aboutsummaryrefslogtreecommitdiff
path: root/gcc/tree-parloops.c
blob: 6558dc110031b5ff5281a71f9a01859ae521045c (plain)
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
/* Loop autoparallelization.
   Copyright (C) 2006-2019 Free Software Foundation, Inc.
   Contributed by Sebastian Pop <pop@cri.ensmp.fr> 
   Zdenek Dvorak <dvorakz@suse.cz> and Razya Ladelsky <razya@il.ibm.com>.

This file is part of GCC.

GCC is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free
Software Foundation; either version 3, or (at your option) any later
version.

GCC is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
for more details.

You should have received a copy of the GNU General Public License
along with GCC; see the file COPYING3.  If not see
<http://www.gnu.org/licenses/>.  */

#include "config.h"
#include "system.h"
#include "coretypes.h"
#include "backend.h"
#include "tree.h"
#include "gimple.h"
#include "cfghooks.h"
#include "tree-pass.h"
#include "ssa.h"
#include "cgraph.h"
#include "gimple-pretty-print.h"
#include "fold-const.h"
#include "gimplify.h"
#include "gimple-iterator.h"
#include "gimplify-me.h"
#include "gimple-walk.h"
#include "stor-layout.h"
#include "tree-nested.h"
#include "tree-cfg.h"
#include "tree-ssa-loop-ivopts.h"
#include "tree-ssa-loop-manip.h"
#include "tree-ssa-loop-niter.h"
#include "tree-ssa-loop.h"
#include "tree-into-ssa.h"
#include "cfgloop.h"
#include "tree-scalar-evolution.h"
#include "langhooks.h"
#include "tree-vectorizer.h"
#include "tree-hasher.h"
#include "tree-parloops.h"
#include "omp-general.h"
#include "omp-low.h"
#include "tree-ssa.h"
#include "tree-ssa-alias.h"
#include "tree-eh.h"
#include "gomp-constants.h"
#include "tree-dfa.h"
#include "stringpool.h"
#include "attribs.h"

/* This pass tries to distribute iterations of loops into several threads.
   The implementation is straightforward -- for each loop we test whether its
   iterations are independent, and if it is the case (and some additional
   conditions regarding profitability and correctness are satisfied), we
   add GIMPLE_OMP_PARALLEL and GIMPLE_OMP_FOR codes and let omp expansion
   machinery do its job.

   The most of the complexity is in bringing the code into shape expected
   by the omp expanders:
   -- for GIMPLE_OMP_FOR, ensuring that the loop has only one induction
      variable and that the exit test is at the start of the loop body
   -- for GIMPLE_OMP_PARALLEL, replacing the references to local addressable
      variables by accesses through pointers, and breaking up ssa chains
      by storing the values incoming to the parallelized loop to a structure
      passed to the new function as an argument (something similar is done
      in omp gimplification, unfortunately only a small part of the code
      can be shared).

   TODO:
   -- if there are several parallelizable loops in a function, it may be
      possible to generate the threads just once (using synchronization to
      ensure that cross-loop dependences are obeyed).
   -- handling of common reduction patterns for outer loops.  
    
   More info can also be found at http://gcc.gnu.org/wiki/AutoParInGCC  */
/*
  Reduction handling:
  currently we use code inspired by vect_force_simple_reduction to detect
  reduction patterns.
  The code transformation will be introduced by an example.


parloop
{
  int sum=1;

  for (i = 0; i < N; i++)
   {
    x[i] = i + 3;
    sum+=x[i];
   }
}

gimple-like code:
header_bb:

  # sum_29 = PHI <sum_11(5), 1(3)>
  # i_28 = PHI <i_12(5), 0(3)>
  D.1795_8 = i_28 + 3;
  x[i_28] = D.1795_8;
  sum_11 = D.1795_8 + sum_29;
  i_12 = i_28 + 1;
  if (N_6(D) > i_12)
    goto header_bb;


exit_bb:

  # sum_21 = PHI <sum_11(4)>
  printf (&"%d"[0], sum_21);


after reduction transformation (only relevant parts):

parloop
{

....


  # Storing the initial value given by the user.  #

  .paral_data_store.32.sum.27 = 1;

  #pragma omp parallel num_threads(4)

  #pragma omp for schedule(static)

  # The neutral element corresponding to the particular
  reduction's operation, e.g. 0 for PLUS_EXPR,
  1 for MULT_EXPR, etc. replaces the user's initial value.  #

  # sum.27_29 = PHI <sum.27_11, 0>

  sum.27_11 = D.1827_8 + sum.27_29;

  GIMPLE_OMP_CONTINUE

  # Adding this reduction phi is done at create_phi_for_local_result() #
  # sum.27_56 = PHI <sum.27_11, 0>
  GIMPLE_OMP_RETURN

  # Creating the atomic operation is done at
  create_call_for_reduction_1()  #

  #pragma omp atomic_load
  D.1839_59 = *&.paral_data_load.33_51->reduction.23;
  D.1840_60 = sum.27_56 + D.1839_59;
  #pragma omp atomic_store (D.1840_60);

  GIMPLE_OMP_RETURN

 # collecting the result after the join of the threads is done at
  create_loads_for_reductions().
  The value computed by the threads is loaded from the
  shared struct.  #


  .paral_data_load.33_52 = &.paral_data_store.32;
  sum_37 =  .paral_data_load.33_52->sum.27;
  sum_43 = D.1795_41 + sum_37;

  exit bb:
  # sum_21 = PHI <sum_43, sum_26>
  printf (&"%d"[0], sum_21);

...

}

*/

/* Error reporting helper for parloops_is_simple_reduction below.  GIMPLE
   statement STMT is printed with a message MSG. */

static void
report_ploop_op (dump_flags_t msg_type, gimple *stmt, const char *msg)
{
  dump_printf_loc (msg_type, vect_location, "%s%G", msg, stmt);
}

/* DEF_STMT_INFO occurs in a loop that contains a potential reduction
   operation.  Return true if the results of DEF_STMT_INFO are something
   that can be accumulated by such a reduction.  */

static bool
parloops_valid_reduction_input_p (stmt_vec_info def_stmt_info)
{
  return (is_gimple_assign (def_stmt_info->stmt)
	  || is_gimple_call (def_stmt_info->stmt)
	  || STMT_VINFO_DEF_TYPE (def_stmt_info) == vect_induction_def
	  || (gimple_code (def_stmt_info->stmt) == GIMPLE_PHI
	      && STMT_VINFO_DEF_TYPE (def_stmt_info) == vect_internal_def
	      && !is_loop_header_bb_p (gimple_bb (def_stmt_info->stmt))));
}

/* Detect SLP reduction of the form:

   #a1 = phi <a5, a0>
   a2 = operation (a1)
   a3 = operation (a2)
   a4 = operation (a3)
   a5 = operation (a4)

   #a = phi <a5>

   PHI is the reduction phi node (#a1 = phi <a5, a0> above)
   FIRST_STMT is the first reduction stmt in the chain
   (a2 = operation (a1)).

   Return TRUE if a reduction chain was detected.  */

static bool
parloops_is_slp_reduction (loop_vec_info loop_info, gimple *phi,
			   gimple *first_stmt)
{
  class loop *loop = (gimple_bb (phi))->loop_father;
  class loop *vect_loop = LOOP_VINFO_LOOP (loop_info);
  enum tree_code code;
  gimple *loop_use_stmt = NULL;
  stmt_vec_info use_stmt_info;
  tree lhs;
  imm_use_iterator imm_iter;
  use_operand_p use_p;
  int nloop_uses, size = 0, n_out_of_loop_uses;
  bool found = false;

  if (loop != vect_loop)
    return false;

  auto_vec<stmt_vec_info, 8> reduc_chain;
  lhs = PHI_RESULT (phi);
  code = gimple_assign_rhs_code (first_stmt);
  while (1)
    {
      nloop_uses = 0;
      n_out_of_loop_uses = 0;
      FOR_EACH_IMM_USE_FAST (use_p, imm_iter, lhs)
        {
	  gimple *use_stmt = USE_STMT (use_p);
	  if (is_gimple_debug (use_stmt))
	    continue;

          /* Check if we got back to the reduction phi.  */
	  if (use_stmt == phi)
            {
	      loop_use_stmt = use_stmt;
              found = true;
              break;
            }

          if (flow_bb_inside_loop_p (loop, gimple_bb (use_stmt)))
            {
	      loop_use_stmt = use_stmt;
	      nloop_uses++;
            }
           else
             n_out_of_loop_uses++;

           /* There are can be either a single use in the loop or two uses in
              phi nodes.  */
           if (nloop_uses > 1 || (n_out_of_loop_uses && nloop_uses))
             return false;
        }

      if (found)
        break;

      /* We reached a statement with no loop uses.  */
      if (nloop_uses == 0)
	return false;

      /* This is a loop exit phi, and we haven't reached the reduction phi.  */
      if (gimple_code (loop_use_stmt) == GIMPLE_PHI)
        return false;

      if (!is_gimple_assign (loop_use_stmt)
	  || code != gimple_assign_rhs_code (loop_use_stmt)
	  || !flow_bb_inside_loop_p (loop, gimple_bb (loop_use_stmt)))
        return false;

      /* Insert USE_STMT into reduction chain.  */
      use_stmt_info = loop_info->lookup_stmt (loop_use_stmt);
      reduc_chain.safe_push (use_stmt_info);

      lhs = gimple_assign_lhs (loop_use_stmt);
      size++;
   }

  if (!found || loop_use_stmt != phi || size < 2)
    return false;

  /* Swap the operands, if needed, to make the reduction operand be the second
     operand.  */
  lhs = PHI_RESULT (phi);
  for (unsigned i = 0; i < reduc_chain.length (); ++i)
    {
      gassign *next_stmt = as_a <gassign *> (reduc_chain[i]->stmt);
      if (gimple_assign_rhs2 (next_stmt) == lhs)
	{
	  tree op = gimple_assign_rhs1 (next_stmt);
	  stmt_vec_info def_stmt_info = loop_info->lookup_def (op);

	  /* Check that the other def is either defined in the loop
	     ("vect_internal_def"), or it's an induction (defined by a
	     loop-header phi-node).  */
	  if (def_stmt_info
	      && flow_bb_inside_loop_p (loop, gimple_bb (def_stmt_info->stmt))
	      && parloops_valid_reduction_input_p (def_stmt_info))
	    {
	      lhs = gimple_assign_lhs (next_stmt);
	      continue;
	    }

	  return false;
	}
      else
	{
          tree op = gimple_assign_rhs2 (next_stmt);
	  stmt_vec_info def_stmt_info = loop_info->lookup_def (op);

          /* Check that the other def is either defined in the loop
            ("vect_internal_def"), or it's an induction (defined by a
            loop-header phi-node).  */
	  if (def_stmt_info
	      && flow_bb_inside_loop_p (loop, gimple_bb (def_stmt_info->stmt))
	      && parloops_valid_reduction_input_p (def_stmt_info))
	    {
	      if (dump_enabled_p ())
		dump_printf_loc (MSG_NOTE, vect_location, "swapping oprnds: %G",
				 next_stmt);

	      swap_ssa_operands (next_stmt,
				 gimple_assign_rhs1_ptr (next_stmt),
                                 gimple_assign_rhs2_ptr (next_stmt));
	      update_stmt (next_stmt);
	    }
	  else
	    return false;
        }

      lhs = gimple_assign_lhs (next_stmt);
    }

  /* Build up the actual chain.  */
  for (unsigned i = 0; i < reduc_chain.length () - 1; ++i)
    {
      REDUC_GROUP_FIRST_ELEMENT (reduc_chain[i]) = reduc_chain[0];
      REDUC_GROUP_NEXT_ELEMENT (reduc_chain[i]) = reduc_chain[i+1];
    }
  REDUC_GROUP_FIRST_ELEMENT (reduc_chain.last ()) = reduc_chain[0];
  REDUC_GROUP_NEXT_ELEMENT (reduc_chain.last ()) = NULL;

  /* Save the chain for further analysis in SLP detection.  */
  LOOP_VINFO_REDUCTION_CHAINS (loop_info).safe_push (reduc_chain[0]);
  REDUC_GROUP_SIZE (reduc_chain[0]) = size;

  return true;
}

/* Return true if we need an in-order reduction for operation CODE
   on type TYPE.  NEED_WRAPPING_INTEGRAL_OVERFLOW is true if integer
   overflow must wrap.  */

static bool
parloops_needs_fold_left_reduction_p (tree type, tree_code code,
				      bool need_wrapping_integral_overflow)
{
  /* CHECKME: check for !flag_finite_math_only too?  */
  if (SCALAR_FLOAT_TYPE_P (type))
    switch (code)
      {
      case MIN_EXPR:
      case MAX_EXPR:
	return false;

      default:
	return !flag_associative_math;
      }

  if (INTEGRAL_TYPE_P (type))
    {
      if (!operation_no_trapping_overflow (type, code))
	return true;
      if (need_wrapping_integral_overflow
	  && !TYPE_OVERFLOW_WRAPS (type)
	  && operation_can_overflow (code))
	return true;
      return false;
    }

  if (SAT_FIXED_POINT_TYPE_P (type))
    return true;

  return false;
}


/* Function parloops_is_simple_reduction

   (1) Detect a cross-iteration def-use cycle that represents a simple
   reduction computation.  We look for the following pattern:

   loop_header:
     a1 = phi < a0, a2 >
     a3 = ...
     a2 = operation (a3, a1)

   or

   a3 = ...
   loop_header:
     a1 = phi < a0, a2 >
     a2 = operation (a3, a1)

   such that:
   1. operation is commutative and associative and it is safe to
      change the order of the computation
   2. no uses for a2 in the loop (a2 is used out of the loop)
   3. no uses of a1 in the loop besides the reduction operation
   4. no uses of a1 outside the loop.

   Conditions 1,4 are tested here.
   Conditions 2,3 are tested in vect_mark_stmts_to_be_vectorized.

   (2) Detect a cross-iteration def-use cycle in nested loops, i.e.,
   nested cycles.

   (3) Detect cycles of phi nodes in outer-loop vectorization, i.e., double
   reductions:

     a1 = phi < a0, a2 >
     inner loop (def of a3)
     a2 = phi < a3 >

   (4) Detect condition expressions, ie:
     for (int i = 0; i < N; i++)
       if (a[i] < val)
	ret_val = a[i];

*/

static stmt_vec_info
parloops_is_simple_reduction (loop_vec_info loop_info, stmt_vec_info phi_info,
			  bool *double_reduc,
			  bool need_wrapping_integral_overflow,
			  enum vect_reduction_type *v_reduc_type)
{
  gphi *phi = as_a <gphi *> (phi_info->stmt);
  class loop *loop = (gimple_bb (phi))->loop_father;
  class loop *vect_loop = LOOP_VINFO_LOOP (loop_info);
  bool nested_in_vect_loop = flow_loop_nested_p (vect_loop, loop);
  gimple *phi_use_stmt = NULL;
  enum tree_code orig_code, code;
  tree op1, op2, op3 = NULL_TREE, op4 = NULL_TREE;
  tree type;
  tree name;
  imm_use_iterator imm_iter;
  use_operand_p use_p;
  bool phi_def;

  *double_reduc = false;
  *v_reduc_type = TREE_CODE_REDUCTION;

  tree phi_name = PHI_RESULT (phi);
  /* ???  If there are no uses of the PHI result the inner loop reduction
     won't be detected as possibly double-reduction by vectorizable_reduction
     because that tries to walk the PHI arg from the preheader edge which
     can be constant.  See PR60382.  */
  if (has_zero_uses (phi_name))
    return NULL;
  unsigned nphi_def_loop_uses = 0;
  FOR_EACH_IMM_USE_FAST (use_p, imm_iter, phi_name)
    {
      gimple *use_stmt = USE_STMT (use_p);
      if (is_gimple_debug (use_stmt))
	continue;

      if (!flow_bb_inside_loop_p (loop, gimple_bb (use_stmt)))
        {
          if (dump_enabled_p ())
	    dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
			     "intermediate value used outside loop.\n");

          return NULL;
        }

      nphi_def_loop_uses++;
      phi_use_stmt = use_stmt;
    }

  edge latch_e = loop_latch_edge (loop);
  tree loop_arg = PHI_ARG_DEF_FROM_EDGE (phi, latch_e);
  if (TREE_CODE (loop_arg) != SSA_NAME)
    {
      if (dump_enabled_p ())
	dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
			 "reduction: not ssa_name: %T\n", loop_arg);
      return NULL;
    }

  stmt_vec_info def_stmt_info = loop_info->lookup_def (loop_arg);
  if (!def_stmt_info
      || !flow_bb_inside_loop_p (loop, gimple_bb (def_stmt_info->stmt)))
    return NULL;

  if (gassign *def_stmt = dyn_cast <gassign *> (def_stmt_info->stmt))
    {
      name = gimple_assign_lhs (def_stmt);
      phi_def = false;
    }
  else if (gphi *def_stmt = dyn_cast <gphi *> (def_stmt_info->stmt))
    {
      name = PHI_RESULT (def_stmt);
      phi_def = true;
    }
  else
    {
      if (dump_enabled_p ())
	dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
			 "reduction: unhandled reduction operation: %G",
			 def_stmt_info->stmt);
      return NULL;
    }

  unsigned nlatch_def_loop_uses = 0;
  auto_vec<gphi *, 3> lcphis;
  bool inner_loop_of_double_reduc = false;
  FOR_EACH_IMM_USE_FAST (use_p, imm_iter, name)
    {
      gimple *use_stmt = USE_STMT (use_p);
      if (is_gimple_debug (use_stmt))
	continue;
      if (flow_bb_inside_loop_p (loop, gimple_bb (use_stmt)))
	nlatch_def_loop_uses++;
      else
	{
	  /* We can have more than one loop-closed PHI.  */
	  lcphis.safe_push (as_a <gphi *> (use_stmt));
	  if (nested_in_vect_loop
	      && (STMT_VINFO_DEF_TYPE (loop_info->lookup_stmt (use_stmt))
		  == vect_double_reduction_def))
	    inner_loop_of_double_reduc = true;
	}
    }

  /* If this isn't a nested cycle or if the nested cycle reduction value
     is used ouside of the inner loop we cannot handle uses of the reduction
     value.  */
  if ((!nested_in_vect_loop || inner_loop_of_double_reduc)
      && (nlatch_def_loop_uses > 1 || nphi_def_loop_uses > 1))
    {
      if (dump_enabled_p ())
	dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
			 "reduction used in loop.\n");
      return NULL;
    }

  /* If DEF_STMT is a phi node itself, we expect it to have a single argument
     defined in the inner loop.  */
  if (phi_def)
    {
      gphi *def_stmt = as_a <gphi *> (def_stmt_info->stmt);
      op1 = PHI_ARG_DEF (def_stmt, 0);

      if (gimple_phi_num_args (def_stmt) != 1
          || TREE_CODE (op1) != SSA_NAME)
        {
          if (dump_enabled_p ())
	    dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
			     "unsupported phi node definition.\n");

          return NULL;
        }

      gimple *def1 = SSA_NAME_DEF_STMT (op1);
      if (gimple_bb (def1)
	  && flow_bb_inside_loop_p (loop, gimple_bb (def_stmt))
          && loop->inner
          && flow_bb_inside_loop_p (loop->inner, gimple_bb (def1))
          && is_gimple_assign (def1)
	  && is_a <gphi *> (phi_use_stmt)
	  && flow_bb_inside_loop_p (loop->inner, gimple_bb (phi_use_stmt)))
        {
          if (dump_enabled_p ())
            report_ploop_op (MSG_NOTE, def_stmt,
			     "detected double reduction: ");

          *double_reduc = true;
	  return def_stmt_info;
        }

      return NULL;
    }

  /* If we are vectorizing an inner reduction we are executing that
     in the original order only in case we are not dealing with a
     double reduction.  */
  bool check_reduction = true;
  if (flow_loop_nested_p (vect_loop, loop))
    {
      gphi *lcphi;
      unsigned i;
      check_reduction = false;
      FOR_EACH_VEC_ELT (lcphis, i, lcphi)
	FOR_EACH_IMM_USE_FAST (use_p, imm_iter, gimple_phi_result (lcphi))
	  {
	    gimple *use_stmt = USE_STMT (use_p);
	    if (is_gimple_debug (use_stmt))
	      continue;
	    if (! flow_bb_inside_loop_p (vect_loop, gimple_bb (use_stmt)))
	      check_reduction = true;
	  }
    }

  gassign *def_stmt = as_a <gassign *> (def_stmt_info->stmt);
  code = orig_code = gimple_assign_rhs_code (def_stmt);

  if (nested_in_vect_loop && !check_reduction)
    {
      /* FIXME: Even for non-reductions code generation is funneled
	 through vectorizable_reduction for the stmt defining the
	 PHI latch value.  So we have to artificially restrict ourselves
	 for the supported operations.  */
      switch (get_gimple_rhs_class (code))
	{
	case GIMPLE_BINARY_RHS:
	case GIMPLE_TERNARY_RHS:
	  break;
	default:
	  /* Not supported by vectorizable_reduction.  */
	  if (dump_enabled_p ())
	    report_ploop_op (MSG_MISSED_OPTIMIZATION, def_stmt,
			     "nested cycle: not handled operation: ");
	  return NULL;
	}
      if (dump_enabled_p ())
	report_ploop_op (MSG_NOTE, def_stmt, "detected nested cycle: ");
      return def_stmt_info;
    }

  /* We can handle "res -= x[i]", which is non-associative by
     simply rewriting this into "res += -x[i]".  Avoid changing
     gimple instruction for the first simple tests and only do this
     if we're allowed to change code at all.  */
  if (code == MINUS_EXPR && gimple_assign_rhs2 (def_stmt) != phi_name)
    code = PLUS_EXPR;

  if (code == COND_EXPR)
    {
      if (! nested_in_vect_loop)
	*v_reduc_type = COND_REDUCTION;

      op3 = gimple_assign_rhs1 (def_stmt);
      if (COMPARISON_CLASS_P (op3))
        {
          op4 = TREE_OPERAND (op3, 1);
          op3 = TREE_OPERAND (op3, 0);
        }
      if (op3 == phi_name || op4 == phi_name)
	{
	  if (dump_enabled_p ())
	    report_ploop_op (MSG_MISSED_OPTIMIZATION, def_stmt,
			     "reduction: condition depends on previous"
			     " iteration: ");
	  return NULL;
	}

      op1 = gimple_assign_rhs2 (def_stmt);
      op2 = gimple_assign_rhs3 (def_stmt);
    }
  else if (!commutative_tree_code (code) || !associative_tree_code (code))
    {
      if (dump_enabled_p ())
	report_ploop_op (MSG_MISSED_OPTIMIZATION, def_stmt,
			 "reduction: not commutative/associative: ");
      return NULL;
    }
  else if (get_gimple_rhs_class (code) == GIMPLE_BINARY_RHS)
    {
      op1 = gimple_assign_rhs1 (def_stmt);
      op2 = gimple_assign_rhs2 (def_stmt);
    }
  else
    {
      if (dump_enabled_p ())
	report_ploop_op (MSG_MISSED_OPTIMIZATION, def_stmt,
			 "reduction: not handled operation: ");
      return NULL;
    }

  if (TREE_CODE (op1) != SSA_NAME && TREE_CODE (op2) != SSA_NAME)
    {
      if (dump_enabled_p ())
	report_ploop_op (MSG_MISSED_OPTIMIZATION, def_stmt,
			 "reduction: both uses not ssa_names: ");

      return NULL;
    }

  type = TREE_TYPE (gimple_assign_lhs (def_stmt));
  if ((TREE_CODE (op1) == SSA_NAME
       && !types_compatible_p (type,TREE_TYPE (op1)))
      || (TREE_CODE (op2) == SSA_NAME
          && !types_compatible_p (type, TREE_TYPE (op2)))
      || (op3 && TREE_CODE (op3) == SSA_NAME
          && !types_compatible_p (type, TREE_TYPE (op3)))
      || (op4 && TREE_CODE (op4) == SSA_NAME
          && !types_compatible_p (type, TREE_TYPE (op4))))
    {
      if (dump_enabled_p ())
        {
          dump_printf_loc (MSG_NOTE, vect_location,
			   "reduction: multiple types: operation type: "
			   "%T, operands types: %T,%T",
			   type,  TREE_TYPE (op1), TREE_TYPE (op2));
          if (op3)
	    dump_printf (MSG_NOTE, ",%T", TREE_TYPE (op3));

          if (op4)
	    dump_printf (MSG_NOTE, ",%T", TREE_TYPE (op4));
          dump_printf (MSG_NOTE, "\n");
        }

      return NULL;
    }

  /* Check whether it's ok to change the order of the computation.
     Generally, when vectorizing a reduction we change the order of the
     computation.  This may change the behavior of the program in some
     cases, so we need to check that this is ok.  One exception is when
     vectorizing an outer-loop: the inner-loop is executed sequentially,
     and therefore vectorizing reductions in the inner-loop during
     outer-loop vectorization is safe.  */
  if (check_reduction
      && *v_reduc_type == TREE_CODE_REDUCTION
      && parloops_needs_fold_left_reduction_p (type, code,
					       need_wrapping_integral_overflow))
    *v_reduc_type = FOLD_LEFT_REDUCTION;

  /* Reduction is safe. We're dealing with one of the following:
     1) integer arithmetic and no trapv
     2) floating point arithmetic, and special flags permit this optimization
     3) nested cycle (i.e., outer loop vectorization).  */
  stmt_vec_info def1_info = loop_info->lookup_def (op1);
  stmt_vec_info def2_info = loop_info->lookup_def (op2);
  if (code != COND_EXPR && !def1_info && !def2_info)
    {
      if (dump_enabled_p ())
	report_ploop_op (MSG_NOTE, def_stmt,
			 "reduction: no defs for operands: ");
      return NULL;
    }

  /* Check that one def is the reduction def, defined by PHI,
     the other def is either defined in the loop ("vect_internal_def"),
     or it's an induction (defined by a loop-header phi-node).  */

  if (def2_info
      && def2_info->stmt == phi
      && (code == COND_EXPR
	  || !def1_info
	  || !flow_bb_inside_loop_p (loop, gimple_bb (def1_info->stmt))
	  || parloops_valid_reduction_input_p (def1_info)))
    {
      if (dump_enabled_p ())
	report_ploop_op (MSG_NOTE, def_stmt, "detected reduction: ");
      return def_stmt_info;
    }

  if (def1_info
      && def1_info->stmt == phi
      && (code == COND_EXPR
	  || !def2_info
	  || !flow_bb_inside_loop_p (loop, gimple_bb (def2_info->stmt))
	  || parloops_valid_reduction_input_p (def2_info)))
    {
      if (! nested_in_vect_loop && orig_code != MINUS_EXPR)
	{
	  /* Check if we can swap operands (just for simplicity - so that
	     the rest of the code can assume that the reduction variable
	     is always the last (second) argument).  */
	  if (code == COND_EXPR)
	    {
	      /* Swap cond_expr by inverting the condition.  */
	      tree cond_expr = gimple_assign_rhs1 (def_stmt);
	      enum tree_code invert_code = ERROR_MARK;
	      enum tree_code cond_code = TREE_CODE (cond_expr);

	      if (TREE_CODE_CLASS (cond_code) == tcc_comparison)
		{
		  bool honor_nans = HONOR_NANS (TREE_OPERAND (cond_expr, 0));
		  invert_code = invert_tree_comparison (cond_code, honor_nans);
		}
	      if (invert_code != ERROR_MARK)
		{
		  TREE_SET_CODE (cond_expr, invert_code);
		  swap_ssa_operands (def_stmt,
				     gimple_assign_rhs2_ptr (def_stmt),
				     gimple_assign_rhs3_ptr (def_stmt));
		}
	      else
		{
		  if (dump_enabled_p ())
		    report_ploop_op (MSG_NOTE, def_stmt,
				     "detected reduction: cannot swap operands "
				     "for cond_expr");
		  return NULL;
		}
	    }
	  else
	    swap_ssa_operands (def_stmt, gimple_assign_rhs1_ptr (def_stmt),
			       gimple_assign_rhs2_ptr (def_stmt));

	  if (dump_enabled_p ())
	    report_ploop_op (MSG_NOTE, def_stmt,
			     "detected reduction: need to swap operands: ");
        }
      else
        {
          if (dump_enabled_p ())
            report_ploop_op (MSG_NOTE, def_stmt, "detected reduction: ");
        }

      return def_stmt_info;
    }

  /* Try to find SLP reduction chain.  */
  if (! nested_in_vect_loop
      && code != COND_EXPR
      && orig_code != MINUS_EXPR
      && parloops_is_slp_reduction (loop_info, phi, def_stmt))
    {
      if (dump_enabled_p ())
        report_ploop_op (MSG_NOTE, def_stmt,
			 "reduction: detected reduction chain: ");

      return def_stmt_info;
    }

  /* Look for the expression computing loop_arg from loop PHI result.  */
  if (check_reduction_path (vect_location, loop, phi, loop_arg, code))
    return def_stmt_info;

  if (dump_enabled_p ())
    {
      report_ploop_op (MSG_MISSED_OPTIMIZATION, def_stmt,
		       "reduction: unknown pattern: ");
    }

  return NULL;
}

/* Wrapper around vect_is_simple_reduction, which will modify code
   in-place if it enables detection of more reductions.  Arguments
   as there.  */

stmt_vec_info
parloops_force_simple_reduction (loop_vec_info loop_info, stmt_vec_info phi_info,
			     bool *double_reduc,
			     bool need_wrapping_integral_overflow)
{
  enum vect_reduction_type v_reduc_type;
  stmt_vec_info def_info
    = parloops_is_simple_reduction (loop_info, phi_info, double_reduc,
				need_wrapping_integral_overflow,
				&v_reduc_type);
  if (def_info)
    {
      STMT_VINFO_REDUC_TYPE (phi_info) = v_reduc_type;
      STMT_VINFO_REDUC_DEF (phi_info) = def_info;
      STMT_VINFO_REDUC_TYPE (def_info) = v_reduc_type;
      STMT_VINFO_REDUC_DEF (def_info) = phi_info;
    }
  return def_info;
}

/* Minimal number of iterations of a loop that should be executed in each
   thread.  */
#define MIN_PER_THREAD param_parloops_min_per_thread

/* Element of the hashtable, representing a
   reduction in the current loop.  */
struct reduction_info
{
  gimple *reduc_stmt;		/* reduction statement.  */
  gimple *reduc_phi;		/* The phi node defining the reduction.  */
  enum tree_code reduction_code;/* code for the reduction operation.  */
  unsigned reduc_version;	/* SSA_NAME_VERSION of original reduc_phi
				   result.  */
  gphi *keep_res;		/* The PHI_RESULT of this phi is the resulting value
				   of the reduction variable when existing the loop. */
  tree initial_value;		/* The initial value of the reduction var before entering the loop.  */
  tree field;			/*  the name of the field in the parloop data structure intended for reduction.  */
  tree reduc_addr;		/* The address of the reduction variable for
				   openacc reductions.  */
  tree init;			/* reduction initialization value.  */
  gphi *new_phi;		/* (helper field) Newly created phi node whose result
				   will be passed to the atomic operation.  Represents
				   the local result each thread computed for the reduction
				   operation.  */
};

/* Reduction info hashtable helpers.  */

struct reduction_hasher : free_ptr_hash <reduction_info>
{
  static inline hashval_t hash (const reduction_info *);
  static inline bool equal (const reduction_info *, const reduction_info *);
};

/* Equality and hash functions for hashtab code.  */

inline bool
reduction_hasher::equal (const reduction_info *a, const reduction_info *b)
{
  return (a->reduc_phi == b->reduc_phi);
}

inline hashval_t
reduction_hasher::hash (const reduction_info *a)
{
  return a->reduc_version;
}

typedef hash_table<reduction_hasher> reduction_info_table_type;


static struct reduction_info *
reduction_phi (reduction_info_table_type *reduction_list, gimple *phi)
{
  struct reduction_info tmpred, *red;

  if (reduction_list->is_empty () || phi == NULL)
    return NULL;

  if (gimple_uid (phi) == (unsigned int)-1
      || gimple_uid (phi) == 0)
    return NULL;

  tmpred.reduc_phi = phi;
  tmpred.reduc_version = gimple_uid (phi);
  red = reduction_list->find (&tmpred);
  gcc_assert (red == NULL || red->reduc_phi == phi);

  return red;
}

/* Element of hashtable of names to copy.  */

struct name_to_copy_elt
{
  unsigned version;	/* The version of the name to copy.  */
  tree new_name;	/* The new name used in the copy.  */
  tree field;		/* The field of the structure used to pass the
			   value.  */
};

/* Name copies hashtable helpers.  */

struct name_to_copy_hasher : free_ptr_hash <name_to_copy_elt>
{
  static inline hashval_t hash (const name_to_copy_elt *);
  static inline bool equal (const name_to_copy_elt *, const name_to_copy_elt *);
};

/* Equality and hash functions for hashtab code.  */

inline bool
name_to_copy_hasher::equal (const name_to_copy_elt *a, const name_to_copy_elt *b)
{
  return a->version == b->version;
}

inline hashval_t
name_to_copy_hasher::hash (const name_to_copy_elt *a)
{
  return (hashval_t) a->version;
}

typedef hash_table<name_to_copy_hasher> name_to_copy_table_type;

/* A transformation matrix, which is a self-contained ROWSIZE x COLSIZE
   matrix.  Rather than use floats, we simply keep a single DENOMINATOR that
   represents the denominator for every element in the matrix.  */
typedef struct lambda_trans_matrix_s
{
  lambda_matrix matrix;
  int rowsize;
  int colsize;
  int denominator;
} *lambda_trans_matrix;
#define LTM_MATRIX(T) ((T)->matrix)
#define LTM_ROWSIZE(T) ((T)->rowsize)
#define LTM_COLSIZE(T) ((T)->colsize)
#define LTM_DENOMINATOR(T) ((T)->denominator)

/* Allocate a new transformation matrix.  */

static lambda_trans_matrix
lambda_trans_matrix_new (int colsize, int rowsize,
			 struct obstack * lambda_obstack)
{
  lambda_trans_matrix ret;

  ret = (lambda_trans_matrix)
    obstack_alloc (lambda_obstack, sizeof (struct lambda_trans_matrix_s));
  LTM_MATRIX (ret) = lambda_matrix_new (rowsize, colsize, lambda_obstack);
  LTM_ROWSIZE (ret) = rowsize;
  LTM_COLSIZE (ret) = colsize;
  LTM_DENOMINATOR (ret) = 1;
  return ret;
}

/* Multiply a vector VEC by a matrix MAT.
   MAT is an M*N matrix, and VEC is a vector with length N.  The result
   is stored in DEST which must be a vector of length M.  */

static void
lambda_matrix_vector_mult (lambda_matrix matrix, int m, int n,
			   lambda_vector vec, lambda_vector dest)
{
  int i, j;

  lambda_vector_clear (dest, m);
  for (i = 0; i < m; i++)
    for (j = 0; j < n; j++)
      dest[i] += matrix[i][j] * vec[j];
}

/* Return true if TRANS is a legal transformation matrix that respects
   the dependence vectors in DISTS and DIRS.  The conservative answer
   is false.

   "Wolfe proves that a unimodular transformation represented by the
   matrix T is legal when applied to a loop nest with a set of
   lexicographically non-negative distance vectors RDG if and only if
   for each vector d in RDG, (T.d >= 0) is lexicographically positive.
   i.e.: if and only if it transforms the lexicographically positive
   distance vectors to lexicographically positive vectors.  Note that
   a unimodular matrix must transform the zero vector (and only it) to
   the zero vector." S.Muchnick.  */

static bool
lambda_transform_legal_p (lambda_trans_matrix trans,
			  int nb_loops,
			  vec<ddr_p> dependence_relations)
{
  unsigned int i, j;
  lambda_vector distres;
  struct data_dependence_relation *ddr;

  gcc_assert (LTM_COLSIZE (trans) == nb_loops
	      && LTM_ROWSIZE (trans) == nb_loops);

  /* When there are no dependences, the transformation is correct.  */
  if (dependence_relations.length () == 0)
    return true;

  ddr = dependence_relations[0];
  if (ddr == NULL)
    return true;

  /* When there is an unknown relation in the dependence_relations, we
     know that it is no worth looking at this loop nest: give up.  */
  if (DDR_ARE_DEPENDENT (ddr) == chrec_dont_know)
    return false;

  distres = lambda_vector_new (nb_loops);

  /* For each distance vector in the dependence graph.  */
  FOR_EACH_VEC_ELT (dependence_relations, i, ddr)
    {
      /* Don't care about relations for which we know that there is no
	 dependence, nor about read-read (aka. output-dependences):
	 these data accesses can happen in any order.  */
      if (DDR_ARE_DEPENDENT (ddr) == chrec_known
	  || (DR_IS_READ (DDR_A (ddr)) && DR_IS_READ (DDR_B (ddr))))
	continue;

      /* Conservatively answer: "this transformation is not valid".  */
      if (DDR_ARE_DEPENDENT (ddr) == chrec_dont_know)
	return false;

      /* If the dependence could not be captured by a distance vector,
	 conservatively answer that the transform is not valid.  */
      if (DDR_NUM_DIST_VECTS (ddr) == 0)
	return false;

      /* Compute trans.dist_vect */
      for (j = 0; j < DDR_NUM_DIST_VECTS (ddr); j++)
	{
	  lambda_matrix_vector_mult (LTM_MATRIX (trans), nb_loops, nb_loops,
				     DDR_DIST_VECT (ddr, j), distres);

	  if (!lambda_vector_lexico_pos (distres, nb_loops))
	    return false;
	}
    }
  return true;
}

/* Data dependency analysis. Returns true if the iterations of LOOP
   are independent on each other (that is, if we can execute them
   in parallel).  */

static bool
loop_parallel_p (class loop *loop, struct obstack * parloop_obstack)
{
  vec<ddr_p> dependence_relations;
  vec<data_reference_p> datarefs;
  lambda_trans_matrix trans;
  bool ret = false;

  if (dump_file && (dump_flags & TDF_DETAILS))
  {
    fprintf (dump_file, "Considering loop %d\n", loop->num);
    if (!loop->inner)
      fprintf (dump_file, "loop is innermost\n");
    else
      fprintf (dump_file, "loop NOT innermost\n");
   }

  /* Check for problems with dependences.  If the loop can be reversed,
     the iterations are independent.  */
  auto_vec<loop_p, 3> loop_nest;
  datarefs.create (10);
  dependence_relations.create (100);
  if (! compute_data_dependences_for_loop (loop, true, &loop_nest, &datarefs,
					   &dependence_relations))
    {
      if (dump_file && (dump_flags & TDF_DETAILS))
	fprintf (dump_file, "  FAILED: cannot analyze data dependencies\n");
      ret = false;
      goto end;
    }
  if (dump_file && (dump_flags & TDF_DETAILS))
    dump_data_dependence_relations (dump_file, dependence_relations);

  trans = lambda_trans_matrix_new (1, 1, parloop_obstack);
  LTM_MATRIX (trans)[0][0] = -1;

  if (lambda_transform_legal_p (trans, 1, dependence_relations))
    {
      ret = true;
      if (dump_file && (dump_flags & TDF_DETAILS))
	fprintf (dump_file, "  SUCCESS: may be parallelized\n");
    }
  else if (dump_file && (dump_flags & TDF_DETAILS))
    fprintf (dump_file,
	     "  FAILED: data dependencies exist across iterations\n");

 end:
  free_dependence_relations (dependence_relations);
  free_data_refs (datarefs);

  return ret;
}

/* Return true when LOOP contains basic blocks marked with the
   BB_IRREDUCIBLE_LOOP flag.  */

static inline bool
loop_has_blocks_with_irreducible_flag (class loop *loop)
{
  unsigned i;
  basic_block *bbs = get_loop_body_in_dom_order (loop);
  bool res = true;

  for (i = 0; i < loop->num_nodes; i++)
    if (bbs[i]->flags & BB_IRREDUCIBLE_LOOP)
      goto end;

  res = false;
 end:
  free (bbs);
  return res;
}

/* Assigns the address of OBJ in TYPE to an ssa name, and returns this name.
   The assignment statement is placed on edge ENTRY.  DECL_ADDRESS maps decls
   to their addresses that can be reused.  The address of OBJ is known to
   be invariant in the whole function.  Other needed statements are placed
   right before GSI.  */

static tree
take_address_of (tree obj, tree type, edge entry,
		 int_tree_htab_type *decl_address, gimple_stmt_iterator *gsi)
{
  int uid;
  tree *var_p, name, addr;
  gassign *stmt;
  gimple_seq stmts;

  /* Since the address of OBJ is invariant, the trees may be shared.
     Avoid rewriting unrelated parts of the code.  */
  obj = unshare_expr (obj);
  for (var_p = &obj;
       handled_component_p (*var_p);
       var_p = &TREE_OPERAND (*var_p, 0))
    continue;

  /* Canonicalize the access to base on a MEM_REF.  */
  if (DECL_P (*var_p))
    *var_p = build_simple_mem_ref (build_fold_addr_expr (*var_p));

  /* Assign a canonical SSA name to the address of the base decl used
     in the address and share it for all accesses and addresses based
     on it.  */
  uid = DECL_UID (TREE_OPERAND (TREE_OPERAND (*var_p, 0), 0));
  int_tree_map elt;
  elt.uid = uid;
  int_tree_map *slot = decl_address->find_slot (elt, INSERT);
  if (!slot->to)
    {
      if (gsi == NULL)
	return NULL;
      addr = TREE_OPERAND (*var_p, 0);
      const char *obj_name
	= get_name (TREE_OPERAND (TREE_OPERAND (*var_p, 0), 0));
      if (obj_name)
	name = make_temp_ssa_name (TREE_TYPE (addr), NULL, obj_name);
      else
	name = make_ssa_name (TREE_TYPE (addr));
      stmt = gimple_build_assign (name, addr);
      gsi_insert_on_edge_immediate (entry, stmt);

      slot->uid = uid;
      slot->to = name;
    }
  else
    name = slot->to;

  /* Express the address in terms of the canonical SSA name.  */
  TREE_OPERAND (*var_p, 0) = name;
  if (gsi == NULL)
    return build_fold_addr_expr_with_type (obj, type);

  name = force_gimple_operand (build_addr (obj),
			       &stmts, true, NULL_TREE);
  if (!gimple_seq_empty_p (stmts))
    gsi_insert_seq_before (gsi, stmts, GSI_SAME_STMT);

  if (!useless_type_conversion_p (type, TREE_TYPE (name)))
    {
      name = force_gimple_operand (fold_convert (type, name), &stmts, true,
				   NULL_TREE);
      if (!gimple_seq_empty_p (stmts))
	gsi_insert_seq_before (gsi, stmts, GSI_SAME_STMT);
    }

  return name;
}

static tree
reduc_stmt_res (gimple *stmt)
{
  return (gimple_code (stmt) == GIMPLE_PHI
	  ? gimple_phi_result (stmt)
	  : gimple_assign_lhs (stmt));
}

/* Callback for htab_traverse.  Create the initialization statement
   for reduction described in SLOT, and place it at the preheader of
   the loop described in DATA.  */

int
initialize_reductions (reduction_info **slot, class loop *loop)
{
  tree init;
  tree type, arg;
  edge e;

  struct reduction_info *const reduc = *slot;

  /* Create initialization in preheader:
     reduction_variable = initialization value of reduction.  */

  /* In the phi node at the header, replace the argument coming
     from the preheader with the reduction initialization value.  */

  /* Initialize the reduction.  */
  type = TREE_TYPE (PHI_RESULT (reduc->reduc_phi));
  init = omp_reduction_init_op (gimple_location (reduc->reduc_stmt),
				reduc->reduction_code, type);
  reduc->init = init;

  /* Replace the argument representing the initialization value
     with the initialization value for the reduction (neutral
     element for the particular operation, e.g. 0 for PLUS_EXPR,
     1 for MULT_EXPR, etc).
     Keep the old value in a new variable "reduction_initial",
     that will be taken in consideration after the parallel
     computing is done.  */

  e = loop_preheader_edge (loop);
  arg = PHI_ARG_DEF_FROM_EDGE (reduc->reduc_phi, e);
  /* Create new variable to hold the initial value.  */

  SET_USE (PHI_ARG_DEF_PTR_FROM_EDGE
	   (reduc->reduc_phi, loop_preheader_edge (loop)), init);
  reduc->initial_value = arg;
  return 1;
}

struct elv_data
{
  struct walk_stmt_info info;
  edge entry;
  int_tree_htab_type *decl_address;
  gimple_stmt_iterator *gsi;
  bool changed;
  bool reset;
};

/* Eliminates references to local variables in *TP out of the single
   entry single exit region starting at DTA->ENTRY.
   DECL_ADDRESS contains addresses of the references that had their
   address taken already.  If the expression is changed, CHANGED is
   set to true.  Callback for walk_tree.  */

static tree
eliminate_local_variables_1 (tree *tp, int *walk_subtrees, void *data)
{
  struct elv_data *const dta = (struct elv_data *) data;
  tree t = *tp, var, addr, addr_type, type, obj;

  if (DECL_P (t))
    {
      *walk_subtrees = 0;

      if (!SSA_VAR_P (t) || DECL_EXTERNAL (t))
	return NULL_TREE;

      type = TREE_TYPE (t);
      addr_type = build_pointer_type (type);
      addr = take_address_of (t, addr_type, dta->entry, dta->decl_address,
			      dta->gsi);
      if (dta->gsi == NULL && addr == NULL_TREE)
	{
	  dta->reset = true;
	  return NULL_TREE;
	}

      *tp = build_simple_mem_ref (addr);

      dta->changed = true;
      return NULL_TREE;
    }

  if (TREE_CODE (t) == ADDR_EXPR)
    {
      /* ADDR_EXPR may appear in two contexts:
	 -- as a gimple operand, when the address taken is a function invariant
	 -- as gimple rhs, when the resulting address in not a function
	    invariant
	 We do not need to do anything special in the latter case (the base of
	 the memory reference whose address is taken may be replaced in the
	 DECL_P case).  The former case is more complicated, as we need to
	 ensure that the new address is still a gimple operand.  Thus, it
	 is not sufficient to replace just the base of the memory reference --
	 we need to move the whole computation of the address out of the
	 loop.  */
      if (!is_gimple_val (t))
	return NULL_TREE;

      *walk_subtrees = 0;
      obj = TREE_OPERAND (t, 0);
      var = get_base_address (obj);
      if (!var || !SSA_VAR_P (var) || DECL_EXTERNAL (var))
	return NULL_TREE;

      addr_type = TREE_TYPE (t);
      addr = take_address_of (obj, addr_type, dta->entry, dta->decl_address,
			      dta->gsi);
      if (dta->gsi == NULL && addr == NULL_TREE)
	{
	  dta->reset = true;
	  return NULL_TREE;
	}
      *tp = addr;

      dta->changed = true;
      return NULL_TREE;
    }

  if (!EXPR_P (t))
    *walk_subtrees = 0;

  return NULL_TREE;
}

/* Moves the references to local variables in STMT at *GSI out of the single
   entry single exit region starting at ENTRY.  DECL_ADDRESS contains
   addresses of the references that had their address taken
   already.  */

static void
eliminate_local_variables_stmt (edge entry, gimple_stmt_iterator *gsi,
				int_tree_htab_type *decl_address)
{
  struct elv_data dta;
  gimple *stmt = gsi_stmt (*gsi);

  memset (&dta.info, '\0', sizeof (dta.info));
  dta.entry = entry;
  dta.decl_address = decl_address;
  dta.changed = false;
  dta.reset = false;

  if (gimple_debug_bind_p (stmt))
    {
      dta.gsi = NULL;
      walk_tree (gimple_debug_bind_get_value_ptr (stmt),
		 eliminate_local_variables_1, &dta.info, NULL);
      if (dta.reset)
	{
	  gimple_debug_bind_reset_value (stmt);
	  dta.changed = true;
	}
    }
  else if (gimple_clobber_p (stmt))
    {
      unlink_stmt_vdef (stmt);
      stmt = gimple_build_nop ();
      gsi_replace (gsi, stmt, false);
      dta.changed = true;
    }
  else
    {
      dta.gsi = gsi;
      walk_gimple_op (stmt, eliminate_local_variables_1, &dta.info);
    }

  if (dta.changed)
    update_stmt (stmt);
}

/* Eliminates the references to local variables from the single entry
   single exit region between the ENTRY and EXIT edges.

   This includes:
   1) Taking address of a local variable -- these are moved out of the
   region (and temporary variable is created to hold the address if
   necessary).

   2) Dereferencing a local variable -- these are replaced with indirect
   references.  */

static void
eliminate_local_variables (edge entry, edge exit)
{
  basic_block bb;
  auto_vec<basic_block, 3> body;
  unsigned i;
  gimple_stmt_iterator gsi;
  bool has_debug_stmt = false;
  int_tree_htab_type decl_address (10);
  basic_block entry_bb = entry->src;
  basic_block exit_bb = exit->dest;

  gather_blocks_in_sese_region (entry_bb, exit_bb, &body);

  FOR_EACH_VEC_ELT (body, i, bb)
    if (bb != entry_bb && bb != exit_bb)
      {
        for (gsi = gsi_start_bb (bb); !gsi_end_p (gsi); gsi_next (&gsi))
	  if (is_gimple_debug (gsi_stmt (gsi)))
	    {
	      if (gimple_debug_bind_p (gsi_stmt (gsi)))
	        has_debug_stmt = true;
	    }
	  else
	    eliminate_local_variables_stmt (entry, &gsi, &decl_address);
      }

  if (has_debug_stmt)
    FOR_EACH_VEC_ELT (body, i, bb)
      if (bb != entry_bb && bb != exit_bb)
	for (gsi = gsi_start_bb (bb); !gsi_end_p (gsi); gsi_next (&gsi))
	  if (gimple_debug_bind_p (gsi_stmt (gsi)))
	    eliminate_local_variables_stmt (entry, &gsi, &decl_address);
}

/* Returns true if expression EXPR is not defined between ENTRY and
   EXIT, i.e. if all its operands are defined outside of the region.  */

static bool
expr_invariant_in_region_p (edge entry, edge exit, tree expr)
{
  basic_block entry_bb = entry->src;
  basic_block exit_bb = exit->dest;
  basic_block def_bb;

  if (is_gimple_min_invariant (expr))
    return true;

  if (TREE_CODE (expr) == SSA_NAME)
    {
      def_bb = gimple_bb (SSA_NAME_DEF_STMT (expr));
      if (def_bb
	  && dominated_by_p (CDI_DOMINATORS, def_bb, entry_bb)
	  && !dominated_by_p (CDI_DOMINATORS, def_bb, exit_bb))
	return false;

      return true;
    }

  return false;
}

/* If COPY_NAME_P is true, creates and returns a duplicate of NAME.
   The copies are stored to NAME_COPIES, if NAME was already duplicated,
   its duplicate stored in NAME_COPIES is returned.

   Regardless of COPY_NAME_P, the decl used as a base of the ssa name is also
   duplicated, storing the copies in DECL_COPIES.  */

static tree
separate_decls_in_region_name (tree name, name_to_copy_table_type *name_copies,
			       int_tree_htab_type *decl_copies,
			       bool copy_name_p)
{
  tree copy, var, var_copy;
  unsigned idx, uid, nuid;
  struct int_tree_map ielt;
  struct name_to_copy_elt elt, *nelt;
  name_to_copy_elt **slot;
  int_tree_map *dslot;

  if (TREE_CODE (name) != SSA_NAME)
    return name;

  idx = SSA_NAME_VERSION (name);
  elt.version = idx;
  slot = name_copies->find_slot_with_hash (&elt, idx,
					   copy_name_p ? INSERT : NO_INSERT);
  if (slot && *slot)
    return (*slot)->new_name;

  if (copy_name_p)
    {
      copy = duplicate_ssa_name (name, NULL);
      nelt = XNEW (struct name_to_copy_elt);
      nelt->version = idx;
      nelt->new_name = copy;
      nelt->field = NULL_TREE;
      *slot = nelt;
    }
  else
    {
      gcc_assert (!slot);
      copy = name;
    }

  var = SSA_NAME_VAR (name);
  if (!var)
    return copy;

  uid = DECL_UID (var);
  ielt.uid = uid;
  dslot = decl_copies->find_slot_with_hash (ielt, uid, INSERT);
  if (!dslot->to)
    {
      var_copy = create_tmp_var (TREE_TYPE (var), get_name (var));
      DECL_GIMPLE_REG_P (var_copy) = DECL_GIMPLE_REG_P (var);
      dslot->uid = uid;
      dslot->to = var_copy;

      /* Ensure that when we meet this decl next time, we won't duplicate
         it again.  */
      nuid = DECL_UID (var_copy);
      ielt.uid = nuid;
      dslot = decl_copies->find_slot_with_hash (ielt, nuid, INSERT);
      gcc_assert (!dslot->to);
      dslot->uid = nuid;
      dslot->to = var_copy;
    }
  else
    var_copy = dslot->to;

  replace_ssa_name_symbol (copy, var_copy);
  return copy;
}

/* Finds the ssa names used in STMT that are defined outside the
   region between ENTRY and EXIT and replaces such ssa names with
   their duplicates.  The duplicates are stored to NAME_COPIES.  Base
   decls of all ssa names used in STMT (including those defined in
   LOOP) are replaced with the new temporary variables; the
   replacement decls are stored in DECL_COPIES.  */

static void
separate_decls_in_region_stmt (edge entry, edge exit, gimple *stmt,
			       name_to_copy_table_type *name_copies,
			       int_tree_htab_type *decl_copies)
{
  use_operand_p use;
  def_operand_p def;
  ssa_op_iter oi;
  tree name, copy;
  bool copy_name_p;

  FOR_EACH_PHI_OR_STMT_DEF (def, stmt, oi, SSA_OP_DEF)
  {
    name = DEF_FROM_PTR (def);
    gcc_assert (TREE_CODE (name) == SSA_NAME);
    copy = separate_decls_in_region_name (name, name_copies, decl_copies,
					  false);
    gcc_assert (copy == name);
  }

  FOR_EACH_PHI_OR_STMT_USE (use, stmt, oi, SSA_OP_USE)
  {
    name = USE_FROM_PTR (use);
    if (TREE_CODE (name) != SSA_NAME)
      continue;

    copy_name_p = expr_invariant_in_region_p (entry, exit, name);
    copy = separate_decls_in_region_name (name, name_copies, decl_copies,
					  copy_name_p);
    SET_USE (use, copy);
  }
}

/* Finds the ssa names used in STMT that are defined outside the
   region between ENTRY and EXIT and replaces such ssa names with
   their duplicates.  The duplicates are stored to NAME_COPIES.  Base
   decls of all ssa names used in STMT (including those defined in
   LOOP) are replaced with the new temporary variables; the
   replacement decls are stored in DECL_COPIES.  */

static bool
separate_decls_in_region_debug (gimple *stmt,
				name_to_copy_table_type *name_copies,
				int_tree_htab_type *decl_copies)
{
  use_operand_p use;
  ssa_op_iter oi;
  tree var, name;
  struct int_tree_map ielt;
  struct name_to_copy_elt elt;
  name_to_copy_elt **slot;
  int_tree_map *dslot;

  if (gimple_debug_bind_p (stmt))
    var = gimple_debug_bind_get_var (stmt);
  else if (gimple_debug_source_bind_p (stmt))
    var = gimple_debug_source_bind_get_var (stmt);
  else
    return true;
  if (TREE_CODE (var) == DEBUG_EXPR_DECL || TREE_CODE (var) == LABEL_DECL)
    return true;
  gcc_assert (DECL_P (var) && SSA_VAR_P (var));
  ielt.uid = DECL_UID (var);
  dslot = decl_copies->find_slot_with_hash (ielt, ielt.uid, NO_INSERT);
  if (!dslot)
    return true;
  if (gimple_debug_bind_p (stmt))
    gimple_debug_bind_set_var (stmt, dslot->to);
  else if (gimple_debug_source_bind_p (stmt))
    gimple_debug_source_bind_set_var (stmt, dslot->to);

  FOR_EACH_PHI_OR_STMT_USE (use, stmt, oi, SSA_OP_USE)
  {
    name = USE_FROM_PTR (use);
    if (TREE_CODE (name) != SSA_NAME)
      continue;

    elt.version = SSA_NAME_VERSION (name);
    slot = name_copies->find_slot_with_hash (&elt, elt.version, NO_INSERT);
    if (!slot)
      {
	gimple_debug_bind_reset_value (stmt);
	update_stmt (stmt);
	break;
      }

    SET_USE (use, (*slot)->new_name);
  }

  return false;
}

/* Callback for htab_traverse.  Adds a field corresponding to the reduction
   specified in SLOT. The type is passed in DATA.  */

int
add_field_for_reduction (reduction_info **slot, tree type)
{

  struct reduction_info *const red = *slot;
  tree var = reduc_stmt_res (red->reduc_stmt);
  tree field = build_decl (gimple_location (red->reduc_stmt), FIELD_DECL,
			   SSA_NAME_IDENTIFIER (var), TREE_TYPE (var));

  insert_field_into_struct (type, field);

  red->field = field;

  return 1;
}

/* Callback for htab_traverse.  Adds a field corresponding to a ssa name
   described in SLOT. The type is passed in DATA.  */

int
add_field_for_name (name_to_copy_elt **slot, tree type)
{
  struct name_to_copy_elt *const elt = *slot;
  tree name = ssa_name (elt->version);
  tree field = build_decl (UNKNOWN_LOCATION,
			   FIELD_DECL, SSA_NAME_IDENTIFIER (name),
			   TREE_TYPE (name));

  insert_field_into_struct (type, field);
  elt->field = field;

  return 1;
}

/* Callback for htab_traverse.  A local result is the intermediate result
   computed by a single
   thread, or the initial value in case no iteration was executed.
   This function creates a phi node reflecting these values.
   The phi's result will be stored in NEW_PHI field of the
   reduction's data structure.  */

int
create_phi_for_local_result (reduction_info **slot, class loop *loop)
{
  struct reduction_info *const reduc = *slot;
  edge e;
  gphi *new_phi;
  basic_block store_bb, continue_bb;
  tree local_res;
  location_t locus;

  /* STORE_BB is the block where the phi
     should be stored.  It is the destination of the loop exit.
     (Find the fallthru edge from GIMPLE_OMP_CONTINUE).  */
  continue_bb = single_pred (loop->latch);
  store_bb = FALLTHRU_EDGE (continue_bb)->dest;

  /* STORE_BB has two predecessors.  One coming from  the loop
     (the reduction's result is computed at the loop),
     and another coming from a block preceding the loop,
     when no iterations
     are executed (the initial value should be taken).  */
  if (EDGE_PRED (store_bb, 0) == FALLTHRU_EDGE (continue_bb))
    e = EDGE_PRED (store_bb, 1);
  else
    e = EDGE_PRED (store_bb, 0);
  tree lhs = reduc_stmt_res (reduc->reduc_stmt);
  local_res = copy_ssa_name (lhs);
  locus = gimple_location (reduc->reduc_stmt);
  new_phi = create_phi_node (local_res, store_bb);
  add_phi_arg (new_phi, reduc->init, e, locus);
  add_phi_arg (new_phi, lhs, FALLTHRU_EDGE (continue_bb), locus);
  reduc->new_phi = new_phi;

  return 1;
}

struct clsn_data
{
  tree store;
  tree load;

  basic_block store_bb;
  basic_block load_bb;
};

/* Callback for htab_traverse.  Create an atomic instruction for the
   reduction described in SLOT.
   DATA annotates the place in memory the atomic operation relates to,
   and the basic block it needs to be generated in.  */

int
create_call_for_reduction_1 (reduction_info **slot, struct clsn_data *clsn_data)
{
  struct reduction_info *const reduc = *slot;
  gimple_stmt_iterator gsi;
  tree type = TREE_TYPE (PHI_RESULT (reduc->reduc_phi));
  tree load_struct;
  basic_block bb;
  basic_block new_bb;
  edge e;
  tree t, addr, ref, x;
  tree tmp_load, name;
  gimple *load;

  if (reduc->reduc_addr == NULL_TREE)
    {
      load_struct = build_simple_mem_ref (clsn_data->load);
      t = build3 (COMPONENT_REF, type, load_struct, reduc->field, NULL_TREE);

      addr = build_addr (t);
    }
  else
    {
      /* Set the address for the atomic store.  */
      addr = reduc->reduc_addr;

      /* Remove the non-atomic store '*addr = sum'.  */
      tree res = PHI_RESULT (reduc->keep_res);
      use_operand_p use_p;
      gimple *stmt;
      bool single_use_p = single_imm_use (res, &use_p, &stmt);
      gcc_assert (single_use_p);
      replace_uses_by (gimple_vdef (stmt),
		       gimple_vuse (stmt));
      gimple_stmt_iterator gsi = gsi_for_stmt (stmt);
      gsi_remove (&gsi, true);
    }

  /* Create phi node.  */
  bb = clsn_data->load_bb;

  gsi = gsi_last_bb (bb);
  e = split_block (bb, gsi_stmt (gsi));
  new_bb = e->dest;

  tmp_load = create_tmp_var (TREE_TYPE (TREE_TYPE (addr)));
  tmp_load = make_ssa_name (tmp_load);
  load = gimple_build_omp_atomic_load (tmp_load, addr,
				       OMP_MEMORY_ORDER_RELAXED);
  SSA_NAME_DEF_STMT (tmp_load) = load;
  gsi = gsi_start_bb (new_bb);
  gsi_insert_after (&gsi, load, GSI_NEW_STMT);

  e = split_block (new_bb, load);
  new_bb = e->dest;
  gsi = gsi_start_bb (new_bb);
  ref = tmp_load;
  x = fold_build2 (reduc->reduction_code,
		   TREE_TYPE (PHI_RESULT (reduc->new_phi)), ref,
		   PHI_RESULT (reduc->new_phi));

  name = force_gimple_operand_gsi (&gsi, x, true, NULL_TREE, true,
				   GSI_CONTINUE_LINKING);

  gimple *store = gimple_build_omp_atomic_store (name,
						 OMP_MEMORY_ORDER_RELAXED);
  gsi_insert_after (&gsi, store, GSI_NEW_STMT);
  return 1;
}

/* Create the atomic operation at the join point of the threads.
   REDUCTION_LIST describes the reductions in the LOOP.
   LD_ST_DATA describes the shared data structure where
   shared data is stored in and loaded from.  */
static void
create_call_for_reduction (class loop *loop,
			   reduction_info_table_type *reduction_list,
			   struct clsn_data *ld_st_data)
{
  reduction_list->traverse <class loop *, create_phi_for_local_result> (loop);
  /* Find the fallthru edge from GIMPLE_OMP_CONTINUE.  */
  basic_block continue_bb = single_pred (loop->latch);
  ld_st_data->load_bb = FALLTHRU_EDGE (continue_bb)->dest;
  reduction_list
    ->traverse <struct clsn_data *, create_call_for_reduction_1> (ld_st_data);
}

/* Callback for htab_traverse.  Loads the final reduction value at the
   join point of all threads, and inserts it in the right place.  */

int
create_loads_for_reductions (reduction_info **slot, struct clsn_data *clsn_data)
{
  struct reduction_info *const red = *slot;
  gimple *stmt;
  gimple_stmt_iterator gsi;
  tree type = TREE_TYPE (reduc_stmt_res (red->reduc_stmt));
  tree load_struct;
  tree name;
  tree x;

  /* If there's no exit phi, the result of the reduction is unused.  */
  if (red->keep_res == NULL)
    return 1;

  gsi = gsi_after_labels (clsn_data->load_bb);
  load_struct = build_simple_mem_ref (clsn_data->load);
  load_struct = build3 (COMPONENT_REF, type, load_struct, red->field,
			NULL_TREE);

  x = load_struct;
  name = PHI_RESULT (red->keep_res);
  stmt = gimple_build_assign (name, x);

  gsi_insert_after (&gsi, stmt, GSI_NEW_STMT);

  for (gsi = gsi_start_phis (gimple_bb (red->keep_res));
       !gsi_end_p (gsi); gsi_next (&gsi))
    if (gsi_stmt (gsi) == red->keep_res)
      {
	remove_phi_node (&gsi, false);
	return 1;
      }
  gcc_unreachable ();
}

/* Load the reduction result that was stored in LD_ST_DATA.
   REDUCTION_LIST describes the list of reductions that the
   loads should be generated for.  */
static void
create_final_loads_for_reduction (reduction_info_table_type *reduction_list,
				  struct clsn_data *ld_st_data)
{
  gimple_stmt_iterator gsi;
  tree t;
  gimple *stmt;

  gsi = gsi_after_labels (ld_st_data->load_bb);
  t = build_fold_addr_expr (ld_st_data->store);
  stmt = gimple_build_assign (ld_st_data->load, t);

  gsi_insert_before (&gsi, stmt, GSI_NEW_STMT);

  reduction_list
    ->traverse <struct clsn_data *, create_loads_for_reductions> (ld_st_data);

}

/* Callback for htab_traverse.  Store the neutral value for the
  particular reduction's operation, e.g. 0 for PLUS_EXPR,
  1 for MULT_EXPR, etc. into the reduction field.
  The reduction is specified in SLOT. The store information is
  passed in DATA.  */

int
create_stores_for_reduction (reduction_info **slot, struct clsn_data *clsn_data)
{
  struct reduction_info *const red = *slot;
  tree t;
  gimple *stmt;
  gimple_stmt_iterator gsi;
  tree type = TREE_TYPE (reduc_stmt_res (red->reduc_stmt));

  gsi = gsi_last_bb (clsn_data->store_bb);
  t = build3 (COMPONENT_REF, type, clsn_data->store, red->field, NULL_TREE);
  stmt = gimple_build_assign (t, red->initial_value);
  gsi_insert_after (&gsi, stmt, GSI_NEW_STMT);

  return 1;
}

/* Callback for htab_traverse.  Creates loads to a field of LOAD in LOAD_BB and
   store to a field of STORE in STORE_BB for the ssa name and its duplicate
   specified in SLOT.  */

int
create_loads_and_stores_for_name (name_to_copy_elt **slot,
				  struct clsn_data *clsn_data)
{
  struct name_to_copy_elt *const elt = *slot;
  tree t;
  gimple *stmt;
  gimple_stmt_iterator gsi;
  tree type = TREE_TYPE (elt->new_name);
  tree load_struct;

  gsi = gsi_last_bb (clsn_data->store_bb);
  t = build3 (COMPONENT_REF, type, clsn_data->store, elt->field, NULL_TREE);
  stmt = gimple_build_assign (t, ssa_name (elt->version));
  gsi_insert_after (&gsi, stmt, GSI_NEW_STMT);

  gsi = gsi_last_bb (clsn_data->load_bb);
  load_struct = build_simple_mem_ref (clsn_data->load);
  t = build3 (COMPONENT_REF, type, load_struct, elt->field, NULL_TREE);
  stmt = gimple_build_assign (elt->new_name, t);
  gsi_insert_after (&gsi, stmt, GSI_NEW_STMT);

  return 1;
}

/* Moves all the variables used in LOOP and defined outside of it (including
   the initial values of loop phi nodes, and *PER_THREAD if it is a ssa
   name) to a structure created for this purpose.  The code

   while (1)
     {
       use (a);
       use (b);
     }

   is transformed this way:

   bb0:
   old.a = a;
   old.b = b;

   bb1:
   a' = new->a;
   b' = new->b;
   while (1)
     {
       use (a');
       use (b');
     }

   `old' is stored to *ARG_STRUCT and `new' is stored to NEW_ARG_STRUCT.  The
   pointer `new' is intentionally not initialized (the loop will be split to a
   separate function later, and `new' will be initialized from its arguments).
   LD_ST_DATA holds information about the shared data structure used to pass
   information among the threads.  It is initialized here, and
   gen_parallel_loop will pass it to create_call_for_reduction that
   needs this information.  REDUCTION_LIST describes the reductions
   in LOOP.  */

static void
separate_decls_in_region (edge entry, edge exit,
			  reduction_info_table_type *reduction_list,
			  tree *arg_struct, tree *new_arg_struct,
			  struct clsn_data *ld_st_data)

{
  basic_block bb1 = split_edge (entry);
  basic_block bb0 = single_pred (bb1);
  name_to_copy_table_type name_copies (10);
  int_tree_htab_type decl_copies (10);
  unsigned i;
  tree type, type_name, nvar;
  gimple_stmt_iterator gsi;
  struct clsn_data clsn_data;
  auto_vec<basic_block, 3> body;
  basic_block bb;
  basic_block entry_bb = bb1;
  basic_block exit_bb = exit->dest;
  bool has_debug_stmt = false;

  entry = single_succ_edge (entry_bb);
  gather_blocks_in_sese_region (entry_bb, exit_bb, &body);

  FOR_EACH_VEC_ELT (body, i, bb)
    {
      if (bb != entry_bb && bb != exit_bb)
	{
	  for (gsi = gsi_start_phis (bb); !gsi_end_p (gsi); gsi_next (&gsi))
	    separate_decls_in_region_stmt (entry, exit, gsi_stmt (gsi),
					   &name_copies, &decl_copies);

	  for (gsi = gsi_start_bb (bb); !gsi_end_p (gsi); gsi_next (&gsi))
	    {
	      gimple *stmt = gsi_stmt (gsi);

	      if (is_gimple_debug (stmt))
		has_debug_stmt = true;
	      else
		separate_decls_in_region_stmt (entry, exit, stmt,
					       &name_copies, &decl_copies);
	    }
	}
    }

  /* Now process debug bind stmts.  We must not create decls while
     processing debug stmts, so we defer their processing so as to
     make sure we will have debug info for as many variables as
     possible (all of those that were dealt with in the loop above),
     and discard those for which we know there's nothing we can
     do.  */
  if (has_debug_stmt)
    FOR_EACH_VEC_ELT (body, i, bb)
      if (bb != entry_bb && bb != exit_bb)
	{
	  for (gsi = gsi_start_bb (bb); !gsi_end_p (gsi);)
	    {
	      gimple *stmt = gsi_stmt (gsi);

	      if (is_gimple_debug (stmt))
		{
		  if (separate_decls_in_region_debug (stmt, &name_copies,
						      &decl_copies))
		    {
		      gsi_remove (&gsi, true);
		      continue;
		    }
		}

	      gsi_next (&gsi);
	    }
	}

  if (name_copies.is_empty () && reduction_list->is_empty ())
    {
      /* It may happen that there is nothing to copy (if there are only
         loop carried and external variables in the loop).  */
      *arg_struct = NULL;
      *new_arg_struct = NULL;
    }
  else
    {
      /* Create the type for the structure to store the ssa names to.  */
      type = lang_hooks.types.make_type (RECORD_TYPE);
      type_name = build_decl (UNKNOWN_LOCATION,
			      TYPE_DECL, create_tmp_var_name (".paral_data"),
			      type);
      TYPE_NAME (type) = type_name;

      name_copies.traverse <tree, add_field_for_name> (type);
      if (reduction_list && !reduction_list->is_empty ())
	{
	  /* Create the fields for reductions.  */
	  reduction_list->traverse <tree, add_field_for_reduction> (type);
	}
      layout_type (type);

      /* Create the loads and stores.  */
      *arg_struct = create_tmp_var (type, ".paral_data_store");
      nvar = create_tmp_var (build_pointer_type (type), ".paral_data_load");
      *new_arg_struct = make_ssa_name (nvar);

      ld_st_data->store = *arg_struct;
      ld_st_data->load = *new_arg_struct;
      ld_st_data->store_bb = bb0;
      ld_st_data->load_bb = bb1;

      name_copies
	.traverse <struct clsn_data *, create_loads_and_stores_for_name>
		  (ld_st_data);

      /* Load the calculation from memory (after the join of the threads).  */

      if (reduction_list && !reduction_list->is_empty ())
	{
	  reduction_list
	    ->traverse <struct clsn_data *, create_stores_for_reduction>
	    (ld_st_data);
	  clsn_data.load = make_ssa_name (nvar);
	  clsn_data.load_bb = exit->dest;
	  clsn_data.store = ld_st_data->store;
	  create_final_loads_for_reduction (reduction_list, &clsn_data);
	}
    }
}

/* Returns true if FN was created to run in parallel.  */

bool
parallelized_function_p (tree fndecl)
{
  cgraph_node *node = cgraph_node::get (fndecl);
  gcc_assert (node != NULL);
  return node->parallelized_function;
}

/* Creates and returns an empty function that will receive the body of
   a parallelized loop.  */

static tree
create_loop_fn (location_t loc)
{
  char buf[100];
  char *tname;
  tree decl, type, name, t;
  struct function *act_cfun = cfun;
  static unsigned loopfn_num;

  loc = LOCATION_LOCUS (loc);
  snprintf (buf, 100, "%s.$loopfn", current_function_name ());
  ASM_FORMAT_PRIVATE_NAME (tname, buf, loopfn_num++);
  clean_symbol_name (tname);
  name = get_identifier (tname);
  type = build_function_type_list (void_type_node, ptr_type_node, NULL_TREE);

  decl = build_decl (loc, FUNCTION_DECL, name, type);
  TREE_STATIC (decl) = 1;
  TREE_USED (decl) = 1;
  DECL_ARTIFICIAL (decl) = 1;
  DECL_IGNORED_P (decl) = 0;
  TREE_PUBLIC (decl) = 0;
  DECL_UNINLINABLE (decl) = 1;
  DECL_EXTERNAL (decl) = 0;
  DECL_CONTEXT (decl) = NULL_TREE;
  DECL_INITIAL (decl) = make_node (BLOCK);
  BLOCK_SUPERCONTEXT (DECL_INITIAL (decl)) = decl;

  t = build_decl (loc, RESULT_DECL, NULL_TREE, void_type_node);
  DECL_ARTIFICIAL (t) = 1;
  DECL_IGNORED_P (t) = 1;
  DECL_RESULT (decl) = t;

  t = build_decl (loc, PARM_DECL, get_identifier (".paral_data_param"),
		  ptr_type_node);
  DECL_ARTIFICIAL (t) = 1;
  DECL_ARG_TYPE (t) = ptr_type_node;
  DECL_CONTEXT (t) = decl;
  TREE_USED (t) = 1;
  DECL_ARGUMENTS (decl) = t;

  allocate_struct_function (decl, false);
  DECL_STRUCT_FUNCTION (decl)->last_clique = act_cfun->last_clique;

  /* The call to allocate_struct_function clobbers CFUN, so we need to restore
     it.  */
  set_cfun (act_cfun);

  return decl;
}

/* Replace uses of NAME by VAL in block BB.  */

static void
replace_uses_in_bb_by (tree name, tree val, basic_block bb)
{
  gimple *use_stmt;
  imm_use_iterator imm_iter;

  FOR_EACH_IMM_USE_STMT (use_stmt, imm_iter, name)
    {
      if (gimple_bb (use_stmt) != bb)
	continue;

      use_operand_p use_p;
      FOR_EACH_IMM_USE_ON_STMT (use_p, imm_iter)
	SET_USE (use_p, val);
    }
}

/* Do transformation from:

     <bb preheader>:
     ...
     goto <bb header>

     <bb header>:
     ivtmp_a = PHI <ivtmp_init (preheader), ivtmp_b (latch)>
     sum_a = PHI <sum_init (preheader), sum_b (latch)>
     ...
     use (ivtmp_a)
     ...
     sum_b = sum_a + sum_update
     ...
     if (ivtmp_a < n)
       goto <bb latch>;
     else
       goto <bb exit>;

     <bb latch>:
     ivtmp_b = ivtmp_a + 1;
     goto <bb header>

     <bb exit>:
     sum_z = PHI <sum_b (cond[1]), ...>

     [1] Where <bb cond> is single_pred (bb latch); In the simplest case,
	 that's <bb header>.

   to:

     <bb preheader>:
     ...
     goto <bb newheader>

     <bb header>:
     ivtmp_a = PHI <ivtmp_c (latch)>
     sum_a = PHI <sum_c (latch)>
     ...
     use (ivtmp_a)
     ...
     sum_b = sum_a + sum_update
     ...
     goto <bb latch>;

     <bb newheader>:
     ivtmp_c = PHI <ivtmp_init (preheader), ivtmp_b (latch)>
     sum_c = PHI <sum_init (preheader), sum_b (latch)>
     if (ivtmp_c < n + 1)
       goto <bb header>;
     else
       goto <bb newexit>;

     <bb latch>:
     ivtmp_b = ivtmp_a + 1;
     goto <bb newheader>

     <bb newexit>:
     sum_y = PHI <sum_c (newheader)>

     <bb exit>:
     sum_z = PHI <sum_y (newexit), ...>


   In unified diff format:

      <bb preheader>:
      ...
-     goto <bb header>
+     goto <bb newheader>

      <bb header>:
-     ivtmp_a = PHI <ivtmp_init (preheader), ivtmp_b (latch)>
-     sum_a = PHI <sum_init (preheader), sum_b (latch)>
+     ivtmp_a = PHI <ivtmp_c (latch)>
+     sum_a = PHI <sum_c (latch)>
      ...
      use (ivtmp_a)
      ...
      sum_b = sum_a + sum_update
      ...
-     if (ivtmp_a < n)
-       goto <bb latch>;
+     goto <bb latch>;
+
+     <bb newheader>:
+     ivtmp_c = PHI <ivtmp_init (preheader), ivtmp_b (latch)>
+     sum_c = PHI <sum_init (preheader), sum_b (latch)>
+     if (ivtmp_c < n + 1)
+       goto <bb header>;
      else
	goto <bb exit>;

      <bb latch>:
      ivtmp_b = ivtmp_a + 1;
-     goto <bb header>
+     goto <bb newheader>

+    <bb newexit>:
+    sum_y = PHI <sum_c (newheader)>

      <bb exit>:
-     sum_z = PHI <sum_b (cond[1]), ...>
+     sum_z = PHI <sum_y (newexit), ...>

   Note: the example does not show any virtual phis, but these are handled more
   or less as reductions.


   Moves the exit condition of LOOP to the beginning of its header.
   REDUCTION_LIST describes the reductions in LOOP.  BOUND is the new loop
   bound.  */

static void
transform_to_exit_first_loop_alt (class loop *loop,
				  reduction_info_table_type *reduction_list,
				  tree bound)
{
  basic_block header = loop->header;
  basic_block latch = loop->latch;
  edge exit = single_dom_exit (loop);
  basic_block exit_block = exit->dest;
  gcond *cond_stmt = as_a <gcond *> (last_stmt (exit->src));
  tree control = gimple_cond_lhs (cond_stmt);
  edge e;

  /* Rewriting virtuals into loop-closed ssa normal form makes this
     transformation simpler.  It also ensures that the virtuals are in
     loop-closed ssa normal from after the transformation, which is required by
     create_parallel_loop.  */
  rewrite_virtuals_into_loop_closed_ssa (loop);

  /* Create the new_header block.  */
  basic_block new_header = split_block_before_cond_jump (exit->src);
  edge edge_at_split = single_pred_edge (new_header);

  /* Redirect entry edge to new_header.  */
  edge entry = loop_preheader_edge (loop);
  e = redirect_edge_and_branch (entry, new_header);
  gcc_assert (e == entry);

  /* Redirect post_inc_edge to new_header.  */
  edge post_inc_edge = single_succ_edge (latch);
  e = redirect_edge_and_branch (post_inc_edge, new_header);
  gcc_assert (e == post_inc_edge);

  /* Redirect post_cond_edge to header.  */
  edge post_cond_edge = single_pred_edge (latch);
  e = redirect_edge_and_branch (post_cond_edge, header);
  gcc_assert (e == post_cond_edge);

  /* Redirect edge_at_split to latch.  */
  e = redirect_edge_and_branch (edge_at_split, latch);
  gcc_assert (e == edge_at_split);

  /* Set the new loop bound.  */
  gimple_cond_set_rhs (cond_stmt, bound);
  update_stmt (cond_stmt);

  /* Repair the ssa.  */
  vec<edge_var_map> *v = redirect_edge_var_map_vector (post_inc_edge);
  edge_var_map *vm;
  gphi_iterator gsi;
  int i;
  for (gsi = gsi_start_phis (header), i = 0;
       !gsi_end_p (gsi) && v->iterate (i, &vm);
       gsi_next (&gsi), i++)
    {
      gphi *phi = gsi.phi ();
      tree res_a = PHI_RESULT (phi);

      /* Create new phi.  */
      tree res_c = copy_ssa_name (res_a, phi);
      gphi *nphi = create_phi_node (res_c, new_header);

      /* Replace ivtmp_a with ivtmp_c in condition 'if (ivtmp_a < n)'.  */
      replace_uses_in_bb_by (res_a, res_c, new_header);

      /* Replace ivtmp/sum_b with ivtmp/sum_c in header phi.  */
      add_phi_arg (phi, res_c, post_cond_edge, UNKNOWN_LOCATION);

      /* Replace sum_b with sum_c in exit phi.  */
      tree res_b = redirect_edge_var_map_def (vm);
      replace_uses_in_bb_by (res_b, res_c, exit_block);

      struct reduction_info *red = reduction_phi (reduction_list, phi);
      gcc_assert (virtual_operand_p (res_a)
		  || res_a == control
		  || red != NULL);

      if (red)
	{
	  /* Register the new reduction phi.  */
	  red->reduc_phi = nphi;
	  gimple_set_uid (red->reduc_phi, red->reduc_version);
	}
    }
  gcc_assert (gsi_end_p (gsi) && !v->iterate (i, &vm));

  /* Set the preheader argument of the new phis to ivtmp/sum_init.  */
  flush_pending_stmts (entry);

  /* Set the latch arguments of the new phis to ivtmp/sum_b.  */
  flush_pending_stmts (post_inc_edge);


  basic_block new_exit_block = NULL;
  if (!single_pred_p (exit->dest))
    {
      /* Create a new empty exit block, inbetween the new loop header and the
	 old exit block.  The function separate_decls_in_region needs this block
	 to insert code that is active on loop exit, but not any other path.  */
      new_exit_block = split_edge (exit);
    }

  /* Insert and register the reduction exit phis.  */
  for (gphi_iterator gsi = gsi_start_phis (exit_block);
       !gsi_end_p (gsi);
       gsi_next (&gsi))
    {
      gphi *phi = gsi.phi ();
      gphi *nphi = NULL;
      tree res_z = PHI_RESULT (phi);
      tree res_c;

      if (new_exit_block != NULL)
	{
	  /* Now that we have a new exit block, duplicate the phi of the old
	     exit block in the new exit block to preserve loop-closed ssa.  */
	  edge succ_new_exit_block = single_succ_edge (new_exit_block);
	  edge pred_new_exit_block = single_pred_edge (new_exit_block);
	  tree res_y = copy_ssa_name (res_z, phi);
	  nphi = create_phi_node (res_y, new_exit_block);
	  res_c = PHI_ARG_DEF_FROM_EDGE (phi, succ_new_exit_block);
	  add_phi_arg (nphi, res_c, pred_new_exit_block, UNKNOWN_LOCATION);
	  add_phi_arg (phi, res_y, succ_new_exit_block, UNKNOWN_LOCATION);
	}
      else
	res_c = PHI_ARG_DEF_FROM_EDGE (phi, exit);

      if (virtual_operand_p (res_z))
	continue;

      gimple *reduc_phi = SSA_NAME_DEF_STMT (res_c);
      struct reduction_info *red = reduction_phi (reduction_list, reduc_phi);
      if (red != NULL)
	red->keep_res = (nphi != NULL
			 ? nphi
			 : phi);
    }

  /* We're going to cancel the loop at the end of gen_parallel_loop, but until
     then we're still using some fields, so only bother about fields that are
     still used: header and latch.
     The loop has a new header bb, so we update it.  The latch bb stays the
     same.  */
  loop->header = new_header;

  /* Recalculate dominance info.  */
  free_dominance_info (CDI_DOMINATORS);
  calculate_dominance_info (CDI_DOMINATORS);

  checking_verify_ssa (true, true);
}

/* Tries to moves the exit condition of LOOP to the beginning of its header
   without duplication of the loop body.  NIT is the number of iterations of the
   loop.  REDUCTION_LIST describes the reductions in LOOP.  Return true if
   transformation is successful.  */

static bool
try_transform_to_exit_first_loop_alt (class loop *loop,
				      reduction_info_table_type *reduction_list,
				      tree nit)
{
  /* Check whether the latch contains a single statement.  */
  if (!gimple_seq_nondebug_singleton_p (bb_seq (loop->latch)))
    return false;

  /* Check whether the latch contains no phis.  */
  if (phi_nodes (loop->latch) != NULL)
    return false;

  /* Check whether the latch contains the loop iv increment.  */
  edge back = single_succ_edge (loop->latch);
  edge exit = single_dom_exit (loop);
  gcond *cond_stmt = as_a <gcond *> (last_stmt (exit->src));
  tree control = gimple_cond_lhs (cond_stmt);
  gphi *phi = as_a <gphi *> (SSA_NAME_DEF_STMT (control));
  tree inc_res = gimple_phi_arg_def (phi, back->dest_idx);
  if (gimple_bb (SSA_NAME_DEF_STMT (inc_res)) != loop->latch)
    return false;

  /* Check whether there's no code between the loop condition and the latch.  */
  if (!single_pred_p (loop->latch)
      || single_pred (loop->latch) != exit->src)
    return false;

  tree alt_bound = NULL_TREE;
  tree nit_type = TREE_TYPE (nit);

  /* Figure out whether nit + 1 overflows.  */
  if (TREE_CODE (nit) == INTEGER_CST)
    {
      if (!tree_int_cst_equal (nit, TYPE_MAX_VALUE (nit_type)))
	{
	  alt_bound = fold_build2_loc (UNKNOWN_LOCATION, PLUS_EXPR, nit_type,
				       nit, build_one_cst (nit_type));

	  gcc_assert (TREE_CODE (alt_bound) == INTEGER_CST);
	  transform_to_exit_first_loop_alt (loop, reduction_list, alt_bound);
	  return true;
	}
      else
	{
	  /* Todo: Figure out if we can trigger this, if it's worth to handle
	     optimally, and if we can handle it optimally.  */
	  return false;
	}
    }

  gcc_assert (TREE_CODE (nit) == SSA_NAME);

  /* Variable nit is the loop bound as returned by canonicalize_loop_ivs, for an
     iv with base 0 and step 1 that is incremented in the latch, like this:

     <bb header>:
     # iv_1 = PHI <0 (preheader), iv_2 (latch)>
     ...
     if (iv_1 < nit)
       goto <bb latch>;
     else
       goto <bb exit>;

     <bb latch>:
     iv_2 = iv_1 + 1;
     goto <bb header>;

     The range of iv_1 is [0, nit].  The latch edge is taken for
     iv_1 == [0, nit - 1] and the exit edge is taken for iv_1 == nit.  So the
     number of latch executions is equal to nit.

     The function max_loop_iterations gives us the maximum number of latch
     executions, so it gives us the maximum value of nit.  */
  widest_int nit_max;
  if (!max_loop_iterations (loop, &nit_max))
    return false;

  /* Check if nit + 1 overflows.  */
  widest_int type_max = wi::to_widest (TYPE_MAX_VALUE (nit_type));
  if (nit_max >= type_max)
    return false;

  gimple *def = SSA_NAME_DEF_STMT (nit);

  /* Try to find nit + 1, in the form of n in an assignment nit = n - 1.  */
  if (def
      && is_gimple_assign (def)
      && gimple_assign_rhs_code (def) == PLUS_EXPR)
    {
      tree op1 = gimple_assign_rhs1 (def);
      tree op2 = gimple_assign_rhs2 (def);
      if (integer_minus_onep (op1))
	alt_bound = op2;
      else if (integer_minus_onep (op2))
	alt_bound = op1;
    }

  /* If not found, insert nit + 1.  */
  if (alt_bound == NULL_TREE)
    {
      alt_bound = fold_build2 (PLUS_EXPR, nit_type, nit,
			       build_int_cst_type (nit_type, 1));

      gimple_stmt_iterator gsi = gsi_last_bb (loop_preheader_edge (loop)->src);

      alt_bound
	= force_gimple_operand_gsi (&gsi, alt_bound, true, NULL_TREE, false,
				    GSI_CONTINUE_LINKING);
    }

  transform_to_exit_first_loop_alt (loop, reduction_list, alt_bound);
  return true;
}

/* Moves the exit condition of LOOP to the beginning of its header.  NIT is the
   number of iterations of the loop.  REDUCTION_LIST describes the reductions in
   LOOP.  */

static void
transform_to_exit_first_loop (class loop *loop,
			      reduction_info_table_type *reduction_list,
			      tree nit)
{
  basic_block *bbs, *nbbs, ex_bb, orig_header;
  unsigned n;
  bool ok;
  edge exit = single_dom_exit (loop), hpred;
  tree control, control_name, res, t;
  gphi *phi, *nphi;
  gassign *stmt;
  gcond *cond_stmt, *cond_nit;
  tree nit_1;

  split_block_after_labels (loop->header);
  orig_header = single_succ (loop->header);
  hpred = single_succ_edge (loop->header);

  cond_stmt = as_a <gcond *> (last_stmt (exit->src));
  control = gimple_cond_lhs (cond_stmt);
  gcc_assert (gimple_cond_rhs (cond_stmt) == nit);

  /* Make sure that we have phi nodes on exit for all loop header phis
     (create_parallel_loop requires that).  */
  for (gphi_iterator gsi = gsi_start_phis (loop->header);
       !gsi_end_p (gsi);
       gsi_next (&gsi))
    {
      phi = gsi.phi ();
      res = PHI_RESULT (phi);
      t = copy_ssa_name (res, phi);
      SET_PHI_RESULT (phi, t);
      nphi = create_phi_node (res, orig_header);
      add_phi_arg (nphi, t, hpred, UNKNOWN_LOCATION);

      if (res == control)
	{
	  gimple_cond_set_lhs (cond_stmt, t);
	  update_stmt (cond_stmt);
	  control = t;
	}
    }

  bbs = get_loop_body_in_dom_order (loop);

  for (n = 0; bbs[n] != exit->src; n++)
   continue;
  nbbs = XNEWVEC (basic_block, n);
  ok = gimple_duplicate_sese_tail (single_succ_edge (loop->header), exit,
				   bbs + 1, n, nbbs);
  gcc_assert (ok);
  free (bbs);
  ex_bb = nbbs[0];
  free (nbbs);

  /* Other than reductions, the only gimple reg that should be copied
     out of the loop is the control variable.  */
  exit = single_dom_exit (loop);
  control_name = NULL_TREE;
  for (gphi_iterator gsi = gsi_start_phis (ex_bb);
       !gsi_end_p (gsi); )
    {
      phi = gsi.phi ();
      res = PHI_RESULT (phi);
      if (virtual_operand_p (res))
	{
	  gsi_next (&gsi);
	  continue;
	}

      /* Check if it is a part of reduction.  If it is,
         keep the phi at the reduction's keep_res field.  The
         PHI_RESULT of this phi is the resulting value of the reduction
         variable when exiting the loop.  */

      if (!reduction_list->is_empty ())
	{
	  struct reduction_info *red;

	  tree val = PHI_ARG_DEF_FROM_EDGE (phi, exit);
	  red = reduction_phi (reduction_list, SSA_NAME_DEF_STMT (val));
	  if (red)
	    {
	      red->keep_res = phi;
	      gsi_next (&gsi);
	      continue;
	    }
	}
      gcc_assert (control_name == NULL_TREE
		  && SSA_NAME_VAR (res) == SSA_NAME_VAR (control));
      control_name = res;
      remove_phi_node (&gsi, false);
    }
  gcc_assert (control_name != NULL_TREE);

  /* Initialize the control variable to number of iterations
     according to the rhs of the exit condition.  */
  gimple_stmt_iterator gsi = gsi_after_labels (ex_bb);
  cond_nit = as_a <gcond *> (last_stmt (exit->src));
  nit_1 =  gimple_cond_rhs (cond_nit);
  nit_1 = force_gimple_operand_gsi (&gsi,
				  fold_convert (TREE_TYPE (control_name), nit_1),
				  false, NULL_TREE, false, GSI_SAME_STMT);
  stmt = gimple_build_assign (control_name, nit_1);
  gsi_insert_before (&gsi, stmt, GSI_NEW_STMT);
}

/* Create the parallel constructs for LOOP as described in gen_parallel_loop.
   LOOP_FN and DATA are the arguments of GIMPLE_OMP_PARALLEL.
   NEW_DATA is the variable that should be initialized from the argument
   of LOOP_FN.  N_THREADS is the requested number of threads, which can be 0 if
   that number is to be determined later.  */

static void
create_parallel_loop (class loop *loop, tree loop_fn, tree data,
		      tree new_data, unsigned n_threads, location_t loc,
		      bool oacc_kernels_p)
{
  gimple_stmt_iterator gsi;
  basic_block for_bb, ex_bb, continue_bb;
  tree t, param;
  gomp_parallel *omp_par_stmt;
  gimple *omp_return_stmt1, *omp_return_stmt2;
  gimple *phi;
  gcond *cond_stmt;
  gomp_for *for_stmt;
  gomp_continue *omp_cont_stmt;
  tree cvar, cvar_init, initvar, cvar_next, cvar_base, type;
  edge exit, nexit, guard, end, e;

  if (oacc_kernels_p)
    {
      gcc_checking_assert (lookup_attribute ("oacc kernels",
					     DECL_ATTRIBUTES (cfun->decl)));
      /* Indicate to later processing that this is a parallelized OpenACC
	 kernels construct.  */
      DECL_ATTRIBUTES (cfun->decl)
	= tree_cons (get_identifier ("oacc kernels parallelized"),
		     NULL_TREE, DECL_ATTRIBUTES (cfun->decl));
    }
  else
    {
      /* Prepare the GIMPLE_OMP_PARALLEL statement.  */

      basic_block bb = loop_preheader_edge (loop)->src;
      basic_block paral_bb = single_pred (bb);
      gsi = gsi_last_bb (paral_bb);

      gcc_checking_assert (n_threads != 0);
      t = build_omp_clause (loc, OMP_CLAUSE_NUM_THREADS);
      OMP_CLAUSE_NUM_THREADS_EXPR (t)
	= build_int_cst (integer_type_node, n_threads);
      omp_par_stmt = gimple_build_omp_parallel (NULL, t, loop_fn, data);
      gimple_set_location (omp_par_stmt, loc);

      gsi_insert_after (&gsi, omp_par_stmt, GSI_NEW_STMT);

      /* Initialize NEW_DATA.  */
      if (data)
	{
	  gassign *assign_stmt;

	  gsi = gsi_after_labels (bb);

	  param = make_ssa_name (DECL_ARGUMENTS (loop_fn));
	  assign_stmt = gimple_build_assign (param, build_fold_addr_expr (data));
	  gsi_insert_before (&gsi, assign_stmt, GSI_SAME_STMT);

	  assign_stmt = gimple_build_assign (new_data,
					     fold_convert (TREE_TYPE (new_data), param));
	  gsi_insert_before (&gsi, assign_stmt, GSI_SAME_STMT);
	}

      /* Emit GIMPLE_OMP_RETURN for GIMPLE_OMP_PARALLEL.  */
      bb = split_loop_exit_edge (single_dom_exit (loop));
      gsi = gsi_last_bb (bb);
      omp_return_stmt1 = gimple_build_omp_return (false);
      gimple_set_location (omp_return_stmt1, loc);
      gsi_insert_after (&gsi, omp_return_stmt1, GSI_NEW_STMT);
    }

  /* Extract data for GIMPLE_OMP_FOR.  */
  gcc_assert (loop->header == single_dom_exit (loop)->src);
  cond_stmt = as_a <gcond *> (last_stmt (loop->header));

  cvar = gimple_cond_lhs (cond_stmt);
  cvar_base = SSA_NAME_VAR (cvar);
  phi = SSA_NAME_DEF_STMT (cvar);
  cvar_init = PHI_ARG_DEF_FROM_EDGE (phi, loop_preheader_edge (loop));
  initvar = copy_ssa_name (cvar);
  SET_USE (PHI_ARG_DEF_PTR_FROM_EDGE (phi, loop_preheader_edge (loop)),
	   initvar);
  cvar_next = PHI_ARG_DEF_FROM_EDGE (phi, loop_latch_edge (loop));

  gsi = gsi_last_nondebug_bb (loop->latch);
  gcc_assert (gsi_stmt (gsi) == SSA_NAME_DEF_STMT (cvar_next));
  gsi_remove (&gsi, true);

  /* Prepare cfg.  */
  for_bb = split_edge (loop_preheader_edge (loop));
  ex_bb = split_loop_exit_edge (single_dom_exit (loop));
  extract_true_false_edges_from_block (loop->header, &nexit, &exit);
  gcc_assert (exit == single_dom_exit (loop));

  guard = make_edge (for_bb, ex_bb, 0);
  /* FIXME: What is the probability?  */
  guard->probability = profile_probability::guessed_never ();
  /* Split the latch edge, so LOOPS_HAVE_SIMPLE_LATCHES is still valid.  */
  loop->latch = split_edge (single_succ_edge (loop->latch));
  single_pred_edge (loop->latch)->flags = 0;
  end = make_single_succ_edge (single_pred (loop->latch), ex_bb, EDGE_FALLTHRU);
  rescan_loop_exit (end, true, false);

  for (gphi_iterator gpi = gsi_start_phis (ex_bb);
       !gsi_end_p (gpi); gsi_next (&gpi))
    {
      location_t locus;
      gphi *phi = gpi.phi ();
      tree def = PHI_ARG_DEF_FROM_EDGE (phi, exit);
      gimple *def_stmt = SSA_NAME_DEF_STMT (def);

      /* If the exit phi is not connected to a header phi in the same loop, this
	 value is not modified in the loop, and we're done with this phi.  */
      if (!(gimple_code (def_stmt) == GIMPLE_PHI
	    && gimple_bb (def_stmt) == loop->header))
	{
	  locus = gimple_phi_arg_location_from_edge (phi, exit);
	  add_phi_arg (phi, def, guard, locus);
	  add_phi_arg (phi, def, end, locus);
	  continue;
	}

      gphi *stmt = as_a <gphi *> (def_stmt);
      def = PHI_ARG_DEF_FROM_EDGE (stmt, loop_preheader_edge (loop));
      locus = gimple_phi_arg_location_from_edge (stmt,
						 loop_preheader_edge (loop));
      add_phi_arg (phi, def, guard, locus);

      def = PHI_ARG_DEF_FROM_EDGE (stmt, loop_latch_edge (loop));
      locus = gimple_phi_arg_location_from_edge (stmt, loop_latch_edge (loop));
      add_phi_arg (phi, def, end, locus);
    }
  e = redirect_edge_and_branch (exit, nexit->dest);
  PENDING_STMT (e) = NULL;

  /* Emit GIMPLE_OMP_FOR.  */
  if (oacc_kernels_p)
    /* Parallelized OpenACC kernels constructs use gang parallelism.  See also
       omp-offload.c:execute_oacc_device_lower.  */
    t = build_omp_clause (loc, OMP_CLAUSE_GANG);
  else
    {
      t = build_omp_clause (loc, OMP_CLAUSE_SCHEDULE);
      int chunk_size = param_parloops_chunk_size;
      switch (param_parloops_schedule)
	{
	case PARLOOPS_SCHEDULE_STATIC:
	  OMP_CLAUSE_SCHEDULE_KIND (t) = OMP_CLAUSE_SCHEDULE_STATIC;
	  break;
	case PARLOOPS_SCHEDULE_DYNAMIC:
	  OMP_CLAUSE_SCHEDULE_KIND (t) = OMP_CLAUSE_SCHEDULE_DYNAMIC;
	  break;
	case PARLOOPS_SCHEDULE_GUIDED:
	  OMP_CLAUSE_SCHEDULE_KIND (t) = OMP_CLAUSE_SCHEDULE_GUIDED;
	  break;
	case PARLOOPS_SCHEDULE_AUTO:
	  OMP_CLAUSE_SCHEDULE_KIND (t) = OMP_CLAUSE_SCHEDULE_AUTO;
	  chunk_size = 0;
	  break;
	case PARLOOPS_SCHEDULE_RUNTIME:
	  OMP_CLAUSE_SCHEDULE_KIND (t) = OMP_CLAUSE_SCHEDULE_RUNTIME;
	  chunk_size = 0;
	  break;
	default:
	  gcc_unreachable ();
	}
      if (chunk_size != 0)
	OMP_CLAUSE_SCHEDULE_CHUNK_EXPR (t)
	  = build_int_cst (integer_type_node, chunk_size);
    }

  for_stmt = gimple_build_omp_for (NULL,
				   (oacc_kernels_p
				    ? GF_OMP_FOR_KIND_OACC_LOOP
				    : GF_OMP_FOR_KIND_FOR),
				   t, 1, NULL);

  gimple_cond_set_lhs (cond_stmt, cvar_base);
  type = TREE_TYPE (cvar);
  gimple_set_location (for_stmt, loc);
  gimple_omp_for_set_index (for_stmt, 0, initvar);
  gimple_omp_for_set_initial (for_stmt, 0, cvar_init);
  gimple_omp_for_set_final (for_stmt, 0, gimple_cond_rhs (cond_stmt));
  gimple_omp_for_set_cond (for_stmt, 0, gimple_cond_code (cond_stmt));
  gimple_omp_for_set_incr (for_stmt, 0, build2 (PLUS_EXPR, type,
						cvar_base,
						build_int_cst (type, 1)));

  gsi = gsi_last_bb (for_bb);
  gsi_insert_after (&gsi, for_stmt, GSI_NEW_STMT);
  SSA_NAME_DEF_STMT (initvar) = for_stmt;

  /* Emit GIMPLE_OMP_CONTINUE.  */
  continue_bb = single_pred (loop->latch);
  gsi = gsi_last_bb (continue_bb);
  omp_cont_stmt = gimple_build_omp_continue (cvar_next, cvar);
  gimple_set_location (omp_cont_stmt, loc);
  gsi_insert_after (&gsi, omp_cont_stmt, GSI_NEW_STMT);
  SSA_NAME_DEF_STMT (cvar_next) = omp_cont_stmt;

  /* Emit GIMPLE_OMP_RETURN for GIMPLE_OMP_FOR.  */
  gsi = gsi_last_bb (ex_bb);
  omp_return_stmt2 = gimple_build_omp_return (true);
  gimple_set_location (omp_return_stmt2, loc);
  gsi_insert_after (&gsi, omp_return_stmt2, GSI_NEW_STMT);

  /* After the above dom info is hosed.  Re-compute it.  */
  free_dominance_info (CDI_DOMINATORS);
  calculate_dominance_info (CDI_DOMINATORS);
}

/* Return number of phis in bb.  If COUNT_VIRTUAL_P is false, don't count the
   virtual phi.  */

static unsigned int
num_phis (basic_block bb, bool count_virtual_p)
{
  unsigned int nr_phis = 0;
  gphi_iterator gsi;
  for (gsi = gsi_start_phis (bb); !gsi_end_p (gsi); gsi_next (&gsi))
    {
      if (!count_virtual_p && virtual_operand_p (PHI_RESULT (gsi.phi ())))
	continue;

      nr_phis++;
    }

  return nr_phis;
}

/* Generates code to execute the iterations of LOOP in N_THREADS
   threads in parallel, which can be 0 if that number is to be determined
   later.

   NITER describes number of iterations of LOOP.
   REDUCTION_LIST describes the reductions existent in the LOOP.  */

static void
gen_parallel_loop (class loop *loop,
		   reduction_info_table_type *reduction_list,
		   unsigned n_threads, class tree_niter_desc *niter,
		   bool oacc_kernels_p)
{
  tree many_iterations_cond, type, nit;
  tree arg_struct, new_arg_struct;
  gimple_seq stmts;
  edge entry, exit;
  struct clsn_data clsn_data;
  location_t loc;
  gimple *cond_stmt;
  unsigned int m_p_thread=2;

  /* From

     ---------------------------------------------------------------------
     loop
       {
	 IV = phi (INIT, IV + STEP)
	 BODY1;
	 if (COND)
	   break;
	 BODY2;
       }
     ---------------------------------------------------------------------

     with # of iterations NITER (possibly with MAY_BE_ZERO assumption),
     we generate the following code:

     ---------------------------------------------------------------------

     if (MAY_BE_ZERO
     || NITER < MIN_PER_THREAD * N_THREADS)
     goto original;

     BODY1;
     store all local loop-invariant variables used in body of the loop to DATA.
     GIMPLE_OMP_PARALLEL (OMP_CLAUSE_NUM_THREADS (N_THREADS), LOOPFN, DATA);
     load the variables from DATA.
     GIMPLE_OMP_FOR (IV = INIT; COND; IV += STEP) (OMP_CLAUSE_SCHEDULE (static))
     BODY2;
     BODY1;
     GIMPLE_OMP_CONTINUE;
     GIMPLE_OMP_RETURN         -- GIMPLE_OMP_FOR
     GIMPLE_OMP_RETURN         -- GIMPLE_OMP_PARALLEL
     goto end;

     original:
     loop
       {
	 IV = phi (INIT, IV + STEP)
	 BODY1;
	 if (COND)
	   break;
	 BODY2;
       }

     end:

   */

  /* Create two versions of the loop -- in the old one, we know that the
     number of iterations is large enough, and we will transform it into the
     loop that will be split to loop_fn, the new one will be used for the
     remaining iterations.  */

  /* We should compute a better number-of-iterations value for outer loops.
     That is, if we have
 
    for (i = 0; i < n; ++i)
      for (j = 0; j < m; ++j)
        ...

    we should compute nit = n * m, not nit = n.  
    Also may_be_zero handling would need to be adjusted.  */

  type = TREE_TYPE (niter->niter);
  nit = force_gimple_operand (unshare_expr (niter->niter), &stmts, true,
			      NULL_TREE);
  if (stmts)
    gsi_insert_seq_on_edge_immediate (loop_preheader_edge (loop), stmts);

  if (!oacc_kernels_p)
    {
      if (loop->inner)
	m_p_thread=2;
      else
	m_p_thread=MIN_PER_THREAD;

      gcc_checking_assert (n_threads != 0);
      many_iterations_cond =
	fold_build2 (GE_EXPR, boolean_type_node,
		     nit, build_int_cst (type, m_p_thread * n_threads - 1));

      many_iterations_cond
	= fold_build2 (TRUTH_AND_EXPR, boolean_type_node,
		       invert_truthvalue (unshare_expr (niter->may_be_zero)),
		       many_iterations_cond);
      many_iterations_cond
	= force_gimple_operand (many_iterations_cond, &stmts, false, NULL_TREE);
      if (stmts)
	gsi_insert_seq_on_edge_immediate (loop_preheader_edge (loop), stmts);
      if (!is_gimple_condexpr (many_iterations_cond))
	{
	  many_iterations_cond
	    = force_gimple_operand (many_iterations_cond, &stmts,
				    true, NULL_TREE);
	  if (stmts)
	    gsi_insert_seq_on_edge_immediate (loop_preheader_edge (loop),
					      stmts);
	}

      initialize_original_copy_tables ();

      /* We assume that the loop usually iterates a lot.  */
      loop_version (loop, many_iterations_cond, NULL,
		    profile_probability::likely (),
		    profile_probability::unlikely (),
		    profile_probability::likely (),
		    profile_probability::unlikely (), true);
      update_ssa (TODO_update_ssa);
      free_original_copy_tables ();
    }

  /* Base all the induction variables in LOOP on a single control one.  */
  canonicalize_loop_ivs (loop, &nit, true);
  if (num_phis (loop->header, false) != reduction_list->elements () + 1)
    {
      /* The call to canonicalize_loop_ivs above failed to "base all the
	 induction variables in LOOP on a single control one".  Do damage
	 control.  */
      basic_block preheader = loop_preheader_edge (loop)->src;
      basic_block cond_bb = single_pred (preheader);
      gcond *cond = as_a <gcond *> (gsi_stmt (gsi_last_bb (cond_bb)));
      gimple_cond_make_true (cond);
      update_stmt (cond);
      /* We've gotten rid of the duplicate loop created by loop_version, but
	 we can't undo whatever canonicalize_loop_ivs has done.
	 TODO: Fix this properly by ensuring that the call to
	 canonicalize_loop_ivs succeeds.  */
      if (dump_file
	  && (dump_flags & TDF_DETAILS))
	fprintf (dump_file, "canonicalize_loop_ivs failed for loop %d,"
		 " aborting transformation\n", loop->num);
      return;
    }

  /* Ensure that the exit condition is the first statement in the loop.
     The common case is that latch of the loop is empty (apart from the
     increment) and immediately follows the loop exit test.  Attempt to move the
     entry of the loop directly before the exit check and increase the number of
     iterations of the loop by one.  */
  if (try_transform_to_exit_first_loop_alt (loop, reduction_list, nit))
    {
      if (dump_file
	  && (dump_flags & TDF_DETAILS))
	fprintf (dump_file,
		 "alternative exit-first loop transform succeeded"
		 " for loop %d\n", loop->num);
    }
  else
    {
      if (oacc_kernels_p)
	n_threads = 1;

      /* Fall back on the method that handles more cases, but duplicates the
	 loop body: move the exit condition of LOOP to the beginning of its
	 header, and duplicate the part of the last iteration that gets disabled
	 to the exit of the loop.  */
      transform_to_exit_first_loop (loop, reduction_list, nit);
    }

  /* Generate initializations for reductions.  */
  if (!reduction_list->is_empty ())
    reduction_list->traverse <class loop *, initialize_reductions> (loop);

  /* Eliminate the references to local variables from the loop.  */
  gcc_assert (single_exit (loop));
  entry = loop_preheader_edge (loop);
  exit = single_dom_exit (loop);

  /* This rewrites the body in terms of new variables.  This has already
     been done for oacc_kernels_p in pass_lower_omp/lower_omp ().  */
  if (!oacc_kernels_p)
    {
      eliminate_local_variables (entry, exit);
      /* In the old loop, move all variables non-local to the loop to a
	 structure and back, and create separate decls for the variables used in
	 loop.  */
      separate_decls_in_region (entry, exit, reduction_list, &arg_struct,
				&new_arg_struct, &clsn_data);
    }
  else
    {
      arg_struct = NULL_TREE;
      new_arg_struct = NULL_TREE;
      clsn_data.load = NULL_TREE;
      clsn_data.load_bb = exit->dest;
      clsn_data.store = NULL_TREE;
      clsn_data.store_bb = NULL;
    }

  /* Create the parallel constructs.  */
  loc = UNKNOWN_LOCATION;
  cond_stmt = last_stmt (loop->header);
  if (cond_stmt)
    loc = gimple_location (cond_stmt);
  create_parallel_loop (loop, create_loop_fn (loc), arg_struct, new_arg_struct,
			n_threads, loc, oacc_kernels_p);
  if (!reduction_list->is_empty ())
    create_call_for_reduction (loop, reduction_list, &clsn_data);

  scev_reset ();

  /* Free loop bound estimations that could contain references to
     removed statements.  */
  free_numbers_of_iterations_estimates (cfun);
}

/* Returns true when LOOP contains vector phi nodes.  */

static bool
loop_has_vector_phi_nodes (class loop *loop ATTRIBUTE_UNUSED)
{
  unsigned i;
  basic_block *bbs = get_loop_body_in_dom_order (loop);
  gphi_iterator gsi;
  bool res = true;

  for (i = 0; i < loop->num_nodes; i++)
    for (gsi = gsi_start_phis (bbs[i]); !gsi_end_p (gsi); gsi_next (&gsi))
      if (TREE_CODE (TREE_TYPE (PHI_RESULT (gsi.phi ()))) == VECTOR_TYPE)
	goto end;

  res = false;
 end:
  free (bbs);
  return res;
}

/* Create a reduction_info struct, initialize it with REDUC_STMT
   and PHI, insert it to the REDUCTION_LIST.  */

static void
build_new_reduction (reduction_info_table_type *reduction_list,
		     gimple *reduc_stmt, gphi *phi)
{
  reduction_info **slot;
  struct reduction_info *new_reduction;
  enum tree_code reduction_code;

  gcc_assert (reduc_stmt);

  if (gimple_code (reduc_stmt) == GIMPLE_PHI)
    {
      tree op1 = PHI_ARG_DEF (reduc_stmt, 0);
      gimple *def1 = SSA_NAME_DEF_STMT (op1);
      reduction_code = gimple_assign_rhs_code (def1);
    }
  else
    reduction_code = gimple_assign_rhs_code (reduc_stmt);
  /* Check for OpenMP supported reduction.  */
  switch (reduction_code)
    {
    case PLUS_EXPR:
    case MULT_EXPR:
    case MAX_EXPR:
    case MIN_EXPR:
    case BIT_IOR_EXPR:
    case BIT_XOR_EXPR:
    case BIT_AND_EXPR:
    case TRUTH_OR_EXPR:
    case TRUTH_XOR_EXPR:
    case TRUTH_AND_EXPR:
      break;
    default:
      return;
    }

  if (dump_file && (dump_flags & TDF_DETAILS))
    {
      fprintf (dump_file,
	       "Detected reduction. reduction stmt is:\n");
      print_gimple_stmt (dump_file, reduc_stmt, 0);
      fprintf (dump_file, "\n");
    }

  new_reduction = XCNEW (struct reduction_info);

  new_reduction->reduc_stmt = reduc_stmt;
  new_reduction->reduc_phi = phi;
  new_reduction->reduc_version = SSA_NAME_VERSION (gimple_phi_result (phi));
  new_reduction->reduction_code = reduction_code;
  slot = reduction_list->find_slot (new_reduction, INSERT);
  *slot = new_reduction;
}

/* Callback for htab_traverse.  Sets gimple_uid of reduc_phi stmts.  */

int
set_reduc_phi_uids (reduction_info **slot, void *data ATTRIBUTE_UNUSED)
{
  struct reduction_info *const red = *slot;
  gimple_set_uid (red->reduc_phi, red->reduc_version);
  return 1;
}

/* Return true if the type of reduction performed by STMT_INFO is suitable
   for this pass.  */

static bool
valid_reduction_p (stmt_vec_info stmt_info)
{
  /* Parallelization would reassociate the operation, which isn't
     allowed for in-order reductions.  */
  vect_reduction_type reduc_type = STMT_VINFO_REDUC_TYPE (stmt_info);
  return reduc_type != FOLD_LEFT_REDUCTION;
}

/* Detect all reductions in the LOOP, insert them into REDUCTION_LIST.  */

static void
gather_scalar_reductions (loop_p loop, reduction_info_table_type *reduction_list)
{
  gphi_iterator gsi;
  loop_vec_info simple_loop_info;
  auto_vec<gphi *, 4> double_reduc_phis;
  auto_vec<gimple *, 4> double_reduc_stmts;

  vec_info_shared shared;
  simple_loop_info = vect_analyze_loop_form (loop, &shared);
  if (simple_loop_info == NULL)
    goto gather_done;

  for (gsi = gsi_start_phis (loop->header); !gsi_end_p (gsi); gsi_next (&gsi))
    {
      gphi *phi = gsi.phi ();
      affine_iv iv;
      tree res = PHI_RESULT (phi);
      bool double_reduc;

      if (virtual_operand_p (res))
	continue;

      if (simple_iv (loop, loop, res, &iv, true))
	continue;

      stmt_vec_info reduc_stmt_info
	= parloops_force_simple_reduction (simple_loop_info,
					   simple_loop_info->lookup_stmt (phi),
					   &double_reduc, true);
      if (!reduc_stmt_info || !valid_reduction_p (reduc_stmt_info))
	continue;

      if (double_reduc)
	{
	  if (loop->inner->inner != NULL)
	    continue;

	  double_reduc_phis.safe_push (phi);
	  double_reduc_stmts.safe_push (reduc_stmt_info->stmt);
	  continue;
	}

      build_new_reduction (reduction_list, reduc_stmt_info->stmt, phi);
    }
  delete simple_loop_info;

  if (!double_reduc_phis.is_empty ())
    {
      vec_info_shared shared;
      simple_loop_info = vect_analyze_loop_form (loop->inner, &shared);
      if (simple_loop_info)
	{
	  gphi *phi;
	  unsigned int i;

	  FOR_EACH_VEC_ELT (double_reduc_phis, i, phi)
	    {
	      affine_iv iv;
	      tree res = PHI_RESULT (phi);
	      bool double_reduc;

	      use_operand_p use_p;
	      gimple *inner_stmt;
	      bool single_use_p = single_imm_use (res, &use_p, &inner_stmt);
	      gcc_assert (single_use_p);
	      if (gimple_code (inner_stmt) != GIMPLE_PHI)
		continue;
	      gphi *inner_phi = as_a <gphi *> (inner_stmt);
	      if (simple_iv (loop->inner, loop->inner, PHI_RESULT (inner_phi),
			     &iv, true))
		continue;

	      stmt_vec_info inner_phi_info
		= simple_loop_info->lookup_stmt (inner_phi);
	      stmt_vec_info inner_reduc_stmt_info
		= parloops_force_simple_reduction (simple_loop_info,
						   inner_phi_info,
						   &double_reduc, true);
	      gcc_assert (!double_reduc);
	      if (!inner_reduc_stmt_info
		  || !valid_reduction_p (inner_reduc_stmt_info))
		continue;

	      build_new_reduction (reduction_list, double_reduc_stmts[i], phi);
	    }
	  delete simple_loop_info;
	}
    }

 gather_done:
  if (reduction_list->is_empty ())
    return;

  /* As gimple_uid is used by the vectorizer in between vect_analyze_loop_form
     and delete simple_loop_info, we can set gimple_uid of reduc_phi stmts only
     now.  */
  basic_block bb;
  FOR_EACH_BB_FN (bb, cfun)
    for (gsi = gsi_start_phis (bb); !gsi_end_p (gsi); gsi_next (&gsi))
      gimple_set_uid (gsi_stmt (gsi), (unsigned int)-1);
  reduction_list->traverse <void *, set_reduc_phi_uids> (NULL);
}

/* Try to initialize NITER for code generation part.  */

static bool
try_get_loop_niter (loop_p loop, class tree_niter_desc *niter)
{
  edge exit = single_dom_exit (loop);

  gcc_assert (exit);

  /* We need to know # of iterations, and there should be no uses of values
     defined inside loop outside of it, unless the values are invariants of
     the loop.  */
  if (!number_of_iterations_exit (loop, exit, niter, false))
    {
      if (dump_file && (dump_flags & TDF_DETAILS))
	fprintf (dump_file, "  FAILED: number of iterations not known\n");
      return false;
    }

  return true;
}

/* Return the default def of the first function argument.  */

static tree
get_omp_data_i_param (void)
{
  tree decl = DECL_ARGUMENTS (cfun->decl);
  gcc_assert (DECL_CHAIN (decl) == NULL_TREE);
  return ssa_default_def (cfun, decl);
}

/* For PHI in loop header of LOOP, look for pattern:

   <bb preheader>
   .omp_data_i = &.omp_data_arr;
   addr = .omp_data_i->sum;
   sum_a = *addr;

   <bb header>:
   sum_b = PHI <sum_a (preheader), sum_c (latch)>

   and return addr.  Otherwise, return NULL_TREE.  */

static tree
find_reduc_addr (class loop *loop, gphi *phi)
{
  edge e = loop_preheader_edge (loop);
  tree arg = PHI_ARG_DEF_FROM_EDGE (phi, e);
  gimple *stmt = SSA_NAME_DEF_STMT (arg);
  if (!gimple_assign_single_p (stmt))
    return NULL_TREE;
  tree memref = gimple_assign_rhs1 (stmt);
  if (TREE_CODE (memref) != MEM_REF)
    return NULL_TREE;
  tree addr = TREE_OPERAND (memref, 0);

  gimple *stmt2 = SSA_NAME_DEF_STMT (addr);
  if (!gimple_assign_single_p (stmt2))
    return NULL_TREE;
  tree compref = gimple_assign_rhs1 (stmt2);
  if (TREE_CODE (compref) != COMPONENT_REF)
    return NULL_TREE;
  tree addr2 = TREE_OPERAND (compref, 0);
  if (TREE_CODE (addr2) != MEM_REF)
    return NULL_TREE;
  addr2 = TREE_OPERAND (addr2, 0);
  if (TREE_CODE (addr2) != SSA_NAME
      || addr2 != get_omp_data_i_param ())
    return NULL_TREE;

  return addr;
}

/* Try to initialize REDUCTION_LIST for code generation part.
   REDUCTION_LIST describes the reductions.  */

static bool
try_create_reduction_list (loop_p loop,
			   reduction_info_table_type *reduction_list,
			   bool oacc_kernels_p)
{
  edge exit = single_dom_exit (loop);
  gphi_iterator gsi;

  gcc_assert (exit);

  /* Try to get rid of exit phis.  */
  final_value_replacement_loop (loop);

  gather_scalar_reductions (loop, reduction_list);


  for (gsi = gsi_start_phis (exit->dest); !gsi_end_p (gsi); gsi_next (&gsi))
    {
      gphi *phi = gsi.phi ();
      struct reduction_info *red;
      imm_use_iterator imm_iter;
      use_operand_p use_p;
      gimple *reduc_phi;
      tree val = PHI_ARG_DEF_FROM_EDGE (phi, exit);

      if (!virtual_operand_p (val))
	{
	  if (TREE_CODE (val) != SSA_NAME)
	    {
	      if (dump_file && (dump_flags & TDF_DETAILS))
		fprintf (dump_file,
			 "  FAILED: exit PHI argument invariant.\n");
	      return false;
	    }

	  if (dump_file && (dump_flags & TDF_DETAILS))
	    {
	      fprintf (dump_file, "phi is ");
	      print_gimple_stmt (dump_file, phi, 0);
	      fprintf (dump_file, "arg of phi to exit:   value ");
	      print_generic_expr (dump_file, val);
	      fprintf (dump_file, " used outside loop\n");
	      fprintf (dump_file,
		       "  checking if it is part of reduction pattern:\n");
	    }
	  if (reduction_list->is_empty ())
	    {
	      if (dump_file && (dump_flags & TDF_DETAILS))
		fprintf (dump_file,
			 "  FAILED: it is not a part of reduction.\n");
	      return false;
	    }
	  reduc_phi = NULL;
	  FOR_EACH_IMM_USE_FAST (use_p, imm_iter, val)
	    {
	      if (!gimple_debug_bind_p (USE_STMT (use_p))
		  && flow_bb_inside_loop_p (loop, gimple_bb (USE_STMT (use_p))))
		{
		  reduc_phi = USE_STMT (use_p);
		  break;
		}
	    }
	  red = reduction_phi (reduction_list, reduc_phi);
	  if (red == NULL)
	    {
	      if (dump_file && (dump_flags & TDF_DETAILS))
		fprintf (dump_file,
			 "  FAILED: it is not a part of reduction.\n");
	      return false;
	    }
	  if (red->keep_res != NULL)
	    {
	      if (dump_file && (dump_flags & TDF_DETAILS))
		fprintf (dump_file,
			 "  FAILED: reduction has multiple exit phis.\n");
	      return false;
	    }
	  red->keep_res = phi;
	  if (dump_file && (dump_flags & TDF_DETAILS))
	    {
	      fprintf (dump_file, "reduction phi is  ");
	      print_gimple_stmt (dump_file, red->reduc_phi, 0);
	      fprintf (dump_file, "reduction stmt is  ");
	      print_gimple_stmt (dump_file, red->reduc_stmt, 0);
	    }
	}
    }

  /* The iterations of the loop may communicate only through bivs whose
     iteration space can be distributed efficiently.  */
  for (gsi = gsi_start_phis (loop->header); !gsi_end_p (gsi); gsi_next (&gsi))
    {
      gphi *phi = gsi.phi ();
      tree def = PHI_RESULT (phi);
      affine_iv iv;

      if (!virtual_operand_p (def) && !simple_iv (loop, loop, def, &iv, true))
	{
	  struct reduction_info *red;

	  red = reduction_phi (reduction_list, phi);
	  if (red == NULL)
	    {
	      if (dump_file && (dump_flags & TDF_DETAILS))
		fprintf (dump_file,
			 "  FAILED: scalar dependency between iterations\n");
	      return false;
	    }
	}
    }

  if (oacc_kernels_p)
    {
      for (gsi = gsi_start_phis (loop->header); !gsi_end_p (gsi);
	   gsi_next (&gsi))
	{
	  gphi *phi = gsi.phi ();
	  tree def = PHI_RESULT (phi);
	  affine_iv iv;

	  if (!virtual_operand_p (def)
	      && !simple_iv (loop, loop, def, &iv, true))
	    {
	      tree addr = find_reduc_addr (loop, phi);
	      if (addr == NULL_TREE)
		return false;
	      struct reduction_info *red = reduction_phi (reduction_list, phi);
	      red->reduc_addr = addr;
	    }
	}
    }

  return true;
}

/* Return true if LOOP contains phis with ADDR_EXPR in args.  */

static bool
loop_has_phi_with_address_arg (class loop *loop)
{
  basic_block *bbs = get_loop_body (loop);
  bool res = false;

  unsigned i, j;
  gphi_iterator gsi;
  for (i = 0; i < loop->num_nodes; i++)
    for (gsi = gsi_start_phis (bbs[i]); !gsi_end_p (gsi); gsi_next (&gsi))
      {
	gphi *phi = gsi.phi ();
	for (j = 0; j < gimple_phi_num_args (phi); j++)
	  {
	    tree arg = gimple_phi_arg_def (phi, j);
	    if (TREE_CODE (arg) == ADDR_EXPR)
	      {
		/* This should be handled by eliminate_local_variables, but that
		   function currently ignores phis.  */
		res = true;
		goto end;
	      }
	  }
      }
 end:
  free (bbs);

  return res;
}

/* Return true if memory ref REF (corresponding to the stmt at GSI in
   REGIONS_BB[I]) conflicts with the statements in REGIONS_BB[I] after gsi,
   or the statements in REGIONS_BB[I + n].  REF_IS_STORE indicates if REF is a
   store.  Ignore conflicts with SKIP_STMT.  */

static bool
ref_conflicts_with_region (gimple_stmt_iterator gsi, ao_ref *ref,
			   bool ref_is_store, vec<basic_block> region_bbs,
			   unsigned int i, gimple *skip_stmt)
{
  basic_block bb = region_bbs[i];
  gsi_next (&gsi);

  while (true)
    {
      for (; !gsi_end_p (gsi);
	   gsi_next (&gsi))
	{
	  gimple *stmt = gsi_stmt (gsi);
	  if (stmt == skip_stmt)
	    {
	      if (dump_file)
		{
		  fprintf (dump_file, "skipping reduction store: ");
		  print_gimple_stmt (dump_file, stmt, 0);
		}
	      continue;
	    }

	  if (!gimple_vdef (stmt)
	      && !gimple_vuse (stmt))
	    continue;

	  if (gimple_code (stmt) == GIMPLE_RETURN)
	    continue;

	  if (ref_is_store)
	    {
	      if (ref_maybe_used_by_stmt_p (stmt, ref))
		{
		  if (dump_file)
		    {
		      fprintf (dump_file, "Stmt ");
		      print_gimple_stmt (dump_file, stmt, 0);
		    }
		  return true;
		}
	    }
	  else
	    {
	      if (stmt_may_clobber_ref_p_1 (stmt, ref))
		{
		  if (dump_file)
		    {
		      fprintf (dump_file, "Stmt ");
		      print_gimple_stmt (dump_file, stmt, 0);
		    }
		  return true;
		}
	    }
	}
      i++;
      if (i == region_bbs.length ())
	break;
      bb = region_bbs[i];
      gsi = gsi_start_bb (bb);
    }

  return false;
}

/* Return true if the bbs in REGION_BBS but not in in_loop_bbs can be executed
   in parallel with REGION_BBS containing the loop.  Return the stores of
   reduction results in REDUCTION_STORES.  */

static bool
oacc_entry_exit_ok_1 (bitmap in_loop_bbs, vec<basic_block> region_bbs,
		      reduction_info_table_type *reduction_list,
		      bitmap reduction_stores)
{
  tree omp_data_i = get_omp_data_i_param ();

  unsigned i;
  basic_block bb;
  FOR_EACH_VEC_ELT (region_bbs, i, bb)
    {
      if (bitmap_bit_p (in_loop_bbs, bb->index))
	continue;

      gimple_stmt_iterator gsi;
      for (gsi = gsi_start_bb (bb); !gsi_end_p (gsi);
	   gsi_next (&gsi))
	{
	  gimple *stmt = gsi_stmt (gsi);
	  gimple *skip_stmt = NULL;

	  if (is_gimple_debug (stmt)
	      || gimple_code (stmt) == GIMPLE_COND)
	    continue;

	  ao_ref ref;
	  bool ref_is_store = false;
	  if (gimple_assign_load_p (stmt))
	    {
	      tree rhs = gimple_assign_rhs1 (stmt);
	      tree base = get_base_address (rhs);
	      if (TREE_CODE (base) == MEM_REF
		  && operand_equal_p (TREE_OPERAND (base, 0), omp_data_i, 0))
		continue;

	      tree lhs = gimple_assign_lhs (stmt);
	      if (TREE_CODE (lhs) == SSA_NAME
		  && has_single_use (lhs))
		{
		  use_operand_p use_p;
		  gimple *use_stmt;
		  struct reduction_info *red;
		  single_imm_use (lhs, &use_p, &use_stmt);
		  if (gimple_code (use_stmt) == GIMPLE_PHI
		      && (red = reduction_phi (reduction_list, use_stmt)))
		    {
		      tree val = PHI_RESULT (red->keep_res);
		      if (has_single_use (val))
			{
			  single_imm_use (val, &use_p, &use_stmt);
			  if (gimple_store_p (use_stmt))
			    {
			      unsigned int id
				= SSA_NAME_VERSION (gimple_vdef (use_stmt));
			      bitmap_set_bit (reduction_stores, id);
			      skip_stmt = use_stmt;
			      if (dump_file)
				{
				  fprintf (dump_file, "found reduction load: ");
				  print_gimple_stmt (dump_file, stmt, 0);
				}
			    }
			}
		    }
		}

	      ao_ref_init (&ref, rhs);
	    }
	  else if (gimple_store_p (stmt))
	    {
	      ao_ref_init (&ref, gimple_assign_lhs (stmt));
	      ref_is_store = true;
	    }
	  else if (gimple_code (stmt) == GIMPLE_OMP_RETURN)
	    continue;
	  else if (!gimple_has_side_effects (stmt)
		   && !gimple_could_trap_p (stmt)
		   && !stmt_could_throw_p (cfun, stmt)
		   && !gimple_vdef (stmt)
		   && !gimple_vuse (stmt))
	    continue;
	  else if (gimple_call_internal_p (stmt, IFN_GOACC_DIM_POS))
	    continue;
	  else if (gimple_code (stmt) == GIMPLE_RETURN)
	    continue;
	  else
	    {
	      if (dump_file)
		{
		  fprintf (dump_file, "Unhandled stmt in entry/exit: ");
		  print_gimple_stmt (dump_file, stmt, 0);
		}
	      return false;
	    }

	  if (ref_conflicts_with_region (gsi, &ref, ref_is_store, region_bbs,
					 i, skip_stmt))
	    {
	      if (dump_file)
		{
		  fprintf (dump_file, "conflicts with entry/exit stmt: ");
		  print_gimple_stmt (dump_file, stmt, 0);
		}
	      return false;
	    }
	}
    }

  return true;
}

/* Find stores inside REGION_BBS and outside IN_LOOP_BBS, and guard them with
   gang_pos == 0, except when the stores are REDUCTION_STORES.  Return true
   if any changes were made.  */

static bool
oacc_entry_exit_single_gang (bitmap in_loop_bbs, vec<basic_block> region_bbs,
			     bitmap reduction_stores)
{
  tree gang_pos = NULL_TREE;
  bool changed = false;

  unsigned i;
  basic_block bb;
  FOR_EACH_VEC_ELT (region_bbs, i, bb)
    {
      if (bitmap_bit_p (in_loop_bbs, bb->index))
	continue;

      gimple_stmt_iterator gsi;
      for (gsi = gsi_start_bb (bb); !gsi_end_p (gsi);)
	{
	  gimple *stmt = gsi_stmt (gsi);

	  if (!gimple_store_p (stmt))
	    {
	      /* Update gsi to point to next stmt.  */
	      gsi_next (&gsi);
	      continue;
	    }

	  if (bitmap_bit_p (reduction_stores,
			    SSA_NAME_VERSION (gimple_vdef (stmt))))
	    {
	      if (dump_file)
		{
		  fprintf (dump_file,
			   "skipped reduction store for single-gang"
			   " neutering: ");
		  print_gimple_stmt (dump_file, stmt, 0);
		}

	      /* Update gsi to point to next stmt.  */
	      gsi_next (&gsi);
	      continue;
	    }

	  changed = true;

	  if (gang_pos == NULL_TREE)
	    {
	      tree arg = build_int_cst (integer_type_node, GOMP_DIM_GANG);
	      gcall *gang_single
		= gimple_build_call_internal (IFN_GOACC_DIM_POS, 1, arg);
	      gang_pos = make_ssa_name (integer_type_node);
	      gimple_call_set_lhs (gang_single, gang_pos);
	      gimple_stmt_iterator start
		= gsi_start_bb (single_succ (ENTRY_BLOCK_PTR_FOR_FN (cfun)));
	      tree vuse = ssa_default_def (cfun, gimple_vop (cfun));
	      gimple_set_vuse (gang_single, vuse);
	      gsi_insert_before (&start, gang_single, GSI_SAME_STMT);
	    }

	  if (dump_file)
	    {
	      fprintf (dump_file,
		       "found store that needs single-gang neutering: ");
	      print_gimple_stmt (dump_file, stmt, 0);
	    }

	  {
	    /* Split block before store.  */
	    gimple_stmt_iterator gsi2 = gsi;
	    gsi_prev (&gsi2);
	    edge e;
	    if (gsi_end_p (gsi2))
	      {
		e = split_block_after_labels (bb);
		gsi2 = gsi_last_bb (bb);
	      }
	    else
	      e = split_block (bb, gsi_stmt (gsi2));
	    basic_block bb2 = e->dest;

	    /* Split block after store.  */
	    gimple_stmt_iterator gsi3 = gsi_start_bb (bb2);
	    edge e2 = split_block (bb2, gsi_stmt (gsi3));
	    basic_block bb3 = e2->dest;

	    gimple *cond
	      = gimple_build_cond (EQ_EXPR, gang_pos, integer_zero_node,
				   NULL_TREE, NULL_TREE);
	    gsi_insert_after (&gsi2, cond, GSI_NEW_STMT);

	    edge e3 = make_edge (bb, bb3, EDGE_FALSE_VALUE);
	    /* FIXME: What is the probability?  */
	    e3->probability = profile_probability::guessed_never ();
	    e->flags = EDGE_TRUE_VALUE;

	    tree vdef = gimple_vdef (stmt);
	    tree vuse = gimple_vuse (stmt);

	    tree phi_res = copy_ssa_name (vdef);
	    gphi *new_phi = create_phi_node (phi_res, bb3);
	    replace_uses_by (vdef, phi_res);
	    add_phi_arg (new_phi, vuse, e3, UNKNOWN_LOCATION);
	    add_phi_arg (new_phi, vdef, e2, UNKNOWN_LOCATION);

	    /* Update gsi to point to next stmt.  */
	    bb = bb3;
	    gsi = gsi_start_bb (bb);
	  }
	}
    }

  return changed;
}

/* Return true if the statements before and after the LOOP can be executed in
   parallel with the function containing the loop.  Resolve conflicting stores
   outside LOOP by guarding them such that only a single gang executes them.  */

static bool
oacc_entry_exit_ok (class loop *loop,
		    reduction_info_table_type *reduction_list)
{
  basic_block *loop_bbs = get_loop_body_in_dom_order (loop);
  vec<basic_block> region_bbs
    = get_all_dominated_blocks (CDI_DOMINATORS, ENTRY_BLOCK_PTR_FOR_FN (cfun));

  bitmap in_loop_bbs = BITMAP_ALLOC (NULL);
  bitmap_clear (in_loop_bbs);
  for (unsigned int i = 0; i < loop->num_nodes; i++)
    bitmap_set_bit (in_loop_bbs, loop_bbs[i]->index);

  bitmap reduction_stores = BITMAP_ALLOC (NULL);
  bool res = oacc_entry_exit_ok_1 (in_loop_bbs, region_bbs, reduction_list,
				   reduction_stores);

  if (res)
    {
      bool changed = oacc_entry_exit_single_gang (in_loop_bbs, region_bbs,
						  reduction_stores);
      if (changed)
	{
	  free_dominance_info (CDI_DOMINATORS);
	  calculate_dominance_info (CDI_DOMINATORS);
	}
    }

  region_bbs.release ();
  free (loop_bbs);

  BITMAP_FREE (in_loop_bbs);
  BITMAP_FREE (reduction_stores);

  return res;
}

/* Detect parallel loops and generate parallel code using libgomp
   primitives.  Returns true if some loop was parallelized, false
   otherwise.  */

static bool
parallelize_loops (bool oacc_kernels_p)
{
  unsigned n_threads;
  bool changed = false;
  class loop *loop;
  class loop *skip_loop = NULL;
  class tree_niter_desc niter_desc;
  struct obstack parloop_obstack;
  HOST_WIDE_INT estimated;

  /* Do not parallelize loops in the functions created by parallelization.  */
  if (!oacc_kernels_p
      && parallelized_function_p (cfun->decl))
    return false;

  /* Do not parallelize loops in offloaded functions.  */
  if (!oacc_kernels_p
      && oacc_get_fn_attrib (cfun->decl) != NULL)
     return false;

  if (cfun->has_nonlocal_label)
    return false;

  /* For OpenACC kernels, n_threads will be determined later; otherwise, it's
     the argument to -ftree-parallelize-loops.  */
  if (oacc_kernels_p)
    n_threads = 0;
  else
    n_threads = flag_tree_parallelize_loops;

  gcc_obstack_init (&parloop_obstack);
  reduction_info_table_type reduction_list (10);

  calculate_dominance_info (CDI_DOMINATORS);

  FOR_EACH_LOOP (loop, 0)
    {
      if (loop == skip_loop)
	{
	  if (!loop->in_oacc_kernels_region
	      && dump_file && (dump_flags & TDF_DETAILS))
	    fprintf (dump_file,
		     "Skipping loop %d as inner loop of parallelized loop\n",
		     loop->num);

	  skip_loop = loop->inner;
	  continue;
	}
      else
	skip_loop = NULL;

      reduction_list.empty ();

      if (oacc_kernels_p)
	{
	  if (!loop->in_oacc_kernels_region)
	    continue;

	  /* Don't try to parallelize inner loops in an oacc kernels region.  */
	  if (loop->inner)
	    skip_loop = loop->inner;

	  if (dump_file && (dump_flags & TDF_DETAILS))
	    fprintf (dump_file,
		     "Trying loop %d with header bb %d in oacc kernels"
		     " region\n", loop->num, loop->header->index);
	}

      if (dump_file && (dump_flags & TDF_DETAILS))
      {
        fprintf (dump_file, "Trying loop %d as candidate\n",loop->num);
	if (loop->inner)
	  fprintf (dump_file, "loop %d is not innermost\n",loop->num);
	else
	  fprintf (dump_file, "loop %d is innermost\n",loop->num);
      }

      if (!single_dom_exit (loop))
      {

        if (dump_file && (dump_flags & TDF_DETAILS))
	  fprintf (dump_file, "loop is !single_dom_exit\n");

	continue;
      }

      if (/* And of course, the loop must be parallelizable.  */
	  !can_duplicate_loop_p (loop)
	  || loop_has_blocks_with_irreducible_flag (loop)
	  || (loop_preheader_edge (loop)->src->flags & BB_IRREDUCIBLE_LOOP)
	  /* FIXME: the check for vector phi nodes could be removed.  */
	  || loop_has_vector_phi_nodes (loop))
	continue;

      estimated = estimated_loop_iterations_int (loop);
      if (estimated == -1)
	estimated = get_likely_max_loop_iterations_int (loop);
      /* FIXME: Bypass this check as graphite doesn't update the
	 count and frequency correctly now.  */
      if (!flag_loop_parallelize_all
	  && !oacc_kernels_p
	  && ((estimated != -1
	       && (estimated
		   < ((HOST_WIDE_INT) n_threads
		      * (loop->inner ? 2 : MIN_PER_THREAD) - 1)))
	      /* Do not bother with loops in cold areas.  */
	      || optimize_loop_nest_for_size_p (loop)))
	continue;

      if (!try_get_loop_niter (loop, &niter_desc))
	continue;

      if (!try_create_reduction_list (loop, &reduction_list, oacc_kernels_p))
	continue;

      if (loop_has_phi_with_address_arg (loop))
	continue;

      if (!loop->can_be_parallel
	  && !loop_parallel_p (loop, &parloop_obstack))
	continue;

      if (oacc_kernels_p
	&& !oacc_entry_exit_ok (loop, &reduction_list))
	{
	  if (dump_file)
	    fprintf (dump_file, "entry/exit not ok: FAILED\n");
	  continue;
	}

      changed = true;
      skip_loop = loop->inner;

      if (dump_enabled_p ())
	{
	  dump_user_location_t loop_loc = find_loop_location (loop);
	  if (loop->inner)
	    dump_printf_loc (MSG_OPTIMIZED_LOCATIONS, loop_loc,
			     "parallelizing outer loop %d\n", loop->num);
	  else
	    dump_printf_loc (MSG_OPTIMIZED_LOCATIONS, loop_loc,
			     "parallelizing inner loop %d\n", loop->num);
	}

      gen_parallel_loop (loop, &reduction_list,
			 n_threads, &niter_desc, oacc_kernels_p);
    }

  obstack_free (&parloop_obstack, NULL);

  /* Parallelization will cause new function calls to be inserted through
     which local variables will escape.  Reset the points-to solution
     for ESCAPED.  */
  if (changed)
    pt_solution_reset (&cfun->gimple_df->escaped);

  return changed;
}

/* Parallelization.  */

namespace {

const pass_data pass_data_parallelize_loops =
{
  GIMPLE_PASS, /* type */
  "parloops", /* name */
  OPTGROUP_LOOP, /* optinfo_flags */
  TV_TREE_PARALLELIZE_LOOPS, /* tv_id */
  ( PROP_cfg | PROP_ssa ), /* properties_required */
  0, /* properties_provided */
  0, /* properties_destroyed */
  0, /* todo_flags_start */
  0, /* todo_flags_finish */
};

class pass_parallelize_loops : public gimple_opt_pass
{
public:
  pass_parallelize_loops (gcc::context *ctxt)
    : gimple_opt_pass (pass_data_parallelize_loops, ctxt),
      oacc_kernels_p (false)
  {}

  /* opt_pass methods: */
  virtual bool gate (function *)
  {
    if (oacc_kernels_p)
      return flag_openacc;
    else
      return flag_tree_parallelize_loops > 1;
  }
  virtual unsigned int execute (function *);
  opt_pass * clone () { return new pass_parallelize_loops (m_ctxt); }
  void set_pass_param (unsigned int n, bool param)
    {
      gcc_assert (n == 0);
      oacc_kernels_p = param;
    }

 private:
  bool oacc_kernels_p;
}; // class pass_parallelize_loops

unsigned
pass_parallelize_loops::execute (function *fun)
{
  tree nthreads = builtin_decl_explicit (BUILT_IN_OMP_GET_NUM_THREADS);
  if (nthreads == NULL_TREE)
    return 0;

  bool in_loop_pipeline = scev_initialized_p ();
  if (!in_loop_pipeline)
    loop_optimizer_init (LOOPS_NORMAL
			 | LOOPS_HAVE_RECORDED_EXITS);

  if (number_of_loops (fun) <= 1)
    return 0;

  if (!in_loop_pipeline)
    {
      rewrite_into_loop_closed_ssa (NULL, TODO_update_ssa);
      scev_initialize ();
    }

  unsigned int todo = 0;
  if (parallelize_loops (oacc_kernels_p))
    {
      fun->curr_properties &= ~(PROP_gimple_eomp);

      checking_verify_loop_structure ();

      todo |= TODO_update_ssa;
    }

  if (!in_loop_pipeline)
    {
      scev_finalize ();
      loop_optimizer_finalize ();
    }

  return todo;
}

} // anon namespace

gimple_opt_pass *
make_pass_parallelize_loops (gcc::context *ctxt)
{
  return new pass_parallelize_loops (ctxt);
}