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
|
/* Optimization of PHI nodes by converting them into straightline code.
Copyright (C) 2004-2018 Free Software Foundation, Inc.
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 "insn-codes.h"
#include "rtl.h"
#include "tree.h"
#include "gimple.h"
#include "cfghooks.h"
#include "tree-pass.h"
#include "ssa.h"
#include "optabs-tree.h"
#include "insn-config.h"
#include "gimple-pretty-print.h"
#include "fold-const.h"
#include "stor-layout.h"
#include "cfganal.h"
#include "gimplify.h"
#include "gimple-iterator.h"
#include "gimplify-me.h"
#include "tree-cfg.h"
#include "tree-dfa.h"
#include "domwalk.h"
#include "cfgloop.h"
#include "tree-data-ref.h"
#include "tree-scalar-evolution.h"
#include "tree-inline.h"
#include "params.h"
#include "case-cfn-macros.h"
static unsigned int tree_ssa_phiopt_worker (bool, bool, bool);
static bool conditional_replacement (basic_block, basic_block,
edge, edge, gphi *, tree, tree);
static gphi *factor_out_conditional_conversion (edge, edge, gphi *, tree, tree,
gimple *);
static int value_replacement (basic_block, basic_block,
edge, edge, gimple *, tree, tree);
static bool minmax_replacement (basic_block, basic_block,
edge, edge, gimple *, tree, tree);
static bool abs_replacement (basic_block, basic_block,
edge, edge, gimple *, tree, tree);
static bool cond_removal_in_popcount_pattern (basic_block, basic_block,
edge, edge, gimple *, tree, tree);
static bool cond_store_replacement (basic_block, basic_block, edge, edge,
hash_set<tree> *);
static bool cond_if_else_store_replacement (basic_block, basic_block, basic_block);
static hash_set<tree> * get_non_trapping ();
static void replace_phi_edge_with_variable (basic_block, edge, gimple *, tree);
static void hoist_adjacent_loads (basic_block, basic_block,
basic_block, basic_block);
static bool gate_hoist_loads (void);
/* This pass tries to transform conditional stores into unconditional
ones, enabling further simplifications with the simpler then and else
blocks. In particular it replaces this:
bb0:
if (cond) goto bb2; else goto bb1;
bb1:
*p = RHS;
bb2:
with
bb0:
if (cond) goto bb1; else goto bb2;
bb1:
condtmp' = *p;
bb2:
condtmp = PHI <RHS, condtmp'>
*p = condtmp;
This transformation can only be done under several constraints,
documented below. It also replaces:
bb0:
if (cond) goto bb2; else goto bb1;
bb1:
*p = RHS1;
goto bb3;
bb2:
*p = RHS2;
bb3:
with
bb0:
if (cond) goto bb3; else goto bb1;
bb1:
bb3:
condtmp = PHI <RHS1, RHS2>
*p = condtmp; */
static unsigned int
tree_ssa_cs_elim (void)
{
unsigned todo;
/* ??? We are not interested in loop related info, but the following
will create it, ICEing as we didn't init loops with pre-headers.
An interfacing issue of find_data_references_in_bb. */
loop_optimizer_init (LOOPS_NORMAL);
scev_initialize ();
todo = tree_ssa_phiopt_worker (true, false, false);
scev_finalize ();
loop_optimizer_finalize ();
return todo;
}
/* Return the singleton PHI in the SEQ of PHIs for edges E0 and E1. */
static gphi *
single_non_singleton_phi_for_edges (gimple_seq seq, edge e0, edge e1)
{
gimple_stmt_iterator i;
gphi *phi = NULL;
if (gimple_seq_singleton_p (seq))
return as_a <gphi *> (gsi_stmt (gsi_start (seq)));
for (i = gsi_start (seq); !gsi_end_p (i); gsi_next (&i))
{
gphi *p = as_a <gphi *> (gsi_stmt (i));
/* If the PHI arguments are equal then we can skip this PHI. */
if (operand_equal_for_phi_arg_p (gimple_phi_arg_def (p, e0->dest_idx),
gimple_phi_arg_def (p, e1->dest_idx)))
continue;
/* If we already have a PHI that has the two edge arguments are
different, then return it is not a singleton for these PHIs. */
if (phi)
return NULL;
phi = p;
}
return phi;
}
/* The core routine of conditional store replacement and normal
phi optimizations. Both share much of the infrastructure in how
to match applicable basic block patterns. DO_STORE_ELIM is true
when we want to do conditional store replacement, false otherwise.
DO_HOIST_LOADS is true when we want to hoist adjacent loads out
of diamond control flow patterns, false otherwise. */
static unsigned int
tree_ssa_phiopt_worker (bool do_store_elim, bool do_hoist_loads, bool early_p)
{
basic_block bb;
basic_block *bb_order;
unsigned n, i;
bool cfgchanged = false;
hash_set<tree> *nontrap = 0;
if (do_store_elim)
/* Calculate the set of non-trapping memory accesses. */
nontrap = get_non_trapping ();
/* Search every basic block for COND_EXPR we may be able to optimize.
We walk the blocks in order that guarantees that a block with
a single predecessor is processed before the predecessor.
This ensures that we collapse inner ifs before visiting the
outer ones, and also that we do not try to visit a removed
block. */
bb_order = single_pred_before_succ_order ();
n = n_basic_blocks_for_fn (cfun) - NUM_FIXED_BLOCKS;
for (i = 0; i < n; i++)
{
gimple *cond_stmt;
gphi *phi;
basic_block bb1, bb2;
edge e1, e2;
tree arg0, arg1;
bb = bb_order[i];
cond_stmt = last_stmt (bb);
/* Check to see if the last statement is a GIMPLE_COND. */
if (!cond_stmt
|| gimple_code (cond_stmt) != GIMPLE_COND)
continue;
e1 = EDGE_SUCC (bb, 0);
bb1 = e1->dest;
e2 = EDGE_SUCC (bb, 1);
bb2 = e2->dest;
/* We cannot do the optimization on abnormal edges. */
if ((e1->flags & EDGE_ABNORMAL) != 0
|| (e2->flags & EDGE_ABNORMAL) != 0)
continue;
/* If either bb1's succ or bb2 or bb2's succ is non NULL. */
if (EDGE_COUNT (bb1->succs) == 0
|| bb2 == NULL
|| EDGE_COUNT (bb2->succs) == 0)
continue;
/* Find the bb which is the fall through to the other. */
if (EDGE_SUCC (bb1, 0)->dest == bb2)
;
else if (EDGE_SUCC (bb2, 0)->dest == bb1)
{
std::swap (bb1, bb2);
std::swap (e1, e2);
}
else if (do_store_elim
&& EDGE_SUCC (bb1, 0)->dest == EDGE_SUCC (bb2, 0)->dest)
{
basic_block bb3 = EDGE_SUCC (bb1, 0)->dest;
if (!single_succ_p (bb1)
|| (EDGE_SUCC (bb1, 0)->flags & EDGE_FALLTHRU) == 0
|| !single_succ_p (bb2)
|| (EDGE_SUCC (bb2, 0)->flags & EDGE_FALLTHRU) == 0
|| EDGE_COUNT (bb3->preds) != 2)
continue;
if (cond_if_else_store_replacement (bb1, bb2, bb3))
cfgchanged = true;
continue;
}
else if (do_hoist_loads
&& EDGE_SUCC (bb1, 0)->dest == EDGE_SUCC (bb2, 0)->dest)
{
basic_block bb3 = EDGE_SUCC (bb1, 0)->dest;
if (!FLOAT_TYPE_P (TREE_TYPE (gimple_cond_lhs (cond_stmt)))
&& single_succ_p (bb1)
&& single_succ_p (bb2)
&& single_pred_p (bb1)
&& single_pred_p (bb2)
&& EDGE_COUNT (bb->succs) == 2
&& EDGE_COUNT (bb3->preds) == 2
/* If one edge or the other is dominant, a conditional move
is likely to perform worse than the well-predicted branch. */
&& !predictable_edge_p (EDGE_SUCC (bb, 0))
&& !predictable_edge_p (EDGE_SUCC (bb, 1)))
hoist_adjacent_loads (bb, bb1, bb2, bb3);
continue;
}
else
continue;
e1 = EDGE_SUCC (bb1, 0);
/* Make sure that bb1 is just a fall through. */
if (!single_succ_p (bb1)
|| (e1->flags & EDGE_FALLTHRU) == 0)
continue;
/* Also make sure that bb1 only have one predecessor and that it
is bb. */
if (!single_pred_p (bb1)
|| single_pred (bb1) != bb)
continue;
if (do_store_elim)
{
/* bb1 is the middle block, bb2 the join block, bb the split block,
e1 the fallthrough edge from bb1 to bb2. We can't do the
optimization if the join block has more than two predecessors. */
if (EDGE_COUNT (bb2->preds) > 2)
continue;
if (cond_store_replacement (bb1, bb2, e1, e2, nontrap))
cfgchanged = true;
}
else
{
gimple_seq phis = phi_nodes (bb2);
gimple_stmt_iterator gsi;
bool candorest = true;
/* Value replacement can work with more than one PHI
so try that first. */
if (!early_p)
for (gsi = gsi_start (phis); !gsi_end_p (gsi); gsi_next (&gsi))
{
phi = as_a <gphi *> (gsi_stmt (gsi));
arg0 = gimple_phi_arg_def (phi, e1->dest_idx);
arg1 = gimple_phi_arg_def (phi, e2->dest_idx);
if (value_replacement (bb, bb1, e1, e2, phi, arg0, arg1) == 2)
{
candorest = false;
cfgchanged = true;
break;
}
}
if (!candorest)
continue;
phi = single_non_singleton_phi_for_edges (phis, e1, e2);
if (!phi)
continue;
arg0 = gimple_phi_arg_def (phi, e1->dest_idx);
arg1 = gimple_phi_arg_def (phi, e2->dest_idx);
/* Something is wrong if we cannot find the arguments in the PHI
node. */
gcc_assert (arg0 != NULL_TREE && arg1 != NULL_TREE);
gphi *newphi = factor_out_conditional_conversion (e1, e2, phi,
arg0, arg1,
cond_stmt);
if (newphi != NULL)
{
phi = newphi;
/* factor_out_conditional_conversion may create a new PHI in
BB2 and eliminate an existing PHI in BB2. Recompute values
that may be affected by that change. */
arg0 = gimple_phi_arg_def (phi, e1->dest_idx);
arg1 = gimple_phi_arg_def (phi, e2->dest_idx);
gcc_assert (arg0 != NULL_TREE && arg1 != NULL_TREE);
}
/* Do the replacement of conditional if it can be done. */
if (!early_p
&& conditional_replacement (bb, bb1, e1, e2, phi, arg0, arg1))
cfgchanged = true;
else if (abs_replacement (bb, bb1, e1, e2, phi, arg0, arg1))
cfgchanged = true;
else if (!early_p
&& cond_removal_in_popcount_pattern (bb, bb1, e1, e2,
phi, arg0, arg1))
cfgchanged = true;
else if (minmax_replacement (bb, bb1, e1, e2, phi, arg0, arg1))
cfgchanged = true;
}
}
free (bb_order);
if (do_store_elim)
delete nontrap;
/* If the CFG has changed, we should cleanup the CFG. */
if (cfgchanged && do_store_elim)
{
/* In cond-store replacement we have added some loads on edges
and new VOPS (as we moved the store, and created a load). */
gsi_commit_edge_inserts ();
return TODO_cleanup_cfg | TODO_update_ssa_only_virtuals;
}
else if (cfgchanged)
return TODO_cleanup_cfg;
return 0;
}
/* Replace PHI node element whose edge is E in block BB with variable NEW.
Remove the edge from COND_BLOCK which does not lead to BB (COND_BLOCK
is known to have two edges, one of which must reach BB). */
static void
replace_phi_edge_with_variable (basic_block cond_block,
edge e, gimple *phi, tree new_tree)
{
basic_block bb = gimple_bb (phi);
basic_block block_to_remove;
gimple_stmt_iterator gsi;
/* Change the PHI argument to new. */
SET_USE (PHI_ARG_DEF_PTR (phi, e->dest_idx), new_tree);
/* Remove the empty basic block. */
if (EDGE_SUCC (cond_block, 0)->dest == bb)
{
EDGE_SUCC (cond_block, 0)->flags |= EDGE_FALLTHRU;
EDGE_SUCC (cond_block, 0)->flags &= ~(EDGE_TRUE_VALUE | EDGE_FALSE_VALUE);
EDGE_SUCC (cond_block, 0)->probability = profile_probability::always ();
block_to_remove = EDGE_SUCC (cond_block, 1)->dest;
}
else
{
EDGE_SUCC (cond_block, 1)->flags |= EDGE_FALLTHRU;
EDGE_SUCC (cond_block, 1)->flags
&= ~(EDGE_TRUE_VALUE | EDGE_FALSE_VALUE);
EDGE_SUCC (cond_block, 1)->probability = profile_probability::always ();
block_to_remove = EDGE_SUCC (cond_block, 0)->dest;
}
delete_basic_block (block_to_remove);
/* Eliminate the COND_EXPR at the end of COND_BLOCK. */
gsi = gsi_last_bb (cond_block);
gsi_remove (&gsi, true);
if (dump_file && (dump_flags & TDF_DETAILS))
fprintf (dump_file,
"COND_EXPR in block %d and PHI in block %d converted to straightline code.\n",
cond_block->index,
bb->index);
}
/* PR66726: Factor conversion out of COND_EXPR. If the arguments of the PHI
stmt are CONVERT_STMT, factor out the conversion and perform the conversion
to the result of PHI stmt. COND_STMT is the controlling predicate.
Return the newly-created PHI, if any. */
static gphi *
factor_out_conditional_conversion (edge e0, edge e1, gphi *phi,
tree arg0, tree arg1, gimple *cond_stmt)
{
gimple *arg0_def_stmt = NULL, *arg1_def_stmt = NULL, *new_stmt;
tree new_arg0 = NULL_TREE, new_arg1 = NULL_TREE;
tree temp, result;
gphi *newphi;
gimple_stmt_iterator gsi, gsi_for_def;
location_t locus = gimple_location (phi);
enum tree_code convert_code;
/* Handle only PHI statements with two arguments. TODO: If all
other arguments to PHI are INTEGER_CST or if their defining
statement have the same unary operation, we can handle more
than two arguments too. */
if (gimple_phi_num_args (phi) != 2)
return NULL;
/* First canonicalize to simplify tests. */
if (TREE_CODE (arg0) != SSA_NAME)
{
std::swap (arg0, arg1);
std::swap (e0, e1);
}
if (TREE_CODE (arg0) != SSA_NAME
|| (TREE_CODE (arg1) != SSA_NAME
&& TREE_CODE (arg1) != INTEGER_CST))
return NULL;
/* Check if arg0 is an SSA_NAME and the stmt which defines arg0 is
a conversion. */
arg0_def_stmt = SSA_NAME_DEF_STMT (arg0);
if (!gimple_assign_cast_p (arg0_def_stmt))
return NULL;
/* Use the RHS as new_arg0. */
convert_code = gimple_assign_rhs_code (arg0_def_stmt);
new_arg0 = gimple_assign_rhs1 (arg0_def_stmt);
if (convert_code == VIEW_CONVERT_EXPR)
{
new_arg0 = TREE_OPERAND (new_arg0, 0);
if (!is_gimple_reg_type (TREE_TYPE (new_arg0)))
return NULL;
}
if (TREE_CODE (arg1) == SSA_NAME)
{
/* Check if arg1 is an SSA_NAME and the stmt which defines arg1
is a conversion. */
arg1_def_stmt = SSA_NAME_DEF_STMT (arg1);
if (!is_gimple_assign (arg1_def_stmt)
|| gimple_assign_rhs_code (arg1_def_stmt) != convert_code)
return NULL;
/* Use the RHS as new_arg1. */
new_arg1 = gimple_assign_rhs1 (arg1_def_stmt);
if (convert_code == VIEW_CONVERT_EXPR)
new_arg1 = TREE_OPERAND (new_arg1, 0);
}
else
{
/* If arg1 is an INTEGER_CST, fold it to new type. */
if (INTEGRAL_TYPE_P (TREE_TYPE (new_arg0))
&& int_fits_type_p (arg1, TREE_TYPE (new_arg0)))
{
if (gimple_assign_cast_p (arg0_def_stmt))
{
/* For the INTEGER_CST case, we are just moving the
conversion from one place to another, which can often
hurt as the conversion moves further away from the
statement that computes the value. So, perform this
only if new_arg0 is an operand of COND_STMT, or
if arg0_def_stmt is the only non-debug stmt in
its basic block, because then it is possible this
could enable further optimizations (minmax replacement
etc.). See PR71016. */
if (new_arg0 != gimple_cond_lhs (cond_stmt)
&& new_arg0 != gimple_cond_rhs (cond_stmt)
&& gimple_bb (arg0_def_stmt) == e0->src)
{
gsi = gsi_for_stmt (arg0_def_stmt);
gsi_prev_nondebug (&gsi);
if (!gsi_end_p (gsi))
return NULL;
gsi = gsi_for_stmt (arg0_def_stmt);
gsi_next_nondebug (&gsi);
if (!gsi_end_p (gsi))
return NULL;
}
new_arg1 = fold_convert (TREE_TYPE (new_arg0), arg1);
}
else
return NULL;
}
else
return NULL;
}
/* If arg0/arg1 have > 1 use, then this transformation actually increases
the number of expressions evaluated at runtime. */
if (!has_single_use (arg0)
|| (arg1_def_stmt && !has_single_use (arg1)))
return NULL;
/* If types of new_arg0 and new_arg1 are different bailout. */
if (!types_compatible_p (TREE_TYPE (new_arg0), TREE_TYPE (new_arg1)))
return NULL;
/* Create a new PHI stmt. */
result = PHI_RESULT (phi);
temp = make_ssa_name (TREE_TYPE (new_arg0), NULL);
newphi = create_phi_node (temp, gimple_bb (phi));
if (dump_file && (dump_flags & TDF_DETAILS))
{
fprintf (dump_file, "PHI ");
print_generic_expr (dump_file, gimple_phi_result (phi));
fprintf (dump_file,
" changed to factor conversion out from COND_EXPR.\n");
fprintf (dump_file, "New stmt with CAST that defines ");
print_generic_expr (dump_file, result);
fprintf (dump_file, ".\n");
}
/* Remove the old cast(s) that has single use. */
gsi_for_def = gsi_for_stmt (arg0_def_stmt);
gsi_remove (&gsi_for_def, true);
release_defs (arg0_def_stmt);
if (arg1_def_stmt)
{
gsi_for_def = gsi_for_stmt (arg1_def_stmt);
gsi_remove (&gsi_for_def, true);
release_defs (arg1_def_stmt);
}
add_phi_arg (newphi, new_arg0, e0, locus);
add_phi_arg (newphi, new_arg1, e1, locus);
/* Create the conversion stmt and insert it. */
if (convert_code == VIEW_CONVERT_EXPR)
{
temp = fold_build1 (VIEW_CONVERT_EXPR, TREE_TYPE (result), temp);
new_stmt = gimple_build_assign (result, temp);
}
else
new_stmt = gimple_build_assign (result, convert_code, temp);
gsi = gsi_after_labels (gimple_bb (phi));
gsi_insert_before (&gsi, new_stmt, GSI_SAME_STMT);
/* Remove the original PHI stmt. */
gsi = gsi_for_stmt (phi);
gsi_remove (&gsi, true);
return newphi;
}
/* The function conditional_replacement does the main work of doing the
conditional replacement. Return true if the replacement is done.
Otherwise return false.
BB is the basic block where the replacement is going to be done on. ARG0
is argument 0 from PHI. Likewise for ARG1. */
static bool
conditional_replacement (basic_block cond_bb, basic_block middle_bb,
edge e0, edge e1, gphi *phi,
tree arg0, tree arg1)
{
tree result;
gimple *stmt;
gassign *new_stmt;
tree cond;
gimple_stmt_iterator gsi;
edge true_edge, false_edge;
tree new_var, new_var2;
bool neg;
/* FIXME: Gimplification of complex type is too hard for now. */
/* We aren't prepared to handle vectors either (and it is a question
if it would be worthwhile anyway). */
if (!(INTEGRAL_TYPE_P (TREE_TYPE (arg0))
|| POINTER_TYPE_P (TREE_TYPE (arg0)))
|| !(INTEGRAL_TYPE_P (TREE_TYPE (arg1))
|| POINTER_TYPE_P (TREE_TYPE (arg1))))
return false;
/* The PHI arguments have the constants 0 and 1, or 0 and -1, then
convert it to the conditional. */
if ((integer_zerop (arg0) && integer_onep (arg1))
|| (integer_zerop (arg1) && integer_onep (arg0)))
neg = false;
else if ((integer_zerop (arg0) && integer_all_onesp (arg1))
|| (integer_zerop (arg1) && integer_all_onesp (arg0)))
neg = true;
else
return false;
if (!empty_block_p (middle_bb))
return false;
/* At this point we know we have a GIMPLE_COND with two successors.
One successor is BB, the other successor is an empty block which
falls through into BB.
There is a single PHI node at the join point (BB) and its arguments
are constants (0, 1) or (0, -1).
So, given the condition COND, and the two PHI arguments, we can
rewrite this PHI into non-branching code:
dest = (COND) or dest = COND'
We use the condition as-is if the argument associated with the
true edge has the value one or the argument associated with the
false edge as the value zero. Note that those conditions are not
the same since only one of the outgoing edges from the GIMPLE_COND
will directly reach BB and thus be associated with an argument. */
stmt = last_stmt (cond_bb);
result = PHI_RESULT (phi);
/* To handle special cases like floating point comparison, it is easier and
less error-prone to build a tree and gimplify it on the fly though it is
less efficient. */
cond = fold_build2_loc (gimple_location (stmt),
gimple_cond_code (stmt), boolean_type_node,
gimple_cond_lhs (stmt), gimple_cond_rhs (stmt));
/* We need to know which is the true edge and which is the false
edge so that we know when to invert the condition below. */
extract_true_false_edges_from_block (cond_bb, &true_edge, &false_edge);
if ((e0 == true_edge && integer_zerop (arg0))
|| (e0 == false_edge && !integer_zerop (arg0))
|| (e1 == true_edge && integer_zerop (arg1))
|| (e1 == false_edge && !integer_zerop (arg1)))
cond = fold_build1_loc (gimple_location (stmt),
TRUTH_NOT_EXPR, TREE_TYPE (cond), cond);
if (neg)
{
cond = fold_convert_loc (gimple_location (stmt),
TREE_TYPE (result), cond);
cond = fold_build1_loc (gimple_location (stmt),
NEGATE_EXPR, TREE_TYPE (cond), cond);
}
/* Insert our new statements at the end of conditional block before the
COND_STMT. */
gsi = gsi_for_stmt (stmt);
new_var = force_gimple_operand_gsi (&gsi, cond, true, NULL, true,
GSI_SAME_STMT);
if (!useless_type_conversion_p (TREE_TYPE (result), TREE_TYPE (new_var)))
{
location_t locus_0, locus_1;
new_var2 = make_ssa_name (TREE_TYPE (result));
new_stmt = gimple_build_assign (new_var2, CONVERT_EXPR, new_var);
gsi_insert_before (&gsi, new_stmt, GSI_SAME_STMT);
new_var = new_var2;
/* Set the locus to the first argument, unless is doesn't have one. */
locus_0 = gimple_phi_arg_location (phi, 0);
locus_1 = gimple_phi_arg_location (phi, 1);
if (locus_0 == UNKNOWN_LOCATION)
locus_0 = locus_1;
gimple_set_location (new_stmt, locus_0);
}
replace_phi_edge_with_variable (cond_bb, e1, phi, new_var);
/* Note that we optimized this PHI. */
return true;
}
/* Update *ARG which is defined in STMT so that it contains the
computed value if that seems profitable. Return true if the
statement is made dead by that rewriting. */
static bool
jump_function_from_stmt (tree *arg, gimple *stmt)
{
enum tree_code code = gimple_assign_rhs_code (stmt);
if (code == ADDR_EXPR)
{
/* For arg = &p->i transform it to p, if possible. */
tree rhs1 = gimple_assign_rhs1 (stmt);
poly_int64 offset;
tree tem = get_addr_base_and_unit_offset (TREE_OPERAND (rhs1, 0),
&offset);
if (tem
&& TREE_CODE (tem) == MEM_REF
&& known_eq (mem_ref_offset (tem) + offset, 0))
{
*arg = TREE_OPERAND (tem, 0);
return true;
}
}
/* TODO: Much like IPA-CP jump-functions we want to handle constant
additions symbolically here, and we'd need to update the comparison
code that compares the arg + cst tuples in our caller. For now the
code above exactly handles the VEC_BASE pattern from vec.h. */
return false;
}
/* RHS is a source argument in a BIT_AND_EXPR which feeds a conditional
of the form SSA_NAME NE 0.
If RHS is fed by a simple EQ_EXPR comparison of two values, see if
the two input values of the EQ_EXPR match arg0 and arg1.
If so update *code and return TRUE. Otherwise return FALSE. */
static bool
rhs_is_fed_for_value_replacement (const_tree arg0, const_tree arg1,
enum tree_code *code, const_tree rhs)
{
/* Obviously if RHS is not an SSA_NAME, we can't look at the defining
statement. */
if (TREE_CODE (rhs) == SSA_NAME)
{
gimple *def1 = SSA_NAME_DEF_STMT (rhs);
/* Verify the defining statement has an EQ_EXPR on the RHS. */
if (is_gimple_assign (def1) && gimple_assign_rhs_code (def1) == EQ_EXPR)
{
/* Finally verify the source operands of the EQ_EXPR are equal
to arg0 and arg1. */
tree op0 = gimple_assign_rhs1 (def1);
tree op1 = gimple_assign_rhs2 (def1);
if ((operand_equal_for_phi_arg_p (arg0, op0)
&& operand_equal_for_phi_arg_p (arg1, op1))
|| (operand_equal_for_phi_arg_p (arg0, op1)
&& operand_equal_for_phi_arg_p (arg1, op0)))
{
/* We will perform the optimization. */
*code = gimple_assign_rhs_code (def1);
return true;
}
}
}
return false;
}
/* Return TRUE if arg0/arg1 are equal to the rhs/lhs or lhs/rhs of COND.
Also return TRUE if arg0/arg1 are equal to the source arguments of a
an EQ comparison feeding a BIT_AND_EXPR which feeds COND.
Return FALSE otherwise. */
static bool
operand_equal_for_value_replacement (const_tree arg0, const_tree arg1,
enum tree_code *code, gimple *cond)
{
gimple *def;
tree lhs = gimple_cond_lhs (cond);
tree rhs = gimple_cond_rhs (cond);
if ((operand_equal_for_phi_arg_p (arg0, lhs)
&& operand_equal_for_phi_arg_p (arg1, rhs))
|| (operand_equal_for_phi_arg_p (arg1, lhs)
&& operand_equal_for_phi_arg_p (arg0, rhs)))
return true;
/* Now handle more complex case where we have an EQ comparison
which feeds a BIT_AND_EXPR which feeds COND.
First verify that COND is of the form SSA_NAME NE 0. */
if (*code != NE_EXPR || !integer_zerop (rhs)
|| TREE_CODE (lhs) != SSA_NAME)
return false;
/* Now ensure that SSA_NAME is set by a BIT_AND_EXPR. */
def = SSA_NAME_DEF_STMT (lhs);
if (!is_gimple_assign (def) || gimple_assign_rhs_code (def) != BIT_AND_EXPR)
return false;
/* Now verify arg0/arg1 correspond to the source arguments of an
EQ comparison feeding the BIT_AND_EXPR. */
tree tmp = gimple_assign_rhs1 (def);
if (rhs_is_fed_for_value_replacement (arg0, arg1, code, tmp))
return true;
tmp = gimple_assign_rhs2 (def);
if (rhs_is_fed_for_value_replacement (arg0, arg1, code, tmp))
return true;
return false;
}
/* Returns true if ARG is a neutral element for operation CODE
on the RIGHT side. */
static bool
neutral_element_p (tree_code code, tree arg, bool right)
{
switch (code)
{
case PLUS_EXPR:
case BIT_IOR_EXPR:
case BIT_XOR_EXPR:
return integer_zerop (arg);
case LROTATE_EXPR:
case RROTATE_EXPR:
case LSHIFT_EXPR:
case RSHIFT_EXPR:
case MINUS_EXPR:
case POINTER_PLUS_EXPR:
return right && integer_zerop (arg);
case MULT_EXPR:
return integer_onep (arg);
case TRUNC_DIV_EXPR:
case CEIL_DIV_EXPR:
case FLOOR_DIV_EXPR:
case ROUND_DIV_EXPR:
case EXACT_DIV_EXPR:
return right && integer_onep (arg);
case BIT_AND_EXPR:
return integer_all_onesp (arg);
default:
return false;
}
}
/* Returns true if ARG is an absorbing element for operation CODE. */
static bool
absorbing_element_p (tree_code code, tree arg, bool right, tree rval)
{
switch (code)
{
case BIT_IOR_EXPR:
return integer_all_onesp (arg);
case MULT_EXPR:
case BIT_AND_EXPR:
return integer_zerop (arg);
case LSHIFT_EXPR:
case RSHIFT_EXPR:
case LROTATE_EXPR:
case RROTATE_EXPR:
return !right && integer_zerop (arg);
case TRUNC_DIV_EXPR:
case CEIL_DIV_EXPR:
case FLOOR_DIV_EXPR:
case ROUND_DIV_EXPR:
case EXACT_DIV_EXPR:
case TRUNC_MOD_EXPR:
case CEIL_MOD_EXPR:
case FLOOR_MOD_EXPR:
case ROUND_MOD_EXPR:
return (!right
&& integer_zerop (arg)
&& tree_single_nonzero_warnv_p (rval, NULL));
default:
return false;
}
}
/* The function value_replacement does the main work of doing the value
replacement. Return non-zero if the replacement is done. Otherwise return
0. If we remove the middle basic block, return 2.
BB is the basic block where the replacement is going to be done on. ARG0
is argument 0 from the PHI. Likewise for ARG1. */
static int
value_replacement (basic_block cond_bb, basic_block middle_bb,
edge e0, edge e1, gimple *phi,
tree arg0, tree arg1)
{
gimple_stmt_iterator gsi;
gimple *cond;
edge true_edge, false_edge;
enum tree_code code;
bool emtpy_or_with_defined_p = true;
/* If the type says honor signed zeros we cannot do this
optimization. */
if (HONOR_SIGNED_ZEROS (arg1))
return 0;
/* If there is a statement in MIDDLE_BB that defines one of the PHI
arguments, then adjust arg0 or arg1. */
gsi = gsi_start_nondebug_after_labels_bb (middle_bb);
while (!gsi_end_p (gsi))
{
gimple *stmt = gsi_stmt (gsi);
tree lhs;
gsi_next_nondebug (&gsi);
if (!is_gimple_assign (stmt))
{
if (gimple_code (stmt) != GIMPLE_PREDICT
&& gimple_code (stmt) != GIMPLE_NOP)
emtpy_or_with_defined_p = false;
continue;
}
/* Now try to adjust arg0 or arg1 according to the computation
in the statement. */
lhs = gimple_assign_lhs (stmt);
if (!(lhs == arg0
&& jump_function_from_stmt (&arg0, stmt))
|| (lhs == arg1
&& jump_function_from_stmt (&arg1, stmt)))
emtpy_or_with_defined_p = false;
}
cond = last_stmt (cond_bb);
code = gimple_cond_code (cond);
/* This transformation is only valid for equality comparisons. */
if (code != NE_EXPR && code != EQ_EXPR)
return 0;
/* We need to know which is the true edge and which is the false
edge so that we know if have abs or negative abs. */
extract_true_false_edges_from_block (cond_bb, &true_edge, &false_edge);
/* At this point we know we have a COND_EXPR with two successors.
One successor is BB, the other successor is an empty block which
falls through into BB.
The condition for the COND_EXPR is known to be NE_EXPR or EQ_EXPR.
There is a single PHI node at the join point (BB) with two arguments.
We now need to verify that the two arguments in the PHI node match
the two arguments to the equality comparison. */
if (operand_equal_for_value_replacement (arg0, arg1, &code, cond))
{
edge e;
tree arg;
/* For NE_EXPR, we want to build an assignment result = arg where
arg is the PHI argument associated with the true edge. For
EQ_EXPR we want the PHI argument associated with the false edge. */
e = (code == NE_EXPR ? true_edge : false_edge);
/* Unfortunately, E may not reach BB (it may instead have gone to
OTHER_BLOCK). If that is the case, then we want the single outgoing
edge from OTHER_BLOCK which reaches BB and represents the desired
path from COND_BLOCK. */
if (e->dest == middle_bb)
e = single_succ_edge (e->dest);
/* Now we know the incoming edge to BB that has the argument for the
RHS of our new assignment statement. */
if (e0 == e)
arg = arg0;
else
arg = arg1;
/* If the middle basic block was empty or is defining the
PHI arguments and this is a single phi where the args are different
for the edges e0 and e1 then we can remove the middle basic block. */
if (emtpy_or_with_defined_p
&& single_non_singleton_phi_for_edges (phi_nodes (gimple_bb (phi)),
e0, e1) == phi)
{
replace_phi_edge_with_variable (cond_bb, e1, phi, arg);
/* Note that we optimized this PHI. */
return 2;
}
else
{
/* Replace the PHI arguments with arg. */
SET_PHI_ARG_DEF (phi, e0->dest_idx, arg);
SET_PHI_ARG_DEF (phi, e1->dest_idx, arg);
if (dump_file && (dump_flags & TDF_DETAILS))
{
fprintf (dump_file, "PHI ");
print_generic_expr (dump_file, gimple_phi_result (phi));
fprintf (dump_file, " reduced for COND_EXPR in block %d to ",
cond_bb->index);
print_generic_expr (dump_file, arg);
fprintf (dump_file, ".\n");
}
return 1;
}
}
/* Now optimize (x != 0) ? x + y : y to just x + y. */
gsi = gsi_last_nondebug_bb (middle_bb);
if (gsi_end_p (gsi))
return 0;
gimple *assign = gsi_stmt (gsi);
if (!is_gimple_assign (assign)
|| gimple_assign_rhs_class (assign) != GIMPLE_BINARY_RHS
|| (!INTEGRAL_TYPE_P (TREE_TYPE (arg0))
&& !POINTER_TYPE_P (TREE_TYPE (arg0))))
return 0;
/* Punt if there are (degenerate) PHIs in middle_bb, there should not be. */
if (!gimple_seq_empty_p (phi_nodes (middle_bb)))
return 0;
/* Allow up to 2 cheap preparation statements that prepare argument
for assign, e.g.:
if (y_4 != 0)
goto <bb 3>;
else
goto <bb 4>;
<bb 3>:
_1 = (int) y_4;
iftmp.0_6 = x_5(D) r<< _1;
<bb 4>:
# iftmp.0_2 = PHI <iftmp.0_6(3), x_5(D)(2)>
or:
if (y_3(D) == 0)
goto <bb 4>;
else
goto <bb 3>;
<bb 3>:
y_4 = y_3(D) & 31;
_1 = (int) y_4;
_6 = x_5(D) r<< _1;
<bb 4>:
# _2 = PHI <x_5(D)(2), _6(3)> */
gimple *prep_stmt[2] = { NULL, NULL };
int prep_cnt;
for (prep_cnt = 0; ; prep_cnt++)
{
gsi_prev_nondebug (&gsi);
if (gsi_end_p (gsi))
break;
gimple *g = gsi_stmt (gsi);
if (gimple_code (g) == GIMPLE_LABEL)
break;
if (prep_cnt == 2 || !is_gimple_assign (g))
return 0;
tree lhs = gimple_assign_lhs (g);
tree rhs1 = gimple_assign_rhs1 (g);
use_operand_p use_p;
gimple *use_stmt;
if (TREE_CODE (lhs) != SSA_NAME
|| TREE_CODE (rhs1) != SSA_NAME
|| !INTEGRAL_TYPE_P (TREE_TYPE (lhs))
|| !INTEGRAL_TYPE_P (TREE_TYPE (rhs1))
|| !single_imm_use (lhs, &use_p, &use_stmt)
|| use_stmt != (prep_cnt ? prep_stmt[prep_cnt - 1] : assign))
return 0;
switch (gimple_assign_rhs_code (g))
{
CASE_CONVERT:
break;
case PLUS_EXPR:
case BIT_AND_EXPR:
case BIT_IOR_EXPR:
case BIT_XOR_EXPR:
if (TREE_CODE (gimple_assign_rhs2 (g)) != INTEGER_CST)
return 0;
break;
default:
return 0;
}
prep_stmt[prep_cnt] = g;
}
/* Only transform if it removes the condition. */
if (!single_non_singleton_phi_for_edges (phi_nodes (gimple_bb (phi)), e0, e1))
return 0;
/* Size-wise, this is always profitable. */
if (optimize_bb_for_speed_p (cond_bb)
/* The special case is useless if it has a low probability. */
&& profile_status_for_fn (cfun) != PROFILE_ABSENT
&& EDGE_PRED (middle_bb, 0)->probability < profile_probability::even ()
/* If assign is cheap, there is no point avoiding it. */
&& estimate_num_insns (bb_seq (middle_bb), &eni_time_weights)
>= 3 * estimate_num_insns (cond, &eni_time_weights))
return 0;
tree lhs = gimple_assign_lhs (assign);
tree rhs1 = gimple_assign_rhs1 (assign);
tree rhs2 = gimple_assign_rhs2 (assign);
enum tree_code code_def = gimple_assign_rhs_code (assign);
tree cond_lhs = gimple_cond_lhs (cond);
tree cond_rhs = gimple_cond_rhs (cond);
/* Propagate the cond_rhs constant through preparation stmts,
make sure UB isn't invoked while doing that. */
for (int i = prep_cnt - 1; i >= 0; --i)
{
gimple *g = prep_stmt[i];
tree grhs1 = gimple_assign_rhs1 (g);
if (!operand_equal_for_phi_arg_p (cond_lhs, grhs1))
return 0;
cond_lhs = gimple_assign_lhs (g);
cond_rhs = fold_convert (TREE_TYPE (grhs1), cond_rhs);
if (TREE_CODE (cond_rhs) != INTEGER_CST
|| TREE_OVERFLOW (cond_rhs))
return 0;
if (gimple_assign_rhs_class (g) == GIMPLE_BINARY_RHS)
{
cond_rhs = int_const_binop (gimple_assign_rhs_code (g), cond_rhs,
gimple_assign_rhs2 (g));
if (TREE_OVERFLOW (cond_rhs))
return 0;
}
cond_rhs = fold_convert (TREE_TYPE (cond_lhs), cond_rhs);
if (TREE_CODE (cond_rhs) != INTEGER_CST
|| TREE_OVERFLOW (cond_rhs))
return 0;
}
if (((code == NE_EXPR && e1 == false_edge)
|| (code == EQ_EXPR && e1 == true_edge))
&& arg0 == lhs
&& ((arg1 == rhs1
&& operand_equal_for_phi_arg_p (rhs2, cond_lhs)
&& neutral_element_p (code_def, cond_rhs, true))
|| (arg1 == rhs2
&& operand_equal_for_phi_arg_p (rhs1, cond_lhs)
&& neutral_element_p (code_def, cond_rhs, false))
|| (operand_equal_for_phi_arg_p (arg1, cond_rhs)
&& ((operand_equal_for_phi_arg_p (rhs2, cond_lhs)
&& absorbing_element_p (code_def, cond_rhs, true, rhs2))
|| (operand_equal_for_phi_arg_p (rhs1, cond_lhs)
&& absorbing_element_p (code_def,
cond_rhs, false, rhs2))))))
{
gsi = gsi_for_stmt (cond);
/* Moving ASSIGN might change VR of lhs, e.g. when moving u_6
def-stmt in:
if (n_5 != 0)
goto <bb 3>;
else
goto <bb 4>;
<bb 3>:
# RANGE [0, 4294967294]
u_6 = n_5 + 4294967295;
<bb 4>:
# u_3 = PHI <u_6(3), 4294967295(2)> */
reset_flow_sensitive_info (lhs);
if (INTEGRAL_TYPE_P (TREE_TYPE (lhs)))
{
/* If available, we can use VR of phi result at least. */
tree phires = gimple_phi_result (phi);
struct range_info_def *phires_range_info
= SSA_NAME_RANGE_INFO (phires);
if (phires_range_info)
duplicate_ssa_name_range_info (lhs, SSA_NAME_RANGE_TYPE (phires),
phires_range_info);
}
gimple_stmt_iterator gsi_from;
for (int i = prep_cnt - 1; i >= 0; --i)
{
tree plhs = gimple_assign_lhs (prep_stmt[i]);
reset_flow_sensitive_info (plhs);
gsi_from = gsi_for_stmt (prep_stmt[i]);
gsi_move_before (&gsi_from, &gsi);
}
gsi_from = gsi_for_stmt (assign);
gsi_move_before (&gsi_from, &gsi);
replace_phi_edge_with_variable (cond_bb, e1, phi, lhs);
return 2;
}
return 0;
}
/* The function minmax_replacement does the main work of doing the minmax
replacement. Return true if the replacement is done. Otherwise return
false.
BB is the basic block where the replacement is going to be done on. ARG0
is argument 0 from the PHI. Likewise for ARG1. */
static bool
minmax_replacement (basic_block cond_bb, basic_block middle_bb,
edge e0, edge e1, gimple *phi,
tree arg0, tree arg1)
{
tree result, type, rhs;
gcond *cond;
gassign *new_stmt;
edge true_edge, false_edge;
enum tree_code cmp, minmax, ass_code;
tree smaller, alt_smaller, larger, alt_larger, arg_true, arg_false;
gimple_stmt_iterator gsi, gsi_from;
type = TREE_TYPE (PHI_RESULT (phi));
/* The optimization may be unsafe due to NaNs. */
if (HONOR_NANS (type) || HONOR_SIGNED_ZEROS (type))
return false;
cond = as_a <gcond *> (last_stmt (cond_bb));
cmp = gimple_cond_code (cond);
rhs = gimple_cond_rhs (cond);
/* Turn EQ/NE of extreme values to order comparisons. */
if ((cmp == NE_EXPR || cmp == EQ_EXPR)
&& TREE_CODE (rhs) == INTEGER_CST)
{
if (wi::eq_p (wi::to_wide (rhs), wi::min_value (TREE_TYPE (rhs))))
{
cmp = (cmp == EQ_EXPR) ? LT_EXPR : GE_EXPR;
rhs = wide_int_to_tree (TREE_TYPE (rhs),
wi::min_value (TREE_TYPE (rhs)) + 1);
}
else if (wi::eq_p (wi::to_wide (rhs), wi::max_value (TREE_TYPE (rhs))))
{
cmp = (cmp == EQ_EXPR) ? GT_EXPR : LE_EXPR;
rhs = wide_int_to_tree (TREE_TYPE (rhs),
wi::max_value (TREE_TYPE (rhs)) - 1);
}
}
/* This transformation is only valid for order comparisons. Record which
operand is smaller/larger if the result of the comparison is true. */
alt_smaller = NULL_TREE;
alt_larger = NULL_TREE;
if (cmp == LT_EXPR || cmp == LE_EXPR)
{
smaller = gimple_cond_lhs (cond);
larger = rhs;
/* If we have smaller < CST it is equivalent to smaller <= CST-1.
Likewise smaller <= CST is equivalent to smaller < CST+1. */
if (TREE_CODE (larger) == INTEGER_CST)
{
if (cmp == LT_EXPR)
{
wi::overflow_type overflow;
wide_int alt = wi::sub (wi::to_wide (larger), 1,
TYPE_SIGN (TREE_TYPE (larger)),
&overflow);
if (! overflow)
alt_larger = wide_int_to_tree (TREE_TYPE (larger), alt);
}
else
{
wi::overflow_type overflow;
wide_int alt = wi::add (wi::to_wide (larger), 1,
TYPE_SIGN (TREE_TYPE (larger)),
&overflow);
if (! overflow)
alt_larger = wide_int_to_tree (TREE_TYPE (larger), alt);
}
}
}
else if (cmp == GT_EXPR || cmp == GE_EXPR)
{
smaller = rhs;
larger = gimple_cond_lhs (cond);
/* If we have larger > CST it is equivalent to larger >= CST+1.
Likewise larger >= CST is equivalent to larger > CST-1. */
if (TREE_CODE (smaller) == INTEGER_CST)
{
wi::overflow_type overflow;
if (cmp == GT_EXPR)
{
wide_int alt = wi::add (wi::to_wide (smaller), 1,
TYPE_SIGN (TREE_TYPE (smaller)),
&overflow);
if (! overflow)
alt_smaller = wide_int_to_tree (TREE_TYPE (smaller), alt);
}
else
{
wide_int alt = wi::sub (wi::to_wide (smaller), 1,
TYPE_SIGN (TREE_TYPE (smaller)),
&overflow);
if (! overflow)
alt_smaller = wide_int_to_tree (TREE_TYPE (smaller), alt);
}
}
}
else
return false;
/* We need to know which is the true edge and which is the false
edge so that we know if have abs or negative abs. */
extract_true_false_edges_from_block (cond_bb, &true_edge, &false_edge);
/* Forward the edges over the middle basic block. */
if (true_edge->dest == middle_bb)
true_edge = EDGE_SUCC (true_edge->dest, 0);
if (false_edge->dest == middle_bb)
false_edge = EDGE_SUCC (false_edge->dest, 0);
if (true_edge == e0)
{
gcc_assert (false_edge == e1);
arg_true = arg0;
arg_false = arg1;
}
else
{
gcc_assert (false_edge == e0);
gcc_assert (true_edge == e1);
arg_true = arg1;
arg_false = arg0;
}
if (empty_block_p (middle_bb))
{
if ((operand_equal_for_phi_arg_p (arg_true, smaller)
|| (alt_smaller
&& operand_equal_for_phi_arg_p (arg_true, alt_smaller)))
&& (operand_equal_for_phi_arg_p (arg_false, larger)
|| (alt_larger
&& operand_equal_for_phi_arg_p (arg_true, alt_larger))))
{
/* Case
if (smaller < larger)
rslt = smaller;
else
rslt = larger; */
minmax = MIN_EXPR;
}
else if ((operand_equal_for_phi_arg_p (arg_false, smaller)
|| (alt_smaller
&& operand_equal_for_phi_arg_p (arg_false, alt_smaller)))
&& (operand_equal_for_phi_arg_p (arg_true, larger)
|| (alt_larger
&& operand_equal_for_phi_arg_p (arg_true, alt_larger))))
minmax = MAX_EXPR;
else
return false;
}
else
{
/* Recognize the following case, assuming d <= u:
if (a <= u)
b = MAX (a, d);
x = PHI <b, u>
This is equivalent to
b = MAX (a, d);
x = MIN (b, u); */
gimple *assign = last_and_only_stmt (middle_bb);
tree lhs, op0, op1, bound;
if (!assign
|| gimple_code (assign) != GIMPLE_ASSIGN)
return false;
lhs = gimple_assign_lhs (assign);
ass_code = gimple_assign_rhs_code (assign);
if (ass_code != MAX_EXPR && ass_code != MIN_EXPR)
return false;
op0 = gimple_assign_rhs1 (assign);
op1 = gimple_assign_rhs2 (assign);
if (true_edge->src == middle_bb)
{
/* We got here if the condition is true, i.e., SMALLER < LARGER. */
if (!operand_equal_for_phi_arg_p (lhs, arg_true))
return false;
if (operand_equal_for_phi_arg_p (arg_false, larger)
|| (alt_larger
&& operand_equal_for_phi_arg_p (arg_false, alt_larger)))
{
/* Case
if (smaller < larger)
{
r' = MAX_EXPR (smaller, bound)
}
r = PHI <r', larger> --> to be turned to MIN_EXPR. */
if (ass_code != MAX_EXPR)
return false;
minmax = MIN_EXPR;
if (operand_equal_for_phi_arg_p (op0, smaller)
|| (alt_smaller
&& operand_equal_for_phi_arg_p (op0, alt_smaller)))
bound = op1;
else if (operand_equal_for_phi_arg_p (op1, smaller)
|| (alt_smaller
&& operand_equal_for_phi_arg_p (op1, alt_smaller)))
bound = op0;
else
return false;
/* We need BOUND <= LARGER. */
if (!integer_nonzerop (fold_build2 (LE_EXPR, boolean_type_node,
bound, larger)))
return false;
}
else if (operand_equal_for_phi_arg_p (arg_false, smaller)
|| (alt_smaller
&& operand_equal_for_phi_arg_p (arg_false, alt_smaller)))
{
/* Case
if (smaller < larger)
{
r' = MIN_EXPR (larger, bound)
}
r = PHI <r', smaller> --> to be turned to MAX_EXPR. */
if (ass_code != MIN_EXPR)
return false;
minmax = MAX_EXPR;
if (operand_equal_for_phi_arg_p (op0, larger)
|| (alt_larger
&& operand_equal_for_phi_arg_p (op0, alt_larger)))
bound = op1;
else if (operand_equal_for_phi_arg_p (op1, larger)
|| (alt_larger
&& operand_equal_for_phi_arg_p (op1, alt_larger)))
bound = op0;
else
return false;
/* We need BOUND >= SMALLER. */
if (!integer_nonzerop (fold_build2 (GE_EXPR, boolean_type_node,
bound, smaller)))
return false;
}
else
return false;
}
else
{
/* We got here if the condition is false, i.e., SMALLER > LARGER. */
if (!operand_equal_for_phi_arg_p (lhs, arg_false))
return false;
if (operand_equal_for_phi_arg_p (arg_true, larger)
|| (alt_larger
&& operand_equal_for_phi_arg_p (arg_true, alt_larger)))
{
/* Case
if (smaller > larger)
{
r' = MIN_EXPR (smaller, bound)
}
r = PHI <r', larger> --> to be turned to MAX_EXPR. */
if (ass_code != MIN_EXPR)
return false;
minmax = MAX_EXPR;
if (operand_equal_for_phi_arg_p (op0, smaller)
|| (alt_smaller
&& operand_equal_for_phi_arg_p (op0, alt_smaller)))
bound = op1;
else if (operand_equal_for_phi_arg_p (op1, smaller)
|| (alt_smaller
&& operand_equal_for_phi_arg_p (op1, alt_smaller)))
bound = op0;
else
return false;
/* We need BOUND >= LARGER. */
if (!integer_nonzerop (fold_build2 (GE_EXPR, boolean_type_node,
bound, larger)))
return false;
}
else if (operand_equal_for_phi_arg_p (arg_true, smaller)
|| (alt_smaller
&& operand_equal_for_phi_arg_p (arg_true, alt_smaller)))
{
/* Case
if (smaller > larger)
{
r' = MAX_EXPR (larger, bound)
}
r = PHI <r', smaller> --> to be turned to MIN_EXPR. */
if (ass_code != MAX_EXPR)
return false;
minmax = MIN_EXPR;
if (operand_equal_for_phi_arg_p (op0, larger))
bound = op1;
else if (operand_equal_for_phi_arg_p (op1, larger))
bound = op0;
else
return false;
/* We need BOUND <= SMALLER. */
if (!integer_nonzerop (fold_build2 (LE_EXPR, boolean_type_node,
bound, smaller)))
return false;
}
else
return false;
}
/* Move the statement from the middle block. */
gsi = gsi_last_bb (cond_bb);
gsi_from = gsi_last_nondebug_bb (middle_bb);
reset_flow_sensitive_info (SINGLE_SSA_TREE_OPERAND (gsi_stmt (gsi_from),
SSA_OP_DEF));
gsi_move_before (&gsi_from, &gsi);
}
/* Create an SSA var to hold the min/max result. If we're the only
things setting the target PHI, then we can clone the PHI
variable. Otherwise we must create a new one. */
result = PHI_RESULT (phi);
if (EDGE_COUNT (gimple_bb (phi)->preds) == 2)
result = duplicate_ssa_name (result, NULL);
else
result = make_ssa_name (TREE_TYPE (result));
/* Emit the statement to compute min/max. */
new_stmt = gimple_build_assign (result, minmax, arg0, arg1);
gsi = gsi_last_bb (cond_bb);
gsi_insert_before (&gsi, new_stmt, GSI_NEW_STMT);
replace_phi_edge_with_variable (cond_bb, e1, phi, result);
return true;
}
/* Convert
<bb 2>
if (b_4(D) != 0)
goto <bb 3>
else
goto <bb 4>
<bb 3>
_2 = (unsigned long) b_4(D);
_9 = __builtin_popcountl (_2);
OR
_9 = __builtin_popcountl (b_4(D));
<bb 4>
c_12 = PHI <0(2), _9(3)>
Into
<bb 2>
_2 = (unsigned long) b_4(D);
_9 = __builtin_popcountl (_2);
OR
_9 = __builtin_popcountl (b_4(D));
<bb 4>
c_12 = PHI <_9(2)>
*/
static bool
cond_removal_in_popcount_pattern (basic_block cond_bb, basic_block middle_bb,
edge e1, edge e2,
gimple *phi, tree arg0, tree arg1)
{
gimple *cond;
gimple_stmt_iterator gsi, gsi_from;
gimple *popcount;
gimple *cast = NULL;
tree lhs, arg;
/* Check that
_2 = (unsigned long) b_4(D);
_9 = __builtin_popcountl (_2);
OR
_9 = __builtin_popcountl (b_4(D));
are the only stmts in the middle_bb. */
gsi = gsi_start_nondebug_after_labels_bb (middle_bb);
if (gsi_end_p (gsi))
return false;
cast = gsi_stmt (gsi);
gsi_next_nondebug (&gsi);
if (!gsi_end_p (gsi))
{
popcount = gsi_stmt (gsi);
gsi_next_nondebug (&gsi);
if (!gsi_end_p (gsi))
return false;
}
else
{
popcount = cast;
cast = NULL;
}
/* Check that we have a popcount builtin. */
if (!is_gimple_call (popcount))
return false;
combined_fn cfn = gimple_call_combined_fn (popcount);
switch (cfn)
{
CASE_CFN_POPCOUNT:
break;
default:
return false;
}
arg = gimple_call_arg (popcount, 0);
lhs = gimple_get_lhs (popcount);
if (cast)
{
/* We have a cast stmt feeding popcount builtin. */
/* Check that we have a cast prior to that. */
if (gimple_code (cast) != GIMPLE_ASSIGN
|| !CONVERT_EXPR_CODE_P (gimple_assign_rhs_code (cast)))
return false;
/* Result of the cast stmt is the argument to the builtin. */
if (arg != gimple_assign_lhs (cast))
return false;
arg = gimple_assign_rhs1 (cast);
}
cond = last_stmt (cond_bb);
/* Cond_bb has a check for b_4 [!=|==] 0 before calling the popcount
builtin. */
if (gimple_code (cond) != GIMPLE_COND
|| (gimple_cond_code (cond) != NE_EXPR
&& gimple_cond_code (cond) != EQ_EXPR)
|| !integer_zerop (gimple_cond_rhs (cond))
|| arg != gimple_cond_lhs (cond))
return false;
/* Canonicalize. */
if ((e2->flags & EDGE_TRUE_VALUE
&& gimple_cond_code (cond) == NE_EXPR)
|| (e1->flags & EDGE_TRUE_VALUE
&& gimple_cond_code (cond) == EQ_EXPR))
{
std::swap (arg0, arg1);
std::swap (e1, e2);
}
/* Check PHI arguments. */
if (lhs != arg0 || !integer_zerop (arg1))
return false;
/* And insert the popcount builtin and cast stmt before the cond_bb. */
gsi = gsi_last_bb (cond_bb);
if (cast)
{
gsi_from = gsi_for_stmt (cast);
gsi_move_before (&gsi_from, &gsi);
reset_flow_sensitive_info (gimple_get_lhs (cast));
}
gsi_from = gsi_for_stmt (popcount);
gsi_move_before (&gsi_from, &gsi);
reset_flow_sensitive_info (gimple_get_lhs (popcount));
/* Now update the PHI and remove unneeded bbs. */
replace_phi_edge_with_variable (cond_bb, e2, phi, lhs);
return true;
}
/* The function absolute_replacement does the main work of doing the absolute
replacement. Return true if the replacement is done. Otherwise return
false.
bb is the basic block where the replacement is going to be done on. arg0
is argument 0 from the phi. Likewise for arg1. */
static bool
abs_replacement (basic_block cond_bb, basic_block middle_bb,
edge e0 ATTRIBUTE_UNUSED, edge e1,
gimple *phi, tree arg0, tree arg1)
{
tree result;
gassign *new_stmt;
gimple *cond;
gimple_stmt_iterator gsi;
edge true_edge, false_edge;
gimple *assign;
edge e;
tree rhs, lhs;
bool negate;
enum tree_code cond_code;
/* If the type says honor signed zeros we cannot do this
optimization. */
if (HONOR_SIGNED_ZEROS (arg1))
return false;
/* OTHER_BLOCK must have only one executable statement which must have the
form arg0 = -arg1 or arg1 = -arg0. */
assign = last_and_only_stmt (middle_bb);
/* If we did not find the proper negation assignment, then we can not
optimize. */
if (assign == NULL)
return false;
/* If we got here, then we have found the only executable statement
in OTHER_BLOCK. If it is anything other than arg = -arg1 or
arg1 = -arg0, then we can not optimize. */
if (gimple_code (assign) != GIMPLE_ASSIGN)
return false;
lhs = gimple_assign_lhs (assign);
if (gimple_assign_rhs_code (assign) != NEGATE_EXPR)
return false;
rhs = gimple_assign_rhs1 (assign);
/* The assignment has to be arg0 = -arg1 or arg1 = -arg0. */
if (!(lhs == arg0 && rhs == arg1)
&& !(lhs == arg1 && rhs == arg0))
return false;
cond = last_stmt (cond_bb);
result = PHI_RESULT (phi);
/* Only relationals comparing arg[01] against zero are interesting. */
cond_code = gimple_cond_code (cond);
if (cond_code != GT_EXPR && cond_code != GE_EXPR
&& cond_code != LT_EXPR && cond_code != LE_EXPR)
return false;
/* Make sure the conditional is arg[01] OP y. */
if (gimple_cond_lhs (cond) != rhs)
return false;
if (FLOAT_TYPE_P (TREE_TYPE (gimple_cond_rhs (cond)))
? real_zerop (gimple_cond_rhs (cond))
: integer_zerop (gimple_cond_rhs (cond)))
;
else
return false;
/* We need to know which is the true edge and which is the false
edge so that we know if have abs or negative abs. */
extract_true_false_edges_from_block (cond_bb, &true_edge, &false_edge);
/* For GT_EXPR/GE_EXPR, if the true edge goes to OTHER_BLOCK, then we
will need to negate the result. Similarly for LT_EXPR/LE_EXPR if
the false edge goes to OTHER_BLOCK. */
if (cond_code == GT_EXPR || cond_code == GE_EXPR)
e = true_edge;
else
e = false_edge;
if (e->dest == middle_bb)
negate = true;
else
negate = false;
/* If the code negates only iff positive then make sure to not
introduce undefined behavior when negating or computing the absolute.
??? We could use range info if present to check for arg1 == INT_MIN. */
if (negate
&& (ANY_INTEGRAL_TYPE_P (TREE_TYPE (arg1))
&& ! TYPE_OVERFLOW_WRAPS (TREE_TYPE (arg1))))
return false;
result = duplicate_ssa_name (result, NULL);
if (negate)
lhs = make_ssa_name (TREE_TYPE (result));
else
lhs = result;
/* Build the modify expression with abs expression. */
new_stmt = gimple_build_assign (lhs, ABS_EXPR, rhs);
gsi = gsi_last_bb (cond_bb);
gsi_insert_before (&gsi, new_stmt, GSI_NEW_STMT);
if (negate)
{
/* Get the right GSI. We want to insert after the recently
added ABS_EXPR statement (which we know is the first statement
in the block. */
new_stmt = gimple_build_assign (result, NEGATE_EXPR, lhs);
gsi_insert_after (&gsi, new_stmt, GSI_NEW_STMT);
}
replace_phi_edge_with_variable (cond_bb, e1, phi, result);
/* Note that we optimized this PHI. */
return true;
}
/* Auxiliary functions to determine the set of memory accesses which
can't trap because they are preceded by accesses to the same memory
portion. We do that for MEM_REFs, so we only need to track
the SSA_NAME of the pointer indirectly referenced. The algorithm
simply is a walk over all instructions in dominator order. When
we see an MEM_REF we determine if we've already seen a same
ref anywhere up to the root of the dominator tree. If we do the
current access can't trap. If we don't see any dominating access
the current access might trap, but might also make later accesses
non-trapping, so we remember it. We need to be careful with loads
or stores, for instance a load might not trap, while a store would,
so if we see a dominating read access this doesn't mean that a later
write access would not trap. Hence we also need to differentiate the
type of access(es) seen.
??? We currently are very conservative and assume that a load might
trap even if a store doesn't (write-only memory). This probably is
overly conservative. */
/* A hash-table of SSA_NAMEs, and in which basic block an MEM_REF
through it was seen, which would constitute a no-trap region for
same accesses. */
struct name_to_bb
{
unsigned int ssa_name_ver;
unsigned int phase;
bool store;
HOST_WIDE_INT offset, size;
basic_block bb;
};
/* Hashtable helpers. */
struct ssa_names_hasher : free_ptr_hash <name_to_bb>
{
static inline hashval_t hash (const name_to_bb *);
static inline bool equal (const name_to_bb *, const name_to_bb *);
};
/* Used for quick clearing of the hash-table when we see calls.
Hash entries with phase < nt_call_phase are invalid. */
static unsigned int nt_call_phase;
/* The hash function. */
inline hashval_t
ssa_names_hasher::hash (const name_to_bb *n)
{
return n->ssa_name_ver ^ (((hashval_t) n->store) << 31)
^ (n->offset << 6) ^ (n->size << 3);
}
/* The equality function of *P1 and *P2. */
inline bool
ssa_names_hasher::equal (const name_to_bb *n1, const name_to_bb *n2)
{
return n1->ssa_name_ver == n2->ssa_name_ver
&& n1->store == n2->store
&& n1->offset == n2->offset
&& n1->size == n2->size;
}
class nontrapping_dom_walker : public dom_walker
{
public:
nontrapping_dom_walker (cdi_direction direction, hash_set<tree> *ps)
: dom_walker (direction), m_nontrapping (ps), m_seen_ssa_names (128) {}
virtual edge before_dom_children (basic_block);
virtual void after_dom_children (basic_block);
private:
/* We see the expression EXP in basic block BB. If it's an interesting
expression (an MEM_REF through an SSA_NAME) possibly insert the
expression into the set NONTRAP or the hash table of seen expressions.
STORE is true if this expression is on the LHS, otherwise it's on
the RHS. */
void add_or_mark_expr (basic_block, tree, bool);
hash_set<tree> *m_nontrapping;
/* The hash table for remembering what we've seen. */
hash_table<ssa_names_hasher> m_seen_ssa_names;
};
/* Called by walk_dominator_tree, when entering the block BB. */
edge
nontrapping_dom_walker::before_dom_children (basic_block bb)
{
edge e;
edge_iterator ei;
gimple_stmt_iterator gsi;
/* If we haven't seen all our predecessors, clear the hash-table. */
FOR_EACH_EDGE (e, ei, bb->preds)
if ((((size_t)e->src->aux) & 2) == 0)
{
nt_call_phase++;
break;
}
/* Mark this BB as being on the path to dominator root and as visited. */
bb->aux = (void*)(1 | 2);
/* And walk the statements in order. */
for (gsi = gsi_start_bb (bb); !gsi_end_p (gsi); gsi_next (&gsi))
{
gimple *stmt = gsi_stmt (gsi);
if ((gimple_code (stmt) == GIMPLE_ASM && gimple_vdef (stmt))
|| (is_gimple_call (stmt)
&& (!nonfreeing_call_p (stmt) || !nonbarrier_call_p (stmt))))
nt_call_phase++;
else if (gimple_assign_single_p (stmt) && !gimple_has_volatile_ops (stmt))
{
add_or_mark_expr (bb, gimple_assign_lhs (stmt), true);
add_or_mark_expr (bb, gimple_assign_rhs1 (stmt), false);
}
}
return NULL;
}
/* Called by walk_dominator_tree, when basic block BB is exited. */
void
nontrapping_dom_walker::after_dom_children (basic_block bb)
{
/* This BB isn't on the path to dominator root anymore. */
bb->aux = (void*)2;
}
/* We see the expression EXP in basic block BB. If it's an interesting
expression (an MEM_REF through an SSA_NAME) possibly insert the
expression into the set NONTRAP or the hash table of seen expressions.
STORE is true if this expression is on the LHS, otherwise it's on
the RHS. */
void
nontrapping_dom_walker::add_or_mark_expr (basic_block bb, tree exp, bool store)
{
HOST_WIDE_INT size;
if (TREE_CODE (exp) == MEM_REF
&& TREE_CODE (TREE_OPERAND (exp, 0)) == SSA_NAME
&& tree_fits_shwi_p (TREE_OPERAND (exp, 1))
&& (size = int_size_in_bytes (TREE_TYPE (exp))) > 0)
{
tree name = TREE_OPERAND (exp, 0);
struct name_to_bb map;
name_to_bb **slot;
struct name_to_bb *n2bb;
basic_block found_bb = 0;
/* Try to find the last seen MEM_REF through the same
SSA_NAME, which can trap. */
map.ssa_name_ver = SSA_NAME_VERSION (name);
map.phase = 0;
map.bb = 0;
map.store = store;
map.offset = tree_to_shwi (TREE_OPERAND (exp, 1));
map.size = size;
slot = m_seen_ssa_names.find_slot (&map, INSERT);
n2bb = *slot;
if (n2bb && n2bb->phase >= nt_call_phase)
found_bb = n2bb->bb;
/* If we've found a trapping MEM_REF, _and_ it dominates EXP
(it's in a basic block on the path from us to the dominator root)
then we can't trap. */
if (found_bb && (((size_t)found_bb->aux) & 1) == 1)
{
m_nontrapping->add (exp);
}
else
{
/* EXP might trap, so insert it into the hash table. */
if (n2bb)
{
n2bb->phase = nt_call_phase;
n2bb->bb = bb;
}
else
{
n2bb = XNEW (struct name_to_bb);
n2bb->ssa_name_ver = SSA_NAME_VERSION (name);
n2bb->phase = nt_call_phase;
n2bb->bb = bb;
n2bb->store = store;
n2bb->offset = map.offset;
n2bb->size = size;
*slot = n2bb;
}
}
}
}
/* This is the entry point of gathering non trapping memory accesses.
It will do a dominator walk over the whole function, and it will
make use of the bb->aux pointers. It returns a set of trees
(the MEM_REFs itself) which can't trap. */
static hash_set<tree> *
get_non_trapping (void)
{
nt_call_phase = 0;
hash_set<tree> *nontrap = new hash_set<tree>;
/* We're going to do a dominator walk, so ensure that we have
dominance information. */
calculate_dominance_info (CDI_DOMINATORS);
nontrapping_dom_walker (CDI_DOMINATORS, nontrap)
.walk (cfun->cfg->x_entry_block_ptr);
clear_aux_for_blocks ();
return nontrap;
}
/* Do the main work of conditional store replacement. We already know
that the recognized pattern looks like so:
split:
if (cond) goto MIDDLE_BB; else goto JOIN_BB (edge E1)
MIDDLE_BB:
something
fallthrough (edge E0)
JOIN_BB:
some more
We check that MIDDLE_BB contains only one store, that that store
doesn't trap (not via NOTRAP, but via checking if an access to the same
memory location dominates us) and that the store has a "simple" RHS. */
static bool
cond_store_replacement (basic_block middle_bb, basic_block join_bb,
edge e0, edge e1, hash_set<tree> *nontrap)
{
gimple *assign = last_and_only_stmt (middle_bb);
tree lhs, rhs, name, name2;
gphi *newphi;
gassign *new_stmt;
gimple_stmt_iterator gsi;
location_t locus;
/* Check if middle_bb contains of only one store. */
if (!assign
|| !gimple_assign_single_p (assign)
|| gimple_has_volatile_ops (assign))
return false;
locus = gimple_location (assign);
lhs = gimple_assign_lhs (assign);
rhs = gimple_assign_rhs1 (assign);
if (TREE_CODE (lhs) != MEM_REF
|| TREE_CODE (TREE_OPERAND (lhs, 0)) != SSA_NAME
|| !is_gimple_reg_type (TREE_TYPE (lhs)))
return false;
/* Prove that we can move the store down. We could also check
TREE_THIS_NOTRAP here, but in that case we also could move stores,
whose value is not available readily, which we want to avoid. */
if (!nontrap->contains (lhs))
return false;
/* Now we've checked the constraints, so do the transformation:
1) Remove the single store. */
gsi = gsi_for_stmt (assign);
unlink_stmt_vdef (assign);
gsi_remove (&gsi, true);
release_defs (assign);
/* Make both store and load use alias-set zero as we have to
deal with the case of the store being a conditional change
of the dynamic type. */
lhs = unshare_expr (lhs);
tree *basep = &lhs;
while (handled_component_p (*basep))
basep = &TREE_OPERAND (*basep, 0);
if (TREE_CODE (*basep) == MEM_REF
|| TREE_CODE (*basep) == TARGET_MEM_REF)
TREE_OPERAND (*basep, 1)
= fold_convert (ptr_type_node, TREE_OPERAND (*basep, 1));
else
*basep = build2 (MEM_REF, TREE_TYPE (*basep),
build_fold_addr_expr (*basep),
build_zero_cst (ptr_type_node));
/* 2) Insert a load from the memory of the store to the temporary
on the edge which did not contain the store. */
name = make_temp_ssa_name (TREE_TYPE (lhs), NULL, "cstore");
new_stmt = gimple_build_assign (name, lhs);
gimple_set_location (new_stmt, locus);
gsi_insert_on_edge (e1, new_stmt);
/* 3) Create a PHI node at the join block, with one argument
holding the old RHS, and the other holding the temporary
where we stored the old memory contents. */
name2 = make_temp_ssa_name (TREE_TYPE (lhs), NULL, "cstore");
newphi = create_phi_node (name2, join_bb);
add_phi_arg (newphi, rhs, e0, locus);
add_phi_arg (newphi, name, e1, locus);
lhs = unshare_expr (lhs);
new_stmt = gimple_build_assign (lhs, PHI_RESULT (newphi));
/* 4) Insert that PHI node. */
gsi = gsi_after_labels (join_bb);
if (gsi_end_p (gsi))
{
gsi = gsi_last_bb (join_bb);
gsi_insert_after (&gsi, new_stmt, GSI_NEW_STMT);
}
else
gsi_insert_before (&gsi, new_stmt, GSI_NEW_STMT);
return true;
}
/* Do the main work of conditional store replacement. */
static bool
cond_if_else_store_replacement_1 (basic_block then_bb, basic_block else_bb,
basic_block join_bb, gimple *then_assign,
gimple *else_assign)
{
tree lhs_base, lhs, then_rhs, else_rhs, name;
location_t then_locus, else_locus;
gimple_stmt_iterator gsi;
gphi *newphi;
gassign *new_stmt;
if (then_assign == NULL
|| !gimple_assign_single_p (then_assign)
|| gimple_clobber_p (then_assign)
|| gimple_has_volatile_ops (then_assign)
|| else_assign == NULL
|| !gimple_assign_single_p (else_assign)
|| gimple_clobber_p (else_assign)
|| gimple_has_volatile_ops (else_assign))
return false;
lhs = gimple_assign_lhs (then_assign);
if (!is_gimple_reg_type (TREE_TYPE (lhs))
|| !operand_equal_p (lhs, gimple_assign_lhs (else_assign), 0))
return false;
lhs_base = get_base_address (lhs);
if (lhs_base == NULL_TREE
|| (!DECL_P (lhs_base) && TREE_CODE (lhs_base) != MEM_REF))
return false;
then_rhs = gimple_assign_rhs1 (then_assign);
else_rhs = gimple_assign_rhs1 (else_assign);
then_locus = gimple_location (then_assign);
else_locus = gimple_location (else_assign);
/* Now we've checked the constraints, so do the transformation:
1) Remove the stores. */
gsi = gsi_for_stmt (then_assign);
unlink_stmt_vdef (then_assign);
gsi_remove (&gsi, true);
release_defs (then_assign);
gsi = gsi_for_stmt (else_assign);
unlink_stmt_vdef (else_assign);
gsi_remove (&gsi, true);
release_defs (else_assign);
/* 2) Create a PHI node at the join block, with one argument
holding the old RHS, and the other holding the temporary
where we stored the old memory contents. */
name = make_temp_ssa_name (TREE_TYPE (lhs), NULL, "cstore");
newphi = create_phi_node (name, join_bb);
add_phi_arg (newphi, then_rhs, EDGE_SUCC (then_bb, 0), then_locus);
add_phi_arg (newphi, else_rhs, EDGE_SUCC (else_bb, 0), else_locus);
new_stmt = gimple_build_assign (lhs, PHI_RESULT (newphi));
/* 3) Insert that PHI node. */
gsi = gsi_after_labels (join_bb);
if (gsi_end_p (gsi))
{
gsi = gsi_last_bb (join_bb);
gsi_insert_after (&gsi, new_stmt, GSI_NEW_STMT);
}
else
gsi_insert_before (&gsi, new_stmt, GSI_NEW_STMT);
return true;
}
/* Return the single store in BB with VDEF or NULL if there are
other stores in the BB or loads following the store. */
static gimple *
single_trailing_store_in_bb (basic_block bb, tree vdef)
{
if (SSA_NAME_IS_DEFAULT_DEF (vdef))
return NULL;
gimple *store = SSA_NAME_DEF_STMT (vdef);
if (gimple_bb (store) != bb
|| gimple_code (store) == GIMPLE_PHI)
return NULL;
/* Verify there is no other store in this BB. */
if (!SSA_NAME_IS_DEFAULT_DEF (gimple_vuse (store))
&& gimple_bb (SSA_NAME_DEF_STMT (gimple_vuse (store))) == bb
&& gimple_code (SSA_NAME_DEF_STMT (gimple_vuse (store))) != GIMPLE_PHI)
return NULL;
/* Verify there is no load or store after the store. */
use_operand_p use_p;
imm_use_iterator imm_iter;
FOR_EACH_IMM_USE_FAST (use_p, imm_iter, gimple_vdef (store))
if (USE_STMT (use_p) != store
&& gimple_bb (USE_STMT (use_p)) == bb)
return NULL;
return store;
}
/* Conditional store replacement. We already know
that the recognized pattern looks like so:
split:
if (cond) goto THEN_BB; else goto ELSE_BB (edge E1)
THEN_BB:
...
X = Y;
...
goto JOIN_BB;
ELSE_BB:
...
X = Z;
...
fallthrough (edge E0)
JOIN_BB:
some more
We check that it is safe to sink the store to JOIN_BB by verifying that
there are no read-after-write or write-after-write dependencies in
THEN_BB and ELSE_BB. */
static bool
cond_if_else_store_replacement (basic_block then_bb, basic_block else_bb,
basic_block join_bb)
{
vec<data_reference_p> then_datarefs, else_datarefs;
vec<ddr_p> then_ddrs, else_ddrs;
gimple *then_store, *else_store;
bool found, ok = false, res;
struct data_dependence_relation *ddr;
data_reference_p then_dr, else_dr;
int i, j;
tree then_lhs, else_lhs;
basic_block blocks[3];
/* Handle the case with single store in THEN_BB and ELSE_BB. That is
cheap enough to always handle as it allows us to elide dependence
checking. */
gphi *vphi = NULL;
for (gphi_iterator si = gsi_start_phis (join_bb); !gsi_end_p (si);
gsi_next (&si))
if (virtual_operand_p (gimple_phi_result (si.phi ())))
{
vphi = si.phi ();
break;
}
if (!vphi)
return false;
tree then_vdef = PHI_ARG_DEF_FROM_EDGE (vphi, single_succ_edge (then_bb));
tree else_vdef = PHI_ARG_DEF_FROM_EDGE (vphi, single_succ_edge (else_bb));
gimple *then_assign = single_trailing_store_in_bb (then_bb, then_vdef);
if (then_assign)
{
gimple *else_assign = single_trailing_store_in_bb (else_bb, else_vdef);
if (else_assign)
return cond_if_else_store_replacement_1 (then_bb, else_bb, join_bb,
then_assign, else_assign);
}
if (MAX_STORES_TO_SINK == 0)
return false;
/* Find data references. */
then_datarefs.create (1);
else_datarefs.create (1);
if ((find_data_references_in_bb (NULL, then_bb, &then_datarefs)
== chrec_dont_know)
|| !then_datarefs.length ()
|| (find_data_references_in_bb (NULL, else_bb, &else_datarefs)
== chrec_dont_know)
|| !else_datarefs.length ())
{
free_data_refs (then_datarefs);
free_data_refs (else_datarefs);
return false;
}
/* Find pairs of stores with equal LHS. */
auto_vec<gimple *, 1> then_stores, else_stores;
FOR_EACH_VEC_ELT (then_datarefs, i, then_dr)
{
if (DR_IS_READ (then_dr))
continue;
then_store = DR_STMT (then_dr);
then_lhs = gimple_get_lhs (then_store);
if (then_lhs == NULL_TREE)
continue;
found = false;
FOR_EACH_VEC_ELT (else_datarefs, j, else_dr)
{
if (DR_IS_READ (else_dr))
continue;
else_store = DR_STMT (else_dr);
else_lhs = gimple_get_lhs (else_store);
if (else_lhs == NULL_TREE)
continue;
if (operand_equal_p (then_lhs, else_lhs, 0))
{
found = true;
break;
}
}
if (!found)
continue;
then_stores.safe_push (then_store);
else_stores.safe_push (else_store);
}
/* No pairs of stores found. */
if (!then_stores.length ()
|| then_stores.length () > (unsigned) MAX_STORES_TO_SINK)
{
free_data_refs (then_datarefs);
free_data_refs (else_datarefs);
return false;
}
/* Compute and check data dependencies in both basic blocks. */
then_ddrs.create (1);
else_ddrs.create (1);
if (!compute_all_dependences (then_datarefs, &then_ddrs,
vNULL, false)
|| !compute_all_dependences (else_datarefs, &else_ddrs,
vNULL, false))
{
free_dependence_relations (then_ddrs);
free_dependence_relations (else_ddrs);
free_data_refs (then_datarefs);
free_data_refs (else_datarefs);
return false;
}
blocks[0] = then_bb;
blocks[1] = else_bb;
blocks[2] = join_bb;
renumber_gimple_stmt_uids_in_blocks (blocks, 3);
/* Check that there are no read-after-write or write-after-write dependencies
in THEN_BB. */
FOR_EACH_VEC_ELT (then_ddrs, i, ddr)
{
struct data_reference *dra = DDR_A (ddr);
struct data_reference *drb = DDR_B (ddr);
if (DDR_ARE_DEPENDENT (ddr) != chrec_known
&& ((DR_IS_READ (dra) && DR_IS_WRITE (drb)
&& gimple_uid (DR_STMT (dra)) > gimple_uid (DR_STMT (drb)))
|| (DR_IS_READ (drb) && DR_IS_WRITE (dra)
&& gimple_uid (DR_STMT (drb)) > gimple_uid (DR_STMT (dra)))
|| (DR_IS_WRITE (dra) && DR_IS_WRITE (drb))))
{
free_dependence_relations (then_ddrs);
free_dependence_relations (else_ddrs);
free_data_refs (then_datarefs);
free_data_refs (else_datarefs);
return false;
}
}
/* Check that there are no read-after-write or write-after-write dependencies
in ELSE_BB. */
FOR_EACH_VEC_ELT (else_ddrs, i, ddr)
{
struct data_reference *dra = DDR_A (ddr);
struct data_reference *drb = DDR_B (ddr);
if (DDR_ARE_DEPENDENT (ddr) != chrec_known
&& ((DR_IS_READ (dra) && DR_IS_WRITE (drb)
&& gimple_uid (DR_STMT (dra)) > gimple_uid (DR_STMT (drb)))
|| (DR_IS_READ (drb) && DR_IS_WRITE (dra)
&& gimple_uid (DR_STMT (drb)) > gimple_uid (DR_STMT (dra)))
|| (DR_IS_WRITE (dra) && DR_IS_WRITE (drb))))
{
free_dependence_relations (then_ddrs);
free_dependence_relations (else_ddrs);
free_data_refs (then_datarefs);
free_data_refs (else_datarefs);
return false;
}
}
/* Sink stores with same LHS. */
FOR_EACH_VEC_ELT (then_stores, i, then_store)
{
else_store = else_stores[i];
res = cond_if_else_store_replacement_1 (then_bb, else_bb, join_bb,
then_store, else_store);
ok = ok || res;
}
free_dependence_relations (then_ddrs);
free_dependence_relations (else_ddrs);
free_data_refs (then_datarefs);
free_data_refs (else_datarefs);
return ok;
}
/* Return TRUE if STMT has a VUSE whose corresponding VDEF is in BB. */
static bool
local_mem_dependence (gimple *stmt, basic_block bb)
{
tree vuse = gimple_vuse (stmt);
gimple *def;
if (!vuse)
return false;
def = SSA_NAME_DEF_STMT (vuse);
return (def && gimple_bb (def) == bb);
}
/* Given a "diamond" control-flow pattern where BB0 tests a condition,
BB1 and BB2 are "then" and "else" blocks dependent on this test,
and BB3 rejoins control flow following BB1 and BB2, look for
opportunities to hoist loads as follows. If BB3 contains a PHI of
two loads, one each occurring in BB1 and BB2, and the loads are
provably of adjacent fields in the same structure, then move both
loads into BB0. Of course this can only be done if there are no
dependencies preventing such motion.
One of the hoisted loads will always be speculative, so the
transformation is currently conservative:
- The fields must be strictly adjacent.
- The two fields must occupy a single memory block that is
guaranteed to not cross a page boundary.
The last is difficult to prove, as such memory blocks should be
aligned on the minimum of the stack alignment boundary and the
alignment guaranteed by heap allocation interfaces. Thus we rely
on a parameter for the alignment value.
Provided a good value is used for the last case, the first
restriction could possibly be relaxed. */
static void
hoist_adjacent_loads (basic_block bb0, basic_block bb1,
basic_block bb2, basic_block bb3)
{
int param_align = PARAM_VALUE (PARAM_L1_CACHE_LINE_SIZE);
unsigned param_align_bits = (unsigned) (param_align * BITS_PER_UNIT);
gphi_iterator gsi;
/* Walk the phis in bb3 looking for an opportunity. We are looking
for phis of two SSA names, one each of which is defined in bb1 and
bb2. */
for (gsi = gsi_start_phis (bb3); !gsi_end_p (gsi); gsi_next (&gsi))
{
gphi *phi_stmt = gsi.phi ();
gimple *def1, *def2;
tree arg1, arg2, ref1, ref2, field1, field2;
tree tree_offset1, tree_offset2, tree_size2, next;
int offset1, offset2, size2;
unsigned align1;
gimple_stmt_iterator gsi2;
basic_block bb_for_def1, bb_for_def2;
if (gimple_phi_num_args (phi_stmt) != 2
|| virtual_operand_p (gimple_phi_result (phi_stmt)))
continue;
arg1 = gimple_phi_arg_def (phi_stmt, 0);
arg2 = gimple_phi_arg_def (phi_stmt, 1);
if (TREE_CODE (arg1) != SSA_NAME
|| TREE_CODE (arg2) != SSA_NAME
|| SSA_NAME_IS_DEFAULT_DEF (arg1)
|| SSA_NAME_IS_DEFAULT_DEF (arg2))
continue;
def1 = SSA_NAME_DEF_STMT (arg1);
def2 = SSA_NAME_DEF_STMT (arg2);
if ((gimple_bb (def1) != bb1 || gimple_bb (def2) != bb2)
&& (gimple_bb (def2) != bb1 || gimple_bb (def1) != bb2))
continue;
/* Check the mode of the arguments to be sure a conditional move
can be generated for it. */
if (optab_handler (movcc_optab, TYPE_MODE (TREE_TYPE (arg1)))
== CODE_FOR_nothing)
continue;
/* Both statements must be assignments whose RHS is a COMPONENT_REF. */
if (!gimple_assign_single_p (def1)
|| !gimple_assign_single_p (def2)
|| gimple_has_volatile_ops (def1)
|| gimple_has_volatile_ops (def2))
continue;
ref1 = gimple_assign_rhs1 (def1);
ref2 = gimple_assign_rhs1 (def2);
if (TREE_CODE (ref1) != COMPONENT_REF
|| TREE_CODE (ref2) != COMPONENT_REF)
continue;
/* The zeroth operand of the two component references must be
identical. It is not sufficient to compare get_base_address of
the two references, because this could allow for different
elements of the same array in the two trees. It is not safe to
assume that the existence of one array element implies the
existence of a different one. */
if (!operand_equal_p (TREE_OPERAND (ref1, 0), TREE_OPERAND (ref2, 0), 0))
continue;
field1 = TREE_OPERAND (ref1, 1);
field2 = TREE_OPERAND (ref2, 1);
/* Check for field adjacency, and ensure field1 comes first. */
for (next = DECL_CHAIN (field1);
next && TREE_CODE (next) != FIELD_DECL;
next = DECL_CHAIN (next))
;
if (next != field2)
{
for (next = DECL_CHAIN (field2);
next && TREE_CODE (next) != FIELD_DECL;
next = DECL_CHAIN (next))
;
if (next != field1)
continue;
std::swap (field1, field2);
std::swap (def1, def2);
}
bb_for_def1 = gimple_bb (def1);
bb_for_def2 = gimple_bb (def2);
/* Check for proper alignment of the first field. */
tree_offset1 = bit_position (field1);
tree_offset2 = bit_position (field2);
tree_size2 = DECL_SIZE (field2);
if (!tree_fits_uhwi_p (tree_offset1)
|| !tree_fits_uhwi_p (tree_offset2)
|| !tree_fits_uhwi_p (tree_size2))
continue;
offset1 = tree_to_uhwi (tree_offset1);
offset2 = tree_to_uhwi (tree_offset2);
size2 = tree_to_uhwi (tree_size2);
align1 = DECL_ALIGN (field1) % param_align_bits;
if (offset1 % BITS_PER_UNIT != 0)
continue;
/* For profitability, the two field references should fit within
a single cache line. */
if (align1 + offset2 - offset1 + size2 > param_align_bits)
continue;
/* The two expressions cannot be dependent upon vdefs defined
in bb1/bb2. */
if (local_mem_dependence (def1, bb_for_def1)
|| local_mem_dependence (def2, bb_for_def2))
continue;
/* The conditions are satisfied; hoist the loads from bb1 and bb2 into
bb0. We hoist the first one first so that a cache miss is handled
efficiently regardless of hardware cache-fill policy. */
gsi2 = gsi_for_stmt (def1);
gsi_move_to_bb_end (&gsi2, bb0);
gsi2 = gsi_for_stmt (def2);
gsi_move_to_bb_end (&gsi2, bb0);
if (dump_file && (dump_flags & TDF_DETAILS))
{
fprintf (dump_file,
"\nHoisting adjacent loads from %d and %d into %d: \n",
bb_for_def1->index, bb_for_def2->index, bb0->index);
print_gimple_stmt (dump_file, def1, 0, TDF_VOPS|TDF_MEMSYMS);
print_gimple_stmt (dump_file, def2, 0, TDF_VOPS|TDF_MEMSYMS);
}
}
}
/* Determine whether we should attempt to hoist adjacent loads out of
diamond patterns in pass_phiopt. Always hoist loads if
-fhoist-adjacent-loads is specified and the target machine has
both a conditional move instruction and a defined cache line size. */
static bool
gate_hoist_loads (void)
{
return (flag_hoist_adjacent_loads == 1
&& PARAM_VALUE (PARAM_L1_CACHE_LINE_SIZE)
&& HAVE_conditional_move);
}
/* This pass tries to replaces an if-then-else block with an
assignment. We have four kinds of transformations. Some of these
transformations are also performed by the ifcvt RTL optimizer.
Conditional Replacement
-----------------------
This transformation, implemented in conditional_replacement,
replaces
bb0:
if (cond) goto bb2; else goto bb1;
bb1:
bb2:
x = PHI <0 (bb1), 1 (bb0), ...>;
with
bb0:
x' = cond;
goto bb2;
bb2:
x = PHI <x' (bb0), ...>;
We remove bb1 as it becomes unreachable. This occurs often due to
gimplification of conditionals.
Value Replacement
-----------------
This transformation, implemented in value_replacement, replaces
bb0:
if (a != b) goto bb2; else goto bb1;
bb1:
bb2:
x = PHI <a (bb1), b (bb0), ...>;
with
bb0:
bb2:
x = PHI <b (bb0), ...>;
This opportunity can sometimes occur as a result of other
optimizations.
Another case caught by value replacement looks like this:
bb0:
t1 = a == CONST;
t2 = b > c;
t3 = t1 & t2;
if (t3 != 0) goto bb1; else goto bb2;
bb1:
bb2:
x = PHI (CONST, a)
Gets replaced with:
bb0:
bb2:
t1 = a == CONST;
t2 = b > c;
t3 = t1 & t2;
x = a;
ABS Replacement
---------------
This transformation, implemented in abs_replacement, replaces
bb0:
if (a >= 0) goto bb2; else goto bb1;
bb1:
x = -a;
bb2:
x = PHI <x (bb1), a (bb0), ...>;
with
bb0:
x' = ABS_EXPR< a >;
bb2:
x = PHI <x' (bb0), ...>;
MIN/MAX Replacement
-------------------
This transformation, minmax_replacement replaces
bb0:
if (a <= b) goto bb2; else goto bb1;
bb1:
bb2:
x = PHI <b (bb1), a (bb0), ...>;
with
bb0:
x' = MIN_EXPR (a, b)
bb2:
x = PHI <x' (bb0), ...>;
A similar transformation is done for MAX_EXPR.
This pass also performs a fifth transformation of a slightly different
flavor.
Factor conversion in COND_EXPR
------------------------------
This transformation factors the conversion out of COND_EXPR with
factor_out_conditional_conversion.
For example:
if (a <= CST) goto <bb 3>; else goto <bb 4>;
<bb 3>:
tmp = (int) a;
<bb 4>:
tmp = PHI <tmp, CST>
Into:
if (a <= CST) goto <bb 3>; else goto <bb 4>;
<bb 3>:
<bb 4>:
a = PHI <a, CST>
tmp = (int) a;
Adjacent Load Hoisting
----------------------
This transformation replaces
bb0:
if (...) goto bb2; else goto bb1;
bb1:
x1 = (<expr>).field1;
goto bb3;
bb2:
x2 = (<expr>).field2;
bb3:
# x = PHI <x1, x2>;
with
bb0:
x1 = (<expr>).field1;
x2 = (<expr>).field2;
if (...) goto bb2; else goto bb1;
bb1:
goto bb3;
bb2:
bb3:
# x = PHI <x1, x2>;
The purpose of this transformation is to enable generation of conditional
move instructions such as Intel CMOVE or PowerPC ISEL. Because one of
the loads is speculative, the transformation is restricted to very
specific cases to avoid introducing a page fault. We are looking for
the common idiom:
if (...)
x = y->left;
else
x = y->right;
where left and right are typically adjacent pointers in a tree structure. */
namespace {
const pass_data pass_data_phiopt =
{
GIMPLE_PASS, /* type */
"phiopt", /* name */
OPTGROUP_NONE, /* optinfo_flags */
TV_TREE_PHIOPT, /* tv_id */
( PROP_cfg | PROP_ssa ), /* properties_required */
0, /* properties_provided */
0, /* properties_destroyed */
0, /* todo_flags_start */
0, /* todo_flags_finish */
};
class pass_phiopt : public gimple_opt_pass
{
public:
pass_phiopt (gcc::context *ctxt)
: gimple_opt_pass (pass_data_phiopt, ctxt), early_p (false)
{}
/* opt_pass methods: */
opt_pass * clone () { return new pass_phiopt (m_ctxt); }
void set_pass_param (unsigned n, bool param)
{
gcc_assert (n == 0);
early_p = param;
}
virtual bool gate (function *) { return flag_ssa_phiopt; }
virtual unsigned int execute (function *)
{
return tree_ssa_phiopt_worker (false,
!early_p ? gate_hoist_loads () : false,
early_p);
}
private:
bool early_p;
}; // class pass_phiopt
} // anon namespace
gimple_opt_pass *
make_pass_phiopt (gcc::context *ctxt)
{
return new pass_phiopt (ctxt);
}
namespace {
const pass_data pass_data_cselim =
{
GIMPLE_PASS, /* type */
"cselim", /* name */
OPTGROUP_NONE, /* optinfo_flags */
TV_TREE_PHIOPT, /* tv_id */
( PROP_cfg | PROP_ssa ), /* properties_required */
0, /* properties_provided */
0, /* properties_destroyed */
0, /* todo_flags_start */
0, /* todo_flags_finish */
};
class pass_cselim : public gimple_opt_pass
{
public:
pass_cselim (gcc::context *ctxt)
: gimple_opt_pass (pass_data_cselim, ctxt)
{}
/* opt_pass methods: */
virtual bool gate (function *) { return flag_tree_cselim; }
virtual unsigned int execute (function *) { return tree_ssa_cs_elim (); }
}; // class pass_cselim
} // anon namespace
gimple_opt_pass *
make_pass_cselim (gcc::context *ctxt)
{
return new pass_cselim (ctxt);
}
|