1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
|
/* Inlining decision heuristics.
Copyright (C) 2003-2015 Free Software Foundation, Inc.
Contributed by Jan Hubicka
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/>. */
/* Analysis used by the inliner and other passes limiting code size growth.
We estimate for each function
- function body size
- average function execution time
- inlining size benefit (that is how much of function body size
and its call sequence is expected to disappear by inlining)
- inlining time benefit
- function frame size
For each call
- call statement size and time
inlinie_summary datastructures store above information locally (i.e.
parameters of the function itself) and globally (i.e. parameters of
the function created by applying all the inline decisions already
present in the callgraph).
We provide accestor to the inline_summary datastructure and
basic logic updating the parameters when inlining is performed.
The summaries are context sensitive. Context means
1) partial assignment of known constant values of operands
2) whether function is inlined into the call or not.
It is easy to add more variants. To represent function size and time
that depends on context (i.e. it is known to be optimized away when
context is known either by inlining or from IP-CP and clonning),
we use predicates. Predicates are logical formulas in
conjunctive-disjunctive form consisting of clauses. Clauses are bitmaps
specifying what conditions must be true. Conditions are simple test
of the form described above.
In order to make predicate (possibly) true, all of its clauses must
be (possibly) true. To make clause (possibly) true, one of conditions
it mentions must be (possibly) true. There are fixed bounds on
number of clauses and conditions and all the manipulation functions
are conservative in positive direction. I.e. we may lose precision
by thinking that predicate may be true even when it is not.
estimate_edge_size and estimate_edge_growth can be used to query
function size/time in the given context. inline_merge_summary merges
properties of caller and callee after inlining.
Finally pass_inline_parameters is exported. This is used to drive
computation of function parameters used by the early inliner. IPA
inlined performs analysis via its analyze_function method. */
#include "config.h"
#include "system.h"
#include "coretypes.h"
#include "tm.h"
#include "hash-set.h"
#include "machmode.h"
#include "vec.h"
#include "double-int.h"
#include "input.h"
#include "alias.h"
#include "symtab.h"
#include "wide-int.h"
#include "inchash.h"
#include "real.h"
#include "tree.h"
#include "fold-const.h"
#include "stor-layout.h"
#include "stringpool.h"
#include "print-tree.h"
#include "tree-inline.h"
#include "langhooks.h"
#include "flags.h"
#include "diagnostic.h"
#include "gimple-pretty-print.h"
#include "params.h"
#include "tree-pass.h"
#include "coverage.h"
#include "predict.h"
#include "hard-reg-set.h"
#include "input.h"
#include "function.h"
#include "dominance.h"
#include "cfg.h"
#include "cfganal.h"
#include "basic-block.h"
#include "tree-ssa-alias.h"
#include "internal-fn.h"
#include "gimple-expr.h"
#include "is-a.h"
#include "gimple.h"
#include "gimple-iterator.h"
#include "gimple-ssa.h"
#include "tree-cfg.h"
#include "tree-phinodes.h"
#include "ssa-iterators.h"
#include "tree-ssanames.h"
#include "tree-ssa-loop-niter.h"
#include "tree-ssa-loop.h"
#include "hash-map.h"
#include "plugin-api.h"
#include "ipa-ref.h"
#include "cgraph.h"
#include "alloc-pool.h"
#include "symbol-summary.h"
#include "ipa-prop.h"
#include "lto-streamer.h"
#include "data-streamer.h"
#include "tree-streamer.h"
#include "ipa-inline.h"
#include "cfgloop.h"
#include "tree-scalar-evolution.h"
#include "ipa-utils.h"
#include "cilk.h"
#include "cfgexpand.h"
/* Estimate runtime of function can easilly run into huge numbers with many
nested loops. Be sure we can compute time * INLINE_SIZE_SCALE * 2 in an
integer. For anything larger we use gcov_type. */
#define MAX_TIME 500000
/* Number of bits in integer, but we really want to be stable across different
hosts. */
#define NUM_CONDITIONS 32
enum predicate_conditions
{
predicate_false_condition = 0,
predicate_not_inlined_condition = 1,
predicate_first_dynamic_condition = 2
};
/* Special condition code we use to represent test that operand is compile time
constant. */
#define IS_NOT_CONSTANT ERROR_MARK
/* Special condition code we use to represent test that operand is not changed
across invocation of the function. When operand IS_NOT_CONSTANT it is always
CHANGED, however i.e. loop invariants can be NOT_CHANGED given percentage
of executions even when they are not compile time constants. */
#define CHANGED IDENTIFIER_NODE
/* Holders of ipa cgraph hooks: */
static struct cgraph_2edge_hook_list *edge_duplication_hook_holder;
static struct cgraph_edge_hook_list *edge_removal_hook_holder;
static void inline_edge_removal_hook (struct cgraph_edge *, void *);
static void inline_edge_duplication_hook (struct cgraph_edge *,
struct cgraph_edge *, void *);
/* VECtor holding inline summaries.
In GGC memory because conditions might point to constant trees. */
function_summary <inline_summary *> *inline_summaries;
vec<inline_edge_summary_t> inline_edge_summary_vec;
/* Cached node/edge growths. */
vec<edge_growth_cache_entry> edge_growth_cache;
/* Edge predicates goes here. */
static alloc_pool edge_predicate_pool;
/* Return true predicate (tautology).
We represent it by empty list of clauses. */
static inline struct predicate
true_predicate (void)
{
struct predicate p;
p.clause[0] = 0;
return p;
}
/* Return predicate testing single condition number COND. */
static inline struct predicate
single_cond_predicate (int cond)
{
struct predicate p;
p.clause[0] = 1 << cond;
p.clause[1] = 0;
return p;
}
/* Return false predicate. First clause require false condition. */
static inline struct predicate
false_predicate (void)
{
return single_cond_predicate (predicate_false_condition);
}
/* Return true if P is (true). */
static inline bool
true_predicate_p (struct predicate *p)
{
return !p->clause[0];
}
/* Return true if P is (false). */
static inline bool
false_predicate_p (struct predicate *p)
{
if (p->clause[0] == (1 << predicate_false_condition))
{
gcc_checking_assert (!p->clause[1]
&& p->clause[0] == 1 << predicate_false_condition);
return true;
}
return false;
}
/* Return predicate that is set true when function is not inlined. */
static inline struct predicate
not_inlined_predicate (void)
{
return single_cond_predicate (predicate_not_inlined_condition);
}
/* Simple description of whether a memory load or a condition refers to a load
from an aggregate and if so, how and where from in the aggregate.
Individual fields have the same meaning like fields with the same name in
struct condition. */
struct agg_position_info
{
HOST_WIDE_INT offset;
bool agg_contents;
bool by_ref;
};
/* Add condition to condition list CONDS. AGGPOS describes whether the used
oprand is loaded from an aggregate and where in the aggregate it is. It can
be NULL, which means this not a load from an aggregate. */
static struct predicate
add_condition (struct inline_summary *summary, int operand_num,
struct agg_position_info *aggpos,
enum tree_code code, tree val)
{
int i;
struct condition *c;
struct condition new_cond;
HOST_WIDE_INT offset;
bool agg_contents, by_ref;
if (aggpos)
{
offset = aggpos->offset;
agg_contents = aggpos->agg_contents;
by_ref = aggpos->by_ref;
}
else
{
offset = 0;
agg_contents = false;
by_ref = false;
}
gcc_checking_assert (operand_num >= 0);
for (i = 0; vec_safe_iterate (summary->conds, i, &c); i++)
{
if (c->operand_num == operand_num
&& c->code == code
&& c->val == val
&& c->agg_contents == agg_contents
&& (!agg_contents || (c->offset == offset && c->by_ref == by_ref)))
return single_cond_predicate (i + predicate_first_dynamic_condition);
}
/* Too many conditions. Give up and return constant true. */
if (i == NUM_CONDITIONS - predicate_first_dynamic_condition)
return true_predicate ();
new_cond.operand_num = operand_num;
new_cond.code = code;
new_cond.val = val;
new_cond.agg_contents = agg_contents;
new_cond.by_ref = by_ref;
new_cond.offset = offset;
vec_safe_push (summary->conds, new_cond);
return single_cond_predicate (i + predicate_first_dynamic_condition);
}
/* Add clause CLAUSE into the predicate P. */
static inline void
add_clause (conditions conditions, struct predicate *p, clause_t clause)
{
int i;
int i2;
int insert_here = -1;
int c1, c2;
/* True clause. */
if (!clause)
return;
/* False clause makes the whole predicate false. Kill the other variants. */
if (clause == (1 << predicate_false_condition))
{
p->clause[0] = (1 << predicate_false_condition);
p->clause[1] = 0;
return;
}
if (false_predicate_p (p))
return;
/* No one should be silly enough to add false into nontrivial clauses. */
gcc_checking_assert (!(clause & (1 << predicate_false_condition)));
/* Look where to insert the clause. At the same time prune out
clauses of P that are implied by the new clause and thus
redundant. */
for (i = 0, i2 = 0; i <= MAX_CLAUSES; i++)
{
p->clause[i2] = p->clause[i];
if (!p->clause[i])
break;
/* If p->clause[i] implies clause, there is nothing to add. */
if ((p->clause[i] & clause) == p->clause[i])
{
/* We had nothing to add, none of clauses should've become
redundant. */
gcc_checking_assert (i == i2);
return;
}
if (p->clause[i] < clause && insert_here < 0)
insert_here = i2;
/* If clause implies p->clause[i], then p->clause[i] becomes redundant.
Otherwise the p->clause[i] has to stay. */
if ((p->clause[i] & clause) != clause)
i2++;
}
/* Look for clauses that are obviously true. I.e.
op0 == 5 || op0 != 5. */
for (c1 = predicate_first_dynamic_condition; c1 < NUM_CONDITIONS; c1++)
{
condition *cc1;
if (!(clause & (1 << c1)))
continue;
cc1 = &(*conditions)[c1 - predicate_first_dynamic_condition];
/* We have no way to represent !CHANGED and !IS_NOT_CONSTANT
and thus there is no point for looking for them. */
if (cc1->code == CHANGED || cc1->code == IS_NOT_CONSTANT)
continue;
for (c2 = c1 + 1; c2 < NUM_CONDITIONS; c2++)
if (clause & (1 << c2))
{
condition *cc1 =
&(*conditions)[c1 - predicate_first_dynamic_condition];
condition *cc2 =
&(*conditions)[c2 - predicate_first_dynamic_condition];
if (cc1->operand_num == cc2->operand_num
&& cc1->val == cc2->val
&& cc2->code != IS_NOT_CONSTANT
&& cc2->code != CHANGED
&& cc1->code == invert_tree_comparison (cc2->code,
HONOR_NANS (cc1->val)))
return;
}
}
/* We run out of variants. Be conservative in positive direction. */
if (i2 == MAX_CLAUSES)
return;
/* Keep clauses in decreasing order. This makes equivalence testing easy. */
p->clause[i2 + 1] = 0;
if (insert_here >= 0)
for (; i2 > insert_here; i2--)
p->clause[i2] = p->clause[i2 - 1];
else
insert_here = i2;
p->clause[insert_here] = clause;
}
/* Return P & P2. */
static struct predicate
and_predicates (conditions conditions,
struct predicate *p, struct predicate *p2)
{
struct predicate out = *p;
int i;
/* Avoid busy work. */
if (false_predicate_p (p2) || true_predicate_p (p))
return *p2;
if (false_predicate_p (p) || true_predicate_p (p2))
return *p;
/* See how far predicates match. */
for (i = 0; p->clause[i] && p->clause[i] == p2->clause[i]; i++)
{
gcc_checking_assert (i < MAX_CLAUSES);
}
/* Combine the predicates rest. */
for (; p2->clause[i]; i++)
{
gcc_checking_assert (i < MAX_CLAUSES);
add_clause (conditions, &out, p2->clause[i]);
}
return out;
}
/* Return true if predicates are obviously equal. */
static inline bool
predicates_equal_p (struct predicate *p, struct predicate *p2)
{
int i;
for (i = 0; p->clause[i]; i++)
{
gcc_checking_assert (i < MAX_CLAUSES);
gcc_checking_assert (p->clause[i] > p->clause[i + 1]);
gcc_checking_assert (!p2->clause[i]
|| p2->clause[i] > p2->clause[i + 1]);
if (p->clause[i] != p2->clause[i])
return false;
}
return !p2->clause[i];
}
/* Return P | P2. */
static struct predicate
or_predicates (conditions conditions,
struct predicate *p, struct predicate *p2)
{
struct predicate out = true_predicate ();
int i, j;
/* Avoid busy work. */
if (false_predicate_p (p2) || true_predicate_p (p))
return *p;
if (false_predicate_p (p) || true_predicate_p (p2))
return *p2;
if (predicates_equal_p (p, p2))
return *p;
/* OK, combine the predicates. */
for (i = 0; p->clause[i]; i++)
for (j = 0; p2->clause[j]; j++)
{
gcc_checking_assert (i < MAX_CLAUSES && j < MAX_CLAUSES);
add_clause (conditions, &out, p->clause[i] | p2->clause[j]);
}
return out;
}
/* Having partial truth assignment in POSSIBLE_TRUTHS, return false
if predicate P is known to be false. */
static bool
evaluate_predicate (struct predicate *p, clause_t possible_truths)
{
int i;
/* True remains true. */
if (true_predicate_p (p))
return true;
gcc_assert (!(possible_truths & (1 << predicate_false_condition)));
/* See if we can find clause we can disprove. */
for (i = 0; p->clause[i]; i++)
{
gcc_checking_assert (i < MAX_CLAUSES);
if (!(p->clause[i] & possible_truths))
return false;
}
return true;
}
/* Return the probability in range 0...REG_BR_PROB_BASE that the predicated
instruction will be recomputed per invocation of the inlined call. */
static int
predicate_probability (conditions conds,
struct predicate *p, clause_t possible_truths,
vec<inline_param_summary> inline_param_summary)
{
int i;
int combined_prob = REG_BR_PROB_BASE;
/* True remains true. */
if (true_predicate_p (p))
return REG_BR_PROB_BASE;
if (false_predicate_p (p))
return 0;
gcc_assert (!(possible_truths & (1 << predicate_false_condition)));
/* See if we can find clause we can disprove. */
for (i = 0; p->clause[i]; i++)
{
gcc_checking_assert (i < MAX_CLAUSES);
if (!(p->clause[i] & possible_truths))
return 0;
else
{
int this_prob = 0;
int i2;
if (!inline_param_summary.exists ())
return REG_BR_PROB_BASE;
for (i2 = 0; i2 < NUM_CONDITIONS; i2++)
if ((p->clause[i] & possible_truths) & (1 << i2))
{
if (i2 >= predicate_first_dynamic_condition)
{
condition *c =
&(*conds)[i2 - predicate_first_dynamic_condition];
if (c->code == CHANGED
&& (c->operand_num <
(int) inline_param_summary.length ()))
{
int iprob =
inline_param_summary[c->operand_num].change_prob;
this_prob = MAX (this_prob, iprob);
}
else
this_prob = REG_BR_PROB_BASE;
}
else
this_prob = REG_BR_PROB_BASE;
}
combined_prob = MIN (this_prob, combined_prob);
if (!combined_prob)
return 0;
}
}
return combined_prob;
}
/* Dump conditional COND. */
static void
dump_condition (FILE *f, conditions conditions, int cond)
{
condition *c;
if (cond == predicate_false_condition)
fprintf (f, "false");
else if (cond == predicate_not_inlined_condition)
fprintf (f, "not inlined");
else
{
c = &(*conditions)[cond - predicate_first_dynamic_condition];
fprintf (f, "op%i", c->operand_num);
if (c->agg_contents)
fprintf (f, "[%soffset: " HOST_WIDE_INT_PRINT_DEC "]",
c->by_ref ? "ref " : "", c->offset);
if (c->code == IS_NOT_CONSTANT)
{
fprintf (f, " not constant");
return;
}
if (c->code == CHANGED)
{
fprintf (f, " changed");
return;
}
fprintf (f, " %s ", op_symbol_code (c->code));
print_generic_expr (f, c->val, 1);
}
}
/* Dump clause CLAUSE. */
static void
dump_clause (FILE *f, conditions conds, clause_t clause)
{
int i;
bool found = false;
fprintf (f, "(");
if (!clause)
fprintf (f, "true");
for (i = 0; i < NUM_CONDITIONS; i++)
if (clause & (1 << i))
{
if (found)
fprintf (f, " || ");
found = true;
dump_condition (f, conds, i);
}
fprintf (f, ")");
}
/* Dump predicate PREDICATE. */
static void
dump_predicate (FILE *f, conditions conds, struct predicate *pred)
{
int i;
if (true_predicate_p (pred))
dump_clause (f, conds, 0);
else
for (i = 0; pred->clause[i]; i++)
{
if (i)
fprintf (f, " && ");
dump_clause (f, conds, pred->clause[i]);
}
fprintf (f, "\n");
}
/* Dump inline hints. */
void
dump_inline_hints (FILE *f, inline_hints hints)
{
if (!hints)
return;
fprintf (f, "inline hints:");
if (hints & INLINE_HINT_indirect_call)
{
hints &= ~INLINE_HINT_indirect_call;
fprintf (f, " indirect_call");
}
if (hints & INLINE_HINT_loop_iterations)
{
hints &= ~INLINE_HINT_loop_iterations;
fprintf (f, " loop_iterations");
}
if (hints & INLINE_HINT_loop_stride)
{
hints &= ~INLINE_HINT_loop_stride;
fprintf (f, " loop_stride");
}
if (hints & INLINE_HINT_same_scc)
{
hints &= ~INLINE_HINT_same_scc;
fprintf (f, " same_scc");
}
if (hints & INLINE_HINT_in_scc)
{
hints &= ~INLINE_HINT_in_scc;
fprintf (f, " in_scc");
}
if (hints & INLINE_HINT_cross_module)
{
hints &= ~INLINE_HINT_cross_module;
fprintf (f, " cross_module");
}
if (hints & INLINE_HINT_declared_inline)
{
hints &= ~INLINE_HINT_declared_inline;
fprintf (f, " declared_inline");
}
if (hints & INLINE_HINT_array_index)
{
hints &= ~INLINE_HINT_array_index;
fprintf (f, " array_index");
}
if (hints & INLINE_HINT_known_hot)
{
hints &= ~INLINE_HINT_known_hot;
fprintf (f, " known_hot");
}
gcc_assert (!hints);
}
/* Record SIZE and TIME under condition PRED into the inline summary. */
static void
account_size_time (struct inline_summary *summary, int size, int time,
struct predicate *pred)
{
size_time_entry *e;
bool found = false;
int i;
if (false_predicate_p (pred))
return;
/* We need to create initial empty unconitional clause, but otherwie
we don't need to account empty times and sizes. */
if (!size && !time && summary->entry)
return;
/* Watch overflow that might result from insane profiles. */
if (time > MAX_TIME * INLINE_TIME_SCALE)
time = MAX_TIME * INLINE_TIME_SCALE;
gcc_assert (time >= 0);
for (i = 0; vec_safe_iterate (summary->entry, i, &e); i++)
if (predicates_equal_p (&e->predicate, pred))
{
found = true;
break;
}
if (i == 256)
{
i = 0;
found = true;
e = &(*summary->entry)[0];
gcc_assert (!e->predicate.clause[0]);
if (dump_file && (dump_flags & TDF_DETAILS))
fprintf (dump_file,
"\t\tReached limit on number of entries, "
"ignoring the predicate.");
}
if (dump_file && (dump_flags & TDF_DETAILS) && (time || size))
{
fprintf (dump_file,
"\t\tAccounting size:%3.2f, time:%3.2f on %spredicate:",
((double) size) / INLINE_SIZE_SCALE,
((double) time) / INLINE_TIME_SCALE, found ? "" : "new ");
dump_predicate (dump_file, summary->conds, pred);
}
if (!found)
{
struct size_time_entry new_entry;
new_entry.size = size;
new_entry.time = time;
new_entry.predicate = *pred;
vec_safe_push (summary->entry, new_entry);
}
else
{
e->size += size;
e->time += time;
if (e->time > MAX_TIME * INLINE_TIME_SCALE)
e->time = MAX_TIME * INLINE_TIME_SCALE;
}
}
/* We proved E to be unreachable, redirect it to __bultin_unreachable. */
static struct cgraph_edge *
redirect_to_unreachable (struct cgraph_edge *e)
{
struct cgraph_node *callee = !e->inline_failed ? e->callee : NULL;
struct cgraph_node *target = cgraph_node::get_create
(builtin_decl_implicit (BUILT_IN_UNREACHABLE));
if (e->speculative)
e = e->resolve_speculation (target->decl);
else if (!e->callee)
e->make_direct (target);
else
e->redirect_callee (target);
struct inline_edge_summary *es = inline_edge_summary (e);
e->inline_failed = CIF_UNREACHABLE;
e->frequency = 0;
e->count = 0;
es->call_stmt_size = 0;
es->call_stmt_time = 0;
if (callee)
callee->remove_symbol_and_inline_clones ();
return e;
}
/* Set predicate for edge E. */
static void
edge_set_predicate (struct cgraph_edge *e, struct predicate *predicate)
{
/* If the edge is determined to be never executed, redirect it
to BUILTIN_UNREACHABLE to save inliner from inlining into it. */
if (predicate && false_predicate_p (predicate)
/* When handling speculative edges, we need to do the redirection
just once. Do it always on the direct edge, so we do not
attempt to resolve speculation while duplicating the edge. */
&& (!e->speculative || e->callee))
e = redirect_to_unreachable (e);
struct inline_edge_summary *es = inline_edge_summary (e);
if (predicate && !true_predicate_p (predicate))
{
if (!es->predicate)
es->predicate = (struct predicate *) pool_alloc (edge_predicate_pool);
*es->predicate = *predicate;
}
else
{
if (es->predicate)
pool_free (edge_predicate_pool, es->predicate);
es->predicate = NULL;
}
}
/* Set predicate for hint *P. */
static void
set_hint_predicate (struct predicate **p, struct predicate new_predicate)
{
if (false_predicate_p (&new_predicate) || true_predicate_p (&new_predicate))
{
if (*p)
pool_free (edge_predicate_pool, *p);
*p = NULL;
}
else
{
if (!*p)
*p = (struct predicate *) pool_alloc (edge_predicate_pool);
**p = new_predicate;
}
}
/* KNOWN_VALS is partial mapping of parameters of NODE to constant values.
KNOWN_AGGS is a vector of aggreggate jump functions for each parameter.
Return clause of possible truths. When INLINE_P is true, assume that we are
inlining.
ERROR_MARK means compile time invariant. */
static clause_t
evaluate_conditions_for_known_args (struct cgraph_node *node,
bool inline_p,
vec<tree> known_vals,
vec<ipa_agg_jump_function_p>
known_aggs)
{
clause_t clause = inline_p ? 0 : 1 << predicate_not_inlined_condition;
struct inline_summary *info = inline_summaries->get (node);
int i;
struct condition *c;
for (i = 0; vec_safe_iterate (info->conds, i, &c); i++)
{
tree val;
tree res;
/* We allow call stmt to have fewer arguments than the callee function
(especially for K&R style programs). So bound check here (we assume
known_aggs vector, if non-NULL, has the same length as
known_vals). */
gcc_checking_assert (!known_aggs.exists ()
|| (known_vals.length () == known_aggs.length ()));
if (c->operand_num >= (int) known_vals.length ())
{
clause |= 1 << (i + predicate_first_dynamic_condition);
continue;
}
if (c->agg_contents)
{
struct ipa_agg_jump_function *agg;
if (c->code == CHANGED
&& !c->by_ref
&& (known_vals[c->operand_num] == error_mark_node))
continue;
if (known_aggs.exists ())
{
agg = known_aggs[c->operand_num];
val = ipa_find_agg_cst_for_param (agg, c->offset, c->by_ref);
}
else
val = NULL_TREE;
}
else
{
val = known_vals[c->operand_num];
if (val == error_mark_node && c->code != CHANGED)
val = NULL_TREE;
}
if (!val)
{
clause |= 1 << (i + predicate_first_dynamic_condition);
continue;
}
if (c->code == IS_NOT_CONSTANT || c->code == CHANGED)
continue;
if (operand_equal_p (TYPE_SIZE (TREE_TYPE (c->val)),
TYPE_SIZE (TREE_TYPE (val)), 0))
{
val = fold_unary (VIEW_CONVERT_EXPR, TREE_TYPE (c->val), val);
res = val
? fold_binary_to_constant (c->code, boolean_type_node, val, c->val)
: NULL;
if (res && integer_zerop (res))
continue;
}
clause |= 1 << (i + predicate_first_dynamic_condition);
}
return clause;
}
/* Work out what conditions might be true at invocation of E. */
static void
evaluate_properties_for_edge (struct cgraph_edge *e, bool inline_p,
clause_t *clause_ptr,
vec<tree> *known_vals_ptr,
vec<ipa_polymorphic_call_context>
*known_contexts_ptr,
vec<ipa_agg_jump_function_p> *known_aggs_ptr)
{
struct cgraph_node *callee = e->callee->ultimate_alias_target ();
struct inline_summary *info = inline_summaries->get (callee);
vec<tree> known_vals = vNULL;
vec<ipa_agg_jump_function_p> known_aggs = vNULL;
if (clause_ptr)
*clause_ptr = inline_p ? 0 : 1 << predicate_not_inlined_condition;
if (known_vals_ptr)
known_vals_ptr->create (0);
if (known_contexts_ptr)
known_contexts_ptr->create (0);
if (ipa_node_params_sum
&& !e->call_stmt_cannot_inline_p
&& ((clause_ptr && info->conds) || known_vals_ptr || known_contexts_ptr))
{
struct ipa_node_params *parms_info;
struct ipa_edge_args *args = IPA_EDGE_REF (e);
struct inline_edge_summary *es = inline_edge_summary (e);
int i, count = ipa_get_cs_argument_count (args);
if (e->caller->global.inlined_to)
parms_info = IPA_NODE_REF (e->caller->global.inlined_to);
else
parms_info = IPA_NODE_REF (e->caller);
if (count && (info->conds || known_vals_ptr))
known_vals.safe_grow_cleared (count);
if (count && (info->conds || known_aggs_ptr))
known_aggs.safe_grow_cleared (count);
if (count && known_contexts_ptr)
known_contexts_ptr->safe_grow_cleared (count);
for (i = 0; i < count; i++)
{
struct ipa_jump_func *jf = ipa_get_ith_jump_func (args, i);
tree cst = ipa_value_from_jfunc (parms_info, jf);
if (!cst && e->call_stmt
&& i < (int)gimple_call_num_args (e->call_stmt))
{
cst = gimple_call_arg (e->call_stmt, i);
if (!is_gimple_min_invariant (cst))
cst = NULL;
}
if (cst)
{
gcc_checking_assert (TREE_CODE (cst) != TREE_BINFO);
if (known_vals.exists ())
known_vals[i] = cst;
}
else if (inline_p && !es->param[i].change_prob)
known_vals[i] = error_mark_node;
if (known_contexts_ptr)
(*known_contexts_ptr)[i] = ipa_context_from_jfunc (parms_info, e,
i, jf);
/* TODO: When IPA-CP starts propagating and merging aggregate jump
functions, use its knowledge of the caller too, just like the
scalar case above. */
known_aggs[i] = &jf->agg;
}
}
else if (e->call_stmt && !e->call_stmt_cannot_inline_p
&& ((clause_ptr && info->conds) || known_vals_ptr))
{
int i, count = (int)gimple_call_num_args (e->call_stmt);
if (count && (info->conds || known_vals_ptr))
known_vals.safe_grow_cleared (count);
for (i = 0; i < count; i++)
{
tree cst = gimple_call_arg (e->call_stmt, i);
if (!is_gimple_min_invariant (cst))
cst = NULL;
if (cst)
known_vals[i] = cst;
}
}
if (clause_ptr)
*clause_ptr = evaluate_conditions_for_known_args (callee, inline_p,
known_vals, known_aggs);
if (known_vals_ptr)
*known_vals_ptr = known_vals;
else
known_vals.release ();
if (known_aggs_ptr)
*known_aggs_ptr = known_aggs;
else
known_aggs.release ();
}
/* Allocate the inline summary vector or resize it to cover all cgraph nodes. */
static void
inline_summary_alloc (void)
{
if (!edge_removal_hook_holder)
edge_removal_hook_holder =
symtab->add_edge_removal_hook (&inline_edge_removal_hook, NULL);
if (!edge_duplication_hook_holder)
edge_duplication_hook_holder =
symtab->add_edge_duplication_hook (&inline_edge_duplication_hook, NULL);
if (!inline_summaries)
inline_summaries = (inline_summary_t*) inline_summary_t::create_ggc (symtab);
if (inline_edge_summary_vec.length () <= (unsigned) symtab->edges_max_uid)
inline_edge_summary_vec.safe_grow_cleared (symtab->edges_max_uid + 1);
if (!edge_predicate_pool)
edge_predicate_pool = create_alloc_pool ("edge predicates",
sizeof (struct predicate), 10);
}
/* We are called multiple time for given function; clear
data from previous run so they are not cumulated. */
static void
reset_inline_edge_summary (struct cgraph_edge *e)
{
if (e->uid < (int) inline_edge_summary_vec.length ())
{
struct inline_edge_summary *es = inline_edge_summary (e);
es->call_stmt_size = es->call_stmt_time = 0;
if (es->predicate)
pool_free (edge_predicate_pool, es->predicate);
es->predicate = NULL;
es->param.release ();
}
}
/* We are called multiple time for given function; clear
data from previous run so they are not cumulated. */
static void
reset_inline_summary (struct cgraph_node *node,
inline_summary *info)
{
struct cgraph_edge *e;
info->self_size = info->self_time = 0;
info->estimated_stack_size = 0;
info->estimated_self_stack_size = 0;
info->stack_frame_offset = 0;
info->size = 0;
info->time = 0;
info->growth = 0;
info->scc_no = 0;
if (info->loop_iterations)
{
pool_free (edge_predicate_pool, info->loop_iterations);
info->loop_iterations = NULL;
}
if (info->loop_stride)
{
pool_free (edge_predicate_pool, info->loop_stride);
info->loop_stride = NULL;
}
if (info->array_index)
{
pool_free (edge_predicate_pool, info->array_index);
info->array_index = NULL;
}
vec_free (info->conds);
vec_free (info->entry);
for (e = node->callees; e; e = e->next_callee)
reset_inline_edge_summary (e);
for (e = node->indirect_calls; e; e = e->next_callee)
reset_inline_edge_summary (e);
}
/* Hook that is called by cgraph.c when a node is removed. */
void
inline_summary_t::remove (cgraph_node *node, inline_summary *info)
{
reset_inline_summary (node, info);
}
/* Remap predicate P of former function to be predicate of duplicated function.
POSSIBLE_TRUTHS is clause of possible truths in the duplicated node,
INFO is inline summary of the duplicated node. */
static struct predicate
remap_predicate_after_duplication (struct predicate *p,
clause_t possible_truths,
struct inline_summary *info)
{
struct predicate new_predicate = true_predicate ();
int j;
for (j = 0; p->clause[j]; j++)
if (!(possible_truths & p->clause[j]))
{
new_predicate = false_predicate ();
break;
}
else
add_clause (info->conds, &new_predicate,
possible_truths & p->clause[j]);
return new_predicate;
}
/* Same as remap_predicate_after_duplication but handle hint predicate *P.
Additionally care about allocating new memory slot for updated predicate
and set it to NULL when it becomes true or false (and thus uninteresting).
*/
static void
remap_hint_predicate_after_duplication (struct predicate **p,
clause_t possible_truths,
struct inline_summary *info)
{
struct predicate new_predicate;
if (!*p)
return;
new_predicate = remap_predicate_after_duplication (*p,
possible_truths, info);
/* We do not want to free previous predicate; it is used by node origin. */
*p = NULL;
set_hint_predicate (p, new_predicate);
}
/* Hook that is called by cgraph.c when a node is duplicated. */
void
inline_summary_t::duplicate (cgraph_node *src,
cgraph_node *dst,
inline_summary *,
inline_summary *info)
{
inline_summary_alloc ();
memcpy (info, inline_summaries->get (src), sizeof (inline_summary));
/* TODO: as an optimization, we may avoid copying conditions
that are known to be false or true. */
info->conds = vec_safe_copy (info->conds);
/* When there are any replacements in the function body, see if we can figure
out that something was optimized out. */
if (ipa_node_params_sum && dst->clone.tree_map)
{
vec<size_time_entry, va_gc> *entry = info->entry;
/* Use SRC parm info since it may not be copied yet. */
struct ipa_node_params *parms_info = IPA_NODE_REF (src);
vec<tree> known_vals = vNULL;
int count = ipa_get_param_count (parms_info);
int i, j;
clause_t possible_truths;
struct predicate true_pred = true_predicate ();
size_time_entry *e;
int optimized_out_size = 0;
bool inlined_to_p = false;
struct cgraph_edge *edge, *next;
info->entry = 0;
known_vals.safe_grow_cleared (count);
for (i = 0; i < count; i++)
{
struct ipa_replace_map *r;
for (j = 0; vec_safe_iterate (dst->clone.tree_map, j, &r); j++)
{
if (((!r->old_tree && r->parm_num == i)
|| (r->old_tree && r->old_tree == ipa_get_param (parms_info, i)))
&& r->replace_p && !r->ref_p)
{
known_vals[i] = r->new_tree;
break;
}
}
}
possible_truths = evaluate_conditions_for_known_args (dst, false,
known_vals,
vNULL);
known_vals.release ();
account_size_time (info, 0, 0, &true_pred);
/* Remap size_time vectors.
Simplify the predicate by prunning out alternatives that are known
to be false.
TODO: as on optimization, we can also eliminate conditions known
to be true. */
for (i = 0; vec_safe_iterate (entry, i, &e); i++)
{
struct predicate new_predicate;
new_predicate = remap_predicate_after_duplication (&e->predicate,
possible_truths,
info);
if (false_predicate_p (&new_predicate))
optimized_out_size += e->size;
else
account_size_time (info, e->size, e->time, &new_predicate);
}
/* Remap edge predicates with the same simplification as above.
Also copy constantness arrays. */
for (edge = dst->callees; edge; edge = next)
{
struct predicate new_predicate;
struct inline_edge_summary *es = inline_edge_summary (edge);
next = edge->next_callee;
if (!edge->inline_failed)
inlined_to_p = true;
if (!es->predicate)
continue;
new_predicate = remap_predicate_after_duplication (es->predicate,
possible_truths,
info);
if (false_predicate_p (&new_predicate)
&& !false_predicate_p (es->predicate))
optimized_out_size += es->call_stmt_size * INLINE_SIZE_SCALE;
edge_set_predicate (edge, &new_predicate);
}
/* Remap indirect edge predicates with the same simplificaiton as above.
Also copy constantness arrays. */
for (edge = dst->indirect_calls; edge; edge = next)
{
struct predicate new_predicate;
struct inline_edge_summary *es = inline_edge_summary (edge);
next = edge->next_callee;
gcc_checking_assert (edge->inline_failed);
if (!es->predicate)
continue;
new_predicate = remap_predicate_after_duplication (es->predicate,
possible_truths,
info);
if (false_predicate_p (&new_predicate)
&& !false_predicate_p (es->predicate))
optimized_out_size += es->call_stmt_size * INLINE_SIZE_SCALE;
edge_set_predicate (edge, &new_predicate);
}
remap_hint_predicate_after_duplication (&info->loop_iterations,
possible_truths, info);
remap_hint_predicate_after_duplication (&info->loop_stride,
possible_truths, info);
remap_hint_predicate_after_duplication (&info->array_index,
possible_truths, info);
/* If inliner or someone after inliner will ever start producing
non-trivial clones, we will get trouble with lack of information
about updating self sizes, because size vectors already contains
sizes of the calees. */
gcc_assert (!inlined_to_p || !optimized_out_size);
}
else
{
info->entry = vec_safe_copy (info->entry);
if (info->loop_iterations)
{
predicate p = *info->loop_iterations;
info->loop_iterations = NULL;
set_hint_predicate (&info->loop_iterations, p);
}
if (info->loop_stride)
{
predicate p = *info->loop_stride;
info->loop_stride = NULL;
set_hint_predicate (&info->loop_stride, p);
}
if (info->array_index)
{
predicate p = *info->array_index;
info->array_index = NULL;
set_hint_predicate (&info->array_index, p);
}
}
if (!dst->global.inlined_to)
inline_update_overall_summary (dst);
}
/* Hook that is called by cgraph.c when a node is duplicated. */
static void
inline_edge_duplication_hook (struct cgraph_edge *src,
struct cgraph_edge *dst,
ATTRIBUTE_UNUSED void *data)
{
struct inline_edge_summary *info;
struct inline_edge_summary *srcinfo;
inline_summary_alloc ();
info = inline_edge_summary (dst);
srcinfo = inline_edge_summary (src);
memcpy (info, srcinfo, sizeof (struct inline_edge_summary));
info->predicate = NULL;
edge_set_predicate (dst, srcinfo->predicate);
info->param = srcinfo->param.copy ();
if (!dst->indirect_unknown_callee && src->indirect_unknown_callee)
{
info->call_stmt_size -= (eni_size_weights.indirect_call_cost
- eni_size_weights.call_cost);
info->call_stmt_time -= (eni_time_weights.indirect_call_cost
- eni_time_weights.call_cost);
}
}
/* Keep edge cache consistent across edge removal. */
static void
inline_edge_removal_hook (struct cgraph_edge *edge,
void *data ATTRIBUTE_UNUSED)
{
if (edge_growth_cache.exists ())
reset_edge_growth_cache (edge);
reset_inline_edge_summary (edge);
}
/* Initialize growth caches. */
void
initialize_growth_caches (void)
{
if (symtab->edges_max_uid)
edge_growth_cache.safe_grow_cleared (symtab->edges_max_uid);
}
/* Free growth caches. */
void
free_growth_caches (void)
{
edge_growth_cache.release ();
}
/* Dump edge summaries associated to NODE and recursively to all clones.
Indent by INDENT. */
static void
dump_inline_edge_summary (FILE *f, int indent, struct cgraph_node *node,
struct inline_summary *info)
{
struct cgraph_edge *edge;
for (edge = node->callees; edge; edge = edge->next_callee)
{
struct inline_edge_summary *es = inline_edge_summary (edge);
struct cgraph_node *callee = edge->callee->ultimate_alias_target ();
int i;
fprintf (f,
"%*s%s/%i %s\n%*s loop depth:%2i freq:%4i size:%2i"
" time: %2i callee size:%2i stack:%2i",
indent, "", callee->name (), callee->order,
!edge->inline_failed
? "inlined" : cgraph_inline_failed_string (edge-> inline_failed),
indent, "", es->loop_depth, edge->frequency,
es->call_stmt_size, es->call_stmt_time,
(int) inline_summaries->get (callee)->size / INLINE_SIZE_SCALE,
(int) inline_summaries->get (callee)->estimated_stack_size);
if (es->predicate)
{
fprintf (f, " predicate: ");
dump_predicate (f, info->conds, es->predicate);
}
else
fprintf (f, "\n");
if (es->param.exists ())
for (i = 0; i < (int) es->param.length (); i++)
{
int prob = es->param[i].change_prob;
if (!prob)
fprintf (f, "%*s op%i is compile time invariant\n",
indent + 2, "", i);
else if (prob != REG_BR_PROB_BASE)
fprintf (f, "%*s op%i change %f%% of time\n", indent + 2, "", i,
prob * 100.0 / REG_BR_PROB_BASE);
}
if (!edge->inline_failed)
{
fprintf (f, "%*sStack frame offset %i, callee self size %i,"
" callee size %i\n",
indent + 2, "",
(int) inline_summaries->get (callee)->stack_frame_offset,
(int) inline_summaries->get (callee)->estimated_self_stack_size,
(int) inline_summaries->get (callee)->estimated_stack_size);
dump_inline_edge_summary (f, indent + 2, callee, info);
}
}
for (edge = node->indirect_calls; edge; edge = edge->next_callee)
{
struct inline_edge_summary *es = inline_edge_summary (edge);
fprintf (f, "%*sindirect call loop depth:%2i freq:%4i size:%2i"
" time: %2i",
indent, "",
es->loop_depth,
edge->frequency, es->call_stmt_size, es->call_stmt_time);
if (es->predicate)
{
fprintf (f, "predicate: ");
dump_predicate (f, info->conds, es->predicate);
}
else
fprintf (f, "\n");
}
}
void
dump_inline_summary (FILE *f, struct cgraph_node *node)
{
if (node->definition)
{
struct inline_summary *s = inline_summaries->get (node);
size_time_entry *e;
int i;
fprintf (f, "Inline summary for %s/%i", node->name (),
node->order);
if (DECL_DISREGARD_INLINE_LIMITS (node->decl))
fprintf (f, " always_inline");
if (s->inlinable)
fprintf (f, " inlinable");
if (s->contains_cilk_spawn)
fprintf (f, " contains_cilk_spawn");
fprintf (f, "\n self time: %i\n", s->self_time);
fprintf (f, " global time: %i\n", s->time);
fprintf (f, " self size: %i\n", s->self_size);
fprintf (f, " global size: %i\n", s->size);
fprintf (f, " min size: %i\n", s->min_size);
fprintf (f, " self stack: %i\n",
(int) s->estimated_self_stack_size);
fprintf (f, " global stack: %i\n", (int) s->estimated_stack_size);
if (s->growth)
fprintf (f, " estimated growth:%i\n", (int) s->growth);
if (s->scc_no)
fprintf (f, " In SCC: %i\n", (int) s->scc_no);
for (i = 0; vec_safe_iterate (s->entry, i, &e); i++)
{
fprintf (f, " size:%f, time:%f, predicate:",
(double) e->size / INLINE_SIZE_SCALE,
(double) e->time / INLINE_TIME_SCALE);
dump_predicate (f, s->conds, &e->predicate);
}
if (s->loop_iterations)
{
fprintf (f, " loop iterations:");
dump_predicate (f, s->conds, s->loop_iterations);
}
if (s->loop_stride)
{
fprintf (f, " loop stride:");
dump_predicate (f, s->conds, s->loop_stride);
}
if (s->array_index)
{
fprintf (f, " array index:");
dump_predicate (f, s->conds, s->array_index);
}
fprintf (f, " calls:\n");
dump_inline_edge_summary (f, 4, node, s);
fprintf (f, "\n");
}
}
DEBUG_FUNCTION void
debug_inline_summary (struct cgraph_node *node)
{
dump_inline_summary (stderr, node);
}
void
dump_inline_summaries (FILE *f)
{
struct cgraph_node *node;
FOR_EACH_DEFINED_FUNCTION (node)
if (!node->global.inlined_to)
dump_inline_summary (f, node);
}
/* Give initial reasons why inlining would fail on EDGE. This gets either
nullified or usually overwritten by more precise reasons later. */
void
initialize_inline_failed (struct cgraph_edge *e)
{
struct cgraph_node *callee = e->callee;
if (e->indirect_unknown_callee)
e->inline_failed = CIF_INDIRECT_UNKNOWN_CALL;
else if (!callee->definition)
e->inline_failed = CIF_BODY_NOT_AVAILABLE;
else if (callee->local.redefined_extern_inline)
e->inline_failed = CIF_REDEFINED_EXTERN_INLINE;
else if (e->call_stmt_cannot_inline_p)
e->inline_failed = CIF_MISMATCHED_ARGUMENTS;
else if (cfun && fn_contains_cilk_spawn_p (cfun))
/* We can't inline if the function is spawing a function. */
e->inline_failed = CIF_FUNCTION_NOT_INLINABLE;
else
e->inline_failed = CIF_FUNCTION_NOT_CONSIDERED;
}
/* Callback of walk_aliased_vdefs. Flags that it has been invoked to the
boolean variable pointed to by DATA. */
static bool
mark_modified (ao_ref *ao ATTRIBUTE_UNUSED, tree vdef ATTRIBUTE_UNUSED,
void *data)
{
bool *b = (bool *) data;
*b = true;
return true;
}
/* If OP refers to value of function parameter, return the corresponding
parameter. */
static tree
unmodified_parm_1 (gimple stmt, tree op)
{
/* SSA_NAME referring to parm default def? */
if (TREE_CODE (op) == SSA_NAME
&& SSA_NAME_IS_DEFAULT_DEF (op)
&& TREE_CODE (SSA_NAME_VAR (op)) == PARM_DECL)
return SSA_NAME_VAR (op);
/* Non-SSA parm reference? */
if (TREE_CODE (op) == PARM_DECL)
{
bool modified = false;
ao_ref refd;
ao_ref_init (&refd, op);
walk_aliased_vdefs (&refd, gimple_vuse (stmt), mark_modified, &modified,
NULL);
if (!modified)
return op;
}
return NULL_TREE;
}
/* If OP refers to value of function parameter, return the corresponding
parameter. Also traverse chains of SSA register assignments. */
static tree
unmodified_parm (gimple stmt, tree op)
{
tree res = unmodified_parm_1 (stmt, op);
if (res)
return res;
if (TREE_CODE (op) == SSA_NAME
&& !SSA_NAME_IS_DEFAULT_DEF (op)
&& gimple_assign_single_p (SSA_NAME_DEF_STMT (op)))
return unmodified_parm (SSA_NAME_DEF_STMT (op),
gimple_assign_rhs1 (SSA_NAME_DEF_STMT (op)));
return NULL_TREE;
}
/* If OP refers to a value of a function parameter or value loaded from an
aggregate passed to a parameter (either by value or reference), return TRUE
and store the number of the parameter to *INDEX_P and information whether
and how it has been loaded from an aggregate into *AGGPOS. INFO describes
the function parameters, STMT is the statement in which OP is used or
loaded. */
static bool
unmodified_parm_or_parm_agg_item (struct ipa_func_body_info *fbi,
gimple stmt, tree op, int *index_p,
struct agg_position_info *aggpos)
{
tree res = unmodified_parm_1 (stmt, op);
gcc_checking_assert (aggpos);
if (res)
{
*index_p = ipa_get_param_decl_index (fbi->info, res);
if (*index_p < 0)
return false;
aggpos->agg_contents = false;
aggpos->by_ref = false;
return true;
}
if (TREE_CODE (op) == SSA_NAME)
{
if (SSA_NAME_IS_DEFAULT_DEF (op)
|| !gimple_assign_single_p (SSA_NAME_DEF_STMT (op)))
return false;
stmt = SSA_NAME_DEF_STMT (op);
op = gimple_assign_rhs1 (stmt);
if (!REFERENCE_CLASS_P (op))
return unmodified_parm_or_parm_agg_item (fbi, stmt, op, index_p,
aggpos);
}
aggpos->agg_contents = true;
return ipa_load_from_parm_agg (fbi, fbi->info->descriptors,
stmt, op, index_p, &aggpos->offset,
NULL, &aggpos->by_ref);
}
/* See if statement might disappear after inlining.
0 - means not eliminated
1 - half of statements goes away
2 - for sure it is eliminated.
We are not terribly sophisticated, basically looking for simple abstraction
penalty wrappers. */
static int
eliminated_by_inlining_prob (gimple stmt)
{
enum gimple_code code = gimple_code (stmt);
enum tree_code rhs_code;
if (!optimize)
return 0;
switch (code)
{
case GIMPLE_RETURN:
return 2;
case GIMPLE_ASSIGN:
if (gimple_num_ops (stmt) != 2)
return 0;
rhs_code = gimple_assign_rhs_code (stmt);
/* Casts of parameters, loads from parameters passed by reference
and stores to return value or parameters are often free after
inlining dua to SRA and further combining.
Assume that half of statements goes away. */
if (CONVERT_EXPR_CODE_P (rhs_code)
|| rhs_code == VIEW_CONVERT_EXPR
|| rhs_code == ADDR_EXPR
|| gimple_assign_rhs_class (stmt) == GIMPLE_SINGLE_RHS)
{
tree rhs = gimple_assign_rhs1 (stmt);
tree lhs = gimple_assign_lhs (stmt);
tree inner_rhs = get_base_address (rhs);
tree inner_lhs = get_base_address (lhs);
bool rhs_free = false;
bool lhs_free = false;
if (!inner_rhs)
inner_rhs = rhs;
if (!inner_lhs)
inner_lhs = lhs;
/* Reads of parameter are expected to be free. */
if (unmodified_parm (stmt, inner_rhs))
rhs_free = true;
/* Match expressions of form &this->field. Those will most likely
combine with something upstream after inlining. */
else if (TREE_CODE (inner_rhs) == ADDR_EXPR)
{
tree op = get_base_address (TREE_OPERAND (inner_rhs, 0));
if (TREE_CODE (op) == PARM_DECL)
rhs_free = true;
else if (TREE_CODE (op) == MEM_REF
&& unmodified_parm (stmt, TREE_OPERAND (op, 0)))
rhs_free = true;
}
/* When parameter is not SSA register because its address is taken
and it is just copied into one, the statement will be completely
free after inlining (we will copy propagate backward). */
if (rhs_free && is_gimple_reg (lhs))
return 2;
/* Reads of parameters passed by reference
expected to be free (i.e. optimized out after inlining). */
if (TREE_CODE (inner_rhs) == MEM_REF
&& unmodified_parm (stmt, TREE_OPERAND (inner_rhs, 0)))
rhs_free = true;
/* Copying parameter passed by reference into gimple register is
probably also going to copy propagate, but we can't be quite
sure. */
if (rhs_free && is_gimple_reg (lhs))
lhs_free = true;
/* Writes to parameters, parameters passed by value and return value
(either dirrectly or passed via invisible reference) are free.
TODO: We ought to handle testcase like
struct a {int a,b;};
struct a
retrurnsturct (void)
{
struct a a ={1,2};
return a;
}
This translate into:
retrurnsturct ()
{
int a$b;
int a$a;
struct a a;
struct a D.2739;
<bb 2>:
D.2739.a = 1;
D.2739.b = 2;
return D.2739;
}
For that we either need to copy ipa-split logic detecting writes
to return value. */
if (TREE_CODE (inner_lhs) == PARM_DECL
|| TREE_CODE (inner_lhs) == RESULT_DECL
|| (TREE_CODE (inner_lhs) == MEM_REF
&& (unmodified_parm (stmt, TREE_OPERAND (inner_lhs, 0))
|| (TREE_CODE (TREE_OPERAND (inner_lhs, 0)) == SSA_NAME
&& SSA_NAME_VAR (TREE_OPERAND (inner_lhs, 0))
&& TREE_CODE (SSA_NAME_VAR (TREE_OPERAND
(inner_lhs,
0))) == RESULT_DECL))))
lhs_free = true;
if (lhs_free
&& (is_gimple_reg (rhs) || is_gimple_min_invariant (rhs)))
rhs_free = true;
if (lhs_free && rhs_free)
return 1;
}
return 0;
default:
return 0;
}
}
/* If BB ends by a conditional we can turn into predicates, attach corresponding
predicates to the CFG edges. */
static void
set_cond_stmt_execution_predicate (struct ipa_func_body_info *fbi,
struct inline_summary *summary,
basic_block bb)
{
gimple last;
tree op;
int index;
struct agg_position_info aggpos;
enum tree_code code, inverted_code;
edge e;
edge_iterator ei;
gimple set_stmt;
tree op2;
last = last_stmt (bb);
if (!last || gimple_code (last) != GIMPLE_COND)
return;
if (!is_gimple_ip_invariant (gimple_cond_rhs (last)))
return;
op = gimple_cond_lhs (last);
/* TODO: handle conditionals like
var = op0 < 4;
if (var != 0). */
if (unmodified_parm_or_parm_agg_item (fbi, last, op, &index, &aggpos))
{
code = gimple_cond_code (last);
inverted_code = invert_tree_comparison (code, HONOR_NANS (op));
FOR_EACH_EDGE (e, ei, bb->succs)
{
enum tree_code this_code = (e->flags & EDGE_TRUE_VALUE
? code : inverted_code);
/* invert_tree_comparison will return ERROR_MARK on FP
comparsions that are not EQ/NE instead of returning proper
unordered one. Be sure it is not confused with NON_CONSTANT. */
if (this_code != ERROR_MARK)
{
struct predicate p = add_condition (summary, index, &aggpos,
this_code,
gimple_cond_rhs (last));
e->aux = pool_alloc (edge_predicate_pool);
*(struct predicate *) e->aux = p;
}
}
}
if (TREE_CODE (op) != SSA_NAME)
return;
/* Special case
if (builtin_constant_p (op))
constant_code
else
nonconstant_code.
Here we can predicate nonconstant_code. We can't
really handle constant_code since we have no predicate
for this and also the constant code is not known to be
optimized away when inliner doen't see operand is constant.
Other optimizers might think otherwise. */
if (gimple_cond_code (last) != NE_EXPR
|| !integer_zerop (gimple_cond_rhs (last)))
return;
set_stmt = SSA_NAME_DEF_STMT (op);
if (!gimple_call_builtin_p (set_stmt, BUILT_IN_CONSTANT_P)
|| gimple_call_num_args (set_stmt) != 1)
return;
op2 = gimple_call_arg (set_stmt, 0);
if (!unmodified_parm_or_parm_agg_item (fbi, set_stmt, op2, &index, &aggpos))
return;
FOR_EACH_EDGE (e, ei, bb->succs) if (e->flags & EDGE_FALSE_VALUE)
{
struct predicate p = add_condition (summary, index, &aggpos,
IS_NOT_CONSTANT, NULL_TREE);
e->aux = pool_alloc (edge_predicate_pool);
*(struct predicate *) e->aux = p;
}
}
/* If BB ends by a switch we can turn into predicates, attach corresponding
predicates to the CFG edges. */
static void
set_switch_stmt_execution_predicate (struct ipa_func_body_info *fbi,
struct inline_summary *summary,
basic_block bb)
{
gimple lastg;
tree op;
int index;
struct agg_position_info aggpos;
edge e;
edge_iterator ei;
size_t n;
size_t case_idx;
lastg = last_stmt (bb);
if (!lastg || gimple_code (lastg) != GIMPLE_SWITCH)
return;
gswitch *last = as_a <gswitch *> (lastg);
op = gimple_switch_index (last);
if (!unmodified_parm_or_parm_agg_item (fbi, last, op, &index, &aggpos))
return;
FOR_EACH_EDGE (e, ei, bb->succs)
{
e->aux = pool_alloc (edge_predicate_pool);
*(struct predicate *) e->aux = false_predicate ();
}
n = gimple_switch_num_labels (last);
for (case_idx = 0; case_idx < n; ++case_idx)
{
tree cl = gimple_switch_label (last, case_idx);
tree min, max;
struct predicate p;
e = find_edge (bb, label_to_block (CASE_LABEL (cl)));
min = CASE_LOW (cl);
max = CASE_HIGH (cl);
/* For default we might want to construct predicate that none
of cases is met, but it is bit hard to do not having negations
of conditionals handy. */
if (!min && !max)
p = true_predicate ();
else if (!max)
p = add_condition (summary, index, &aggpos, EQ_EXPR, min);
else
{
struct predicate p1, p2;
p1 = add_condition (summary, index, &aggpos, GE_EXPR, min);
p2 = add_condition (summary, index, &aggpos, LE_EXPR, max);
p = and_predicates (summary->conds, &p1, &p2);
}
*(struct predicate *) e->aux
= or_predicates (summary->conds, &p, (struct predicate *) e->aux);
}
}
/* For each BB in NODE attach to its AUX pointer predicate under
which it is executable. */
static void
compute_bb_predicates (struct ipa_func_body_info *fbi,
struct cgraph_node *node,
struct inline_summary *summary)
{
struct function *my_function = DECL_STRUCT_FUNCTION (node->decl);
bool done = false;
basic_block bb;
FOR_EACH_BB_FN (bb, my_function)
{
set_cond_stmt_execution_predicate (fbi, summary, bb);
set_switch_stmt_execution_predicate (fbi, summary, bb);
}
/* Entry block is always executable. */
ENTRY_BLOCK_PTR_FOR_FN (my_function)->aux
= pool_alloc (edge_predicate_pool);
*(struct predicate *) ENTRY_BLOCK_PTR_FOR_FN (my_function)->aux
= true_predicate ();
/* A simple dataflow propagation of predicates forward in the CFG.
TODO: work in reverse postorder. */
while (!done)
{
done = true;
FOR_EACH_BB_FN (bb, my_function)
{
struct predicate p = false_predicate ();
edge e;
edge_iterator ei;
FOR_EACH_EDGE (e, ei, bb->preds)
{
if (e->src->aux)
{
struct predicate this_bb_predicate
= *(struct predicate *) e->src->aux;
if (e->aux)
this_bb_predicate
= and_predicates (summary->conds, &this_bb_predicate,
(struct predicate *) e->aux);
p = or_predicates (summary->conds, &p, &this_bb_predicate);
if (true_predicate_p (&p))
break;
}
}
if (false_predicate_p (&p))
gcc_assert (!bb->aux);
else
{
if (!bb->aux)
{
done = false;
bb->aux = pool_alloc (edge_predicate_pool);
*((struct predicate *) bb->aux) = p;
}
else if (!predicates_equal_p (&p, (struct predicate *) bb->aux))
{
/* This OR operation is needed to ensure monotonous data flow
in the case we hit the limit on number of clauses and the
and/or operations above give approximate answers. */
p = or_predicates (summary->conds, &p, (struct predicate *)bb->aux);
if (!predicates_equal_p (&p, (struct predicate *) bb->aux))
{
done = false;
*((struct predicate *) bb->aux) = p;
}
}
}
}
}
}
/* We keep info about constantness of SSA names. */
typedef struct predicate predicate_t;
/* Return predicate specifying when the STMT might have result that is not
a compile time constant. */
static struct predicate
will_be_nonconstant_expr_predicate (struct ipa_node_params *info,
struct inline_summary *summary,
tree expr,
vec<predicate_t> nonconstant_names)
{
tree parm;
int index;
while (UNARY_CLASS_P (expr))
expr = TREE_OPERAND (expr, 0);
parm = unmodified_parm (NULL, expr);
if (parm && (index = ipa_get_param_decl_index (info, parm)) >= 0)
return add_condition (summary, index, NULL, CHANGED, NULL_TREE);
if (is_gimple_min_invariant (expr))
return false_predicate ();
if (TREE_CODE (expr) == SSA_NAME)
return nonconstant_names[SSA_NAME_VERSION (expr)];
if (BINARY_CLASS_P (expr) || COMPARISON_CLASS_P (expr))
{
struct predicate p1 = will_be_nonconstant_expr_predicate
(info, summary, TREE_OPERAND (expr, 0),
nonconstant_names);
struct predicate p2;
if (true_predicate_p (&p1))
return p1;
p2 = will_be_nonconstant_expr_predicate (info, summary,
TREE_OPERAND (expr, 1),
nonconstant_names);
return or_predicates (summary->conds, &p1, &p2);
}
else if (TREE_CODE (expr) == COND_EXPR)
{
struct predicate p1 = will_be_nonconstant_expr_predicate
(info, summary, TREE_OPERAND (expr, 0),
nonconstant_names);
struct predicate p2;
if (true_predicate_p (&p1))
return p1;
p2 = will_be_nonconstant_expr_predicate (info, summary,
TREE_OPERAND (expr, 1),
nonconstant_names);
if (true_predicate_p (&p2))
return p2;
p1 = or_predicates (summary->conds, &p1, &p2);
p2 = will_be_nonconstant_expr_predicate (info, summary,
TREE_OPERAND (expr, 2),
nonconstant_names);
return or_predicates (summary->conds, &p1, &p2);
}
else
{
debug_tree (expr);
gcc_unreachable ();
}
return false_predicate ();
}
/* Return predicate specifying when the STMT might have result that is not
a compile time constant. */
static struct predicate
will_be_nonconstant_predicate (struct ipa_func_body_info *fbi,
struct inline_summary *summary,
gimple stmt,
vec<predicate_t> nonconstant_names)
{
struct predicate p = true_predicate ();
ssa_op_iter iter;
tree use;
struct predicate op_non_const;
bool is_load;
int base_index;
struct agg_position_info aggpos;
/* What statments might be optimized away
when their arguments are constant. */
if (gimple_code (stmt) != GIMPLE_ASSIGN
&& gimple_code (stmt) != GIMPLE_COND
&& gimple_code (stmt) != GIMPLE_SWITCH
&& (gimple_code (stmt) != GIMPLE_CALL
|| !(gimple_call_flags (stmt) & ECF_CONST)))
return p;
/* Stores will stay anyway. */
if (gimple_store_p (stmt))
return p;
is_load = gimple_assign_load_p (stmt);
/* Loads can be optimized when the value is known. */
if (is_load)
{
tree op;
gcc_assert (gimple_assign_single_p (stmt));
op = gimple_assign_rhs1 (stmt);
if (!unmodified_parm_or_parm_agg_item (fbi, stmt, op, &base_index,
&aggpos))
return p;
}
else
base_index = -1;
/* See if we understand all operands before we start
adding conditionals. */
FOR_EACH_SSA_TREE_OPERAND (use, stmt, iter, SSA_OP_USE)
{
tree parm = unmodified_parm (stmt, use);
/* For arguments we can build a condition. */
if (parm && ipa_get_param_decl_index (fbi->info, parm) >= 0)
continue;
if (TREE_CODE (use) != SSA_NAME)
return p;
/* If we know when operand is constant,
we still can say something useful. */
if (!true_predicate_p (&nonconstant_names[SSA_NAME_VERSION (use)]))
continue;
return p;
}
if (is_load)
op_non_const =
add_condition (summary, base_index, &aggpos, CHANGED, NULL);
else
op_non_const = false_predicate ();
FOR_EACH_SSA_TREE_OPERAND (use, stmt, iter, SSA_OP_USE)
{
tree parm = unmodified_parm (stmt, use);
int index;
if (parm && (index = ipa_get_param_decl_index (fbi->info, parm)) >= 0)
{
if (index != base_index)
p = add_condition (summary, index, NULL, CHANGED, NULL_TREE);
else
continue;
}
else
p = nonconstant_names[SSA_NAME_VERSION (use)];
op_non_const = or_predicates (summary->conds, &p, &op_non_const);
}
if ((gimple_code (stmt) == GIMPLE_ASSIGN || gimple_code (stmt) == GIMPLE_CALL)
&& gimple_op (stmt, 0)
&& TREE_CODE (gimple_op (stmt, 0)) == SSA_NAME)
nonconstant_names[SSA_NAME_VERSION (gimple_op (stmt, 0))]
= op_non_const;
return op_non_const;
}
struct record_modified_bb_info
{
bitmap bb_set;
gimple stmt;
};
/* Callback of walk_aliased_vdefs. Records basic blocks where the value may be
set except for info->stmt. */
static bool
record_modified (ao_ref *ao ATTRIBUTE_UNUSED, tree vdef, void *data)
{
struct record_modified_bb_info *info =
(struct record_modified_bb_info *) data;
if (SSA_NAME_DEF_STMT (vdef) == info->stmt)
return false;
bitmap_set_bit (info->bb_set,
SSA_NAME_IS_DEFAULT_DEF (vdef)
? ENTRY_BLOCK_PTR_FOR_FN (cfun)->index
: gimple_bb (SSA_NAME_DEF_STMT (vdef))->index);
return false;
}
/* Return probability (based on REG_BR_PROB_BASE) that I-th parameter of STMT
will change since last invocation of STMT.
Value 0 is reserved for compile time invariants.
For common parameters it is REG_BR_PROB_BASE. For loop invariants it
ought to be REG_BR_PROB_BASE / estimated_iters. */
static int
param_change_prob (gimple stmt, int i)
{
tree op = gimple_call_arg (stmt, i);
basic_block bb = gimple_bb (stmt);
tree base;
/* Global invariants neve change. */
if (is_gimple_min_invariant (op))
return 0;
/* We would have to do non-trivial analysis to really work out what
is the probability of value to change (i.e. when init statement
is in a sibling loop of the call).
We do an conservative estimate: when call is executed N times more often
than the statement defining value, we take the frequency 1/N. */
if (TREE_CODE (op) == SSA_NAME)
{
int init_freq;
if (!bb->frequency)
return REG_BR_PROB_BASE;
if (SSA_NAME_IS_DEFAULT_DEF (op))
init_freq = ENTRY_BLOCK_PTR_FOR_FN (cfun)->frequency;
else
init_freq = gimple_bb (SSA_NAME_DEF_STMT (op))->frequency;
if (!init_freq)
init_freq = 1;
if (init_freq < bb->frequency)
return MAX (GCOV_COMPUTE_SCALE (init_freq, bb->frequency), 1);
else
return REG_BR_PROB_BASE;
}
base = get_base_address (op);
if (base)
{
ao_ref refd;
int max;
struct record_modified_bb_info info;
bitmap_iterator bi;
unsigned index;
tree init = ctor_for_folding (base);
if (init != error_mark_node)
return 0;
if (!bb->frequency)
return REG_BR_PROB_BASE;
ao_ref_init (&refd, op);
info.stmt = stmt;
info.bb_set = BITMAP_ALLOC (NULL);
walk_aliased_vdefs (&refd, gimple_vuse (stmt), record_modified, &info,
NULL);
if (bitmap_bit_p (info.bb_set, bb->index))
{
BITMAP_FREE (info.bb_set);
return REG_BR_PROB_BASE;
}
/* Assume that every memory is initialized at entry.
TODO: Can we easilly determine if value is always defined
and thus we may skip entry block? */
if (ENTRY_BLOCK_PTR_FOR_FN (cfun)->frequency)
max = ENTRY_BLOCK_PTR_FOR_FN (cfun)->frequency;
else
max = 1;
EXECUTE_IF_SET_IN_BITMAP (info.bb_set, 0, index, bi)
max = MIN (max, BASIC_BLOCK_FOR_FN (cfun, index)->frequency);
BITMAP_FREE (info.bb_set);
if (max < bb->frequency)
return MAX (GCOV_COMPUTE_SCALE (max, bb->frequency), 1);
else
return REG_BR_PROB_BASE;
}
return REG_BR_PROB_BASE;
}
/* Find whether a basic block BB is the final block of a (half) diamond CFG
sub-graph and if the predicate the condition depends on is known. If so,
return true and store the pointer the predicate in *P. */
static bool
phi_result_unknown_predicate (struct ipa_node_params *info,
inline_summary *summary, basic_block bb,
struct predicate *p,
vec<predicate_t> nonconstant_names)
{
edge e;
edge_iterator ei;
basic_block first_bb = NULL;
gimple stmt;
if (single_pred_p (bb))
{
*p = false_predicate ();
return true;
}
FOR_EACH_EDGE (e, ei, bb->preds)
{
if (single_succ_p (e->src))
{
if (!single_pred_p (e->src))
return false;
if (!first_bb)
first_bb = single_pred (e->src);
else if (single_pred (e->src) != first_bb)
return false;
}
else
{
if (!first_bb)
first_bb = e->src;
else if (e->src != first_bb)
return false;
}
}
if (!first_bb)
return false;
stmt = last_stmt (first_bb);
if (!stmt
|| gimple_code (stmt) != GIMPLE_COND
|| !is_gimple_ip_invariant (gimple_cond_rhs (stmt)))
return false;
*p = will_be_nonconstant_expr_predicate (info, summary,
gimple_cond_lhs (stmt),
nonconstant_names);
if (true_predicate_p (p))
return false;
else
return true;
}
/* Given a PHI statement in a function described by inline properties SUMMARY
and *P being the predicate describing whether the selected PHI argument is
known, store a predicate for the result of the PHI statement into
NONCONSTANT_NAMES, if possible. */
static void
predicate_for_phi_result (struct inline_summary *summary, gphi *phi,
struct predicate *p,
vec<predicate_t> nonconstant_names)
{
unsigned i;
for (i = 0; i < gimple_phi_num_args (phi); i++)
{
tree arg = gimple_phi_arg (phi, i)->def;
if (!is_gimple_min_invariant (arg))
{
gcc_assert (TREE_CODE (arg) == SSA_NAME);
*p = or_predicates (summary->conds, p,
&nonconstant_names[SSA_NAME_VERSION (arg)]);
if (true_predicate_p (p))
return;
}
}
if (dump_file && (dump_flags & TDF_DETAILS))
{
fprintf (dump_file, "\t\tphi predicate: ");
dump_predicate (dump_file, summary->conds, p);
}
nonconstant_names[SSA_NAME_VERSION (gimple_phi_result (phi))] = *p;
}
/* Return predicate specifying when array index in access OP becomes non-constant. */
static struct predicate
array_index_predicate (inline_summary *info,
vec< predicate_t> nonconstant_names, tree op)
{
struct predicate p = false_predicate ();
while (handled_component_p (op))
{
if (TREE_CODE (op) == ARRAY_REF || TREE_CODE (op) == ARRAY_RANGE_REF)
{
if (TREE_CODE (TREE_OPERAND (op, 1)) == SSA_NAME)
p = or_predicates (info->conds, &p,
&nonconstant_names[SSA_NAME_VERSION
(TREE_OPERAND (op, 1))]);
}
op = TREE_OPERAND (op, 0);
}
return p;
}
/* For a typical usage of __builtin_expect (a<b, 1), we
may introduce an extra relation stmt:
With the builtin, we have
t1 = a <= b;
t2 = (long int) t1;
t3 = __builtin_expect (t2, 1);
if (t3 != 0)
goto ...
Without the builtin, we have
if (a<=b)
goto...
This affects the size/time estimation and may have
an impact on the earlier inlining.
Here find this pattern and fix it up later. */
static gimple
find_foldable_builtin_expect (basic_block bb)
{
gimple_stmt_iterator bsi;
for (bsi = gsi_start_bb (bb); !gsi_end_p (bsi); gsi_next (&bsi))
{
gimple stmt = gsi_stmt (bsi);
if (gimple_call_builtin_p (stmt, BUILT_IN_EXPECT)
|| (is_gimple_call (stmt)
&& gimple_call_internal_p (stmt)
&& gimple_call_internal_fn (stmt) == IFN_BUILTIN_EXPECT))
{
tree var = gimple_call_lhs (stmt);
tree arg = gimple_call_arg (stmt, 0);
use_operand_p use_p;
gimple use_stmt;
bool match = false;
bool done = false;
if (!var || !arg)
continue;
gcc_assert (TREE_CODE (var) == SSA_NAME);
while (TREE_CODE (arg) == SSA_NAME)
{
gimple stmt_tmp = SSA_NAME_DEF_STMT (arg);
if (!is_gimple_assign (stmt_tmp))
break;
switch (gimple_assign_rhs_code (stmt_tmp))
{
case LT_EXPR:
case LE_EXPR:
case GT_EXPR:
case GE_EXPR:
case EQ_EXPR:
case NE_EXPR:
match = true;
done = true;
break;
CASE_CONVERT:
break;
default:
done = true;
break;
}
if (done)
break;
arg = gimple_assign_rhs1 (stmt_tmp);
}
if (match && single_imm_use (var, &use_p, &use_stmt)
&& gimple_code (use_stmt) == GIMPLE_COND)
return use_stmt;
}
}
return NULL;
}
/* Return true when the basic blocks contains only clobbers followed by RESX.
Such BBs are kept around to make removal of dead stores possible with
presence of EH and will be optimized out by optimize_clobbers later in the
game.
NEED_EH is used to recurse in case the clobber has non-EH predecestors
that can be clobber only, too.. When it is false, the RESX is not necessary
on the end of basic block. */
static bool
clobber_only_eh_bb_p (basic_block bb, bool need_eh = true)
{
gimple_stmt_iterator gsi = gsi_last_bb (bb);
edge_iterator ei;
edge e;
if (need_eh)
{
if (gsi_end_p (gsi))
return false;
if (gimple_code (gsi_stmt (gsi)) != GIMPLE_RESX)
return false;
gsi_prev (&gsi);
}
else if (!single_succ_p (bb))
return false;
for (; !gsi_end_p (gsi); gsi_prev (&gsi))
{
gimple stmt = gsi_stmt (gsi);
if (is_gimple_debug (stmt))
continue;
if (gimple_clobber_p (stmt))
continue;
if (gimple_code (stmt) == GIMPLE_LABEL)
break;
return false;
}
/* See if all predecestors are either throws or clobber only BBs. */
FOR_EACH_EDGE (e, ei, bb->preds)
if (!(e->flags & EDGE_EH)
&& !clobber_only_eh_bb_p (e->src, false))
return false;
return true;
}
/* Compute function body size parameters for NODE.
When EARLY is true, we compute only simple summaries without
non-trivial predicates to drive the early inliner. */
static void
estimate_function_body_sizes (struct cgraph_node *node, bool early)
{
gcov_type time = 0;
/* Estimate static overhead for function prologue/epilogue and alignment. */
int size = 2;
/* Benefits are scaled by probability of elimination that is in range
<0,2>. */
basic_block bb;
struct function *my_function = DECL_STRUCT_FUNCTION (node->decl);
int freq;
struct inline_summary *info = inline_summaries->get (node);
struct predicate bb_predicate;
struct ipa_func_body_info fbi;
vec<predicate_t> nonconstant_names = vNULL;
int nblocks, n;
int *order;
predicate array_index = true_predicate ();
gimple fix_builtin_expect_stmt;
gcc_assert (my_function && my_function->cfg);
gcc_assert (cfun == my_function);
memset(&fbi, 0, sizeof(fbi));
info->conds = NULL;
info->entry = NULL;
/* When optimizing and analyzing for IPA inliner, initialize loop optimizer
so we can produce proper inline hints.
When optimizing and analyzing for early inliner, initialize node params
so we can produce correct BB predicates. */
if (opt_for_fn (node->decl, optimize))
{
calculate_dominance_info (CDI_DOMINATORS);
if (!early)
loop_optimizer_init (LOOPS_NORMAL | LOOPS_HAVE_RECORDED_EXITS);
else
{
ipa_check_create_node_params ();
ipa_initialize_node_params (node);
}
if (ipa_node_params_sum)
{
fbi.node = node;
fbi.info = IPA_NODE_REF (node);
fbi.bb_infos = vNULL;
fbi.bb_infos.safe_grow_cleared (last_basic_block_for_fn (cfun));
fbi.param_count = count_formal_params(node->decl);
nonconstant_names.safe_grow_cleared
(SSANAMES (my_function)->length ());
}
}
if (dump_file)
fprintf (dump_file, "\nAnalyzing function body size: %s\n",
node->name ());
/* When we run into maximal number of entries, we assign everything to the
constant truth case. Be sure to have it in list. */
bb_predicate = true_predicate ();
account_size_time (info, 0, 0, &bb_predicate);
bb_predicate = not_inlined_predicate ();
account_size_time (info, 2 * INLINE_SIZE_SCALE, 0, &bb_predicate);
if (fbi.info)
compute_bb_predicates (&fbi, node, info);
order = XNEWVEC (int, n_basic_blocks_for_fn (cfun));
nblocks = pre_and_rev_post_order_compute (NULL, order, false);
for (n = 0; n < nblocks; n++)
{
bb = BASIC_BLOCK_FOR_FN (cfun, order[n]);
freq = compute_call_stmt_bb_frequency (node->decl, bb);
if (clobber_only_eh_bb_p (bb))
{
if (dump_file && (dump_flags & TDF_DETAILS))
fprintf (dump_file, "\n Ignoring BB %i;"
" it will be optimized away by cleanup_clobbers\n",
bb->index);
continue;
}
/* TODO: Obviously predicates can be propagated down across CFG. */
if (fbi.info)
{
if (bb->aux)
bb_predicate = *(struct predicate *) bb->aux;
else
bb_predicate = false_predicate ();
}
else
bb_predicate = true_predicate ();
if (dump_file && (dump_flags & TDF_DETAILS))
{
fprintf (dump_file, "\n BB %i predicate:", bb->index);
dump_predicate (dump_file, info->conds, &bb_predicate);
}
if (fbi.info && nonconstant_names.exists ())
{
struct predicate phi_predicate;
bool first_phi = true;
for (gphi_iterator bsi = gsi_start_phis (bb); !gsi_end_p (bsi);
gsi_next (&bsi))
{
if (first_phi
&& !phi_result_unknown_predicate (fbi.info, info, bb,
&phi_predicate,
nonconstant_names))
break;
first_phi = false;
if (dump_file && (dump_flags & TDF_DETAILS))
{
fprintf (dump_file, " ");
print_gimple_stmt (dump_file, gsi_stmt (bsi), 0, 0);
}
predicate_for_phi_result (info, bsi.phi (), &phi_predicate,
nonconstant_names);
}
}
fix_builtin_expect_stmt = find_foldable_builtin_expect (bb);
for (gimple_stmt_iterator bsi = gsi_start_bb (bb); !gsi_end_p (bsi);
gsi_next (&bsi))
{
gimple stmt = gsi_stmt (bsi);
int this_size = estimate_num_insns (stmt, &eni_size_weights);
int this_time = estimate_num_insns (stmt, &eni_time_weights);
int prob;
struct predicate will_be_nonconstant;
/* This relation stmt should be folded after we remove
buildin_expect call. Adjust the cost here. */
if (stmt == fix_builtin_expect_stmt)
{
this_size--;
this_time--;
}
if (dump_file && (dump_flags & TDF_DETAILS))
{
fprintf (dump_file, " ");
print_gimple_stmt (dump_file, stmt, 0, 0);
fprintf (dump_file, "\t\tfreq:%3.2f size:%3i time:%3i\n",
((double) freq) / CGRAPH_FREQ_BASE, this_size,
this_time);
}
if (gimple_assign_load_p (stmt) && nonconstant_names.exists ())
{
struct predicate this_array_index;
this_array_index =
array_index_predicate (info, nonconstant_names,
gimple_assign_rhs1 (stmt));
if (!false_predicate_p (&this_array_index))
array_index =
and_predicates (info->conds, &array_index,
&this_array_index);
}
if (gimple_store_p (stmt) && nonconstant_names.exists ())
{
struct predicate this_array_index;
this_array_index =
array_index_predicate (info, nonconstant_names,
gimple_get_lhs (stmt));
if (!false_predicate_p (&this_array_index))
array_index =
and_predicates (info->conds, &array_index,
&this_array_index);
}
if (is_gimple_call (stmt)
&& !gimple_call_internal_p (stmt))
{
struct cgraph_edge *edge = node->get_edge (stmt);
struct inline_edge_summary *es = inline_edge_summary (edge);
/* Special case: results of BUILT_IN_CONSTANT_P will be always
resolved as constant. We however don't want to optimize
out the cgraph edges. */
if (nonconstant_names.exists ()
&& gimple_call_builtin_p (stmt, BUILT_IN_CONSTANT_P)
&& gimple_call_lhs (stmt)
&& TREE_CODE (gimple_call_lhs (stmt)) == SSA_NAME)
{
struct predicate false_p = false_predicate ();
nonconstant_names[SSA_NAME_VERSION (gimple_call_lhs (stmt))]
= false_p;
}
if (ipa_node_params_sum)
{
int count = gimple_call_num_args (stmt);
int i;
if (count)
es->param.safe_grow_cleared (count);
for (i = 0; i < count; i++)
{
int prob = param_change_prob (stmt, i);
gcc_assert (prob >= 0 && prob <= REG_BR_PROB_BASE);
es->param[i].change_prob = prob;
}
}
es->call_stmt_size = this_size;
es->call_stmt_time = this_time;
es->loop_depth = bb_loop_depth (bb);
edge_set_predicate (edge, &bb_predicate);
}
/* TODO: When conditional jump or swithc is known to be constant, but
we did not translate it into the predicates, we really can account
just maximum of the possible paths. */
if (fbi.info)
will_be_nonconstant
= will_be_nonconstant_predicate (&fbi, info,
stmt, nonconstant_names);
if (this_time || this_size)
{
struct predicate p;
this_time *= freq;
prob = eliminated_by_inlining_prob (stmt);
if (prob == 1 && dump_file && (dump_flags & TDF_DETAILS))
fprintf (dump_file,
"\t\t50%% will be eliminated by inlining\n");
if (prob == 2 && dump_file && (dump_flags & TDF_DETAILS))
fprintf (dump_file, "\t\tWill be eliminated by inlining\n");
if (fbi.info)
p = and_predicates (info->conds, &bb_predicate,
&will_be_nonconstant);
else
p = true_predicate ();
if (!false_predicate_p (&p)
|| (is_gimple_call (stmt)
&& !false_predicate_p (&bb_predicate)))
{
time += this_time;
size += this_size;
if (time > MAX_TIME * INLINE_TIME_SCALE)
time = MAX_TIME * INLINE_TIME_SCALE;
}
/* We account everything but the calls. Calls have their own
size/time info attached to cgraph edges. This is necessary
in order to make the cost disappear after inlining. */
if (!is_gimple_call (stmt))
{
if (prob)
{
struct predicate ip = not_inlined_predicate ();
ip = and_predicates (info->conds, &ip, &p);
account_size_time (info, this_size * prob,
this_time * prob, &ip);
}
if (prob != 2)
account_size_time (info, this_size * (2 - prob),
this_time * (2 - prob), &p);
}
gcc_assert (time >= 0);
gcc_assert (size >= 0);
}
}
}
set_hint_predicate (&inline_summaries->get (node)->array_index, array_index);
time = (time + CGRAPH_FREQ_BASE / 2) / CGRAPH_FREQ_BASE;
if (time > MAX_TIME)
time = MAX_TIME;
free (order);
if (nonconstant_names.exists () && !early)
{
struct loop *loop;
predicate loop_iterations = true_predicate ();
predicate loop_stride = true_predicate ();
if (dump_file && (dump_flags & TDF_DETAILS))
flow_loops_dump (dump_file, NULL, 0);
scev_initialize ();
FOR_EACH_LOOP (loop, 0)
{
vec<edge> exits;
edge ex;
unsigned int j;
struct tree_niter_desc niter_desc;
bb_predicate = *(struct predicate *) loop->header->aux;
exits = get_loop_exit_edges (loop);
FOR_EACH_VEC_ELT (exits, j, ex)
if (number_of_iterations_exit (loop, ex, &niter_desc, false)
&& !is_gimple_min_invariant (niter_desc.niter))
{
predicate will_be_nonconstant
= will_be_nonconstant_expr_predicate (fbi.info, info,
niter_desc.niter,
nonconstant_names);
if (!true_predicate_p (&will_be_nonconstant))
will_be_nonconstant = and_predicates (info->conds,
&bb_predicate,
&will_be_nonconstant);
if (!true_predicate_p (&will_be_nonconstant)
&& !false_predicate_p (&will_be_nonconstant))
/* This is slightly inprecise. We may want to represent each
loop with independent predicate. */
loop_iterations =
and_predicates (info->conds, &loop_iterations,
&will_be_nonconstant);
}
exits.release ();
for (gphi_iterator gsi = gsi_start_phis (loop->header);
!gsi_end_p (gsi); gsi_next (&gsi))
{
gphi *phi = gsi.phi ();
tree use = gimple_phi_result (phi);
affine_iv iv;
predicate will_be_nonconstant;
if (virtual_operand_p (use)
|| !simple_iv (loop, loop, use, &iv, true)
|| is_gimple_min_invariant (iv.step))
continue;
will_be_nonconstant
= will_be_nonconstant_expr_predicate (fbi.info, info,
iv.step,
nonconstant_names);
if (!true_predicate_p (&will_be_nonconstant))
will_be_nonconstant = and_predicates (info->conds,
&bb_predicate,
&will_be_nonconstant);
if (!true_predicate_p (&will_be_nonconstant)
&& !false_predicate_p (&will_be_nonconstant))
/* This is slightly inprecise. We may want to represent
each loop with independent predicate. */
loop_stride = and_predicates (info->conds, &loop_stride,
&will_be_nonconstant);
}
}
set_hint_predicate (&inline_summaries->get (node)->loop_iterations,
loop_iterations);
set_hint_predicate (&inline_summaries->get (node)->loop_stride, loop_stride);
scev_finalize ();
}
FOR_ALL_BB_FN (bb, my_function)
{
edge e;
edge_iterator ei;
if (bb->aux)
pool_free (edge_predicate_pool, bb->aux);
bb->aux = NULL;
FOR_EACH_EDGE (e, ei, bb->succs)
{
if (e->aux)
pool_free (edge_predicate_pool, e->aux);
e->aux = NULL;
}
}
inline_summaries->get (node)->self_time = time;
inline_summaries->get (node)->self_size = size;
nonconstant_names.release ();
if (opt_for_fn (node->decl, optimize))
{
if (!early)
loop_optimizer_finalize ();
else if (!ipa_edge_args_vector)
ipa_free_all_node_params ();
free_dominance_info (CDI_DOMINATORS);
}
if (dump_file)
{
fprintf (dump_file, "\n");
dump_inline_summary (dump_file, node);
}
}
/* Compute parameters of functions used by inliner.
EARLY is true when we compute parameters for the early inliner */
void
compute_inline_parameters (struct cgraph_node *node, bool early)
{
HOST_WIDE_INT self_stack_size;
struct cgraph_edge *e;
struct inline_summary *info;
gcc_assert (!node->global.inlined_to);
inline_summary_alloc ();
info = inline_summaries->get (node);
reset_inline_summary (node, info);
/* FIXME: Thunks are inlinable, but tree-inline don't know how to do that.
Once this happen, we will need to more curefully predict call
statement size. */
if (node->thunk.thunk_p)
{
struct inline_edge_summary *es = inline_edge_summary (node->callees);
struct predicate t = true_predicate ();
info->inlinable = 0;
node->callees->call_stmt_cannot_inline_p = true;
node->local.can_change_signature = false;
es->call_stmt_time = 1;
es->call_stmt_size = 1;
account_size_time (info, 0, 0, &t);
return;
}
/* Even is_gimple_min_invariant rely on current_function_decl. */
push_cfun (DECL_STRUCT_FUNCTION (node->decl));
/* Estimate the stack size for the function if we're optimizing. */
self_stack_size = optimize ? estimated_stack_frame_size (node) : 0;
info->estimated_self_stack_size = self_stack_size;
info->estimated_stack_size = self_stack_size;
info->stack_frame_offset = 0;
/* Can this function be inlined at all? */
if (!opt_for_fn (node->decl, optimize)
&& !lookup_attribute ("always_inline",
DECL_ATTRIBUTES (node->decl)))
info->inlinable = false;
else
info->inlinable = tree_inlinable_function_p (node->decl);
info->contains_cilk_spawn = fn_contains_cilk_spawn_p (cfun);
/* Type attributes can use parameter indices to describe them. */
if (TYPE_ATTRIBUTES (TREE_TYPE (node->decl)))
node->local.can_change_signature = false;
else
{
/* Otherwise, inlinable functions always can change signature. */
if (info->inlinable)
node->local.can_change_signature = true;
else
{
/* Functions calling builtin_apply can not change signature. */
for (e = node->callees; e; e = e->next_callee)
{
tree cdecl = e->callee->decl;
if (DECL_BUILT_IN (cdecl)
&& DECL_BUILT_IN_CLASS (cdecl) == BUILT_IN_NORMAL
&& (DECL_FUNCTION_CODE (cdecl) == BUILT_IN_APPLY_ARGS
|| DECL_FUNCTION_CODE (cdecl) == BUILT_IN_VA_START))
break;
}
node->local.can_change_signature = !e;
}
}
estimate_function_body_sizes (node, early);
for (e = node->callees; e; e = e->next_callee)
if (e->callee->comdat_local_p ())
break;
node->calls_comdat_local = (e != NULL);
/* Inlining characteristics are maintained by the cgraph_mark_inline. */
info->time = info->self_time;
info->size = info->self_size;
info->stack_frame_offset = 0;
info->estimated_stack_size = info->estimated_self_stack_size;
#ifdef ENABLE_CHECKING
inline_update_overall_summary (node);
gcc_assert (info->time == info->self_time && info->size == info->self_size);
#endif
pop_cfun ();
}
/* Compute parameters of functions used by inliner using
current_function_decl. */
static unsigned int
compute_inline_parameters_for_current (void)
{
compute_inline_parameters (cgraph_node::get (current_function_decl), true);
return 0;
}
namespace {
const pass_data pass_data_inline_parameters =
{
GIMPLE_PASS, /* type */
"inline_param", /* name */
OPTGROUP_INLINE, /* optinfo_flags */
TV_INLINE_PARAMETERS, /* tv_id */
0, /* properties_required */
0, /* properties_provided */
0, /* properties_destroyed */
0, /* todo_flags_start */
0, /* todo_flags_finish */
};
class pass_inline_parameters : public gimple_opt_pass
{
public:
pass_inline_parameters (gcc::context *ctxt)
: gimple_opt_pass (pass_data_inline_parameters, ctxt)
{}
/* opt_pass methods: */
opt_pass * clone () { return new pass_inline_parameters (m_ctxt); }
virtual unsigned int execute (function *)
{
return compute_inline_parameters_for_current ();
}
}; // class pass_inline_parameters
} // anon namespace
gimple_opt_pass *
make_pass_inline_parameters (gcc::context *ctxt)
{
return new pass_inline_parameters (ctxt);
}
/* Estimate benefit devirtualizing indirect edge IE, provided KNOWN_VALS,
KNOWN_CONTEXTS and KNOWN_AGGS. */
static bool
estimate_edge_devirt_benefit (struct cgraph_edge *ie,
int *size, int *time,
vec<tree> known_vals,
vec<ipa_polymorphic_call_context> known_contexts,
vec<ipa_agg_jump_function_p> known_aggs)
{
tree target;
struct cgraph_node *callee;
struct inline_summary *isummary;
enum availability avail;
bool speculative;
if (!known_vals.exists () && !known_contexts.exists ())
return false;
if (!opt_for_fn (ie->caller->decl, flag_indirect_inlining))
return false;
target = ipa_get_indirect_edge_target (ie, known_vals, known_contexts,
known_aggs, &speculative);
if (!target || speculative)
return false;
/* Account for difference in cost between indirect and direct calls. */
*size -= (eni_size_weights.indirect_call_cost - eni_size_weights.call_cost);
*time -= (eni_time_weights.indirect_call_cost - eni_time_weights.call_cost);
gcc_checking_assert (*time >= 0);
gcc_checking_assert (*size >= 0);
callee = cgraph_node::get (target);
if (!callee || !callee->definition)
return false;
callee = callee->function_symbol (&avail);
if (avail < AVAIL_AVAILABLE)
return false;
isummary = inline_summaries->get (callee);
return isummary->inlinable;
}
/* Increase SIZE, MIN_SIZE (if non-NULL) and TIME for size and time needed to
handle edge E with probability PROB.
Set HINTS if edge may be devirtualized.
KNOWN_VALS, KNOWN_AGGS and KNOWN_CONTEXTS describe context of the call
site. */
static inline void
estimate_edge_size_and_time (struct cgraph_edge *e, int *size, int *min_size,
int *time,
int prob,
vec<tree> known_vals,
vec<ipa_polymorphic_call_context> known_contexts,
vec<ipa_agg_jump_function_p> known_aggs,
inline_hints *hints)
{
struct inline_edge_summary *es = inline_edge_summary (e);
int call_size = es->call_stmt_size;
int call_time = es->call_stmt_time;
int cur_size;
if (!e->callee
&& estimate_edge_devirt_benefit (e, &call_size, &call_time,
known_vals, known_contexts, known_aggs)
&& hints && e->maybe_hot_p ())
*hints |= INLINE_HINT_indirect_call;
cur_size = call_size * INLINE_SIZE_SCALE;
*size += cur_size;
if (min_size)
*min_size += cur_size;
*time += apply_probability ((gcov_type) call_time, prob)
* e->frequency * (INLINE_TIME_SCALE / CGRAPH_FREQ_BASE);
if (*time > MAX_TIME * INLINE_TIME_SCALE)
*time = MAX_TIME * INLINE_TIME_SCALE;
}
/* Increase SIZE, MIN_SIZE and TIME for size and time needed to handle all
calls in NODE. POSSIBLE_TRUTHS, KNOWN_VALS, KNOWN_AGGS and KNOWN_CONTEXTS
describe context of the call site. */
static void
estimate_calls_size_and_time (struct cgraph_node *node, int *size,
int *min_size, int *time,
inline_hints *hints,
clause_t possible_truths,
vec<tree> known_vals,
vec<ipa_polymorphic_call_context> known_contexts,
vec<ipa_agg_jump_function_p> known_aggs)
{
struct cgraph_edge *e;
for (e = node->callees; e; e = e->next_callee)
{
if (inline_edge_summary_vec.length () <= (unsigned) e->uid)
continue;
struct inline_edge_summary *es = inline_edge_summary (e);
/* Do not care about zero sized builtins. */
if (e->inline_failed && !es->call_stmt_size)
{
gcc_checking_assert (!es->call_stmt_time);
continue;
}
if (!es->predicate
|| evaluate_predicate (es->predicate, possible_truths))
{
if (e->inline_failed)
{
/* Predicates of calls shall not use NOT_CHANGED codes,
sowe do not need to compute probabilities. */
estimate_edge_size_and_time (e, size,
es->predicate ? NULL : min_size,
time, REG_BR_PROB_BASE,
known_vals, known_contexts,
known_aggs, hints);
}
else
estimate_calls_size_and_time (e->callee, size, min_size, time,
hints,
possible_truths,
known_vals, known_contexts,
known_aggs);
}
}
for (e = node->indirect_calls; e; e = e->next_callee)
{
if (inline_edge_summary_vec.length () <= (unsigned) e->uid)
continue;
struct inline_edge_summary *es = inline_edge_summary (e);
if (!es->predicate
|| evaluate_predicate (es->predicate, possible_truths))
estimate_edge_size_and_time (e, size,
es->predicate ? NULL : min_size,
time, REG_BR_PROB_BASE,
known_vals, known_contexts, known_aggs,
hints);
}
}
/* Estimate size and time needed to execute NODE assuming
POSSIBLE_TRUTHS clause, and KNOWN_VALS, KNOWN_AGGS and KNOWN_CONTEXTS
information about NODE's arguments. If non-NULL use also probability
information present in INLINE_PARAM_SUMMARY vector.
Additionally detemine hints determined by the context. Finally compute
minimal size needed for the call that is independent on the call context and
can be used for fast estimates. Return the values in RET_SIZE,
RET_MIN_SIZE, RET_TIME and RET_HINTS. */
static void
estimate_node_size_and_time (struct cgraph_node *node,
clause_t possible_truths,
vec<tree> known_vals,
vec<ipa_polymorphic_call_context> known_contexts,
vec<ipa_agg_jump_function_p> known_aggs,
int *ret_size, int *ret_min_size, int *ret_time,
inline_hints *ret_hints,
vec<inline_param_summary>
inline_param_summary)
{
struct inline_summary *info = inline_summaries->get (node);
size_time_entry *e;
int size = 0;
int time = 0;
int min_size = 0;
inline_hints hints = 0;
int i;
if (dump_file && (dump_flags & TDF_DETAILS))
{
bool found = false;
fprintf (dump_file, " Estimating body: %s/%i\n"
" Known to be false: ", node->name (),
node->order);
for (i = predicate_not_inlined_condition;
i < (predicate_first_dynamic_condition
+ (int) vec_safe_length (info->conds)); i++)
if (!(possible_truths & (1 << i)))
{
if (found)
fprintf (dump_file, ", ");
found = true;
dump_condition (dump_file, info->conds, i);
}
}
for (i = 0; vec_safe_iterate (info->entry, i, &e); i++)
if (evaluate_predicate (&e->predicate, possible_truths))
{
size += e->size;
gcc_checking_assert (e->time >= 0);
gcc_checking_assert (time >= 0);
if (!inline_param_summary.exists ())
time += e->time;
else
{
int prob = predicate_probability (info->conds,
&e->predicate,
possible_truths,
inline_param_summary);
gcc_checking_assert (prob >= 0);
gcc_checking_assert (prob <= REG_BR_PROB_BASE);
time += apply_probability ((gcov_type) e->time, prob);
}
if (time > MAX_TIME * INLINE_TIME_SCALE)
time = MAX_TIME * INLINE_TIME_SCALE;
gcc_checking_assert (time >= 0);
}
gcc_checking_assert (true_predicate_p (&(*info->entry)[0].predicate));
min_size = (*info->entry)[0].size;
gcc_checking_assert (size >= 0);
gcc_checking_assert (time >= 0);
if (info->loop_iterations
&& !evaluate_predicate (info->loop_iterations, possible_truths))
hints |= INLINE_HINT_loop_iterations;
if (info->loop_stride
&& !evaluate_predicate (info->loop_stride, possible_truths))
hints |= INLINE_HINT_loop_stride;
if (info->array_index
&& !evaluate_predicate (info->array_index, possible_truths))
hints |= INLINE_HINT_array_index;
if (info->scc_no)
hints |= INLINE_HINT_in_scc;
if (DECL_DECLARED_INLINE_P (node->decl))
hints |= INLINE_HINT_declared_inline;
estimate_calls_size_and_time (node, &size, &min_size, &time, &hints, possible_truths,
known_vals, known_contexts, known_aggs);
gcc_checking_assert (size >= 0);
gcc_checking_assert (time >= 0);
time = RDIV (time, INLINE_TIME_SCALE);
size = RDIV (size, INLINE_SIZE_SCALE);
min_size = RDIV (min_size, INLINE_SIZE_SCALE);
if (dump_file && (dump_flags & TDF_DETAILS))
fprintf (dump_file, "\n size:%i time:%i\n", (int) size, (int) time);
if (ret_time)
*ret_time = time;
if (ret_size)
*ret_size = size;
if (ret_min_size)
*ret_min_size = min_size;
if (ret_hints)
*ret_hints = hints;
return;
}
/* Estimate size and time needed to execute callee of EDGE assuming that
parameters known to be constant at caller of EDGE are propagated.
KNOWN_VALS and KNOWN_CONTEXTS are vectors of assumed known constant values
and types for parameters. */
void
estimate_ipcp_clone_size_and_time (struct cgraph_node *node,
vec<tree> known_vals,
vec<ipa_polymorphic_call_context>
known_contexts,
vec<ipa_agg_jump_function_p> known_aggs,
int *ret_size, int *ret_time,
inline_hints *hints)
{
clause_t clause;
clause = evaluate_conditions_for_known_args (node, false, known_vals,
known_aggs);
estimate_node_size_and_time (node, clause, known_vals, known_contexts,
known_aggs, ret_size, NULL, ret_time, hints, vNULL);
}
/* Translate all conditions from callee representation into caller
representation and symbolically evaluate predicate P into new predicate.
INFO is inline_summary of function we are adding predicate into, CALLEE_INFO
is summary of function predicate P is from. OPERAND_MAP is array giving
callee formal IDs the caller formal IDs. POSSSIBLE_TRUTHS is clausule of all
callee conditions that may be true in caller context. TOPLEV_PREDICATE is
predicate under which callee is executed. OFFSET_MAP is an array of of
offsets that need to be added to conditions, negative offset means that
conditions relying on values passed by reference have to be discarded
because they might not be preserved (and should be considered offset zero
for other purposes). */
static struct predicate
remap_predicate (struct inline_summary *info,
struct inline_summary *callee_info,
struct predicate *p,
vec<int> operand_map,
vec<int> offset_map,
clause_t possible_truths, struct predicate *toplev_predicate)
{
int i;
struct predicate out = true_predicate ();
/* True predicate is easy. */
if (true_predicate_p (p))
return *toplev_predicate;
for (i = 0; p->clause[i]; i++)
{
clause_t clause = p->clause[i];
int cond;
struct predicate clause_predicate = false_predicate ();
gcc_assert (i < MAX_CLAUSES);
for (cond = 0; cond < NUM_CONDITIONS; cond++)
/* Do we have condition we can't disprove? */
if (clause & possible_truths & (1 << cond))
{
struct predicate cond_predicate;
/* Work out if the condition can translate to predicate in the
inlined function. */
if (cond >= predicate_first_dynamic_condition)
{
struct condition *c;
c = &(*callee_info->conds)[cond
-
predicate_first_dynamic_condition];
/* See if we can remap condition operand to caller's operand.
Otherwise give up. */
if (!operand_map.exists ()
|| (int) operand_map.length () <= c->operand_num
|| operand_map[c->operand_num] == -1
/* TODO: For non-aggregate conditions, adding an offset is
basically an arithmetic jump function processing which
we should support in future. */
|| ((!c->agg_contents || !c->by_ref)
&& offset_map[c->operand_num] > 0)
|| (c->agg_contents && c->by_ref
&& offset_map[c->operand_num] < 0))
cond_predicate = true_predicate ();
else
{
struct agg_position_info ap;
HOST_WIDE_INT offset_delta = offset_map[c->operand_num];
if (offset_delta < 0)
{
gcc_checking_assert (!c->agg_contents || !c->by_ref);
offset_delta = 0;
}
gcc_assert (!c->agg_contents
|| c->by_ref || offset_delta == 0);
ap.offset = c->offset + offset_delta;
ap.agg_contents = c->agg_contents;
ap.by_ref = c->by_ref;
cond_predicate = add_condition (info,
operand_map[c->operand_num],
&ap, c->code, c->val);
}
}
/* Fixed conditions remains same, construct single
condition predicate. */
else
{
cond_predicate.clause[0] = 1 << cond;
cond_predicate.clause[1] = 0;
}
clause_predicate = or_predicates (info->conds, &clause_predicate,
&cond_predicate);
}
out = and_predicates (info->conds, &out, &clause_predicate);
}
return and_predicates (info->conds, &out, toplev_predicate);
}
/* Update summary information of inline clones after inlining.
Compute peak stack usage. */
static void
inline_update_callee_summaries (struct cgraph_node *node, int depth)
{
struct cgraph_edge *e;
struct inline_summary *callee_info = inline_summaries->get (node);
struct inline_summary *caller_info = inline_summaries->get (node->callers->caller);
HOST_WIDE_INT peak;
callee_info->stack_frame_offset
= caller_info->stack_frame_offset
+ caller_info->estimated_self_stack_size;
peak = callee_info->stack_frame_offset
+ callee_info->estimated_self_stack_size;
if (inline_summaries->get (node->global.inlined_to)->estimated_stack_size < peak)
inline_summaries->get (node->global.inlined_to)->estimated_stack_size = peak;
ipa_propagate_frequency (node);
for (e = node->callees; e; e = e->next_callee)
{
if (!e->inline_failed)
inline_update_callee_summaries (e->callee, depth);
inline_edge_summary (e)->loop_depth += depth;
}
for (e = node->indirect_calls; e; e = e->next_callee)
inline_edge_summary (e)->loop_depth += depth;
}
/* Update change_prob of EDGE after INLINED_EDGE has been inlined.
When functoin A is inlined in B and A calls C with parameter that
changes with probability PROB1 and C is known to be passthroug
of argument if B that change with probability PROB2, the probability
of change is now PROB1*PROB2. */
static void
remap_edge_change_prob (struct cgraph_edge *inlined_edge,
struct cgraph_edge *edge)
{
if (ipa_node_params_sum)
{
int i;
struct ipa_edge_args *args = IPA_EDGE_REF (edge);
struct inline_edge_summary *es = inline_edge_summary (edge);
struct inline_edge_summary *inlined_es
= inline_edge_summary (inlined_edge);
for (i = 0; i < ipa_get_cs_argument_count (args); i++)
{
struct ipa_jump_func *jfunc = ipa_get_ith_jump_func (args, i);
if (jfunc->type == IPA_JF_PASS_THROUGH
&& (ipa_get_jf_pass_through_formal_id (jfunc)
< (int) inlined_es->param.length ()))
{
int jf_formal_id = ipa_get_jf_pass_through_formal_id (jfunc);
int prob1 = es->param[i].change_prob;
int prob2 = inlined_es->param[jf_formal_id].change_prob;
int prob = combine_probabilities (prob1, prob2);
if (prob1 && prob2 && !prob)
prob = 1;
es->param[i].change_prob = prob;
}
}
}
}
/* Update edge summaries of NODE after INLINED_EDGE has been inlined.
Remap predicates of callees of NODE. Rest of arguments match
remap_predicate.
Also update change probabilities. */
static void
remap_edge_summaries (struct cgraph_edge *inlined_edge,
struct cgraph_node *node,
struct inline_summary *info,
struct inline_summary *callee_info,
vec<int> operand_map,
vec<int> offset_map,
clause_t possible_truths,
struct predicate *toplev_predicate)
{
struct cgraph_edge *e, *next;
for (e = node->callees; e; e = next)
{
struct inline_edge_summary *es = inline_edge_summary (e);
struct predicate p;
next = e->next_callee;
if (e->inline_failed)
{
remap_edge_change_prob (inlined_edge, e);
if (es->predicate)
{
p = remap_predicate (info, callee_info,
es->predicate, operand_map, offset_map,
possible_truths, toplev_predicate);
edge_set_predicate (e, &p);
}
else
edge_set_predicate (e, toplev_predicate);
}
else
remap_edge_summaries (inlined_edge, e->callee, info, callee_info,
operand_map, offset_map, possible_truths,
toplev_predicate);
}
for (e = node->indirect_calls; e; e = next)
{
struct inline_edge_summary *es = inline_edge_summary (e);
struct predicate p;
next = e->next_callee;
remap_edge_change_prob (inlined_edge, e);
if (es->predicate)
{
p = remap_predicate (info, callee_info,
es->predicate, operand_map, offset_map,
possible_truths, toplev_predicate);
edge_set_predicate (e, &p);
}
else
edge_set_predicate (e, toplev_predicate);
}
}
/* Same as remap_predicate, but set result into hint *HINT. */
static void
remap_hint_predicate (struct inline_summary *info,
struct inline_summary *callee_info,
struct predicate **hint,
vec<int> operand_map,
vec<int> offset_map,
clause_t possible_truths,
struct predicate *toplev_predicate)
{
predicate p;
if (!*hint)
return;
p = remap_predicate (info, callee_info,
*hint,
operand_map, offset_map,
possible_truths, toplev_predicate);
if (!false_predicate_p (&p) && !true_predicate_p (&p))
{
if (!*hint)
set_hint_predicate (hint, p);
else
**hint = and_predicates (info->conds, *hint, &p);
}
}
/* We inlined EDGE. Update summary of the function we inlined into. */
void
inline_merge_summary (struct cgraph_edge *edge)
{
struct inline_summary *callee_info = inline_summaries->get (edge->callee);
struct cgraph_node *to = (edge->caller->global.inlined_to
? edge->caller->global.inlined_to : edge->caller);
struct inline_summary *info = inline_summaries->get (to);
clause_t clause = 0; /* not_inline is known to be false. */
size_time_entry *e;
vec<int> operand_map = vNULL;
vec<int> offset_map = vNULL;
int i;
struct predicate toplev_predicate;
struct predicate true_p = true_predicate ();
struct inline_edge_summary *es = inline_edge_summary (edge);
if (es->predicate)
toplev_predicate = *es->predicate;
else
toplev_predicate = true_predicate ();
if (callee_info->conds)
evaluate_properties_for_edge (edge, true, &clause, NULL, NULL, NULL);
if (ipa_node_params_sum && callee_info->conds)
{
struct ipa_edge_args *args = IPA_EDGE_REF (edge);
int count = ipa_get_cs_argument_count (args);
int i;
if (count)
{
operand_map.safe_grow_cleared (count);
offset_map.safe_grow_cleared (count);
}
for (i = 0; i < count; i++)
{
struct ipa_jump_func *jfunc = ipa_get_ith_jump_func (args, i);
int map = -1;
/* TODO: handle non-NOPs when merging. */
if (jfunc->type == IPA_JF_PASS_THROUGH)
{
if (ipa_get_jf_pass_through_operation (jfunc) == NOP_EXPR)
map = ipa_get_jf_pass_through_formal_id (jfunc);
if (!ipa_get_jf_pass_through_agg_preserved (jfunc))
offset_map[i] = -1;
}
else if (jfunc->type == IPA_JF_ANCESTOR)
{
HOST_WIDE_INT offset = ipa_get_jf_ancestor_offset (jfunc);
if (offset >= 0 && offset < INT_MAX)
{
map = ipa_get_jf_ancestor_formal_id (jfunc);
if (!ipa_get_jf_ancestor_agg_preserved (jfunc))
offset = -1;
offset_map[i] = offset;
}
}
operand_map[i] = map;
gcc_assert (map < ipa_get_param_count (IPA_NODE_REF (to)));
}
}
for (i = 0; vec_safe_iterate (callee_info->entry, i, &e); i++)
{
struct predicate p = remap_predicate (info, callee_info,
&e->predicate, operand_map,
offset_map, clause,
&toplev_predicate);
if (!false_predicate_p (&p))
{
gcov_type add_time = ((gcov_type) e->time * edge->frequency
+ CGRAPH_FREQ_BASE / 2) / CGRAPH_FREQ_BASE;
int prob = predicate_probability (callee_info->conds,
&e->predicate,
clause, es->param);
add_time = apply_probability ((gcov_type) add_time, prob);
if (add_time > MAX_TIME * INLINE_TIME_SCALE)
add_time = MAX_TIME * INLINE_TIME_SCALE;
if (prob != REG_BR_PROB_BASE
&& dump_file && (dump_flags & TDF_DETAILS))
{
fprintf (dump_file, "\t\tScaling time by probability:%f\n",
(double) prob / REG_BR_PROB_BASE);
}
account_size_time (info, e->size, add_time, &p);
}
}
remap_edge_summaries (edge, edge->callee, info, callee_info, operand_map,
offset_map, clause, &toplev_predicate);
remap_hint_predicate (info, callee_info,
&callee_info->loop_iterations,
operand_map, offset_map, clause, &toplev_predicate);
remap_hint_predicate (info, callee_info,
&callee_info->loop_stride,
operand_map, offset_map, clause, &toplev_predicate);
remap_hint_predicate (info, callee_info,
&callee_info->array_index,
operand_map, offset_map, clause, &toplev_predicate);
inline_update_callee_summaries (edge->callee,
inline_edge_summary (edge)->loop_depth);
/* We do not maintain predicates of inlined edges, free it. */
edge_set_predicate (edge, &true_p);
/* Similarly remove param summaries. */
es->param.release ();
operand_map.release ();
offset_map.release ();
}
/* For performance reasons inline_merge_summary is not updating overall size
and time. Recompute it. */
void
inline_update_overall_summary (struct cgraph_node *node)
{
struct inline_summary *info = inline_summaries->get (node);
size_time_entry *e;
int i;
info->size = 0;
info->time = 0;
for (i = 0; vec_safe_iterate (info->entry, i, &e); i++)
{
info->size += e->size, info->time += e->time;
if (info->time > MAX_TIME * INLINE_TIME_SCALE)
info->time = MAX_TIME * INLINE_TIME_SCALE;
}
estimate_calls_size_and_time (node, &info->size, &info->min_size,
&info->time, NULL,
~(clause_t) (1 << predicate_false_condition),
vNULL, vNULL, vNULL);
info->time = (info->time + INLINE_TIME_SCALE / 2) / INLINE_TIME_SCALE;
info->size = (info->size + INLINE_SIZE_SCALE / 2) / INLINE_SIZE_SCALE;
}
/* Return hints derrived from EDGE. */
int
simple_edge_hints (struct cgraph_edge *edge)
{
int hints = 0;
struct cgraph_node *to = (edge->caller->global.inlined_to
? edge->caller->global.inlined_to : edge->caller);
struct cgraph_node *callee = edge->callee->ultimate_alias_target ();
if (inline_summaries->get (to)->scc_no
&& inline_summaries->get (to)->scc_no
== inline_summaries->get (callee)->scc_no
&& !edge->recursive_p ())
hints |= INLINE_HINT_same_scc;
if (callee->lto_file_data && edge->caller->lto_file_data
&& edge->caller->lto_file_data != callee->lto_file_data
&& !callee->merged)
hints |= INLINE_HINT_cross_module;
return hints;
}
/* Estimate the time cost for the caller when inlining EDGE.
Only to be called via estimate_edge_time, that handles the
caching mechanism.
When caching, also update the cache entry. Compute both time and
size, since we always need both metrics eventually. */
int
do_estimate_edge_time (struct cgraph_edge *edge)
{
int time;
int size;
inline_hints hints;
struct cgraph_node *callee;
clause_t clause;
vec<tree> known_vals;
vec<ipa_polymorphic_call_context> known_contexts;
vec<ipa_agg_jump_function_p> known_aggs;
struct inline_edge_summary *es = inline_edge_summary (edge);
int min_size;
callee = edge->callee->ultimate_alias_target ();
gcc_checking_assert (edge->inline_failed);
evaluate_properties_for_edge (edge, true,
&clause, &known_vals, &known_contexts,
&known_aggs);
estimate_node_size_and_time (callee, clause, known_vals, known_contexts,
known_aggs, &size, &min_size, &time, &hints, es->param);
/* When we have profile feedback, we can quite safely identify hot
edges and for those we disable size limits. Don't do that when
probability that caller will call the callee is low however, since it
may hurt optimization of the caller's hot path. */
if (edge->count && edge->maybe_hot_p ()
&& (edge->count * 2
> (edge->caller->global.inlined_to
? edge->caller->global.inlined_to->count : edge->caller->count)))
hints |= INLINE_HINT_known_hot;
known_vals.release ();
known_contexts.release ();
known_aggs.release ();
gcc_checking_assert (size >= 0);
gcc_checking_assert (time >= 0);
/* When caching, update the cache entry. */
if (edge_growth_cache.exists ())
{
inline_summaries->get (edge->callee)->min_size = min_size;
if ((int) edge_growth_cache.length () <= edge->uid)
edge_growth_cache.safe_grow_cleared (symtab->edges_max_uid);
edge_growth_cache[edge->uid].time = time + (time >= 0);
edge_growth_cache[edge->uid].size = size + (size >= 0);
hints |= simple_edge_hints (edge);
edge_growth_cache[edge->uid].hints = hints + 1;
}
return time;
}
/* Return estimated callee growth after inlining EDGE.
Only to be called via estimate_edge_size. */
int
do_estimate_edge_size (struct cgraph_edge *edge)
{
int size;
struct cgraph_node *callee;
clause_t clause;
vec<tree> known_vals;
vec<ipa_polymorphic_call_context> known_contexts;
vec<ipa_agg_jump_function_p> known_aggs;
/* When we do caching, use do_estimate_edge_time to populate the entry. */
if (edge_growth_cache.exists ())
{
do_estimate_edge_time (edge);
size = edge_growth_cache[edge->uid].size;
gcc_checking_assert (size);
return size - (size > 0);
}
callee = edge->callee->ultimate_alias_target ();
/* Early inliner runs without caching, go ahead and do the dirty work. */
gcc_checking_assert (edge->inline_failed);
evaluate_properties_for_edge (edge, true,
&clause, &known_vals, &known_contexts,
&known_aggs);
estimate_node_size_and_time (callee, clause, known_vals, known_contexts,
known_aggs, &size, NULL, NULL, NULL, vNULL);
known_vals.release ();
known_contexts.release ();
known_aggs.release ();
return size;
}
/* Estimate the growth of the caller when inlining EDGE.
Only to be called via estimate_edge_size. */
inline_hints
do_estimate_edge_hints (struct cgraph_edge *edge)
{
inline_hints hints;
struct cgraph_node *callee;
clause_t clause;
vec<tree> known_vals;
vec<ipa_polymorphic_call_context> known_contexts;
vec<ipa_agg_jump_function_p> known_aggs;
/* When we do caching, use do_estimate_edge_time to populate the entry. */
if (edge_growth_cache.exists ())
{
do_estimate_edge_time (edge);
hints = edge_growth_cache[edge->uid].hints;
gcc_checking_assert (hints);
return hints - 1;
}
callee = edge->callee->ultimate_alias_target ();
/* Early inliner runs without caching, go ahead and do the dirty work. */
gcc_checking_assert (edge->inline_failed);
evaluate_properties_for_edge (edge, true,
&clause, &known_vals, &known_contexts,
&known_aggs);
estimate_node_size_and_time (callee, clause, known_vals, known_contexts,
known_aggs, NULL, NULL, NULL, &hints, vNULL);
known_vals.release ();
known_contexts.release ();
known_aggs.release ();
hints |= simple_edge_hints (edge);
return hints;
}
/* Estimate self time of the function NODE after inlining EDGE. */
int
estimate_time_after_inlining (struct cgraph_node *node,
struct cgraph_edge *edge)
{
struct inline_edge_summary *es = inline_edge_summary (edge);
if (!es->predicate || !false_predicate_p (es->predicate))
{
gcov_type time =
inline_summaries->get (node)->time + estimate_edge_time (edge);
if (time < 0)
time = 0;
if (time > MAX_TIME)
time = MAX_TIME;
return time;
}
return inline_summaries->get (node)->time;
}
/* Estimate the size of NODE after inlining EDGE which should be an
edge to either NODE or a call inlined into NODE. */
int
estimate_size_after_inlining (struct cgraph_node *node,
struct cgraph_edge *edge)
{
struct inline_edge_summary *es = inline_edge_summary (edge);
if (!es->predicate || !false_predicate_p (es->predicate))
{
int size = inline_summaries->get (node)->size + estimate_edge_growth (edge);
gcc_assert (size >= 0);
return size;
}
return inline_summaries->get (node)->size;
}
struct growth_data
{
struct cgraph_node *node;
bool self_recursive;
bool uninlinable;
int growth;
};
/* Worker for do_estimate_growth. Collect growth for all callers. */
static bool
do_estimate_growth_1 (struct cgraph_node *node, void *data)
{
struct cgraph_edge *e;
struct growth_data *d = (struct growth_data *) data;
for (e = node->callers; e; e = e->next_caller)
{
gcc_checking_assert (e->inline_failed);
if (cgraph_inline_failed_type (e->inline_failed) == CIF_FINAL_ERROR)
{
d->uninlinable = true;
continue;
}
if (e->recursive_p ())
{
d->self_recursive = true;
continue;
}
d->growth += estimate_edge_growth (e);
}
return false;
}
/* Estimate the growth caused by inlining NODE into all callees. */
int
estimate_growth (struct cgraph_node *node)
{
struct growth_data d = { node, false, false, 0 };
struct inline_summary *info = inline_summaries->get (node);
node->call_for_symbol_and_aliases (do_estimate_growth_1, &d, true);
/* For self recursive functions the growth estimation really should be
infinity. We don't want to return very large values because the growth
plays various roles in badness computation fractions. Be sure to not
return zero or negative growths. */
if (d.self_recursive)
d.growth = d.growth < info->size ? info->size : d.growth;
else if (DECL_EXTERNAL (node->decl) || d.uninlinable)
;
else
{
if (node->will_be_removed_from_program_if_no_direct_calls_p ())
d.growth -= info->size;
/* COMDAT functions are very often not shared across multiple units
since they come from various template instantiations.
Take this into account. */
else if (DECL_COMDAT (node->decl)
&& node->can_remove_if_no_direct_calls_p ())
d.growth -= (info->size
* (100 - PARAM_VALUE (PARAM_COMDAT_SHARING_PROBABILITY))
+ 50) / 100;
}
return d.growth;
}
/* Verify if there are fewer than MAX_CALLERS. */
static bool
check_callers (cgraph_node *node, int *max_callers)
{
ipa_ref *ref;
if (!node->can_remove_if_no_direct_calls_and_refs_p ())
return true;
for (cgraph_edge *e = node->callers; e; e = e->next_caller)
{
(*max_callers)--;
if (!*max_callers
|| cgraph_inline_failed_type (e->inline_failed) == CIF_FINAL_ERROR)
return true;
}
FOR_EACH_ALIAS (node, ref)
if (check_callers (dyn_cast <cgraph_node *> (ref->referring), max_callers))
return true;
return false;
}
/* Make cheap estimation if growth of NODE is likely positive knowing
EDGE_GROWTH of one particular edge.
We assume that most of other edges will have similar growth
and skip computation if there are too many callers. */
bool
growth_likely_positive (struct cgraph_node *node,
int edge_growth)
{
int max_callers;
struct cgraph_edge *e;
gcc_checking_assert (edge_growth > 0);
/* First quickly check if NODE is removable at all. */
if (DECL_EXTERNAL (node->decl))
return true;
if (!node->can_remove_if_no_direct_calls_and_refs_p ()
|| node->address_taken)
return true;
max_callers = inline_summaries->get (node)->size * 4 / edge_growth + 2;
for (e = node->callers; e; e = e->next_caller)
{
max_callers--;
if (!max_callers
|| cgraph_inline_failed_type (e->inline_failed) == CIF_FINAL_ERROR)
return true;
}
ipa_ref *ref;
FOR_EACH_ALIAS (node, ref)
if (check_callers (dyn_cast <cgraph_node *> (ref->referring), &max_callers))
return true;
/* Unlike for functions called once, we play unsafe with
COMDATs. We can allow that since we know functions
in consideration are small (and thus risk is small) and
moreover grow estimates already accounts that COMDAT
functions may or may not disappear when eliminated from
current unit. With good probability making aggressive
choice in all units is going to make overall program
smaller. */
if (DECL_COMDAT (node->decl))
{
if (!node->can_remove_if_no_direct_calls_p ())
return true;
}
else if (!node->will_be_removed_from_program_if_no_direct_calls_p ())
return true;
return estimate_growth (node) > 0;
}
/* This function performs intraprocedural analysis in NODE that is required to
inline indirect calls. */
static void
inline_indirect_intraprocedural_analysis (struct cgraph_node *node)
{
ipa_analyze_node (node);
if (dump_file && (dump_flags & TDF_DETAILS))
{
ipa_print_node_params (dump_file, node);
ipa_print_node_jump_functions (dump_file, node);
}
}
/* Note function body size. */
void
inline_analyze_function (struct cgraph_node *node)
{
push_cfun (DECL_STRUCT_FUNCTION (node->decl));
if (dump_file)
fprintf (dump_file, "\nAnalyzing function: %s/%u\n",
node->name (), node->order);
if (opt_for_fn (node->decl, optimize) && !node->thunk.thunk_p)
inline_indirect_intraprocedural_analysis (node);
compute_inline_parameters (node, false);
if (!optimize)
{
struct cgraph_edge *e;
for (e = node->callees; e; e = e->next_callee)
{
if (e->inline_failed == CIF_FUNCTION_NOT_CONSIDERED)
e->inline_failed = CIF_FUNCTION_NOT_OPTIMIZED;
e->call_stmt_cannot_inline_p = true;
}
for (e = node->indirect_calls; e; e = e->next_callee)
{
if (e->inline_failed == CIF_FUNCTION_NOT_CONSIDERED)
e->inline_failed = CIF_FUNCTION_NOT_OPTIMIZED;
e->call_stmt_cannot_inline_p = true;
}
}
pop_cfun ();
}
/* Called when new function is inserted to callgraph late. */
void
inline_summary_t::insert (struct cgraph_node *node, inline_summary *)
{
inline_analyze_function (node);
}
/* Note function body size. */
void
inline_generate_summary (void)
{
struct cgraph_node *node;
/* When not optimizing, do not bother to analyze. Inlining is still done
because edge redirection needs to happen there. */
if (!optimize && !flag_generate_lto && !flag_generate_offload && !flag_wpa)
return;
if (!inline_summaries)
inline_summaries = (inline_summary_t*) inline_summary_t::create_ggc (symtab);
inline_summaries->enable_insertion_hook ();
ipa_register_cgraph_hooks ();
inline_free_summary ();
FOR_EACH_DEFINED_FUNCTION (node)
if (!node->alias)
inline_analyze_function (node);
}
/* Read predicate from IB. */
static struct predicate
read_predicate (struct lto_input_block *ib)
{
struct predicate out;
clause_t clause;
int k = 0;
do
{
gcc_assert (k <= MAX_CLAUSES);
clause = out.clause[k++] = streamer_read_uhwi (ib);
}
while (clause);
/* Zero-initialize the remaining clauses in OUT. */
while (k <= MAX_CLAUSES)
out.clause[k++] = 0;
return out;
}
/* Write inline summary for edge E to OB. */
static void
read_inline_edge_summary (struct lto_input_block *ib, struct cgraph_edge *e)
{
struct inline_edge_summary *es = inline_edge_summary (e);
struct predicate p;
int length, i;
es->call_stmt_size = streamer_read_uhwi (ib);
es->call_stmt_time = streamer_read_uhwi (ib);
es->loop_depth = streamer_read_uhwi (ib);
p = read_predicate (ib);
edge_set_predicate (e, &p);
length = streamer_read_uhwi (ib);
if (length)
{
es->param.safe_grow_cleared (length);
for (i = 0; i < length; i++)
es->param[i].change_prob = streamer_read_uhwi (ib);
}
}
/* Stream in inline summaries from the section. */
static void
inline_read_section (struct lto_file_decl_data *file_data, const char *data,
size_t len)
{
const struct lto_function_header *header =
(const struct lto_function_header *) data;
const int cfg_offset = sizeof (struct lto_function_header);
const int main_offset = cfg_offset + header->cfg_size;
const int string_offset = main_offset + header->main_size;
struct data_in *data_in;
unsigned int i, count2, j;
unsigned int f_count;
lto_input_block ib ((const char *) data + main_offset, header->main_size,
file_data->mode_table);
data_in =
lto_data_in_create (file_data, (const char *) data + string_offset,
header->string_size, vNULL);
f_count = streamer_read_uhwi (&ib);
for (i = 0; i < f_count; i++)
{
unsigned int index;
struct cgraph_node *node;
struct inline_summary *info;
lto_symtab_encoder_t encoder;
struct bitpack_d bp;
struct cgraph_edge *e;
predicate p;
index = streamer_read_uhwi (&ib);
encoder = file_data->symtab_node_encoder;
node = dyn_cast<cgraph_node *> (lto_symtab_encoder_deref (encoder,
index));
info = inline_summaries->get (node);
info->estimated_stack_size
= info->estimated_self_stack_size = streamer_read_uhwi (&ib);
info->size = info->self_size = streamer_read_uhwi (&ib);
info->time = info->self_time = streamer_read_uhwi (&ib);
bp = streamer_read_bitpack (&ib);
info->inlinable = bp_unpack_value (&bp, 1);
info->contains_cilk_spawn = bp_unpack_value (&bp, 1);
count2 = streamer_read_uhwi (&ib);
gcc_assert (!info->conds);
for (j = 0; j < count2; j++)
{
struct condition c;
c.operand_num = streamer_read_uhwi (&ib);
c.code = (enum tree_code) streamer_read_uhwi (&ib);
c.val = stream_read_tree (&ib, data_in);
bp = streamer_read_bitpack (&ib);
c.agg_contents = bp_unpack_value (&bp, 1);
c.by_ref = bp_unpack_value (&bp, 1);
if (c.agg_contents)
c.offset = streamer_read_uhwi (&ib);
vec_safe_push (info->conds, c);
}
count2 = streamer_read_uhwi (&ib);
gcc_assert (!info->entry);
for (j = 0; j < count2; j++)
{
struct size_time_entry e;
e.size = streamer_read_uhwi (&ib);
e.time = streamer_read_uhwi (&ib);
e.predicate = read_predicate (&ib);
vec_safe_push (info->entry, e);
}
p = read_predicate (&ib);
set_hint_predicate (&info->loop_iterations, p);
p = read_predicate (&ib);
set_hint_predicate (&info->loop_stride, p);
p = read_predicate (&ib);
set_hint_predicate (&info->array_index, p);
for (e = node->callees; e; e = e->next_callee)
read_inline_edge_summary (&ib, e);
for (e = node->indirect_calls; e; e = e->next_callee)
read_inline_edge_summary (&ib, e);
}
lto_free_section_data (file_data, LTO_section_inline_summary, NULL, data,
len);
lto_data_in_delete (data_in);
}
/* Read inline summary. Jump functions are shared among ipa-cp
and inliner, so when ipa-cp is active, we don't need to write them
twice. */
void
inline_read_summary (void)
{
struct lto_file_decl_data **file_data_vec = lto_get_file_decl_data ();
struct lto_file_decl_data *file_data;
unsigned int j = 0;
inline_summary_alloc ();
while ((file_data = file_data_vec[j++]))
{
size_t len;
const char *data = lto_get_section_data (file_data,
LTO_section_inline_summary,
NULL, &len);
if (data)
inline_read_section (file_data, data, len);
else
/* Fatal error here. We do not want to support compiling ltrans units
with different version of compiler or different flags than the WPA
unit, so this should never happen. */
fatal_error (input_location,
"ipa inline summary is missing in input file");
}
if (optimize)
{
ipa_register_cgraph_hooks ();
if (!flag_ipa_cp)
ipa_prop_read_jump_functions ();
}
gcc_assert (inline_summaries);
inline_summaries->enable_insertion_hook ();
}
/* Write predicate P to OB. */
static void
write_predicate (struct output_block *ob, struct predicate *p)
{
int j;
if (p)
for (j = 0; p->clause[j]; j++)
{
gcc_assert (j < MAX_CLAUSES);
streamer_write_uhwi (ob, p->clause[j]);
}
streamer_write_uhwi (ob, 0);
}
/* Write inline summary for edge E to OB. */
static void
write_inline_edge_summary (struct output_block *ob, struct cgraph_edge *e)
{
struct inline_edge_summary *es = inline_edge_summary (e);
int i;
streamer_write_uhwi (ob, es->call_stmt_size);
streamer_write_uhwi (ob, es->call_stmt_time);
streamer_write_uhwi (ob, es->loop_depth);
write_predicate (ob, es->predicate);
streamer_write_uhwi (ob, es->param.length ());
for (i = 0; i < (int) es->param.length (); i++)
streamer_write_uhwi (ob, es->param[i].change_prob);
}
/* Write inline summary for node in SET.
Jump functions are shared among ipa-cp and inliner, so when ipa-cp is
active, we don't need to write them twice. */
void
inline_write_summary (void)
{
struct cgraph_node *node;
struct output_block *ob = create_output_block (LTO_section_inline_summary);
lto_symtab_encoder_t encoder = ob->decl_state->symtab_node_encoder;
unsigned int count = 0;
int i;
for (i = 0; i < lto_symtab_encoder_size (encoder); i++)
{
symtab_node *snode = lto_symtab_encoder_deref (encoder, i);
cgraph_node *cnode = dyn_cast <cgraph_node *> (snode);
if (cnode && cnode->definition && !cnode->alias)
count++;
}
streamer_write_uhwi (ob, count);
for (i = 0; i < lto_symtab_encoder_size (encoder); i++)
{
symtab_node *snode = lto_symtab_encoder_deref (encoder, i);
cgraph_node *cnode = dyn_cast <cgraph_node *> (snode);
if (cnode && (node = cnode)->definition && !node->alias)
{
struct inline_summary *info = inline_summaries->get (node);
struct bitpack_d bp;
struct cgraph_edge *edge;
int i;
size_time_entry *e;
struct condition *c;
streamer_write_uhwi (ob,
lto_symtab_encoder_encode (encoder,
node));
streamer_write_hwi (ob, info->estimated_self_stack_size);
streamer_write_hwi (ob, info->self_size);
streamer_write_hwi (ob, info->self_time);
bp = bitpack_create (ob->main_stream);
bp_pack_value (&bp, info->inlinable, 1);
bp_pack_value (&bp, info->contains_cilk_spawn, 1);
streamer_write_bitpack (&bp);
streamer_write_uhwi (ob, vec_safe_length (info->conds));
for (i = 0; vec_safe_iterate (info->conds, i, &c); i++)
{
streamer_write_uhwi (ob, c->operand_num);
streamer_write_uhwi (ob, c->code);
stream_write_tree (ob, c->val, true);
bp = bitpack_create (ob->main_stream);
bp_pack_value (&bp, c->agg_contents, 1);
bp_pack_value (&bp, c->by_ref, 1);
streamer_write_bitpack (&bp);
if (c->agg_contents)
streamer_write_uhwi (ob, c->offset);
}
streamer_write_uhwi (ob, vec_safe_length (info->entry));
for (i = 0; vec_safe_iterate (info->entry, i, &e); i++)
{
streamer_write_uhwi (ob, e->size);
streamer_write_uhwi (ob, e->time);
write_predicate (ob, &e->predicate);
}
write_predicate (ob, info->loop_iterations);
write_predicate (ob, info->loop_stride);
write_predicate (ob, info->array_index);
for (edge = node->callees; edge; edge = edge->next_callee)
write_inline_edge_summary (ob, edge);
for (edge = node->indirect_calls; edge; edge = edge->next_callee)
write_inline_edge_summary (ob, edge);
}
}
streamer_write_char_stream (ob->main_stream, 0);
produce_asm (ob, NULL);
destroy_output_block (ob);
if (optimize && !flag_ipa_cp)
ipa_prop_write_jump_functions ();
}
/* Release inline summary. */
void
inline_free_summary (void)
{
struct cgraph_node *node;
if (edge_removal_hook_holder)
symtab->remove_edge_removal_hook (edge_removal_hook_holder);
edge_removal_hook_holder = NULL;
if (edge_duplication_hook_holder)
symtab->remove_edge_duplication_hook (edge_duplication_hook_holder);
edge_duplication_hook_holder = NULL;
if (!inline_edge_summary_vec.exists ())
return;
FOR_EACH_DEFINED_FUNCTION (node)
if (!node->alias)
reset_inline_summary (node, inline_summaries->get (node));
inline_summaries->release ();
inline_summaries = NULL;
inline_edge_summary_vec.release ();
if (edge_predicate_pool)
free_alloc_pool (edge_predicate_pool);
edge_predicate_pool = 0;
}
|