!neededInt && !neededSSE)
return EmitVAArgFromMemory(VAListAddr, Ty, CGF);
// AMD64-ABI 3.5.7p5: Step 2. Compute num_gp to hold the number of
// general purpose registers needed to pass type and num_fp to hold
// the number of floating point registers needed.
// AMD64-ABI 3.5.7p5: Step 3. Verify whether arguments fit into
// registers. In the case: l->gp_offset > 48 - num_gp * 8 or
// l->fp_offset > 304 - num_fp * 16 go to step 7.
//
// NOTE: 304 is a typo, there are (6 * 8 + 8 * 16) = 176 bytes of
// register save space).
llvm::Value *InRegs = 0;
llvm::Value *gp_offset_p = 0, *gp_offset = 0;
llvm::Value *fp_offset_p = 0, *fp_offset = 0;
if (neededInt) {
gp_offset_p = CGF.Builder.CreateStructGEP(VAListAddr, 0, "gp_offset_p");
gp_offset = CGF.Builder.CreateLoad(gp_offset_p, "gp_offset");
InRegs = llvm::ConstantInt::get(CGF.Int32Ty, 48 - neededInt * 8);
InRegs = CGF.Builder.CreateICmpULE(gp_offset, InRegs, "fits_in_gp");
}
if (neededSSE) {
fp_offset_p = CGF.Builder.CreateStructGEP(VAListAddr, 1, "fp_offset_p");
fp_offset = CGF.Builder.CreateLoad(fp_offset_p, "fp_offset");
llvm::Value *FitsInFP =
llvm::ConstantInt::get(CGF.Int32Ty, 176 - neededSSE * 16);
FitsInFP = CGF.Builder.CreateICmpULE(fp_offset, FitsInFP, "fits_in_fp");
InRegs = InRegs ? CGF.Builder.CreateAnd(InRegs, FitsInFP) : FitsInFP;
}
llvm::BasicBlock *InRegBlock = CGF.createBasicBlock("vaarg.in_reg");
llvm::BasicBlock *InMemBlock = CGF.createBasicBlock("vaarg.in_mem");
llvm::BasicBlock *ContBlock = CGF.createBasicBlock("vaarg.end");
CGF.Builder.CreateCondBr(InRegs, InRegBlock, InMemBlock);
// Emit code to load the value if it was passed in registers.
CGF.EmitBlock(InRegBlock);
// AMD64-ABI 3.5.7p5: Step 4. Fetch type from l->reg_save_area with
// an offset of l->gp_offset and/or l->fp_offset. This may require
// copying to a temporary location in case the parameter is passed
// in different register classes or requires an alignment greater
// than 8 for general purpose registers and 16 for XMM registers.
//
// FIXME: This really results in shameful code when we end up needing to
// collect arguments from different places; often what should result in a
// simple assembling of a structure from scattered addresses has many more
// loads than necessary. Can we clean this up?
llvm::Type *LTy = CGF.ConvertTypeForMem(Ty);
llvm::Value *RegAddr =
CGF.Builder.CreateLoad(CGF.Builder.CreateStructGEP(VAListAddr, 3),
"reg_save_area");
if (neededInt && neededSSE) {
// FIXME: Cleanup.
assert(AI.isDirect() && "Unexpected ABI info for mixed regs");
llvm::StructType *ST = cast<llvm::StructType>(AI.getCoerceToType());
llvm::Value *Tmp = CGF.CreateTempAlloca(ST);
assert(ST->getNumElements() == 2 && "Unexpected ABI info for mixed regs");
llvm::Type *TyLo = ST->getElementType(0);
llvm::Type *TyHi = ST->getElementType(1);
assert((TyLo->isFPOrFPVectorTy() ^ TyHi->isFPOrFPVectorTy()) &&
"Unexpected ABI info for mixed regs");
llvm::Type *PTyLo = llvm::PointerType::getUnqual(TyLo);
llvm::Type *PTyHi = llvm::PointerType::getUnqual(TyHi);
llvm::Value *GPAddr = CGF.Builder.CreateGEP(RegAddr, gp_offset);
llvm::Value *FPAddr = CGF.Builder.CreateGEP(RegAddr, fp_offset);
llvm::Value *RegLoAddr = TyLo->isFloatingPointTy() ? FPAddr : GPAddr;
llvm::Value *RegHiAddr = TyLo->isFloatingPointTy() ? GPAddr : FPAddr;
llvm::Value *V =
CGF.Builder.CreateLoad(CGF.Builder.CreateBitCast(RegLoAddr, PTyLo));
CGF.Builder.CreateStore(V, CGF.Builder.CreateStructGEP(Tmp, 0));
V = CGF.Builder.CreateLoad(CGF.Builder.CreateBitCast(RegHiAddr, PTyHi));
CGF.Builder.CreateStore(V, CGF.Builder.CreateStructGEP(Tmp, 1));
RegAddr = CGF.Builder.CreateBitCast(Tmp,
llvm::PointerType::getUnqual(LTy));
} else if (neededInt) {
RegAddr = CGF.Builder.CreateGEP(RegAddr, gp_offset);
RegAddr = CGF.Builder.CreateBitCast(RegAddr,
llvm::PointerType::getUnqual(LTy));
} else if (neededSSE == 1) {
RegAddr = CGF.Builder.CreateGEP(RegAddr, fp_offset);
RegAddr = CGF.Builder.CreateBitCast(RegAddr,
llvm::PointerType::getUnqual(LTy));
} else {
assert(neededSSE == 2 && "Invalid number of needed registers!");
// SSE registers are spaced 16 bytes apart in the register save
// area, we need to collect the two eightbytes together.
llvm::Value *RegAddrLo = CGF.Builder.CreateGEP(RegAddr, fp_offset);
llvm::Value *RegAddrHi = CGF.Builder.CreateConstGEP1_32(RegAddrLo, 16);
llvm::Type *DoubleTy = CGF.DoubleTy;
llvm::Type *DblPtrTy =
llvm::PointerType::getUnqual(DoubleTy);
llvm::StructType *ST = llvm::StructType::get(DoubleTy,
DoubleTy, NULL);
llvm::Value *V, *Tmp = CGF.CreateTempAlloca(ST);
V = CGF.Builder.CreateLoad(CGF.Builder.CreateBitCast(RegAddrLo,
DblPtrTy));
CGF.Builder.CreateStore(V, CGF.Builder.CreateStructGEP(Tmp, 0));
V = CGF.Builder.CreateLoad(CGF.Builder.CreateBitCast(RegAddrHi,
DblPtrTy));
CGF.Builder.CreateStore(V, CGF.Builder.CreateStructGEP(Tmp, 1));
RegAddr = CGF.Builder.CreateBitCast(Tmp,
llvm::PointerType::getUnqual(LTy));
}
// AMD64-ABI 3.5.7p5: Step 5. Set:
// l->gp_offset = l->gp_offset + num_gp * 8
// l->fp_offset = l->fp_offset + num_fp * 16.
if (neededInt) {
llvm::Value *Offset = llvm::ConstantInt::get(CGF.Int32Ty, neededInt * 8);
CGF.Builder.CreateStore(CGF.Builder.CreateAdd(gp_offset, Offset),
gp_offset_p);
}
if (neededSSE) {
llvm::Value *Offset = llvm::ConstantInt::get(CGF.Int32Ty, neededSSE * 16);
CGF.Builder.CreateStore(CGF.Builder.CreateAdd(fp_offset, Offset),
fp_offset_p);
}
CGF.EmitBranch(ContBlock);
// Emit code to load the value if it was passed in memory.
CGF.EmitBlock(InMemBlock);
llvm::Value *MemAddr = EmitVAArgFromMemory(VAListAddr, Ty, CGF);
// Return the appropriate result.
CGF.EmitBlock(ContBlock);
llvm::PHINode *ResAddr = CGF.Builder.CreatePHI(RegAddr->getType(), 2,
"vaarg.addr");
ResAddr->addIncoming(RegAddr, InRegBlock);
ResAddr->addIncoming(MemAddr, InMemBlock);
return ResAddr;
}
ABIArgInfo WinX86_64ABIInfo::classify(QualType Ty, bool IsReturnType) const {
if (Ty->isVoidType())
return ABIArgInfo::getIgnore();
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
uint64_t Size = getContext().getTypeSize(Ty);
if (const RecordType *RT = Ty->getAs<RecordType>()) {
if (IsReturnType) {
if (isRecordReturnIndirect(RT, CGT))
return ABIArgInfo::getIndirect(0, false);
} else {
if (CGCXXABI::RecordArgABI RAA = getRecordArgABI(RT, CGT))
return ABIArgInfo::getIndirect(0, RAA == CGCXXABI::RAA_DirectInMemory);
}
if (RT->getDecl()->hasFlexibleArrayMember())
return ABIArgInfo::getIndirect(0, /*ByVal=*/false);
// FIXME: mingw-w64-gcc emits 128-bit struct as i128
if (Size == 128 && getTarget().getTriple().getOS() == llvm::Triple::MinGW32)
return ABIArgInfo::getDirect(llvm::IntegerType::get(getVMContext(),
Size));
// MS x64 ABI requirement: "Any argument that doesn't fit in 8 bytes, or is
// not 1, 2, 4, or 8 bytes, must be passed by reference."
if (Size <= 64 &&
(Size & (Size - 1)) == 0)
return ABIArgInfo::getDirect(llvm::IntegerType::get(getVMContext(),
Size));
return ABIArgInfo::getIndirect(0, /*ByVal=*/false);
}
if (Ty->isPromotableIntegerType())
return ABIArgInfo::getExtend();
return ABIArgInfo::getDirect();
}
void WinX86_64ABIInfo::computeInfo(CGFunctionInfo &FI) const {
QualType RetTy = FI.getReturnType();
FI.getReturnInfo() = classify(RetTy, true);
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it)
it->info = classify(it->type, false);
}
llvm::Value *WinX86_64ABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
llvm::Type *BPP = CGF.Int8PtrPtrTy;
CGBuilderTy &Builder = CGF.Builder;
llvm::Value *VAListAddrAsBPP = Builder.CreateBitCast(VAListAddr, BPP,
"ap");
llvm::Value *Addr = Builder.CreateLoad(VAListAddrAsBPP, "ap.cur");
llvm::Type *PTy =
llvm::PointerType::getUnqual(CGF.ConvertType(Ty));
llvm::Value *AddrTyped = Builder.CreateBitCast(Addr, PTy);
uint64_t Offset =
llvm::RoundUpToAlignment(CGF.getContext().getTypeSize(Ty) / 8, 8);
llvm::Value *NextAddr =
Builder.CreateGEP(Addr, llvm::ConstantInt::get(CGF.Int32Ty, Offset),
"ap.next");
Builder.CreateStore(NextAddr, VAListAddrAsBPP);
return AddrTyped;
}
namespace {
class NaClX86_64ABIInfo : public ABIInfo {
public:
NaClX86_64ABIInfo(CodeGen::CodeGenTypes &CGT, bool HasAVX)
: ABIInfo(CGT), PInfo(CGT), NInfo(CGT, HasAVX) {}
virtual void computeInfo(CGFunctionInfo &FI) const;
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
private:
PNaClABIInfo PInfo; // Used for generating calls with pnaclcall callingconv.
X86_64ABIInfo NInfo; // Used for everything else.
};
class NaClX86_64TargetCodeGenInfo : public TargetCodeGenInfo {
public:
NaClX86_64TargetCodeGenInfo(CodeGen::CodeGenTypes &CGT, bool HasAVX)
: TargetCodeGenInfo(new NaClX86_64ABIInfo(CGT, HasAVX)) {}
};
}
void NaClX86_64ABIInfo::computeInfo(CGFunctionInfo &FI) const {
if (FI.getASTCallingConvention() == CC_PnaclCall)
PInfo.computeInfo(FI);
else
NInfo.computeInfo(FI);
}
llvm::Value *NaClX86_64ABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
// Always use the native convention; calling pnacl-style varargs functions
// is unuspported.
return NInfo.EmitVAArg(VAListAddr, Ty, CGF);
}
// PowerPC-32
namespace {
class PPC32TargetCodeGenInfo : public DefaultTargetCodeGenInfo {
public:
PPC32TargetCodeGenInfo(CodeGenTypes &CGT) : DefaultTargetCodeGenInfo(CGT) {}
int getDwarfEHStackPointer(CodeGen::CodeGenModule &M) const {
// This is recovered from gcc output.
return 1; // r1 is the dedicated stack pointer
}
bool initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const;
};
}
bool
PPC32TargetCodeGenInfo::initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const {
// This is calculated from the LLVM and GCC tables and verified
// against gcc output. AFAIK all ABIs use the same encoding.
CodeGen::CGBuilderTy &Builder = CGF.Builder;
llvm::IntegerType *i8 = CGF.Int8Ty;
llvm::Value *Four8 = llvm::ConstantInt::get(i8, 4);
llvm::Value *Eight8 = llvm::ConstantInt::get(i8, 8);
llvm::Value *Sixteen8 = llvm::ConstantInt::get(i8, 16);
// 0-31: r0-31, the 4-byte general-purpose registers
AssignToArrayRange(Builder, Address, Four8, 0, 31);
// 32-63: fp0-31, the 8-byte floating-point registers
AssignToArrayRange(Builder, Address, Eight8, 32, 63);
// 64-76 are various 4-byte special-purpose registers:
// 64: mq
// 65: lr
// 66: ctr
// 67: ap
// 68-75 cr0-7
// 76: xer
AssignToArrayRange(Builder, Address, Four8, 64, 76);
// 77-108: v0-31, the 16-byte vector registers
AssignToArrayRange(Builder, Address, Sixteen8, 77, 108);
// 109: vrsave
// 110: vscr
// 111: spe_acc
// 112: spefscr
// 113: sfp
AssignToArrayRange(Builder, Address, Four8, 109, 113);
return false;
}
// PowerPC-64
namespace {
/// PPC64_SVR4_ABIInfo - The 64-bit PowerPC ELF (SVR4) ABI information.
class PPC64_SVR4_ABIInfo : public DefaultABIInfo {
public:
PPC64_SVR4_ABIInfo(CodeGen::CodeGenTypes &CGT) : DefaultABIInfo(CGT) {}
bool isPromotableTypeForABI(QualType Ty) const;
ABIArgInfo classifyReturnType(QualType RetTy) const;
ABIArgInfo classifyArgumentType(QualType Ty) const;
// TODO: We can add more logic to computeInfo to improve performance.
// Example: For aggregate arguments that fit in a register, we could
// use getDirectInReg (as is done below for structs containing a single
// floating-point value) to avoid pushing them to memory on function
// entry. This would require changing the logic in PPCISelLowering
// when lowering the parameters in the caller and args in the callee.
virtual void computeInfo(CGFunctionInfo &FI) const {
FI.getReturnInfo() = classifyReturnType(FI.getReturnType());
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it) {
// We rely on the default argument classification for the most part.
// One exception: An aggregate containing a single floating-point
// item must be passed in a register if one is available.
const Type *T = isSingleElementStruct(it->type, getContext());
if (T) {
const BuiltinType *BT = T->getAs<BuiltinType>();
if (BT && BT->isFloatingPoint()) {
QualType QT(T, 0);
it->info = ABIArgInfo::getDirectInReg(CGT.ConvertType(QT));
continue;
}
}
it->info = classifyArgumentType(it->type);
}
}
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr,
QualType Ty,
CodeGenFunction &CGF) const;
};
class PPC64_SVR4_TargetCodeGenInfo : public TargetCodeGenInfo {
public:
PPC64_SVR4_TargetCodeGenInfo(CodeGenTypes &CGT)
: TargetCodeGenInfo(new PPC64_SVR4_ABIInfo(CGT)) {}
int getDwarfEHStackPointer(CodeGen::CodeGenModule &M) const {
// This is recovered from gcc output.
return 1; // r1 is the dedicated stack pointer
}
bool initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const;
};
class PPC64TargetCodeGenInfo : public DefaultTargetCodeGenInfo {
public:
PPC64TargetCodeGenInfo(CodeGenTypes &CGT) : DefaultTargetCodeGenInfo(CGT) {}
int getDwarfEHStackPointer(CodeGen::CodeGenModule &M) const {
// This is recovered from gcc output.
return 1; // r1 is the dedicated stack pointer
}
bool initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const;
};
}
// Return true if the ABI requires Ty to be passed sign- or zero-
// extended to 64 bits.
bool
PPC64_SVR4_ABIInfo::isPromotableTypeForABI(QualType Ty) const {
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
// Promotable integer types are required to be promoted by the ABI.
if (Ty->isPromotableIntegerType())
return true;
// In addition to the usual promotable integer types, we also need to
// extend all 32-bit types, since the ABI requires promotion to 64 bits.
if (const BuiltinType *BT = Ty->getAs<BuiltinType>())
switch (BT->getKind()) {
case BuiltinType::Int:
case BuiltinType::UInt:
return true;
default:
break;
}
return false;
}
ABIArgInfo
PPC64_SVR4_ABIInfo::classifyArgumentType(QualType Ty) const {
if (Ty->isAnyComplexType())
return ABIArgInfo::getDirect();
if (isAggregateTypeForABI(Ty)) {
if (CGCXXABI::RecordArgABI RAA = getRecordArgABI(Ty, CGT))
return ABIArgInfo::getIndirect(0, RAA == CGCXXABI::RAA_DirectInMemory);
return ABIArgInfo::getIndirect(0);
}
return (isPromotableTypeForABI(Ty) ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
ABIArgInfo
PPC64_SVR4_ABIInfo::classifyReturnType(QualType RetTy) const {
if (RetTy->isVoidType())
return ABIArgInfo::getIgnore();
if (RetTy->isAnyComplexType())
return ABIArgInfo::getDirect();
if (isAggregateTypeForABI(RetTy))
return ABIArgInfo::getIndirect(0);
return (isPromotableTypeForABI(RetTy) ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
// Based on ARMABIInfo::EmitVAArg, adjusted for 64-bit machine.
llvm::Value *PPC64_SVR4_ABIInfo::EmitVAArg(llvm::Value *VAListAddr,
QualType Ty,
CodeGenFunction &CGF) const {
llvm::Type *BP = CGF.Int8PtrTy;
llvm::Type *BPP = CGF.Int8PtrPtrTy;
CGBuilderTy &Builder = CGF.Builder;
llvm::Value *VAListAddrAsBPP = Builder.CreateBitCast(VAListAddr, BPP, "ap");
llvm::Value *Addr = Builder.CreateLoad(VAListAddrAsBPP, "ap.cur");
// Update the va_list pointer. The pointer should be bumped by the
// size of the object. We can trust getTypeSize() except for a complex
// type whose base type is smaller than a doubleword. For these, the
// size of the object is 16 bytes; see below for further explanation.
unsigned SizeInBytes = CGF.getContext().getTypeSize(Ty) / 8;
QualType BaseTy;
unsigned CplxBaseSize = 0;
if (const ComplexType *CTy = Ty->getAs<ComplexType>()) {
BaseTy = CTy->getElementType();
CplxBaseSize = CGF.getContext().getTypeSize(BaseTy) / 8;
if (CplxBaseSize < 8)
SizeInBytes = 16;
}
unsigned Offset = llvm::RoundUpToAlignment(SizeInBytes, 8);
llvm::Value *NextAddr =
Builder.CreateGEP(Addr, llvm::ConstantInt::get(CGF.Int64Ty, Offset),
"ap.next");
Builder.CreateStore(NextAddr, VAListAddrAsBPP);
// If we have a complex type and the base type is smaller than 8 bytes,
// the ABI calls for the real and imaginary parts to be right-adjusted
// in separate doublewords. However, Clang expects us to produce a
// pointer to a structure with the two parts packed tightly. So generate
// loads of the real and imaginary parts relative to the va_list pointer,
// and store them to a temporary structure.
if (CplxBaseSize && CplxBaseSize < 8) {
llvm::Value *RealAddr = Builder.CreatePtrToInt(Addr, CGF.Int64Ty);
llvm::Value *ImagAddr = RealAddr;
RealAddr = Builder.CreateAdd(RealAddr, Builder.getInt64(8 - CplxBaseSize));
ImagAddr = Builder.CreateAdd(ImagAddr, Builder.getInt64(16 - CplxBaseSize));
llvm::Type *PBaseTy = llvm::PointerType::getUnqual(CGF.ConvertType(BaseTy));
RealAddr = Builder.CreateIntToPtr(RealAddr, PBaseTy);
ImagAddr = Builder.CreateIntToPtr(ImagAddr, PBaseTy);
llvm::Value *Real = Builder.CreateLoad(RealAddr, false, ".vareal");
llvm::Value *Imag = Builder.CreateLoad(ImagAddr, false, ".vaimag");
llvm::Value *Ptr = CGF.CreateTempAlloca(CGT.ConvertTypeForMem(Ty),
"vacplx");
llvm::Value *RealPtr = Builder.CreateStructGEP(Ptr, 0, ".real");
llvm::Value *ImagPtr = Builder.CreateStructGEP(Ptr, 1, ".imag");
Builder.CreateStore(Real, RealPtr, false);
Builder.CreateStore(Imag, ImagPtr, false);
return Ptr;
}
// If the argument is smaller than 8 bytes, it is right-adjusted in
// its doubleword slot. Adjust the pointer to pick it up from the
// correct offset.
if (SizeInBytes < 8) {
llvm::Value *AddrAsInt = Builder.CreatePtrToInt(Addr, CGF.Int64Ty);
AddrAsInt = Builder.CreateAdd(AddrAsInt, Builder.getInt64(8 - SizeInBytes));
Addr = Builder.CreateIntToPtr(AddrAsInt, BP);
}
llvm::Type *PTy = llvm::PointerType::getUnqual(CGF.ConvertType(Ty));
return Builder.CreateBitCast(Addr, PTy);
}
static bool
PPC64_initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) {
// This is calculated from the LLVM and GCC tables and verified
// against gcc output. AFAIK all ABIs use the same encoding.
CodeGen::CGBuilderTy &Builder = CGF.Builder;
llvm::IntegerType *i8 = CGF.Int8Ty;
llvm::Value *Four8 = llvm::ConstantInt::get(i8, 4);
llvm::Value *Eight8 = llvm::ConstantInt::get(i8, 8);
llvm::Value *Sixteen8 = llvm::ConstantInt::get(i8, 16);
// 0-31: r0-31, the 8-byte general-purpose registers
AssignToArrayRange(Builder, Address, Eight8, 0, 31);
// 32-63: fp0-31, the 8-byte floating-point registers
AssignToArrayRange(Builder, Address, Eight8, 32, 63);
// 64-76 are various 4-byte special-purpose registers:
// 64: mq
// 65: lr
// 66: ctr
// 67: ap
// 68-75 cr0-7
// 76: xer
AssignToArrayRange(Builder, Address, Four8, 64, 76);
// 77-108: v0-31, the 16-byte vector registers
AssignToArrayRange(Builder, Address, Sixteen8, 77, 108);
// 109: vrsave
// 110: vscr
// 111: spe_acc
// 112: spefscr
// 113: sfp
AssignToArrayRange(Builder, Address, Four8, 109, 113);
return false;
}
bool
PPC64_SVR4_TargetCodeGenInfo::initDwarfEHRegSizeTable(
CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const {
return PPC64_initDwarfEHRegSizeTable(CGF, Address);
}
bool
PPC64TargetCodeGenInfo::initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const {
return PPC64_initDwarfEHRegSizeTable(CGF, Address);
}
//===----------------------------------------------------------------------===//
// ARM ABI Implementation
//===----------------------------------------------------------------------===//
namespace {
class ARMABIInfo : public ABIInfo {
public:
enum ABIKind {
APCS = 0,
AAPCS = 1,
AAPCS_VFP
};
private:
ABIKind Kind;
public:
ARMABIInfo(CodeGenTypes &CGT, ABIKind _Kind) : ABIInfo(CGT), Kind(_Kind) {
setRuntimeCC();
}
bool isEABI() const {
StringRef Env = getTarget().getTriple().getEnvironmentName();
return (Env == "gnueabi" || Env == "eabi" ||
Env == "android" || Env == "androideabi");
}
private:
ABIKind getABIKind() const { return Kind; }
ABIArgInfo classifyReturnType(QualType RetTy) const;
ABIArgInfo classifyArgumentType(QualType RetTy, int *VFPRegs,
unsigned &AllocatedVFP,
bool &IsHA) const;
bool isIllegalVectorType(QualType Ty) const;
virtual void computeInfo(CGFunctionInfo &FI) const;
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
llvm::CallingConv::ID getLLVMDefaultCC() const;
llvm::CallingConv::ID getABIDefaultCC() const;
void setRuntimeCC();
};
class ARMTargetCodeGenInfo : public TargetCodeGenInfo {
public:
ARMTargetCodeGenInfo(CodeGenTypes &CGT, ARMABIInfo::ABIKind K)
:TargetCodeGenInfo(new ARMABIInfo(CGT, K)) {}
const ARMABIInfo &getABIInfo() const {
return static_cast<const ARMABIInfo&>(TargetCodeGenInfo::getABIInfo());
}
int getDwarfEHStackPointer(CodeGen::CodeGenModule &M) const {
return 13;
}
StringRef getARCRetainAutoreleasedReturnValueMarker() const {
return "mov\tr7, r7\t\t@ marker for objc_retainAutoreleaseReturnValue";
}
bool initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const {
llvm::Value *Four8 = llvm::ConstantInt::get(CGF.Int8Ty, 4);
// 0-15 are the 16 integer registers.
AssignToArrayRange(CGF.Builder, Address, Four8, 0, 15);
return false;
}
unsigned getSizeOfUnwindException() const {
if (getABIInfo().isEABI()) return 88;
return TargetCodeGenInfo::getSizeOfUnwindException();
}
};
}
void ARMABIInfo::computeInfo(CGFunctionInfo &FI) const {
// To correctly handle Homogeneous Aggregate, we need to keep track of the
// VFP registers allocated so far.
// C.1.vfp If the argument is a VFP CPRC and there are sufficient consecutive
// VFP registers of the appropriate type unallocated then the argument is
// allocated to the lowest-numbered sequence of such registers.
// C.2.vfp If the argument is a VFP CPRC then any VFP registers that are
// unallocated are marked as unavailable.
unsigned AllocatedVFP = 0;
int VFPRegs[16] = { 0 };
FI.getReturnInfo() = classifyReturnType(FI.getReturnType());
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it) {
unsigned PreAllocation = AllocatedVFP;
bool IsHA = false;
// 6.1.2.3 There is one VFP co-processor register class using registers
// s0-s15 (d0-d7) for passing arguments.
const unsigned NumVFPs = 16;
it->info = classifyArgumentType(it->type, VFPRegs, AllocatedVFP, IsHA);
// If we do not have enough VFP registers for the HA, any VFP registers
// that are unallocated are marked as unavailable. To achieve this, we add
// padding of (NumVFPs - PreAllocation) floats.
if (IsHA && AllocatedVFP > NumVFPs && PreAllocation < NumVFPs) {
llvm::Type *PaddingTy = llvm::ArrayType::get(
llvm::Type::getFloatTy(getVMContext()), NumVFPs - PreAllocation);
it->info = ABIArgInfo::getExpandWithPadding(false, PaddingTy);
}
}
// Always honor user-specified calling convention.
if (FI.getCallingConvention() != llvm::CallingConv::C)
return;
llvm::CallingConv::ID cc = getRuntimeCC();
if (cc != llvm::CallingConv::C)
FI.setEffectiveCallingConvention(cc);
}
/// Return the default calling convention that LLVM will use.
llvm::CallingConv::ID ARMABIInfo::getLLVMDefaultCC() const {
// The default calling convention that LLVM will infer.
if (getTarget().getTriple().getEnvironmentName()=="gnueabihf")
return llvm::CallingConv::ARM_AAPCS_VFP;
else if (isEABI())
return llvm::CallingConv::ARM_AAPCS;
else
return llvm::CallingConv::ARM_APCS;
}
/// Return the calling convention that our ABI would like us to use
/// as the C calling convention.
llvm::CallingConv::ID ARMABIInfo::getABIDefaultCC() const {
switch (getABIKind()) {
case APCS: return llvm::CallingConv::ARM_APCS;
case AAPCS: return llvm::CallingConv::ARM_AAPCS;
case AAPCS_VFP: return llvm::CallingConv::ARM_AAPCS_VFP;
}
llvm_unreachable("bad ABI kind");
}
void ARMABIInfo::setRuntimeCC() {
assert(getRuntimeCC() == llvm::CallingConv::C);
// Don't muddy up the IR with a ton of explicit annotations if
// they'd just match what LLVM will infer from the triple.
llvm::CallingConv::ID abiCC = getABIDefaultCC();
if (abiCC != getLLVMDefaultCC())
RuntimeCC = abiCC;
}
/// isHomogeneousAggregate - Return true if a type is an AAPCS-VFP homogeneous
/// aggregate. If HAMembers is non-null, the number of base elements
/// contained in the type is returned through it; this is used for the
/// recursive calls that check aggregate component types.
static bool isHomogeneousAggregate(QualType Ty, const Type *&Base,
ASTContext &Context,
uint64_t *HAMembers = 0) {
uint64_t Members = 0;
if (const ConstantArrayType *AT = Context.getAsConstantArrayType(Ty)) {
if (!isHomogeneousAggregate(AT->getElementType(), Base, Context, &Members))
return false;
Members *= AT->getSize().getZExtValue();
} else if (const RecordType *RT = Ty->getAs<RecordType>()) {
const RecordDecl *RD = RT->getDecl();
if (RD->hasFlexibleArrayMember())
return false;
Members = 0;
for (RecordDecl::field_iterator i = RD->field_begin(), e = RD->field_end();
i != e; ++i) {
const FieldDecl *FD = *i;
uint64_t FldMembers;
if (!isHomogeneousAggregate(FD->getType(), Base, Context, &FldMembers))
return false;
Members = (RD->isUnion() ?
std::max(Members, FldMembers) : Members + FldMembers);
}
} else {
Members = 1;
if (const ComplexType *CT = Ty->getAs<ComplexType>()) {
Members = 2;
Ty = CT->getElementType();
}
// Homogeneous aggregates for AAPCS-VFP must have base types of float,
// double, or 64-bit or 128-bit vectors.
if (const BuiltinType *BT = Ty->getAs<BuiltinType>()) {
if (BT->getKind() != BuiltinType::Float &&
BT->getKind() != BuiltinType::Double &&
BT->getKind() != BuiltinType::LongDouble)
return false;
} else if (const VectorType *VT = Ty->getAs<VectorType>()) {
unsigned VecSize = Context.getTypeSize(VT);
if (VecSize != 64 && VecSize != 128)
return false;
} else {
return false;
}
// The base type must be the same for all members. Vector types of the
// same total size are treated as being equivalent here.
const Type *TyPtr = Ty.getTypePtr();
if (!Base)
Base = TyPtr;
if (Base != TyPtr &&
(!Base->isVectorType() || !TyPtr->isVectorType() ||
Context.getTypeSize(Base) != Context.getTypeSize(TyPtr)))
return false;
}
// Homogeneous Aggregates can have at most 4 members of the base type.
if (HAMembers)
*HAMembers = Members;
return (Members > 0 && Members <= 4);
}
/// markAllocatedVFPs - update VFPRegs according to the alignment and
/// number of VFP registers (unit is S register) requested.
static void markAllocatedVFPs(int *VFPRegs, unsigned &AllocatedVFP,
unsigned Alignment,
unsigned NumRequired) {
// Early Exit.
if (AllocatedVFP >= 16)
return;
// C.1.vfp If the argument is a VFP CPRC and there are sufficient consecutive
// VFP registers of the appropriate type unallocated then the argument is
// allocated to the lowest-numbered sequence of such registers.
for (unsigned I = 0; I < 16; I += Alignment) {
bool FoundSlot = true;
for (unsigned J = I, JEnd = I + NumRequired; J < JEnd; J++)
if (J >= 16 || VFPRegs[J]) {
FoundSlot = false;
break;
}
if (FoundSlot) {
for (unsigned J = I, JEnd = I + NumRequired; J < JEnd; J++)
VFPRegs[J] = 1;
AllocatedVFP += NumRequired;
return;
}
}
// C.2.vfp If the argument is a VFP CPRC then any VFP registers that are
// unallocated are marked as unavailable.
for (unsigned I = 0; I < 16; I++)
VFPRegs[I] = 1;
AllocatedVFP = 17; // We do not have enough VFP registers.
}
ABIArgInfo ARMABIInfo::classifyArgumentType(QualType Ty, int *VFPRegs,
unsigned &AllocatedVFP,
bool &IsHA) const {
// We update number of allocated VFPs according to
// 6.1.2.1 The following argument types are VFP CPRCs:
// A single-precision floating-point type (including promoted
// half-precision types); A double-precision floating-point type;
// A 64-bit or 128-bit containerized vector type; Homogeneous Aggregate
// with a Base Type of a single- or double-precision floating-point type,
// 64-bit containerized vectors or 128-bit containerized vectors with one
// to four Elements.
// Handle illegal vector types here.
if (isIllegalVectorType(Ty)) {
uint64_t Size = getContext().getTypeSize(Ty);
if (Size <= 32) {
llvm::Type *ResType =
llvm::Type::getInt32Ty(getVMContext());
return ABIArgInfo::getDirect(ResType);
}
if (Size == 64) {
llvm::Type *ResType = llvm::VectorType::get(
llvm::Type::getInt32Ty(getVMContext()), 2);
markAllocatedVFPs(VFPRegs, AllocatedVFP, 2, 2);
return ABIArgInfo::getDirect(ResType);
}
if (Size == 128) {
llvm::Type *ResType = llvm::VectorType::get(
llvm::Type::getInt32Ty(getVMContext()), 4);
markAllocatedVFPs(VFPRegs, AllocatedVFP, 4, 4);
return ABIArgInfo::getDirect(ResType);
}
return ABIArgInfo::getIndirect(0, /*ByVal=*/false);
}
// Update VFPRegs for legal vector types.
if (const VectorType *VT = Ty->getAs<VectorType>()) {
uint64_t Size = getContext().getTypeSize(VT);
// Size of a legal vector should be power of 2 and above 64.
markAllocatedVFPs(VFPRegs, AllocatedVFP, Size >= 128 ? 4 : 2, Size / 32);
}
// Update VFPRegs for floating point types.
if (const BuiltinType *BT = Ty->getAs<BuiltinType>()) {
if (BT->getKind() == BuiltinType::Half ||
BT->getKind() == BuiltinType::Float)
markAllocatedVFPs(VFPRegs, AllocatedVFP, 1, 1);
if (BT->getKind() == BuiltinType::Double ||
BT->getKind() == BuiltinType::LongDouble)
markAllocatedVFPs(VFPRegs, AllocatedVFP, 2, 2);
}
if (!isAggregateTypeForABI(Ty)) {
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
return (Ty->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
// Ignore empty records.
if (isEmptyRecord(getContext(), Ty, true))
return ABIArgInfo::getIgnore();
if (CGCXXABI::RecordArgABI RAA = getRecordArgABI(Ty, CGT))
return ABIArgInfo::getIndirect(0, RAA == CGCXXABI::RAA_DirectInMemory);
if (getABIKind() == ARMABIInfo::AAPCS_VFP) {
// Homogeneous Aggregates need to be expanded when we can fit the aggregate
// into VFP registers.
const Type *Base = 0;
uint64_t Members = 0;
if (isHomogeneousAggregate(Ty, Base, getContext(), &Members)) {
assert(Base && "Base class should be set for homogeneous aggregate");
// Base can be a floating-point or a vector.
if (Base->isVectorType()) {
// ElementSize is in number of floats.
unsigned ElementSize = getContext().getTypeSize(Base) == 64 ? 2 : 4;
markAllocatedVFPs(VFPRegs, AllocatedVFP, ElementSize,
Members * ElementSize);
} else if (Base->isSpecificBuiltinType(BuiltinType::Float))
markAllocatedVFPs(VFPRegs, AllocatedVFP, 1, Members);
else {
assert(Base->isSpecificBuiltinType(BuiltinType::Double) ||
Base->isSpecificBuiltinType(BuiltinType::LongDouble));
markAllocatedVFPs(VFPRegs, AllocatedVFP, 2, Members * 2);
}
IsHA = true;
return ABIArgInfo::getExpand();
}
}
// Support byval for ARM.
// The ABI alignment for APCS is 4-byte and for AAPCS at least 4-byte and at
// most 8-byte. We realign the indirect argument if type alignment is bigger
// than ABI alignment.
uint64_t ABIAlign = 4;
uint64_t TyAlign = getContext().getTypeAlign(Ty) / 8;
if (getABIKind() == ARMABIInfo::AAPCS_VFP ||
getABIKind() == ARMABIInfo::AAPCS)
ABIAlign = std::min(std::max(TyAlign, (uint64_t)4), (uint64_t)8);
if (getContext().getTypeSizeInChars(Ty) > CharUnits::fromQuantity(64)) {
return ABIArgInfo::getIndirect(0, /*ByVal=*/true,
/*Realign=*/TyAlign > ABIAlign);
}
// Otherwise, pass by coercing to a structure of the appropriate size.
llvm::Type* ElemTy;
unsigned SizeRegs;
// FIXME: Try to match the types of the arguments more accurately where
// we can.
if (getContext().getTypeAlign(Ty) <= 32) {
ElemTy = llvm::Type::getInt32Ty(getVMContext());
SizeRegs = (getContext().getTypeSize(Ty) + 31) / 32;
} else {
ElemTy = llvm::Type::getInt64Ty(getVMContext());
SizeRegs = (getContext().getTypeSize(Ty) + 63) / 64;
}
llvm::Type *STy =
llvm::StructType::get(llvm::ArrayType::get(ElemTy, SizeRegs), NULL);
return ABIArgInfo::getDirect(STy);
}
static bool isIntegerLikeType(QualType Ty, ASTContext &Context,
llvm::LLVMContext &VMContext) {
// APCS, C Language Calling Conventions, Non-Simple Return Values: A structure
// is called integer-like if its size is less than or equal to one word, and
// the offset of each of its addressable sub-fields is zero.
uint64_t Size = Context.getTypeSize(Ty);
// Check that the type fits in a word.
if (Size > 32)
return false;
// FIXME: Handle vector types!
if (Ty->isVectorType())
return false;
// Float types are never treated as "integer like".
if (Ty->isRealFloatingType())
return false;
// If this is a builtin or pointer type then it is ok.
if (Ty->getAs<BuiltinType>() || Ty->isPointerType())
return true;
// Small complex integer types are "integer like".
if (const ComplexType *CT = Ty->getAs<ComplexType>())
return isIntegerLikeType(CT->getElementType(), Context, VMContext);
// Single element and zero sized arrays should be allowed, by the definition
// above, but they are not.
// Otherwise, it must be a record type.
const RecordType *RT = Ty->getAs<RecordType>();
if (!RT) return false;
// Ignore records with flexible arrays.
const RecordDecl *RD = RT->getDecl();
if (RD->hasFlexibleArrayMember())
return false;
// Check that all sub-fields are at offset 0, and are themselves "integer
// like".
const ASTRecordLayout &Layout = Context.getASTRecordLayout(RD);
bool HadField = false;
unsigned idx = 0;
for (RecordDecl::field_iterator i = RD->field_begin(), e = RD->field_end();
i != e; ++i, ++idx) {
const FieldDecl *FD = *i;
// Bit-fields are not addressable, we only need to verify they are "integer
// like". We still have to disallow a subsequent non-bitfield, for example:
// struct { int : 0; int x }
// is non-integer like according to gcc.
if (FD->isBitField()) {
if (!RD->isUnion())
HadField = true;
if (!isIntegerLikeType(FD->getType(), Context, VMContext))
return false;
continue;
}
// Check if this field is at offset 0.
if (Layout.getFieldOffset(idx) != 0)
return false;
if (!isIntegerLikeType(FD->getType(), Context, VMContext))
return false;
// Only allow at most one field in a structure. This doesn't match the
// wording above, but follows gcc in situations with a field following an
// empty structure.
if (!RD->isUnion()) {
if (HadField)
return false;
HadField = true;
}
}
return true;
}
ABIArgInfo ARMABIInfo::classifyReturnType(QualType RetTy) const {
if (RetTy->isVoidType())
return ABIArgInfo::getIgnore();
// Large vector types should be returned via memory.
if (RetTy->isVectorType() && getContext().getTypeSize(RetTy) > 128)
return ABIArgInfo::getIndirect(0);
if (!isAggregateTypeForABI(RetTy)) {
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = RetTy->getAs<EnumType>())
RetTy = EnumTy->getDecl()->getIntegerType();
return (RetTy->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
// Structures with either a non-trivial destructor or a non-trivial
// copy constructor are always indirect.
if (isRecordReturnIndirect(RetTy, CGT))
return ABIArgInfo::getIndirect(0, /*ByVal=*/false);
// Are we following APCS?
if (getABIKind() == APCS) {
if (isEmptyRecord(getContext(), RetTy, false))
return ABIArgInfo::getIgnore();
// Complex types are all returned as packed integers.
//
// FIXME: Consider using 2 x vector types if the back end handles them
// correctly.
if (RetTy->isAnyComplexType())
return ABIArgInfo::getDirect(llvm::IntegerType::get(getVMContext(),
getContext().getTypeSize(RetTy)));
// Integer like structures are returned in r0.
if (isIntegerLikeType(RetTy, getContext(), getVMContext())) {
// Return in the smallest viable integer type.
uint64_t Size = getContext().getTypeSize(RetTy);
if (Size <= 8)
return ABIArgInfo::getDirect(llvm::Type::getInt8Ty(getVMContext()));
if (Size <= 16)
return ABIArgInfo::getDirect(llvm::Type::getInt16Ty(getVMContext()));
return ABIArgInfo::getDirect(llvm::Type::getInt32Ty(getVMContext()));
}
// Otherwise return in memory.
return ABIArgInfo::getIndirect(0);
}
// Otherwise this is an AAPCS variant.
if (isEmptyRecord(getContext(), RetTy, true))
return ABIArgInfo::getIgnore();
// Check for homogeneous aggregates with AAPCS-VFP.
if (getABIKind() == AAPCS_VFP) {
const Type *Base = 0;
if (isHomogeneousAggregate(RetTy, Base, getContext())) {
assert(Base && "Base class should be set for homogeneous aggregate");
// Homogeneous Aggregates are returned directly.
return ABIArgInfo::getDirect();
}
}
// Aggregates <= 4 bytes are returned in r0; other aggregates
// are returned indirectly.
uint64_t Size = getContext().getTypeSize(RetTy);
if (Size <= 32) {
// Return in the smallest viable integer type.
if (Size <= 8)
return ABIArgInfo::getDirect(llvm::Type::getInt8Ty(getVMContext()));
if (Size <= 16)
return ABIArgInfo::getDirect(llvm::Type::getInt16Ty(getVMContext()));
return ABIArgInfo::getDirect(llvm::Type::getInt32Ty(getVMContext()));
}
return ABIArgInfo::getIndirect(0);
}
/// isIllegalVector - check whether Ty is an illegal vector type.
bool ARMABIInfo::isIllegalVectorType(QualType Ty) const {
if (const VectorType *VT = Ty->getAs<VectorType>()) {
// Check whether VT is legal.
unsigned NumElements = VT->getNumElements();
uint64_t Size = getContext().getTypeSize(VT);
// NumElements should be power of 2.
if ((NumElements & (NumElements - 1)) != 0)
return true;
// Size should be greater than 32 bits.
return Size <= 32;
}
return false;
}
llvm::Value *ARMABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
llvm::Type *BP = CGF.Int8PtrTy;
llvm::Type *BPP = CGF.Int8PtrPtrTy;
CGBuilderTy &Builder = CGF.Builder;
llvm::Value *VAListAddrAsBPP = Builder.CreateBitCast(VAListAddr, BPP, "ap");
llvm::Value *Addr = Builder.CreateLoad(VAListAddrAsBPP, "ap.cur");
uint64_t Size = CGF.getContext().getTypeSize(Ty) / 8;
uint64_t TyAlign = CGF.getContext().getTypeAlign(Ty) / 8;
bool IsIndirect = false;
// The ABI alignment for 64-bit or 128-bit vectors is 8 for AAPCS and 4 for
// APCS. For AAPCS, the ABI alignment is at least 4-byte and at most 8-byte.
if (getABIKind() == ARMABIInfo::AAPCS_VFP ||
getABIKind() == ARMABIInfo::AAPCS)
TyAlign = std::min(std::max(TyAlign, (uint64_t)4), (uint64_t)8);
else
TyAlign = 4;
// Use indirect if size of the illegal vector is bigger than 16 bytes.
if (isIllegalVectorType(Ty) && Size > 16) {
IsIndirect = true;
Size = 4;
TyAlign = 4;
}
// Handle address alignment for ABI alignment > 4 bytes.
if (TyAlign > 4) {
assert((TyAlign & (TyAlign - 1)) == 0 &&
"Alignment is not power of 2!");
llvm::Value *AddrAsInt = Builder.CreatePtrToInt(Addr, CGF.Int32Ty);
AddrAsInt = Builder.CreateAdd(AddrAsInt, Builder.getInt32(TyAlign - 1));
AddrAsInt = Builder.CreateAnd(AddrAsInt, Builder.getInt32(~(TyAlign - 1)));
Addr = Builder.CreateIntToPtr(AddrAsInt, BP, "ap.align");
}
uint64_t Offset =
llvm::RoundUpToAlignment(Size, 4);
llvm::Value *NextAddr =
Builder.CreateGEP(Addr, llvm::ConstantInt::get(CGF.Int32Ty, Offset),
"ap.next");
Builder.CreateStore(NextAddr, VAListAddrAsBPP);
if (IsIndirect)
Addr = Builder.CreateLoad(Builder.CreateBitCast(Addr, BPP));
else if (TyAlign < CGF.getContext().getTypeAlign(Ty) / 8) {
// We can't directly cast ap.cur to pointer to a vector type, since ap.cur
// may not be correctly aligned for the vector type. We create an aligned
// temporary space and copy the content over from ap.cur to the temporary
// space. This is necessary if the natural alignment of the type is greater
// than the ABI alignment.
llvm::Type *I8PtrTy = Builder.getInt8PtrTy();
CharUnits CharSize = getContext().getTypeSizeInChars(Ty);
llvm::Value *AlignedTemp = CGF.CreateTempAlloca(CGF.ConvertType(Ty),
"var.align");
llvm::Value *Dst = Builder.CreateBitCast(AlignedTemp, I8PtrTy);
llvm::Value *Src = Builder.CreateBitCast(Addr, I8PtrTy);
Builder.CreateMemCpy(Dst, Src,
llvm::ConstantInt::get(CGF.IntPtrTy, CharSize.getQuantity()),
TyAlign, false);
Addr = AlignedTemp; //The content is in aligned location.
}
llvm::Type *PTy =
llvm::PointerType::getUnqual(CGF.ConvertType(Ty));
llvm::Value *AddrTyped = Builder.CreateBitCast(Addr, PTy);
return AddrTyped;
}
namespace {
class NaClARMABIInfo : public ABIInfo {
public:
NaClARMABIInfo(CodeGen::CodeGenTypes &CGT, ARMABIInfo::ABIKind Kind)
: ABIInfo(CGT), PInfo(CGT), NInfo(CGT, Kind) {}
virtual void computeInfo(CGFunctionInfo &FI) const;
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
private:
PNaClABIInfo PInfo; // Used for generating calls with pnaclcall callingconv.
ARMABIInfo NInfo; // Used for everything else.
};
class NaClARMTargetCodeGenInfo : public TargetCodeGenInfo {
public:
NaClARMTargetCodeGenInfo(CodeGen::CodeGenTypes &CGT, ARMABIInfo::ABIKind Kind)
: TargetCodeGenInfo(new NaClARMABIInfo(CGT, Kind)) {}
};
}
void NaClARMABIInfo::computeInfo(CGFunctionInfo &FI) const {
if (FI.getASTCallingConvention() == CC_PnaclCall)
PInfo.computeInfo(FI);
else
static_cast<const ABIInfo&>(NInfo).computeInfo(FI);
}
llvm::Value *NaClARMABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
// Always use the native convention; calling pnacl-style varargs functions
// is unsupported.
return static_cast<const ABIInfo&>(NInfo).EmitVAArg(VAListAddr, Ty, CGF);
}
//===----------------------------------------------------------------------===//
// AArch64 ABI Implementation
//===----------------------------------------------------------------------===//
namespace {
class AArch64ABIInfo : public ABIInfo {
public:
AArch64ABIInfo(CodeGenTypes &CGT) : ABIInfo(CGT) {}
private:
// The AArch64 PCS is explicit about return types and argument types being
// handled identically, so we don't need to draw a distinction between
// Argument and Return classification.
ABIArgInfo classifyGenericType(QualType Ty, int &FreeIntRegs,
int &FreeVFPRegs) const;
ABIArgInfo tryUseRegs(QualType Ty, int &FreeRegs, int RegsNeeded, bool IsInt,
llvm::Type *DirectTy = 0) const;
virtual void computeInfo(CGFunctionInfo &FI) const;
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
};
class AArch64TargetCodeGenInfo : public TargetCodeGenInfo {
public:
AArch64TargetCodeGenInfo(CodeGenTypes &CGT)
:TargetCodeGenInfo(new AArch64ABIInfo(CGT)) {}
const AArch64ABIInfo &getABIInfo() const {
return static_cast<const AArch64ABIInfo&>(TargetCodeGenInfo::getABIInfo());
}
int getDwarfEHStackPointer(CodeGen::CodeGenModule &M) const {
return 31;
}
bool initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const {
// 0-31 are x0-x30 and sp: 8 bytes each
llvm::Value *Eight8 = llvm::ConstantInt::get(CGF.Int8Ty, 8);
AssignToArrayRange(CGF.Builder, Address, Eight8, 0, 31);
// 64-95 are v0-v31: 16 bytes each
llvm::Value *Sixteen8 = llvm::ConstantInt::get(CGF.Int8Ty, 16);
AssignToArrayRange(CGF.Builder, Address, Sixteen8, 64, 95);
return false;
}
};
}
void AArch64ABIInfo::computeInfo(CGFunctionInfo &FI) const {
int FreeIntRegs = 8, FreeVFPRegs = 8;
FI.getReturnInfo() = classifyGenericType(FI.getReturnType(),
FreeIntRegs, FreeVFPRegs);
FreeIntRegs = FreeVFPRegs = 8;
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it) {
it->info = classifyGenericType(it->type, FreeIntRegs, FreeVFPRegs);
}
}
ABIArgInfo
AArch64ABIInfo::tryUseRegs(QualType Ty, int &FreeRegs, int RegsNeeded,
bool IsInt, llvm::Type *DirectTy) const {
if (FreeRegs >= RegsNeeded) {
FreeRegs -= RegsNeeded;
return ABIArgInfo::getDirect(DirectTy);
}
llvm::Type *Padding = 0;
// We need padding so that later arguments don't get filled in anyway. That
// wouldn't happen if only ByVal arguments followed in the same category, but
// a large structure will simply seem to be a pointer as far as LLVM is
// concerned.
if (FreeRegs > 0) {
if (IsInt)
Padding = llvm::Type::getInt64Ty(getVMContext());
else
Padding = llvm::Type::getFloatTy(getVMContext());
// Either [N x i64] or [N x float].
Padding = llvm::ArrayType::get(Padding, FreeRegs);
FreeRegs = 0;
}
return ABIArgInfo::getIndirect(getContext().getTypeAlign(Ty) / 8,
/*IsByVal=*/ true, /*Realign=*/ false,
Padding);
}
ABIArgInfo AArch64ABIInfo::classifyGenericType(QualType Ty,
int &FreeIntRegs,
int &FreeVFPRegs) const {
// Can only occurs for return, but harmless otherwise.
if (Ty->isVoidType())
return ABIArgInfo::getIgnore();
// Large vector types should be returned via memory. There's no such concept
// in the ABI, but they'd be over 16 bytes anyway so no matter how they're
// classified they'd go into memory (see B.3).
if (Ty->isVectorType() && getContext().getTypeSize(Ty) > 128) {
if (FreeIntRegs > 0)
--FreeIntRegs;
return ABIArgInfo::getIndirect(0, /*ByVal=*/false);
}
// All non-aggregate LLVM types have a concrete ABI representation so they can
// be passed directly. After this block we're guaranteed to be in a
// complicated case.
if (!isAggregateTypeForABI(Ty)) {
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
if (Ty->isFloatingType() || Ty->isVectorType())
return tryUseRegs(Ty, FreeVFPRegs, /*RegsNeeded=*/ 1, /*IsInt=*/ false);
assert(getContext().getTypeSize(Ty) <= 128 &&
"unexpectedly large scalar type");
int RegsNeeded = getContext().getTypeSize(Ty) > 64 ? 2 : 1;
// If the type may need padding registers to ensure "alignment", we must be
// careful when this is accounted for. Increasing the effective size covers
// all cases.
if (getContext().getTypeAlign(Ty) == 128)
RegsNeeded += FreeIntRegs % 2 != 0;
return tryUseRegs(Ty, FreeIntRegs, RegsNeeded, /*IsInt=*/ true);
}
if (CGCXXABI::RecordArgABI RAA = getRecordArgABI(Ty, CGT)) {
if (FreeIntRegs > 0 && RAA == CGCXXABI::RAA_Indirect)
--FreeIntRegs;
return ABIArgInfo::getIndirect(0, RAA == CGCXXABI::RAA_DirectInMemory);
}
if (isEmptyRecord(getContext(), Ty, true)) {
if (!getContext().getLangOpts().CPlusPlus) {
// Empty structs outside C++ mode are a GNU extension, so no ABI can
// possibly tell us what to do. It turns out (I believe) that GCC ignores
// the object for parameter-passsing purposes.
return ABIArgInfo::getIgnore();
}
// The combination of C++98 9p5 (sizeof(struct) != 0) and the pseudocode
// description of va_arg in the PCS require that an empty struct does
// actually occupy space for parameter-passing. I'm hoping for a
// clarification giving an explicit paragraph to point to in future.
return tryUseRegs(Ty, FreeIntRegs, /*RegsNeeded=*/ 1, /*IsInt=*/ true,
llvm::Type::getInt8Ty(getVMContext()));
}
// Homogeneous vector aggregates get passed in registers or on the stack.
const Type *Base = 0;
uint64_t NumMembers = 0;
if (isHomogeneousAggregate(Ty, Base, getContext(), &NumMembers)) {
assert(Base && "Base class should be set for homogeneous aggregate");
// Homogeneous aggregates are passed and returned directly.
return tryUseRegs(Ty, FreeVFPRegs, /*RegsNeeded=*/ NumMembers,
/*IsInt=*/ false);
}
uint64_t Size = getContext().getTypeSize(Ty);
if (Size <= 128) {
// Small structs can use the same direct type whether they're in registers
// or on the stack.
llvm::Type *BaseTy;
unsigned NumBases;
int SizeInRegs = (Size + 63) / 64;
if (getContext().getTypeAlign(Ty) == 128) {
BaseTy = llvm::Type::getIntNTy(getVMContext(), 128);
NumBases = 1;
// If the type may need padding registers to ensure "alignment", we must
// be careful when this is accounted for. Increasing the effective size
// covers all cases.
SizeInRegs += FreeIntRegs % 2 != 0;
} else {
BaseTy = llvm::Type::getInt64Ty(getVMContext());
NumBases = SizeInRegs;
}
llvm::Type *DirectTy = llvm::ArrayType::get(BaseTy, NumBases);
return tryUseRegs(Ty, FreeIntRegs, /*RegsNeeded=*/ SizeInRegs,
/*IsInt=*/ true, DirectTy);
}
// If the aggregate is > 16 bytes, it's passed and returned indirectly. In
// LLVM terms the return uses an "sret" pointer, but that's handled elsewhere.
--FreeIntRegs;
return ABIArgInfo::getIndirect(0, /* byVal = */ false);
}
llvm::Value *AArch64ABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
// The AArch64 va_list type and handling is specified in the Procedure Call
// Standard, section B.4:
//
// struct {
// void *__stack;
// void *__gr_top;
// void *__vr_top;
// int __gr_offs;
// int __vr_offs;
// };
assert(!CGF.CGM.getDataLayout().isBigEndian()
&& "va_arg not implemented for big-endian AArch64");
int FreeIntRegs = 8, FreeVFPRegs = 8;
Ty = CGF.getContext().getCanonicalType(Ty);
ABIArgInfo AI = classifyGenericType(Ty, FreeIntRegs, FreeVFPRegs);
llvm::BasicBlock *MaybeRegBlock = CGF.createBasicBlock("vaarg.maybe_reg");
llvm::BasicBlock *InRegBlock = CGF.createBasicBlock("vaarg.in_reg");
llvm::BasicBlock *OnStackBlock = CGF.createBasicBlock("vaarg.on_stack");
llvm::BasicBlock *ContBlock = CGF.createBasicBlock("vaarg.end");
llvm::Value *reg_offs_p = 0, *reg_offs = 0;
int reg_top_index;
int RegSize;
if (FreeIntRegs < 8) {
assert(FreeVFPRegs == 8 && "Arguments never split between int & VFP regs");
// 3 is the field number of __gr_offs
reg_offs_p = CGF.Builder.CreateStructGEP(VAListAddr, 3, "gr_offs_p");
reg_offs = CGF.Builder.CreateLoad(reg_offs_p, "gr_offs");
reg_top_index = 1; // field number for __gr_top
RegSize = 8 * (8 - FreeIntRegs);
} else {
assert(FreeVFPRegs < 8 && "Argument must go in VFP or int regs");
// 4 is the field number of __vr_offs.
reg_offs_p = CGF.Builder.CreateStructGEP(VAListAddr, 4, "vr_offs_p");
reg_offs = CGF.Builder.CreateLoad(reg_offs_p, "vr_offs");
reg_top_index = 2; // field number for __vr_top
RegSize = 16 * (8 - FreeVFPRegs);
}
//=======================================
// Find out where argument was passed
//=======================================
// If reg_offs >= 0 we're already using the stack for this type of
// argument. We don't want to keep updating reg_offs (in case it overflows,
// though anyone passing 2GB of arguments, each at most 16 bytes, deserves
// whatever they get).
llvm::Value *UsingStack = 0;
UsingStack = CGF.Builder.CreateICmpSGE(reg_offs,
llvm::ConstantInt::get(CGF.Int32Ty, 0));
CGF.Builder.CreateCondBr(UsingStack, OnStackBlock, MaybeRegBlock);
// Otherwise, at least some kind of argument could go in these registers, the
// quesiton is whether this particular type is too big.
CGF.EmitBlock(MaybeRegBlock);
// Integer arguments may need to correct register alignment (for example a
// "struct { __int128 a; };" gets passed in x_2N, x_{2N+1}). In this case we
// align __gr_offs to calculate the potential address.
if (FreeIntRegs < 8 && AI.isDirect() && getContext().getTypeAlign(Ty) > 64) {
int Align = getContext().getTypeAlign(Ty) / 8;
reg_offs = CGF.Builder.CreateAdd(reg_offs,
llvm::ConstantInt::get(CGF.Int32Ty, Align - 1),
"align_regoffs");
reg_offs = CGF.Builder.CreateAnd(reg_offs,
llvm::ConstantInt::get(CGF.Int32Ty, -Align),
"aligned_regoffs");
}
// Update the gr_offs/vr_offs pointer for next call to va_arg on this va_list.
llvm::Value *NewOffset = 0;
NewOffset = CGF.Builder.CreateAdd(reg_offs,
llvm::ConstantInt::get(CGF.Int32Ty, RegSize),
"new_reg_offs");
CGF.Builder.CreateStore(NewOffset, reg_offs_p);
// Now we're in a position to decide whether this argument really was in
// registers or not.
llvm::Value *InRegs = 0;
InRegs = CGF.Builder.CreateICmpSLE(NewOffset,
llvm::ConstantInt::get(CGF.Int32Ty, 0),
"inreg");
CGF.Builder.CreateCondBr(InRegs, InRegBlock, OnStackBlock);
//=======================================
// Argument was in registers
//=======================================
// Now we emit the code for if the argument was originally passed in
// registers. First start the appropriate block:
CGF.EmitBlock(InRegBlock);
llvm::Value *reg_top_p = 0, *reg_top = 0;
reg_top_p = CGF.Builder.CreateStructGEP(VAListAddr, reg_top_index, "reg_top_p");
reg_top = CGF.Builder.CreateLoad(reg_top_p, "reg_top");
llvm::Value *BaseAddr = CGF.Builder.CreateGEP(reg_top, reg_offs);
llvm::Value *RegAddr = 0;
llvm::Type *MemTy = llvm::PointerType::getUnqual(CGF.ConvertTypeForMem(Ty));
if (!AI.isDirect()) {
// If it's been passed indirectly (actually a struct), whatever we find from
// stored registers or on the stack will actually be a struct **.
MemTy = llvm::PointerType::getUnqual(MemTy);
}
const Type *Base = 0;
uint64_t NumMembers;
if (isHomogeneousAggregate(Ty, Base, getContext(), &NumMembers)
&& NumMembers > 1) {
// Homogeneous aggregates passed in registers will have their elements split
// and stored 16-bytes apart regardless of size (they're notionally in qN,
// qN+1, ...). We reload and store into a temporary local variable
// contiguously.
assert(AI.isDirect() && "Homogeneous aggregates should be passed directly");
llvm::Type *BaseTy = CGF.ConvertType(QualType(Base, 0));
llvm::Type *HFATy = llvm::ArrayType::get(BaseTy, NumMembers);
llvm::Value *Tmp = CGF.CreateTempAlloca(HFATy);
for (unsigned i = 0; i < NumMembers; ++i) {
llvm::Value *BaseOffset = llvm::ConstantInt::get(CGF.Int32Ty, 16 * i);
llvm::Value *LoadAddr = CGF.Builder.CreateGEP(BaseAddr, BaseOffset);
LoadAddr = CGF.Builder.CreateBitCast(LoadAddr,
llvm::PointerType::getUnqual(BaseTy));
llvm::Value *StoreAddr = CGF.Builder.CreateStructGEP(Tmp, i);
llvm::Value *Elem = CGF.Builder.CreateLoad(LoadAddr);
CGF.Builder.CreateStore(Elem, StoreAddr);
}
RegAddr = CGF.Builder.CreateBitCast(Tmp, MemTy);
} else {
// Otherwise the object is contiguous in memory
RegAddr = CGF.Builder.CreateBitCast(BaseAddr, MemTy);
}
CGF.EmitBranch(ContBlock);
//=======================================
// Argument was on the stack
//=======================================
CGF.EmitBlock(OnStackBlock);
llvm::Value *stack_p = 0, *OnStackAddr = 0;
stack_p = CGF.Builder.CreateStructGEP(VAListAddr, 0, "stack_p");
OnStackAddr = CGF.Builder.CreateLoad(stack_p, "stack");
// Again, stack arguments may need realigmnent. In this case both integer and
// floating-point ones might be affected.
if (AI.isDirect() && getContext().getTypeAlign(Ty) > 64) {
int Align = getContext().getTypeAlign(Ty) / 8;
OnStackAddr = CGF.Builder.CreatePtrToInt(OnStackAddr, CGF.Int64Ty);
OnStackAddr = CGF.Builder.CreateAdd(OnStackAddr,
llvm::ConstantInt::get(CGF.Int64Ty, Align - 1),
"align_stack");
OnStackAddr = CGF.Builder.CreateAnd(OnStackAddr,
llvm::ConstantInt::get(CGF.Int64Ty, -Align),
"align_stack");
OnStackAddr = CGF.Builder.CreateIntToPtr(OnStackAddr, CGF.Int8PtrTy);
}
uint64_t StackSize;
if (AI.isDirect())
StackSize = getContext().getTypeSize(Ty) / 8;
else
StackSize = 8;
// All stack slots are 8 bytes
StackSize = llvm::RoundUpToAlignment(StackSize, 8);
llvm::Value *StackSizeC = llvm::ConstantInt::get(CGF.Int32Ty, StackSize);
llvm::Value *NewStack = CGF.Builder.CreateGEP(OnStackAddr, StackSizeC,
"new_stack");
// Write the new value of __stack for the next call to va_arg
CGF.Builder.CreateStore(NewStack, stack_p);
OnStackAddr = CGF.Builder.CreateBitCast(OnStackAddr, MemTy);
CGF.EmitBranch(ContBlock);
//=======================================
// Tidy up
//=======================================
CGF.EmitBlock(ContBlock);
llvm::PHINode *ResAddr = CGF.Builder.CreatePHI(MemTy, 2, "vaarg.addr");
ResAddr->addIncoming(RegAddr, InRegBlock);
ResAddr->addIncoming(OnStackAddr, OnStackBlock);
if (AI.isDirect())
return ResAddr;
return CGF.Builder.CreateLoad(ResAddr, "vaarg.addr");
}
//===----------------------------------------------------------------------===//
// NVPTX ABI Implementation
//===----------------------------------------------------------------------===//
namespace {
class NVPTXABIInfo : public ABIInfo {
public:
NVPTXABIInfo(CodeGenTypes &CGT) : ABIInfo(CGT) {}
ABIArgInfo classifyReturnType(QualType RetTy) const;
ABIArgInfo classifyArgumentType(QualType Ty) const;
virtual void computeInfo(CGFunctionInfo &FI) const;
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CFG) const;
};
class NVPTXTargetCodeGenInfo : public TargetCodeGenInfo {
public:
NVPTXTargetCodeGenInfo(CodeGenTypes &CGT)
: TargetCodeGenInfo(new NVPTXABIInfo(CGT)) {}
virtual void SetTargetAttributes(const Decl *D, llvm::GlobalValue *GV,
CodeGen::CodeGenModule &M) const;
private:
static void addKernelMetadata(llvm::Function *F);
};
ABIArgInfo NVPTXABIInfo::classifyReturnType(QualType RetTy) const {
if (RetTy->isVoidType())
return ABIArgInfo::getIgnore();
if (isAggregateTypeForABI(RetTy))
return ABIArgInfo::getIndirect(0);
return ABIArgInfo::getDirect();
}
ABIArgInfo NVPTXABIInfo::classifyArgumentType(QualType Ty) const {
if (isAggregateTypeForABI(Ty))
return ABIArgInfo::getIndirect(0);
return ABIArgInfo::getDirect();
}
void NVPTXABIInfo::computeInfo(CGFunctionInfo &FI) const {
FI.getReturnInfo() = classifyReturnType(FI.getReturnType());
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it)
it->info = classifyArgumentType(it->type);
// Always honor user-specified calling convention.
if (FI.getCallingConvention() != llvm::CallingConv::C)
return;
FI.setEffectiveCallingConvention(getRuntimeCC());
}
llvm::Value *NVPTXABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CFG) const {
llvm_unreachable("NVPTX does not support varargs");
}
void NVPTXTargetCodeGenInfo::
SetTargetAttributes(const Decl *D, llvm::GlobalValue *GV,
CodeGen::CodeGenModule &M) const{
const FunctionDecl *FD = dyn_cast<FunctionDecl>(D);
if (!FD) return;
llvm::Function *F = cast<llvm::Function>(GV);
// Perform special handling in OpenCL mode
if (M.getLangOpts().OpenCL) {
// Use OpenCL function attributes to check for kernel functions
// By default, all functions are device functions
if (FD->hasAttr<OpenCLKernelAttr>()) {
// OpenCL __kernel functions get kernel metadata
addKernelMetadata(F);
// And kernel functions are not subject to inlining
F->addFnAttr(llvm::Attribute::NoInline);
}
}
// Perform special handling in CUDA mode.
if (M.getLangOpts().CUDA) {
// CUDA __global__ functions get a kernel metadata entry. Since
// __global__ functions cannot be called from the device, we do not
// need to set the noinline attribute.
if (FD->getAttr<CUDAGlobalAttr>())
addKernelMetadata(F);
}
}
void NVPTXTargetCodeGenInfo::addKernelMetadata(llvm::Function *F) {
llvm::Module *M = F->getParent();
llvm::LLVMContext &Ctx = M->getContext();
// Get "nvvm.annotations" metadata node
llvm::NamedMDNode *MD = M->getOrInsertNamedMetadata("nvvm.annotations");
// Create !{<func-ref>, metadata !"kernel", i32 1} node
llvm::SmallVector<llvm::Value *, 3> MDVals;
MDVals.push_back(F);
MDVals.push_back(llvm::MDString::get(Ctx, "kernel"));
MDVals.push_back(llvm::ConstantInt::get(llvm::Type::getInt32Ty(Ctx), 1));
// Append metadata to nvvm.annotations
MD->addOperand(llvm::MDNode::get(Ctx, MDVals));
}
}
//===----------------------------------------------------------------------===//
// SystemZ ABI Implementation
//===----------------------------------------------------------------------===//
namespace {
class SystemZABIInfo : public ABIInfo {
public:
SystemZABIInfo(CodeGenTypes &CGT) : ABIInfo(CGT) {}
bool isPromotableIntegerType(QualType Ty) const;
bool isCompoundType(QualType Ty) const;
bool isFPArgumentType(QualType Ty) const;
ABIArgInfo classifyReturnType(QualType RetTy) const;
ABIArgInfo classifyArgumentType(QualType ArgTy) const;
virtual void computeInfo(CGFunctionInfo &FI) const {
FI.getReturnInfo() = classifyReturnType(FI.getReturnType());
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it)
it->info = classifyArgumentType(it->type);
}
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
};
class SystemZTargetCodeGenInfo : public TargetCodeGenInfo {
public:
SystemZTargetCodeGenInfo(CodeGenTypes &CGT)
: TargetCodeGenInfo(new SystemZABIInfo(CGT)) {}
};
}
bool SystemZABIInfo::isPromotableIntegerType(QualType Ty) const {
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
// Promotable integer types are required to be promoted by the ABI.
if (Ty->isPromotableIntegerType())
return true;
// 32-bit values must also be promoted.
if (const BuiltinType *BT = Ty->getAs<BuiltinType>())
switch (BT->getKind()) {
case BuiltinType::Int:
case BuiltinType::UInt:
return true;
default:
return false;
}
return false;
}
bool SystemZABIInfo::isCompoundType(QualType Ty) const {
return Ty->isAnyComplexType() || isAggregateTypeForABI(Ty);
}
bool SystemZABIInfo::isFPArgumentType(QualType Ty) const {
if (const BuiltinType *BT = Ty->getAs<BuiltinType>())
switch (BT->getKind()) {
case BuiltinType::Float:
case BuiltinType::Double:
return true;
default:
return false;
}
if (const RecordType *RT = Ty->getAsStructureType()) {
const RecordDecl *RD = RT->getDecl();
bool Found = false;
// If this is a C++ record, check the bases first.
if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD))
for (CXXRecordDecl::base_class_const_iterator I = CXXRD->bases_begin(),
E = CXXRD->bases_end(); I != E; ++I) {
QualType Base = I->getType();
// Empty bases don't affect things either way.
if (isEmptyRecord(getContext(), Base, true))
continue;
if (Found)
return false;
Found = isFPArgumentType(Base);
if (!Found)
return false;
}
// Check the fields.
for (RecordDecl::field_iterator I = RD->field_begin(),
E = RD->field_end(); I != E; ++I) {
const FieldDecl *FD = *I;
// Empty bitfields don't affect things either way.
// Unlike isSingleElementStruct(), empty structure and array fields
// do count. So do anonymous bitfields that aren't zero-sized.
if (FD->isBitField() && FD->getBitWidthValue(getContext()) == 0)
return true;
// Unlike isSingleElementStruct(), arrays do not count.
// Nested isFPArgumentType structures still do though.
if (Found)
return false;
Found = isFPArgumentType(FD->getType());
if (!Found)
return false;
}
// Unlike isSingleElementStruct(), trailing padding is allowed.
// An 8-byte aligned struct s { float f; } is passed as a double.
return Found;
}
return false;
}
llvm::Value *SystemZABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
// Assume that va_list type is correct; should be pointer to LLVM type:
// struct {
// i64 __gpr;
// i64 __fpr;
// i8 *__overflow_arg_area;
// i8 *__reg_save_area;
// };
// Every argument occupies 8 bytes and is passed by preference in either
// GPRs or FPRs.
Ty = CGF.getContext().getCanonicalType(Ty);
ABIArgInfo AI = classifyArgumentType(Ty);
bool InFPRs = isFPArgumentType(Ty);
llvm::Type *APTy = llvm::PointerType::getUnqual(CGF.ConvertTypeForMem(Ty));
bool IsIndirect = AI.isIndirect();
unsigned UnpaddedBitSize;
if (IsIndirect) {
APTy = llvm::PointerType::getUnqual(APTy);
UnpaddedBitSize = 64;
} else
UnpaddedBitSize = getContext().getTypeSize(Ty);
unsigned PaddedBitSize = 64;
assert((UnpaddedBitSize <= PaddedBitSize) && "Invalid argument size.");
unsigned PaddedSize = PaddedBitSize / 8;
unsigned Padding = (PaddedBitSize - UnpaddedBitSize) / 8;
unsigned MaxRegs, RegCountField, RegSaveIndex, RegPadding;
if (InFPRs) {
MaxRegs = 4; // Maximum of 4 FPR arguments
RegCountField = 1; // __fpr
RegSaveIndex = 16; // save offset for f0
RegPadding = 0; // floats are passed in the high bits of an FPR
} else {
MaxRegs = 5; // Maximum of 5 GPR arguments
RegCountField = 0; // __gpr
RegSaveIndex = 2; // save offset for r2
RegPadding = Padding; // values are passed in the low bits of a GPR
}
llvm::Value *RegCountPtr =
CGF.Builder.CreateStructGEP(VAListAddr, RegCountField, "reg_count_ptr");
llvm::Value *RegCount = CGF.Builder.CreateLoad(RegCountPtr, "reg_count");
llvm::Type *IndexTy = RegCount->getType();
llvm::Value *MaxRegsV = llvm::ConstantInt::get(IndexTy, MaxRegs);
llvm::Value *InRegs = CGF.Builder.CreateICmpULT(RegCount, MaxRegsV,
"fits_in_regs");
llvm::BasicBlock *InRegBlock = CGF.createBasicBlock("vaarg.in_reg");
llvm::BasicBlock *InMemBlock = CGF.createBasicBlock("vaarg.in_mem");
llvm::BasicBlock *ContBlock = CGF.createBasicBlock("vaarg.end");
CGF.Builder.CreateCondBr(InRegs, InRegBlock, InMemBlock);
// Emit code to load the value if it was passed in registers.
CGF.EmitBlock(InRegBlock);
// Work out the address of an argument register.
llvm::Value *PaddedSizeV = llvm::ConstantInt::get(IndexTy, PaddedSize);
llvm::Value *ScaledRegCount =
CGF.Builder.CreateMul(RegCount, PaddedSizeV, "scaled_reg_count");
llvm::Value *RegBase =
llvm::ConstantInt::get(IndexTy, RegSaveIndex * PaddedSize + RegPadding);
llvm::Value *RegOffset =
CGF.Builder.CreateAdd(ScaledRegCount, RegBase, "reg_offset");
llvm::Value *RegSaveAreaPtr =
CGF.Builder.CreateStructGEP(VAListAddr, 3, "reg_save_area_ptr");
llvm::Value *RegSaveArea =
CGF.Builder.CreateLoad(RegSaveAreaPtr, "reg_save_area");
llvm::Value *RawRegAddr =
CGF.Builder.CreateGEP(RegSaveArea, RegOffset, "raw_reg_addr");
llvm::Value *RegAddr =
CGF.Builder.CreateBitCast(RawRegAddr, APTy, "reg_addr");
// Update the register count
llvm::Value *One = llvm::ConstantInt::get(IndexTy, 1);
llvm::Value *NewRegCount =
CGF.Builder.CreateAdd(RegCount, One, "reg_count");
CGF.Builder.CreateStore(NewRegCount, RegCountPtr);
CGF.EmitBranch(ContBlock);
// Emit code to load the value if it was passed in memory.
CGF.EmitBlock(InMemBlock);
// Work out the address of a stack argument.
llvm::Value *OverflowArgAreaPtr =
CGF.Builder.CreateStructGEP(VAListAddr, 2, "overflow_arg_area_ptr");
llvm::Value *OverflowArgArea =
CGF.Builder.CreateLoad(OverflowArgAreaPtr, "overflow_arg_area");
llvm::Value *PaddingV = llvm::ConstantInt::get(IndexTy, Padding);
llvm::Value *RawMemAddr =
CGF.Builder.CreateGEP(OverflowArgArea, PaddingV, "raw_mem_addr");
llvm::Value *MemAddr =
CGF.Builder.CreateBitCast(RawMemAddr, APTy, "mem_addr");
// Update overflow_arg_area_ptr pointer
llvm::Value *NewOverflowArgArea =
CGF.Builder.CreateGEP(OverflowArgArea, PaddedSizeV, "overflow_arg_area");
CGF.Builder.CreateStore(NewOverflowArgArea, OverflowArgAreaPtr);
CGF.EmitBranch(ContBlock);
// Return the appropriate result.
CGF.EmitBlock(ContBlock);
llvm::PHINode *ResAddr = CGF.Builder.CreatePHI(APTy, 2, "va_arg.addr");
ResAddr->addIncoming(RegAddr, InRegBlock);
ResAddr->addIncoming(MemAddr, InMemBlock);
if (IsIndirect)
return CGF.Builder.CreateLoad(ResAddr, "indirect_arg");
return ResAddr;
}
ABIArgInfo SystemZABIInfo::classifyReturnType(QualType RetTy) const {
if (RetTy->isVoidType())
return ABIArgInfo::getIgnore();
if (isCompoundType(RetTy) || getContext().getTypeSize(RetTy) > 64)
return ABIArgInfo::getIndirect(0);
return (isPromotableIntegerType(RetTy) ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
ABIArgInfo SystemZABIInfo::classifyArgumentType(QualType Ty) const {
// Handle the generic C++ ABI.
if (CGCXXABI::RecordArgABI RAA = getRecordArgABI(Ty, CGT))
return ABIArgInfo::getIndirect(0, RAA == CGCXXABI::RAA_DirectInMemory);
// Integers and enums are extended to full register width.
if (isPromotableIntegerType(Ty))
return ABIArgInfo::getExtend();
// Values that are not 1, 2, 4 or 8 bytes in size are passed indirectly.
uint64_t Size = getContext().getTypeSize(Ty);
if (Size != 8 && Size != 16 && Size != 32 && Size != 64)
return ABIArgInfo::getIndirect(0);
// Handle small structures.
if (const RecordType *RT = Ty->getAs<RecordType>()) {
// Structures with flexible arrays have variable length, so really
// fail the size test above.
const RecordDecl *RD = RT->getDecl();
if (RD->hasFlexibleArrayMember())
return ABIArgInfo::getIndirect(0);
// The structure is passed as an unextended integer, a float, or a double.
llvm::Type *PassTy;
if (isFPArgumentType(Ty)) {
assert(Size == 32 || Size == 64);
if (Size == 32)
PassTy = llvm::Type::getFloatTy(getVMContext());
else
PassTy = llvm::Type::getDoubleTy(getVMContext());
} else
PassTy = llvm::IntegerType::get(getVMContext(), Size);
return ABIArgInfo::getDirect(PassTy);
}
// Non-structure compounds are passed indirectly.
if (isCompoundType(Ty))
return ABIArgInfo::getIndirect(0);
return ABIArgInfo::getDirect(0);
}
//===----------------------------------------------------------------------===//
// MBlaze ABI Implementation
//===----------------------------------------------------------------------===//
namespace {
class MBlazeABIInfo : public ABIInfo {
public:
MBlazeABIInfo(CodeGenTypes &CGT) : ABIInfo(CGT) {}
bool isPromotableIntegerType(QualType Ty) const;
ABIArgInfo classifyReturnType(QualType RetTy) const;
ABIArgInfo classifyArgumentType(QualType RetTy) const;
virtual void computeInfo(CGFunctionInfo &FI) const {
FI.getReturnInfo() = classifyReturnType(FI.getReturnType());
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it)
it->info = classifyArgumentType(it->type);
}
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
};
class MBlazeTargetCodeGenInfo : public TargetCodeGenInfo {
public:
MBlazeTargetCodeGenInfo(CodeGenTypes &CGT)
: TargetCodeGenInfo(new MBlazeABIInfo(CGT)) {}
void SetTargetAttributes(const Decl *D, llvm::GlobalValue *GV,
CodeGen::CodeGenModule &M) const;
};
}
bool MBlazeABIInfo::isPromotableIntegerType(QualType Ty) const {
// MBlaze ABI requires all 8 and 16 bit quantities to be extended.
if (const BuiltinType *BT = Ty->getAs<BuiltinType>())
switch (BT->getKind()) {
case BuiltinType::Bool:
case BuiltinType::Char_S:
case BuiltinType::Char_U:
case BuiltinType::SChar:
case BuiltinType::UChar:
case BuiltinType::Short:
case BuiltinType::UShort:
return true;
default:
return false;
}
return false;
}
llvm::Value *MBlazeABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
// FIXME: Implement
return 0;
}
ABIArgInfo MBlazeABIInfo::classifyReturnType(QualType RetTy) const {
if (RetTy->isVoidType())
return ABIArgInfo::getIgnore();
if (isAggregateTypeForABI(RetTy))
return ABIArgInfo::getIndirect(0);
return (isPromotableIntegerType(RetTy) ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
ABIArgInfo MBlazeABIInfo::classifyArgumentType(QualType Ty) const {
if (isAggregateTypeForABI(Ty))
return ABIArgInfo::getIndirect(0);
return (isPromotableIntegerType(Ty) ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
void MBlazeTargetCodeGenInfo::SetTargetAttributes(const Decl *D,
llvm::GlobalValue *GV,
CodeGen::CodeGenModule &M)
const {
const FunctionDecl *FD = dyn_cast<FunctionDecl>(D);
if (!FD) return;
llvm::CallingConv::ID CC = llvm::CallingConv::C;
if (FD->hasAttr<MBlazeInterruptHandlerAttr>())
CC = llvm::CallingConv::MBLAZE_INTR;
else if (FD->hasAttr<MBlazeSaveVolatilesAttr>())
CC = llvm::CallingConv::MBLAZE_SVOL;
if (CC != llvm::CallingConv::C) {
// Handle 'interrupt_handler' attribute:
llvm::Function *F = cast<llvm::Function>(GV);
// Step 1: Set ISR calling convention.
F->setCallingConv(CC);
// Step 2: Add attributes goodness.
F->addFnAttr(llvm::Attribute::NoInline);
}
// Step 3: Emit _interrupt_handler alias.
if (CC == llvm::CallingConv::MBLAZE_INTR)
new llvm::GlobalAlias(GV->getType(), llvm::Function::ExternalLinkage,
"_interrupt_handler", GV, &M.getModule());
}
//===----------------------------------------------------------------------===//
// MSP430 ABI Implementation
//===----------------------------------------------------------------------===//
namespace {
class MSP430TargetCodeGenInfo : public TargetCodeGenInfo {
public:
MSP430TargetCodeGenInfo(CodeGenTypes &CGT)
: TargetCodeGenInfo(new DefaultABIInfo(CGT)) {}
void SetTargetAttributes(const Decl *D, llvm::GlobalValue *GV,
CodeGen::CodeGenModule &M) const;
};
}
void MSP430TargetCodeGenInfo::SetTargetAttributes(const Decl *D,
llvm::GlobalValue *GV,
CodeGen::CodeGenModule &M) const {
if (const FunctionDecl *FD = dyn_cast<FunctionDecl>(D)) {
if (const MSP430InterruptAttr *attr = FD->getAttr<MSP430InterruptAttr>()) {
// Handle 'interrupt' attribute:
llvm::Function *F = cast<llvm::Function>(GV);
// Step 1: Set ISR calling convention.
F->setCallingConv(llvm::CallingConv::MSP430_INTR);
// Step 2: Add attributes goodness.
F->addFnAttr(llvm::Attribute::NoInline);
// Step 3: Emit ISR vector alias.
unsigned Num = attr->getNumber() / 2;
new llvm::GlobalAlias(GV->getType(), llvm::Function::ExternalLinkage,
"__isr_" + Twine(Num),
GV, &M.getModule());
}
}
}
//===----------------------------------------------------------------------===//
// MIPS ABI Implementation. This works for both little-endian and
// big-endian variants.
//===----------------------------------------------------------------------===//
namespace {
class MipsABIInfo : public ABIInfo {
bool IsO32;
unsigned MinABIStackAlignInBytes, StackAlignInBytes;
void CoerceToIntArgs(uint64_t TySize,
SmallVector<llvm::Type*, 8> &ArgList) const;
llvm::Type* HandleAggregates(QualType Ty, uint64_t TySize) const;
llvm::Type* returnAggregateInRegs(QualType RetTy, uint64_t Size) const;
llvm::Type* getPaddingType(uint64_t Align, uint64_t Offset) const;
public:
MipsABIInfo(CodeGenTypes &CGT, bool _IsO32) :
ABIInfo(CGT), IsO32(_IsO32), MinABIStackAlignInBytes(IsO32 ? 4 : 8),
StackAlignInBytes(IsO32 ? 8 : 16) {}
ABIArgInfo classifyReturnType(QualType RetTy) const;
ABIArgInfo classifyArgumentType(QualType RetTy, uint64_t &Offset) const;
virtual void computeInfo(CGFunctionInfo &FI) const;
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
};
class MIPSTargetCodeGenInfo : public TargetCodeGenInfo {
unsigned SizeOfUnwindException;
public:
MIPSTargetCodeGenInfo(CodeGenTypes &CGT, bool IsO32)
: TargetCodeGenInfo(new MipsABIInfo(CGT, IsO32)),
SizeOfUnwindException(IsO32 ? 24 : 32) {}
int getDwarfEHStackPointer(CodeGen::CodeGenModule &CGM) const {
return 29;
}
void SetTargetAttributes(const Decl *D, llvm::GlobalValue *GV,
CodeGen::CodeGenModule &CGM) const {
const FunctionDecl *FD = dyn_cast<FunctionDecl>(D);
if (!FD) return;
llvm::Function *Fn = cast<llvm::Function>(GV);
if (FD->hasAttr<Mips16Attr>()) {
Fn->addFnAttr("mips16");
}
else if (FD->hasAttr<NoMips16Attr>()) {
Fn->addFnAttr("nomips16");
}
}
bool initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const;
unsigned getSizeOfUnwindException() const {
return SizeOfUnwindException;
}
};
}
void MipsABIInfo::CoerceToIntArgs(uint64_t TySize,
SmallVector<llvm::Type*, 8> &ArgList) const {
llvm::IntegerType *IntTy =
llvm::IntegerType::get(getVMContext(), MinABIStackAlignInBytes * 8);
// Add (TySize / MinABIStackAlignInBytes) args of IntTy.
for (unsigned N = TySize / (MinABIStackAlignInBytes * 8); N; --N)
ArgList.push_back(IntTy);
// If necessary, add one more integer type to ArgList.
unsigned R = TySize % (MinABIStackAlignInBytes * 8);
if (R)
ArgList.push_back(llvm::IntegerType::get(getVMContext(), R));
}
// In N32/64, an aligned double precision floating point field is passed in
// a register.
llvm::Type* MipsABIInfo::HandleAggregates(QualType Ty, uint64_t TySize) const {
SmallVector<llvm::Type*, 8> ArgList, IntArgList;
if (IsO32) {
CoerceToIntArgs(TySize, ArgList);
return llvm::StructType::get(getVMContext(), ArgList);
}
if (Ty->isComplexType())
return CGT.ConvertType(Ty);
const RecordType *RT = Ty->getAs<RecordType>();
// Unions/vectors are passed in integer registers.
if (!RT || !RT->isStructureOrClassType()) {
CoerceToIntArgs(TySize, ArgList);
return llvm::StructType::get(getVMContext(), ArgList);
}
const RecordDecl *RD = RT->getDecl();
const ASTRecordLayout &Layout = getContext().getASTRecordLayout(RD);
assert(!(TySize % 8) && "Size of structure must be multiple of 8.");
uint64_t LastOffset = 0;
unsigned idx = 0;
llvm::IntegerType *I64 = llvm::IntegerType::get(getVMContext(), 64);
// Iterate over fields in the struct/class and check if there are any aligned
// double fields.
for (RecordDecl::field_iterator i = RD->field_begin(), e = RD->field_end();
i != e; ++i, ++idx) {
const QualType Ty = i->getType();
const BuiltinType *BT = Ty->getAs<BuiltinType>();
if (!BT || BT->getKind() != BuiltinType::Double)
continue;
uint64_t Offset = Layout.getFieldOffset(idx);
if (Offset % 64) // Ignore doubles that are not aligned.
continue;
// Add ((Offset - LastOffset) / 64) args of type i64.
for (unsigned j = (Offset - LastOffset) / 64; j > 0; --j)
ArgList.push_back(I64);
// Add double type.
ArgList.push_back(llvm::Type::getDoubleTy(getVMContext()));
LastOffset = Offset + 64;
}
CoerceToIntArgs(TySize - LastOffset, IntArgList);
ArgList.append(IntArgList.begin(), IntArgList.end());
return llvm::StructType::get(getVMContext(), ArgList);
}
llvm::Type *MipsABIInfo::getPaddingType(uint64_t Align, uint64_t Offset) const {
assert((Offset % MinABIStackAlignInBytes) == 0);
if ((Align - 1) & Offset)
return llvm::IntegerType::get(getVMContext(), MinABIStackAlignInBytes * 8);
return 0;
}
ABIArgInfo
MipsABIInfo::classifyArgumentType(QualType Ty, uint64_t &Offset) const {
uint64_t OrigOffset = Offset;
uint64_t TySize = getContext().getTypeSize(Ty);
uint64_t Align = getContext().getTypeAlign(Ty) / 8;
Align = std::min(std::max(Align, (uint64_t)MinABIStackAlignInBytes),
(uint64_t)StackAlignInBytes);
Offset = llvm::RoundUpToAlignment(Offset, Align);
Offset += llvm::RoundUpToAlignment(TySize, Align * 8) / 8;
if (isAggregateTypeForABI(Ty) || Ty->isVectorType()) {
// Ignore empty aggregates.
if (TySize == 0)
return ABIArgInfo::getIgnore();
if (CGCXXABI::RecordArgABI RAA = getRecordArgABI(Ty, CGT)) {
Offset = OrigOffset + MinABIStackAlignInBytes;
return ABIArgInfo::getIndirect(0, RAA == CGCXXABI::RAA_DirectInMemory);
}
// If we have reached here, aggregates are passed directly by coercing to
// another structure type. Padding is inserted if the offset of the
// aggregate is unaligned.
return ABIArgInfo::getDirect(HandleAggregates(Ty, TySize), 0,
getPaddingType(Align, OrigOffset));
}
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
if (Ty->isPromotableIntegerType())
return ABIArgInfo::getExtend();
return ABIArgInfo::getDirect(0, 0,
IsO32 ? 0 : getPaddingType(Align, OrigOffset));
}
llvm::Type*
MipsABIInfo::returnAggregateInRegs(QualType RetTy, uint64_t Size) const {
const RecordType *RT = RetTy->getAs<RecordType>();
SmallVector<llvm::Type*, 8> RTList;
if (RT && RT->isStructureOrClassType()) {
const RecordDecl *RD = RT->getDecl();
const ASTRecordLayout &Layout = getContext().getASTRecordLayout(RD);
unsigned FieldCnt = Layout.getFieldCount();
// N32/64 returns struct/classes in floating point registers if the
// following conditions are met:
// 1. The size of the struct/class is no larger than 128-bit.
// 2. The struct/class has one or two fields all of which are floating
// point types.
// 3. The offset of the first field is zero (this follows what gcc does).
//
// Any other composite results are returned in integer registers.
//
if (FieldCnt && (FieldCnt <= 2) && !Layout.getFieldOffset(0)) {
RecordDecl::field_iterator b = RD->field_begin(), e = RD->field_end();
for (; b != e; ++b) {
const BuiltinType *BT = b->getType()->getAs<BuiltinType>();
if (!BT || !BT->isFloatingPoint())
break;
RTList.push_back(CGT.ConvertType(b->getType()));
}
if (b == e)
return llvm::StructType::get(getVMContext(), RTList,
RD->hasAttr<PackedAttr>());
RTList.clear();
}
}
CoerceToIntArgs(Size, RTList);
return llvm::StructType::get(getVMContext(), RTList);
}
ABIArgInfo MipsABIInfo::classifyReturnType(QualType RetTy) const {
uint64_t Size = getContext().getTypeSize(RetTy);
if (RetTy->isVoidType() || Size == 0)
return ABIArgInfo::getIgnore();
if (isAggregateTypeForABI(RetTy) || RetTy->isVectorType()) {
if (isRecordReturnIndirect(RetTy, CGT))
return ABIArgInfo::getIndirect(0);
if (Size <= 128) {
if (RetTy->isAnyComplexType())
return ABIArgInfo::getDirect();
// O32 returns integer vectors in registers.
if (IsO32 && RetTy->isVectorType() && !RetTy->hasFloatingRepresentation())
return ABIArgInfo::getDirect(returnAggregateInRegs(RetTy, Size));
if (!IsO32)
return ABIArgInfo::getDirect(returnAggregateInRegs(RetTy, Size));
}
return ABIArgInfo::getIndirect(0);
}
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = RetTy->getAs<EnumType>())
RetTy = EnumTy->getDecl()->getIntegerType();
return (RetTy->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
void MipsABIInfo::computeInfo(CGFunctionInfo &FI) const {
ABIArgInfo &RetInfo = FI.getReturnInfo();
RetInfo = classifyReturnType(FI.getReturnType());
// Check if a pointer to an aggregate is passed as a hidden argument.
uint64_t Offset = RetInfo.isIndirect() ? MinABIStackAlignInBytes : 0;
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it)
it->info = classifyArgumentType(it->type, Offset);
}
llvm::Value* MipsABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
llvm::Type *BP = CGF.Int8PtrTy;
llvm::Type *BPP = CGF.Int8PtrPtrTy;
CGBuilderTy &Builder = CGF.Builder;
llvm::Value *VAListAddrAsBPP = Builder.CreateBitCast(VAListAddr, BPP, "ap");
llvm::Value *Addr = Builder.CreateLoad(VAListAddrAsBPP, "ap.cur");
int64_t TypeAlign = getContext().getTypeAlign(Ty) / 8;
llvm::Type *PTy = llvm::PointerType::getUnqual(CGF.ConvertType(Ty));
llvm::Value *AddrTyped;
unsigned PtrWidth = getTarget().getPointerWidth(0);
llvm::IntegerType *IntTy = (PtrWidth == 32) ? CGF.Int32Ty : CGF.Int64Ty;
if (TypeAlign > MinABIStackAlignInBytes) {
llvm::Value *AddrAsInt = CGF.Builder.CreatePtrToInt(Addr, IntTy);
llvm::Value *Inc = llvm::ConstantInt::get(IntTy, TypeAlign - 1);
llvm::Value *Mask = llvm::ConstantInt::get(IntTy, -TypeAlign);
llvm::Value *Add = CGF.Builder.CreateAdd(AddrAsInt, Inc);
llvm::Value *And = CGF.Builder.CreateAnd(Add, Mask);
AddrTyped = CGF.Builder.CreateIntToPtr(And, PTy);
}
else
AddrTyped = Builder.CreateBitCast(Addr, PTy);
llvm::Value *AlignedAddr = Builder.CreateBitCast(AddrTyped, BP);
TypeAlign = std::max((unsigned)TypeAlign, MinABIStackAlignInBytes);
uint64_t Offset =
llvm::RoundUpToAlignment(CGF.getContext().getTypeSize(Ty) / 8, TypeAlign);
llvm::Value *NextAddr =
Builder.CreateGEP(AlignedAddr, llvm::ConstantInt::get(IntTy, Offset),
"ap.next");
Builder.CreateStore(NextAddr, VAListAddrAsBPP);
return AddrTyped;
}
bool
MIPSTargetCodeGenInfo::initDwarfEHRegSizeTable(CodeGen::CodeGenFunction &CGF,
llvm::Value *Address) const {
// This information comes from gcc's implementation, which seems to
// as canonical as it gets.
// Everything on MIPS is 4 bytes. Double-precision FP registers
// are aliased to pairs of single-precision FP registers.
llvm::Value *Four8 = llvm::ConstantInt::get(CGF.Int8Ty, 4);
// 0-31 are the general purpose registers, $0 - $31.
// 32-63 are the floating-point registers, $f0 - $f31.
// 64 and 65 are the multiply/divide registers, $hi and $lo.
// 66 is the (notional, I think) register for signal-handler return.
AssignToArrayRange(CGF.Builder, Address, Four8, 0, 65);
// 67-74 are the floating-point status registers, $fcc0 - $fcc7.
// They are one bit wide and ignored here.
// 80-111 are the coprocessor 0 registers, $c0r0 - $c0r31.
// (coprocessor 1 is the FP unit)
// 112-143 are the coprocessor 2 registers, $c2r0 - $c2r31.
// 144-175 are the coprocessor 3 registers, $c3r0 - $c3r31.
// 176-181 are the DSP accumulator registers.
AssignToArrayRange(CGF.Builder, Address, Four8, 80, 181);
return false;
}
//===----------------------------------------------------------------------===//
// TCE ABI Implementation (see http://tce.cs.tut.fi). Uses mostly the defaults.
// Currently subclassed only to implement custom OpenCL C function attribute
// handling.
//===----------------------------------------------------------------------===//
namespace {
class TCETargetCodeGenInfo : public DefaultTargetCodeGenInfo {
public:
TCETargetCodeGenInfo(CodeGenTypes &CGT)
: DefaultTargetCodeGenInfo(CGT) {}
virtual void SetTargetAttributes(const Decl *D, llvm::GlobalValue *GV,
CodeGen::CodeGenModule &M) const;
};
void TCETargetCodeGenInfo::SetTargetAttributes(const Decl *D,
llvm::GlobalValue *GV,
CodeGen::CodeGenModule &M) const {
const FunctionDecl *FD = dyn_cast<FunctionDecl>(D);
if (!FD) return;
llvm::Function *F = cast<llvm::Function>(GV);
if (M.getLangOpts().OpenCL) {
if (FD->hasAttr<OpenCLKernelAttr>()) {
// OpenCL C Kernel functions are not subject to inlining
F->addFnAttr(llvm::Attribute::NoInline);
if (FD->hasAttr<ReqdWorkGroupSizeAttr>()) {
// Convert the reqd_work_group_size() attributes to metadata.
llvm::LLVMContext &Context = F->getContext();
llvm::NamedMDNode *OpenCLMetadata =
M.getModule().getOrInsertNamedMetadata("opencl.kernel_wg_size_info");
SmallVector<llvm::Value*, 5> Operands;
Operands.push_back(F);
Operands.push_back(llvm::Constant::getIntegerValue(M.Int32Ty,
llvm::APInt(32,
FD->getAttr<ReqdWorkGroupSizeAttr>()->getXDim())));
Operands.push_back(llvm::Constant::getIntegerValue(M.Int32Ty,
llvm::APInt(32,
FD->getAttr<ReqdWorkGroupSizeAttr>()->getYDim())));
Operands.push_back(llvm::Constant::getIntegerValue(M.Int32Ty,
llvm::APInt(32,
FD->getAttr<ReqdWorkGroupSizeAttr>()->getZDim())));
// Add a boolean constant operand for "required" (true) or "hint" (false)
// for implementing the work_group_size_hint attr later. Currently
// always true as the hint is not yet implemented.
Operands.push_back(llvm::ConstantInt::getTrue(Context));
OpenCLMetadata->addOperand(llvm::MDNode::get(Context, Operands));
}
}
}
}
}
//===----------------------------------------------------------------------===//
// Hexagon ABI Implementation
//===----------------------------------------------------------------------===//
namespace {
class HexagonABIInfo : public ABIInfo {
public:
HexagonABIInfo(CodeGenTypes &CGT) : ABIInfo(CGT) {}
private:
ABIArgInfo classifyReturnType(QualType RetTy) const;
ABIArgInfo classifyArgumentType(QualType RetTy) const;
virtual void computeInfo(CGFunctionInfo &FI) const;
virtual llvm::Value *EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const;
};
class HexagonTargetCodeGenInfo : public TargetCodeGenInfo {
public:
HexagonTargetCodeGenInfo(CodeGenTypes &CGT)
:TargetCodeGenInfo(new HexagonABIInfo(CGT)) {}
int getDwarfEHStackPointer(CodeGen::CodeGenModule &M) const {
return 29;
}
};
}
void HexagonABIInfo::computeInfo(CGFunctionInfo &FI) const {
FI.getReturnInfo() = classifyReturnType(FI.getReturnType());
for (CGFunctionInfo::arg_iterator it = FI.arg_begin(), ie = FI.arg_end();
it != ie; ++it)
it->info = classifyArgumentType(it->type);
}
ABIArgInfo HexagonABIInfo::classifyArgumentType(QualType Ty) const {
if (!isAggregateTypeForABI(Ty)) {
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = Ty->getAs<EnumType>())
Ty = EnumTy->getDecl()->getIntegerType();
return (Ty->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
// Ignore empty records.
if (isEmptyRecord(getContext(), Ty, true))
return ABIArgInfo::getIgnore();
if (CGCXXABI::RecordArgABI RAA = getRecordArgABI(Ty, CGT))
return ABIArgInfo::getIndirect(0, RAA == CGCXXABI::RAA_DirectInMemory);
uint64_t Size = getContext().getTypeSize(Ty);
if (Size > 64)
return ABIArgInfo::getIndirect(0, /*ByVal=*/true);
// Pass in the smallest viable integer type.
else if (Size > 32)
return ABIArgInfo::getDirect(llvm::Type::getInt64Ty(getVMContext()));
else if (Size > 16)
return ABIArgInfo::getDirect(llvm::Type::getInt32Ty(getVMContext()));
else if (Size > 8)
return ABIArgInfo::getDirect(llvm::Type::getInt16Ty(getVMContext()));
else
return ABIArgInfo::getDirect(llvm::Type::getInt8Ty(getVMContext()));
}
ABIArgInfo HexagonABIInfo::classifyReturnType(QualType RetTy) const {
if (RetTy->isVoidType())
return ABIArgInfo::getIgnore();
// Large vector types should be returned via memory.
if (RetTy->isVectorType() && getContext().getTypeSize(RetTy) > 64)
return ABIArgInfo::getIndirect(0);
if (!isAggregateTypeForABI(RetTy)) {
// Treat an enum type as its underlying type.
if (const EnumType *EnumTy = RetTy->getAs<EnumType>())
RetTy = EnumTy->getDecl()->getIntegerType();
return (RetTy->isPromotableIntegerType() ?
ABIArgInfo::getExtend() : ABIArgInfo::getDirect());
}
// Structures with either a non-trivial destructor or a non-trivial
// copy constructor are always indirect.
if (isRecordReturnIndirect(RetTy, CGT))
return ABIArgInfo::getIndirect(0, /*ByVal=*/false);
if (isEmptyRecord(getContext(), RetTy, true))
return ABIArgInfo::getIgnore();
// Aggregates <= 8 bytes are returned in r0; other aggregates
// are returned indirectly.
uint64_t Size = getContext().getTypeSize(RetTy);
if (Size <= 64) {
// Return in the smallest viable integer type.
if (Size <= 8)
return ABIArgInfo::getDirect(llvm::Type::getInt8Ty(getVMContext()));
if (Size <= 16)
return ABIArgInfo::getDirect(llvm::Type::getInt16Ty(getVMContext()));
if (Size <= 32)
return ABIArgInfo::getDirect(llvm::Type::getInt32Ty(getVMContext()));
return ABIArgInfo::getDirect(llvm::Type::getInt64Ty(getVMContext()));
}
return ABIArgInfo::getIndirect(0, /*ByVal=*/true);
}
llvm::Value *HexagonABIInfo::EmitVAArg(llvm::Value *VAListAddr, QualType Ty,
CodeGenFunction &CGF) const {
// FIXME: Need to handle alignment
llvm::Type *BPP = CGF.Int8PtrPtrTy;
CGBuilderTy &Builder = CGF.Builder;
llvm::Value *VAListAddrAsBPP = Builder.CreateBitCast(VAListAddr, BPP,
"ap");
llvm::Value *Addr = Builder.CreateLoad(VAListAddrAsBPP, "ap.cur");
llvm::Type *PTy =
llvm::PointerType::getUnqual(CGF.ConvertType(Ty));
llvm::Value *AddrTyped = Builder.CreateBitCast(Addr, PTy);
uint64_t Offset =
llvm::RoundUpToAlignment(CGF.getContext().getTypeSize(Ty) / 8, 4);
llvm::Value *NextAddr =
Builder.CreateGEP(Addr, llvm::ConstantInt::get(CGF.Int32Ty, Offset),
"ap.next");
Builder.CreateStore(NextAddr, VAListAddrAsBPP);
return AddrTyped;
}
const TargetCodeGenInfo &CodeGenModule::getTargetCodeGenInfo() {
if (TheTargetCodeGenInfo)
return *TheTargetCodeGenInfo;
const llvm::Triple &Triple = getTarget().getTriple();
switch (Triple.getArch()) {
default:
return *(TheTargetCodeGenInfo = new DefaultTargetCodeGenInfo(Types));
case llvm::Triple::le32:
return *(TheTargetCodeGenInfo = new PNaClTargetCodeGenInfo(Types));
case llvm::Triple::mips:
case llvm::Triple::mipsel:
return *(TheTargetCodeGenInfo = new MIPSTargetCodeGenInfo(Types, true));
case llvm::Triple::mips64:
case llvm::Triple::mips64el:
return *(TheTargetCodeGenInfo = new MIPSTargetCodeGenInfo(Types, false));
case llvm::Triple::aarch64:
return *(TheTargetCodeGenInfo = new AArch64TargetCodeGenInfo(Types));
case llvm::Triple::arm:
case llvm::Triple::thumb:
{
ARMABIInfo::ABIKind Kind = ARMABIInfo::AAPCS;
if (strcmp(getTarget().getABI(), "apcs-gnu") == 0)
Kind = ARMABIInfo::APCS;
else if (CodeGenOpts.FloatABI == "hard" ||
(CodeGenOpts.FloatABI != "soft" &&
Triple.getEnvironment() == llvm::Triple::GNUEABIHF))
Kind = ARMABIInfo::AAPCS_VFP;
switch (Triple.getOS()) {
case llvm::Triple::NaCl:
return *(TheTargetCodeGenInfo =
new NaClARMTargetCodeGenInfo(Types, Kind));
default:
return *(TheTargetCodeGenInfo =
new ARMTargetCodeGenInfo(Types, Kind));
}
}
case llvm::Triple::ppc:
return *(TheTargetCodeGenInfo = new PPC32TargetCodeGenInfo(Types));
case llvm::Triple::ppc64:
if (Triple.isOSBinFormatELF())
return *(TheTargetCodeGenInfo = new PPC64_SVR4_TargetCodeGenInfo(Types));
else
return *(TheTargetCodeGenInfo = new PPC64TargetCodeGenInfo(Types));
case llvm::Triple::nvptx:
case llvm::Triple::nvptx64:
return *(TheTargetCodeGenInfo = new NVPTXTargetCodeGenInfo(Types));
case llvm::Triple::mblaze:
return *(TheTargetCodeGenInfo = new MBlazeTargetCodeGenInfo(Types));
case llvm::Triple::msp430:
return *(TheTargetCodeGenInfo = new MSP430TargetCodeGenInfo(Types));
case llvm::Triple::systemz:
return *(TheTargetCodeGenInfo = new SystemZTargetCodeGenInfo(Types));
case llvm::Triple::tce:
return *(TheTargetCodeGenInfo = new TCETargetCodeGenInfo(Types));
case llvm::Triple::x86: {
if (Triple.isOSDarwin())
return *(TheTargetCodeGenInfo =
new X86_32TargetCodeGenInfo(Types, true, true, false,
CodeGenOpts.NumRegisterParameters));
switch (Triple.getOS()) {
case llvm::Triple::Cygwin:
case llvm::Triple::MinGW32:
case llvm::Triple::AuroraUX:
case llvm::Triple::DragonFly:
case llvm::Triple::FreeBSD:
case llvm::Triple::OpenBSD:
case llvm::Triple::Bitrig:
return *(TheTargetCodeGenInfo =
new X86_32TargetCodeGenInfo(Types, false, true, false,
CodeGenOpts.NumRegisterParameters));
case llvm::Triple::Win32:
return *(TheTargetCodeGenInfo =
new X86_32TargetCodeGenInfo(Types, false, true, true,
CodeGenOpts.NumRegisterParameters));
default:
return *(TheTargetCodeGenInfo =
new X86_32TargetCodeGenInfo(Types, false, false, false,
CodeGenOpts.NumRegisterParameters));
}
}
case llvm::Triple::x86_64: {
bool HasAVX = strcmp(getTarget().getABI(), "avx") == 0;
switch (Triple.getOS()) {
case llvm::Triple::Win32:
case llvm::Triple::MinGW32:
case llvm::Triple::Cygwin:
return *(TheTargetCodeGenInfo = new WinX86_64TargetCodeGenInfo(Types));
case llvm::Triple::NaCl:
return *(TheTargetCodeGenInfo = new NaClX86_64TargetCodeGenInfo(Types,
HasAVX));
default:
return *(TheTargetCodeGenInfo = new X86_64TargetCodeGenInfo(Types,
HasAVX));
}
}
case llvm::Triple::hexagon:
return *(TheTargetCodeGenInfo = new HexagonTargetCodeGenInfo(Types));
}
}
|