aboutsummaryrefslogtreecommitdiffstats
path: root/contrib/libs/llvm16/lib/Support/APInt.cpp
blob: afe7478a8b2a5fef573cc5b910a36aa7d442fe27 (plain) (blame)
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
//===-- APInt.cpp - Implement APInt class ---------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// This file implements a class to represent arbitrary precision integer
// constant values and provide a variety of arithmetic operations on them.
//
//===----------------------------------------------------------------------===//

#include "llvm/ADT/APInt.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/FoldingSet.h"
#include "llvm/ADT/Hashing.h"
#include "llvm/ADT/SmallString.h"
#include "llvm/ADT/StringRef.h"
#include "llvm/ADT/bit.h"
#include "llvm/Config/llvm-config.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/MathExtras.h"
#include "llvm/Support/raw_ostream.h"
#include <cmath>
#include <optional>

using namespace llvm;

#define DEBUG_TYPE "apint"

/// A utility function for allocating memory, checking for allocation failures,
/// and ensuring the contents are zeroed.
inline static uint64_t* getClearedMemory(unsigned numWords) {
  uint64_t *result = new uint64_t[numWords];
  memset(result, 0, numWords * sizeof(uint64_t));
  return result;
}

/// A utility function for allocating memory and checking for allocation
/// failure.  The content is not zeroed.
inline static uint64_t* getMemory(unsigned numWords) {
  return new uint64_t[numWords];
}

/// A utility function that converts a character to a digit.
inline static unsigned getDigit(char cdigit, uint8_t radix) {
  unsigned r;

  if (radix == 16 || radix == 36) {
    r = cdigit - '0';
    if (r <= 9)
      return r;

    r = cdigit - 'A';
    if (r <= radix - 11U)
      return r + 10;

    r = cdigit - 'a';
    if (r <= radix - 11U)
      return r + 10;

    radix = 10;
  }

  r = cdigit - '0';
  if (r < radix)
    return r;

  return -1U;
}


void APInt::initSlowCase(uint64_t val, bool isSigned) {
  U.pVal = getClearedMemory(getNumWords());
  U.pVal[0] = val;
  if (isSigned && int64_t(val) < 0)
    for (unsigned i = 1; i < getNumWords(); ++i)
      U.pVal[i] = WORDTYPE_MAX;
  clearUnusedBits();
}

void APInt::initSlowCase(const APInt& that) {
  U.pVal = getMemory(getNumWords());
  memcpy(U.pVal, that.U.pVal, getNumWords() * APINT_WORD_SIZE);
}

void APInt::initFromArray(ArrayRef<uint64_t> bigVal) {
  assert(bigVal.data() && "Null pointer detected!");
  if (isSingleWord())
    U.VAL = bigVal[0];
  else {
    // Get memory, cleared to 0
    U.pVal = getClearedMemory(getNumWords());
    // Calculate the number of words to copy
    unsigned words = std::min<unsigned>(bigVal.size(), getNumWords());
    // Copy the words from bigVal to pVal
    memcpy(U.pVal, bigVal.data(), words * APINT_WORD_SIZE);
  }
  // Make sure unused high bits are cleared
  clearUnusedBits();
}

APInt::APInt(unsigned numBits, ArrayRef<uint64_t> bigVal) : BitWidth(numBits) {
  initFromArray(bigVal);
}

APInt::APInt(unsigned numBits, unsigned numWords, const uint64_t bigVal[])
    : BitWidth(numBits) {
  initFromArray(ArrayRef(bigVal, numWords));
}

APInt::APInt(unsigned numbits, StringRef Str, uint8_t radix)
    : BitWidth(numbits) {
  fromString(numbits, Str, radix);
}

void APInt::reallocate(unsigned NewBitWidth) {
  // If the number of words is the same we can just change the width and stop.
  if (getNumWords() == getNumWords(NewBitWidth)) {
    BitWidth = NewBitWidth;
    return;
  }

  // If we have an allocation, delete it.
  if (!isSingleWord())
    delete [] U.pVal;

  // Update BitWidth.
  BitWidth = NewBitWidth;

  // If we are supposed to have an allocation, create it.
  if (!isSingleWord())
    U.pVal = getMemory(getNumWords());
}

void APInt::assignSlowCase(const APInt &RHS) {
  // Don't do anything for X = X
  if (this == &RHS)
    return;

  // Adjust the bit width and handle allocations as necessary.
  reallocate(RHS.getBitWidth());

  // Copy the data.
  if (isSingleWord())
    U.VAL = RHS.U.VAL;
  else
    memcpy(U.pVal, RHS.U.pVal, getNumWords() * APINT_WORD_SIZE);
}

/// This method 'profiles' an APInt for use with FoldingSet.
void APInt::Profile(FoldingSetNodeID& ID) const {
  ID.AddInteger(BitWidth);

  if (isSingleWord()) {
    ID.AddInteger(U.VAL);
    return;
  }

  unsigned NumWords = getNumWords();
  for (unsigned i = 0; i < NumWords; ++i)
    ID.AddInteger(U.pVal[i]);
}

/// Prefix increment operator. Increments the APInt by one.
APInt& APInt::operator++() {
  if (isSingleWord())
    ++U.VAL;
  else
    tcIncrement(U.pVal, getNumWords());
  return clearUnusedBits();
}

/// Prefix decrement operator. Decrements the APInt by one.
APInt& APInt::operator--() {
  if (isSingleWord())
    --U.VAL;
  else
    tcDecrement(U.pVal, getNumWords());
  return clearUnusedBits();
}

/// Adds the RHS APInt to this APInt.
/// @returns this, after addition of RHS.
/// Addition assignment operator.
APInt& APInt::operator+=(const APInt& RHS) {
  assert(BitWidth == RHS.BitWidth && "Bit widths must be the same");
  if (isSingleWord())
    U.VAL += RHS.U.VAL;
  else
    tcAdd(U.pVal, RHS.U.pVal, 0, getNumWords());
  return clearUnusedBits();
}

APInt& APInt::operator+=(uint64_t RHS) {
  if (isSingleWord())
    U.VAL += RHS;
  else
    tcAddPart(U.pVal, RHS, getNumWords());
  return clearUnusedBits();
}

/// Subtracts the RHS APInt from this APInt
/// @returns this, after subtraction
/// Subtraction assignment operator.
APInt& APInt::operator-=(const APInt& RHS) {
  assert(BitWidth == RHS.BitWidth && "Bit widths must be the same");
  if (isSingleWord())
    U.VAL -= RHS.U.VAL;
  else
    tcSubtract(U.pVal, RHS.U.pVal, 0, getNumWords());
  return clearUnusedBits();
}

APInt& APInt::operator-=(uint64_t RHS) {
  if (isSingleWord())
    U.VAL -= RHS;
  else
    tcSubtractPart(U.pVal, RHS, getNumWords());
  return clearUnusedBits();
}

APInt APInt::operator*(const APInt& RHS) const {
  assert(BitWidth == RHS.BitWidth && "Bit widths must be the same");
  if (isSingleWord())
    return APInt(BitWidth, U.VAL * RHS.U.VAL);

  APInt Result(getMemory(getNumWords()), getBitWidth());
  tcMultiply(Result.U.pVal, U.pVal, RHS.U.pVal, getNumWords());
  Result.clearUnusedBits();
  return Result;
}

void APInt::andAssignSlowCase(const APInt &RHS) {
  WordType *dst = U.pVal, *rhs = RHS.U.pVal;
  for (size_t i = 0, e = getNumWords(); i != e; ++i)
    dst[i] &= rhs[i];
}

void APInt::orAssignSlowCase(const APInt &RHS) {
  WordType *dst = U.pVal, *rhs = RHS.U.pVal;
  for (size_t i = 0, e = getNumWords(); i != e; ++i)
    dst[i] |= rhs[i];
}

void APInt::xorAssignSlowCase(const APInt &RHS) {
  WordType *dst = U.pVal, *rhs = RHS.U.pVal;
  for (size_t i = 0, e = getNumWords(); i != e; ++i)
    dst[i] ^= rhs[i];
}

APInt &APInt::operator*=(const APInt &RHS) {
  *this = *this * RHS;
  return *this;
}

APInt& APInt::operator*=(uint64_t RHS) {
  if (isSingleWord()) {
    U.VAL *= RHS;
  } else {
    unsigned NumWords = getNumWords();
    tcMultiplyPart(U.pVal, U.pVal, RHS, 0, NumWords, NumWords, false);
  }
  return clearUnusedBits();
}

bool APInt::equalSlowCase(const APInt &RHS) const {
  return std::equal(U.pVal, U.pVal + getNumWords(), RHS.U.pVal);
}

int APInt::compare(const APInt& RHS) const {
  assert(BitWidth == RHS.BitWidth && "Bit widths must be same for comparison");
  if (isSingleWord())
    return U.VAL < RHS.U.VAL ? -1 : U.VAL > RHS.U.VAL;

  return tcCompare(U.pVal, RHS.U.pVal, getNumWords());
}

int APInt::compareSigned(const APInt& RHS) const {
  assert(BitWidth == RHS.BitWidth && "Bit widths must be same for comparison");
  if (isSingleWord()) {
    int64_t lhsSext = SignExtend64(U.VAL, BitWidth);
    int64_t rhsSext = SignExtend64(RHS.U.VAL, BitWidth);
    return lhsSext < rhsSext ? -1 : lhsSext > rhsSext;
  }

  bool lhsNeg = isNegative();
  bool rhsNeg = RHS.isNegative();

  // If the sign bits don't match, then (LHS < RHS) if LHS is negative
  if (lhsNeg != rhsNeg)
    return lhsNeg ? -1 : 1;

  // Otherwise we can just use an unsigned comparison, because even negative
  // numbers compare correctly this way if both have the same signed-ness.
  return tcCompare(U.pVal, RHS.U.pVal, getNumWords());
}

void APInt::setBitsSlowCase(unsigned loBit, unsigned hiBit) {
  unsigned loWord = whichWord(loBit);
  unsigned hiWord = whichWord(hiBit);

  // Create an initial mask for the low word with zeros below loBit.
  uint64_t loMask = WORDTYPE_MAX << whichBit(loBit);

  // If hiBit is not aligned, we need a high mask.
  unsigned hiShiftAmt = whichBit(hiBit);
  if (hiShiftAmt != 0) {
    // Create a high mask with zeros above hiBit.
    uint64_t hiMask = WORDTYPE_MAX >> (APINT_BITS_PER_WORD - hiShiftAmt);
    // If loWord and hiWord are equal, then we combine the masks. Otherwise,
    // set the bits in hiWord.
    if (hiWord == loWord)
      loMask &= hiMask;
    else
      U.pVal[hiWord] |= hiMask;
  }
  // Apply the mask to the low word.
  U.pVal[loWord] |= loMask;

  // Fill any words between loWord and hiWord with all ones.
  for (unsigned word = loWord + 1; word < hiWord; ++word)
    U.pVal[word] = WORDTYPE_MAX;
}

// Complement a bignum in-place.
static void tcComplement(APInt::WordType *dst, unsigned parts) {
  for (unsigned i = 0; i < parts; i++)
    dst[i] = ~dst[i];
}

/// Toggle every bit to its opposite value.
void APInt::flipAllBitsSlowCase() {
  tcComplement(U.pVal, getNumWords());
  clearUnusedBits();
}

/// Concatenate the bits from "NewLSB" onto the bottom of *this.  This is
/// equivalent to:
///   (this->zext(NewWidth) << NewLSB.getBitWidth()) | NewLSB.zext(NewWidth)
/// In the slow case, we know the result is large.
APInt APInt::concatSlowCase(const APInt &NewLSB) const {
  unsigned NewWidth = getBitWidth() + NewLSB.getBitWidth();
  APInt Result = NewLSB.zext(NewWidth);
  Result.insertBits(*this, NewLSB.getBitWidth());
  return Result;
}

/// Toggle a given bit to its opposite value whose position is given
/// as "bitPosition".
/// Toggles a given bit to its opposite value.
void APInt::flipBit(unsigned bitPosition) {
  assert(bitPosition < BitWidth && "Out of the bit-width range!");
  setBitVal(bitPosition, !(*this)[bitPosition]);
}

void APInt::insertBits(const APInt &subBits, unsigned bitPosition) {
  unsigned subBitWidth = subBits.getBitWidth();
  assert((subBitWidth + bitPosition) <= BitWidth && "Illegal bit insertion");

  // inserting no bits is a noop.
  if (subBitWidth == 0)
    return;

  // Insertion is a direct copy.
  if (subBitWidth == BitWidth) {
    *this = subBits;
    return;
  }

  // Single word result can be done as a direct bitmask.
  if (isSingleWord()) {
    uint64_t mask = WORDTYPE_MAX >> (APINT_BITS_PER_WORD - subBitWidth);
    U.VAL &= ~(mask << bitPosition);
    U.VAL |= (subBits.U.VAL << bitPosition);
    return;
  }

  unsigned loBit = whichBit(bitPosition);
  unsigned loWord = whichWord(bitPosition);
  unsigned hi1Word = whichWord(bitPosition + subBitWidth - 1);

  // Insertion within a single word can be done as a direct bitmask.
  if (loWord == hi1Word) {
    uint64_t mask = WORDTYPE_MAX >> (APINT_BITS_PER_WORD - subBitWidth);
    U.pVal[loWord] &= ~(mask << loBit);
    U.pVal[loWord] |= (subBits.U.VAL << loBit);
    return;
  }

  // Insert on word boundaries.
  if (loBit == 0) {
    // Direct copy whole words.
    unsigned numWholeSubWords = subBitWidth / APINT_BITS_PER_WORD;
    memcpy(U.pVal + loWord, subBits.getRawData(),
           numWholeSubWords * APINT_WORD_SIZE);

    // Mask+insert remaining bits.
    unsigned remainingBits = subBitWidth % APINT_BITS_PER_WORD;
    if (remainingBits != 0) {
      uint64_t mask = WORDTYPE_MAX >> (APINT_BITS_PER_WORD - remainingBits);
      U.pVal[hi1Word] &= ~mask;
      U.pVal[hi1Word] |= subBits.getWord(subBitWidth - 1);
    }
    return;
  }

  // General case - set/clear individual bits in dst based on src.
  // TODO - there is scope for optimization here, but at the moment this code
  // path is barely used so prefer readability over performance.
  for (unsigned i = 0; i != subBitWidth; ++i)
    setBitVal(bitPosition + i, subBits[i]);
}

void APInt::insertBits(uint64_t subBits, unsigned bitPosition, unsigned numBits) {
  uint64_t maskBits = maskTrailingOnes<uint64_t>(numBits);
  subBits &= maskBits;
  if (isSingleWord()) {
    U.VAL &= ~(maskBits << bitPosition);
    U.VAL |= subBits << bitPosition;
    return;
  }

  unsigned loBit = whichBit(bitPosition);
  unsigned loWord = whichWord(bitPosition);
  unsigned hiWord = whichWord(bitPosition + numBits - 1);
  if (loWord == hiWord) {
    U.pVal[loWord] &= ~(maskBits << loBit);
    U.pVal[loWord] |= subBits << loBit;
    return;
  }

  static_assert(8 * sizeof(WordType) <= 64, "This code assumes only two words affected");
  unsigned wordBits = 8 * sizeof(WordType);
  U.pVal[loWord] &= ~(maskBits << loBit);
  U.pVal[loWord] |= subBits << loBit;

  U.pVal[hiWord] &= ~(maskBits >> (wordBits - loBit));
  U.pVal[hiWord] |= subBits >> (wordBits - loBit);
}

APInt APInt::extractBits(unsigned numBits, unsigned bitPosition) const {
  assert(bitPosition < BitWidth && (numBits + bitPosition) <= BitWidth &&
         "Illegal bit extraction");

  if (isSingleWord())
    return APInt(numBits, U.VAL >> bitPosition);

  unsigned loBit = whichBit(bitPosition);
  unsigned loWord = whichWord(bitPosition);
  unsigned hiWord = whichWord(bitPosition + numBits - 1);

  // Single word result extracting bits from a single word source.
  if (loWord == hiWord)
    return APInt(numBits, U.pVal[loWord] >> loBit);

  // Extracting bits that start on a source word boundary can be done
  // as a fast memory copy.
  if (loBit == 0)
    return APInt(numBits, ArrayRef(U.pVal + loWord, 1 + hiWord - loWord));

  // General case - shift + copy source words directly into place.
  APInt Result(numBits, 0);
  unsigned NumSrcWords = getNumWords();
  unsigned NumDstWords = Result.getNumWords();

  uint64_t *DestPtr = Result.isSingleWord() ? &Result.U.VAL : Result.U.pVal;
  for (unsigned word = 0; word < NumDstWords; ++word) {
    uint64_t w0 = U.pVal[loWord + word];
    uint64_t w1 =
        (loWord + word + 1) < NumSrcWords ? U.pVal[loWord + word + 1] : 0;
    DestPtr[word] = (w0 >> loBit) | (w1 << (APINT_BITS_PER_WORD - loBit));
  }

  return Result.clearUnusedBits();
}

uint64_t APInt::extractBitsAsZExtValue(unsigned numBits,
                                       unsigned bitPosition) const {
  assert(numBits > 0 && "Can't extract zero bits");
  assert(bitPosition < BitWidth && (numBits + bitPosition) <= BitWidth &&
         "Illegal bit extraction");
  assert(numBits <= 64 && "Illegal bit extraction");

  uint64_t maskBits = maskTrailingOnes<uint64_t>(numBits);
  if (isSingleWord())
    return (U.VAL >> bitPosition) & maskBits;

  unsigned loBit = whichBit(bitPosition);
  unsigned loWord = whichWord(bitPosition);
  unsigned hiWord = whichWord(bitPosition + numBits - 1);
  if (loWord == hiWord)
    return (U.pVal[loWord] >> loBit) & maskBits;

  static_assert(8 * sizeof(WordType) <= 64, "This code assumes only two words affected");
  unsigned wordBits = 8 * sizeof(WordType);
  uint64_t retBits = U.pVal[loWord] >> loBit;
  retBits |= U.pVal[hiWord] << (wordBits - loBit);
  retBits &= maskBits;
  return retBits;
}

unsigned APInt::getSufficientBitsNeeded(StringRef Str, uint8_t Radix) {
  assert(!Str.empty() && "Invalid string length");
  size_t StrLen = Str.size();

  // Each computation below needs to know if it's negative.
  unsigned IsNegative = false;
  if (Str[0] == '-' || Str[0] == '+') {
    IsNegative = Str[0] == '-';
    StrLen--;
    assert(StrLen && "String is only a sign, needs a value.");
  }

  // For radixes of power-of-two values, the bits required is accurately and
  // easily computed.
  if (Radix == 2)
    return StrLen + IsNegative;
  if (Radix == 8)
    return StrLen * 3 + IsNegative;
  if (Radix == 16)
    return StrLen * 4 + IsNegative;

  // Compute a sufficient number of bits that is always large enough but might
  // be too large. This avoids the assertion in the constructor. This
  // calculation doesn't work appropriately for the numbers 0-9, so just use 4
  // bits in that case.
  if (Radix == 10)
    return (StrLen == 1 ? 4 : StrLen * 64 / 18) + IsNegative;

  assert(Radix == 36);
  return (StrLen == 1 ? 7 : StrLen * 16 / 3) + IsNegative;
}

unsigned APInt::getBitsNeeded(StringRef str, uint8_t radix) {
  // Compute a sufficient number of bits that is always large enough but might
  // be too large.
  unsigned sufficient = getSufficientBitsNeeded(str, radix);

  // For bases 2, 8, and 16, the sufficient number of bits is exact and we can
  // return the value directly. For bases 10 and 36, we need to do extra work.
  if (radix == 2 || radix == 8 || radix == 16)
    return sufficient;

  // This is grossly inefficient but accurate. We could probably do something
  // with a computation of roughly slen*64/20 and then adjust by the value of
  // the first few digits. But, I'm not sure how accurate that could be.
  size_t slen = str.size();

  // Each computation below needs to know if it's negative.
  StringRef::iterator p = str.begin();
  unsigned isNegative = *p == '-';
  if (*p == '-' || *p == '+') {
    p++;
    slen--;
    assert(slen && "String is only a sign, needs a value.");
  }


  // Convert to the actual binary value.
  APInt tmp(sufficient, StringRef(p, slen), radix);

  // Compute how many bits are required. If the log is infinite, assume we need
  // just bit. If the log is exact and value is negative, then the value is
  // MinSignedValue with (log + 1) bits.
  unsigned log = tmp.logBase2();
  if (log == (unsigned)-1) {
    return isNegative + 1;
  } else if (isNegative && tmp.isPowerOf2()) {
    return isNegative + log;
  } else {
    return isNegative + log + 1;
  }
}

hash_code llvm::hash_value(const APInt &Arg) {
  if (Arg.isSingleWord())
    return hash_combine(Arg.BitWidth, Arg.U.VAL);

  return hash_combine(
      Arg.BitWidth,
      hash_combine_range(Arg.U.pVal, Arg.U.pVal + Arg.getNumWords()));
}

unsigned DenseMapInfo<APInt, void>::getHashValue(const APInt &Key) {
  return static_cast<unsigned>(hash_value(Key));
}

bool APInt::isSplat(unsigned SplatSizeInBits) const {
  assert(getBitWidth() % SplatSizeInBits == 0 &&
         "SplatSizeInBits must divide width!");
  // We can check that all parts of an integer are equal by making use of a
  // little trick: rotate and check if it's still the same value.
  return *this == rotl(SplatSizeInBits);
}

/// This function returns the high "numBits" bits of this APInt.
APInt APInt::getHiBits(unsigned numBits) const {
  return this->lshr(BitWidth - numBits);
}

/// This function returns the low "numBits" bits of this APInt.
APInt APInt::getLoBits(unsigned numBits) const {
  APInt Result(getLowBitsSet(BitWidth, numBits));
  Result &= *this;
  return Result;
}

/// Return a value containing V broadcasted over NewLen bits.
APInt APInt::getSplat(unsigned NewLen, const APInt &V) {
  assert(NewLen >= V.getBitWidth() && "Can't splat to smaller bit width!");

  APInt Val = V.zext(NewLen);
  for (unsigned I = V.getBitWidth(); I < NewLen; I <<= 1)
    Val |= Val << I;

  return Val;
}

unsigned APInt::countLeadingZerosSlowCase() const {
  unsigned Count = 0;
  for (int i = getNumWords()-1; i >= 0; --i) {
    uint64_t V = U.pVal[i];
    if (V == 0)
      Count += APINT_BITS_PER_WORD;
    else {
      Count += llvm::countLeadingZeros(V);
      break;
    }
  }
  // Adjust for unused bits in the most significant word (they are zero).
  unsigned Mod = BitWidth % APINT_BITS_PER_WORD;
  Count -= Mod > 0 ? APINT_BITS_PER_WORD - Mod : 0;
  return Count;
}

unsigned APInt::countLeadingOnesSlowCase() const {
  unsigned highWordBits = BitWidth % APINT_BITS_PER_WORD;
  unsigned shift;
  if (!highWordBits) {
    highWordBits = APINT_BITS_PER_WORD;
    shift = 0;
  } else {
    shift = APINT_BITS_PER_WORD - highWordBits;
  }
  int i = getNumWords() - 1;
  unsigned Count = llvm::countLeadingOnes(U.pVal[i] << shift);
  if (Count == highWordBits) {
    for (i--; i >= 0; --i) {
      if (U.pVal[i] == WORDTYPE_MAX)
        Count += APINT_BITS_PER_WORD;
      else {
        Count += llvm::countLeadingOnes(U.pVal[i]);
        break;
      }
    }
  }
  return Count;
}

unsigned APInt::countTrailingZerosSlowCase() const {
  unsigned Count = 0;
  unsigned i = 0;
  for (; i < getNumWords() && U.pVal[i] == 0; ++i)
    Count += APINT_BITS_PER_WORD;
  if (i < getNumWords())
    Count += llvm::countTrailingZeros(U.pVal[i]);
  return std::min(Count, BitWidth);
}

unsigned APInt::countTrailingOnesSlowCase() const {
  unsigned Count = 0;
  unsigned i = 0;
  for (; i < getNumWords() && U.pVal[i] == WORDTYPE_MAX; ++i)
    Count += APINT_BITS_PER_WORD;
  if (i < getNumWords())
    Count += llvm::countTrailingOnes(U.pVal[i]);
  assert(Count <= BitWidth);
  return Count;
}

unsigned APInt::countPopulationSlowCase() const {
  unsigned Count = 0;
  for (unsigned i = 0; i < getNumWords(); ++i)
    Count += llvm::popcount(U.pVal[i]);
  return Count;
}

bool APInt::intersectsSlowCase(const APInt &RHS) const {
  for (unsigned i = 0, e = getNumWords(); i != e; ++i)
    if ((U.pVal[i] & RHS.U.pVal[i]) != 0)
      return true;

  return false;
}

bool APInt::isSubsetOfSlowCase(const APInt &RHS) const {
  for (unsigned i = 0, e = getNumWords(); i != e; ++i)
    if ((U.pVal[i] & ~RHS.U.pVal[i]) != 0)
      return false;

  return true;
}

APInt APInt::byteSwap() const {
  assert(BitWidth >= 16 && BitWidth % 8 == 0 && "Cannot byteswap!");
  if (BitWidth == 16)
    return APInt(BitWidth, ByteSwap_16(uint16_t(U.VAL)));
  if (BitWidth == 32)
    return APInt(BitWidth, ByteSwap_32(unsigned(U.VAL)));
  if (BitWidth <= 64) {
    uint64_t Tmp1 = ByteSwap_64(U.VAL);
    Tmp1 >>= (64 - BitWidth);
    return APInt(BitWidth, Tmp1);
  }

  APInt Result(getNumWords() * APINT_BITS_PER_WORD, 0);
  for (unsigned I = 0, N = getNumWords(); I != N; ++I)
    Result.U.pVal[I] = ByteSwap_64(U.pVal[N - I - 1]);
  if (Result.BitWidth != BitWidth) {
    Result.lshrInPlace(Result.BitWidth - BitWidth);
    Result.BitWidth = BitWidth;
  }
  return Result;
}

APInt APInt::reverseBits() const {
  switch (BitWidth) {
  case 64:
    return APInt(BitWidth, llvm::reverseBits<uint64_t>(U.VAL));
  case 32:
    return APInt(BitWidth, llvm::reverseBits<uint32_t>(U.VAL));
  case 16:
    return APInt(BitWidth, llvm::reverseBits<uint16_t>(U.VAL));
  case 8:
    return APInt(BitWidth, llvm::reverseBits<uint8_t>(U.VAL));
  case 0:
    return *this;
  default:
    break;
  }

  APInt Val(*this);
  APInt Reversed(BitWidth, 0);
  unsigned S = BitWidth;

  for (; Val != 0; Val.lshrInPlace(1)) {
    Reversed <<= 1;
    Reversed |= Val[0];
    --S;
  }

  Reversed <<= S;
  return Reversed;
}

APInt llvm::APIntOps::GreatestCommonDivisor(APInt A, APInt B) {
  // Fast-path a common case.
  if (A == B) return A;

  // Corner cases: if either operand is zero, the other is the gcd.
  if (!A) return B;
  if (!B) return A;

  // Count common powers of 2 and remove all other powers of 2.
  unsigned Pow2;
  {
    unsigned Pow2_A = A.countTrailingZeros();
    unsigned Pow2_B = B.countTrailingZeros();
    if (Pow2_A > Pow2_B) {
      A.lshrInPlace(Pow2_A - Pow2_B);
      Pow2 = Pow2_B;
    } else if (Pow2_B > Pow2_A) {
      B.lshrInPlace(Pow2_B - Pow2_A);
      Pow2 = Pow2_A;
    } else {
      Pow2 = Pow2_A;
    }
  }

  // Both operands are odd multiples of 2^Pow_2:
  //
  //   gcd(a, b) = gcd(|a - b| / 2^i, min(a, b))
  //
  // This is a modified version of Stein's algorithm, taking advantage of
  // efficient countTrailingZeros().
  while (A != B) {
    if (A.ugt(B)) {
      A -= B;
      A.lshrInPlace(A.countTrailingZeros() - Pow2);
    } else {
      B -= A;
      B.lshrInPlace(B.countTrailingZeros() - Pow2);
    }
  }

  return A;
}

APInt llvm::APIntOps::RoundDoubleToAPInt(double Double, unsigned width) {
  uint64_t I = bit_cast<uint64_t>(Double);

  // Get the sign bit from the highest order bit
  bool isNeg = I >> 63;

  // Get the 11-bit exponent and adjust for the 1023 bit bias
  int64_t exp = ((I >> 52) & 0x7ff) - 1023;

  // If the exponent is negative, the value is < 0 so just return 0.
  if (exp < 0)
    return APInt(width, 0u);

  // Extract the mantissa by clearing the top 12 bits (sign + exponent).
  uint64_t mantissa = (I & (~0ULL >> 12)) | 1ULL << 52;

  // If the exponent doesn't shift all bits out of the mantissa
  if (exp < 52)
    return isNeg ? -APInt(width, mantissa >> (52 - exp)) :
                    APInt(width, mantissa >> (52 - exp));

  // If the client didn't provide enough bits for us to shift the mantissa into
  // then the result is undefined, just return 0
  if (width <= exp - 52)
    return APInt(width, 0);

  // Otherwise, we have to shift the mantissa bits up to the right location
  APInt Tmp(width, mantissa);
  Tmp <<= (unsigned)exp - 52;
  return isNeg ? -Tmp : Tmp;
}

/// This function converts this APInt to a double.
/// The layout for double is as following (IEEE Standard 754):
///  --------------------------------------
/// |  Sign    Exponent    Fraction    Bias |
/// |-------------------------------------- |
/// |  1[63]   11[62-52]   52[51-00]   1023 |
///  --------------------------------------
double APInt::roundToDouble(bool isSigned) const {

  // Handle the simple case where the value is contained in one uint64_t.
  // It is wrong to optimize getWord(0) to VAL; there might be more than one word.
  if (isSingleWord() || getActiveBits() <= APINT_BITS_PER_WORD) {
    if (isSigned) {
      int64_t sext = SignExtend64(getWord(0), BitWidth);
      return double(sext);
    } else
      return double(getWord(0));
  }

  // Determine if the value is negative.
  bool isNeg = isSigned ? (*this)[BitWidth-1] : false;

  // Construct the absolute value if we're negative.
  APInt Tmp(isNeg ? -(*this) : (*this));

  // Figure out how many bits we're using.
  unsigned n = Tmp.getActiveBits();

  // The exponent (without bias normalization) is just the number of bits
  // we are using. Note that the sign bit is gone since we constructed the
  // absolute value.
  uint64_t exp = n;

  // Return infinity for exponent overflow
  if (exp > 1023) {
    if (!isSigned || !isNeg)
      return std::numeric_limits<double>::infinity();
    else
      return -std::numeric_limits<double>::infinity();
  }
  exp += 1023; // Increment for 1023 bias

  // Number of bits in mantissa is 52. To obtain the mantissa value, we must
  // extract the high 52 bits from the correct words in pVal.
  uint64_t mantissa;
  unsigned hiWord = whichWord(n-1);
  if (hiWord == 0) {
    mantissa = Tmp.U.pVal[0];
    if (n > 52)
      mantissa >>= n - 52; // shift down, we want the top 52 bits.
  } else {
    assert(hiWord > 0 && "huh?");
    uint64_t hibits = Tmp.U.pVal[hiWord] << (52 - n % APINT_BITS_PER_WORD);
    uint64_t lobits = Tmp.U.pVal[hiWord-1] >> (11 + n % APINT_BITS_PER_WORD);
    mantissa = hibits | lobits;
  }

  // The leading bit of mantissa is implicit, so get rid of it.
  uint64_t sign = isNeg ? (1ULL << (APINT_BITS_PER_WORD - 1)) : 0;
  uint64_t I = sign | (exp << 52) | mantissa;
  return bit_cast<double>(I);
}

// Truncate to new width.
APInt APInt::trunc(unsigned width) const {
  assert(width <= BitWidth && "Invalid APInt Truncate request");

  if (width <= APINT_BITS_PER_WORD)
    return APInt(width, getRawData()[0]);

  if (width == BitWidth)
    return *this;

  APInt Result(getMemory(getNumWords(width)), width);

  // Copy full words.
  unsigned i;
  for (i = 0; i != width / APINT_BITS_PER_WORD; i++)
    Result.U.pVal[i] = U.pVal[i];

  // Truncate and copy any partial word.
  unsigned bits = (0 - width) % APINT_BITS_PER_WORD;
  if (bits != 0)
    Result.U.pVal[i] = U.pVal[i] << bits >> bits;

  return Result;
}

// Truncate to new width with unsigned saturation.
APInt APInt::truncUSat(unsigned width) const {
  assert(width <= BitWidth && "Invalid APInt Truncate request");

  // Can we just losslessly truncate it?
  if (isIntN(width))
    return trunc(width);
  // If not, then just return the new limit.
  return APInt::getMaxValue(width);
}

// Truncate to new width with signed saturation.
APInt APInt::truncSSat(unsigned width) const {
  assert(width <= BitWidth && "Invalid APInt Truncate request");

  // Can we just losslessly truncate it?
  if (isSignedIntN(width))
    return trunc(width);
  // If not, then just return the new limits.
  return isNegative() ? APInt::getSignedMinValue(width)
                      : APInt::getSignedMaxValue(width);
}

// Sign extend to a new width.
APInt APInt::sext(unsigned Width) const {
  assert(Width >= BitWidth && "Invalid APInt SignExtend request");

  if (Width <= APINT_BITS_PER_WORD)
    return APInt(Width, SignExtend64(U.VAL, BitWidth));

  if (Width == BitWidth)
    return *this;

  APInt Result(getMemory(getNumWords(Width)), Width);

  // Copy words.
  std::memcpy(Result.U.pVal, getRawData(), getNumWords() * APINT_WORD_SIZE);

  // Sign extend the last word since there may be unused bits in the input.
  Result.U.pVal[getNumWords() - 1] =
      SignExtend64(Result.U.pVal[getNumWords() - 1],
                   ((BitWidth - 1) % APINT_BITS_PER_WORD) + 1);

  // Fill with sign bits.
  std::memset(Result.U.pVal + getNumWords(), isNegative() ? -1 : 0,
              (Result.getNumWords() - getNumWords()) * APINT_WORD_SIZE);
  Result.clearUnusedBits();
  return Result;
}

//  Zero extend to a new width.
APInt APInt::zext(unsigned width) const {
  assert(width >= BitWidth && "Invalid APInt ZeroExtend request");

  if (width <= APINT_BITS_PER_WORD)
    return APInt(width, U.VAL);

  if (width == BitWidth)
    return *this;

  APInt Result(getMemory(getNumWords(width)), width);

  // Copy words.
  std::memcpy(Result.U.pVal, getRawData(), getNumWords() * APINT_WORD_SIZE);

  // Zero remaining words.
  std::memset(Result.U.pVal + getNumWords(), 0,
              (Result.getNumWords() - getNumWords()) * APINT_WORD_SIZE);

  return Result;
}

APInt APInt::zextOrTrunc(unsigned width) const {
  if (BitWidth < width)
    return zext(width);
  if (BitWidth > width)
    return trunc(width);
  return *this;
}

APInt APInt::sextOrTrunc(unsigned width) const {
  if (BitWidth < width)
    return sext(width);
  if (BitWidth > width)
    return trunc(width);
  return *this;
}

/// Arithmetic right-shift this APInt by shiftAmt.
/// Arithmetic right-shift function.
void APInt::ashrInPlace(const APInt &shiftAmt) {
  ashrInPlace((unsigned)shiftAmt.getLimitedValue(BitWidth));
}

/// Arithmetic right-shift this APInt by shiftAmt.
/// Arithmetic right-shift function.
void APInt::ashrSlowCase(unsigned ShiftAmt) {
  // Don't bother performing a no-op shift.
  if (!ShiftAmt)
    return;

  // Save the original sign bit for later.
  bool Negative = isNegative();

  // WordShift is the inter-part shift; BitShift is intra-part shift.
  unsigned WordShift = ShiftAmt / APINT_BITS_PER_WORD;
  unsigned BitShift = ShiftAmt % APINT_BITS_PER_WORD;

  unsigned WordsToMove = getNumWords() - WordShift;
  if (WordsToMove != 0) {
    // Sign extend the last word to fill in the unused bits.
    U.pVal[getNumWords() - 1] = SignExtend64(
        U.pVal[getNumWords() - 1], ((BitWidth - 1) % APINT_BITS_PER_WORD) + 1);

    // Fastpath for moving by whole words.
    if (BitShift == 0) {
      std::memmove(U.pVal, U.pVal + WordShift, WordsToMove * APINT_WORD_SIZE);
    } else {
      // Move the words containing significant bits.
      for (unsigned i = 0; i != WordsToMove - 1; ++i)
        U.pVal[i] = (U.pVal[i + WordShift] >> BitShift) |
                    (U.pVal[i + WordShift + 1] << (APINT_BITS_PER_WORD - BitShift));

      // Handle the last word which has no high bits to copy.
      U.pVal[WordsToMove - 1] = U.pVal[WordShift + WordsToMove - 1] >> BitShift;
      // Sign extend one more time.
      U.pVal[WordsToMove - 1] =
          SignExtend64(U.pVal[WordsToMove - 1], APINT_BITS_PER_WORD - BitShift);
    }
  }

  // Fill in the remainder based on the original sign.
  std::memset(U.pVal + WordsToMove, Negative ? -1 : 0,
              WordShift * APINT_WORD_SIZE);
  clearUnusedBits();
}

/// Logical right-shift this APInt by shiftAmt.
/// Logical right-shift function.
void APInt::lshrInPlace(const APInt &shiftAmt) {
  lshrInPlace((unsigned)shiftAmt.getLimitedValue(BitWidth));
}

/// Logical right-shift this APInt by shiftAmt.
/// Logical right-shift function.
void APInt::lshrSlowCase(unsigned ShiftAmt) {
  tcShiftRight(U.pVal, getNumWords(), ShiftAmt);
}

/// Left-shift this APInt by shiftAmt.
/// Left-shift function.
APInt &APInt::operator<<=(const APInt &shiftAmt) {
  // It's undefined behavior in C to shift by BitWidth or greater.
  *this <<= (unsigned)shiftAmt.getLimitedValue(BitWidth);
  return *this;
}

void APInt::shlSlowCase(unsigned ShiftAmt) {
  tcShiftLeft(U.pVal, getNumWords(), ShiftAmt);
  clearUnusedBits();
}

// Calculate the rotate amount modulo the bit width.
static unsigned rotateModulo(unsigned BitWidth, const APInt &rotateAmt) {
  if (LLVM_UNLIKELY(BitWidth == 0))
    return 0;
  unsigned rotBitWidth = rotateAmt.getBitWidth();
  APInt rot = rotateAmt;
  if (rotBitWidth < BitWidth) {
    // Extend the rotate APInt, so that the urem doesn't divide by 0.
    // e.g. APInt(1, 32) would give APInt(1, 0).
    rot = rotateAmt.zext(BitWidth);
  }
  rot = rot.urem(APInt(rot.getBitWidth(), BitWidth));
  return rot.getLimitedValue(BitWidth);
}

APInt APInt::rotl(const APInt &rotateAmt) const {
  return rotl(rotateModulo(BitWidth, rotateAmt));
}

APInt APInt::rotl(unsigned rotateAmt) const {
  if (LLVM_UNLIKELY(BitWidth == 0))
    return *this;
  rotateAmt %= BitWidth;
  if (rotateAmt == 0)
    return *this;
  return shl(rotateAmt) | lshr(BitWidth - rotateAmt);
}

APInt APInt::rotr(const APInt &rotateAmt) const {
  return rotr(rotateModulo(BitWidth, rotateAmt));
}

APInt APInt::rotr(unsigned rotateAmt) const {
  if (BitWidth == 0)
    return *this;
  rotateAmt %= BitWidth;
  if (rotateAmt == 0)
    return *this;
  return lshr(rotateAmt) | shl(BitWidth - rotateAmt);
}

/// \returns the nearest log base 2 of this APInt. Ties round up.
///
/// NOTE: When we have a BitWidth of 1, we define:
///
///   log2(0) = UINT32_MAX
///   log2(1) = 0
///
/// to get around any mathematical concerns resulting from
/// referencing 2 in a space where 2 does no exist.
unsigned APInt::nearestLogBase2() const {
  // Special case when we have a bitwidth of 1. If VAL is 1, then we
  // get 0. If VAL is 0, we get WORDTYPE_MAX which gets truncated to
  // UINT32_MAX.
  if (BitWidth == 1)
    return U.VAL - 1;

  // Handle the zero case.
  if (isZero())
    return UINT32_MAX;

  // The non-zero case is handled by computing:
  //
  //   nearestLogBase2(x) = logBase2(x) + x[logBase2(x)-1].
  //
  // where x[i] is referring to the value of the ith bit of x.
  unsigned lg = logBase2();
  return lg + unsigned((*this)[lg - 1]);
}

// Square Root - this method computes and returns the square root of "this".
// Three mechanisms are used for computation. For small values (<= 5 bits),
// a table lookup is done. This gets some performance for common cases. For
// values using less than 52 bits, the value is converted to double and then
// the libc sqrt function is called. The result is rounded and then converted
// back to a uint64_t which is then used to construct the result. Finally,
// the Babylonian method for computing square roots is used.
APInt APInt::sqrt() const {

  // Determine the magnitude of the value.
  unsigned magnitude = getActiveBits();

  // Use a fast table for some small values. This also gets rid of some
  // rounding errors in libc sqrt for small values.
  if (magnitude <= 5) {
    static const uint8_t results[32] = {
      /*     0 */ 0,
      /*  1- 2 */ 1, 1,
      /*  3- 6 */ 2, 2, 2, 2,
      /*  7-12 */ 3, 3, 3, 3, 3, 3,
      /* 13-20 */ 4, 4, 4, 4, 4, 4, 4, 4,
      /* 21-30 */ 5, 5, 5, 5, 5, 5, 5, 5, 5, 5,
      /*    31 */ 6
    };
    return APInt(BitWidth, results[ (isSingleWord() ? U.VAL : U.pVal[0]) ]);
  }

  // If the magnitude of the value fits in less than 52 bits (the precision of
  // an IEEE double precision floating point value), then we can use the
  // libc sqrt function which will probably use a hardware sqrt computation.
  // This should be faster than the algorithm below.
  if (magnitude < 52) {
    return APInt(BitWidth,
                 uint64_t(::round(::sqrt(double(isSingleWord() ? U.VAL
                                                               : U.pVal[0])))));
  }

  // Okay, all the short cuts are exhausted. We must compute it. The following
  // is a classical Babylonian method for computing the square root. This code
  // was adapted to APInt from a wikipedia article on such computations.
  // See http://www.wikipedia.org/ and go to the page named
  // Calculate_an_integer_square_root.
  unsigned nbits = BitWidth, i = 4;
  APInt testy(BitWidth, 16);
  APInt x_old(BitWidth, 1);
  APInt x_new(BitWidth, 0);
  APInt two(BitWidth, 2);

  // Select a good starting value using binary logarithms.
  for (;; i += 2, testy = testy.shl(2))
    if (i >= nbits || this->ule(testy)) {
      x_old = x_old.shl(i / 2);
      break;
    }

  // Use the Babylonian method to arrive at the integer square root:
  for (;;) {
    x_new = (this->udiv(x_old) + x_old).udiv(two);
    if (x_old.ule(x_new))
      break;
    x_old = x_new;
  }

  // Make sure we return the closest approximation
  // NOTE: The rounding calculation below is correct. It will produce an
  // off-by-one discrepancy with results from pari/gp. That discrepancy has been
  // determined to be a rounding issue with pari/gp as it begins to use a
  // floating point representation after 192 bits. There are no discrepancies
  // between this algorithm and pari/gp for bit widths < 192 bits.
  APInt square(x_old * x_old);
  APInt nextSquare((x_old + 1) * (x_old +1));
  if (this->ult(square))
    return x_old;
  assert(this->ule(nextSquare) && "Error in APInt::sqrt computation");
  APInt midpoint((nextSquare - square).udiv(two));
  APInt offset(*this - square);
  if (offset.ult(midpoint))
    return x_old;
  return x_old + 1;
}

/// Computes the multiplicative inverse of this APInt for a given modulo. The
/// iterative extended Euclidean algorithm is used to solve for this value,
/// however we simplify it to speed up calculating only the inverse, and take
/// advantage of div+rem calculations. We also use some tricks to avoid copying
/// (potentially large) APInts around.
/// WARNING: a value of '0' may be returned,
///          signifying that no multiplicative inverse exists!
APInt APInt::multiplicativeInverse(const APInt& modulo) const {
  assert(ult(modulo) && "This APInt must be smaller than the modulo");

  // Using the properties listed at the following web page (accessed 06/21/08):
  //   http://www.numbertheory.org/php/euclid.html
  // (especially the properties numbered 3, 4 and 9) it can be proved that
  // BitWidth bits suffice for all the computations in the algorithm implemented
  // below. More precisely, this number of bits suffice if the multiplicative
  // inverse exists, but may not suffice for the general extended Euclidean
  // algorithm.

  APInt r[2] = { modulo, *this };
  APInt t[2] = { APInt(BitWidth, 0), APInt(BitWidth, 1) };
  APInt q(BitWidth, 0);

  unsigned i;
  for (i = 0; r[i^1] != 0; i ^= 1) {
    // An overview of the math without the confusing bit-flipping:
    // q = r[i-2] / r[i-1]
    // r[i] = r[i-2] % r[i-1]
    // t[i] = t[i-2] - t[i-1] * q
    udivrem(r[i], r[i^1], q, r[i]);
    t[i] -= t[i^1] * q;
  }

  // If this APInt and the modulo are not coprime, there is no multiplicative
  // inverse, so return 0. We check this by looking at the next-to-last
  // remainder, which is the gcd(*this,modulo) as calculated by the Euclidean
  // algorithm.
  if (r[i] != 1)
    return APInt(BitWidth, 0);

  // The next-to-last t is the multiplicative inverse.  However, we are
  // interested in a positive inverse. Calculate a positive one from a negative
  // one if necessary. A simple addition of the modulo suffices because
  // abs(t[i]) is known to be less than *this/2 (see the link above).
  if (t[i].isNegative())
    t[i] += modulo;

  return std::move(t[i]);
}

/// Implementation of Knuth's Algorithm D (Division of nonnegative integers)
/// from "Art of Computer Programming, Volume 2", section 4.3.1, p. 272. The
/// variables here have the same names as in the algorithm. Comments explain
/// the algorithm and any deviation from it.
static void KnuthDiv(uint32_t *u, uint32_t *v, uint32_t *q, uint32_t* r,
                     unsigned m, unsigned n) {
  assert(u && "Must provide dividend");
  assert(v && "Must provide divisor");
  assert(q && "Must provide quotient");
  assert(u != v && u != q && v != q && "Must use different memory");
  assert(n>1 && "n must be > 1");

  // b denotes the base of the number system. In our case b is 2^32.
  const uint64_t b = uint64_t(1) << 32;

// The DEBUG macros here tend to be spam in the debug output if you're not
// debugging this code. Disable them unless KNUTH_DEBUG is defined.
#ifdef KNUTH_DEBUG
#define DEBUG_KNUTH(X) LLVM_DEBUG(X)
#else
#define DEBUG_KNUTH(X) do {} while(false)
#endif

  DEBUG_KNUTH(dbgs() << "KnuthDiv: m=" << m << " n=" << n << '\n');
  DEBUG_KNUTH(dbgs() << "KnuthDiv: original:");
  DEBUG_KNUTH(for (int i = m + n; i >= 0; i--) dbgs() << " " << u[i]);
  DEBUG_KNUTH(dbgs() << " by");
  DEBUG_KNUTH(for (int i = n; i > 0; i--) dbgs() << " " << v[i - 1]);
  DEBUG_KNUTH(dbgs() << '\n');
  // D1. [Normalize.] Set d = b / (v[n-1] + 1) and multiply all the digits of
  // u and v by d. Note that we have taken Knuth's advice here to use a power
  // of 2 value for d such that d * v[n-1] >= b/2 (b is the base). A power of
  // 2 allows us to shift instead of multiply and it is easy to determine the
  // shift amount from the leading zeros.  We are basically normalizing the u
  // and v so that its high bits are shifted to the top of v's range without
  // overflow. Note that this can require an extra word in u so that u must
  // be of length m+n+1.
  unsigned shift = countLeadingZeros(v[n-1]);
  uint32_t v_carry = 0;
  uint32_t u_carry = 0;
  if (shift) {
    for (unsigned i = 0; i < m+n; ++i) {
      uint32_t u_tmp = u[i] >> (32 - shift);
      u[i] = (u[i] << shift) | u_carry;
      u_carry = u_tmp;
    }
    for (unsigned i = 0; i < n; ++i) {
      uint32_t v_tmp = v[i] >> (32 - shift);
      v[i] = (v[i] << shift) | v_carry;
      v_carry = v_tmp;
    }
  }
  u[m+n] = u_carry;

  DEBUG_KNUTH(dbgs() << "KnuthDiv:   normal:");
  DEBUG_KNUTH(for (int i = m + n; i >= 0; i--) dbgs() << " " << u[i]);
  DEBUG_KNUTH(dbgs() << " by");
  DEBUG_KNUTH(for (int i = n; i > 0; i--) dbgs() << " " << v[i - 1]);
  DEBUG_KNUTH(dbgs() << '\n');

  // D2. [Initialize j.]  Set j to m. This is the loop counter over the places.
  int j = m;
  do {
    DEBUG_KNUTH(dbgs() << "KnuthDiv: quotient digit #" << j << '\n');
    // D3. [Calculate q'.].
    //     Set qp = (u[j+n]*b + u[j+n-1]) / v[n-1]. (qp=qprime=q')
    //     Set rp = (u[j+n]*b + u[j+n-1]) % v[n-1]. (rp=rprime=r')
    // Now test if qp == b or qp*v[n-2] > b*rp + u[j+n-2]; if so, decrease
    // qp by 1, increase rp by v[n-1], and repeat this test if rp < b. The test
    // on v[n-2] determines at high speed most of the cases in which the trial
    // value qp is one too large, and it eliminates all cases where qp is two
    // too large.
    uint64_t dividend = Make_64(u[j+n], u[j+n-1]);
    DEBUG_KNUTH(dbgs() << "KnuthDiv: dividend == " << dividend << '\n');
    uint64_t qp = dividend / v[n-1];
    uint64_t rp = dividend % v[n-1];
    if (qp == b || qp*v[n-2] > b*rp + u[j+n-2]) {
      qp--;
      rp += v[n-1];
      if (rp < b && (qp == b || qp*v[n-2] > b*rp + u[j+n-2]))
        qp--;
    }
    DEBUG_KNUTH(dbgs() << "KnuthDiv: qp == " << qp << ", rp == " << rp << '\n');

    // D4. [Multiply and subtract.] Replace (u[j+n]u[j+n-1]...u[j]) with
    // (u[j+n]u[j+n-1]..u[j]) - qp * (v[n-1]...v[1]v[0]). This computation
    // consists of a simple multiplication by a one-place number, combined with
    // a subtraction.
    // The digits (u[j+n]...u[j]) should be kept positive; if the result of
    // this step is actually negative, (u[j+n]...u[j]) should be left as the
    // true value plus b**(n+1), namely as the b's complement of
    // the true value, and a "borrow" to the left should be remembered.
    int64_t borrow = 0;
    for (unsigned i = 0; i < n; ++i) {
      uint64_t p = uint64_t(qp) * uint64_t(v[i]);
      int64_t subres = int64_t(u[j+i]) - borrow - Lo_32(p);
      u[j+i] = Lo_32(subres);
      borrow = Hi_32(p) - Hi_32(subres);
      DEBUG_KNUTH(dbgs() << "KnuthDiv: u[j+i] = " << u[j + i]
                        << ", borrow = " << borrow << '\n');
    }
    bool isNeg = u[j+n] < borrow;
    u[j+n] -= Lo_32(borrow);

    DEBUG_KNUTH(dbgs() << "KnuthDiv: after subtraction:");
    DEBUG_KNUTH(for (int i = m + n; i >= 0; i--) dbgs() << " " << u[i]);
    DEBUG_KNUTH(dbgs() << '\n');

    // D5. [Test remainder.] Set q[j] = qp. If the result of step D4 was
    // negative, go to step D6; otherwise go on to step D7.
    q[j] = Lo_32(qp);
    if (isNeg) {
      // D6. [Add back]. The probability that this step is necessary is very
      // small, on the order of only 2/b. Make sure that test data accounts for
      // this possibility. Decrease q[j] by 1
      q[j]--;
      // and add (0v[n-1]...v[1]v[0]) to (u[j+n]u[j+n-1]...u[j+1]u[j]).
      // A carry will occur to the left of u[j+n], and it should be ignored
      // since it cancels with the borrow that occurred in D4.
      bool carry = false;
      for (unsigned i = 0; i < n; i++) {
        uint32_t limit = std::min(u[j+i],v[i]);
        u[j+i] += v[i] + carry;
        carry = u[j+i] < limit || (carry && u[j+i] == limit);
      }
      u[j+n] += carry;
    }
    DEBUG_KNUTH(dbgs() << "KnuthDiv: after correction:");
    DEBUG_KNUTH(for (int i = m + n; i >= 0; i--) dbgs() << " " << u[i]);
    DEBUG_KNUTH(dbgs() << "\nKnuthDiv: digit result = " << q[j] << '\n');

    // D7. [Loop on j.]  Decrease j by one. Now if j >= 0, go back to D3.
  } while (--j >= 0);

  DEBUG_KNUTH(dbgs() << "KnuthDiv: quotient:");
  DEBUG_KNUTH(for (int i = m; i >= 0; i--) dbgs() << " " << q[i]);
  DEBUG_KNUTH(dbgs() << '\n');

  // D8. [Unnormalize]. Now q[...] is the desired quotient, and the desired
  // remainder may be obtained by dividing u[...] by d. If r is non-null we
  // compute the remainder (urem uses this).
  if (r) {
    // The value d is expressed by the "shift" value above since we avoided
    // multiplication by d by using a shift left. So, all we have to do is
    // shift right here.
    if (shift) {
      uint32_t carry = 0;
      DEBUG_KNUTH(dbgs() << "KnuthDiv: remainder:");
      for (int i = n-1; i >= 0; i--) {
        r[i] = (u[i] >> shift) | carry;
        carry = u[i] << (32 - shift);
        DEBUG_KNUTH(dbgs() << " " << r[i]);
      }
    } else {
      for (int i = n-1; i >= 0; i--) {
        r[i] = u[i];
        DEBUG_KNUTH(dbgs() << " " << r[i]);
      }
    }
    DEBUG_KNUTH(dbgs() << '\n');
  }
  DEBUG_KNUTH(dbgs() << '\n');
}

void APInt::divide(const WordType *LHS, unsigned lhsWords, const WordType *RHS,
                   unsigned rhsWords, WordType *Quotient, WordType *Remainder) {
  assert(lhsWords >= rhsWords && "Fractional result");

  // First, compose the values into an array of 32-bit words instead of
  // 64-bit words. This is a necessity of both the "short division" algorithm
  // and the Knuth "classical algorithm" which requires there to be native
  // operations for +, -, and * on an m bit value with an m*2 bit result. We
  // can't use 64-bit operands here because we don't have native results of
  // 128-bits. Furthermore, casting the 64-bit values to 32-bit values won't
  // work on large-endian machines.
  unsigned n = rhsWords * 2;
  unsigned m = (lhsWords * 2) - n;

  // Allocate space for the temporary values we need either on the stack, if
  // it will fit, or on the heap if it won't.
  uint32_t SPACE[128];
  uint32_t *U = nullptr;
  uint32_t *V = nullptr;
  uint32_t *Q = nullptr;
  uint32_t *R = nullptr;
  if ((Remainder?4:3)*n+2*m+1 <= 128) {
    U = &SPACE[0];
    V = &SPACE[m+n+1];
    Q = &SPACE[(m+n+1) + n];
    if (Remainder)
      R = &SPACE[(m+n+1) + n + (m+n)];
  } else {
    U = new uint32_t[m + n + 1];
    V = new uint32_t[n];
    Q = new uint32_t[m+n];
    if (Remainder)
      R = new uint32_t[n];
  }

  // Initialize the dividend
  memset(U, 0, (m+n+1)*sizeof(uint32_t));
  for (unsigned i = 0; i < lhsWords; ++i) {
    uint64_t tmp = LHS[i];
    U[i * 2] = Lo_32(tmp);
    U[i * 2 + 1] = Hi_32(tmp);
  }
  U[m+n] = 0; // this extra word is for "spill" in the Knuth algorithm.

  // Initialize the divisor
  memset(V, 0, (n)*sizeof(uint32_t));
  for (unsigned i = 0; i < rhsWords; ++i) {
    uint64_t tmp = RHS[i];
    V[i * 2] = Lo_32(tmp);
    V[i * 2 + 1] = Hi_32(tmp);
  }

  // initialize the quotient and remainder
  memset(Q, 0, (m+n) * sizeof(uint32_t));
  if (Remainder)
    memset(R, 0, n * sizeof(uint32_t));

  // Now, adjust m and n for the Knuth division. n is the number of words in
  // the divisor. m is the number of words by which the dividend exceeds the
  // divisor (i.e. m+n is the length of the dividend). These sizes must not
  // contain any zero words or the Knuth algorithm fails.
  for (unsigned i = n; i > 0 && V[i-1] == 0; i--) {
    n--;
    m++;
  }
  for (unsigned i = m+n; i > 0 && U[i-1] == 0; i--)
    m--;

  // If we're left with only a single word for the divisor, Knuth doesn't work
  // so we implement the short division algorithm here. This is much simpler
  // and faster because we are certain that we can divide a 64-bit quantity
  // by a 32-bit quantity at hardware speed and short division is simply a
  // series of such operations. This is just like doing short division but we
  // are using base 2^32 instead of base 10.
  assert(n != 0 && "Divide by zero?");
  if (n == 1) {
    uint32_t divisor = V[0];
    uint32_t remainder = 0;
    for (int i = m; i >= 0; i--) {
      uint64_t partial_dividend = Make_64(remainder, U[i]);
      if (partial_dividend == 0) {
        Q[i] = 0;
        remainder = 0;
      } else if (partial_dividend < divisor) {
        Q[i] = 0;
        remainder = Lo_32(partial_dividend);
      } else if (partial_dividend == divisor) {
        Q[i] = 1;
        remainder = 0;
      } else {
        Q[i] = Lo_32(partial_dividend / divisor);
        remainder = Lo_32(partial_dividend - (Q[i] * divisor));
      }
    }
    if (R)
      R[0] = remainder;
  } else {
    // Now we're ready to invoke the Knuth classical divide algorithm. In this
    // case n > 1.
    KnuthDiv(U, V, Q, R, m, n);
  }

  // If the caller wants the quotient
  if (Quotient) {
    for (unsigned i = 0; i < lhsWords; ++i)
      Quotient[i] = Make_64(Q[i*2+1], Q[i*2]);
  }

  // If the caller wants the remainder
  if (Remainder) {
    for (unsigned i = 0; i < rhsWords; ++i)
      Remainder[i] = Make_64(R[i*2+1], R[i*2]);
  }

  // Clean up the memory we allocated.
  if (U != &SPACE[0]) {
    delete [] U;
    delete [] V;
    delete [] Q;
    delete [] R;
  }
}

APInt APInt::udiv(const APInt &RHS) const {
  assert(BitWidth == RHS.BitWidth && "Bit widths must be the same");

  // First, deal with the easy case
  if (isSingleWord()) {
    assert(RHS.U.VAL != 0 && "Divide by zero?");
    return APInt(BitWidth, U.VAL / RHS.U.VAL);
  }

  // Get some facts about the LHS and RHS number of bits and words
  unsigned lhsWords = getNumWords(getActiveBits());
  unsigned rhsBits  = RHS.getActiveBits();
  unsigned rhsWords = getNumWords(rhsBits);
  assert(rhsWords && "Divided by zero???");

  // Deal with some degenerate cases
  if (!lhsWords)
    // 0 / X ===> 0
    return APInt(BitWidth, 0);
  if (rhsBits == 1)
    // X / 1 ===> X
    return *this;
  if (lhsWords < rhsWords || this->ult(RHS))
    // X / Y ===> 0, iff X < Y
    return APInt(BitWidth, 0);
  if (*this == RHS)
    // X / X ===> 1
    return APInt(BitWidth, 1);
  if (lhsWords == 1) // rhsWords is 1 if lhsWords is 1.
    // All high words are zero, just use native divide
    return APInt(BitWidth, this->U.pVal[0] / RHS.U.pVal[0]);

  // We have to compute it the hard way. Invoke the Knuth divide algorithm.
  APInt Quotient(BitWidth, 0); // to hold result.
  divide(U.pVal, lhsWords, RHS.U.pVal, rhsWords, Quotient.U.pVal, nullptr);
  return Quotient;
}

APInt APInt::udiv(uint64_t RHS) const {
  assert(RHS != 0 && "Divide by zero?");

  // First, deal with the easy case
  if (isSingleWord())
    return APInt(BitWidth, U.VAL / RHS);

  // Get some facts about the LHS words.
  unsigned lhsWords = getNumWords(getActiveBits());

  // Deal with some degenerate cases
  if (!lhsWords)
    // 0 / X ===> 0
    return APInt(BitWidth, 0);
  if (RHS == 1)
    // X / 1 ===> X
    return *this;
  if (this->ult(RHS))
    // X / Y ===> 0, iff X < Y
    return APInt(BitWidth, 0);
  if (*this == RHS)
    // X / X ===> 1
    return APInt(BitWidth, 1);
  if (lhsWords == 1) // rhsWords is 1 if lhsWords is 1.
    // All high words are zero, just use native divide
    return APInt(BitWidth, this->U.pVal[0] / RHS);

  // We have to compute it the hard way. Invoke the Knuth divide algorithm.
  APInt Quotient(BitWidth, 0); // to hold result.
  divide(U.pVal, lhsWords, &RHS, 1, Quotient.U.pVal, nullptr);
  return Quotient;
}

APInt APInt::sdiv(const APInt &RHS) const {
  if (isNegative()) {
    if (RHS.isNegative())
      return (-(*this)).udiv(-RHS);
    return -((-(*this)).udiv(RHS));
  }
  if (RHS.isNegative())
    return -(this->udiv(-RHS));
  return this->udiv(RHS);
}

APInt APInt::sdiv(int64_t RHS) const {
  if (isNegative()) {
    if (RHS < 0)
      return (-(*this)).udiv(-RHS);
    return -((-(*this)).udiv(RHS));
  }
  if (RHS < 0)
    return -(this->udiv(-RHS));
  return this->udiv(RHS);
}

APInt APInt::urem(const APInt &RHS) const {
  assert(BitWidth == RHS.BitWidth && "Bit widths must be the same");
  if (isSingleWord()) {
    assert(RHS.U.VAL != 0 && "Remainder by zero?");
    return APInt(BitWidth, U.VAL % RHS.U.VAL);
  }

  // Get some facts about the LHS
  unsigned lhsWords = getNumWords(getActiveBits());

  // Get some facts about the RHS
  unsigned rhsBits = RHS.getActiveBits();
  unsigned rhsWords = getNumWords(rhsBits);
  assert(rhsWords && "Performing remainder operation by zero ???");

  // Check the degenerate cases
  if (lhsWords == 0)
    // 0 % Y ===> 0
    return APInt(BitWidth, 0);
  if (rhsBits == 1)
    // X % 1 ===> 0
    return APInt(BitWidth, 0);
  if (lhsWords < rhsWords || this->ult(RHS))
    // X % Y ===> X, iff X < Y
    return *this;
  if (*this == RHS)
    // X % X == 0;
    return APInt(BitWidth, 0);
  if (lhsWords == 1)
    // All high words are zero, just use native remainder
    return APInt(BitWidth, U.pVal[0] % RHS.U.pVal[0]);

  // We have to compute it the hard way. Invoke the Knuth divide algorithm.
  APInt Remainder(BitWidth, 0);
  divide(U.pVal, lhsWords, RHS.U.pVal, rhsWords, nullptr, Remainder.U.pVal);
  return Remainder;
}

uint64_t APInt::urem(uint64_t RHS) const {
  assert(RHS != 0 && "Remainder by zero?");

  if (isSingleWord())
    return U.VAL % RHS;

  // Get some facts about the LHS
  unsigned lhsWords = getNumWords(getActiveBits());

  // Check the degenerate cases
  if (lhsWords == 0)
    // 0 % Y ===> 0
    return 0;
  if (RHS == 1)
    // X % 1 ===> 0
    return 0;
  if (this->ult(RHS))
    // X % Y ===> X, iff X < Y
    return getZExtValue();
  if (*this == RHS)
    // X % X == 0;
    return 0;
  if (lhsWords == 1)
    // All high words are zero, just use native remainder
    return U.pVal[0] % RHS;

  // We have to compute it the hard way. Invoke the Knuth divide algorithm.
  uint64_t Remainder;
  divide(U.pVal, lhsWords, &RHS, 1, nullptr, &Remainder);
  return Remainder;
}

APInt APInt::srem(const APInt &RHS) const {
  if (isNegative()) {
    if (RHS.isNegative())
      return -((-(*this)).urem(-RHS));
    return -((-(*this)).urem(RHS));
  }
  if (RHS.isNegative())
    return this->urem(-RHS);
  return this->urem(RHS);
}

int64_t APInt::srem(int64_t RHS) const {
  if (isNegative()) {
    if (RHS < 0)
      return -((-(*this)).urem(-RHS));
    return -((-(*this)).urem(RHS));
  }
  if (RHS < 0)
    return this->urem(-RHS);
  return this->urem(RHS);
}

void APInt::udivrem(const APInt &LHS, const APInt &RHS,
                    APInt &Quotient, APInt &Remainder) {
  assert(LHS.BitWidth == RHS.BitWidth && "Bit widths must be the same");
  unsigned BitWidth = LHS.BitWidth;

  // First, deal with the easy case
  if (LHS.isSingleWord()) {
    assert(RHS.U.VAL != 0 && "Divide by zero?");
    uint64_t QuotVal = LHS.U.VAL / RHS.U.VAL;
    uint64_t RemVal = LHS.U.VAL % RHS.U.VAL;
    Quotient = APInt(BitWidth, QuotVal);
    Remainder = APInt(BitWidth, RemVal);
    return;
  }

  // Get some size facts about the dividend and divisor
  unsigned lhsWords = getNumWords(LHS.getActiveBits());
  unsigned rhsBits  = RHS.getActiveBits();
  unsigned rhsWords = getNumWords(rhsBits);
  assert(rhsWords && "Performing divrem operation by zero ???");

  // Check the degenerate cases
  if (lhsWords == 0) {
    Quotient = APInt(BitWidth, 0);    // 0 / Y ===> 0
    Remainder = APInt(BitWidth, 0);   // 0 % Y ===> 0
    return;
  }

  if (rhsBits == 1) {
    Quotient = LHS;                   // X / 1 ===> X
    Remainder = APInt(BitWidth, 0);   // X % 1 ===> 0
  }

  if (lhsWords < rhsWords || LHS.ult(RHS)) {
    Remainder = LHS;                  // X % Y ===> X, iff X < Y
    Quotient = APInt(BitWidth, 0);    // X / Y ===> 0, iff X < Y
    return;
  }

  if (LHS == RHS) {
    Quotient  = APInt(BitWidth, 1);   // X / X ===> 1
    Remainder = APInt(BitWidth, 0);   // X % X ===> 0;
    return;
  }

  // Make sure there is enough space to hold the results.
  // NOTE: This assumes that reallocate won't affect any bits if it doesn't
  // change the size. This is necessary if Quotient or Remainder is aliased
  // with LHS or RHS.
  Quotient.reallocate(BitWidth);
  Remainder.reallocate(BitWidth);

  if (lhsWords == 1) { // rhsWords is 1 if lhsWords is 1.
    // There is only one word to consider so use the native versions.
    uint64_t lhsValue = LHS.U.pVal[0];
    uint64_t rhsValue = RHS.U.pVal[0];
    Quotient = lhsValue / rhsValue;
    Remainder = lhsValue % rhsValue;
    return;
  }

  // Okay, lets do it the long way
  divide(LHS.U.pVal, lhsWords, RHS.U.pVal, rhsWords, Quotient.U.pVal,
         Remainder.U.pVal);
  // Clear the rest of the Quotient and Remainder.
  std::memset(Quotient.U.pVal + lhsWords, 0,
              (getNumWords(BitWidth) - lhsWords) * APINT_WORD_SIZE);
  std::memset(Remainder.U.pVal + rhsWords, 0,
              (getNumWords(BitWidth) - rhsWords) * APINT_WORD_SIZE);
}

void APInt::udivrem(const APInt &LHS, uint64_t RHS, APInt &Quotient,
                    uint64_t &Remainder) {
  assert(RHS != 0 && "Divide by zero?");
  unsigned BitWidth = LHS.BitWidth;

  // First, deal with the easy case
  if (LHS.isSingleWord()) {
    uint64_t QuotVal = LHS.U.VAL / RHS;
    Remainder = LHS.U.VAL % RHS;
    Quotient = APInt(BitWidth, QuotVal);
    return;
  }

  // Get some size facts about the dividend and divisor
  unsigned lhsWords = getNumWords(LHS.getActiveBits());

  // Check the degenerate cases
  if (lhsWords == 0) {
    Quotient = APInt(BitWidth, 0);    // 0 / Y ===> 0
    Remainder = 0;                    // 0 % Y ===> 0
    return;
  }

  if (RHS == 1) {
    Quotient = LHS;                   // X / 1 ===> X
    Remainder = 0;                    // X % 1 ===> 0
    return;
  }

  if (LHS.ult(RHS)) {
    Remainder = LHS.getZExtValue();   // X % Y ===> X, iff X < Y
    Quotient = APInt(BitWidth, 0);    // X / Y ===> 0, iff X < Y
    return;
  }

  if (LHS == RHS) {
    Quotient  = APInt(BitWidth, 1);   // X / X ===> 1
    Remainder = 0;                    // X % X ===> 0;
    return;
  }

  // Make sure there is enough space to hold the results.
  // NOTE: This assumes that reallocate won't affect any bits if it doesn't
  // change the size. This is necessary if Quotient is aliased with LHS.
  Quotient.reallocate(BitWidth);

  if (lhsWords == 1) { // rhsWords is 1 if lhsWords is 1.
    // There is only one word to consider so use the native versions.
    uint64_t lhsValue = LHS.U.pVal[0];
    Quotient = lhsValue / RHS;
    Remainder = lhsValue % RHS;
    return;
  }

  // Okay, lets do it the long way
  divide(LHS.U.pVal, lhsWords, &RHS, 1, Quotient.U.pVal, &Remainder);
  // Clear the rest of the Quotient.
  std::memset(Quotient.U.pVal + lhsWords, 0,
              (getNumWords(BitWidth) - lhsWords) * APINT_WORD_SIZE);
}

void APInt::sdivrem(const APInt &LHS, const APInt &RHS,
                    APInt &Quotient, APInt &Remainder) {
  if (LHS.isNegative()) {
    if (RHS.isNegative())
      APInt::udivrem(-LHS, -RHS, Quotient, Remainder);
    else {
      APInt::udivrem(-LHS, RHS, Quotient, Remainder);
      Quotient.negate();
    }
    Remainder.negate();
  } else if (RHS.isNegative()) {
    APInt::udivrem(LHS, -RHS, Quotient, Remainder);
    Quotient.negate();
  } else {
    APInt::udivrem(LHS, RHS, Quotient, Remainder);
  }
}

void APInt::sdivrem(const APInt &LHS, int64_t RHS,
                    APInt &Quotient, int64_t &Remainder) {
  uint64_t R = Remainder;
  if (LHS.isNegative()) {
    if (RHS < 0)
      APInt::udivrem(-LHS, -RHS, Quotient, R);
    else {
      APInt::udivrem(-LHS, RHS, Quotient, R);
      Quotient.negate();
    }
    R = -R;
  } else if (RHS < 0) {
    APInt::udivrem(LHS, -RHS, Quotient, R);
    Quotient.negate();
  } else {
    APInt::udivrem(LHS, RHS, Quotient, R);
  }
  Remainder = R;
}

APInt APInt::sadd_ov(const APInt &RHS, bool &Overflow) const {
  APInt Res = *this+RHS;
  Overflow = isNonNegative() == RHS.isNonNegative() &&
             Res.isNonNegative() != isNonNegative();
  return Res;
}

APInt APInt::uadd_ov(const APInt &RHS, bool &Overflow) const {
  APInt Res = *this+RHS;
  Overflow = Res.ult(RHS);
  return Res;
}

APInt APInt::ssub_ov(const APInt &RHS, bool &Overflow) const {
  APInt Res = *this - RHS;
  Overflow = isNonNegative() != RHS.isNonNegative() &&
             Res.isNonNegative() != isNonNegative();
  return Res;
}

APInt APInt::usub_ov(const APInt &RHS, bool &Overflow) const {
  APInt Res = *this-RHS;
  Overflow = Res.ugt(*this);
  return Res;
}

APInt APInt::sdiv_ov(const APInt &RHS, bool &Overflow) const {
  // MININT/-1  -->  overflow.
  Overflow = isMinSignedValue() && RHS.isAllOnes();
  return sdiv(RHS);
}

APInt APInt::smul_ov(const APInt &RHS, bool &Overflow) const {
  APInt Res = *this * RHS;

  if (RHS != 0)
    Overflow = Res.sdiv(RHS) != *this ||
               (isMinSignedValue() && RHS.isAllOnes());
  else
    Overflow = false;
  return Res;
}

APInt APInt::umul_ov(const APInt &RHS, bool &Overflow) const {
  if (countLeadingZeros() + RHS.countLeadingZeros() + 2 <= BitWidth) {
    Overflow = true;
    return *this * RHS;
  }

  APInt Res = lshr(1) * RHS;
  Overflow = Res.isNegative();
  Res <<= 1;
  if ((*this)[0]) {
    Res += RHS;
    if (Res.ult(RHS))
      Overflow = true;
  }
  return Res;
}

APInt APInt::sshl_ov(const APInt &ShAmt, bool &Overflow) const {
  Overflow = ShAmt.uge(getBitWidth());
  if (Overflow)
    return APInt(BitWidth, 0);

  if (isNonNegative()) // Don't allow sign change.
    Overflow = ShAmt.uge(countLeadingZeros());
  else
    Overflow = ShAmt.uge(countLeadingOnes());

  return *this << ShAmt;
}

APInt APInt::ushl_ov(const APInt &ShAmt, bool &Overflow) const {
  Overflow = ShAmt.uge(getBitWidth());
  if (Overflow)
    return APInt(BitWidth, 0);

  Overflow = ShAmt.ugt(countLeadingZeros());

  return *this << ShAmt;
}

APInt APInt::sadd_sat(const APInt &RHS) const {
  bool Overflow;
  APInt Res = sadd_ov(RHS, Overflow);
  if (!Overflow)
    return Res;

  return isNegative() ? APInt::getSignedMinValue(BitWidth)
                      : APInt::getSignedMaxValue(BitWidth);
}

APInt APInt::uadd_sat(const APInt &RHS) const {
  bool Overflow;
  APInt Res = uadd_ov(RHS, Overflow);
  if (!Overflow)
    return Res;

  return APInt::getMaxValue(BitWidth);
}

APInt APInt::ssub_sat(const APInt &RHS) const {
  bool Overflow;
  APInt Res = ssub_ov(RHS, Overflow);
  if (!Overflow)
    return Res;

  return isNegative() ? APInt::getSignedMinValue(BitWidth)
                      : APInt::getSignedMaxValue(BitWidth);
}

APInt APInt::usub_sat(const APInt &RHS) const {
  bool Overflow;
  APInt Res = usub_ov(RHS, Overflow);
  if (!Overflow)
    return Res;

  return APInt(BitWidth, 0);
}

APInt APInt::smul_sat(const APInt &RHS) const {
  bool Overflow;
  APInt Res = smul_ov(RHS, Overflow);
  if (!Overflow)
    return Res;

  // The result is negative if one and only one of inputs is negative.
  bool ResIsNegative = isNegative() ^ RHS.isNegative();

  return ResIsNegative ? APInt::getSignedMinValue(BitWidth)
                       : APInt::getSignedMaxValue(BitWidth);
}

APInt APInt::umul_sat(const APInt &RHS) const {
  bool Overflow;
  APInt Res = umul_ov(RHS, Overflow);
  if (!Overflow)
    return Res;

  return APInt::getMaxValue(BitWidth);
}

APInt APInt::sshl_sat(const APInt &RHS) const {
  bool Overflow;
  APInt Res = sshl_ov(RHS, Overflow);
  if (!Overflow)
    return Res;

  return isNegative() ? APInt::getSignedMinValue(BitWidth)
                      : APInt::getSignedMaxValue(BitWidth);
}

APInt APInt::ushl_sat(const APInt &RHS) const {
  bool Overflow;
  APInt Res = ushl_ov(RHS, Overflow);
  if (!Overflow)
    return Res;

  return APInt::getMaxValue(BitWidth);
}

void APInt::fromString(unsigned numbits, StringRef str, uint8_t radix) {
  // Check our assumptions here
  assert(!str.empty() && "Invalid string length");
  assert((radix == 10 || radix == 8 || radix == 16 || radix == 2 ||
          radix == 36) &&
         "Radix should be 2, 8, 10, 16, or 36!");

  StringRef::iterator p = str.begin();
  size_t slen = str.size();
  bool isNeg = *p == '-';
  if (*p == '-' || *p == '+') {
    p++;
    slen--;
    assert(slen && "String is only a sign, needs a value.");
  }
  assert((slen <= numbits || radix != 2) && "Insufficient bit width");
  assert(((slen-1)*3 <= numbits || radix != 8) && "Insufficient bit width");
  assert(((slen-1)*4 <= numbits || radix != 16) && "Insufficient bit width");
  assert((((slen-1)*64)/22 <= numbits || radix != 10) &&
         "Insufficient bit width");

  // Allocate memory if needed
  if (isSingleWord())
    U.VAL = 0;
  else
    U.pVal = getClearedMemory(getNumWords());

  // Figure out if we can shift instead of multiply
  unsigned shift = (radix == 16 ? 4 : radix == 8 ? 3 : radix == 2 ? 1 : 0);

  // Enter digit traversal loop
  for (StringRef::iterator e = str.end(); p != e; ++p) {
    unsigned digit = getDigit(*p, radix);
    assert(digit < radix && "Invalid character in digit string");

    // Shift or multiply the value by the radix
    if (slen > 1) {
      if (shift)
        *this <<= shift;
      else
        *this *= radix;
    }

    // Add in the digit we just interpreted
    *this += digit;
  }
  // If its negative, put it in two's complement form
  if (isNeg)
    this->negate();
}

void APInt::toString(SmallVectorImpl<char> &Str, unsigned Radix,
                     bool Signed, bool formatAsCLiteral) const {
  assert((Radix == 10 || Radix == 8 || Radix == 16 || Radix == 2 ||
          Radix == 36) &&
         "Radix should be 2, 8, 10, 16, or 36!");

  const char *Prefix = "";
  if (formatAsCLiteral) {
    switch (Radix) {
      case 2:
        // Binary literals are a non-standard extension added in gcc 4.3:
        // http://gcc.gnu.org/onlinedocs/gcc-4.3.0/gcc/Binary-constants.html
        Prefix = "0b";
        break;
      case 8:
        Prefix = "0";
        break;
      case 10:
        break; // No prefix
      case 16:
        Prefix = "0x";
        break;
      default:
        llvm_unreachable("Invalid radix!");
    }
  }

  // First, check for a zero value and just short circuit the logic below.
  if (isZero()) {
    while (*Prefix) {
      Str.push_back(*Prefix);
      ++Prefix;
    };
    Str.push_back('0');
    return;
  }

  static const char Digits[] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ";

  if (isSingleWord()) {
    char Buffer[65];
    char *BufPtr = std::end(Buffer);

    uint64_t N;
    if (!Signed) {
      N = getZExtValue();
    } else {
      int64_t I = getSExtValue();
      if (I >= 0) {
        N = I;
      } else {
        Str.push_back('-');
        N = -(uint64_t)I;
      }
    }

    while (*Prefix) {
      Str.push_back(*Prefix);
      ++Prefix;
    };

    while (N) {
      *--BufPtr = Digits[N % Radix];
      N /= Radix;
    }
    Str.append(BufPtr, std::end(Buffer));
    return;
  }

  APInt Tmp(*this);

  if (Signed && isNegative()) {
    // They want to print the signed version and it is a negative value
    // Flip the bits and add one to turn it into the equivalent positive
    // value and put a '-' in the result.
    Tmp.negate();
    Str.push_back('-');
  }

  while (*Prefix) {
    Str.push_back(*Prefix);
    ++Prefix;
  };

  // We insert the digits backward, then reverse them to get the right order.
  unsigned StartDig = Str.size();

  // For the 2, 8 and 16 bit cases, we can just shift instead of divide
  // because the number of bits per digit (1, 3 and 4 respectively) divides
  // equally.  We just shift until the value is zero.
  if (Radix == 2 || Radix == 8 || Radix == 16) {
    // Just shift tmp right for each digit width until it becomes zero
    unsigned ShiftAmt = (Radix == 16 ? 4 : (Radix == 8 ? 3 : 1));
    unsigned MaskAmt = Radix - 1;

    while (Tmp.getBoolValue()) {
      unsigned Digit = unsigned(Tmp.getRawData()[0]) & MaskAmt;
      Str.push_back(Digits[Digit]);
      Tmp.lshrInPlace(ShiftAmt);
    }
  } else {
    while (Tmp.getBoolValue()) {
      uint64_t Digit;
      udivrem(Tmp, Radix, Tmp, Digit);
      assert(Digit < Radix && "divide failed");
      Str.push_back(Digits[Digit]);
    }
  }

  // Reverse the digits before returning.
  std::reverse(Str.begin()+StartDig, Str.end());
}

#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
LLVM_DUMP_METHOD void APInt::dump() const {
  SmallString<40> S, U;
  this->toStringUnsigned(U);
  this->toStringSigned(S);
  dbgs() << "APInt(" << BitWidth << "b, "
         << U << "u " << S << "s)\n";
}
#endif

void APInt::print(raw_ostream &OS, bool isSigned) const {
  SmallString<40> S;
  this->toString(S, 10, isSigned, /* formatAsCLiteral = */false);
  OS << S;
}

// This implements a variety of operations on a representation of
// arbitrary precision, two's-complement, bignum integer values.

// Assumed by lowHalf, highHalf, partMSB and partLSB.  A fairly safe
// and unrestricting assumption.
static_assert(APInt::APINT_BITS_PER_WORD % 2 == 0,
              "Part width must be divisible by 2!");

// Returns the integer part with the least significant BITS set.
// BITS cannot be zero.
static inline APInt::WordType lowBitMask(unsigned bits) {
  assert(bits != 0 && bits <= APInt::APINT_BITS_PER_WORD);
  return ~(APInt::WordType) 0 >> (APInt::APINT_BITS_PER_WORD - bits);
}

/// Returns the value of the lower half of PART.
static inline APInt::WordType lowHalf(APInt::WordType part) {
  return part & lowBitMask(APInt::APINT_BITS_PER_WORD / 2);
}

/// Returns the value of the upper half of PART.
static inline APInt::WordType highHalf(APInt::WordType part) {
  return part >> (APInt::APINT_BITS_PER_WORD / 2);
}

/// Returns the bit number of the most significant set bit of a part.
/// If the input number has no bits set -1U is returned.
static unsigned partMSB(APInt::WordType value) { return findLastSet(value); }

/// Returns the bit number of the least significant set bit of a part.  If the
/// input number has no bits set -1U is returned.
static unsigned partLSB(APInt::WordType value) { return findFirstSet(value); }

/// Sets the least significant part of a bignum to the input value, and zeroes
/// out higher parts.
void APInt::tcSet(WordType *dst, WordType part, unsigned parts) {
  assert(parts > 0);
  dst[0] = part;
  for (unsigned i = 1; i < parts; i++)
    dst[i] = 0;
}

/// Assign one bignum to another.
void APInt::tcAssign(WordType *dst, const WordType *src, unsigned parts) {
  for (unsigned i = 0; i < parts; i++)
    dst[i] = src[i];
}

/// Returns true if a bignum is zero, false otherwise.
bool APInt::tcIsZero(const WordType *src, unsigned parts) {
  for (unsigned i = 0; i < parts; i++)
    if (src[i])
      return false;

  return true;
}

/// Extract the given bit of a bignum; returns 0 or 1.
int APInt::tcExtractBit(const WordType *parts, unsigned bit) {
  return (parts[whichWord(bit)] & maskBit(bit)) != 0;
}

/// Set the given bit of a bignum.
void APInt::tcSetBit(WordType *parts, unsigned bit) {
  parts[whichWord(bit)] |= maskBit(bit);
}

/// Clears the given bit of a bignum.
void APInt::tcClearBit(WordType *parts, unsigned bit) {
  parts[whichWord(bit)] &= ~maskBit(bit);
}

/// Returns the bit number of the least significant set bit of a number.  If the
/// input number has no bits set -1U is returned.
unsigned APInt::tcLSB(const WordType *parts, unsigned n) {
  for (unsigned i = 0; i < n; i++) {
    if (parts[i] != 0) {
      unsigned lsb = partLSB(parts[i]);
      return lsb + i * APINT_BITS_PER_WORD;
    }
  }

  return -1U;
}

/// Returns the bit number of the most significant set bit of a number.
/// If the input number has no bits set -1U is returned.
unsigned APInt::tcMSB(const WordType *parts, unsigned n) {
  do {
    --n;

    if (parts[n] != 0) {
      unsigned msb = partMSB(parts[n]);

      return msb + n * APINT_BITS_PER_WORD;
    }
  } while (n);

  return -1U;
}

/// Copy the bit vector of width srcBITS from SRC, starting at bit srcLSB, to
/// DST, of dstCOUNT parts, such that the bit srcLSB becomes the least
/// significant bit of DST.  All high bits above srcBITS in DST are zero-filled.
/// */
void
APInt::tcExtract(WordType *dst, unsigned dstCount, const WordType *src,
                 unsigned srcBits, unsigned srcLSB) {
  unsigned dstParts = (srcBits + APINT_BITS_PER_WORD - 1) / APINT_BITS_PER_WORD;
  assert(dstParts <= dstCount);

  unsigned firstSrcPart = srcLSB / APINT_BITS_PER_WORD;
  tcAssign(dst, src + firstSrcPart, dstParts);

  unsigned shift = srcLSB % APINT_BITS_PER_WORD;
  tcShiftRight(dst, dstParts, shift);

  // We now have (dstParts * APINT_BITS_PER_WORD - shift) bits from SRC
  // in DST.  If this is less that srcBits, append the rest, else
  // clear the high bits.
  unsigned n = dstParts * APINT_BITS_PER_WORD - shift;
  if (n < srcBits) {
    WordType mask = lowBitMask (srcBits - n);
    dst[dstParts - 1] |= ((src[firstSrcPart + dstParts] & mask)
                          << n % APINT_BITS_PER_WORD);
  } else if (n > srcBits) {
    if (srcBits % APINT_BITS_PER_WORD)
      dst[dstParts - 1] &= lowBitMask (srcBits % APINT_BITS_PER_WORD);
  }

  // Clear high parts.
  while (dstParts < dstCount)
    dst[dstParts++] = 0;
}

//// DST += RHS + C where C is zero or one.  Returns the carry flag.
APInt::WordType APInt::tcAdd(WordType *dst, const WordType *rhs,
                             WordType c, unsigned parts) {
  assert(c <= 1);

  for (unsigned i = 0; i < parts; i++) {
    WordType l = dst[i];
    if (c) {
      dst[i] += rhs[i] + 1;
      c = (dst[i] <= l);
    } else {
      dst[i] += rhs[i];
      c = (dst[i] < l);
    }
  }

  return c;
}

/// This function adds a single "word" integer, src, to the multiple
/// "word" integer array, dst[]. dst[] is modified to reflect the addition and
/// 1 is returned if there is a carry out, otherwise 0 is returned.
/// @returns the carry of the addition.
APInt::WordType APInt::tcAddPart(WordType *dst, WordType src,
                                 unsigned parts) {
  for (unsigned i = 0; i < parts; ++i) {
    dst[i] += src;
    if (dst[i] >= src)
      return 0; // No need to carry so exit early.
    src = 1; // Carry one to next digit.
  }

  return 1;
}

/// DST -= RHS + C where C is zero or one.  Returns the carry flag.
APInt::WordType APInt::tcSubtract(WordType *dst, const WordType *rhs,
                                  WordType c, unsigned parts) {
  assert(c <= 1);

  for (unsigned i = 0; i < parts; i++) {
    WordType l = dst[i];
    if (c) {
      dst[i] -= rhs[i] + 1;
      c = (dst[i] >= l);
    } else {
      dst[i] -= rhs[i];
      c = (dst[i] > l);
    }
  }

  return c;
}

/// This function subtracts a single "word" (64-bit word), src, from
/// the multi-word integer array, dst[], propagating the borrowed 1 value until
/// no further borrowing is needed or it runs out of "words" in dst.  The result
/// is 1 if "borrowing" exhausted the digits in dst, or 0 if dst was not
/// exhausted. In other words, if src > dst then this function returns 1,
/// otherwise 0.
/// @returns the borrow out of the subtraction
APInt::WordType APInt::tcSubtractPart(WordType *dst, WordType src,
                                      unsigned parts) {
  for (unsigned i = 0; i < parts; ++i) {
    WordType Dst = dst[i];
    dst[i] -= src;
    if (src <= Dst)
      return 0; // No need to borrow so exit early.
    src = 1; // We have to "borrow 1" from next "word"
  }

  return 1;
}

/// Negate a bignum in-place.
void APInt::tcNegate(WordType *dst, unsigned parts) {
  tcComplement(dst, parts);
  tcIncrement(dst, parts);
}

/// DST += SRC * MULTIPLIER + CARRY   if add is true
/// DST  = SRC * MULTIPLIER + CARRY   if add is false
/// Requires 0 <= DSTPARTS <= SRCPARTS + 1.  If DST overlaps SRC
/// they must start at the same point, i.e. DST == SRC.
/// If DSTPARTS == SRCPARTS + 1 no overflow occurs and zero is
/// returned.  Otherwise DST is filled with the least significant
/// DSTPARTS parts of the result, and if all of the omitted higher
/// parts were zero return zero, otherwise overflow occurred and
/// return one.
int APInt::tcMultiplyPart(WordType *dst, const WordType *src,
                          WordType multiplier, WordType carry,
                          unsigned srcParts, unsigned dstParts,
                          bool add) {
  // Otherwise our writes of DST kill our later reads of SRC.
  assert(dst <= src || dst >= src + srcParts);
  assert(dstParts <= srcParts + 1);

  // N loops; minimum of dstParts and srcParts.
  unsigned n = std::min(dstParts, srcParts);

  for (unsigned i = 0; i < n; i++) {
    // [LOW, HIGH] = MULTIPLIER * SRC[i] + DST[i] + CARRY.
    // This cannot overflow, because:
    //   (n - 1) * (n - 1) + 2 (n - 1) = (n - 1) * (n + 1)
    // which is less than n^2.
    WordType srcPart = src[i];
    WordType low, mid, high;
    if (multiplier == 0 || srcPart == 0) {
      low = carry;
      high = 0;
    } else {
      low = lowHalf(srcPart) * lowHalf(multiplier);
      high = highHalf(srcPart) * highHalf(multiplier);

      mid = lowHalf(srcPart) * highHalf(multiplier);
      high += highHalf(mid);
      mid <<= APINT_BITS_PER_WORD / 2;
      if (low + mid < low)
        high++;
      low += mid;

      mid = highHalf(srcPart) * lowHalf(multiplier);
      high += highHalf(mid);
      mid <<= APINT_BITS_PER_WORD / 2;
      if (low + mid < low)
        high++;
      low += mid;

      // Now add carry.
      if (low + carry < low)
        high++;
      low += carry;
    }

    if (add) {
      // And now DST[i], and store the new low part there.
      if (low + dst[i] < low)
        high++;
      dst[i] += low;
    } else
      dst[i] = low;

    carry = high;
  }

  if (srcParts < dstParts) {
    // Full multiplication, there is no overflow.
    assert(srcParts + 1 == dstParts);
    dst[srcParts] = carry;
    return 0;
  }

  // We overflowed if there is carry.
  if (carry)
    return 1;

  // We would overflow if any significant unwritten parts would be
  // non-zero.  This is true if any remaining src parts are non-zero
  // and the multiplier is non-zero.
  if (multiplier)
    for (unsigned i = dstParts; i < srcParts; i++)
      if (src[i])
        return 1;

  // We fitted in the narrow destination.
  return 0;
}

/// DST = LHS * RHS, where DST has the same width as the operands and
/// is filled with the least significant parts of the result.  Returns
/// one if overflow occurred, otherwise zero.  DST must be disjoint
/// from both operands.
int APInt::tcMultiply(WordType *dst, const WordType *lhs,
                      const WordType *rhs, unsigned parts) {
  assert(dst != lhs && dst != rhs);

  int overflow = 0;
  tcSet(dst, 0, parts);

  for (unsigned i = 0; i < parts; i++)
    overflow |= tcMultiplyPart(&dst[i], lhs, rhs[i], 0, parts,
                               parts - i, true);

  return overflow;
}

/// DST = LHS * RHS, where DST has width the sum of the widths of the
/// operands. No overflow occurs. DST must be disjoint from both operands.
void APInt::tcFullMultiply(WordType *dst, const WordType *lhs,
                           const WordType *rhs, unsigned lhsParts,
                           unsigned rhsParts) {
  // Put the narrower number on the LHS for less loops below.
  if (lhsParts > rhsParts)
    return tcFullMultiply (dst, rhs, lhs, rhsParts, lhsParts);

  assert(dst != lhs && dst != rhs);

  tcSet(dst, 0, rhsParts);

  for (unsigned i = 0; i < lhsParts; i++)
    tcMultiplyPart(&dst[i], rhs, lhs[i], 0, rhsParts, rhsParts + 1, true);
}

// If RHS is zero LHS and REMAINDER are left unchanged, return one.
// Otherwise set LHS to LHS / RHS with the fractional part discarded,
// set REMAINDER to the remainder, return zero.  i.e.
//
//   OLD_LHS = RHS * LHS + REMAINDER
//
// SCRATCH is a bignum of the same size as the operands and result for
// use by the routine; its contents need not be initialized and are
// destroyed.  LHS, REMAINDER and SCRATCH must be distinct.
int APInt::tcDivide(WordType *lhs, const WordType *rhs,
                    WordType *remainder, WordType *srhs,
                    unsigned parts) {
  assert(lhs != remainder && lhs != srhs && remainder != srhs);

  unsigned shiftCount = tcMSB(rhs, parts) + 1;
  if (shiftCount == 0)
    return true;

  shiftCount = parts * APINT_BITS_PER_WORD - shiftCount;
  unsigned n = shiftCount / APINT_BITS_PER_WORD;
  WordType mask = (WordType) 1 << (shiftCount % APINT_BITS_PER_WORD);

  tcAssign(srhs, rhs, parts);
  tcShiftLeft(srhs, parts, shiftCount);
  tcAssign(remainder, lhs, parts);
  tcSet(lhs, 0, parts);

  // Loop, subtracting SRHS if REMAINDER is greater and adding that to the
  // total.
  for (;;) {
    int compare = tcCompare(remainder, srhs, parts);
    if (compare >= 0) {
      tcSubtract(remainder, srhs, 0, parts);
      lhs[n] |= mask;
    }

    if (shiftCount == 0)
      break;
    shiftCount--;
    tcShiftRight(srhs, parts, 1);
    if ((mask >>= 1) == 0) {
      mask = (WordType) 1 << (APINT_BITS_PER_WORD - 1);
      n--;
    }
  }

  return false;
}

/// Shift a bignum left Cound bits in-place. Shifted in bits are zero. There are
/// no restrictions on Count.
void APInt::tcShiftLeft(WordType *Dst, unsigned Words, unsigned Count) {
  // Don't bother performing a no-op shift.
  if (!Count)
    return;

  // WordShift is the inter-part shift; BitShift is the intra-part shift.
  unsigned WordShift = std::min(Count / APINT_BITS_PER_WORD, Words);
  unsigned BitShift = Count % APINT_BITS_PER_WORD;

  // Fastpath for moving by whole words.
  if (BitShift == 0) {
    std::memmove(Dst + WordShift, Dst, (Words - WordShift) * APINT_WORD_SIZE);
  } else {
    while (Words-- > WordShift) {
      Dst[Words] = Dst[Words - WordShift] << BitShift;
      if (Words > WordShift)
        Dst[Words] |=
          Dst[Words - WordShift - 1] >> (APINT_BITS_PER_WORD - BitShift);
    }
  }

  // Fill in the remainder with 0s.
  std::memset(Dst, 0, WordShift * APINT_WORD_SIZE);
}

/// Shift a bignum right Count bits in-place. Shifted in bits are zero. There
/// are no restrictions on Count.
void APInt::tcShiftRight(WordType *Dst, unsigned Words, unsigned Count) {
  // Don't bother performing a no-op shift.
  if (!Count)
    return;

  // WordShift is the inter-part shift; BitShift is the intra-part shift.
  unsigned WordShift = std::min(Count / APINT_BITS_PER_WORD, Words);
  unsigned BitShift = Count % APINT_BITS_PER_WORD;

  unsigned WordsToMove = Words - WordShift;
  // Fastpath for moving by whole words.
  if (BitShift == 0) {
    std::memmove(Dst, Dst + WordShift, WordsToMove * APINT_WORD_SIZE);
  } else {
    for (unsigned i = 0; i != WordsToMove; ++i) {
      Dst[i] = Dst[i + WordShift] >> BitShift;
      if (i + 1 != WordsToMove)
        Dst[i] |= Dst[i + WordShift + 1] << (APINT_BITS_PER_WORD - BitShift);
    }
  }

  // Fill in the remainder with 0s.
  std::memset(Dst + WordsToMove, 0, WordShift * APINT_WORD_SIZE);
}

// Comparison (unsigned) of two bignums.
int APInt::tcCompare(const WordType *lhs, const WordType *rhs,
                     unsigned parts) {
  while (parts) {
    parts--;
    if (lhs[parts] != rhs[parts])
      return (lhs[parts] > rhs[parts]) ? 1 : -1;
  }

  return 0;
}

APInt llvm::APIntOps::RoundingUDiv(const APInt &A, const APInt &B,
                                   APInt::Rounding RM) {
  // Currently udivrem always rounds down.
  switch (RM) {
  case APInt::Rounding::DOWN:
  case APInt::Rounding::TOWARD_ZERO:
    return A.udiv(B);
  case APInt::Rounding::UP: {
    APInt Quo, Rem;
    APInt::udivrem(A, B, Quo, Rem);
    if (Rem.isZero())
      return Quo;
    return Quo + 1;
  }
  }
  llvm_unreachable("Unknown APInt::Rounding enum");
}

APInt llvm::APIntOps::RoundingSDiv(const APInt &A, const APInt &B,
                                   APInt::Rounding RM) {
  switch (RM) {
  case APInt::Rounding::DOWN:
  case APInt::Rounding::UP: {
    APInt Quo, Rem;
    APInt::sdivrem(A, B, Quo, Rem);
    if (Rem.isZero())
      return Quo;
    // This algorithm deals with arbitrary rounding mode used by sdivrem.
    // We want to check whether the non-integer part of the mathematical value
    // is negative or not. If the non-integer part is negative, we need to round
    // down from Quo; otherwise, if it's positive or 0, we return Quo, as it's
    // already rounded down.
    if (RM == APInt::Rounding::DOWN) {
      if (Rem.isNegative() != B.isNegative())
        return Quo - 1;
      return Quo;
    }
    if (Rem.isNegative() != B.isNegative())
      return Quo;
    return Quo + 1;
  }
  // Currently sdiv rounds towards zero.
  case APInt::Rounding::TOWARD_ZERO:
    return A.sdiv(B);
  }
  llvm_unreachable("Unknown APInt::Rounding enum");
}

std::optional<APInt>
llvm::APIntOps::SolveQuadraticEquationWrap(APInt A, APInt B, APInt C,
                                           unsigned RangeWidth) {
  unsigned CoeffWidth = A.getBitWidth();
  assert(CoeffWidth == B.getBitWidth() && CoeffWidth == C.getBitWidth());
  assert(RangeWidth <= CoeffWidth &&
         "Value range width should be less than coefficient width");
  assert(RangeWidth > 1 && "Value range bit width should be > 1");

  LLVM_DEBUG(dbgs() << __func__ << ": solving " << A << "x^2 + " << B
                    << "x + " << C << ", rw:" << RangeWidth << '\n');

  // Identify 0 as a (non)solution immediately.
  if (C.sextOrTrunc(RangeWidth).isZero()) {
    LLVM_DEBUG(dbgs() << __func__ << ": zero solution\n");
    return APInt(CoeffWidth, 0);
  }

  // The result of APInt arithmetic has the same bit width as the operands,
  // so it can actually lose high bits. A product of two n-bit integers needs
  // 2n-1 bits to represent the full value.
  // The operation done below (on quadratic coefficients) that can produce
  // the largest value is the evaluation of the equation during bisection,
  // which needs 3 times the bitwidth of the coefficient, so the total number
  // of required bits is 3n.
  //
  // The purpose of this extension is to simulate the set Z of all integers,
  // where n+1 > n for all n in Z. In Z it makes sense to talk about positive
  // and negative numbers (not so much in a modulo arithmetic). The method
  // used to solve the equation is based on the standard formula for real
  // numbers, and uses the concepts of "positive" and "negative" with their
  // usual meanings.
  CoeffWidth *= 3;
  A = A.sext(CoeffWidth);
  B = B.sext(CoeffWidth);
  C = C.sext(CoeffWidth);

  // Make A > 0 for simplicity. Negate cannot overflow at this point because
  // the bit width has increased.
  if (A.isNegative()) {
    A.negate();
    B.negate();
    C.negate();
  }

  // Solving an equation q(x) = 0 with coefficients in modular arithmetic
  // is really solving a set of equations q(x) = kR for k = 0, 1, 2, ...,
  // and R = 2^BitWidth.
  // Since we're trying not only to find exact solutions, but also values
  // that "wrap around", such a set will always have a solution, i.e. an x
  // that satisfies at least one of the equations, or such that |q(x)|
  // exceeds kR, while |q(x-1)| for the same k does not.
  //
  // We need to find a value k, such that Ax^2 + Bx + C = kR will have a
  // positive solution n (in the above sense), and also such that the n
  // will be the least among all solutions corresponding to k = 0, 1, ...
  // (more precisely, the least element in the set
  //   { n(k) | k is such that a solution n(k) exists }).
  //
  // Consider the parabola (over real numbers) that corresponds to the
  // quadratic equation. Since A > 0, the arms of the parabola will point
  // up. Picking different values of k will shift it up and down by R.
  //
  // We want to shift the parabola in such a way as to reduce the problem
  // of solving q(x) = kR to solving shifted_q(x) = 0.
  // (The interesting solutions are the ceilings of the real number
  // solutions.)
  APInt R = APInt::getOneBitSet(CoeffWidth, RangeWidth);
  APInt TwoA = 2 * A;
  APInt SqrB = B * B;
  bool PickLow;

  auto RoundUp = [] (const APInt &V, const APInt &A) -> APInt {
    assert(A.isStrictlyPositive());
    APInt T = V.abs().urem(A);
    if (T.isZero())
      return V;
    return V.isNegative() ? V+T : V+(A-T);
  };

  // The vertex of the parabola is at -B/2A, but since A > 0, it's negative
  // iff B is positive.
  if (B.isNonNegative()) {
    // If B >= 0, the vertex it at a negative location (or at 0), so in
    // order to have a non-negative solution we need to pick k that makes
    // C-kR negative. To satisfy all the requirements for the solution
    // that we are looking for, it needs to be closest to 0 of all k.
    C = C.srem(R);
    if (C.isStrictlyPositive())
      C -= R;
    // Pick the greater solution.
    PickLow = false;
  } else {
    // If B < 0, the vertex is at a positive location. For any solution
    // to exist, the discriminant must be non-negative. This means that
    // C-kR <= B^2/4A is a necessary condition for k, i.e. there is a
    // lower bound on values of k: kR >= C - B^2/4A.
    APInt LowkR = C - SqrB.udiv(2*TwoA); // udiv because all values > 0.
    // Round LowkR up (towards +inf) to the nearest kR.
    LowkR = RoundUp(LowkR, R);

    // If there exists k meeting the condition above, and such that
    // C-kR > 0, there will be two positive real number solutions of
    // q(x) = kR. Out of all such values of k, pick the one that makes
    // C-kR closest to 0, (i.e. pick maximum k such that C-kR > 0).
    // In other words, find maximum k such that LowkR <= kR < C.
    if (C.sgt(LowkR)) {
      // If LowkR < C, then such a k is guaranteed to exist because
      // LowkR itself is a multiple of R.
      C -= -RoundUp(-C, R);      // C = C - RoundDown(C, R)
      // Pick the smaller solution.
      PickLow = true;
    } else {
      // If C-kR < 0 for all potential k's, it means that one solution
      // will be negative, while the other will be positive. The positive
      // solution will shift towards 0 if the parabola is moved up.
      // Pick the kR closest to the lower bound (i.e. make C-kR closest
      // to 0, or in other words, out of all parabolas that have solutions,
      // pick the one that is the farthest "up").
      // Since LowkR is itself a multiple of R, simply take C-LowkR.
      C -= LowkR;
      // Pick the greater solution.
      PickLow = false;
    }
  }

  LLVM_DEBUG(dbgs() << __func__ << ": updated coefficients " << A << "x^2 + "
                    << B << "x + " << C << ", rw:" << RangeWidth << '\n');

  APInt D = SqrB - 4*A*C;
  assert(D.isNonNegative() && "Negative discriminant");
  APInt SQ = D.sqrt();

  APInt Q = SQ * SQ;
  bool InexactSQ = Q != D;
  // The calculated SQ may actually be greater than the exact (non-integer)
  // value. If that's the case, decrement SQ to get a value that is lower.
  if (Q.sgt(D))
    SQ -= 1;

  APInt X;
  APInt Rem;

  // SQ is rounded down (i.e SQ * SQ <= D), so the roots may be inexact.
  // When using the quadratic formula directly, the calculated low root
  // may be greater than the exact one, since we would be subtracting SQ.
  // To make sure that the calculated root is not greater than the exact
  // one, subtract SQ+1 when calculating the low root (for inexact value
  // of SQ).
  if (PickLow)
    APInt::sdivrem(-B - (SQ+InexactSQ), TwoA, X, Rem);
  else
    APInt::sdivrem(-B + SQ, TwoA, X, Rem);

  // The updated coefficients should be such that the (exact) solution is
  // positive. Since APInt division rounds towards 0, the calculated one
  // can be 0, but cannot be negative.
  assert(X.isNonNegative() && "Solution should be non-negative");

  if (!InexactSQ && Rem.isZero()) {
    LLVM_DEBUG(dbgs() << __func__ << ": solution (root): " << X << '\n');
    return X;
  }

  assert((SQ*SQ).sle(D) && "SQ = |_sqrt(D)_|, so SQ*SQ <= D");
  // The exact value of the square root of D should be between SQ and SQ+1.
  // This implies that the solution should be between that corresponding to
  // SQ (i.e. X) and that corresponding to SQ+1.
  //
  // The calculated X cannot be greater than the exact (real) solution.
  // Actually it must be strictly less than the exact solution, while
  // X+1 will be greater than or equal to it.

  APInt VX = (A*X + B)*X + C;
  APInt VY = VX + TwoA*X + A + B;
  bool SignChange =
      VX.isNegative() != VY.isNegative() || VX.isZero() != VY.isZero();
  // If the sign did not change between X and X+1, X is not a valid solution.
  // This could happen when the actual (exact) roots don't have an integer
  // between them, so they would both be contained between X and X+1.
  if (!SignChange) {
    LLVM_DEBUG(dbgs() << __func__ << ": no valid solution\n");
    return std::nullopt;
  }

  X += 1;
  LLVM_DEBUG(dbgs() << __func__ << ": solution (wrap): " << X << '\n');
  return X;
}

std::optional<unsigned>
llvm::APIntOps::GetMostSignificantDifferentBit(const APInt &A, const APInt &B) {
  assert(A.getBitWidth() == B.getBitWidth() && "Must have the same bitwidth");
  if (A == B)
    return std::nullopt;
  return A.getBitWidth() - ((A ^ B).countLeadingZeros() + 1);
}

APInt llvm::APIntOps::ScaleBitMask(const APInt &A, unsigned NewBitWidth,
                                   bool MatchAllBits) {
  unsigned OldBitWidth = A.getBitWidth();
  assert((((OldBitWidth % NewBitWidth) == 0) ||
          ((NewBitWidth % OldBitWidth) == 0)) &&
         "One size should be a multiple of the other one. "
         "Can't do fractional scaling.");

  // Check for matching bitwidths.
  if (OldBitWidth == NewBitWidth)
    return A;

  APInt NewA = APInt::getZero(NewBitWidth);

  // Check for null input.
  if (A.isZero())
    return NewA;

  if (NewBitWidth > OldBitWidth) {
    // Repeat bits.
    unsigned Scale = NewBitWidth / OldBitWidth;
    for (unsigned i = 0; i != OldBitWidth; ++i)
      if (A[i])
        NewA.setBits(i * Scale, (i + 1) * Scale);
  } else {
    unsigned Scale = OldBitWidth / NewBitWidth;
    for (unsigned i = 0; i != NewBitWidth; ++i) {
      if (MatchAllBits) {
        if (A.extractBits(Scale, i * Scale).isAllOnes())
          NewA.setBit(i);
      } else {
        if (!A.extractBits(Scale, i * Scale).isZero())
          NewA.setBit(i);
      }
    }
  }

  return NewA;
}

/// StoreIntToMemory - Fills the StoreBytes bytes of memory starting from Dst
/// with the integer held in IntVal.
void llvm::StoreIntToMemory(const APInt &IntVal, uint8_t *Dst,
                            unsigned StoreBytes) {
  assert((IntVal.getBitWidth()+7)/8 >= StoreBytes && "Integer too small!");
  const uint8_t *Src = (const uint8_t *)IntVal.getRawData();

  if (sys::IsLittleEndianHost) {
    // Little-endian host - the source is ordered from LSB to MSB.  Order the
    // destination from LSB to MSB: Do a straight copy.
    memcpy(Dst, Src, StoreBytes);
  } else {
    // Big-endian host - the source is an array of 64 bit words ordered from
    // LSW to MSW.  Each word is ordered from MSB to LSB.  Order the destination
    // from MSB to LSB: Reverse the word order, but not the bytes in a word.
    while (StoreBytes > sizeof(uint64_t)) {
      StoreBytes -= sizeof(uint64_t);
      // May not be aligned so use memcpy.
      memcpy(Dst + StoreBytes, Src, sizeof(uint64_t));
      Src += sizeof(uint64_t);
    }

    memcpy(Dst, Src + sizeof(uint64_t) - StoreBytes, StoreBytes);
  }
}

/// LoadIntFromMemory - Loads the integer stored in the LoadBytes bytes starting
/// from Src into IntVal, which is assumed to be wide enough and to hold zero.
void llvm::LoadIntFromMemory(APInt &IntVal, const uint8_t *Src,
                             unsigned LoadBytes) {
  assert((IntVal.getBitWidth()+7)/8 >= LoadBytes && "Integer too small!");
  uint8_t *Dst = reinterpret_cast<uint8_t *>(
                   const_cast<uint64_t *>(IntVal.getRawData()));

  if (sys::IsLittleEndianHost)
    // Little-endian host - the destination must be ordered from LSB to MSB.
    // The source is ordered from LSB to MSB: Do a straight copy.
    memcpy(Dst, Src, LoadBytes);
  else {
    // Big-endian - the destination is an array of 64 bit words ordered from
    // LSW to MSW.  Each word must be ordered from MSB to LSB.  The source is
    // ordered from MSB to LSB: Reverse the word order, but not the bytes in
    // a word.
    while (LoadBytes > sizeof(uint64_t)) {
      LoadBytes -= sizeof(uint64_t);
      // May not be aligned so use memcpy.
      memcpy(Dst, Src + LoadBytes, sizeof(uint64_t));
      Dst += sizeof(uint64_t);
    }

    memcpy(Dst + sizeof(uint64_t) - LoadBytes, Src, LoadBytes);
  }
}