aboutsummaryrefslogtreecommitdiff
path: root/sys/arm/arm/pmap-v6.c
blob: f596ace35ba78e4d201fd33ffadd050bc2a4ef66 (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
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
5322
5323
5324
5325
5326
5327
5328
5329
5330
5331
5332
5333
5334
5335
5336
5337
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
5349
5350
5351
5352
5353
5354
5355
5356
5357
5358
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
5377
5378
5379
5380
5381
5382
5383
5384
5385
5386
5387
5388
5389
5390
5391
5392
5393
5394
5395
5396
5397
5398
5399
5400
5401
5402
5403
5404
5405
5406
5407
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
5538
5539
5540
5541
5542
5543
5544
5545
5546
5547
5548
5549
5550
5551
5552
5553
5554
5555
5556
5557
5558
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
5599
5600
5601
5602
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
5636
5637
5638
5639
5640
5641
5642
5643
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
5659
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
5682
5683
5684
5685
5686
5687
5688
5689
5690
5691
5692
5693
5694
5695
5696
5697
5698
5699
5700
5701
5702
5703
5704
5705
5706
5707
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
5723
5724
5725
5726
5727
5728
5729
5730
5731
5732
5733
5734
5735
5736
5737
5738
5739
5740
5741
5742
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
5758
5759
5760
5761
5762
5763
5764
5765
5766
5767
5768
5769
5770
5771
5772
5773
5774
5775
5776
5777
5778
5779
5780
5781
5782
5783
5784
5785
5786
5787
5788
5789
5790
5791
5792
5793
5794
5795
5796
5797
5798
5799
5800
5801
5802
5803
5804
5805
5806
5807
5808
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
5834
5835
5836
5837
5838
5839
5840
5841
5842
5843
5844
5845
5846
5847
5848
5849
5850
5851
5852
5853
5854
5855
5856
5857
5858
5859
5860
5861
5862
5863
5864
5865
5866
5867
5868
5869
5870
5871
5872
5873
5874
5875
5876
5877
5878
5879
5880
5881
5882
5883
5884
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
5900
5901
5902
5903
5904
5905
5906
5907
5908
5909
5910
5911
5912
5913
5914
5915
5916
5917
5918
5919
5920
5921
5922
5923
5924
5925
5926
5927
5928
5929
5930
5931
5932
5933
5934
5935
5936
5937
5938
5939
5940
5941
5942
5943
5944
5945
5946
5947
5948
5949
5950
5951
5952
5953
5954
5955
5956
5957
5958
5959
5960
5961
5962
5963
5964
5965
5966
5967
5968
5969
5970
5971
5972
5973
5974
5975
5976
5977
5978
5979
5980
5981
5982
5983
5984
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
5996
5997
5998
5999
6000
6001
6002
6003
6004
6005
6006
6007
6008
6009
6010
6011
6012
6013
6014
6015
6016
6017
6018
6019
6020
6021
6022
6023
6024
6025
6026
6027
6028
6029
6030
6031
6032
6033
6034
6035
6036
6037
6038
6039
6040
6041
6042
6043
6044
6045
6046
6047
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
6060
6061
6062
6063
6064
6065
6066
6067
6068
6069
6070
6071
6072
6073
6074
6075
6076
6077
6078
6079
6080
6081
6082
6083
6084
6085
6086
6087
6088
6089
6090
6091
6092
6093
6094
6095
6096
6097
6098
6099
6100
6101
6102
6103
6104
6105
6106
6107
6108
6109
6110
6111
6112
6113
6114
6115
6116
6117
6118
6119
6120
6121
6122
6123
6124
6125
6126
6127
6128
6129
6130
6131
6132
6133
6134
6135
6136
6137
6138
6139
6140
6141
6142
6143
6144
6145
6146
6147
6148
6149
6150
6151
6152
6153
6154
6155
6156
6157
6158
6159
6160
6161
6162
6163
6164
6165
6166
6167
6168
6169
6170
6171
6172
6173
6174
6175
6176
6177
6178
6179
6180
6181
6182
6183
6184
6185
6186
6187
6188
6189
6190
6191
6192
6193
6194
6195
6196
6197
6198
6199
6200
6201
6202
6203
6204
6205
6206
6207
6208
6209
6210
6211
6212
6213
6214
6215
6216
6217
6218
6219
6220
6221
6222
6223
6224
6225
6226
6227
6228
6229
6230
6231
6232
6233
6234
6235
6236
6237
6238
6239
6240
6241
6242
6243
6244
6245
6246
6247
6248
6249
6250
6251
6252
6253
6254
6255
6256
6257
6258
6259
6260
6261
6262
6263
6264
6265
6266
6267
6268
6269
6270
6271
6272
6273
6274
6275
6276
6277
6278
6279
6280
6281
6282
6283
6284
6285
6286
6287
6288
6289
6290
6291
6292
6293
6294
6295
6296
6297
6298
6299
6300
6301
6302
6303
6304
6305
6306
6307
6308
6309
6310
6311
6312
6313
6314
6315
6316
6317
6318
6319
6320
6321
6322
6323
6324
6325
6326
6327
6328
6329
6330
6331
6332
6333
6334
6335
6336
6337
6338
6339
6340
6341
6342
6343
6344
6345
6346
6347
6348
6349
6350
6351
6352
6353
6354
6355
6356
6357
6358
6359
6360
6361
6362
6363
6364
6365
6366
6367
6368
6369
6370
6371
6372
6373
6374
6375
6376
6377
6378
6379
6380
6381
6382
6383
6384
6385
6386
6387
6388
6389
6390
6391
6392
6393
6394
6395
6396
6397
6398
6399
6400
6401
6402
6403
6404
6405
6406
6407
6408
6409
6410
6411
6412
6413
6414
6415
6416
6417
6418
6419
6420
6421
6422
6423
6424
6425
6426
6427
6428
6429
6430
6431
6432
6433
6434
6435
6436
6437
6438
6439
6440
6441
6442
6443
6444
6445
6446
6447
6448
6449
6450
6451
6452
6453
6454
6455
6456
6457
6458
6459
6460
6461
6462
6463
6464
6465
6466
6467
6468
6469
6470
6471
6472
6473
6474
6475
6476
6477
6478
6479
6480
6481
6482
6483
6484
6485
6486
6487
6488
6489
6490
6491
6492
6493
6494
6495
6496
6497
6498
6499
6500
6501
6502
6503
6504
6505
6506
6507
6508
6509
6510
6511
6512
6513
6514
6515
6516
6517
6518
6519
6520
6521
6522
6523
6524
6525
6526
6527
6528
6529
6530
6531
6532
6533
6534
6535
6536
6537
6538
6539
6540
6541
6542
6543
6544
6545
6546
6547
6548
6549
6550
6551
6552
6553
6554
6555
6556
6557
6558
6559
6560
6561
6562
6563
6564
6565
6566
6567
6568
6569
6570
6571
6572
6573
6574
6575
6576
6577
6578
6579
6580
6581
6582
6583
6584
6585
6586
6587
6588
6589
6590
6591
6592
6593
6594
6595
6596
6597
6598
6599
6600
6601
6602
6603
6604
6605
6606
6607
6608
6609
6610
6611
6612
6613
6614
6615
6616
6617
6618
6619
6620
6621
6622
6623
6624
6625
6626
6627
6628
6629
6630
6631
6632
6633
6634
6635
6636
6637
6638
6639
6640
6641
6642
6643
6644
6645
6646
6647
6648
6649
6650
6651
6652
6653
6654
6655
6656
6657
6658
6659
6660
6661
6662
6663
6664
6665
6666
6667
6668
6669
6670
6671
6672
6673
6674
6675
6676
6677
6678
6679
6680
6681
6682
6683
6684
6685
6686
6687
6688
6689
6690
6691
6692
6693
6694
6695
6696
6697
6698
6699
6700
6701
6702
6703
6704
6705
6706
6707
6708
6709
6710
6711
6712
6713
6714
6715
6716
6717
6718
6719
6720
6721
6722
6723
6724
6725
6726
6727
6728
6729
6730
6731
6732
6733
6734
6735
6736
6737
6738
6739
6740
6741
6742
6743
6744
6745
6746
6747
6748
6749
6750
6751
6752
6753
6754
6755
6756
6757
6758
6759
6760
6761
6762
6763
6764
6765
6766
6767
6768
6769
6770
6771
6772
6773
6774
6775
6776
6777
6778
6779
6780
6781
6782
6783
6784
6785
6786
6787
6788
6789
6790
6791
6792
6793
6794
6795
6796
6797
6798
6799
6800
6801
6802
6803
6804
6805
6806
6807
6808
6809
6810
6811
6812
6813
6814
6815
6816
6817
6818
6819
6820
6821
6822
6823
6824
6825
6826
6827
6828
6829
6830
6831
6832
6833
6834
6835
6836
6837
6838
6839
6840
6841
6842
6843
6844
6845
6846
6847
6848
6849
6850
6851
6852
6853
6854
6855
6856
6857
6858
6859
6860
6861
6862
6863
6864
6865
6866
6867
6868
6869
6870
6871
6872
6873
6874
6875
6876
6877
6878
6879
6880
6881
6882
6883
6884
6885
6886
6887
6888
6889
6890
6891
6892
6893
6894
6895
6896
6897
6898
6899
6900
6901
6902
6903
6904
6905
6906
6907
6908
6909
6910
6911
6912
6913
6914
6915
6916
6917
6918
6919
6920
6921
6922
6923
6924
6925
6926
6927
6928
6929
6930
6931
6932
6933
6934
6935
6936
6937
6938
6939
6940
6941
6942
6943
6944
6945
6946
6947
6948
6949
6950
6951
6952
6953
6954
6955
/*-
 * SPDX-License-Identifier: BSD-3-Clause AND BSD-2-Clause-FreeBSD
 *
 * Copyright (c) 1991 Regents of the University of California.
 * Copyright (c) 1994 John S. Dyson
 * Copyright (c) 1994 David Greenman
 * Copyright (c) 2005-2010 Alan L. Cox <alc@cs.rice.edu>
 * Copyright (c) 2014-2016 Svatopluk Kraus <skra@FreeBSD.org>
 * Copyright (c) 2014-2016 Michal Meloun <mmel@FreeBSD.org>
 * All rights reserved.
 *
 * This code is derived from software contributed to Berkeley by
 * the Systems Programming Group of the University of Utah Computer
 * Science Department and William Jolitz of UUNET Technologies Inc.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *	from:	@(#)pmap.c	7.7 (Berkeley)	5/12/91
 */
/*-
 * Copyright (c) 2003 Networks Associates Technology, Inc.
 * All rights reserved.
 *
 * This software was developed for the FreeBSD Project by Jake Burkholder,
 * Safeport Network Services, and Network Associates Laboratories, the
 * Security Research Division of Network Associates, Inc. under
 * DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the DARPA
 * CHATS research program.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */

#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");

/*
 *	Manages physical address maps.
 *
 *	Since the information managed by this module is
 *	also stored by the logical address mapping module,
 *	this module may throw away valid virtual-to-physical
 *	mappings at almost any time.  However, invalidations
 *	of virtual-to-physical mappings must be done as
 *	requested.
 *
 *	In order to cope with hardware architectures which
 *	make virtual-to-physical map invalidates expensive,
 *	this module may delay invalidate or reduced protection
 *	operations until such time as they are actually
 *	necessary.  This module is given full information as
 *	to which processors are currently using which maps,
 *	and to when physical maps must be made correct.
 */

#include "opt_vm.h"
#include "opt_pmap.h"
#include "opt_ddb.h"

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/ktr.h>
#include <sys/lock.h>
#include <sys/proc.h>
#include <sys/rwlock.h>
#include <sys/malloc.h>
#include <sys/vmmeter.h>
#include <sys/malloc.h>
#include <sys/mman.h>
#include <sys/sf_buf.h>
#include <sys/smp.h>
#include <sys/sched.h>
#include <sys/sysctl.h>

#ifdef DDB
#include <ddb/ddb.h>
#endif

#include <vm/vm.h>
#include <vm/uma.h>
#include <vm/pmap.h>
#include <vm/vm_param.h>
#include <vm/vm_kern.h>
#include <vm/vm_object.h>
#include <vm/vm_map.h>
#include <vm/vm_page.h>
#include <vm/vm_pageout.h>
#include <vm/vm_phys.h>
#include <vm/vm_extern.h>
#include <vm/vm_reserv.h>
#include <sys/lock.h>
#include <sys/mutex.h>

#include <machine/md_var.h>
#include <machine/pmap_var.h>
#include <machine/cpu.h>
#include <machine/pcb.h>
#include <machine/sf_buf.h>
#ifdef SMP
#include <machine/smp.h>
#endif
#ifndef PMAP_SHPGPERPROC
#define PMAP_SHPGPERPROC 200
#endif

#ifndef DIAGNOSTIC
#define PMAP_INLINE	__inline
#else
#define PMAP_INLINE
#endif

#ifdef PMAP_DEBUG
static void pmap_zero_page_check(vm_page_t m);
void pmap_debug(int level);
int pmap_pid_dump(int pid);

#define PDEBUG(_lev_,_stat_) \
	if (pmap_debug_level >= (_lev_)) \
		((_stat_))
#define dprintf printf
int pmap_debug_level = 1;
#else   /* PMAP_DEBUG */
#define PDEBUG(_lev_,_stat_) /* Nothing */
#define dprintf(x, arg...)
#endif  /* PMAP_DEBUG */

/*
 *  Level 2 page tables map definion ('max' is excluded).
 */

#define PT2V_MIN_ADDRESS	((vm_offset_t)PT2MAP)
#define PT2V_MAX_ADDRESS	((vm_offset_t)PT2MAP + PT2MAP_SIZE)

#define UPT2V_MIN_ADDRESS	((vm_offset_t)PT2MAP)
#define UPT2V_MAX_ADDRESS \
    ((vm_offset_t)(PT2MAP + (KERNBASE >> PT2MAP_SHIFT)))

/*
 *  Promotion to a 1MB (PTE1) page mapping requires that the corresponding
 *  4KB (PTE2) page mappings have identical settings for the following fields:
 */
#define PTE2_PROMOTE	(PTE2_V | PTE2_A | PTE2_NM | PTE2_S | PTE2_NG |	\
			 PTE2_NX | PTE2_RO | PTE2_U | PTE2_W |		\
			 PTE2_ATTR_MASK)

#define PTE1_PROMOTE	(PTE1_V | PTE1_A | PTE1_NM | PTE1_S | PTE1_NG |	\
			 PTE1_NX | PTE1_RO | PTE1_U | PTE1_W |		\
			 PTE1_ATTR_MASK)

#define ATTR_TO_L1(l2_attr)	((((l2_attr) & L2_TEX0) ? L1_S_TEX0 : 0) | \
				 (((l2_attr) & L2_C)    ? L1_S_C    : 0) | \
				 (((l2_attr) & L2_B)    ? L1_S_B    : 0) | \
				 (((l2_attr) & PTE2_A)  ? PTE1_A    : 0) | \
				 (((l2_attr) & PTE2_NM) ? PTE1_NM   : 0) | \
				 (((l2_attr) & PTE2_S)  ? PTE1_S    : 0) | \
				 (((l2_attr) & PTE2_NG) ? PTE1_NG   : 0) | \
				 (((l2_attr) & PTE2_NX) ? PTE1_NX   : 0) | \
				 (((l2_attr) & PTE2_RO) ? PTE1_RO   : 0) | \
				 (((l2_attr) & PTE2_U)  ? PTE1_U    : 0) | \
				 (((l2_attr) & PTE2_W)  ? PTE1_W    : 0))

#define ATTR_TO_L2(l1_attr)	((((l1_attr) & L1_S_TEX0) ? L2_TEX0 : 0) | \
				 (((l1_attr) & L1_S_C)    ? L2_C    : 0) | \
				 (((l1_attr) & L1_S_B)    ? L2_B    : 0) | \
				 (((l1_attr) & PTE1_A)    ? PTE2_A  : 0) | \
				 (((l1_attr) & PTE1_NM)   ? PTE2_NM : 0) | \
				 (((l1_attr) & PTE1_S)    ? PTE2_S  : 0) | \
				 (((l1_attr) & PTE1_NG)   ? PTE2_NG : 0) | \
				 (((l1_attr) & PTE1_NX)   ? PTE2_NX : 0) | \
				 (((l1_attr) & PTE1_RO)   ? PTE2_RO : 0) | \
				 (((l1_attr) & PTE1_U)    ? PTE2_U  : 0) | \
				 (((l1_attr) & PTE1_W)    ? PTE2_W  : 0))

/*
 *  PTE2 descriptors creation macros.
 */
#define PTE2_ATTR_DEFAULT	vm_memattr_to_pte2(VM_MEMATTR_DEFAULT)
#define PTE2_ATTR_PT		vm_memattr_to_pte2(pt_memattr)

#define PTE2_KPT(pa)	PTE2_KERN(pa, PTE2_AP_KRW, PTE2_ATTR_PT)
#define PTE2_KPT_NG(pa)	PTE2_KERN_NG(pa, PTE2_AP_KRW, PTE2_ATTR_PT)

#define PTE2_KRW(pa)	PTE2_KERN(pa, PTE2_AP_KRW, PTE2_ATTR_DEFAULT)
#define PTE2_KRO(pa)	PTE2_KERN(pa, PTE2_AP_KR, PTE2_ATTR_DEFAULT)

#define PV_STATS
#ifdef PV_STATS
#define PV_STAT(x)	do { x ; } while (0)
#else
#define PV_STAT(x)	do { } while (0)
#endif

/*
 *  The boot_pt1 is used temporary in very early boot stage as L1 page table.
 *  We can init many things with no memory allocation thanks to its static
 *  allocation and this brings two main advantages:
 *  (1) other cores can be started very simply,
 *  (2) various boot loaders can be supported as its arguments can be processed
 *      in virtual address space and can be moved to safe location before
 *      first allocation happened.
 *  Only disadvantage is that boot_pt1 is used only in very early boot stage.
 *  However, the table is uninitialized and so lays in bss. Therefore kernel
 *  image size is not influenced.
 *
 *  QQQ: In the future, maybe, boot_pt1 can be used for soft reset and
 *       CPU suspend/resume game.
 */
extern pt1_entry_t boot_pt1[];

vm_paddr_t base_pt1;
pt1_entry_t *kern_pt1;
pt2_entry_t *kern_pt2tab;
pt2_entry_t *PT2MAP;

static uint32_t ttb_flags;
static vm_memattr_t pt_memattr;
ttb_entry_t pmap_kern_ttb;

struct pmap kernel_pmap_store;
LIST_HEAD(pmaplist, pmap);
static struct pmaplist allpmaps;
static struct mtx allpmaps_lock;

vm_offset_t virtual_avail;	/* VA of first avail page (after kernel bss) */
vm_offset_t virtual_end;	/* VA of last avail page (end of kernel AS) */

static vm_offset_t kernel_vm_end_new;
vm_offset_t kernel_vm_end = KERNBASE + NKPT2PG * NPT2_IN_PG * PTE1_SIZE;
vm_offset_t vm_max_kernel_address;
vm_paddr_t kernel_l1pa;

static struct rwlock __aligned(CACHE_LINE_SIZE) pvh_global_lock;

/*
 *  Data for the pv entry allocation mechanism
 */
static TAILQ_HEAD(pch, pv_chunk) pv_chunks = TAILQ_HEAD_INITIALIZER(pv_chunks);
static int pv_entry_count = 0, pv_entry_max = 0, pv_entry_high_water = 0;
static struct md_page *pv_table; /* XXX: Is it used only the list in md_page? */
static int shpgperproc = PMAP_SHPGPERPROC;

struct pv_chunk *pv_chunkbase;		/* KVA block for pv_chunks */
int pv_maxchunks;			/* How many chunks we have KVA for */
vm_offset_t pv_vafree;			/* freelist stored in the PTE */

vm_paddr_t first_managed_pa;
#define	pa_to_pvh(pa)	(&pv_table[pte1_index(pa - first_managed_pa)])

/*
 *  All those kernel PT submaps that BSD is so fond of
 */
caddr_t _tmppt = 0;

/*
 *  Crashdump maps.
 */
static caddr_t crashdumpmap;

static pt2_entry_t *PMAP1 = NULL, *PMAP2;
static pt2_entry_t *PADDR1 = NULL, *PADDR2;
#ifdef DDB
static pt2_entry_t *PMAP3;
static pt2_entry_t *PADDR3;
static int PMAP3cpu __unused; /* for SMP only */
#endif
#ifdef SMP
static int PMAP1cpu;
static int PMAP1changedcpu;
SYSCTL_INT(_debug, OID_AUTO, PMAP1changedcpu, CTLFLAG_RD,
    &PMAP1changedcpu, 0,
    "Number of times pmap_pte2_quick changed CPU with same PMAP1");
#endif
static int PMAP1changed;
SYSCTL_INT(_debug, OID_AUTO, PMAP1changed, CTLFLAG_RD,
    &PMAP1changed, 0,
    "Number of times pmap_pte2_quick changed PMAP1");
static int PMAP1unchanged;
SYSCTL_INT(_debug, OID_AUTO, PMAP1unchanged, CTLFLAG_RD,
    &PMAP1unchanged, 0,
    "Number of times pmap_pte2_quick didn't change PMAP1");
static struct mtx PMAP2mutex;

/*
 * Internal flags for pmap_enter()'s helper functions.
 */
#define	PMAP_ENTER_NORECLAIM	0x1000000	/* Don't reclaim PV entries. */
#define	PMAP_ENTER_NOREPLACE	0x2000000	/* Don't replace mappings. */

static __inline void pt2_wirecount_init(vm_page_t m);
static boolean_t pmap_demote_pte1(pmap_t pmap, pt1_entry_t *pte1p,
    vm_offset_t va);
static int pmap_enter_pte1(pmap_t pmap, vm_offset_t va, pt1_entry_t pte1,
    u_int flags, vm_page_t m);
void cache_icache_sync_fresh(vm_offset_t va, vm_paddr_t pa, vm_size_t size);

/*
 *  Function to set the debug level of the pmap code.
 */
#ifdef PMAP_DEBUG
void
pmap_debug(int level)
{

	pmap_debug_level = level;
	dprintf("pmap_debug: level=%d\n", pmap_debug_level);
}
#endif /* PMAP_DEBUG */

/*
 *  This table must corespond with memory attribute configuration in vm.h.
 *  First entry is used for normal system mapping.
 *
 *  Device memory is always marked as shared.
 *  Normal memory is shared only in SMP .
 *  Not outer shareable bits are not used yet.
 *  Class 6 cannot be used on ARM11.
 */
#define TEXDEF_TYPE_SHIFT	0
#define TEXDEF_TYPE_MASK	0x3
#define TEXDEF_INNER_SHIFT	2
#define TEXDEF_INNER_MASK	0x3
#define TEXDEF_OUTER_SHIFT	4
#define TEXDEF_OUTER_MASK	0x3
#define TEXDEF_NOS_SHIFT	6
#define TEXDEF_NOS_MASK		0x1

#define TEX(t, i, o, s) 			\
		((t) << TEXDEF_TYPE_SHIFT) |	\
		((i) << TEXDEF_INNER_SHIFT) |	\
		((o) << TEXDEF_OUTER_SHIFT | 	\
		((s) << TEXDEF_NOS_SHIFT))

static uint32_t tex_class[8] = {
/*	    type      inner cache outer cache */
	TEX(PRRR_MEM, NMRR_WB_WA, NMRR_WB_WA, 0),  /* 0 - ATTR_WB_WA	*/
	TEX(PRRR_MEM, NMRR_NC,	  NMRR_NC,    0),  /* 1 - ATTR_NOCACHE	*/
	TEX(PRRR_DEV, NMRR_NC,	  NMRR_NC,    0),  /* 2 - ATTR_DEVICE	*/
	TEX(PRRR_SO,  NMRR_NC,	  NMRR_NC,    0),  /* 3 - ATTR_SO	*/
	TEX(PRRR_MEM, NMRR_WT,	  NMRR_WT,    0),  /* 4 - ATTR_WT	*/
	TEX(PRRR_MEM, NMRR_NC,	  NMRR_NC,    0),  /* 5 - NOT USED YET	*/
	TEX(PRRR_MEM, NMRR_NC,	  NMRR_NC,    0),  /* 6 - NOT USED YET	*/
	TEX(PRRR_MEM, NMRR_NC,	  NMRR_NC,    0),  /* 7 - NOT USED YET	*/
};
#undef TEX

static uint32_t pte2_attr_tab[8] = {
	PTE2_ATTR_WB_WA,	/* 0 - VM_MEMATTR_WB_WA */
	PTE2_ATTR_NOCACHE,	/* 1 - VM_MEMATTR_NOCACHE */
	PTE2_ATTR_DEVICE,	/* 2 - VM_MEMATTR_DEVICE */
	PTE2_ATTR_SO,		/* 3 - VM_MEMATTR_SO */
	PTE2_ATTR_WT,		/* 4 - VM_MEMATTR_WRITE_THROUGH */
	0,			/* 5 - NOT USED YET */
	0,			/* 6 - NOT USED YET */
	0			/* 7 - NOT USED YET */
};
CTASSERT(VM_MEMATTR_WB_WA == 0);
CTASSERT(VM_MEMATTR_NOCACHE == 1);
CTASSERT(VM_MEMATTR_DEVICE == 2);
CTASSERT(VM_MEMATTR_SO == 3);
CTASSERT(VM_MEMATTR_WRITE_THROUGH == 4);
#define	VM_MEMATTR_END	(VM_MEMATTR_WRITE_THROUGH + 1)

boolean_t
pmap_is_valid_memattr(pmap_t pmap __unused, vm_memattr_t mode)
{

	return (mode >= 0 && mode < VM_MEMATTR_END);
}

static inline uint32_t
vm_memattr_to_pte2(vm_memattr_t ma)
{

	KASSERT((u_int)ma < VM_MEMATTR_END,
	    ("%s: bad vm_memattr_t %d", __func__, ma));
	return (pte2_attr_tab[(u_int)ma]);
}

static inline uint32_t
vm_page_pte2_attr(vm_page_t m)
{

	return (vm_memattr_to_pte2(m->md.pat_mode));
}

/*
 * Convert TEX definition entry to TTB flags.
 */
static uint32_t
encode_ttb_flags(int idx)
{
	uint32_t inner, outer, nos, reg;

	inner = (tex_class[idx] >> TEXDEF_INNER_SHIFT) &
		TEXDEF_INNER_MASK;
	outer = (tex_class[idx] >> TEXDEF_OUTER_SHIFT) &
		TEXDEF_OUTER_MASK;
	nos = (tex_class[idx] >> TEXDEF_NOS_SHIFT) &
		TEXDEF_NOS_MASK;

	reg = nos << 5;
	reg |= outer << 3;
	if (cpuinfo.coherent_walk)
		reg |= (inner & 0x1) << 6;
	reg |= (inner & 0x2) >> 1;
#ifdef SMP
	ARM_SMP_UP(
		reg |= 1 << 1,
	);
#endif
	return reg;
}

/*
 *  Set TEX remapping registers in current CPU.
 */
void
pmap_set_tex(void)
{
	uint32_t prrr, nmrr;
	uint32_t type, inner, outer, nos;
	int i;

#ifdef PMAP_PTE_NOCACHE
	/* XXX fixme */
	if (cpuinfo.coherent_walk) {
		pt_memattr = VM_MEMATTR_WB_WA;
		ttb_flags = encode_ttb_flags(0);
	}
	else {
		pt_memattr = VM_MEMATTR_NOCACHE;
		ttb_flags = encode_ttb_flags(1);
	}
#else
	pt_memattr = VM_MEMATTR_WB_WA;
	ttb_flags = encode_ttb_flags(0);
#endif

	prrr = 0;
	nmrr = 0;

	/* Build remapping register from TEX classes. */
	for (i = 0; i < 8; i++) {
		type = (tex_class[i] >> TEXDEF_TYPE_SHIFT) &
			TEXDEF_TYPE_MASK;
		inner = (tex_class[i] >> TEXDEF_INNER_SHIFT) &
			TEXDEF_INNER_MASK;
		outer = (tex_class[i] >> TEXDEF_OUTER_SHIFT) &
			TEXDEF_OUTER_MASK;
		nos = (tex_class[i] >> TEXDEF_NOS_SHIFT) &
			TEXDEF_NOS_MASK;

		prrr |= type  << (i * 2);
		prrr |= nos   << (i + 24);
		nmrr |= inner << (i * 2);
		nmrr |= outer << (i * 2 + 16);
	}
	/* Add shareable bits for device memory. */
	prrr |= PRRR_DS0 | PRRR_DS1;

	/* Add shareable bits for normal memory in SMP case. */
#ifdef SMP
	ARM_SMP_UP(
		prrr |= PRRR_NS1,
	);
#endif
	cp15_prrr_set(prrr);
	cp15_nmrr_set(nmrr);

	/* Caches are disabled, so full TLB flush should be enough. */
	tlb_flush_all_local();
}

/*
 * Remap one vm_meattr class to another one. This can be useful as
 * workaround for SOC errata, e.g. if devices must be accessed using
 * SO memory class.
 *
 * !!! Please note that this function is absolutely last resort thing.
 * It should not be used under normal circumstances. !!!
 *
 * Usage rules:
 * - it shall be called after pmap_bootstrap_prepare() and before
 *   cpu_mp_start() (thus only on boot CPU). In practice, it's expected
 *   to be called from platform_attach() or platform_late_init().
 *
 * - if remapping doesn't change caching mode, or until uncached class
 *   is remapped to any kind of cached one, then no other restriction exists.
 *
 * - if pmap_remap_vm_attr() changes caching mode, but both (original and
 *   remapped) remain cached, then caller is resposible for calling
 *   of dcache_wbinv_poc_all().
 *
 * - remapping of any kind of cached class to uncached is not permitted.
 */
void
pmap_remap_vm_attr(vm_memattr_t old_attr, vm_memattr_t new_attr)
{
	int old_idx, new_idx;

	/* Map VM memattrs to indexes to tex_class table. */
	old_idx = PTE2_ATTR2IDX(pte2_attr_tab[(int)old_attr]);
	new_idx = PTE2_ATTR2IDX(pte2_attr_tab[(int)new_attr]);

	/* Replace TEX attribute and apply it. */
	tex_class[old_idx] = tex_class[new_idx];
	pmap_set_tex();
}

/*
 * KERNBASE must be multiple of NPT2_IN_PG * PTE1_SIZE. In other words,
 * KERNBASE is mapped by first L2 page table in L2 page table page. It
 * meets same constrain due to PT2MAP being placed just under KERNBASE.
 */
CTASSERT((KERNBASE & (NPT2_IN_PG * PTE1_SIZE - 1)) == 0);
CTASSERT((KERNBASE - VM_MAXUSER_ADDRESS) >= PT2MAP_SIZE);

/*
 *  In crazy dreams, PAGE_SIZE could be a multiple of PTE2_SIZE in general.
 *  For now, anyhow, the following check must be fulfilled.
 */
CTASSERT(PAGE_SIZE == PTE2_SIZE);
/*
 *  We don't want to mess up MI code with all MMU and PMAP definitions,
 *  so some things, which depend on other ones, are defined independently.
 *  Now, it is time to check that we don't screw up something.
 */
CTASSERT(PDRSHIFT == PTE1_SHIFT);
/*
 *  Check L1 and L2 page table entries definitions consistency.
 */
CTASSERT(NB_IN_PT1 == (sizeof(pt1_entry_t) * NPTE1_IN_PT1));
CTASSERT(NB_IN_PT2 == (sizeof(pt2_entry_t) * NPTE2_IN_PT2));
/*
 *  Check L2 page tables page consistency.
 */
CTASSERT(PAGE_SIZE == (NPT2_IN_PG * NB_IN_PT2));
CTASSERT((1 << PT2PG_SHIFT) == NPT2_IN_PG);
/*
 *  Check PT2TAB consistency.
 *  PT2TAB_ENTRIES is defined as a division of NPTE1_IN_PT1 by NPT2_IN_PG.
 *  This should be done without remainder.
 */
CTASSERT(NPTE1_IN_PT1 == (PT2TAB_ENTRIES * NPT2_IN_PG));

/*
 *	A PT2MAP magic.
 *
 *  All level 2 page tables (PT2s) are mapped continuously and accordingly
 *  into PT2MAP address space. As PT2 size is less than PAGE_SIZE, this can
 *  be done only if PAGE_SIZE is a multiple of PT2 size. All PT2s in one page
 *  must be used together, but not necessary at once. The first PT2 in a page
 *  must map things on correctly aligned address and the others must follow
 *  in right order.
 */
#define NB_IN_PT2TAB	(PT2TAB_ENTRIES * sizeof(pt2_entry_t))
#define NPT2_IN_PT2TAB	(NB_IN_PT2TAB / NB_IN_PT2)
#define NPG_IN_PT2TAB	(NB_IN_PT2TAB / PAGE_SIZE)

/*
 *  Check PT2TAB consistency.
 *  NPT2_IN_PT2TAB is defined as a division of NB_IN_PT2TAB by NB_IN_PT2.
 *  NPG_IN_PT2TAB is defined as a division of NB_IN_PT2TAB by PAGE_SIZE.
 *  The both should be done without remainder.
 */
CTASSERT(NB_IN_PT2TAB == (NPT2_IN_PT2TAB * NB_IN_PT2));
CTASSERT(NB_IN_PT2TAB == (NPG_IN_PT2TAB * PAGE_SIZE));
/*
 *  The implementation was made general, however, with the assumption
 *  bellow in mind. In case of another value of NPG_IN_PT2TAB,
 *  the code should be once more rechecked.
 */
CTASSERT(NPG_IN_PT2TAB == 1);

/*
 *  Get offset of PT2 in a page
 *  associated with given PT1 index.
 */
static __inline u_int
page_pt2off(u_int pt1_idx)
{

	return ((pt1_idx & PT2PG_MASK) * NB_IN_PT2);
}

/*
 *  Get physical address of PT2
 *  associated with given PT2s page and PT1 index.
 */
static __inline vm_paddr_t
page_pt2pa(vm_paddr_t pgpa, u_int pt1_idx)
{

	return (pgpa + page_pt2off(pt1_idx));
}

/*
 *  Get first entry of PT2
 *  associated with given PT2s page and PT1 index.
 */
static __inline pt2_entry_t *
page_pt2(vm_offset_t pgva, u_int pt1_idx)
{

	return ((pt2_entry_t *)(pgva + page_pt2off(pt1_idx)));
}

/*
 *  Get virtual address of PT2s page (mapped in PT2MAP)
 *  which holds PT2 which holds entry which maps given virtual address.
 */
static __inline vm_offset_t
pt2map_pt2pg(vm_offset_t va)
{

	va &= ~(NPT2_IN_PG * PTE1_SIZE - 1);
	return ((vm_offset_t)pt2map_entry(va));
}

/*****************************************************************************
 *
 *     THREE pmap initialization milestones exist:
 *
 *  locore.S
 *    -> fundamental init (including MMU) in ASM
 *
 *  initarm()
 *    -> fundamental init continues in C
 *    -> first available physical address is known
 *
 *    pmap_bootstrap_prepare() -> FIRST PMAP MILESTONE (first epoch begins)
 *      -> basic (safe) interface for physical address allocation is made
 *      -> basic (safe) interface for virtual mapping is made
 *      -> limited not SMP coherent work is possible
 *
 *    -> more fundamental init continues in C
 *    -> locks and some more things are available
 *    -> all fundamental allocations and mappings are done
 *
 *    pmap_bootstrap() -> SECOND PMAP MILESTONE (second epoch begins)
 *      -> phys_avail[] and virtual_avail is set
 *      -> control is passed to vm subsystem
 *      -> physical and virtual address allocation are off limit
 *      -> low level mapping functions, some SMP coherent,
 *         are available, which cannot be used before vm subsystem
 *         is being inited
 *
 *  mi_startup()
 *    -> vm subsystem is being inited
 *
 *      pmap_init() -> THIRD PMAP MILESTONE (third epoch begins)
 *        -> pmap is fully inited
 *
 *****************************************************************************/

/*****************************************************************************
 *
 *	PMAP first stage initialization and utility functions
 *	for pre-bootstrap epoch.
 *
 *  After pmap_bootstrap_prepare() is called, the following functions
 *  can be used:
 *
 *  (1) strictly only for this stage functions for physical page allocations,
 *      virtual space allocations, and mappings:
 *
 *  vm_paddr_t pmap_preboot_get_pages(u_int num);
 *  void pmap_preboot_map_pages(vm_paddr_t pa, vm_offset_t va, u_int num);
 *  vm_offset_t pmap_preboot_reserve_pages(u_int num);
 *  vm_offset_t pmap_preboot_get_vpages(u_int num);
 *  void pmap_preboot_map_attr(vm_paddr_t pa, vm_offset_t va, vm_size_t size,
 *      vm_prot_t prot, vm_memattr_t attr);
 *
 *  (2) for all stages:
 *
 *  vm_paddr_t pmap_kextract(vm_offset_t va);
 *
 *  NOTE: This is not SMP coherent stage.
 *
 *****************************************************************************/

#define KERNEL_P2V(pa) \
    ((vm_offset_t)((pa) - arm_physmem_kernaddr + KERNVIRTADDR))
#define KERNEL_V2P(va) \
    ((vm_paddr_t)((va) - KERNVIRTADDR + arm_physmem_kernaddr))

static vm_paddr_t last_paddr;

/*
 *  Pre-bootstrap epoch page allocator.
 */
vm_paddr_t
pmap_preboot_get_pages(u_int num)
{
	vm_paddr_t ret;

	ret = last_paddr;
	last_paddr += num * PAGE_SIZE;

	return (ret);
}

/*
 *	The fundamental initialization of PMAP stuff.
 *
 *  Some things already happened in locore.S and some things could happen
 *  before pmap_bootstrap_prepare() is called, so let's recall what is done:
 *  1. Caches are disabled.
 *  2. We are running on virtual addresses already with 'boot_pt1'
 *     as L1 page table.
 *  3. So far, all virtual addresses can be converted to physical ones and
 *     vice versa by the following macros:
 *       KERNEL_P2V(pa) .... physical to virtual ones,
 *       KERNEL_V2P(va) .... virtual to physical ones.
 *
 *  What is done herein:
 *  1. The 'boot_pt1' is replaced by real kernel L1 page table 'kern_pt1'.
 *  2. PT2MAP magic is brought to live.
 *  3. Basic preboot functions for page allocations and mappings can be used.
 *  4. Everything is prepared for L1 cache enabling.
 *
 *  Variations:
 *  1. To use second TTB register, so kernel and users page tables will be
 *     separated. This way process forking - pmap_pinit() - could be faster,
 *     it saves physical pages and KVA per a process, and it's simple change.
 *     However, it will lead, due to hardware matter, to the following:
 *     (a) 2G space for kernel and 2G space for users.
 *     (b) 1G space for kernel in low addresses and 3G for users above it.
 *     A question is: Is the case (b) really an option? Note that case (b)
 *     does save neither physical memory and KVA.
 */
void
pmap_bootstrap_prepare(vm_paddr_t last)
{
	vm_paddr_t pt2pg_pa, pt2tab_pa, pa, size;
	vm_offset_t pt2pg_va;
	pt1_entry_t *pte1p;
	pt2_entry_t *pte2p;
	u_int i;
	uint32_t l1_attr;

	/*
	 * Now, we are going to make real kernel mapping. Note that we are
	 * already running on some mapping made in locore.S and we expect
	 * that it's large enough to ensure nofault access to physical memory
	 * allocated herein before switch.
	 *
	 * As kernel image and everything needed before are and will be mapped
	 * by section mappings, we align last physical address to PTE1_SIZE.
	 */
	last_paddr = pte1_roundup(last);

	/*
	 * Allocate and zero page(s) for kernel L1 page table.
	 *
	 * Note that it's first allocation on space which was PTE1_SIZE
	 * aligned and as such base_pt1 is aligned to NB_IN_PT1 too.
	 */
	base_pt1 = pmap_preboot_get_pages(NPG_IN_PT1);
	kern_pt1 = (pt1_entry_t *)KERNEL_P2V(base_pt1);
	bzero((void*)kern_pt1, NB_IN_PT1);
	pte1_sync_range(kern_pt1, NB_IN_PT1);

	/* Allocate and zero page(s) for kernel PT2TAB. */
	pt2tab_pa = pmap_preboot_get_pages(NPG_IN_PT2TAB);
	kern_pt2tab = (pt2_entry_t *)KERNEL_P2V(pt2tab_pa);
	bzero(kern_pt2tab, NB_IN_PT2TAB);
	pte2_sync_range(kern_pt2tab, NB_IN_PT2TAB);

	/* Allocate and zero page(s) for kernel L2 page tables. */
	pt2pg_pa = pmap_preboot_get_pages(NKPT2PG);
	pt2pg_va = KERNEL_P2V(pt2pg_pa);
	size = NKPT2PG * PAGE_SIZE;
	bzero((void*)pt2pg_va, size);
	pte2_sync_range((pt2_entry_t *)pt2pg_va, size);

	/*
	 * Add a physical memory segment (vm_phys_seg) corresponding to the
	 * preallocated pages for kernel L2 page tables so that vm_page
	 * structures representing these pages will be created. The vm_page
	 * structures are required for promotion of the corresponding kernel
	 * virtual addresses to section mappings.
	 */
	vm_phys_add_seg(pt2tab_pa, pmap_preboot_get_pages(0));

	/*
	 * Insert allocated L2 page table pages to PT2TAB and make
	 * link to all PT2s in L1 page table. See how kernel_vm_end
	 * is initialized.
	 *
	 * We play simple and safe. So every KVA will have underlaying
	 * L2 page table, even kernel image mapped by sections.
	 */
	pte2p = kern_pt2tab_entry(KERNBASE);
	for (pa = pt2pg_pa; pa < pt2pg_pa + size; pa += PTE2_SIZE)
		pt2tab_store(pte2p++, PTE2_KPT(pa));

	pte1p = kern_pte1(KERNBASE);
	for (pa = pt2pg_pa; pa < pt2pg_pa + size; pa += NB_IN_PT2)
		pte1_store(pte1p++, PTE1_LINK(pa));

	/* Make section mappings for kernel. */
	l1_attr = ATTR_TO_L1(PTE2_ATTR_DEFAULT);
	pte1p = kern_pte1(KERNBASE);
	for (pa = KERNEL_V2P(KERNBASE); pa < last; pa += PTE1_SIZE)
		pte1_store(pte1p++, PTE1_KERN(pa, PTE1_AP_KRW, l1_attr));

	/*
	 * Get free and aligned space for PT2MAP and make L1 page table links
	 * to L2 page tables held in PT2TAB.
	 *
	 * Note that pages holding PT2s are stored in PT2TAB as pt2_entry_t
	 * descriptors and PT2TAB page(s) itself is(are) used as PT2s. Thus
	 * each entry in PT2TAB maps all PT2s in a page. This implies that
	 * virtual address of PT2MAP must be aligned to NPT2_IN_PG * PTE1_SIZE.
	 */
	PT2MAP = (pt2_entry_t *)(KERNBASE - PT2MAP_SIZE);
	pte1p = kern_pte1((vm_offset_t)PT2MAP);
	for (pa = pt2tab_pa, i = 0; i < NPT2_IN_PT2TAB; i++, pa += NB_IN_PT2) {
		pte1_store(pte1p++, PTE1_LINK(pa));
	}

	/*
	 * Store PT2TAB in PT2TAB itself, i.e. self reference mapping.
	 * Each pmap will hold own PT2TAB, so the mapping should be not global.
	 */
	pte2p = kern_pt2tab_entry((vm_offset_t)PT2MAP);
	for (pa = pt2tab_pa, i = 0; i < NPG_IN_PT2TAB; i++, pa += PTE2_SIZE) {
		pt2tab_store(pte2p++, PTE2_KPT_NG(pa));
	}

	/*
	 * Choose correct L2 page table and make mappings for allocations
	 * made herein which replaces temporary locore.S mappings after a while.
	 * Note that PT2MAP cannot be used until we switch to kern_pt1.
	 *
	 * Note, that these allocations started aligned on 1M section and
	 * kernel PT1 was allocated first. Making of mappings must follow
	 * order of physical allocations as we've used KERNEL_P2V() macro
	 * for virtual addresses resolution.
	 */
	pte2p = kern_pt2tab_entry((vm_offset_t)kern_pt1);
	pt2pg_va = KERNEL_P2V(pte2_pa(pte2_load(pte2p)));

	pte2p = page_pt2(pt2pg_va, pte1_index((vm_offset_t)kern_pt1));

	/* Make mapping for kernel L1 page table. */
	for (pa = base_pt1, i = 0; i < NPG_IN_PT1; i++, pa += PTE2_SIZE)
		pte2_store(pte2p++, PTE2_KPT(pa));

	/* Make mapping for kernel PT2TAB. */
	for (pa = pt2tab_pa, i = 0; i < NPG_IN_PT2TAB; i++, pa += PTE2_SIZE)
		pte2_store(pte2p++, PTE2_KPT(pa));

	/* Finally, switch from 'boot_pt1' to 'kern_pt1'. */
	pmap_kern_ttb = base_pt1 | ttb_flags;
	cpuinfo_reinit_mmu(pmap_kern_ttb);
	/*
	 * Initialize the first available KVA. As kernel image is mapped by
	 * sections, we are leaving some gap behind.
	 */
	virtual_avail = (vm_offset_t)kern_pt2tab + NPG_IN_PT2TAB * PAGE_SIZE;
}

/*
 *  Setup L2 page table page for given KVA.
 *  Used in pre-bootstrap epoch.
 *
 *  Note that we have allocated NKPT2PG pages for L2 page tables in advance
 *  and used them for mapping KVA starting from KERNBASE. However, this is not
 *  enough. Vectors and devices need L2 page tables too. Note that they are
 *  even above VM_MAX_KERNEL_ADDRESS.
 */
static __inline vm_paddr_t
pmap_preboot_pt2pg_setup(vm_offset_t va)
{
	pt2_entry_t *pte2p, pte2;
	vm_paddr_t pt2pg_pa;

	/* Get associated entry in PT2TAB. */
	pte2p = kern_pt2tab_entry(va);

	/* Just return, if PT2s page exists already. */
	pte2 = pt2tab_load(pte2p);
	if (pte2_is_valid(pte2))
		return (pte2_pa(pte2));

	KASSERT(va >= VM_MAX_KERNEL_ADDRESS,
	    ("%s: NKPT2PG too small", __func__));

	/*
	 * Allocate page for PT2s and insert it to PT2TAB.
	 * In other words, map it into PT2MAP space.
	 */
	pt2pg_pa = pmap_preboot_get_pages(1);
	pt2tab_store(pte2p, PTE2_KPT(pt2pg_pa));

	/* Zero all PT2s in allocated page. */
	bzero((void*)pt2map_pt2pg(va), PAGE_SIZE);
	pte2_sync_range((pt2_entry_t *)pt2map_pt2pg(va), PAGE_SIZE);

	return (pt2pg_pa);
}

/*
 *  Setup L2 page table for given KVA.
 *  Used in pre-bootstrap epoch.
 */
static void
pmap_preboot_pt2_setup(vm_offset_t va)
{
	pt1_entry_t *pte1p;
	vm_paddr_t pt2pg_pa, pt2_pa;

	/* Setup PT2's page. */
	pt2pg_pa = pmap_preboot_pt2pg_setup(va);
	pt2_pa = page_pt2pa(pt2pg_pa, pte1_index(va));

	/* Insert PT2 to PT1. */
	pte1p = kern_pte1(va);
	pte1_store(pte1p, PTE1_LINK(pt2_pa));
}

/*
 *  Get L2 page entry associated with given KVA.
 *  Used in pre-bootstrap epoch.
 */
static __inline pt2_entry_t*
pmap_preboot_vtopte2(vm_offset_t va)
{
	pt1_entry_t *pte1p;

	/* Setup PT2 if needed. */
	pte1p = kern_pte1(va);
	if (!pte1_is_valid(pte1_load(pte1p))) /* XXX - sections ?! */
		pmap_preboot_pt2_setup(va);

	return (pt2map_entry(va));
}

/*
 *  Pre-bootstrap epoch page(s) mapping(s).
 */
void
pmap_preboot_map_pages(vm_paddr_t pa, vm_offset_t va, u_int num)
{
	u_int i;
	pt2_entry_t *pte2p;

	/* Map all the pages. */
	for (i = 0; i < num; i++) {
		pte2p = pmap_preboot_vtopte2(va);
		pte2_store(pte2p, PTE2_KRW(pa));
		va += PAGE_SIZE;
		pa += PAGE_SIZE;
	}
}

/*
 *  Pre-bootstrap epoch virtual space alocator.
 */
vm_offset_t
pmap_preboot_reserve_pages(u_int num)
{
	u_int i;
	vm_offset_t start, va;
	pt2_entry_t *pte2p;

	/* Allocate virtual space. */
	start = va = virtual_avail;
	virtual_avail += num * PAGE_SIZE;

	/* Zero the mapping. */
	for (i = 0; i < num; i++) {
		pte2p = pmap_preboot_vtopte2(va);
		pte2_store(pte2p, 0);
		va += PAGE_SIZE;
	}

	return (start);
}

/*
 *  Pre-bootstrap epoch page(s) allocation and mapping(s).
 */
vm_offset_t
pmap_preboot_get_vpages(u_int num)
{
	vm_paddr_t  pa;
	vm_offset_t va;

	/* Allocate physical page(s). */
	pa = pmap_preboot_get_pages(num);

	/* Allocate virtual space. */
	va = virtual_avail;
	virtual_avail += num * PAGE_SIZE;

	/* Map and zero all. */
	pmap_preboot_map_pages(pa, va, num);
	bzero((void *)va, num * PAGE_SIZE);

	return (va);
}

/*
 *  Pre-bootstrap epoch page mapping(s) with attributes.
 */
void
pmap_preboot_map_attr(vm_paddr_t pa, vm_offset_t va, vm_size_t size,
    vm_prot_t prot, vm_memattr_t attr)
{
	u_int num;
	u_int l1_attr, l1_prot, l2_prot, l2_attr;
	pt1_entry_t *pte1p;
	pt2_entry_t *pte2p;

	l2_prot = prot & VM_PROT_WRITE ? PTE2_AP_KRW : PTE2_AP_KR;
	l2_prot |= (prot & VM_PROT_EXECUTE) ? PTE2_X : PTE2_NX;
	l2_attr = vm_memattr_to_pte2(attr);
	l1_prot = ATTR_TO_L1(l2_prot);
	l1_attr = ATTR_TO_L1(l2_attr);

	/* Map all the pages. */
	num = round_page(size);
	while (num > 0) {
		if ((((va | pa) & PTE1_OFFSET) == 0) && (num >= PTE1_SIZE)) {
			pte1p = kern_pte1(va);
			pte1_store(pte1p, PTE1_KERN(pa, l1_prot, l1_attr));
			va += PTE1_SIZE;
			pa += PTE1_SIZE;
			num -= PTE1_SIZE;
		} else {
			pte2p = pmap_preboot_vtopte2(va);
			pte2_store(pte2p, PTE2_KERN(pa, l2_prot, l2_attr));
			va += PAGE_SIZE;
			pa += PAGE_SIZE;
			num -= PAGE_SIZE;
		}
	}
}

/*
 *  Extract from the kernel page table the physical address
 *  that is mapped by the given virtual address "va".
 */
vm_paddr_t
pmap_kextract(vm_offset_t va)
{
	vm_paddr_t pa;
	pt1_entry_t pte1;
	pt2_entry_t pte2;

	pte1 = pte1_load(kern_pte1(va));
	if (pte1_is_section(pte1)) {
		pa = pte1_pa(pte1) | (va & PTE1_OFFSET);
	} else if (pte1_is_link(pte1)) {
		/*
		 * We should beware of concurrent promotion that changes
		 * pte1 at this point. However, it's not a problem as PT2
		 * page is preserved by promotion in PT2TAB. So even if
		 * it happens, using of PT2MAP is still safe.
		 *
		 * QQQ: However, concurrent removing is a problem which
		 *      ends in abort on PT2MAP space. Locking must be used
		 *      to deal with this.
		 */
		pte2 = pte2_load(pt2map_entry(va));
		pa = pte2_pa(pte2) | (va & PTE2_OFFSET);
	}
	else {
		panic("%s: va %#x pte1 %#x", __func__, va, pte1);
	}
	return (pa);
}

/*
 *  Extract from the kernel page table the physical address
 *  that is mapped by the given virtual address "va". Also
 *  return L2 page table entry which maps the address.
 *
 *  This is only intended to be used for panic dumps.
 */
vm_paddr_t
pmap_dump_kextract(vm_offset_t va, pt2_entry_t *pte2p)
{
	vm_paddr_t pa;
	pt1_entry_t pte1;
	pt2_entry_t pte2;

	pte1 = pte1_load(kern_pte1(va));
	if (pte1_is_section(pte1)) {
		pa = pte1_pa(pte1) | (va & PTE1_OFFSET);
		pte2 = pa | ATTR_TO_L2(pte1) | PTE2_V;
	} else if (pte1_is_link(pte1)) {
		pte2 = pte2_load(pt2map_entry(va));
		pa = pte2_pa(pte2);
	} else {
		pte2 = 0;
		pa = 0;
	}
	if (pte2p != NULL)
		*pte2p = pte2;
	return (pa);
}

/*****************************************************************************
 *
 *	PMAP second stage initialization and utility functions
 *	for bootstrap epoch.
 *
 *  After pmap_bootstrap() is called, the following functions for
 *  mappings can be used:
 *
 *  void pmap_kenter(vm_offset_t va, vm_paddr_t pa);
 *  void pmap_kremove(vm_offset_t va);
 *  vm_offset_t pmap_map(vm_offset_t *virt, vm_paddr_t start, vm_paddr_t end,
 *      int prot);
 *
 *  NOTE: This is not SMP coherent stage. And physical page allocation is not
 *        allowed during this stage.
 *
 *****************************************************************************/

/*
 *  Initialize kernel PMAP locks and lists, kernel_pmap itself, and
 *  reserve various virtual spaces for temporary mappings.
 */
void
pmap_bootstrap(vm_offset_t firstaddr)
{
	pt2_entry_t *unused __unused;
	struct pcpu *pc;

	/*
	 * Initialize the kernel pmap (which is statically allocated).
	 */
	PMAP_LOCK_INIT(kernel_pmap);
	kernel_l1pa = (vm_paddr_t)kern_pt1;  /* for libkvm */
	kernel_pmap->pm_pt1 = kern_pt1;
	kernel_pmap->pm_pt2tab = kern_pt2tab;
	CPU_FILL(&kernel_pmap->pm_active);  /* don't allow deactivation */
	TAILQ_INIT(&kernel_pmap->pm_pvchunk);

	/*
	 * Initialize the global pv list lock.
	 */
	rw_init(&pvh_global_lock, "pmap pv global");

	LIST_INIT(&allpmaps);

	/*
	 * Request a spin mutex so that changes to allpmaps cannot be
	 * preempted by smp_rendezvous_cpus().
	 */
	mtx_init(&allpmaps_lock, "allpmaps", NULL, MTX_SPIN);
	mtx_lock_spin(&allpmaps_lock);
	LIST_INSERT_HEAD(&allpmaps, kernel_pmap, pm_list);
	mtx_unlock_spin(&allpmaps_lock);

	/*
	 * Reserve some special page table entries/VA space for temporary
	 * mapping of pages.
	 */
#define	SYSMAP(c, p, v, n)  do {		\
	v = (c)pmap_preboot_reserve_pages(n);	\
	p = pt2map_entry((vm_offset_t)v);	\
	} while (0)

	/*
	 * Local CMAP1/CMAP2 are used for zeroing and copying pages.
	 * Local CMAP2 is also used for data cache cleaning.
	 */
	pc = get_pcpu();
	mtx_init(&pc->pc_cmap_lock, "SYSMAPS", NULL, MTX_DEF);
	SYSMAP(caddr_t, pc->pc_cmap1_pte2p, pc->pc_cmap1_addr, 1);
	SYSMAP(caddr_t, pc->pc_cmap2_pte2p, pc->pc_cmap2_addr, 1);
	SYSMAP(vm_offset_t, pc->pc_qmap_pte2p, pc->pc_qmap_addr, 1);

	/*
	 * Crashdump maps.
	 */
	SYSMAP(caddr_t, unused, crashdumpmap, MAXDUMPPGS);

	/*
	 * _tmppt is used for reading arbitrary physical pages via /dev/mem.
	 */
	SYSMAP(caddr_t, unused, _tmppt, 1);

	/*
	 * PADDR1 and PADDR2 are used by pmap_pte2_quick() and pmap_pte2(),
	 * respectively. PADDR3 is used by pmap_pte2_ddb().
	 */
	SYSMAP(pt2_entry_t *, PMAP1, PADDR1, 1);
	SYSMAP(pt2_entry_t *, PMAP2, PADDR2, 1);
#ifdef DDB
	SYSMAP(pt2_entry_t *, PMAP3, PADDR3, 1);
#endif
	mtx_init(&PMAP2mutex, "PMAP2", NULL, MTX_DEF);

	/*
	 * Note that in very short time in initarm(), we are going to
	 * initialize phys_avail[] array and no further page allocation
	 * can happen after that until vm subsystem will be initialized.
	 */
	kernel_vm_end_new = kernel_vm_end;
	virtual_end = vm_max_kernel_address;
}

static void
pmap_init_reserved_pages(void)
{
	struct pcpu *pc;
	vm_offset_t pages;
	int i;

	CPU_FOREACH(i) {
		pc = pcpu_find(i);
		/*
		 * Skip if the mapping has already been initialized,
		 * i.e. this is the BSP.
		 */
		if (pc->pc_cmap1_addr != 0)
			continue;
		mtx_init(&pc->pc_cmap_lock, "SYSMAPS", NULL, MTX_DEF);
		pages = kva_alloc(PAGE_SIZE * 3);
		if (pages == 0)
			panic("%s: unable to allocate KVA", __func__);
		pc->pc_cmap1_pte2p = pt2map_entry(pages);
		pc->pc_cmap2_pte2p = pt2map_entry(pages + PAGE_SIZE);
		pc->pc_qmap_pte2p = pt2map_entry(pages + (PAGE_SIZE * 2));
		pc->pc_cmap1_addr = (caddr_t)pages;
		pc->pc_cmap2_addr = (caddr_t)(pages + PAGE_SIZE);
		pc->pc_qmap_addr = pages + (PAGE_SIZE * 2);
	}
}
SYSINIT(rpages_init, SI_SUB_CPU, SI_ORDER_ANY, pmap_init_reserved_pages, NULL);

/*
 *  The function can already be use in second initialization stage.
 *  As such, the function DOES NOT call pmap_growkernel() where PT2
 *  allocation can happen. So if used, be sure that PT2 for given
 *  virtual address is allocated already!
 *
 *  Add a wired page to the kva.
 *  Note: not SMP coherent.
 */
static __inline void
pmap_kenter_prot_attr(vm_offset_t va, vm_paddr_t pa, uint32_t prot,
    uint32_t attr)
{
	pt1_entry_t *pte1p;
	pt2_entry_t *pte2p;

	pte1p = kern_pte1(va);
	if (!pte1_is_valid(pte1_load(pte1p))) { /* XXX - sections ?! */
		/*
		 * This is a very low level function, so PT2 and particularly
		 * PT2PG associated with given virtual address must be already
		 * allocated. It's a pain mainly during pmap initialization
		 * stage. However, called after pmap initialization with
		 * virtual address not under kernel_vm_end will lead to
		 * the same misery.
		 */
		if (!pte2_is_valid(pte2_load(kern_pt2tab_entry(va))))
			panic("%s: kernel PT2 not allocated!", __func__);
	}

	pte2p = pt2map_entry(va);
	pte2_store(pte2p, PTE2_KERN(pa, prot, attr));
}

PMAP_INLINE void
pmap_kenter(vm_offset_t va, vm_paddr_t pa)
{

	pmap_kenter_prot_attr(va, pa, PTE2_AP_KRW, PTE2_ATTR_DEFAULT);
}

/*
 *  Remove a page from the kernel pagetables.
 *  Note: not SMP coherent.
 */
PMAP_INLINE void
pmap_kremove(vm_offset_t va)
{
	pt1_entry_t *pte1p;
	pt2_entry_t *pte2p;

	pte1p = kern_pte1(va);
	if (pte1_is_section(pte1_load(pte1p))) {
		pte1_clear(pte1p);
	} else {
		pte2p = pt2map_entry(va);
		pte2_clear(pte2p);
	}
}

/*
 *  Share new kernel PT2PG with all pmaps.
 *  The caller is responsible for maintaining TLB consistency.
 */
static void
pmap_kenter_pt2tab(vm_offset_t va, pt2_entry_t npte2)
{
	pmap_t pmap;
	pt2_entry_t *pte2p;

	mtx_lock_spin(&allpmaps_lock);
	LIST_FOREACH(pmap, &allpmaps, pm_list) {
		pte2p = pmap_pt2tab_entry(pmap, va);
		pt2tab_store(pte2p, npte2);
	}
	mtx_unlock_spin(&allpmaps_lock);
}

/*
 *  Share new kernel PTE1 with all pmaps.
 *  The caller is responsible for maintaining TLB consistency.
 */
static void
pmap_kenter_pte1(vm_offset_t va, pt1_entry_t npte1)
{
	pmap_t pmap;
	pt1_entry_t *pte1p;

	mtx_lock_spin(&allpmaps_lock);
	LIST_FOREACH(pmap, &allpmaps, pm_list) {
		pte1p = pmap_pte1(pmap, va);
		pte1_store(pte1p, npte1);
	}
	mtx_unlock_spin(&allpmaps_lock);
}

/*
 *  Used to map a range of physical addresses into kernel
 *  virtual address space.
 *
 *  The value passed in '*virt' is a suggested virtual address for
 *  the mapping. Architectures which can support a direct-mapped
 *  physical to virtual region can return the appropriate address
 *  within that region, leaving '*virt' unchanged. Other
 *  architectures should map the pages starting at '*virt' and
 *  update '*virt' with the first usable address after the mapped
 *  region.
 *
 *  NOTE: Read the comments above pmap_kenter_prot_attr() as
 *        the function is used herein!
 */
vm_offset_t
pmap_map(vm_offset_t *virt, vm_paddr_t start, vm_paddr_t end, int prot)
{
	vm_offset_t va, sva;
	vm_paddr_t pte1_offset;
	pt1_entry_t npte1;
	uint32_t l1prot, l2prot;
	uint32_t l1attr, l2attr;

	PDEBUG(1, printf("%s: virt = %#x, start = %#x, end = %#x (size = %#x),"
	    " prot = %d\n", __func__, *virt, start, end, end - start,  prot));

	l2prot = (prot & VM_PROT_WRITE) ? PTE2_AP_KRW : PTE2_AP_KR;
	l2prot |= (prot & VM_PROT_EXECUTE) ? PTE2_X : PTE2_NX;
	l1prot = ATTR_TO_L1(l2prot);

	l2attr = PTE2_ATTR_DEFAULT;
	l1attr = ATTR_TO_L1(l2attr);

	va = *virt;
	/*
	 * Does the physical address range's size and alignment permit at
	 * least one section mapping to be created?
	 */
	pte1_offset = start & PTE1_OFFSET;
	if ((end - start) - ((PTE1_SIZE - pte1_offset) & PTE1_OFFSET) >=
	    PTE1_SIZE) {
		/*
		 * Increase the starting virtual address so that its alignment
		 * does not preclude the use of section mappings.
		 */
		if ((va & PTE1_OFFSET) < pte1_offset)
			va = pte1_trunc(va) + pte1_offset;
		else if ((va & PTE1_OFFSET) > pte1_offset)
			va = pte1_roundup(va) + pte1_offset;
	}
	sva = va;
	while (start < end) {
		if ((start & PTE1_OFFSET) == 0 && end - start >= PTE1_SIZE) {
			KASSERT((va & PTE1_OFFSET) == 0,
			    ("%s: misaligned va %#x", __func__, va));
			npte1 = PTE1_KERN(start, l1prot, l1attr);
			pmap_kenter_pte1(va, npte1);
			va += PTE1_SIZE;
			start += PTE1_SIZE;
		} else {
			pmap_kenter_prot_attr(va, start, l2prot, l2attr);
			va += PAGE_SIZE;
			start += PAGE_SIZE;
		}
	}
	tlb_flush_range(sva, va - sva);
	*virt = va;
	return (sva);
}

/*
 *  Make a temporary mapping for a physical address.
 *  This is only intended to be used for panic dumps.
 */
void *
pmap_kenter_temporary(vm_paddr_t pa, int i)
{
	vm_offset_t va;

	/* QQQ: 'i' should be less or equal to MAXDUMPPGS. */

	va = (vm_offset_t)crashdumpmap + (i * PAGE_SIZE);
	pmap_kenter(va, pa);
	tlb_flush_local(va);
	return ((void *)crashdumpmap);
}

/*************************************
 *
 *  TLB & cache maintenance routines.
 *
 *************************************/

/*
 *  We inline these within pmap.c for speed.
 */
PMAP_INLINE void
pmap_tlb_flush(pmap_t pmap, vm_offset_t va)
{

	if (pmap == kernel_pmap || !CPU_EMPTY(&pmap->pm_active))
		tlb_flush(va);
}

PMAP_INLINE void
pmap_tlb_flush_range(pmap_t pmap, vm_offset_t sva, vm_size_t size)
{

	if (pmap == kernel_pmap || !CPU_EMPTY(&pmap->pm_active))
		tlb_flush_range(sva, size);
}

/*
 *  Abuse the pte2 nodes for unmapped kva to thread a kva freelist through.
 *  Requirements:
 *   - Must deal with pages in order to ensure that none of the PTE2_* bits
 *     are ever set, PTE2_V in particular.
 *   - Assumes we can write to pte2s without pte2_store() atomic ops.
 *   - Assumes nothing will ever test these addresses for 0 to indicate
 *     no mapping instead of correctly checking PTE2_V.
 *   - Assumes a vm_offset_t will fit in a pte2 (true for arm).
 *  Because PTE2_V is never set, there can be no mappings to invalidate.
 */
static vm_offset_t
pmap_pte2list_alloc(vm_offset_t *head)
{
	pt2_entry_t *pte2p;
	vm_offset_t va;

	va = *head;
	if (va == 0)
		panic("pmap_ptelist_alloc: exhausted ptelist KVA");
	pte2p = pt2map_entry(va);
	*head = *pte2p;
	if (*head & PTE2_V)
		panic("%s: va with PTE2_V set!", __func__);
	*pte2p = 0;
	return (va);
}

static void
pmap_pte2list_free(vm_offset_t *head, vm_offset_t va)
{
	pt2_entry_t *pte2p;

	if (va & PTE2_V)
		panic("%s: freeing va with PTE2_V set!", __func__);
	pte2p = pt2map_entry(va);
	*pte2p = *head;		/* virtual! PTE2_V is 0 though */
	*head = va;
}

static void
pmap_pte2list_init(vm_offset_t *head, void *base, int npages)
{
	int i;
	vm_offset_t va;

	*head = 0;
	for (i = npages - 1; i >= 0; i--) {
		va = (vm_offset_t)base + i * PAGE_SIZE;
		pmap_pte2list_free(head, va);
	}
}

/*****************************************************************************
 *
 *	PMAP third and final stage initialization.
 *
 *  After pmap_init() is called, PMAP subsystem is fully initialized.
 *
 *****************************************************************************/

SYSCTL_NODE(_vm, OID_AUTO, pmap, CTLFLAG_RD | CTLFLAG_MPSAFE, 0,
    "VM/pmap parameters");

SYSCTL_INT(_vm_pmap, OID_AUTO, pv_entry_max, CTLFLAG_RD, &pv_entry_max, 0,
    "Max number of PV entries");
SYSCTL_INT(_vm_pmap, OID_AUTO, shpgperproc, CTLFLAG_RD, &shpgperproc, 0,
    "Page share factor per proc");

static u_long nkpt2pg = NKPT2PG;
SYSCTL_ULONG(_vm_pmap, OID_AUTO, nkpt2pg, CTLFLAG_RD,
    &nkpt2pg, 0, "Pre-allocated pages for kernel PT2s");

static int sp_enabled = 1;
SYSCTL_INT(_vm_pmap, OID_AUTO, sp_enabled, CTLFLAG_RDTUN | CTLFLAG_NOFETCH,
    &sp_enabled, 0, "Are large page mappings enabled?");

bool
pmap_ps_enabled(pmap_t pmap __unused)
{

	return (sp_enabled != 0);
}

static SYSCTL_NODE(_vm_pmap, OID_AUTO, pte1, CTLFLAG_RD | CTLFLAG_MPSAFE, 0,
    "1MB page mapping counters");

static u_long pmap_pte1_demotions;
SYSCTL_ULONG(_vm_pmap_pte1, OID_AUTO, demotions, CTLFLAG_RD,
    &pmap_pte1_demotions, 0, "1MB page demotions");

static u_long pmap_pte1_mappings;
SYSCTL_ULONG(_vm_pmap_pte1, OID_AUTO, mappings, CTLFLAG_RD,
    &pmap_pte1_mappings, 0, "1MB page mappings");

static u_long pmap_pte1_p_failures;
SYSCTL_ULONG(_vm_pmap_pte1, OID_AUTO, p_failures, CTLFLAG_RD,
    &pmap_pte1_p_failures, 0, "1MB page promotion failures");

static u_long pmap_pte1_promotions;
SYSCTL_ULONG(_vm_pmap_pte1, OID_AUTO, promotions, CTLFLAG_RD,
    &pmap_pte1_promotions, 0, "1MB page promotions");

static u_long pmap_pte1_kern_demotions;
SYSCTL_ULONG(_vm_pmap_pte1, OID_AUTO, kern_demotions, CTLFLAG_RD,
    &pmap_pte1_kern_demotions, 0, "1MB page kernel demotions");

static u_long pmap_pte1_kern_promotions;
SYSCTL_ULONG(_vm_pmap_pte1, OID_AUTO, kern_promotions, CTLFLAG_RD,
    &pmap_pte1_kern_promotions, 0, "1MB page kernel promotions");

static __inline ttb_entry_t
pmap_ttb_get(pmap_t pmap)
{

	return (vtophys(pmap->pm_pt1) | ttb_flags);
}

/*
 *  Initialize a vm_page's machine-dependent fields.
 *
 *  Variations:
 *  1. Pages for L2 page tables are always not managed. So, pv_list and
 *     pt2_wirecount can share same physical space. However, proper
 *     initialization on a page alloc for page tables and reinitialization
 *     on the page free must be ensured.
 */
void
pmap_page_init(vm_page_t m)
{

	TAILQ_INIT(&m->md.pv_list);
	pt2_wirecount_init(m);
	m->md.pat_mode = VM_MEMATTR_DEFAULT;
}

/*
 *  Virtualization for faster way how to zero whole page.
 */
static __inline void
pagezero(void *page)
{

	bzero(page, PAGE_SIZE);
}

/*
 *  Zero L2 page table page.
 *  Use same KVA as in pmap_zero_page().
 */
static __inline vm_paddr_t
pmap_pt2pg_zero(vm_page_t m)
{
	pt2_entry_t *cmap2_pte2p;
	vm_paddr_t pa;
	struct pcpu *pc;

	pa = VM_PAGE_TO_PHYS(m);

	/*
	 * XXX: For now, we map whole page even if it's already zero,
	 *      to sync it even if the sync is only DSB.
	 */
	sched_pin();
	pc = get_pcpu();
	cmap2_pte2p = pc->pc_cmap2_pte2p;
	mtx_lock(&pc->pc_cmap_lock);
	if (pte2_load(cmap2_pte2p) != 0)
		panic("%s: CMAP2 busy", __func__);
	pte2_store(cmap2_pte2p, PTE2_KERN_NG(pa, PTE2_AP_KRW,
	    vm_page_pte2_attr(m)));
	/*  Even VM_ALLOC_ZERO request is only advisory. */
	if ((m->flags & PG_ZERO) == 0)
		pagezero(pc->pc_cmap2_addr);
	pte2_sync_range((pt2_entry_t *)pc->pc_cmap2_addr, PAGE_SIZE);
	pte2_clear(cmap2_pte2p);
	tlb_flush((vm_offset_t)pc->pc_cmap2_addr);

	/*
	 * Unpin the thread before releasing the lock.  Otherwise the thread
	 * could be rescheduled while still bound to the current CPU, only
	 * to unpin itself immediately upon resuming execution.
	 */
	sched_unpin();
	mtx_unlock(&pc->pc_cmap_lock);

	return (pa);
}

/*
 *  Init just allocated page as L2 page table(s) holder
 *  and return its physical address.
 */
static __inline vm_paddr_t
pmap_pt2pg_init(pmap_t pmap, vm_offset_t va, vm_page_t m)
{
	vm_paddr_t pa;
	pt2_entry_t *pte2p;

	/* Check page attributes. */
	if (m->md.pat_mode != pt_memattr)
		pmap_page_set_memattr(m, pt_memattr);

	/* Zero page and init wire counts. */
	pa = pmap_pt2pg_zero(m);
	pt2_wirecount_init(m);

	/*
	 * Map page to PT2MAP address space for given pmap.
	 * Note that PT2MAP space is shared with all pmaps.
	 */
	if (pmap == kernel_pmap)
		pmap_kenter_pt2tab(va, PTE2_KPT(pa));
	else {
		pte2p = pmap_pt2tab_entry(pmap, va);
		pt2tab_store(pte2p, PTE2_KPT_NG(pa));
	}

	return (pa);
}

/*
 *  Initialize the pmap module.
 *  Called by vm_init, to initialize any structures that the pmap
 *  system needs to map virtual memory.
 */
void
pmap_init(void)
{
	vm_size_t s;
	pt2_entry_t *pte2p, pte2;
	u_int i, pte1_idx, pv_npg;

	PDEBUG(1, printf("%s: phys_start = %#x\n", __func__, PHYSADDR));

	/*
	 * Initialize the vm page array entries for kernel pmap's
	 * L2 page table pages allocated in advance.
	 */
	pte1_idx = pte1_index(KERNBASE - PT2MAP_SIZE);
	pte2p = kern_pt2tab_entry(KERNBASE - PT2MAP_SIZE);
	for (i = 0; i < nkpt2pg + NPG_IN_PT2TAB; i++, pte2p++) {
		vm_paddr_t pa;
		vm_page_t m;

		pte2 = pte2_load(pte2p);
		KASSERT(pte2_is_valid(pte2), ("%s: no valid entry", __func__));

		pa = pte2_pa(pte2);
		m = PHYS_TO_VM_PAGE(pa);
		KASSERT(m >= vm_page_array &&
		    m < &vm_page_array[vm_page_array_size],
		    ("%s: L2 page table page is out of range", __func__));

		m->pindex = pte1_idx;
		m->phys_addr = pa;
		pte1_idx += NPT2_IN_PG;
	}

	/*
	 * Initialize the address space (zone) for the pv entries.  Set a
	 * high water mark so that the system can recover from excessive
	 * numbers of pv entries.
	 */
	TUNABLE_INT_FETCH("vm.pmap.shpgperproc", &shpgperproc);
	pv_entry_max = shpgperproc * maxproc + vm_cnt.v_page_count;
	TUNABLE_INT_FETCH("vm.pmap.pv_entries", &pv_entry_max);
	pv_entry_max = roundup(pv_entry_max, _NPCPV);
	pv_entry_high_water = 9 * (pv_entry_max / 10);

	/*
	 * Are large page mappings enabled?
	 */
	TUNABLE_INT_FETCH("vm.pmap.sp_enabled", &sp_enabled);
	if (sp_enabled) {
		KASSERT(MAXPAGESIZES > 1 && pagesizes[1] == 0,
		    ("%s: can't assign to pagesizes[1]", __func__));
		pagesizes[1] = PTE1_SIZE;
	}

	/*
	 * Calculate the size of the pv head table for sections.
	 * Handle the possibility that "vm_phys_segs[...].end" is zero.
	 * Note that the table is only for sections which could be promoted.
	 */
	first_managed_pa = pte1_trunc(vm_phys_segs[0].start);
	pv_npg = (pte1_trunc(vm_phys_segs[vm_phys_nsegs - 1].end - PAGE_SIZE)
	    - first_managed_pa) / PTE1_SIZE + 1;

	/*
	 * Allocate memory for the pv head table for sections.
	 */
	s = (vm_size_t)(pv_npg * sizeof(struct md_page));
	s = round_page(s);
	pv_table = (struct md_page *)kmem_malloc(s, M_WAITOK | M_ZERO);
	for (i = 0; i < pv_npg; i++)
		TAILQ_INIT(&pv_table[i].pv_list);

	pv_maxchunks = MAX(pv_entry_max / _NPCPV, maxproc);
	pv_chunkbase = (struct pv_chunk *)kva_alloc(PAGE_SIZE * pv_maxchunks);
	if (pv_chunkbase == NULL)
		panic("%s: not enough kvm for pv chunks", __func__);
	pmap_pte2list_init(&pv_vafree, pv_chunkbase, pv_maxchunks);
}

/*
 *  Add a list of wired pages to the kva
 *  this routine is only used for temporary
 *  kernel mappings that do not need to have
 *  page modification or references recorded.
 *  Note that old mappings are simply written
 *  over.  The page *must* be wired.
 *  Note: SMP coherent.  Uses a ranged shootdown IPI.
 */
void
pmap_qenter(vm_offset_t sva, vm_page_t *ma, int count)
{
	u_int anychanged;
	pt2_entry_t *epte2p, *pte2p, pte2;
	vm_page_t m;
	vm_paddr_t pa;

	anychanged = 0;
	pte2p = pt2map_entry(sva);
	epte2p = pte2p + count;
	while (pte2p < epte2p) {
		m = *ma++;
		pa = VM_PAGE_TO_PHYS(m);
		pte2 = pte2_load(pte2p);
		if ((pte2_pa(pte2) != pa) ||
		    (pte2_attr(pte2) != vm_page_pte2_attr(m))) {
			anychanged++;
			pte2_store(pte2p, PTE2_KERN(pa, PTE2_AP_KRW,
			    vm_page_pte2_attr(m)));
		}
		pte2p++;
	}
	if (__predict_false(anychanged))
		tlb_flush_range(sva, count * PAGE_SIZE);
}

/*
 *  This routine tears out page mappings from the
 *  kernel -- it is meant only for temporary mappings.
 *  Note: SMP coherent.  Uses a ranged shootdown IPI.
 */
void
pmap_qremove(vm_offset_t sva, int count)
{
	vm_offset_t va;

	va = sva;
	while (count-- > 0) {
		pmap_kremove(va);
		va += PAGE_SIZE;
	}
	tlb_flush_range(sva, va - sva);
}

/*
 *  Are we current address space or kernel?
 */
static __inline int
pmap_is_current(pmap_t pmap)
{

	return (pmap == kernel_pmap ||
		(pmap == vmspace_pmap(curthread->td_proc->p_vmspace)));
}

/*
 *  If the given pmap is not the current or kernel pmap, the returned
 *  pte2 must be released by passing it to pmap_pte2_release().
 */
static pt2_entry_t *
pmap_pte2(pmap_t pmap, vm_offset_t va)
{
	pt1_entry_t pte1;
	vm_paddr_t pt2pg_pa;

	pte1 = pte1_load(pmap_pte1(pmap, va));
	if (pte1_is_section(pte1))
		panic("%s: attempt to map PTE1", __func__);
	if (pte1_is_link(pte1)) {
		/* Are we current address space or kernel? */
		if (pmap_is_current(pmap))
			return (pt2map_entry(va));
		/* Note that L2 page table size is not equal to PAGE_SIZE. */
		pt2pg_pa = trunc_page(pte1_link_pa(pte1));
		mtx_lock(&PMAP2mutex);
		if (pte2_pa(pte2_load(PMAP2)) != pt2pg_pa) {
			pte2_store(PMAP2, PTE2_KPT(pt2pg_pa));
			tlb_flush((vm_offset_t)PADDR2);
		}
		return (PADDR2 + (arm32_btop(va) & (NPTE2_IN_PG - 1)));
	}
	return (NULL);
}

/*
 *  Releases a pte2 that was obtained from pmap_pte2().
 *  Be prepared for the pte2p being NULL.
 */
static __inline void
pmap_pte2_release(pt2_entry_t *pte2p)
{

	if ((pt2_entry_t *)(trunc_page((vm_offset_t)pte2p)) == PADDR2) {
		mtx_unlock(&PMAP2mutex);
	}
}

/*
 *  Super fast pmap_pte2 routine best used when scanning
 *  the pv lists.  This eliminates many coarse-grained
 *  invltlb calls.  Note that many of the pv list
 *  scans are across different pmaps.  It is very wasteful
 *  to do an entire tlb flush for checking a single mapping.
 *
 *  If the given pmap is not the current pmap, pvh_global_lock
 *  must be held and curthread pinned to a CPU.
 */
static pt2_entry_t *
pmap_pte2_quick(pmap_t pmap, vm_offset_t va)
{
	pt1_entry_t pte1;
	vm_paddr_t pt2pg_pa;

	pte1 = pte1_load(pmap_pte1(pmap, va));
	if (pte1_is_section(pte1))
		panic("%s: attempt to map PTE1", __func__);
	if (pte1_is_link(pte1)) {
		/* Are we current address space or kernel? */
		if (pmap_is_current(pmap))
			return (pt2map_entry(va));
		rw_assert(&pvh_global_lock, RA_WLOCKED);
		KASSERT(curthread->td_pinned > 0,
		    ("%s: curthread not pinned", __func__));
		/* Note that L2 page table size is not equal to PAGE_SIZE. */
		pt2pg_pa = trunc_page(pte1_link_pa(pte1));
		if (pte2_pa(pte2_load(PMAP1)) != pt2pg_pa) {
			pte2_store(PMAP1, PTE2_KPT(pt2pg_pa));
#ifdef SMP
			PMAP1cpu = PCPU_GET(cpuid);
#endif
			tlb_flush_local((vm_offset_t)PADDR1);
			PMAP1changed++;
		} else
#ifdef SMP
		if (PMAP1cpu != PCPU_GET(cpuid)) {
			PMAP1cpu = PCPU_GET(cpuid);
			tlb_flush_local((vm_offset_t)PADDR1);
			PMAP1changedcpu++;
		} else
#endif
			PMAP1unchanged++;
		return (PADDR1 + (arm32_btop(va) & (NPTE2_IN_PG - 1)));
	}
	return (NULL);
}

/*
 *  Routine: pmap_extract
 *  Function:
 * 	Extract the physical page address associated
 *	with the given map/virtual_address pair.
 */
vm_paddr_t
pmap_extract(pmap_t pmap, vm_offset_t va)
{
	vm_paddr_t pa;
	pt1_entry_t pte1;
	pt2_entry_t *pte2p;

	PMAP_LOCK(pmap);
	pte1 = pte1_load(pmap_pte1(pmap, va));
	if (pte1_is_section(pte1))
		pa = pte1_pa(pte1) | (va & PTE1_OFFSET);
	else if (pte1_is_link(pte1)) {
		pte2p = pmap_pte2(pmap, va);
		pa = pte2_pa(pte2_load(pte2p)) | (va & PTE2_OFFSET);
		pmap_pte2_release(pte2p);
	} else
		pa = 0;
	PMAP_UNLOCK(pmap);
	return (pa);
}

/*
 *  Routine: pmap_extract_and_hold
 *  Function:
 *	Atomically extract and hold the physical page
 *	with the given pmap and virtual address pair
 *	if that mapping permits the given protection.
 */
vm_page_t
pmap_extract_and_hold(pmap_t pmap, vm_offset_t va, vm_prot_t prot)
{
	vm_paddr_t pa;
	pt1_entry_t pte1;
	pt2_entry_t pte2, *pte2p;
	vm_page_t m;

	m = NULL;
	PMAP_LOCK(pmap);
	pte1 = pte1_load(pmap_pte1(pmap, va));
	if (pte1_is_section(pte1)) {
		if (!(pte1 & PTE1_RO) || !(prot & VM_PROT_WRITE)) {
			pa = pte1_pa(pte1) | (va & PTE1_OFFSET);
			m = PHYS_TO_VM_PAGE(pa);
			if (!vm_page_wire_mapped(m))
				m = NULL;
		}
	} else if (pte1_is_link(pte1)) {
		pte2p = pmap_pte2(pmap, va);
		pte2 = pte2_load(pte2p);
		pmap_pte2_release(pte2p);
		if (pte2_is_valid(pte2) &&
		    (!(pte2 & PTE2_RO) || !(prot & VM_PROT_WRITE))) {
			pa = pte2_pa(pte2);
			m = PHYS_TO_VM_PAGE(pa);
			if (!vm_page_wire_mapped(m))
				m = NULL;
		}
	}
	PMAP_UNLOCK(pmap);
	return (m);
}

/*
 *  Grow the number of kernel L2 page table entries, if needed.
 */
void
pmap_growkernel(vm_offset_t addr)
{
	vm_page_t m;
	vm_paddr_t pt2pg_pa, pt2_pa;
	pt1_entry_t pte1;
	pt2_entry_t pte2;

	PDEBUG(1, printf("%s: addr = %#x\n", __func__, addr));
	/*
	 * All the time kernel_vm_end is first KVA for which underlying
	 * L2 page table is either not allocated or linked from L1 page table
	 * (not considering sections). Except for two possible cases:
	 *
	 *   (1) in the very beginning as long as pmap_growkernel() was
	 *       not called, it could be first unused KVA (which is not
	 *       rounded up to PTE1_SIZE),
	 *
	 *   (2) when all KVA space is mapped and vm_map_max(kernel_map)
	 *       address is not rounded up to PTE1_SIZE. (For example,
	 *       it could be 0xFFFFFFFF.)
	 */
	kernel_vm_end = pte1_roundup(kernel_vm_end);
	mtx_assert(&kernel_map->system_mtx, MA_OWNED);
	addr = roundup2(addr, PTE1_SIZE);
	if (addr - 1 >= vm_map_max(kernel_map))
		addr = vm_map_max(kernel_map);
	while (kernel_vm_end < addr) {
		pte1 = pte1_load(kern_pte1(kernel_vm_end));
		if (pte1_is_valid(pte1)) {
			kernel_vm_end += PTE1_SIZE;
			if (kernel_vm_end - 1 >= vm_map_max(kernel_map)) {
				kernel_vm_end = vm_map_max(kernel_map);
				break;
			}
			continue;
		}

		/*
		 * kernel_vm_end_new is used in pmap_pinit() when kernel
		 * mappings are entered to new pmap all at once to avoid race
		 * between pmap_kenter_pte1() and kernel_vm_end increase.
		 * The same aplies to pmap_kenter_pt2tab().
		 */
		kernel_vm_end_new = kernel_vm_end + PTE1_SIZE;

		pte2 = pt2tab_load(kern_pt2tab_entry(kernel_vm_end));
		if (!pte2_is_valid(pte2)) {
			/*
			 * Install new PT2s page into kernel PT2TAB.
			 */
			m = vm_page_alloc(NULL,
			    pte1_index(kernel_vm_end) & ~PT2PG_MASK,
			    VM_ALLOC_INTERRUPT | VM_ALLOC_NOOBJ |
			    VM_ALLOC_WIRED | VM_ALLOC_ZERO);
			if (m == NULL)
				panic("%s: no memory to grow kernel", __func__);
			/*
			 * QQQ: To link all new L2 page tables from L1 page
			 *      table now and so pmap_kenter_pte1() them
			 *      at once together with pmap_kenter_pt2tab()
			 *      could be nice speed up. However,
			 *      pmap_growkernel() does not happen so often...
			 * QQQ: The other TTBR is another option.
			 */
			pt2pg_pa = pmap_pt2pg_init(kernel_pmap, kernel_vm_end,
			    m);
		} else
			pt2pg_pa = pte2_pa(pte2);

		pt2_pa = page_pt2pa(pt2pg_pa, pte1_index(kernel_vm_end));
		pmap_kenter_pte1(kernel_vm_end, PTE1_LINK(pt2_pa));

		kernel_vm_end = kernel_vm_end_new;
		if (kernel_vm_end - 1 >= vm_map_max(kernel_map)) {
			kernel_vm_end = vm_map_max(kernel_map);
			break;
		}
	}
}

static int
kvm_size(SYSCTL_HANDLER_ARGS)
{
	unsigned long ksize = vm_max_kernel_address - KERNBASE;

	return (sysctl_handle_long(oidp, &ksize, 0, req));
}
SYSCTL_PROC(_vm, OID_AUTO, kvm_size,
    CTLTYPE_LONG | CTLFLAG_RD | CTLFLAG_NEEDGIANT, 0, 0, kvm_size, "IU",
    "Size of KVM");

static int
kvm_free(SYSCTL_HANDLER_ARGS)
{
	unsigned long kfree = vm_max_kernel_address - kernel_vm_end;

	return (sysctl_handle_long(oidp, &kfree, 0, req));
}
SYSCTL_PROC(_vm, OID_AUTO, kvm_free,
    CTLTYPE_LONG | CTLFLAG_RD | CTLFLAG_NEEDGIANT, 0, 0, kvm_free, "IU",
    "Amount of KVM free");

/***********************************************
 *
 *  Pmap allocation/deallocation routines.
 *
 ***********************************************/

/*
 *  Initialize the pmap for the swapper process.
 */
void
pmap_pinit0(pmap_t pmap)
{
	PDEBUG(1, printf("%s: pmap = %p\n", __func__, pmap));

	PMAP_LOCK_INIT(pmap);

	/*
	 * Kernel page table directory and pmap stuff around is already
	 * initialized, we are using it right now and here. So, finish
	 * only PMAP structures initialization for process0 ...
	 *
	 * Since the L1 page table and PT2TAB is shared with the kernel pmap,
	 * which is already included in the list "allpmaps", this pmap does
	 * not need to be inserted into that list.
	 */
	pmap->pm_pt1 = kern_pt1;
	pmap->pm_pt2tab = kern_pt2tab;
	CPU_ZERO(&pmap->pm_active);
	PCPU_SET(curpmap, pmap);
	TAILQ_INIT(&pmap->pm_pvchunk);
	bzero(&pmap->pm_stats, sizeof pmap->pm_stats);
	CPU_SET(0, &pmap->pm_active);
}

static __inline void
pte1_copy_nosync(pt1_entry_t *spte1p, pt1_entry_t *dpte1p, vm_offset_t sva,
    vm_offset_t eva)
{
	u_int idx, count;

	idx = pte1_index(sva);
	count = (pte1_index(eva) - idx + 1) * sizeof(pt1_entry_t);
	bcopy(spte1p + idx, dpte1p + idx, count);
}

static __inline void
pt2tab_copy_nosync(pt2_entry_t *spte2p, pt2_entry_t *dpte2p, vm_offset_t sva,
    vm_offset_t eva)
{
	u_int idx, count;

	idx = pt2tab_index(sva);
	count = (pt2tab_index(eva) - idx + 1) * sizeof(pt2_entry_t);
	bcopy(spte2p + idx, dpte2p + idx, count);
}

/*
 *  Initialize a preallocated and zeroed pmap structure,
 *  such as one in a vmspace structure.
 */
int
pmap_pinit(pmap_t pmap)
{
	pt1_entry_t *pte1p;
	pt2_entry_t *pte2p;
	vm_paddr_t pa, pt2tab_pa;
	u_int i;

	PDEBUG(6, printf("%s: pmap = %p, pm_pt1 = %p\n", __func__, pmap,
	    pmap->pm_pt1));

	/*
	 * No need to allocate L2 page table space yet but we do need
	 * a valid L1 page table and PT2TAB table.
	 *
	 * Install shared kernel mappings to these tables. It's a little
	 * tricky as some parts of KVA are reserved for vectors, devices,
	 * and whatever else. These parts are supposed to be above
	 * vm_max_kernel_address. Thus two regions should be installed:
	 *
	 *   (1) <KERNBASE, kernel_vm_end),
	 *   (2) <vm_max_kernel_address, 0xFFFFFFFF>.
	 *
	 * QQQ: The second region should be stable enough to be installed
	 *      only once in time when the tables are allocated.
	 * QQQ: Maybe copy of both regions at once could be faster ...
	 * QQQ: Maybe the other TTBR is an option.
	 *
	 * Finally, install own PT2TAB table to these tables.
	 */

	if (pmap->pm_pt1 == NULL) {
		pmap->pm_pt1 = (pt1_entry_t *)kmem_alloc_contig(NB_IN_PT1,
		    M_NOWAIT | M_ZERO, 0, -1UL, NB_IN_PT1, 0, pt_memattr);
		if (pmap->pm_pt1 == NULL)
			return (0);
	}
	if (pmap->pm_pt2tab == NULL) {
		/*
		 * QQQ: (1) PT2TAB must be contiguous. If PT2TAB is one page
		 *      only, what should be the only size for 32 bit systems,
		 *      then we could allocate it with vm_page_alloc() and all
		 *      the stuff needed as other L2 page table pages.
		 *      (2) Note that a process PT2TAB is special L2 page table
		 *      page. Its mapping in kernel_arena is permanent and can
		 *      be used no matter which process is current. Its mapping
		 *      in PT2MAP can be used only for current process.
		 */
		pmap->pm_pt2tab = (pt2_entry_t *)kmem_alloc_attr(NB_IN_PT2TAB,
		    M_NOWAIT | M_ZERO, 0, -1UL, pt_memattr);
		if (pmap->pm_pt2tab == NULL) {
			/*
			 * QQQ: As struct pmap is allocated from UMA with
			 *      UMA_ZONE_NOFREE flag, it's important to leave
			 *      no allocation in pmap if initialization failed.
			 */
			kmem_free((vm_offset_t)pmap->pm_pt1, NB_IN_PT1);
			pmap->pm_pt1 = NULL;
			return (0);
		}
		/*
		 * QQQ: Each L2 page table page vm_page_t has pindex set to
		 *      pte1 index of virtual address mapped by this page.
		 *      It's not valid for non kernel PT2TABs themselves.
		 *      The pindex of these pages can not be altered because
		 *      of the way how they are allocated now. However, it
		 *      should not be a problem.
		 */
	}

	mtx_lock_spin(&allpmaps_lock);
	/*
	 * To avoid race with pmap_kenter_pte1() and pmap_kenter_pt2tab(),
	 * kernel_vm_end_new is used here instead of kernel_vm_end.
	 */
	pte1_copy_nosync(kern_pt1, pmap->pm_pt1, KERNBASE,
	    kernel_vm_end_new - 1);
	pte1_copy_nosync(kern_pt1, pmap->pm_pt1, vm_max_kernel_address,
	    0xFFFFFFFF);
	pt2tab_copy_nosync(kern_pt2tab, pmap->pm_pt2tab, KERNBASE,
	    kernel_vm_end_new - 1);
	pt2tab_copy_nosync(kern_pt2tab, pmap->pm_pt2tab, vm_max_kernel_address,
	    0xFFFFFFFF);
	LIST_INSERT_HEAD(&allpmaps, pmap, pm_list);
	mtx_unlock_spin(&allpmaps_lock);

	/*
	 * Store PT2MAP PT2 pages (a.k.a. PT2TAB) in PT2TAB itself.
	 * I.e. self reference mapping.  The PT2TAB is private, however mapped
	 * into shared PT2MAP space, so the mapping should be not global.
	 */
	pt2tab_pa = vtophys(pmap->pm_pt2tab);
	pte2p = pmap_pt2tab_entry(pmap, (vm_offset_t)PT2MAP);
	for (pa = pt2tab_pa, i = 0; i < NPG_IN_PT2TAB; i++, pa += PTE2_SIZE) {
		pt2tab_store(pte2p++, PTE2_KPT_NG(pa));
	}

	/* Insert PT2MAP PT2s into pmap PT1. */
	pte1p = pmap_pte1(pmap, (vm_offset_t)PT2MAP);
	for (pa = pt2tab_pa, i = 0; i < NPT2_IN_PT2TAB; i++, pa += NB_IN_PT2) {
		pte1_store(pte1p++, PTE1_LINK(pa));
	}

	/*
	 * Now synchronize new mapping which was made above.
	 */
	pte1_sync_range(pmap->pm_pt1, NB_IN_PT1);
	pte2_sync_range(pmap->pm_pt2tab, NB_IN_PT2TAB);

	CPU_ZERO(&pmap->pm_active);
	TAILQ_INIT(&pmap->pm_pvchunk);
	bzero(&pmap->pm_stats, sizeof pmap->pm_stats);

	return (1);
}

#ifdef INVARIANTS
static boolean_t
pt2tab_user_is_empty(pt2_entry_t *tab)
{
	u_int i, end;

	end = pt2tab_index(VM_MAXUSER_ADDRESS);
	for (i = 0; i < end; i++)
		if (tab[i] != 0) return (FALSE);
	return (TRUE);
}
#endif
/*
 *  Release any resources held by the given physical map.
 *  Called when a pmap initialized by pmap_pinit is being released.
 *  Should only be called if the map contains no valid mappings.
 */
void
pmap_release(pmap_t pmap)
{
#ifdef INVARIANTS
	vm_offset_t start, end;
#endif
	KASSERT(pmap->pm_stats.resident_count == 0,
	    ("%s: pmap resident count %ld != 0", __func__,
	    pmap->pm_stats.resident_count));
	KASSERT(pt2tab_user_is_empty(pmap->pm_pt2tab),
	    ("%s: has allocated user PT2(s)", __func__));
	KASSERT(CPU_EMPTY(&pmap->pm_active),
	    ("%s: pmap %p is active on some CPU(s)", __func__, pmap));

	mtx_lock_spin(&allpmaps_lock);
	LIST_REMOVE(pmap, pm_list);
	mtx_unlock_spin(&allpmaps_lock);

#ifdef INVARIANTS
	start = pte1_index(KERNBASE) * sizeof(pt1_entry_t);
	end = (pte1_index(0xFFFFFFFF) + 1) * sizeof(pt1_entry_t);
	bzero((char *)pmap->pm_pt1 + start, end - start);

	start = pt2tab_index(KERNBASE) * sizeof(pt2_entry_t);
	end = (pt2tab_index(0xFFFFFFFF) + 1) * sizeof(pt2_entry_t);
	bzero((char *)pmap->pm_pt2tab + start, end - start);
#endif
	/*
	 * We are leaving PT1 and PT2TAB allocated on released pmap,
	 * so hopefully UMA vmspace_zone will always be inited with
	 * UMA_ZONE_NOFREE flag.
	 */
}

/*********************************************************
 *
 *  L2 table pages and their pages management routines.
 *
 *********************************************************/

/*
 *  Virtual interface for L2 page table wire counting.
 *
 *  Each L2 page table in a page has own counter which counts a number of
 *  valid mappings in a table. Global page counter counts mappings in all
 *  tables in a page plus a single itself mapping in PT2TAB.
 *
 *  During a promotion we leave the associated L2 page table counter
 *  untouched, so the table (strictly speaking a page which holds it)
 *  is never freed if promoted.
 *
 *  If a page m->ref_count == 1 then no valid mappings exist in any L2 page
 *  table in the page and the page itself is only mapped in PT2TAB.
 */

static __inline void
pt2_wirecount_init(vm_page_t m)
{
	u_int i;

	/*
	 * Note: A page m is allocated with VM_ALLOC_WIRED flag and
	 *       m->ref_count should be already set correctly.
	 *       So, there is no need to set it again herein.
	 */
	for (i = 0; i < NPT2_IN_PG; i++)
		m->md.pt2_wirecount[i] = 0;
}

static __inline void
pt2_wirecount_inc(vm_page_t m, uint32_t pte1_idx)
{

	/*
	 * Note: A just modificated pte2 (i.e. already allocated)
	 *       is acquiring one extra reference which must be
	 *       explicitly cleared. It influences the KASSERTs herein.
	 *       All L2 page tables in a page always belong to the same
	 *       pmap, so we allow only one extra reference for the page.
	 */
	KASSERT(m->md.pt2_wirecount[pte1_idx & PT2PG_MASK] < (NPTE2_IN_PT2 + 1),
	    ("%s: PT2 is overflowing ...", __func__));
	KASSERT(m->ref_count <= (NPTE2_IN_PG + 1),
	    ("%s: PT2PG is overflowing ...", __func__));

	m->ref_count++;
	m->md.pt2_wirecount[pte1_idx & PT2PG_MASK]++;
}

static __inline void
pt2_wirecount_dec(vm_page_t m, uint32_t pte1_idx)
{

	KASSERT(m->md.pt2_wirecount[pte1_idx & PT2PG_MASK] != 0,
	    ("%s: PT2 is underflowing ...", __func__));
	KASSERT(m->ref_count > 1,
	    ("%s: PT2PG is underflowing ...", __func__));

	m->ref_count--;
	m->md.pt2_wirecount[pte1_idx & PT2PG_MASK]--;
}

static __inline void
pt2_wirecount_set(vm_page_t m, uint32_t pte1_idx, uint16_t count)
{

	KASSERT(count <= NPTE2_IN_PT2,
	    ("%s: invalid count %u", __func__, count));
	KASSERT(m->ref_count >  m->md.pt2_wirecount[pte1_idx & PT2PG_MASK],
	    ("%s: PT2PG corrupting (%u, %u) ...", __func__, m->ref_count,
	    m->md.pt2_wirecount[pte1_idx & PT2PG_MASK]));

	m->ref_count -= m->md.pt2_wirecount[pte1_idx & PT2PG_MASK];
	m->ref_count += count;
	m->md.pt2_wirecount[pte1_idx & PT2PG_MASK] = count;

	KASSERT(m->ref_count <= (NPTE2_IN_PG + 1),
	    ("%s: PT2PG is overflowed (%u) ...", __func__, m->ref_count));
}

static __inline uint32_t
pt2_wirecount_get(vm_page_t m, uint32_t pte1_idx)
{

	return (m->md.pt2_wirecount[pte1_idx & PT2PG_MASK]);
}

static __inline boolean_t
pt2_is_empty(vm_page_t m, vm_offset_t va)
{

	return (m->md.pt2_wirecount[pte1_index(va) & PT2PG_MASK] == 0);
}

static __inline boolean_t
pt2_is_full(vm_page_t m, vm_offset_t va)
{

	return (m->md.pt2_wirecount[pte1_index(va) & PT2PG_MASK] ==
	    NPTE2_IN_PT2);
}

static __inline boolean_t
pt2pg_is_empty(vm_page_t m)
{

	return (m->ref_count == 1);
}

/*
 *  This routine is called if the L2 page table
 *  is not mapped correctly.
 */
static vm_page_t
_pmap_allocpte2(pmap_t pmap, vm_offset_t va, u_int flags)
{
	uint32_t pte1_idx;
	pt1_entry_t *pte1p;
	pt2_entry_t pte2;
	vm_page_t  m;
	vm_paddr_t pt2pg_pa, pt2_pa;

	pte1_idx = pte1_index(va);
	pte1p = pmap->pm_pt1 + pte1_idx;

	KASSERT(pte1_load(pte1p) == 0,
	    ("%s: pm_pt1[%#x] is not zero: %#x", __func__, pte1_idx,
	    pte1_load(pte1p)));

	pte2 = pt2tab_load(pmap_pt2tab_entry(pmap, va));
	if (!pte2_is_valid(pte2)) {
		/*
		 * Install new PT2s page into pmap PT2TAB.
		 */
		m = vm_page_alloc(NULL, pte1_idx & ~PT2PG_MASK,
		    VM_ALLOC_NOOBJ | VM_ALLOC_WIRED | VM_ALLOC_ZERO);
		if (m == NULL) {
			if ((flags & PMAP_ENTER_NOSLEEP) == 0) {
				PMAP_UNLOCK(pmap);
				rw_wunlock(&pvh_global_lock);
				vm_wait(NULL);
				rw_wlock(&pvh_global_lock);
				PMAP_LOCK(pmap);
			}

			/*
			 * Indicate the need to retry.  While waiting,
			 * the L2 page table page may have been allocated.
			 */
			return (NULL);
		}
		pmap->pm_stats.resident_count++;
		pt2pg_pa = pmap_pt2pg_init(pmap, va, m);
	} else {
		pt2pg_pa = pte2_pa(pte2);
		m = PHYS_TO_VM_PAGE(pt2pg_pa);
	}

	pt2_wirecount_inc(m, pte1_idx);
	pt2_pa = page_pt2pa(pt2pg_pa, pte1_idx);
	pte1_store(pte1p, PTE1_LINK(pt2_pa));

	return (m);
}

static vm_page_t
pmap_allocpte2(pmap_t pmap, vm_offset_t va, u_int flags)
{
	u_int pte1_idx;
	pt1_entry_t *pte1p, pte1;
	vm_page_t m;

	pte1_idx = pte1_index(va);
retry:
	pte1p = pmap->pm_pt1 + pte1_idx;
	pte1 = pte1_load(pte1p);

	/*
	 * This supports switching from a 1MB page to a
	 * normal 4K page.
	 */
	if (pte1_is_section(pte1)) {
		(void)pmap_demote_pte1(pmap, pte1p, va);
		/*
		 * Reload pte1 after demotion.
		 *
		 * Note: Demotion can even fail as either PT2 is not find for
		 *       the virtual address or PT2PG can not be allocated.
		 */
		pte1 = pte1_load(pte1p);
	}

	/*
	 * If the L2 page table page is mapped, we just increment the
	 * hold count, and activate it.
	 */
	if (pte1_is_link(pte1)) {
		m = PHYS_TO_VM_PAGE(pte1_link_pa(pte1));
		pt2_wirecount_inc(m, pte1_idx);
	} else  {
		/*
		 * Here if the PT2 isn't mapped, or if it has
		 * been deallocated.
		 */
		m = _pmap_allocpte2(pmap, va, flags);
		if (m == NULL && (flags & PMAP_ENTER_NOSLEEP) == 0)
			goto retry;
	}

	return (m);
}

/*
 *  Schedule the specified unused L2 page table page to be freed. Specifically,
 *  add the page to the specified list of pages that will be released to the
 *  physical memory manager after the TLB has been updated.
 */
static __inline void
pmap_add_delayed_free_list(vm_page_t m, struct spglist *free)
{

	/*
	 * Put page on a list so that it is released after
	 * *ALL* TLB shootdown is done
	 */
#ifdef PMAP_DEBUG
	pmap_zero_page_check(m);
#endif
	m->flags |= PG_ZERO;
	SLIST_INSERT_HEAD(free, m, plinks.s.ss);
}

/*
 *  Unwire L2 page tables page.
 */
static void
pmap_unwire_pt2pg(pmap_t pmap, vm_offset_t va, vm_page_t m)
{
	pt1_entry_t *pte1p, opte1 __unused;
	pt2_entry_t *pte2p;
	uint32_t i;

	KASSERT(pt2pg_is_empty(m),
	    ("%s: pmap %p PT2PG %p wired", __func__, pmap, m));

	/*
	 * Unmap all L2 page tables in the page from L1 page table.
	 *
	 * QQQ: Individual L2 page tables (except the last one) can be unmapped
	 * earlier. However, we are doing that this way.
	 */
	KASSERT(m->pindex == (pte1_index(va) & ~PT2PG_MASK),
	    ("%s: pmap %p va %#x PT2PG %p bad index", __func__, pmap, va, m));
	pte1p = pmap->pm_pt1 + m->pindex;
	for (i = 0; i < NPT2_IN_PG; i++, pte1p++) {
		KASSERT(m->md.pt2_wirecount[i] == 0,
		    ("%s: pmap %p PT2 %u (PG %p) wired", __func__, pmap, i, m));
		opte1 = pte1_load(pte1p);
		if (pte1_is_link(opte1)) {
			pte1_clear(pte1p);
			/*
			 * Flush intermediate TLB cache.
			 */
			pmap_tlb_flush(pmap, (m->pindex + i) << PTE1_SHIFT);
		}
#ifdef INVARIANTS
		else
			KASSERT((opte1 == 0) || pte1_is_section(opte1),
			    ("%s: pmap %p va %#x bad pte1 %x at %u", __func__,
			    pmap, va, opte1, i));
#endif
	}

	/*
	 * Unmap the page from PT2TAB.
	 */
	pte2p = pmap_pt2tab_entry(pmap, va);
	(void)pt2tab_load_clear(pte2p);
	pmap_tlb_flush(pmap, pt2map_pt2pg(va));

	m->ref_count = 0;
	pmap->pm_stats.resident_count--;

	/*
	 * This barrier is so that the ordinary store unmapping
	 * the L2 page table page is globally performed before TLB shoot-
	 * down is begun.
	 */
	wmb();
	vm_wire_sub(1);
}

/*
 *  Decrements a L2 page table page's wire count, which is used to record the
 *  number of valid page table entries within the page.  If the wire count
 *  drops to zero, then the page table page is unmapped.  Returns TRUE if the
 *  page table page was unmapped and FALSE otherwise.
 */
static __inline boolean_t
pmap_unwire_pt2(pmap_t pmap, vm_offset_t va, vm_page_t m, struct spglist *free)
{
	pt2_wirecount_dec(m, pte1_index(va));
	if (pt2pg_is_empty(m)) {
		/*
		 * QQQ: Wire count is zero, so whole page should be zero and
		 *      we can set PG_ZERO flag to it.
		 *      Note that when promotion is enabled, it takes some
		 *      more efforts. See pmap_unwire_pt2_all() below.
		 */
		pmap_unwire_pt2pg(pmap, va, m);
		pmap_add_delayed_free_list(m, free);
		return (TRUE);
	} else
		return (FALSE);
}

/*
 *  Drop a L2 page table page's wire count at once, which is used to record
 *  the number of valid L2 page table entries within the page. If the wire
 *  count drops to zero, then the L2 page table page is unmapped.
 */
static __inline void
pmap_unwire_pt2_all(pmap_t pmap, vm_offset_t va, vm_page_t m,
    struct spglist *free)
{
	u_int pte1_idx = pte1_index(va);

	KASSERT(m->pindex == (pte1_idx & ~PT2PG_MASK),
		("%s: PT2 page's pindex is wrong", __func__));
	KASSERT(m->ref_count > pt2_wirecount_get(m, pte1_idx),
	    ("%s: bad pt2 wire count %u > %u", __func__, m->ref_count,
	    pt2_wirecount_get(m, pte1_idx)));

	/*
	 * It's possible that the L2 page table was never used.
	 * It happened in case that a section was created without promotion.
	 */
	if (pt2_is_full(m, va)) {
		pt2_wirecount_set(m, pte1_idx, 0);

		/*
		 * QQQ: We clear L2 page table now, so when L2 page table page
		 *      is going to be freed, we can set it PG_ZERO flag ...
		 *      This function is called only on section mappings, so
		 *      hopefully it's not to big overload.
		 *
		 * XXX: If pmap is current, existing PT2MAP mapping could be
		 *      used for zeroing.
		 */
		pmap_zero_page_area(m, page_pt2off(pte1_idx), NB_IN_PT2);
	}
#ifdef INVARIANTS
	else
		KASSERT(pt2_is_empty(m, va), ("%s: PT2 is not empty (%u)",
		    __func__, pt2_wirecount_get(m, pte1_idx)));
#endif
	if (pt2pg_is_empty(m)) {
		pmap_unwire_pt2pg(pmap, va, m);
		pmap_add_delayed_free_list(m, free);
	}
}

/*
 *  After removing a L2 page table entry, this routine is used to
 *  conditionally free the page, and manage the hold/wire counts.
 */
static boolean_t
pmap_unuse_pt2(pmap_t pmap, vm_offset_t va, struct spglist *free)
{
	pt1_entry_t pte1;
	vm_page_t mpte;

	if (va >= VM_MAXUSER_ADDRESS)
		return (FALSE);
	pte1 = pte1_load(pmap_pte1(pmap, va));
	mpte = PHYS_TO_VM_PAGE(pte1_link_pa(pte1));
	return (pmap_unwire_pt2(pmap, va, mpte, free));
}

/*************************************
 *
 *  Page management routines.
 *
 *************************************/

CTASSERT(sizeof(struct pv_chunk) == PAGE_SIZE);
CTASSERT(_NPCM == 11);
CTASSERT(_NPCPV == 336);

static __inline struct pv_chunk *
pv_to_chunk(pv_entry_t pv)
{

	return ((struct pv_chunk *)((uintptr_t)pv & ~(uintptr_t)PAGE_MASK));
}

#define PV_PMAP(pv) (pv_to_chunk(pv)->pc_pmap)

#define	PC_FREE0_9	0xfffffffful	/* Free values for index 0 through 9 */
#define	PC_FREE10	0x0000fffful	/* Free values for index 10 */

static const uint32_t pc_freemask[_NPCM] = {
	PC_FREE0_9, PC_FREE0_9, PC_FREE0_9,
	PC_FREE0_9, PC_FREE0_9, PC_FREE0_9,
	PC_FREE0_9, PC_FREE0_9, PC_FREE0_9,
	PC_FREE0_9, PC_FREE10
};

SYSCTL_INT(_vm_pmap, OID_AUTO, pv_entry_count, CTLFLAG_RD, &pv_entry_count, 0,
	"Current number of pv entries");

#ifdef PV_STATS
static int pc_chunk_count, pc_chunk_allocs, pc_chunk_frees, pc_chunk_tryfail;

SYSCTL_INT(_vm_pmap, OID_AUTO, pc_chunk_count, CTLFLAG_RD, &pc_chunk_count, 0,
    "Current number of pv entry chunks");
SYSCTL_INT(_vm_pmap, OID_AUTO, pc_chunk_allocs, CTLFLAG_RD, &pc_chunk_allocs, 0,
    "Current number of pv entry chunks allocated");
SYSCTL_INT(_vm_pmap, OID_AUTO, pc_chunk_frees, CTLFLAG_RD, &pc_chunk_frees, 0,
    "Current number of pv entry chunks frees");
SYSCTL_INT(_vm_pmap, OID_AUTO, pc_chunk_tryfail, CTLFLAG_RD, &pc_chunk_tryfail,
    0, "Number of times tried to get a chunk page but failed.");

static long pv_entry_frees, pv_entry_allocs;
static int pv_entry_spare;

SYSCTL_LONG(_vm_pmap, OID_AUTO, pv_entry_frees, CTLFLAG_RD, &pv_entry_frees, 0,
    "Current number of pv entry frees");
SYSCTL_LONG(_vm_pmap, OID_AUTO, pv_entry_allocs, CTLFLAG_RD, &pv_entry_allocs,
    0, "Current number of pv entry allocs");
SYSCTL_INT(_vm_pmap, OID_AUTO, pv_entry_spare, CTLFLAG_RD, &pv_entry_spare, 0,
    "Current number of spare pv entries");
#endif

/*
 *  Is given page managed?
 */
static __inline bool
is_managed(vm_paddr_t pa)
{
	vm_page_t m;

	m = PHYS_TO_VM_PAGE(pa);
	if (m == NULL)
		return (false);
	return ((m->oflags & VPO_UNMANAGED) == 0);
}

static __inline bool
pte1_is_managed(pt1_entry_t pte1)
{

	return (is_managed(pte1_pa(pte1)));
}

static __inline bool
pte2_is_managed(pt2_entry_t pte2)
{

	return (is_managed(pte2_pa(pte2)));
}

/*
 *  We are in a serious low memory condition.  Resort to
 *  drastic measures to free some pages so we can allocate
 *  another pv entry chunk.
 */
static vm_page_t
pmap_pv_reclaim(pmap_t locked_pmap)
{
	struct pch newtail;
	struct pv_chunk *pc;
	struct md_page *pvh;
	pt1_entry_t *pte1p;
	pmap_t pmap;
	pt2_entry_t *pte2p, tpte2;
	pv_entry_t pv;
	vm_offset_t va;
	vm_page_t m, m_pc;
	struct spglist free;
	uint32_t inuse;
	int bit, field, freed;

	PMAP_LOCK_ASSERT(locked_pmap, MA_OWNED);
	pmap = NULL;
	m_pc = NULL;
	SLIST_INIT(&free);
	TAILQ_INIT(&newtail);
	while ((pc = TAILQ_FIRST(&pv_chunks)) != NULL && (pv_vafree == 0 ||
	    SLIST_EMPTY(&free))) {
		TAILQ_REMOVE(&pv_chunks, pc, pc_lru);
		if (pmap != pc->pc_pmap) {
			if (pmap != NULL) {
				if (pmap != locked_pmap)
					PMAP_UNLOCK(pmap);
			}
			pmap = pc->pc_pmap;
			/* Avoid deadlock and lock recursion. */
			if (pmap > locked_pmap)
				PMAP_LOCK(pmap);
			else if (pmap != locked_pmap && !PMAP_TRYLOCK(pmap)) {
				pmap = NULL;
				TAILQ_INSERT_TAIL(&newtail, pc, pc_lru);
				continue;
			}
		}

		/*
		 * Destroy every non-wired, 4 KB page mapping in the chunk.
		 */
		freed = 0;
		for (field = 0; field < _NPCM; field++) {
			for (inuse = ~pc->pc_map[field] & pc_freemask[field];
			    inuse != 0; inuse &= ~(1UL << bit)) {
				bit = ffs(inuse) - 1;
				pv = &pc->pc_pventry[field * 32 + bit];
				va = pv->pv_va;
				pte1p = pmap_pte1(pmap, va);
				if (pte1_is_section(pte1_load(pte1p)))
					continue;
				pte2p = pmap_pte2(pmap, va);
				tpte2 = pte2_load(pte2p);
				if ((tpte2 & PTE2_W) == 0)
					tpte2 = pte2_load_clear(pte2p);
				pmap_pte2_release(pte2p);
				if ((tpte2 & PTE2_W) != 0)
					continue;
				KASSERT(tpte2 != 0,
				    ("pmap_pv_reclaim: pmap %p va %#x zero pte",
				    pmap, va));
				pmap_tlb_flush(pmap, va);
				m = PHYS_TO_VM_PAGE(pte2_pa(tpte2));
				if (pte2_is_dirty(tpte2))
					vm_page_dirty(m);
				if ((tpte2 & PTE2_A) != 0)
					vm_page_aflag_set(m, PGA_REFERENCED);
				TAILQ_REMOVE(&m->md.pv_list, pv, pv_next);
				if (TAILQ_EMPTY(&m->md.pv_list) &&
				    (m->flags & PG_FICTITIOUS) == 0) {
					pvh = pa_to_pvh(VM_PAGE_TO_PHYS(m));
					if (TAILQ_EMPTY(&pvh->pv_list)) {
						vm_page_aflag_clear(m,
						    PGA_WRITEABLE);
					}
				}
				pc->pc_map[field] |= 1UL << bit;
				pmap_unuse_pt2(pmap, va, &free);
				freed++;
			}
		}
		if (freed == 0) {
			TAILQ_INSERT_TAIL(&newtail, pc, pc_lru);
			continue;
		}
		/* Every freed mapping is for a 4 KB page. */
		pmap->pm_stats.resident_count -= freed;
		PV_STAT(pv_entry_frees += freed);
		PV_STAT(pv_entry_spare += freed);
		pv_entry_count -= freed;
		TAILQ_REMOVE(&pmap->pm_pvchunk, pc, pc_list);
		for (field = 0; field < _NPCM; field++)
			if (pc->pc_map[field] != pc_freemask[field]) {
				TAILQ_INSERT_HEAD(&pmap->pm_pvchunk, pc,
				    pc_list);
				TAILQ_INSERT_TAIL(&newtail, pc, pc_lru);

				/*
				 * One freed pv entry in locked_pmap is
				 * sufficient.
				 */
				if (pmap == locked_pmap)
					goto out;
				break;
			}
		if (field == _NPCM) {
			PV_STAT(pv_entry_spare -= _NPCPV);
			PV_STAT(pc_chunk_count--);
			PV_STAT(pc_chunk_frees++);
			/* Entire chunk is free; return it. */
			m_pc = PHYS_TO_VM_PAGE(pmap_kextract((vm_offset_t)pc));
			pmap_qremove((vm_offset_t)pc, 1);
			pmap_pte2list_free(&pv_vafree, (vm_offset_t)pc);
			break;
		}
	}
out:
	TAILQ_CONCAT(&pv_chunks, &newtail, pc_lru);
	if (pmap != NULL) {
		if (pmap != locked_pmap)
			PMAP_UNLOCK(pmap);
	}
	if (m_pc == NULL && pv_vafree != 0 && SLIST_EMPTY(&free)) {
		m_pc = SLIST_FIRST(&free);
		SLIST_REMOVE_HEAD(&free, plinks.s.ss);
		/* Recycle a freed page table page. */
		m_pc->ref_count = 1;
		vm_wire_add(1);
	}
	vm_page_free_pages_toq(&free, false);
	return (m_pc);
}

static void
free_pv_chunk(struct pv_chunk *pc)
{
	vm_page_t m;

	TAILQ_REMOVE(&pv_chunks, pc, pc_lru);
	PV_STAT(pv_entry_spare -= _NPCPV);
	PV_STAT(pc_chunk_count--);
	PV_STAT(pc_chunk_frees++);
	/* entire chunk is free, return it */
	m = PHYS_TO_VM_PAGE(pmap_kextract((vm_offset_t)pc));
	pmap_qremove((vm_offset_t)pc, 1);
	vm_page_unwire_noq(m);
	vm_page_free(m);
	pmap_pte2list_free(&pv_vafree, (vm_offset_t)pc);
}

/*
 *  Free the pv_entry back to the free list.
 */
static void
free_pv_entry(pmap_t pmap, pv_entry_t pv)
{
	struct pv_chunk *pc;
	int idx, field, bit;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	PMAP_LOCK_ASSERT(pmap, MA_OWNED);
	PV_STAT(pv_entry_frees++);
	PV_STAT(pv_entry_spare++);
	pv_entry_count--;
	pc = pv_to_chunk(pv);
	idx = pv - &pc->pc_pventry[0];
	field = idx / 32;
	bit = idx % 32;
	pc->pc_map[field] |= 1ul << bit;
	for (idx = 0; idx < _NPCM; idx++)
		if (pc->pc_map[idx] != pc_freemask[idx]) {
			/*
			 * 98% of the time, pc is already at the head of the
			 * list.  If it isn't already, move it to the head.
			 */
			if (__predict_false(TAILQ_FIRST(&pmap->pm_pvchunk) !=
			    pc)) {
				TAILQ_REMOVE(&pmap->pm_pvchunk, pc, pc_list);
				TAILQ_INSERT_HEAD(&pmap->pm_pvchunk, pc,
				    pc_list);
			}
			return;
		}
	TAILQ_REMOVE(&pmap->pm_pvchunk, pc, pc_list);
	free_pv_chunk(pc);
}

/*
 *  Get a new pv_entry, allocating a block from the system
 *  when needed.
 */
static pv_entry_t
get_pv_entry(pmap_t pmap, boolean_t try)
{
	static const struct timeval printinterval = { 60, 0 };
	static struct timeval lastprint;
	int bit, field;
	pv_entry_t pv;
	struct pv_chunk *pc;
	vm_page_t m;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	PMAP_LOCK_ASSERT(pmap, MA_OWNED);
	PV_STAT(pv_entry_allocs++);
	pv_entry_count++;
	if (pv_entry_count > pv_entry_high_water)
		if (ratecheck(&lastprint, &printinterval))
			printf("Approaching the limit on PV entries, consider "
			    "increasing either the vm.pmap.shpgperproc or the "
			    "vm.pmap.pv_entries tunable.\n");
retry:
	pc = TAILQ_FIRST(&pmap->pm_pvchunk);
	if (pc != NULL) {
		for (field = 0; field < _NPCM; field++) {
			if (pc->pc_map[field]) {
				bit = ffs(pc->pc_map[field]) - 1;
				break;
			}
		}
		if (field < _NPCM) {
			pv = &pc->pc_pventry[field * 32 + bit];
			pc->pc_map[field] &= ~(1ul << bit);
			/* If this was the last item, move it to tail */
			for (field = 0; field < _NPCM; field++)
				if (pc->pc_map[field] != 0) {
					PV_STAT(pv_entry_spare--);
					return (pv);	/* not full, return */
				}
			TAILQ_REMOVE(&pmap->pm_pvchunk, pc, pc_list);
			TAILQ_INSERT_TAIL(&pmap->pm_pvchunk, pc, pc_list);
			PV_STAT(pv_entry_spare--);
			return (pv);
		}
	}
	/*
	 * Access to the pte2list "pv_vafree" is synchronized by the pvh
	 * global lock.  If "pv_vafree" is currently non-empty, it will
	 * remain non-empty until pmap_pte2list_alloc() completes.
	 */
	if (pv_vafree == 0 || (m = vm_page_alloc(NULL, 0, VM_ALLOC_NORMAL |
	    VM_ALLOC_NOOBJ | VM_ALLOC_WIRED)) == NULL) {
		if (try) {
			pv_entry_count--;
			PV_STAT(pc_chunk_tryfail++);
			return (NULL);
		}
		m = pmap_pv_reclaim(pmap);
		if (m == NULL)
			goto retry;
	}
	PV_STAT(pc_chunk_count++);
	PV_STAT(pc_chunk_allocs++);
	pc = (struct pv_chunk *)pmap_pte2list_alloc(&pv_vafree);
	pmap_qenter((vm_offset_t)pc, &m, 1);
	pc->pc_pmap = pmap;
	pc->pc_map[0] = pc_freemask[0] & ~1ul;	/* preallocated bit 0 */
	for (field = 1; field < _NPCM; field++)
		pc->pc_map[field] = pc_freemask[field];
	TAILQ_INSERT_TAIL(&pv_chunks, pc, pc_lru);
	pv = &pc->pc_pventry[0];
	TAILQ_INSERT_HEAD(&pmap->pm_pvchunk, pc, pc_list);
	PV_STAT(pv_entry_spare += _NPCPV - 1);
	return (pv);
}

/*
 *  Create a pv entry for page at pa for
 *  (pmap, va).
 */
static void
pmap_insert_entry(pmap_t pmap, vm_offset_t va, vm_page_t m)
{
	pv_entry_t pv;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	PMAP_LOCK_ASSERT(pmap, MA_OWNED);
	pv = get_pv_entry(pmap, FALSE);
	pv->pv_va = va;
	TAILQ_INSERT_TAIL(&m->md.pv_list, pv, pv_next);
}

static __inline pv_entry_t
pmap_pvh_remove(struct md_page *pvh, pmap_t pmap, vm_offset_t va)
{
	pv_entry_t pv;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	TAILQ_FOREACH(pv, &pvh->pv_list, pv_next) {
		if (pmap == PV_PMAP(pv) && va == pv->pv_va) {
			TAILQ_REMOVE(&pvh->pv_list, pv, pv_next);
			break;
		}
	}
	return (pv);
}

static void
pmap_pvh_free(struct md_page *pvh, pmap_t pmap, vm_offset_t va)
{
	pv_entry_t pv;

	pv = pmap_pvh_remove(pvh, pmap, va);
	KASSERT(pv != NULL, ("pmap_pvh_free: pv not found"));
	free_pv_entry(pmap, pv);
}

static void
pmap_remove_entry(pmap_t pmap, vm_page_t m, vm_offset_t va)
{
	struct md_page *pvh;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	pmap_pvh_free(&m->md, pmap, va);
	if (TAILQ_EMPTY(&m->md.pv_list) && (m->flags & PG_FICTITIOUS) == 0) {
		pvh = pa_to_pvh(VM_PAGE_TO_PHYS(m));
		if (TAILQ_EMPTY(&pvh->pv_list))
			vm_page_aflag_clear(m, PGA_WRITEABLE);
	}
}

static void
pmap_pv_demote_pte1(pmap_t pmap, vm_offset_t va, vm_paddr_t pa)
{
	struct md_page *pvh;
	pv_entry_t pv;
	vm_offset_t va_last;
	vm_page_t m;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	KASSERT((pa & PTE1_OFFSET) == 0,
	    ("pmap_pv_demote_pte1: pa is not 1mpage aligned"));

	/*
	 * Transfer the 1mpage's pv entry for this mapping to the first
	 * page's pv list.
	 */
	pvh = pa_to_pvh(pa);
	va = pte1_trunc(va);
	pv = pmap_pvh_remove(pvh, pmap, va);
	KASSERT(pv != NULL, ("pmap_pv_demote_pte1: pv not found"));
	m = PHYS_TO_VM_PAGE(pa);
	TAILQ_INSERT_TAIL(&m->md.pv_list, pv, pv_next);
	/* Instantiate the remaining NPTE2_IN_PT2 - 1 pv entries. */
	va_last = va + PTE1_SIZE - PAGE_SIZE;
	do {
		m++;
		KASSERT((m->oflags & VPO_UNMANAGED) == 0,
		    ("pmap_pv_demote_pte1: page %p is not managed", m));
		va += PAGE_SIZE;
		pmap_insert_entry(pmap, va, m);
	} while (va < va_last);
}

#if VM_NRESERVLEVEL > 0
static void
pmap_pv_promote_pte1(pmap_t pmap, vm_offset_t va, vm_paddr_t pa)
{
	struct md_page *pvh;
	pv_entry_t pv;
	vm_offset_t va_last;
	vm_page_t m;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	KASSERT((pa & PTE1_OFFSET) == 0,
	    ("pmap_pv_promote_pte1: pa is not 1mpage aligned"));

	/*
	 * Transfer the first page's pv entry for this mapping to the
	 * 1mpage's pv list.  Aside from avoiding the cost of a call
	 * to get_pv_entry(), a transfer avoids the possibility that
	 * get_pv_entry() calls pmap_pv_reclaim() and that pmap_pv_reclaim()
	 * removes one of the mappings that is being promoted.
	 */
	m = PHYS_TO_VM_PAGE(pa);
	va = pte1_trunc(va);
	pv = pmap_pvh_remove(&m->md, pmap, va);
	KASSERT(pv != NULL, ("pmap_pv_promote_pte1: pv not found"));
	pvh = pa_to_pvh(pa);
	TAILQ_INSERT_TAIL(&pvh->pv_list, pv, pv_next);
	/* Free the remaining NPTE2_IN_PT2 - 1 pv entries. */
	va_last = va + PTE1_SIZE - PAGE_SIZE;
	do {
		m++;
		va += PAGE_SIZE;
		pmap_pvh_free(&m->md, pmap, va);
	} while (va < va_last);
}
#endif

/*
 *  Conditionally create a pv entry.
 */
static boolean_t
pmap_try_insert_pv_entry(pmap_t pmap, vm_offset_t va, vm_page_t m)
{
	pv_entry_t pv;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	PMAP_LOCK_ASSERT(pmap, MA_OWNED);
	if (pv_entry_count < pv_entry_high_water &&
	    (pv = get_pv_entry(pmap, TRUE)) != NULL) {
		pv->pv_va = va;
		TAILQ_INSERT_TAIL(&m->md.pv_list, pv, pv_next);
		return (TRUE);
	} else
		return (FALSE);
}

/*
 *  Create the pv entries for each of the pages within a section.
 */
static bool
pmap_pv_insert_pte1(pmap_t pmap, vm_offset_t va, pt1_entry_t pte1, u_int flags)
{
	struct md_page *pvh;
	pv_entry_t pv;
	bool noreclaim;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	noreclaim = (flags & PMAP_ENTER_NORECLAIM) != 0;
	if ((noreclaim && pv_entry_count >= pv_entry_high_water) ||
	    (pv = get_pv_entry(pmap, noreclaim)) == NULL)
		return (false);
	pv->pv_va = va;
	pvh = pa_to_pvh(pte1_pa(pte1));
	TAILQ_INSERT_TAIL(&pvh->pv_list, pv, pv_next);
	return (true);
}

static inline void
pmap_tlb_flush_pte1(pmap_t pmap, vm_offset_t va, pt1_entry_t npte1)
{

	/* Kill all the small mappings or the big one only. */
	if (pte1_is_section(npte1))
		pmap_tlb_flush_range(pmap, pte1_trunc(va), PTE1_SIZE);
	else
		pmap_tlb_flush(pmap, pte1_trunc(va));
}

/*
 *  Update kernel pte1 on all pmaps.
 *
 *  The following function is called only on one cpu with disabled interrupts.
 *  In SMP case, smp_rendezvous_cpus() is used to stop other cpus. This way
 *  nobody can invoke explicit hardware table walk during the update of pte1.
 *  Unsolicited hardware table walk can still happen, invoked by speculative
 *  data or instruction prefetch or even by speculative hardware table walk.
 *
 *  The break-before-make approach should be implemented here. However, it's
 *  not so easy to do that for kernel mappings as it would be unhappy to unmap
 *  itself unexpectedly but voluntarily.
 */
static void
pmap_update_pte1_kernel(vm_offset_t va, pt1_entry_t npte1)
{
	pmap_t pmap;
	pt1_entry_t *pte1p;

	/*
	 * Get current pmap. Interrupts should be disabled here
	 * so PCPU_GET() is done atomically.
	 */
	pmap = PCPU_GET(curpmap);
	if (pmap == NULL)
		pmap = kernel_pmap;

	/*
	 * (1) Change pte1 on current pmap.
	 * (2) Flush all obsolete TLB entries on current CPU.
	 * (3) Change pte1 on all pmaps.
	 * (4) Flush all obsolete TLB entries on all CPUs in SMP case.
	 */

	pte1p = pmap_pte1(pmap, va);
	pte1_store(pte1p, npte1);

	/* Kill all the small mappings or the big one only. */
	if (pte1_is_section(npte1)) {
		pmap_pte1_kern_promotions++;
		tlb_flush_range_local(pte1_trunc(va), PTE1_SIZE);
	} else {
		pmap_pte1_kern_demotions++;
		tlb_flush_local(pte1_trunc(va));
	}

	/*
	 * In SMP case, this function is called when all cpus are at smp
	 * rendezvous, so there is no need to use 'allpmaps_lock' lock here.
	 * In UP case, the function is called with this lock locked.
	 */
	LIST_FOREACH(pmap, &allpmaps, pm_list) {
		pte1p = pmap_pte1(pmap, va);
		pte1_store(pte1p, npte1);
	}

#ifdef SMP
	/* Kill all the small mappings or the big one only. */
	if (pte1_is_section(npte1))
		tlb_flush_range(pte1_trunc(va), PTE1_SIZE);
	else
		tlb_flush(pte1_trunc(va));
#endif
}

#ifdef SMP
struct pte1_action {
	vm_offset_t va;
	pt1_entry_t npte1;
	u_int update;		/* CPU that updates the PTE1 */
};

static void
pmap_update_pte1_action(void *arg)
{
	struct pte1_action *act = arg;

	if (act->update == PCPU_GET(cpuid))
		pmap_update_pte1_kernel(act->va, act->npte1);
}

/*
 *  Change pte1 on current pmap.
 *  Note that kernel pte1 must be changed on all pmaps.
 *
 *  According to the architecture reference manual published by ARM,
 *  the behaviour is UNPREDICTABLE when two or more TLB entries map the same VA.
 *  According to this manual, UNPREDICTABLE behaviours must never happen in
 *  a viable system. In contrast, on x86 processors, it is not specified which
 *  TLB entry mapping the virtual address will be used, but the MMU doesn't
 *  generate a bogus translation the way it does on Cortex-A8 rev 2 (Beaglebone
 *  Black).
 *
 *  It's a problem when either promotion or demotion is being done. The pte1
 *  update and appropriate TLB flush must be done atomically in general.
 */
static void
pmap_change_pte1(pmap_t pmap, pt1_entry_t *pte1p, vm_offset_t va,
    pt1_entry_t npte1)
{

	if (pmap == kernel_pmap) {
		struct pte1_action act;

		sched_pin();
		act.va = va;
		act.npte1 = npte1;
		act.update = PCPU_GET(cpuid);
		smp_rendezvous_cpus(all_cpus, smp_no_rendezvous_barrier,
		    pmap_update_pte1_action, NULL, &act);
		sched_unpin();
	} else {
		register_t cspr;

		/*
		 * Use break-before-make approach for changing userland
		 * mappings. It can cause L1 translation aborts on other
		 * cores in SMP case. So, special treatment is implemented
		 * in pmap_fault(). To reduce the likelihood that another core
		 * will be affected by the broken mapping, disable interrupts
		 * until the mapping change is completed.
		 */
		cspr = disable_interrupts(PSR_I | PSR_F);
		pte1_clear(pte1p);
		pmap_tlb_flush_pte1(pmap, va, npte1);
		pte1_store(pte1p, npte1);
		restore_interrupts(cspr);
	}
}
#else
static void
pmap_change_pte1(pmap_t pmap, pt1_entry_t *pte1p, vm_offset_t va,
    pt1_entry_t npte1)
{

	if (pmap == kernel_pmap) {
		mtx_lock_spin(&allpmaps_lock);
		pmap_update_pte1_kernel(va, npte1);
		mtx_unlock_spin(&allpmaps_lock);
	} else {
		register_t cspr;

		/*
		 * Use break-before-make approach for changing userland
		 * mappings. It's absolutely safe in UP case when interrupts
		 * are disabled.
		 */
		cspr = disable_interrupts(PSR_I | PSR_F);
		pte1_clear(pte1p);
		pmap_tlb_flush_pte1(pmap, va, npte1);
		pte1_store(pte1p, npte1);
		restore_interrupts(cspr);
	}
}
#endif

#if VM_NRESERVLEVEL > 0
/*
 *  Tries to promote the NPTE2_IN_PT2, contiguous 4KB page mappings that are
 *  within a single page table page (PT2) to a single 1MB page mapping.
 *  For promotion to occur, two conditions must be met: (1) the 4KB page
 *  mappings must map aligned, contiguous physical memory and (2) the 4KB page
 *  mappings must have identical characteristics.
 *
 *  Managed (PG_MANAGED) mappings within the kernel address space are not
 *  promoted.  The reason is that kernel PTE1s are replicated in each pmap but
 *  pmap_remove_write(), pmap_clear_modify(), and pmap_clear_reference() only
 *  read the PTE1 from the kernel pmap.
 */
static void
pmap_promote_pte1(pmap_t pmap, pt1_entry_t *pte1p, vm_offset_t va)
{
	pt1_entry_t npte1;
	pt2_entry_t *fpte2p, fpte2, fpte2_fav;
	pt2_entry_t *pte2p, pte2;
	vm_offset_t pteva __unused;
	vm_page_t m __unused;

	PDEBUG(6, printf("%s(%p): try for va %#x pte1 %#x at %p\n", __func__,
	    pmap, va, pte1_load(pte1p), pte1p));

	PMAP_LOCK_ASSERT(pmap, MA_OWNED);

	/*
	 * Examine the first PTE2 in the specified PT2. Abort if this PTE2 is
	 * either invalid, unused, or does not map the first 4KB physical page
	 * within a 1MB page.
	 */
	fpte2p = pmap_pte2_quick(pmap, pte1_trunc(va));
	fpte2 = pte2_load(fpte2p);
	if ((fpte2 & ((PTE2_FRAME & PTE1_OFFSET) | PTE2_A | PTE2_V)) !=
	    (PTE2_A | PTE2_V)) {
		pmap_pte1_p_failures++;
		CTR3(KTR_PMAP, "%s: failure(1) for va %#x in pmap %p",
		    __func__, va, pmap);
		return;
	}
	if (pte2_is_managed(fpte2) && pmap == kernel_pmap) {
		pmap_pte1_p_failures++;
		CTR3(KTR_PMAP, "%s: failure(2) for va %#x in pmap %p",
		    __func__, va, pmap);
		return;
	}
	if ((fpte2 & (PTE2_NM | PTE2_RO)) == PTE2_NM) {
		/*
		 * When page is not modified, PTE2_RO can be set without
		 * a TLB invalidation.
		 */
		fpte2 |= PTE2_RO;
		pte2_store(fpte2p, fpte2);
	}

	/*
	 * Examine each of the other PTE2s in the specified PT2. Abort if this
	 * PTE2 maps an unexpected 4KB physical page or does not have identical
	 * characteristics to the first PTE2.
	 */
	fpte2_fav = (fpte2 & (PTE2_FRAME | PTE2_A | PTE2_V));
	fpte2_fav += PTE1_SIZE - PTE2_SIZE; /* examine from the end */
	for (pte2p = fpte2p + NPTE2_IN_PT2 - 1; pte2p > fpte2p; pte2p--) {
		pte2 = pte2_load(pte2p);
		if ((pte2 & (PTE2_FRAME | PTE2_A | PTE2_V)) != fpte2_fav) {
			pmap_pte1_p_failures++;
			CTR3(KTR_PMAP, "%s: failure(3) for va %#x in pmap %p",
			    __func__, va, pmap);
			return;
		}
		if ((pte2 & (PTE2_NM | PTE2_RO)) == PTE2_NM) {
			/*
			 * When page is not modified, PTE2_RO can be set
			 * without a TLB invalidation. See note above.
			 */
			pte2 |= PTE2_RO;
			pte2_store(pte2p, pte2);
			pteva = pte1_trunc(va) | (pte2 & PTE1_OFFSET &
			    PTE2_FRAME);
			CTR3(KTR_PMAP, "%s: protect for va %#x in pmap %p",
			    __func__, pteva, pmap);
		}
		if ((pte2 & PTE2_PROMOTE) != (fpte2 & PTE2_PROMOTE)) {
			pmap_pte1_p_failures++;
			CTR3(KTR_PMAP, "%s: failure(4) for va %#x in pmap %p",
			    __func__, va, pmap);
			return;
		}

		fpte2_fav -= PTE2_SIZE;
	}
	/*
	 * The page table page in its current state will stay in PT2TAB
	 * until the PTE1 mapping the section is demoted by pmap_demote_pte1()
	 * or destroyed by pmap_remove_pte1().
	 *
	 * Note that L2 page table size is not equal to PAGE_SIZE.
	 */
	m = PHYS_TO_VM_PAGE(trunc_page(pte1_link_pa(pte1_load(pte1p))));
	KASSERT(m >= vm_page_array && m < &vm_page_array[vm_page_array_size],
	    ("%s: PT2 page is out of range", __func__));
	KASSERT(m->pindex == (pte1_index(va) & ~PT2PG_MASK),
	    ("%s: PT2 page's pindex is wrong", __func__));

	/*
	 * Get pte1 from pte2 format.
	 */
	npte1 = (fpte2 & PTE1_FRAME) | ATTR_TO_L1(fpte2) | PTE1_V;

	/*
	 * Promote the pv entries.
	 */
	if (pte2_is_managed(fpte2))
		pmap_pv_promote_pte1(pmap, va, pte1_pa(npte1));

	/*
	 * Promote the mappings.
	 */
	pmap_change_pte1(pmap, pte1p, va, npte1);

	pmap_pte1_promotions++;
	CTR3(KTR_PMAP, "%s: success for va %#x in pmap %p",
	    __func__, va, pmap);

	PDEBUG(6, printf("%s(%p): success for va %#x pte1 %#x(%#x) at %p\n",
	    __func__, pmap, va, npte1, pte1_load(pte1p), pte1p));
}
#endif /* VM_NRESERVLEVEL > 0 */

/*
 *  Zero L2 page table page.
 */
static __inline void
pmap_clear_pt2(pt2_entry_t *fpte2p)
{
	pt2_entry_t *pte2p;

	for (pte2p = fpte2p; pte2p < fpte2p + NPTE2_IN_PT2; pte2p++)
		pte2_clear(pte2p);

}

/*
 *  Removes a 1MB page mapping from the kernel pmap.
 */
static void
pmap_remove_kernel_pte1(pmap_t pmap, pt1_entry_t *pte1p, vm_offset_t va)
{
	vm_page_t m;
	uint32_t pte1_idx;
	pt2_entry_t *fpte2p;
	vm_paddr_t pt2_pa;

	PMAP_LOCK_ASSERT(pmap, MA_OWNED);
	m = pmap_pt2_page(pmap, va);
	if (m == NULL)
		/*
		 * QQQ: Is this function called only on promoted pte1?
		 *      We certainly do section mappings directly
		 *      (without promotion) in kernel !!!
		 */
		panic("%s: missing pt2 page", __func__);

	pte1_idx = pte1_index(va);

	/*
	 * Initialize the L2 page table.
	 */
	fpte2p = page_pt2(pt2map_pt2pg(va), pte1_idx);
	pmap_clear_pt2(fpte2p);

	/*
	 * Remove the mapping.
	 */
	pt2_pa = page_pt2pa(VM_PAGE_TO_PHYS(m), pte1_idx);
	pmap_kenter_pte1(va, PTE1_LINK(pt2_pa));

	/*
	 * QQQ: We do not need to invalidate PT2MAP mapping
	 * as we did not change it. I.e. the L2 page table page
	 * was and still is mapped the same way.
	 */
}

/*
 *  Do the things to unmap a section in a process
 */
static void
pmap_remove_pte1(pmap_t pmap, pt1_entry_t *pte1p, vm_offset_t sva,
    struct spglist *free)
{
	pt1_entry_t opte1;
	struct md_page *pvh;
	vm_offset_t eva, va;
	vm_page_t m;

	PDEBUG(6, printf("%s(%p): va %#x pte1 %#x at %p\n", __func__, pmap, sva,
	    pte1_load(pte1p), pte1p));

	PMAP_LOCK_ASSERT(pmap, MA_OWNED);
	KASSERT((sva & PTE1_OFFSET) == 0,
	    ("%s: sva is not 1mpage aligned", __func__));

	/*
	 * Clear and invalidate the mapping. It should occupy one and only TLB
	 * entry. So, pmap_tlb_flush() called with aligned address should be
	 * sufficient.
	 */
	opte1 = pte1_load_clear(pte1p);
	pmap_tlb_flush(pmap, sva);

	if (pte1_is_wired(opte1))
		pmap->pm_stats.wired_count -= PTE1_SIZE / PAGE_SIZE;
	pmap->pm_stats.resident_count -= PTE1_SIZE / PAGE_SIZE;
	if (pte1_is_managed(opte1)) {
		pvh = pa_to_pvh(pte1_pa(opte1));
		pmap_pvh_free(pvh, pmap, sva);
		eva = sva + PTE1_SIZE;
		for (va = sva, m = PHYS_TO_VM_PAGE(pte1_pa(opte1));
		    va < eva; va += PAGE_SIZE, m++) {
			if (pte1_is_dirty(opte1))
				vm_page_dirty(m);
			if (opte1 & PTE1_A)
				vm_page_aflag_set(m, PGA_REFERENCED);
			if (TAILQ_EMPTY(&m->md.pv_list) &&
			    TAILQ_EMPTY(&pvh->pv_list))
				vm_page_aflag_clear(m, PGA_WRITEABLE);
		}
	}
	if (pmap == kernel_pmap) {
		/*
		 * L2 page table(s) can't be removed from kernel map as
		 * kernel counts on it (stuff around pmap_growkernel()).
		 */
		 pmap_remove_kernel_pte1(pmap, pte1p, sva);
	} else {
		/*
		 * Get associated L2 page table page.
		 * It's possible that the page was never allocated.
		 */
		m = pmap_pt2_page(pmap, sva);
		if (m != NULL)
			pmap_unwire_pt2_all(pmap, sva, m, free);
	}
}

/*
 *  Fills L2 page table page with mappings to consecutive physical pages.
 */
static __inline void
pmap_fill_pt2(pt2_entry_t *fpte2p, pt2_entry_t npte2)
{
	pt2_entry_t *pte2p;

	for (pte2p = fpte2p; pte2p < fpte2p + NPTE2_IN_PT2; pte2p++) {
		pte2_store(pte2p, npte2);
		npte2 += PTE2_SIZE;
	}
}

/*
 *  Tries to demote a 1MB page mapping. If demotion fails, the
 *  1MB page mapping is invalidated.
 */
static boolean_t
pmap_demote_pte1(pmap_t pmap, pt1_entry_t *pte1p, vm_offset_t va)
{
	pt1_entry_t opte1, npte1;
	pt2_entry_t *fpte2p, npte2;
	vm_paddr_t pt2pg_pa, pt2_pa;
	vm_page_t m;
	struct spglist free;
	uint32_t pte1_idx, isnew = 0;

	PDEBUG(6, printf("%s(%p): try for va %#x pte1 %#x at %p\n", __func__,
	    pmap, va, pte1_load(pte1p), pte1p));

	PMAP_LOCK_ASSERT(pmap, MA_OWNED);

	opte1 = pte1_load(pte1p);
	KASSERT(pte1_is_section(opte1), ("%s: opte1 not a section", __func__));

	if ((opte1 & PTE1_A) == 0 || (m = pmap_pt2_page(pmap, va)) == NULL) {
		KASSERT(!pte1_is_wired(opte1),
		    ("%s: PT2 page for a wired mapping is missing", __func__));

		/*
		 * Invalidate the 1MB page mapping and return
		 * "failure" if the mapping was never accessed or the
		 * allocation of the new page table page fails.
		 */
		if ((opte1 & PTE1_A) == 0 || (m = vm_page_alloc(NULL,
		    pte1_index(va) & ~PT2PG_MASK, VM_ALLOC_NOOBJ |
		    VM_ALLOC_NORMAL | VM_ALLOC_WIRED)) == NULL) {
			SLIST_INIT(&free);
			pmap_remove_pte1(pmap, pte1p, pte1_trunc(va), &free);
			vm_page_free_pages_toq(&free, false);
			CTR3(KTR_PMAP, "%s: failure for va %#x in pmap %p",
			    __func__, va, pmap);
			return (FALSE);
		}
		if (va < VM_MAXUSER_ADDRESS)
			pmap->pm_stats.resident_count++;

		isnew = 1;

		/*
		 * We init all L2 page tables in the page even if
		 * we are going to change everything for one L2 page
		 * table in a while.
		 */
		pt2pg_pa = pmap_pt2pg_init(pmap, va, m);
	} else {
		if (va < VM_MAXUSER_ADDRESS) {
			if (pt2_is_empty(m, va))
				isnew = 1; /* Demoting section w/o promotion. */
#ifdef INVARIANTS
			else
				KASSERT(pt2_is_full(m, va), ("%s: bad PT2 wire"
				    " count %u", __func__,
				    pt2_wirecount_get(m, pte1_index(va))));
#endif
		}
	}

	pt2pg_pa = VM_PAGE_TO_PHYS(m);
	pte1_idx = pte1_index(va);
	/*
	 * If the pmap is current, then the PT2MAP can provide access to
	 * the page table page (promoted L2 page tables are not unmapped).
	 * Otherwise, temporarily map the L2 page table page (m) into
	 * the kernel's address space at either PADDR1 or PADDR2.
	 *
	 * Note that L2 page table size is not equal to PAGE_SIZE.
	 */
	if (pmap_is_current(pmap))
		fpte2p = page_pt2(pt2map_pt2pg(va), pte1_idx);
	else if (curthread->td_pinned > 0 && rw_wowned(&pvh_global_lock)) {
		if (pte2_pa(pte2_load(PMAP1)) != pt2pg_pa) {
			pte2_store(PMAP1, PTE2_KPT(pt2pg_pa));
#ifdef SMP
			PMAP1cpu = PCPU_GET(cpuid);
#endif
			tlb_flush_local((vm_offset_t)PADDR1);
			PMAP1changed++;
		} else
#ifdef SMP
		if (PMAP1cpu != PCPU_GET(cpuid)) {
			PMAP1cpu = PCPU_GET(cpuid);
			tlb_flush_local((vm_offset_t)PADDR1);
			PMAP1changedcpu++;
		} else
#endif
			PMAP1unchanged++;
		fpte2p = page_pt2((vm_offset_t)PADDR1, pte1_idx);
	} else {
		mtx_lock(&PMAP2mutex);
		if (pte2_pa(pte2_load(PMAP2)) != pt2pg_pa) {
			pte2_store(PMAP2, PTE2_KPT(pt2pg_pa));
			tlb_flush((vm_offset_t)PADDR2);
		}
		fpte2p = page_pt2((vm_offset_t)PADDR2, pte1_idx);
	}
	pt2_pa = page_pt2pa(pt2pg_pa, pte1_idx);
	npte1 = PTE1_LINK(pt2_pa);

	KASSERT((opte1 & PTE1_A) != 0,
	    ("%s: opte1 is missing PTE1_A", __func__));
	KASSERT((opte1 & (PTE1_NM | PTE1_RO)) != PTE1_NM,
	    ("%s: opte1 has PTE1_NM", __func__));

	/*
	 *  Get pte2 from pte1 format.
	*/
	npte2 = pte1_pa(opte1) | ATTR_TO_L2(opte1) | PTE2_V;

	/*
	 * If the L2 page table page is new, initialize it. If the mapping
	 * has changed attributes, update the page table entries.
	 */
	if (isnew != 0) {
		pt2_wirecount_set(m, pte1_idx, NPTE2_IN_PT2);
		pmap_fill_pt2(fpte2p, npte2);
	} else if ((pte2_load(fpte2p) & PTE2_PROMOTE) !=
		    (npte2 & PTE2_PROMOTE))
		pmap_fill_pt2(fpte2p, npte2);

	KASSERT(pte2_pa(pte2_load(fpte2p)) == pte2_pa(npte2),
	    ("%s: fpte2p and npte2 map different physical addresses",
	    __func__));

	if (fpte2p == PADDR2)
		mtx_unlock(&PMAP2mutex);

	/*
	 * Demote the mapping. This pmap is locked. The old PTE1 has
	 * PTE1_A set. If the old PTE1 has not PTE1_RO set, it also
	 * has not PTE1_NM set. Thus, there is no danger of a race with
	 * another processor changing the setting of PTE1_A and/or PTE1_NM
	 * between the read above and the store below.
	 */
	pmap_change_pte1(pmap, pte1p, va, npte1);

	/*
	 * Demote the pv entry. This depends on the earlier demotion
	 * of the mapping. Specifically, the (re)creation of a per-
	 * page pv entry might trigger the execution of pmap_pv_reclaim(),
	 * which might reclaim a newly (re)created per-page pv entry
	 * and destroy the associated mapping. In order to destroy
	 * the mapping, the PTE1 must have already changed from mapping
	 * the 1mpage to referencing the page table page.
	 */
	if (pte1_is_managed(opte1))
		pmap_pv_demote_pte1(pmap, va, pte1_pa(opte1));

	pmap_pte1_demotions++;
	CTR3(KTR_PMAP, "%s: success for va %#x in pmap %p",
	    __func__, va, pmap);

	PDEBUG(6, printf("%s(%p): success for va %#x pte1 %#x(%#x) at %p\n",
	    __func__, pmap, va, npte1, pte1_load(pte1p), pte1p));
	return (TRUE);
}

/*
 *	Insert the given physical page (p) at
 *	the specified virtual address (v) in the
 *	target physical map with the protection requested.
 *
 *	If specified, the page will be wired down, meaning
 *	that the related pte can not be reclaimed.
 *
 *	NB:  This is the only routine which MAY NOT lazy-evaluate
 *	or lose information.  That is, this routine must actually
 *	insert this page into the given map NOW.
 */
int
pmap_enter(pmap_t pmap, vm_offset_t va, vm_page_t m, vm_prot_t prot,
    u_int flags, int8_t psind)
{
	pt1_entry_t *pte1p;
	pt2_entry_t *pte2p;
	pt2_entry_t npte2, opte2;
	pv_entry_t pv;
	vm_paddr_t opa, pa;
	vm_page_t mpte2, om;
	int rv;

	va = trunc_page(va);
	KASSERT(va <= vm_max_kernel_address, ("%s: toobig", __func__));
	KASSERT(va < UPT2V_MIN_ADDRESS || va >= UPT2V_MAX_ADDRESS,
	    ("%s: invalid to pmap_enter page table pages (va: 0x%x)", __func__,
	    va));
	KASSERT((m->oflags & VPO_UNMANAGED) != 0 || !VA_IS_CLEANMAP(va),
	    ("%s: managed mapping within the clean submap", __func__));
	if ((m->oflags & VPO_UNMANAGED) == 0)
		VM_PAGE_OBJECT_BUSY_ASSERT(m);
	KASSERT((flags & PMAP_ENTER_RESERVED) == 0,
	    ("%s: flags %u has reserved bits set", __func__, flags));
	pa = VM_PAGE_TO_PHYS(m);
	npte2 = PTE2(pa, PTE2_A, vm_page_pte2_attr(m));
	if ((flags & VM_PROT_WRITE) == 0)
		npte2 |= PTE2_NM;
	if ((prot & VM_PROT_WRITE) == 0)
		npte2 |= PTE2_RO;
	KASSERT((npte2 & (PTE2_NM | PTE2_RO)) != PTE2_RO,
	    ("%s: flags includes VM_PROT_WRITE but prot doesn't", __func__));
	if ((prot & VM_PROT_EXECUTE) == 0)
		npte2 |= PTE2_NX;
	if ((flags & PMAP_ENTER_WIRED) != 0)
		npte2 |= PTE2_W;
	if (va < VM_MAXUSER_ADDRESS)
		npte2 |= PTE2_U;
	if (pmap != kernel_pmap)
		npte2 |= PTE2_NG;

	rw_wlock(&pvh_global_lock);
	PMAP_LOCK(pmap);
	sched_pin();
	if (psind == 1) {
		/* Assert the required virtual and physical alignment. */
		KASSERT((va & PTE1_OFFSET) == 0,
		    ("%s: va unaligned", __func__));
		KASSERT(m->psind > 0, ("%s: m->psind < psind", __func__));
		rv = pmap_enter_pte1(pmap, va, PTE1_PA(pa) | ATTR_TO_L1(npte2) |
		    PTE1_V, flags, m);
		goto out;
	}

	/*
	 * In the case that a page table page is not
	 * resident, we are creating it here.
	 */
	if (va < VM_MAXUSER_ADDRESS) {
		mpte2 = pmap_allocpte2(pmap, va, flags);
		if (mpte2 == NULL) {
			KASSERT((flags & PMAP_ENTER_NOSLEEP) != 0,
			    ("pmap_allocpte2 failed with sleep allowed"));
			rv = KERN_RESOURCE_SHORTAGE;
			goto out;
		}
	} else
		mpte2 = NULL;
	pte1p = pmap_pte1(pmap, va);
	if (pte1_is_section(pte1_load(pte1p)))
		panic("%s: attempted on 1MB page", __func__);
	pte2p = pmap_pte2_quick(pmap, va);
	if (pte2p == NULL)
		panic("%s: invalid L1 page table entry va=%#x", __func__, va);

	om = NULL;
	opte2 = pte2_load(pte2p);
	opa = pte2_pa(opte2);
	/*
	 * Mapping has not changed, must be protection or wiring change.
	 */
	if (pte2_is_valid(opte2) && (opa == pa)) {
		/*
		 * Wiring change, just update stats. We don't worry about
		 * wiring PT2 pages as they remain resident as long as there
		 * are valid mappings in them. Hence, if a user page is wired,
		 * the PT2 page will be also.
		 */
		if (pte2_is_wired(npte2) && !pte2_is_wired(opte2))
			pmap->pm_stats.wired_count++;
		else if (!pte2_is_wired(npte2) && pte2_is_wired(opte2))
			pmap->pm_stats.wired_count--;

		/*
		 * Remove extra pte2 reference
		 */
		if (mpte2)
			pt2_wirecount_dec(mpte2, pte1_index(va));
		if ((m->oflags & VPO_UNMANAGED) == 0)
			om = m;
		goto validate;
	}

	/*
	 * QQQ: We think that changing physical address on writeable mapping
	 *      is not safe. Well, maybe on kernel address space with correct
	 *      locking, it can make a sense. However, we have no idea why
	 *      anyone should do that on user address space. Are we wrong?
	 */
	KASSERT((opa == 0) || (opa == pa) ||
	    !pte2_is_valid(opte2) || ((opte2 & PTE2_RO) != 0),
	    ("%s: pmap %p va %#x(%#x) opa %#x pa %#x - gotcha %#x %#x!",
	    __func__, pmap, va, opte2, opa, pa, flags, prot));

	pv = NULL;

	/*
	 * Mapping has changed, invalidate old range and fall through to
	 * handle validating new mapping.
	 */
	if (opa) {
		if (pte2_is_wired(opte2))
			pmap->pm_stats.wired_count--;
		om = PHYS_TO_VM_PAGE(opa);
		if (om != NULL && (om->oflags & VPO_UNMANAGED) != 0)
			om = NULL;
		if (om != NULL)
			pv = pmap_pvh_remove(&om->md, pmap, va);

		/*
		 * Remove extra pte2 reference
		 */
		if (mpte2 != NULL)
			pt2_wirecount_dec(mpte2, va >> PTE1_SHIFT);
	} else
		pmap->pm_stats.resident_count++;

	/*
	 * Enter on the PV list if part of our managed memory.
	 */
	if ((m->oflags & VPO_UNMANAGED) == 0) {
		if (pv == NULL) {
			pv = get_pv_entry(pmap, FALSE);
			pv->pv_va = va;
		}
		TAILQ_INSERT_TAIL(&m->md.pv_list, pv, pv_next);
	} else if (pv != NULL)
		free_pv_entry(pmap, pv);

	/*
	 * Increment counters
	 */
	if (pte2_is_wired(npte2))
		pmap->pm_stats.wired_count++;

validate:
	/*
	 * Now validate mapping with desired protection/wiring.
	 */
	if (prot & VM_PROT_WRITE) {
		if ((m->oflags & VPO_UNMANAGED) == 0)
			vm_page_aflag_set(m, PGA_WRITEABLE);
	}

	/*
	 * If the mapping or permission bits are different, we need
	 * to update the pte2.
	 *
	 * QQQ: Think again and again what to do
	 *      if the mapping is going to be changed!
	 */
	if ((opte2 & ~(PTE2_NM | PTE2_A)) != (npte2 & ~(PTE2_NM | PTE2_A))) {
		/*
		 * Sync icache if exec permission and attribute VM_MEMATTR_WB_WA
		 * is set. Do it now, before the mapping is stored and made
		 * valid for hardware table walk. If done later, there is a race
		 * for other threads of current process in lazy loading case.
		 * Don't do it for kernel memory which is mapped with exec
		 * permission even if the memory isn't going to hold executable
		 * code. The only time when icache sync is needed is after
		 * kernel module is loaded and the relocation info is processed.
		 * And it's done in elf_cpu_load_file().
		 *
		 * QQQ: (1) Does it exist any better way where
		 *          or how to sync icache?
		 *      (2) Now, we do it on a page basis.
		 */
		if ((prot & VM_PROT_EXECUTE) && pmap != kernel_pmap &&
		    m->md.pat_mode == VM_MEMATTR_WB_WA &&
		    (opa != pa || (opte2 & PTE2_NX)))
			cache_icache_sync_fresh(va, pa, PAGE_SIZE);

		if (opte2 & PTE2_V) {
			/* Change mapping with break-before-make approach. */
			opte2 = pte2_load_clear(pte2p);
			pmap_tlb_flush(pmap, va);
			pte2_store(pte2p, npte2);
			if (om != NULL) {
				KASSERT((om->oflags & VPO_UNMANAGED) == 0,
				    ("%s: om %p unmanaged", __func__, om));
				if ((opte2 & PTE2_A) != 0)
					vm_page_aflag_set(om, PGA_REFERENCED);
				if (pte2_is_dirty(opte2))
					vm_page_dirty(om);
				if (TAILQ_EMPTY(&om->md.pv_list) &&
				    ((om->flags & PG_FICTITIOUS) != 0 ||
				    TAILQ_EMPTY(&pa_to_pvh(opa)->pv_list)))
					vm_page_aflag_clear(om, PGA_WRITEABLE);
			}
		} else
			pte2_store(pte2p, npte2);
	}
#if 0
	else {
		/*
		 * QQQ: In time when both access and not mofified bits are
		 *      emulated by software, this should not happen. Some
		 *      analysis is need, if this really happen. Missing
		 *      tlb flush somewhere could be the reason.
		 */
		panic("%s: pmap %p va %#x opte2 %x npte2 %x !!", __func__, pmap,
		    va, opte2, npte2);
	}
#endif

#if VM_NRESERVLEVEL > 0
	/*
	 * If both the L2 page table page and the reservation are fully
	 * populated, then attempt promotion.
	 */
	if ((mpte2 == NULL || pt2_is_full(mpte2, va)) &&
	    sp_enabled && (m->flags & PG_FICTITIOUS) == 0 &&
	    vm_reserv_level_iffullpop(m) == 0)
		pmap_promote_pte1(pmap, pte1p, va);
#endif

	rv = KERN_SUCCESS;
out:
	sched_unpin();
	rw_wunlock(&pvh_global_lock);
	PMAP_UNLOCK(pmap);
	return (rv);
}

/*
 *  Do the things to unmap a page in a process.
 */
static int
pmap_remove_pte2(pmap_t pmap, pt2_entry_t *pte2p, vm_offset_t va,
    struct spglist *free)
{
	pt2_entry_t opte2;
	vm_page_t m;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	PMAP_LOCK_ASSERT(pmap, MA_OWNED);

	/* Clear and invalidate the mapping. */
	opte2 = pte2_load_clear(pte2p);
	pmap_tlb_flush(pmap, va);

	KASSERT(pte2_is_valid(opte2), ("%s: pmap %p va %#x not link pte2 %#x",
	    __func__, pmap, va, opte2));

	if (opte2 & PTE2_W)
		pmap->pm_stats.wired_count -= 1;
	pmap->pm_stats.resident_count -= 1;
	if (pte2_is_managed(opte2)) {
		m = PHYS_TO_VM_PAGE(pte2_pa(opte2));
		if (pte2_is_dirty(opte2))
			vm_page_dirty(m);
		if (opte2 & PTE2_A)
			vm_page_aflag_set(m, PGA_REFERENCED);
		pmap_remove_entry(pmap, m, va);
	}
	return (pmap_unuse_pt2(pmap, va, free));
}

/*
 *  Remove a single page from a process address space.
 */
static void
pmap_remove_page(pmap_t pmap, vm_offset_t va, struct spglist *free)
{
	pt2_entry_t *pte2p;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	KASSERT(curthread->td_pinned > 0,
	    ("%s: curthread not pinned", __func__));
	PMAP_LOCK_ASSERT(pmap, MA_OWNED);
	if ((pte2p = pmap_pte2_quick(pmap, va)) == NULL ||
	    !pte2_is_valid(pte2_load(pte2p)))
		return;
	pmap_remove_pte2(pmap, pte2p, va, free);
}

/*
 *  Remove the given range of addresses from the specified map.
 *
 *  It is assumed that the start and end are properly
 *  rounded to the page size.
 */
void
pmap_remove(pmap_t pmap, vm_offset_t sva, vm_offset_t eva)
{
	vm_offset_t nextva;
	pt1_entry_t *pte1p, pte1;
	pt2_entry_t *pte2p, pte2;
	struct spglist free;

	/*
	 * Perform an unsynchronized read. This is, however, safe.
	 */
	if (pmap->pm_stats.resident_count == 0)
		return;

	SLIST_INIT(&free);

	rw_wlock(&pvh_global_lock);
	sched_pin();
	PMAP_LOCK(pmap);

	/*
	 * Special handling of removing one page. A very common
	 * operation and easy to short circuit some code.
	 */
	if (sva + PAGE_SIZE == eva) {
		pte1 = pte1_load(pmap_pte1(pmap, sva));
		if (pte1_is_link(pte1)) {
			pmap_remove_page(pmap, sva, &free);
			goto out;
		}
	}

	for (; sva < eva; sva = nextva) {
		/*
		 * Calculate address for next L2 page table.
		 */
		nextva = pte1_trunc(sva + PTE1_SIZE);
		if (nextva < sva)
			nextva = eva;
		if (pmap->pm_stats.resident_count == 0)
			break;

		pte1p = pmap_pte1(pmap, sva);
		pte1 = pte1_load(pte1p);

		/*
		 * Weed out invalid mappings. Note: we assume that the L1 page
		 * table is always allocated, and in kernel virtual.
		 */
		if (pte1 == 0)
			continue;

		if (pte1_is_section(pte1)) {
			/*
			 * Are we removing the entire large page?  If not,
			 * demote the mapping and fall through.
			 */
			if (sva + PTE1_SIZE == nextva && eva >= nextva) {
				pmap_remove_pte1(pmap, pte1p, sva, &free);
				continue;
			} else if (!pmap_demote_pte1(pmap, pte1p, sva)) {
				/* The large page mapping was destroyed. */
				continue;
			}
#ifdef INVARIANTS
			else {
				/* Update pte1 after demotion. */
				pte1 = pte1_load(pte1p);
			}
#endif
		}

		KASSERT(pte1_is_link(pte1), ("%s: pmap %p va %#x pte1 %#x at %p"
		    " is not link", __func__, pmap, sva, pte1, pte1p));

		/*
		 * Limit our scan to either the end of the va represented
		 * by the current L2 page table page, or to the end of the
		 * range being removed.
		 */
		if (nextva > eva)
			nextva = eva;

		for (pte2p = pmap_pte2_quick(pmap, sva); sva != nextva;
		    pte2p++, sva += PAGE_SIZE) {
			pte2 = pte2_load(pte2p);
			if (!pte2_is_valid(pte2))
				continue;
			if (pmap_remove_pte2(pmap, pte2p, sva, &free))
				break;
		}
	}
out:
	sched_unpin();
	rw_wunlock(&pvh_global_lock);
	PMAP_UNLOCK(pmap);
	vm_page_free_pages_toq(&free, false);
}

/*
 *	Routine:	pmap_remove_all
 *	Function:
 *		Removes this physical page from
 *		all physical maps in which it resides.
 *		Reflects back modify bits to the pager.
 *
 *	Notes:
 *		Original versions of this routine were very
 *		inefficient because they iteratively called
 *		pmap_remove (slow...)
 */

void
pmap_remove_all(vm_page_t m)
{
	struct md_page *pvh;
	pv_entry_t pv;
	pmap_t pmap;
	pt2_entry_t *pte2p, opte2;
	pt1_entry_t *pte1p;
	vm_offset_t va;
	struct spglist free;

	KASSERT((m->oflags & VPO_UNMANAGED) == 0,
	    ("%s: page %p is not managed", __func__, m));
	SLIST_INIT(&free);
	rw_wlock(&pvh_global_lock);
	sched_pin();
	if ((m->flags & PG_FICTITIOUS) != 0)
		goto small_mappings;
	pvh = pa_to_pvh(VM_PAGE_TO_PHYS(m));
	while ((pv = TAILQ_FIRST(&pvh->pv_list)) != NULL) {
		va = pv->pv_va;
		pmap = PV_PMAP(pv);
		PMAP_LOCK(pmap);
		pte1p = pmap_pte1(pmap, va);
		(void)pmap_demote_pte1(pmap, pte1p, va);
		PMAP_UNLOCK(pmap);
	}
small_mappings:
	while ((pv = TAILQ_FIRST(&m->md.pv_list)) != NULL) {
		pmap = PV_PMAP(pv);
		PMAP_LOCK(pmap);
		pmap->pm_stats.resident_count--;
		pte1p = pmap_pte1(pmap, pv->pv_va);
		KASSERT(!pte1_is_section(pte1_load(pte1p)), ("%s: found "
		    "a 1mpage in page %p's pv list", __func__, m));
		pte2p = pmap_pte2_quick(pmap, pv->pv_va);
		opte2 = pte2_load_clear(pte2p);
		pmap_tlb_flush(pmap, pv->pv_va);
		KASSERT(pte2_is_valid(opte2), ("%s: pmap %p va %x zero pte2",
		    __func__, pmap, pv->pv_va));
		if (pte2_is_wired(opte2))
			pmap->pm_stats.wired_count--;
		if (opte2 & PTE2_A)
			vm_page_aflag_set(m, PGA_REFERENCED);

		/*
		 * Update the vm_page_t clean and reference bits.
		 */
		if (pte2_is_dirty(opte2))
			vm_page_dirty(m);
		pmap_unuse_pt2(pmap, pv->pv_va, &free);
		TAILQ_REMOVE(&m->md.pv_list, pv, pv_next);
		free_pv_entry(pmap, pv);
		PMAP_UNLOCK(pmap);
	}
	vm_page_aflag_clear(m, PGA_WRITEABLE);
	sched_unpin();
	rw_wunlock(&pvh_global_lock);
	vm_page_free_pages_toq(&free, false);
}

/*
 *  Just subroutine for pmap_remove_pages() to reasonably satisfy
 *  good coding style, a.k.a. 80 character line width limit hell.
 */
static __inline void
pmap_remove_pte1_quick(pmap_t pmap, pt1_entry_t pte1, pv_entry_t pv,
    struct spglist *free)
{
	vm_paddr_t pa;
	vm_page_t m, mt, mpt2pg;
	struct md_page *pvh;

	pa = pte1_pa(pte1);
	m = PHYS_TO_VM_PAGE(pa);

	KASSERT(m->phys_addr == pa, ("%s: vm_page_t %p addr mismatch %#x %#x",
	    __func__, m, m->phys_addr, pa));
	KASSERT((m->flags & PG_FICTITIOUS) != 0 ||
	    m < &vm_page_array[vm_page_array_size],
	    ("%s: bad pte1 %#x", __func__, pte1));

	if (pte1_is_dirty(pte1)) {
		for (mt = m; mt < &m[PTE1_SIZE / PAGE_SIZE]; mt++)
			vm_page_dirty(mt);
	}

	pmap->pm_stats.resident_count -= PTE1_SIZE / PAGE_SIZE;
	pvh = pa_to_pvh(pa);
	TAILQ_REMOVE(&pvh->pv_list, pv, pv_next);
	if (TAILQ_EMPTY(&pvh->pv_list)) {
		for (mt = m; mt < &m[PTE1_SIZE / PAGE_SIZE]; mt++)
			if (TAILQ_EMPTY(&mt->md.pv_list))
				vm_page_aflag_clear(mt, PGA_WRITEABLE);
	}
	mpt2pg = pmap_pt2_page(pmap, pv->pv_va);
	if (mpt2pg != NULL)
		pmap_unwire_pt2_all(pmap, pv->pv_va, mpt2pg, free);
}

/*
 *  Just subroutine for pmap_remove_pages() to reasonably satisfy
 *  good coding style, a.k.a. 80 character line width limit hell.
 */
static __inline void
pmap_remove_pte2_quick(pmap_t pmap, pt2_entry_t pte2, pv_entry_t pv,
    struct spglist *free)
{
	vm_paddr_t pa;
	vm_page_t m;
	struct md_page *pvh;

	pa = pte2_pa(pte2);
	m = PHYS_TO_VM_PAGE(pa);

	KASSERT(m->phys_addr == pa, ("%s: vm_page_t %p addr mismatch %#x %#x",
	    __func__, m, m->phys_addr, pa));
	KASSERT((m->flags & PG_FICTITIOUS) != 0 ||
	    m < &vm_page_array[vm_page_array_size],
	    ("%s: bad pte2 %#x", __func__, pte2));

	if (pte2_is_dirty(pte2))
		vm_page_dirty(m);

	pmap->pm_stats.resident_count--;
	TAILQ_REMOVE(&m->md.pv_list, pv, pv_next);
	if (TAILQ_EMPTY(&m->md.pv_list) && (m->flags & PG_FICTITIOUS) == 0) {
		pvh = pa_to_pvh(pa);
		if (TAILQ_EMPTY(&pvh->pv_list))
			vm_page_aflag_clear(m, PGA_WRITEABLE);
	}
	pmap_unuse_pt2(pmap, pv->pv_va, free);
}

/*
 *  Remove all pages from specified address space this aids process
 *  exit speeds. Also, this code is special cased for current process
 *  only, but can have the more generic (and slightly slower) mode enabled.
 *  This is much faster than pmap_remove in the case of running down
 *  an entire address space.
 */
void
pmap_remove_pages(pmap_t pmap)
{
	pt1_entry_t *pte1p, pte1;
	pt2_entry_t *pte2p, pte2;
	pv_entry_t pv;
	struct pv_chunk *pc, *npc;
	struct spglist free;
	int field, idx;
	int32_t bit;
	uint32_t inuse, bitmask;
	boolean_t allfree;

	/*
	 * Assert that the given pmap is only active on the current
	 * CPU.  Unfortunately, we cannot block another CPU from
	 * activating the pmap while this function is executing.
	 */
	KASSERT(pmap == vmspace_pmap(curthread->td_proc->p_vmspace),
	    ("%s: non-current pmap %p", __func__, pmap));
#if defined(SMP) && defined(INVARIANTS)
	{
		cpuset_t other_cpus;

		sched_pin();
		other_cpus = pmap->pm_active;
		CPU_CLR(PCPU_GET(cpuid), &other_cpus);
		sched_unpin();
		KASSERT(CPU_EMPTY(&other_cpus),
		    ("%s: pmap %p active on other cpus", __func__, pmap));
	}
#endif
	SLIST_INIT(&free);
	rw_wlock(&pvh_global_lock);
	PMAP_LOCK(pmap);
	sched_pin();
	TAILQ_FOREACH_SAFE(pc, &pmap->pm_pvchunk, pc_list, npc) {
		KASSERT(pc->pc_pmap == pmap, ("%s: wrong pmap %p %p",
		    __func__, pmap, pc->pc_pmap));
		allfree = TRUE;
		for (field = 0; field < _NPCM; field++) {
			inuse = (~(pc->pc_map[field])) & pc_freemask[field];
			while (inuse != 0) {
				bit = ffs(inuse) - 1;
				bitmask = 1UL << bit;
				idx = field * 32 + bit;
				pv = &pc->pc_pventry[idx];
				inuse &= ~bitmask;

				/*
				 * Note that we cannot remove wired pages
				 * from a process' mapping at this time
				 */
				pte1p = pmap_pte1(pmap, pv->pv_va);
				pte1 = pte1_load(pte1p);
				if (pte1_is_section(pte1)) {
					if (pte1_is_wired(pte1))  {
						allfree = FALSE;
						continue;
					}
					pte1_clear(pte1p);
					pmap_remove_pte1_quick(pmap, pte1, pv,
					    &free);
				}
				else if (pte1_is_link(pte1)) {
					pte2p = pt2map_entry(pv->pv_va);
					pte2 = pte2_load(pte2p);

					if (!pte2_is_valid(pte2)) {
						printf("%s: pmap %p va %#x "
						    "pte2 %#x\n", __func__,
						    pmap, pv->pv_va, pte2);
						panic("bad pte2");
					}

					if (pte2_is_wired(pte2))   {
						allfree = FALSE;
						continue;
					}
					pte2_clear(pte2p);
					pmap_remove_pte2_quick(pmap, pte2, pv,
					    &free);
				} else {
					printf("%s: pmap %p va %#x pte1 %#x\n",
					    __func__, pmap, pv->pv_va, pte1);
					panic("bad pte1");
				}

				/* Mark free */
				PV_STAT(pv_entry_frees++);
				PV_STAT(pv_entry_spare++);
				pv_entry_count--;
				pc->pc_map[field] |= bitmask;
			}
		}
		if (allfree) {
			TAILQ_REMOVE(&pmap->pm_pvchunk, pc, pc_list);
			free_pv_chunk(pc);
		}
	}
	tlb_flush_all_ng_local();
	sched_unpin();
	rw_wunlock(&pvh_global_lock);
	PMAP_UNLOCK(pmap);
	vm_page_free_pages_toq(&free, false);
}

/*
 *  This code makes some *MAJOR* assumptions:
 *  1. Current pmap & pmap exists.
 *  2. Not wired.
 *  3. Read access.
 *  4. No L2 page table pages.
 *  but is *MUCH* faster than pmap_enter...
 */
static vm_page_t
pmap_enter_quick_locked(pmap_t pmap, vm_offset_t va, vm_page_t m,
    vm_prot_t prot, vm_page_t mpt2pg)
{
	pt2_entry_t *pte2p, pte2;
	vm_paddr_t pa;
	struct spglist free;
	uint32_t l2prot;

	KASSERT(!VA_IS_CLEANMAP(va) ||
	    (m->oflags & VPO_UNMANAGED) != 0,
	    ("%s: managed mapping within the clean submap", __func__));
	rw_assert(&pvh_global_lock, RA_WLOCKED);
	PMAP_LOCK_ASSERT(pmap, MA_OWNED);

	/*
	 * In the case that a L2 page table page is not
	 * resident, we are creating it here.
	 */
	if (va < VM_MAXUSER_ADDRESS) {
		u_int pte1_idx;
		pt1_entry_t pte1, *pte1p;
		vm_paddr_t pt2_pa;

		/*
		 * Get L1 page table things.
		 */
		pte1_idx = pte1_index(va);
		pte1p = pmap_pte1(pmap, va);
		pte1 = pte1_load(pte1p);

		if (mpt2pg && (mpt2pg->pindex == (pte1_idx & ~PT2PG_MASK))) {
			/*
			 * Each of NPT2_IN_PG L2 page tables on the page can
			 * come here. Make sure that associated L1 page table
			 * link is established.
			 *
			 * QQQ: It comes that we don't establish all links to
			 *      L2 page tables for newly allocated L2 page
			 *      tables page.
			 */
			KASSERT(!pte1_is_section(pte1),
			    ("%s: pte1 %#x is section", __func__, pte1));
			if (!pte1_is_link(pte1)) {
				pt2_pa = page_pt2pa(VM_PAGE_TO_PHYS(mpt2pg),
				    pte1_idx);
				pte1_store(pte1p, PTE1_LINK(pt2_pa));
			}
			pt2_wirecount_inc(mpt2pg, pte1_idx);
		} else {
			/*
			 * If the L2 page table page is mapped, we just
			 * increment the hold count, and activate it.
			 */
			if (pte1_is_section(pte1)) {
				return (NULL);
			} else if (pte1_is_link(pte1)) {
				mpt2pg = PHYS_TO_VM_PAGE(pte1_link_pa(pte1));
				pt2_wirecount_inc(mpt2pg, pte1_idx);
			} else {
				mpt2pg = _pmap_allocpte2(pmap, va,
				    PMAP_ENTER_NOSLEEP);
				if (mpt2pg == NULL)
					return (NULL);
			}
		}
	} else {
		mpt2pg = NULL;
	}

	/*
	 * This call to pt2map_entry() makes the assumption that we are
	 * entering the page into the current pmap.  In order to support
	 * quick entry into any pmap, one would likely use pmap_pte2_quick().
	 * But that isn't as quick as pt2map_entry().
	 */
	pte2p = pt2map_entry(va);
	pte2 = pte2_load(pte2p);
	if (pte2_is_valid(pte2)) {
		if (mpt2pg != NULL) {
			/*
			 * Remove extra pte2 reference
			 */
			pt2_wirecount_dec(mpt2pg, pte1_index(va));
			mpt2pg = NULL;
		}
		return (NULL);
	}

	/*
	 * Enter on the PV list if part of our managed memory.
	 */
	if ((m->oflags & VPO_UNMANAGED) == 0 &&
	    !pmap_try_insert_pv_entry(pmap, va, m)) {
		if (mpt2pg != NULL) {
			SLIST_INIT(&free);
			if (pmap_unwire_pt2(pmap, va, mpt2pg, &free)) {
				pmap_tlb_flush(pmap, va);
				vm_page_free_pages_toq(&free, false);
			}

			mpt2pg = NULL;
		}
		return (NULL);
	}

	/*
	 * Increment counters
	 */
	pmap->pm_stats.resident_count++;

	/*
	 * Now validate mapping with RO protection
	 */
	pa = VM_PAGE_TO_PHYS(m);
	l2prot = PTE2_RO | PTE2_NM;
	if (va < VM_MAXUSER_ADDRESS)
		l2prot |= PTE2_U | PTE2_NG;
	if ((prot & VM_PROT_EXECUTE) == 0)
		l2prot |= PTE2_NX;
	else if (m->md.pat_mode == VM_MEMATTR_WB_WA && pmap != kernel_pmap) {
		/*
		 * Sync icache if exec permission and attribute VM_MEMATTR_WB_WA
		 * is set. QQQ: For more info, see comments in pmap_enter().
		 */
		cache_icache_sync_fresh(va, pa, PAGE_SIZE);
	}
	pte2_store(pte2p, PTE2(pa, l2prot, vm_page_pte2_attr(m)));

	return (mpt2pg);
}

void
pmap_enter_quick(pmap_t pmap, vm_offset_t va, vm_page_t m, vm_prot_t prot)
{

	rw_wlock(&pvh_global_lock);
	PMAP_LOCK(pmap);
	(void)pmap_enter_quick_locked(pmap, va, m, prot, NULL);
	rw_wunlock(&pvh_global_lock);
	PMAP_UNLOCK(pmap);
}

/*
 *  Tries to create a read- and/or execute-only 1 MB page mapping.  Returns
 *  true if successful.  Returns false if (1) a mapping already exists at the
 *  specified virtual address or (2) a PV entry cannot be allocated without
 *  reclaiming another PV entry.
 */
static bool
pmap_enter_1mpage(pmap_t pmap, vm_offset_t va, vm_page_t m, vm_prot_t prot)
{
	pt1_entry_t pte1;
	vm_paddr_t pa;

	PMAP_LOCK_ASSERT(pmap, MA_OWNED);
	pa = VM_PAGE_TO_PHYS(m);
	pte1 = PTE1(pa, PTE1_NM | PTE1_RO, ATTR_TO_L1(vm_page_pte2_attr(m)));
	if ((prot & VM_PROT_EXECUTE) == 0)
		pte1 |= PTE1_NX;
	if (va < VM_MAXUSER_ADDRESS)
		pte1 |= PTE1_U;
	if (pmap != kernel_pmap)
		pte1 |= PTE1_NG;
	return (pmap_enter_pte1(pmap, va, pte1, PMAP_ENTER_NOSLEEP |
	    PMAP_ENTER_NOREPLACE | PMAP_ENTER_NORECLAIM, m) == KERN_SUCCESS);
}

/*
 *  Tries to create the specified 1 MB page mapping.  Returns KERN_SUCCESS if
 *  the mapping was created, and either KERN_FAILURE or KERN_RESOURCE_SHORTAGE
 *  otherwise.  Returns KERN_FAILURE if PMAP_ENTER_NOREPLACE was specified and
 *  a mapping already exists at the specified virtual address.  Returns
 *  KERN_RESOURCE_SHORTAGE if PMAP_ENTER_NORECLAIM was specified and PV entry
 *  allocation failed.
 */
static int
pmap_enter_pte1(pmap_t pmap, vm_offset_t va, pt1_entry_t pte1, u_int flags,
    vm_page_t m)
{
	struct spglist free;
	pt1_entry_t opte1, *pte1p;
	pt2_entry_t pte2, *pte2p;
	vm_offset_t cur, end;
	vm_page_t mt;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	KASSERT((pte1 & (PTE1_NM | PTE1_RO)) == 0 ||
	    (pte1 & (PTE1_NM | PTE1_RO)) == (PTE1_NM | PTE1_RO),
	    ("%s: pte1 has inconsistent NM and RO attributes", __func__));
	PMAP_LOCK_ASSERT(pmap, MA_OWNED);
	pte1p = pmap_pte1(pmap, va);
	opte1 = pte1_load(pte1p);
	if (pte1_is_valid(opte1)) {
		if ((flags & PMAP_ENTER_NOREPLACE) != 0) {
			CTR3(KTR_PMAP, "%s: failure for va %#lx in pmap %p",
			    __func__, va, pmap);
			return (KERN_FAILURE);
		}
		/* Break the existing mapping(s). */
		SLIST_INIT(&free);
		if (pte1_is_section(opte1)) {
			/*
			 * If the section resulted from a promotion, then a
			 * reserved PT page could be freed.
			 */
			pmap_remove_pte1(pmap, pte1p, va, &free);
		} else {
			sched_pin();
			end = va + PTE1_SIZE;
			for (cur = va, pte2p = pmap_pte2_quick(pmap, va);
			    cur != end; cur += PAGE_SIZE, pte2p++) {
				pte2 = pte2_load(pte2p);
				if (!pte2_is_valid(pte2))
					continue;
				if (pmap_remove_pte2(pmap, pte2p, cur, &free))
					break;
			}
			sched_unpin();
		}
		vm_page_free_pages_toq(&free, false);
	}
	if ((m->oflags & VPO_UNMANAGED) == 0) {
		/*
		 * Abort this mapping if its PV entry could not be created.
		 */
		if (!pmap_pv_insert_pte1(pmap, va, pte1, flags)) {
			CTR3(KTR_PMAP, "%s: failure for va %#lx in pmap %p",
			    __func__, va, pmap);
			return (KERN_RESOURCE_SHORTAGE);
		}
		if ((pte1 & PTE1_RO) == 0) {
			for (mt = m; mt < &m[PTE1_SIZE / PAGE_SIZE]; mt++)
				vm_page_aflag_set(mt, PGA_WRITEABLE);
		}
	}

	/*
	 * Increment counters.
	 */
	if (pte1_is_wired(pte1))
		pmap->pm_stats.wired_count += PTE1_SIZE / PAGE_SIZE;
	pmap->pm_stats.resident_count += PTE1_SIZE / PAGE_SIZE;

	/*
	 * Sync icache if exec permission and attribute VM_MEMATTR_WB_WA
	 * is set.  QQQ: For more info, see comments in pmap_enter().
	 */
	if ((pte1 & PTE1_NX) == 0 && m->md.pat_mode == VM_MEMATTR_WB_WA &&
	    pmap != kernel_pmap && (!pte1_is_section(opte1) ||
	    pte1_pa(opte1) != VM_PAGE_TO_PHYS(m) || (opte1 & PTE2_NX) != 0))
		cache_icache_sync_fresh(va, VM_PAGE_TO_PHYS(m), PTE1_SIZE);

	/*
	 * Map the section.
	 */
	pte1_store(pte1p, pte1);

	pmap_pte1_mappings++;
	CTR3(KTR_PMAP, "%s: success for va %#lx in pmap %p", __func__, va,
	    pmap);
	return (KERN_SUCCESS);
}

/*
 *  Maps a sequence of resident pages belonging to the same object.
 *  The sequence begins with the given page m_start.  This page is
 *  mapped at the given virtual address start.  Each subsequent page is
 *  mapped at a virtual address that is offset from start by the same
 *  amount as the page is offset from m_start within the object.  The
 *  last page in the sequence is the page with the largest offset from
 *  m_start that can be mapped at a virtual address less than the given
 *  virtual address end.  Not every virtual page between start and end
 *  is mapped; only those for which a resident page exists with the
 *  corresponding offset from m_start are mapped.
 */
void
pmap_enter_object(pmap_t pmap, vm_offset_t start, vm_offset_t end,
    vm_page_t m_start, vm_prot_t prot)
{
	vm_offset_t va;
	vm_page_t m, mpt2pg;
	vm_pindex_t diff, psize;

	PDEBUG(6, printf("%s: pmap %p start %#x end  %#x m %p prot %#x\n",
	    __func__, pmap, start, end, m_start, prot));

	VM_OBJECT_ASSERT_LOCKED(m_start->object);
	psize = atop(end - start);
	mpt2pg = NULL;
	m = m_start;
	rw_wlock(&pvh_global_lock);
	PMAP_LOCK(pmap);
	while (m != NULL && (diff = m->pindex - m_start->pindex) < psize) {
		va = start + ptoa(diff);
		if ((va & PTE1_OFFSET) == 0 && va + PTE1_SIZE <= end &&
		    m->psind == 1 && sp_enabled &&
		    pmap_enter_1mpage(pmap, va, m, prot))
			m = &m[PTE1_SIZE / PAGE_SIZE - 1];
		else
			mpt2pg = pmap_enter_quick_locked(pmap, va, m, prot,
			    mpt2pg);
		m = TAILQ_NEXT(m, listq);
	}
	rw_wunlock(&pvh_global_lock);
	PMAP_UNLOCK(pmap);
}

/*
 *  This code maps large physical mmap regions into the
 *  processor address space.  Note that some shortcuts
 *  are taken, but the code works.
 */
void
pmap_object_init_pt(pmap_t pmap, vm_offset_t addr, vm_object_t object,
    vm_pindex_t pindex, vm_size_t size)
{
	pt1_entry_t *pte1p;
	vm_paddr_t pa, pte2_pa;
	vm_page_t p;
	vm_memattr_t pat_mode;
	u_int l1attr, l1prot;

	VM_OBJECT_ASSERT_WLOCKED(object);
	KASSERT(object->type == OBJT_DEVICE || object->type == OBJT_SG,
	    ("%s: non-device object", __func__));
	if ((addr & PTE1_OFFSET) == 0 && (size & PTE1_OFFSET) == 0) {
		if (!vm_object_populate(object, pindex, pindex + atop(size)))
			return;
		p = vm_page_lookup(object, pindex);
		KASSERT(p->valid == VM_PAGE_BITS_ALL,
		    ("%s: invalid page %p", __func__, p));
		pat_mode = p->md.pat_mode;

		/*
		 * Abort the mapping if the first page is not physically
		 * aligned to a 1MB page boundary.
		 */
		pte2_pa = VM_PAGE_TO_PHYS(p);
		if (pte2_pa & PTE1_OFFSET)
			return;

		/*
		 * Skip the first page. Abort the mapping if the rest of
		 * the pages are not physically contiguous or have differing
		 * memory attributes.
		 */
		p = TAILQ_NEXT(p, listq);
		for (pa = pte2_pa + PAGE_SIZE; pa < pte2_pa + size;
		    pa += PAGE_SIZE) {
			KASSERT(p->valid == VM_PAGE_BITS_ALL,
			    ("%s: invalid page %p", __func__, p));
			if (pa != VM_PAGE_TO_PHYS(p) ||
			    pat_mode != p->md.pat_mode)
				return;
			p = TAILQ_NEXT(p, listq);
		}

		/*
		 * Map using 1MB pages.
		 *
		 * QQQ: Well, we are mapping a section, so same condition must
		 * be hold like during promotion. It looks that only RW mapping
		 * is done here, so readonly mapping must be done elsewhere.
		 */
		l1prot = PTE1_U | PTE1_NG | PTE1_RW | PTE1_M | PTE1_A;
		l1attr = ATTR_TO_L1(vm_memattr_to_pte2(pat_mode));
		PMAP_LOCK(pmap);
		for (pa = pte2_pa; pa < pte2_pa + size; pa += PTE1_SIZE) {
			pte1p = pmap_pte1(pmap, addr);
			if (!pte1_is_valid(pte1_load(pte1p))) {
				pte1_store(pte1p, PTE1(pa, l1prot, l1attr));
				pmap->pm_stats.resident_count += PTE1_SIZE /
				    PAGE_SIZE;
				pmap_pte1_mappings++;
			}
			/* Else continue on if the PTE1 is already valid. */
			addr += PTE1_SIZE;
		}
		PMAP_UNLOCK(pmap);
	}
}

/*
 *  Do the things to protect a 1mpage in a process.
 */
static void
pmap_protect_pte1(pmap_t pmap, pt1_entry_t *pte1p, vm_offset_t sva,
    vm_prot_t prot)
{
	pt1_entry_t npte1, opte1;
	vm_offset_t eva, va;
	vm_page_t m;

	PMAP_LOCK_ASSERT(pmap, MA_OWNED);
	KASSERT((sva & PTE1_OFFSET) == 0,
	    ("%s: sva is not 1mpage aligned", __func__));

	opte1 = npte1 = pte1_load(pte1p);
	if (pte1_is_managed(opte1) && pte1_is_dirty(opte1)) {
		eva = sva + PTE1_SIZE;
		for (va = sva, m = PHYS_TO_VM_PAGE(pte1_pa(opte1));
		    va < eva; va += PAGE_SIZE, m++)
			vm_page_dirty(m);
	}
	if ((prot & VM_PROT_WRITE) == 0)
		npte1 |= PTE1_RO | PTE1_NM;
	if ((prot & VM_PROT_EXECUTE) == 0)
		npte1 |= PTE1_NX;

	/*
	 * QQQ: Herein, execute permission is never set.
	 *      It only can be cleared. So, no icache
	 *      syncing is needed.
	 */

	if (npte1 != opte1) {
		pte1_store(pte1p, npte1);
		pmap_tlb_flush(pmap, sva);
	}
}

/*
 *	Set the physical protection on the
 *	specified range of this map as requested.
 */
void
pmap_protect(pmap_t pmap, vm_offset_t sva, vm_offset_t eva, vm_prot_t prot)
{
	boolean_t pv_lists_locked;
	vm_offset_t nextva;
	pt1_entry_t *pte1p, pte1;
	pt2_entry_t *pte2p, opte2, npte2;

	KASSERT((prot & ~VM_PROT_ALL) == 0, ("invalid prot %x", prot));
	if (prot == VM_PROT_NONE) {
		pmap_remove(pmap, sva, eva);
		return;
	}

	if ((prot & (VM_PROT_WRITE | VM_PROT_EXECUTE)) ==
	    (VM_PROT_WRITE | VM_PROT_EXECUTE))
		return;

	if (pmap_is_current(pmap))
		pv_lists_locked = FALSE;
	else {
		pv_lists_locked = TRUE;
resume:
		rw_wlock(&pvh_global_lock);
		sched_pin();
	}

	PMAP_LOCK(pmap);
	for (; sva < eva; sva = nextva) {
		/*
		 * Calculate address for next L2 page table.
		 */
		nextva = pte1_trunc(sva + PTE1_SIZE);
		if (nextva < sva)
			nextva = eva;

		pte1p = pmap_pte1(pmap, sva);
		pte1 = pte1_load(pte1p);

		/*
		 * Weed out invalid mappings. Note: we assume that L1 page
		 * page table is always allocated, and in kernel virtual.
		 */
		if (pte1 == 0)
			continue;

		if (pte1_is_section(pte1)) {
			/*
			 * Are we protecting the entire large page?  If not,
			 * demote the mapping and fall through.
			 */
			if (sva + PTE1_SIZE == nextva && eva >= nextva) {
				pmap_protect_pte1(pmap, pte1p, sva, prot);
				continue;
			} else {
				if (!pv_lists_locked) {
					pv_lists_locked = TRUE;
					if (!rw_try_wlock(&pvh_global_lock)) {
						PMAP_UNLOCK(pmap);
						goto resume;
					}
					sched_pin();
				}
				if (!pmap_demote_pte1(pmap, pte1p, sva)) {
					/*
					 * The large page mapping
					 * was destroyed.
					 */
					continue;
				}
#ifdef INVARIANTS
				else {
					/* Update pte1 after demotion */
					pte1 = pte1_load(pte1p);
				}
#endif
			}
		}

		KASSERT(pte1_is_link(pte1), ("%s: pmap %p va %#x pte1 %#x at %p"
		    " is not link", __func__, pmap, sva, pte1, pte1p));

		/*
		 * Limit our scan to either the end of the va represented
		 * by the current L2 page table page, or to the end of the
		 * range being protected.
		 */
		if (nextva > eva)
			nextva = eva;

		for (pte2p = pmap_pte2_quick(pmap, sva); sva != nextva; pte2p++,
		    sva += PAGE_SIZE) {
			vm_page_t m;

			opte2 = npte2 = pte2_load(pte2p);
			if (!pte2_is_valid(opte2))
				continue;

			if ((prot & VM_PROT_WRITE) == 0) {
				if (pte2_is_managed(opte2) &&
				    pte2_is_dirty(opte2)) {
					m = PHYS_TO_VM_PAGE(pte2_pa(opte2));
					vm_page_dirty(m);
				}
				npte2 |= PTE2_RO | PTE2_NM;
			}

			if ((prot & VM_PROT_EXECUTE) == 0)
				npte2 |= PTE2_NX;

			/*
			 * QQQ: Herein, execute permission is never set.
			 *      It only can be cleared. So, no icache
			 *      syncing is needed.
			 */

			if (npte2 != opte2) {
				pte2_store(pte2p, npte2);
				pmap_tlb_flush(pmap, sva);
			}
		}
	}
	if (pv_lists_locked) {
		sched_unpin();
		rw_wunlock(&pvh_global_lock);
	}
	PMAP_UNLOCK(pmap);
}

/*
 *	pmap_pvh_wired_mappings:
 *
 *	Return the updated number "count" of managed mappings that are wired.
 */
static int
pmap_pvh_wired_mappings(struct md_page *pvh, int count)
{
	pmap_t pmap;
	pt1_entry_t pte1;
	pt2_entry_t pte2;
	pv_entry_t pv;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	sched_pin();
	TAILQ_FOREACH(pv, &pvh->pv_list, pv_next) {
		pmap = PV_PMAP(pv);
		PMAP_LOCK(pmap);
		pte1 = pte1_load(pmap_pte1(pmap, pv->pv_va));
		if (pte1_is_section(pte1)) {
			if (pte1_is_wired(pte1))
				count++;
		} else {
			KASSERT(pte1_is_link(pte1),
			    ("%s: pte1 %#x is not link", __func__, pte1));
			pte2 = pte2_load(pmap_pte2_quick(pmap, pv->pv_va));
			if (pte2_is_wired(pte2))
				count++;
		}
		PMAP_UNLOCK(pmap);
	}
	sched_unpin();
	return (count);
}

/*
 *	pmap_page_wired_mappings:
 *
 *	Return the number of managed mappings to the given physical page
 *	that are wired.
 */
int
pmap_page_wired_mappings(vm_page_t m)
{
	int count;

	count = 0;
	if ((m->oflags & VPO_UNMANAGED) != 0)
		return (count);
	rw_wlock(&pvh_global_lock);
	count = pmap_pvh_wired_mappings(&m->md, count);
	if ((m->flags & PG_FICTITIOUS) == 0) {
		count = pmap_pvh_wired_mappings(pa_to_pvh(VM_PAGE_TO_PHYS(m)),
		    count);
	}
	rw_wunlock(&pvh_global_lock);
	return (count);
}

/*
 *  Returns TRUE if any of the given mappings were used to modify
 *  physical memory.  Otherwise, returns FALSE.  Both page and 1mpage
 *  mappings are supported.
 */
static boolean_t
pmap_is_modified_pvh(struct md_page *pvh)
{
	pv_entry_t pv;
	pt1_entry_t pte1;
	pt2_entry_t pte2;
	pmap_t pmap;
	boolean_t rv;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	rv = FALSE;
	sched_pin();
	TAILQ_FOREACH(pv, &pvh->pv_list, pv_next) {
		pmap = PV_PMAP(pv);
		PMAP_LOCK(pmap);
		pte1 = pte1_load(pmap_pte1(pmap, pv->pv_va));
		if (pte1_is_section(pte1)) {
			rv = pte1_is_dirty(pte1);
		} else {
			KASSERT(pte1_is_link(pte1),
			    ("%s: pte1 %#x is not link", __func__, pte1));
			pte2 = pte2_load(pmap_pte2_quick(pmap, pv->pv_va));
			rv = pte2_is_dirty(pte2);
		}
		PMAP_UNLOCK(pmap);
		if (rv)
			break;
	}
	sched_unpin();
	return (rv);
}

/*
 *	pmap_is_modified:
 *
 *	Return whether or not the specified physical page was modified
 *	in any physical maps.
 */
boolean_t
pmap_is_modified(vm_page_t m)
{
	boolean_t rv;

	KASSERT((m->oflags & VPO_UNMANAGED) == 0,
	    ("%s: page %p is not managed", __func__, m));

	/*
	 * If the page is not busied then this check is racy.
	 */
	if (!pmap_page_is_write_mapped(m))
		return (FALSE);
	rw_wlock(&pvh_global_lock);
	rv = pmap_is_modified_pvh(&m->md) ||
	    ((m->flags & PG_FICTITIOUS) == 0 &&
	    pmap_is_modified_pvh(pa_to_pvh(VM_PAGE_TO_PHYS(m))));
	rw_wunlock(&pvh_global_lock);
	return (rv);
}

/*
 *	pmap_is_prefaultable:
 *
 *	Return whether or not the specified virtual address is eligible
 *	for prefault.
 */
boolean_t
pmap_is_prefaultable(pmap_t pmap, vm_offset_t addr)
{
	pt1_entry_t pte1;
	pt2_entry_t pte2;
	boolean_t rv;

	rv = FALSE;
	PMAP_LOCK(pmap);
	pte1 = pte1_load(pmap_pte1(pmap, addr));
	if (pte1_is_link(pte1)) {
		pte2 = pte2_load(pt2map_entry(addr));
		rv = !pte2_is_valid(pte2) ;
	}
	PMAP_UNLOCK(pmap);
	return (rv);
}

/*
 *  Returns TRUE if any of the given mappings were referenced and FALSE
 *  otherwise. Both page and 1mpage mappings are supported.
 */
static boolean_t
pmap_is_referenced_pvh(struct md_page *pvh)
{

	pv_entry_t pv;
	pt1_entry_t pte1;
	pt2_entry_t pte2;
	pmap_t pmap;
	boolean_t rv;

	rw_assert(&pvh_global_lock, RA_WLOCKED);
	rv = FALSE;
	sched_pin();
	TAILQ_FOREACH(pv, &pvh->pv_list, pv_next) {
		pmap = PV_PMAP(pv);
		PMAP_LOCK(pmap);
		pte1 = pte1_load(pmap_pte1(pmap, pv->pv_va));
		if (pte1_is_section(pte1)) {
			rv = (pte1 & (PTE1_A | PTE1_V)) == (PTE1_A | PTE1_V);
		} else {
			pte2 = pte2_load(pmap_pte2_quick(pmap, pv->pv_va));
			rv = (pte2 & (PTE2_A | PTE2_V)) == (PTE2_A | PTE2_V);
		}
		PMAP_UNLOCK(pmap);
		if (rv)
			break;
	}
	sched_unpin();
	return (rv);
}

/*
 *	pmap_is_referenced:
 *
 *	Return whether or not the specified physical page was referenced
 *	in any physical maps.
 */
boolean_t
pmap_is_referenced(vm_page_t m)
{
	boolean_t rv;

	KASSERT((m->oflags & VPO_UNMANAGED) == 0,
	    ("%s: page %p is not managed", __func__, m));
	rw_wlock(&pvh_global_lock);
	rv = pmap_is_referenced_pvh(&m->md) ||
	    ((m->flags & PG_FICTITIOUS) == 0 &&
	    pmap_is_referenced_pvh(pa_to_pvh(VM_PAGE_TO_PHYS(m))));
	rw_wunlock(&pvh_global_lock);
	return (rv);
}

/*
 *	pmap_ts_referenced:
 *
 *	Return a count of reference bits for a page, clearing those bits.
 *	It is not necessary for every reference bit to be cleared, but it
 *	is necessary that 0 only be returned when there are truly no
 *	reference bits set.
 *
 *	As an optimization, update the page's dirty field if a modified bit is
 *	found while counting reference bits.  This opportunistic update can be
 *	performed at low cost and can eliminate the need for some future calls
 *	to pmap_is_modified().  However, since this function stops after
 *	finding PMAP_TS_REFERENCED_MAX reference bits, it may not detect some
 *	dirty pages.  Those dirty pages will only be detected by a future call
 *	to pmap_is_modified().
 */
int
pmap_ts_referenced(vm_page_t m)
{
	struct md_page *pvh;
	pv_entry_t pv, pvf;
	pmap_t pmap;
	pt1_entry_t  *pte1p, opte1;
	pt2_entry_t *pte2p, opte2;
	vm_paddr_t pa;
	int rtval = 0;

	KASSERT((m->oflags & VPO_UNMANAGED) == 0,
	    ("%s: page %p is not managed", __func__, m));
	pa = VM_PAGE_TO_PHYS(m);
	pvh = pa_to_pvh(pa);
	rw_wlock(&pvh_global_lock);
	sched_pin();
	if ((m->flags & PG_FICTITIOUS) != 0 ||
	    (pvf = TAILQ_FIRST(&pvh->pv_list)) == NULL)
		goto small_mappings;
	pv = pvf;
	do {
		pmap = PV_PMAP(pv);
		PMAP_LOCK(pmap);
		pte1p = pmap_pte1(pmap, pv->pv_va);
		opte1 = pte1_load(pte1p);
		if (pte1_is_dirty(opte1)) {
			/*
			 * Although "opte1" is mapping a 1MB page, because
			 * this function is called at a 4KB page granularity,
			 * we only update the 4KB page under test.
			 */
			vm_page_dirty(m);
		}
		if ((opte1 & PTE1_A) != 0) {
			/*
			 * Since this reference bit is shared by 256 4KB pages,
			 * it should not be cleared every time it is tested.
			 * Apply a simple "hash" function on the physical page
			 * number, the virtual section number, and the pmap
			 * address to select one 4KB page out of the 256
			 * on which testing the reference bit will result
			 * in clearing that bit. This function is designed
			 * to avoid the selection of the same 4KB page
			 * for every 1MB page mapping.
			 *
			 * On demotion, a mapping that hasn't been referenced
			 * is simply destroyed.  To avoid the possibility of a
			 * subsequent page fault on a demoted wired mapping,
			 * always leave its reference bit set.  Moreover,
			 * since the section is wired, the current state of
			 * its reference bit won't affect page replacement.
			 */
			 if ((((pa >> PAGE_SHIFT) ^ (pv->pv_va >> PTE1_SHIFT) ^
			    (uintptr_t)pmap) & (NPTE2_IN_PG - 1)) == 0 &&
			    !pte1_is_wired(opte1)) {
				pte1_clear_bit(pte1p, PTE1_A);
				pmap_tlb_flush(pmap, pv->pv_va);
			}
			rtval++;
		}
		PMAP_UNLOCK(pmap);
		/* Rotate the PV list if it has more than one entry. */
		if (TAILQ_NEXT(pv, pv_next) != NULL) {
			TAILQ_REMOVE(&pvh->pv_list, pv, pv_next);
			TAILQ_INSERT_TAIL(&pvh->pv_list, pv, pv_next);
		}
		if (rtval >= PMAP_TS_REFERENCED_MAX)
			goto out;
	} while ((pv = TAILQ_FIRST(&pvh->pv_list)) != pvf);
small_mappings:
	if ((pvf = TAILQ_FIRST(&m->md.pv_list)) == NULL)
		goto out;
	pv = pvf;
	do {
		pmap = PV_PMAP(pv);
		PMAP_LOCK(pmap);
		pte1p = pmap_pte1(pmap, pv->pv_va);
		KASSERT(pte1_is_link(pte1_load(pte1p)),
		    ("%s: not found a link in page %p's pv list", __func__, m));

		pte2p = pmap_pte2_quick(pmap, pv->pv_va);
		opte2 = pte2_load(pte2p);
		if (pte2_is_dirty(opte2))
			vm_page_dirty(m);
		if ((opte2 & PTE2_A) != 0) {
			pte2_clear_bit(pte2p, PTE2_A);
			pmap_tlb_flush(pmap, pv->pv_va);
			rtval++;
		}
		PMAP_UNLOCK(pmap);
		/* Rotate the PV list if it has more than one entry. */
		if (TAILQ_NEXT(pv, pv_next) != NULL) {
			TAILQ_REMOVE(&m->md.pv_list, pv, pv_next);
			TAILQ_INSERT_TAIL(&m->md.pv_list, pv, pv_next);
		}
	} while ((pv = TAILQ_FIRST(&m->md.pv_list)) != pvf && rtval <
	    PMAP_TS_REFERENCED_MAX);
out:
	sched_unpin();
	rw_wunlock(&pvh_global_lock);
	return (rtval);
}

/*
 *	Clear the wired attribute from the mappings for the specified range of
 *	addresses in the given pmap.  Every valid mapping within that range
 *	must have the wired attribute set.  In contrast, invalid mappings
 *	cannot have the wired attribute set, so they are ignored.
 *
 *	The wired attribute of the page table entry is not a hardware feature,
 *	so there is no need to invalidate any TLB entries.
 */
void
pmap_unwire(pmap_t pmap, vm_offset_t sva, vm_offset_t eva)
{
	vm_offset_t nextva;
	pt1_entry_t *pte1p, pte1;
	pt2_entry_t *pte2p, pte2;
	boolean_t pv_lists_locked;

	if (pmap_is_current(pmap))
		pv_lists_locked = FALSE;
	else {
		pv_lists_locked = TRUE;
resume:
		rw_wlock(&pvh_global_lock);
		sched_pin();
	}
	PMAP_LOCK(pmap);
	for (; sva < eva; sva = nextva) {
		nextva = pte1_trunc(sva + PTE1_SIZE);
		if (nextva < sva)
			nextva = eva;

		pte1p = pmap_pte1(pmap, sva);
		pte1 = pte1_load(pte1p);

		/*
		 * Weed out invalid mappings. Note: we assume that L1 page
		 * page table is always allocated, and in kernel virtual.
		 */
		if (pte1 == 0)
			continue;

		if (pte1_is_section(pte1)) {
			if (!pte1_is_wired(pte1))
				panic("%s: pte1 %#x not wired", __func__, pte1);

			/*
			 * Are we unwiring the entire large page?  If not,
			 * demote the mapping and fall through.
			 */
			if (sva + PTE1_SIZE == nextva && eva >= nextva) {
				pte1_clear_bit(pte1p, PTE1_W);
				pmap->pm_stats.wired_count -= PTE1_SIZE /
				    PAGE_SIZE;
				continue;
			} else {
				if (!pv_lists_locked) {
					pv_lists_locked = TRUE;
					if (!rw_try_wlock(&pvh_global_lock)) {
						PMAP_UNLOCK(pmap);
						/* Repeat sva. */
						goto resume;
					}
					sched_pin();
				}
				if (!pmap_demote_pte1(pmap, pte1p, sva))
					panic("%s: demotion failed", __func__);
#ifdef INVARIANTS
				else {
					/* Update pte1 after demotion */
					pte1 = pte1_load(pte1p);
				}
#endif
			}
		}

		KASSERT(pte1_is_link(pte1), ("%s: pmap %p va %#x pte1 %#x at %p"
		    " is not link", __func__, pmap, sva, pte1, pte1p));

		/*
		 * Limit our scan to either the end of the va represented
		 * by the current L2 page table page, or to the end of the
		 * range being protected.
		 */
		if (nextva > eva)
			nextva = eva;

		for (pte2p = pmap_pte2_quick(pmap, sva); sva != nextva; pte2p++,
		    sva += PAGE_SIZE) {
			pte2 = pte2_load(pte2p);
			if (!pte2_is_valid(pte2))
				continue;
			if (!pte2_is_wired(pte2))
				panic("%s: pte2 %#x is missing PTE2_W",
				    __func__, pte2);

			/*
			 * PTE2_W must be cleared atomically. Although the pmap
			 * lock synchronizes access to PTE2_W, another processor
			 * could be changing PTE2_NM and/or PTE2_A concurrently.
			 */
			pte2_clear_bit(pte2p, PTE2_W);
			pmap->pm_stats.wired_count--;
		}
	}
	if (pv_lists_locked) {
		sched_unpin();
		rw_wunlock(&pvh_global_lock);
	}
	PMAP_UNLOCK(pmap);
}

/*
 *  Clear the write and modified bits in each of the given page's mappings.
 */
void
pmap_remove_write(vm_page_t m)
{
	struct md_page *pvh;
	pv_entry_t next_pv, pv;
	pmap_t pmap;
	pt1_entry_t *pte1p;
	pt2_entry_t *pte2p, opte2;
	vm_offset_t va;

	KASSERT((m->oflags & VPO_UNMANAGED) == 0,
	    ("%s: page %p is not managed", __func__, m));
	vm_page_assert_busied(m);

	if (!pmap_page_is_write_mapped(m))
		return;
	rw_wlock(&pvh_global_lock);
	sched_pin();
	if ((m->flags & PG_FICTITIOUS) != 0)
		goto small_mappings;
	pvh = pa_to_pvh(VM_PAGE_TO_PHYS(m));
	TAILQ_FOREACH_SAFE(pv, &pvh->pv_list, pv_next, next_pv) {
		va = pv->pv_va;
		pmap = PV_PMAP(pv);
		PMAP_LOCK(pmap);
		pte1p = pmap_pte1(pmap, va);
		if (!(pte1_load(pte1p) & PTE1_RO))
			(void)pmap_demote_pte1(pmap, pte1p, va);
		PMAP_UNLOCK(pmap);
	}
small_mappings:
	TAILQ_FOREACH(pv, &m->md.pv_list, pv_next) {
		pmap = PV_PMAP(pv);
		PMAP_LOCK(pmap);
		pte1p = pmap_pte1(pmap, pv->pv_va);
		KASSERT(!pte1_is_section(pte1_load(pte1p)), ("%s: found"
		    " a section in page %p's pv list", __func__, m));
		pte2p = pmap_pte2_quick(pmap, pv->pv_va);
		opte2 = pte2_load(pte2p);
		if (!(opte2 & PTE2_RO)) {
			pte2_store(pte2p, opte2 | PTE2_RO | PTE2_NM);
			if (pte2_is_dirty(opte2))
				vm_page_dirty(m);
			pmap_tlb_flush(pmap, pv->pv_va);
		}
		PMAP_UNLOCK(pmap);
	}
	vm_page_aflag_clear(m, PGA_WRITEABLE);
	sched_unpin();
	rw_wunlock(&pvh_global_lock);
}

/*
 *	Apply the given advice to the specified range of addresses within the
 *	given pmap.  Depending on the advice, clear the referenced and/or
 *	modified flags in each mapping and set the mapped page's dirty field.
 */
void
pmap_advise(pmap_t pmap, vm_offset_t sva, vm_offset_t eva, int advice)
{
	pt1_entry_t *pte1p, opte1;
	pt2_entry_t *pte2p, pte2;
	vm_offset_t pdnxt;
	vm_page_t m;
	boolean_t pv_lists_locked;

	if (advice != MADV_DONTNEED && advice != MADV_FREE)
		return;
	if (pmap_is_current(pmap))
		pv_lists_locked = FALSE;
	else {
		pv_lists_locked = TRUE;
resume:
		rw_wlock(&pvh_global_lock);
		sched_pin();
	}
	PMAP_LOCK(pmap);
	for (; sva < eva; sva = pdnxt) {
		pdnxt = pte1_trunc(sva + PTE1_SIZE);
		if (pdnxt < sva)
			pdnxt = eva;
		pte1p = pmap_pte1(pmap, sva);
		opte1 = pte1_load(pte1p);
		if (!pte1_is_valid(opte1)) /* XXX */
			continue;
		else if (pte1_is_section(opte1)) {
			if (!pte1_is_managed(opte1))
				continue;
			if (!pv_lists_locked) {
				pv_lists_locked = TRUE;
				if (!rw_try_wlock(&pvh_global_lock)) {
					PMAP_UNLOCK(pmap);
					goto resume;
				}
				sched_pin();
			}
			if (!pmap_demote_pte1(pmap, pte1p, sva)) {
				/*
				 * The large page mapping was destroyed.
				 */
				continue;
			}

			/*
			 * Unless the page mappings are wired, remove the
			 * mapping to a single page so that a subsequent
			 * access may repromote.  Since the underlying L2 page
			 * table is fully populated, this removal never
			 * frees a L2 page table page.
			 */
			if (!pte1_is_wired(opte1)) {
				pte2p = pmap_pte2_quick(pmap, sva);
				KASSERT(pte2_is_valid(pte2_load(pte2p)),
				    ("%s: invalid PTE2", __func__));
				pmap_remove_pte2(pmap, pte2p, sva, NULL);
			}
		}
		if (pdnxt > eva)
			pdnxt = eva;
		for (pte2p = pmap_pte2_quick(pmap, sva); sva != pdnxt; pte2p++,
		    sva += PAGE_SIZE) {
			pte2 = pte2_load(pte2p);
			if (!pte2_is_valid(pte2) || !pte2_is_managed(pte2))
				continue;
			else if (pte2_is_dirty(pte2)) {
				if (advice == MADV_DONTNEED) {
					/*
					 * Future calls to pmap_is_modified()
					 * can be avoided by making the page
					 * dirty now.
					 */
					m = PHYS_TO_VM_PAGE(pte2_pa(pte2));
					vm_page_dirty(m);
				}
				pte2_set_bit(pte2p, PTE2_NM);
				pte2_clear_bit(pte2p, PTE2_A);
			} else if ((pte2 & PTE2_A) != 0)
				pte2_clear_bit(pte2p, PTE2_A);
			else
				continue;
			pmap_tlb_flush(pmap, sva);
		}
	}
	if (pv_lists_locked) {
		sched_unpin();
		rw_wunlock(&pvh_global_lock);
	}
	PMAP_UNLOCK(pmap);
}

/*
 *	Clear the modify bits on the specified physical page.
 */
void
pmap_clear_modify(vm_page_t m)
{
	struct md_page *pvh;
	pv_entry_t next_pv, pv;
	pmap_t pmap;
	pt1_entry_t *pte1p, opte1;
	pt2_entry_t *pte2p, opte2;
	vm_offset_t va;

	KASSERT((m->oflags & VPO_UNMANAGED) == 0,
	    ("%s: page %p is not managed", __func__, m));
	vm_page_assert_busied(m);

	if (!pmap_page_is_write_mapped(m))
		return;
	rw_wlock(&pvh_global_lock);
	sched_pin();
	if ((m->flags & PG_FICTITIOUS) != 0)
		goto small_mappings;
	pvh = pa_to_pvh(VM_PAGE_TO_PHYS(m));
	TAILQ_FOREACH_SAFE(pv, &pvh->pv_list, pv_next, next_pv) {
		va = pv->pv_va;
		pmap = PV_PMAP(pv);
		PMAP_LOCK(pmap);
		pte1p = pmap_pte1(pmap, va);
		opte1 = pte1_load(pte1p);
		if (!(opte1 & PTE1_RO)) {
			if (pmap_demote_pte1(pmap, pte1p, va) &&
			    !pte1_is_wired(opte1)) {
				/*
				 * Write protect the mapping to a
				 * single page so that a subsequent
				 * write access may repromote.
				 */
				va += VM_PAGE_TO_PHYS(m) - pte1_pa(opte1);
				pte2p = pmap_pte2_quick(pmap, va);
				opte2 = pte2_load(pte2p);
				if ((opte2 & PTE2_V)) {
					pte2_set_bit(pte2p, PTE2_NM | PTE2_RO);
					vm_page_dirty(m);
					pmap_tlb_flush(pmap, va);
				}
			}
		}
		PMAP_UNLOCK(pmap);
	}
small_mappings:
	TAILQ_FOREACH(pv, &m->md.pv_list, pv_next) {
		pmap = PV_PMAP(pv);
		PMAP_LOCK(pmap);
		pte1p = pmap_pte1(pmap, pv->pv_va);
		KASSERT(!pte1_is_section(pte1_load(pte1p)), ("%s: found"
		    " a section in page %p's pv list", __func__, m));
		pte2p = pmap_pte2_quick(pmap, pv->pv_va);
		if (pte2_is_dirty(pte2_load(pte2p))) {
			pte2_set_bit(pte2p, PTE2_NM);
			pmap_tlb_flush(pmap, pv->pv_va);
		}
		PMAP_UNLOCK(pmap);
	}
	sched_unpin();
	rw_wunlock(&pvh_global_lock);
}

/*
 *  Sets the memory attribute for the specified page.
 */
void
pmap_page_set_memattr(vm_page_t m, vm_memattr_t ma)
{
	pt2_entry_t *cmap2_pte2p;
	vm_memattr_t oma;
	vm_paddr_t pa;
	struct pcpu *pc;

	oma = m->md.pat_mode;
	m->md.pat_mode = ma;

	CTR5(KTR_PMAP, "%s: page %p - 0x%08X oma: %d, ma: %d", __func__, m,
	    VM_PAGE_TO_PHYS(m), oma, ma);
	if ((m->flags & PG_FICTITIOUS) != 0)
		return;
#if 0
	/*
	 * If "m" is a normal page, flush it from the cache.
	 *
	 * First, try to find an existing mapping of the page by sf
	 * buffer. sf_buf_invalidate_cache() modifies mapping and
	 * flushes the cache.
	 */
	if (sf_buf_invalidate_cache(m, oma))
		return;
#endif
	/*
	 * If page is not mapped by sf buffer, map the page
	 * transient and do invalidation.
	 */
	if (ma != oma) {
		pa = VM_PAGE_TO_PHYS(m);
		sched_pin();
		pc = get_pcpu();
		cmap2_pte2p = pc->pc_cmap2_pte2p;
		mtx_lock(&pc->pc_cmap_lock);
		if (pte2_load(cmap2_pte2p) != 0)
			panic("%s: CMAP2 busy", __func__);
		pte2_store(cmap2_pte2p, PTE2_KERN_NG(pa, PTE2_AP_KRW,
		    vm_memattr_to_pte2(ma)));
		dcache_wbinv_poc((vm_offset_t)pc->pc_cmap2_addr, pa, PAGE_SIZE);
		pte2_clear(cmap2_pte2p);
		tlb_flush((vm_offset_t)pc->pc_cmap2_addr);
		sched_unpin();
		mtx_unlock(&pc->pc_cmap_lock);
	}
}

/*
 *  Miscellaneous support routines follow
 */

/*
 *  Returns TRUE if the given page is mapped individually or as part of
 *  a 1mpage.  Otherwise, returns FALSE.
 */
boolean_t
pmap_page_is_mapped(vm_page_t m)
{
	boolean_t rv;

	if ((m->oflags & VPO_UNMANAGED) != 0)
		return (FALSE);
	rw_wlock(&pvh_global_lock);
	rv = !TAILQ_EMPTY(&m->md.pv_list) ||
	    ((m->flags & PG_FICTITIOUS) == 0 &&
	    !TAILQ_EMPTY(&pa_to_pvh(VM_PAGE_TO_PHYS(m))->pv_list));
	rw_wunlock(&pvh_global_lock);
	return (rv);
}

/*
 *  Returns true if the pmap's pv is one of the first
 *  16 pvs linked to from this page.  This count may
 *  be changed upwards or downwards in the future; it
 *  is only necessary that true be returned for a small
 *  subset of pmaps for proper page aging.
 */
boolean_t
pmap_page_exists_quick(pmap_t pmap, vm_page_t m)
{
	struct md_page *pvh;
	pv_entry_t pv;
	int loops = 0;
	boolean_t rv;

	KASSERT((m->oflags & VPO_UNMANAGED) == 0,
	    ("%s: page %p is not managed", __func__, m));
	rv = FALSE;
	rw_wlock(&pvh_global_lock);
	TAILQ_FOREACH(pv, &m->md.pv_list, pv_next) {
		if (PV_PMAP(pv) == pmap) {
			rv = TRUE;
			break;
		}
		loops++;
		if (loops >= 16)
			break;
	}
	if (!rv && loops < 16 && (m->flags & PG_FICTITIOUS) == 0) {
		pvh = pa_to_pvh(VM_PAGE_TO_PHYS(m));
		TAILQ_FOREACH(pv, &pvh->pv_list, pv_next) {
			if (PV_PMAP(pv) == pmap) {
				rv = TRUE;
				break;
			}
			loops++;
			if (loops >= 16)
				break;
		}
	}
	rw_wunlock(&pvh_global_lock);
	return (rv);
}

/*
 *	pmap_zero_page zeros the specified hardware page by mapping
 *	the page into KVM and using bzero to clear its contents.
 */
void
pmap_zero_page(vm_page_t m)
{
	pt2_entry_t *cmap2_pte2p;
	struct pcpu *pc;

	sched_pin();
	pc = get_pcpu();
	cmap2_pte2p = pc->pc_cmap2_pte2p;
	mtx_lock(&pc->pc_cmap_lock);
	if (pte2_load(cmap2_pte2p) != 0)
		panic("%s: CMAP2 busy", __func__);
	pte2_store(cmap2_pte2p, PTE2_KERN_NG(VM_PAGE_TO_PHYS(m), PTE2_AP_KRW,
	    vm_page_pte2_attr(m)));
	pagezero(pc->pc_cmap2_addr);
	pte2_clear(cmap2_pte2p);
	tlb_flush((vm_offset_t)pc->pc_cmap2_addr);
	sched_unpin();
	mtx_unlock(&pc->pc_cmap_lock);
}

/*
 *	pmap_zero_page_area zeros the specified hardware page by mapping
 *	the page into KVM and using bzero to clear its contents.
 *
 *	off and size may not cover an area beyond a single hardware page.
 */
void
pmap_zero_page_area(vm_page_t m, int off, int size)
{
	pt2_entry_t *cmap2_pte2p;
	struct pcpu *pc;

	sched_pin();
	pc = get_pcpu();
	cmap2_pte2p = pc->pc_cmap2_pte2p;
	mtx_lock(&pc->pc_cmap_lock);
	if (pte2_load(cmap2_pte2p) != 0)
		panic("%s: CMAP2 busy", __func__);
	pte2_store(cmap2_pte2p, PTE2_KERN_NG(VM_PAGE_TO_PHYS(m), PTE2_AP_KRW,
	    vm_page_pte2_attr(m)));
	if (off == 0 && size == PAGE_SIZE)
		pagezero(pc->pc_cmap2_addr);
	else
		bzero(pc->pc_cmap2_addr + off, size);
	pte2_clear(cmap2_pte2p);
	tlb_flush((vm_offset_t)pc->pc_cmap2_addr);
	sched_unpin();
	mtx_unlock(&pc->pc_cmap_lock);
}

/*
 *	pmap_copy_page copies the specified (machine independent)
 *	page by mapping the page into virtual memory and using
 *	bcopy to copy the page, one machine dependent page at a
 *	time.
 */
void
pmap_copy_page(vm_page_t src, vm_page_t dst)
{
	pt2_entry_t *cmap1_pte2p, *cmap2_pte2p;
	struct pcpu *pc;

	sched_pin();
	pc = get_pcpu();
	cmap1_pte2p = pc->pc_cmap1_pte2p;
	cmap2_pte2p = pc->pc_cmap2_pte2p;
	mtx_lock(&pc->pc_cmap_lock);
	if (pte2_load(cmap1_pte2p) != 0)
		panic("%s: CMAP1 busy", __func__);
	if (pte2_load(cmap2_pte2p) != 0)
		panic("%s: CMAP2 busy", __func__);
	pte2_store(cmap1_pte2p, PTE2_KERN_NG(VM_PAGE_TO_PHYS(src),
	    PTE2_AP_KR | PTE2_NM, vm_page_pte2_attr(src)));
	pte2_store(cmap2_pte2p, PTE2_KERN_NG(VM_PAGE_TO_PHYS(dst),
	    PTE2_AP_KRW, vm_page_pte2_attr(dst)));
	bcopy(pc->pc_cmap1_addr, pc->pc_cmap2_addr, PAGE_SIZE);
	pte2_clear(cmap1_pte2p);
	tlb_flush((vm_offset_t)pc->pc_cmap1_addr);
	pte2_clear(cmap2_pte2p);
	tlb_flush((vm_offset_t)pc->pc_cmap2_addr);
	sched_unpin();
	mtx_unlock(&pc->pc_cmap_lock);
}

int unmapped_buf_allowed = 1;

void
pmap_copy_pages(vm_page_t ma[], vm_offset_t a_offset, vm_page_t mb[],
    vm_offset_t b_offset, int xfersize)
{
	pt2_entry_t *cmap1_pte2p, *cmap2_pte2p;
	vm_page_t a_pg, b_pg;
	char *a_cp, *b_cp;
	vm_offset_t a_pg_offset, b_pg_offset;
	struct pcpu *pc;
	int cnt;

	sched_pin();
	pc = get_pcpu();
	cmap1_pte2p = pc->pc_cmap1_pte2p;
	cmap2_pte2p = pc->pc_cmap2_pte2p;
	mtx_lock(&pc->pc_cmap_lock);
	if (pte2_load(cmap1_pte2p) != 0)
		panic("pmap_copy_pages: CMAP1 busy");
	if (pte2_load(cmap2_pte2p) != 0)
		panic("pmap_copy_pages: CMAP2 busy");
	while (xfersize > 0) {
		a_pg = ma[a_offset >> PAGE_SHIFT];
		a_pg_offset = a_offset & PAGE_MASK;
		cnt = min(xfersize, PAGE_SIZE - a_pg_offset);
		b_pg = mb[b_offset >> PAGE_SHIFT];
		b_pg_offset = b_offset & PAGE_MASK;
		cnt = min(cnt, PAGE_SIZE - b_pg_offset);
		pte2_store(cmap1_pte2p, PTE2_KERN_NG(VM_PAGE_TO_PHYS(a_pg),
		    PTE2_AP_KR | PTE2_NM, vm_page_pte2_attr(a_pg)));
		tlb_flush_local((vm_offset_t)pc->pc_cmap1_addr);
		pte2_store(cmap2_pte2p, PTE2_KERN_NG(VM_PAGE_TO_PHYS(b_pg),
		    PTE2_AP_KRW, vm_page_pte2_attr(b_pg)));
		tlb_flush_local((vm_offset_t)pc->pc_cmap2_addr);
		a_cp = pc->pc_cmap1_addr + a_pg_offset;
		b_cp = pc->pc_cmap2_addr + b_pg_offset;
		bcopy(a_cp, b_cp, cnt);
		a_offset += cnt;
		b_offset += cnt;
		xfersize -= cnt;
	}
	pte2_clear(cmap1_pte2p);
	tlb_flush((vm_offset_t)pc->pc_cmap1_addr);
	pte2_clear(cmap2_pte2p);
	tlb_flush((vm_offset_t)pc->pc_cmap2_addr);
	sched_unpin();
	mtx_unlock(&pc->pc_cmap_lock);
}

vm_offset_t
pmap_quick_enter_page(vm_page_t m)
{
	struct pcpu *pc;
	pt2_entry_t *pte2p;

	critical_enter();
	pc = get_pcpu();
	pte2p = pc->pc_qmap_pte2p;

	KASSERT(pte2_load(pte2p) == 0, ("%s: PTE2 busy", __func__));

	pte2_store(pte2p, PTE2_KERN_NG(VM_PAGE_TO_PHYS(m), PTE2_AP_KRW,
	    vm_page_pte2_attr(m)));
	return (pc->pc_qmap_addr);
}

void
pmap_quick_remove_page(vm_offset_t addr)
{
	struct pcpu *pc;
	pt2_entry_t *pte2p;

	pc = get_pcpu();
	pte2p = pc->pc_qmap_pte2p;

	KASSERT(addr == pc->pc_qmap_addr, ("%s: invalid address", __func__));
	KASSERT(pte2_load(pte2p) != 0, ("%s: PTE2 not in use", __func__));

	pte2_clear(pte2p);
	tlb_flush(pc->pc_qmap_addr);
	critical_exit();
}

/*
 *	Copy the range specified by src_addr/len
 *	from the source map to the range dst_addr/len
 *	in the destination map.
 *
 *	This routine is only advisory and need not do anything.
 */
void
pmap_copy(pmap_t dst_pmap, pmap_t src_pmap, vm_offset_t dst_addr, vm_size_t len,
    vm_offset_t src_addr)
{
	struct spglist free;
	vm_offset_t addr;
	vm_offset_t end_addr = src_addr + len;
	vm_offset_t nextva;

	if (dst_addr != src_addr)
		return;

	if (!pmap_is_current(src_pmap))
		return;

	rw_wlock(&pvh_global_lock);
	if (dst_pmap < src_pmap) {
		PMAP_LOCK(dst_pmap);
		PMAP_LOCK(src_pmap);
	} else {
		PMAP_LOCK(src_pmap);
		PMAP_LOCK(dst_pmap);
	}
	sched_pin();
	for (addr = src_addr; addr < end_addr; addr = nextva) {
		pt2_entry_t *src_pte2p, *dst_pte2p;
		vm_page_t dst_mpt2pg, src_mpt2pg;
		pt1_entry_t src_pte1;
		u_int pte1_idx;

		KASSERT(addr < VM_MAXUSER_ADDRESS,
		    ("%s: invalid to pmap_copy page tables", __func__));

		nextva = pte1_trunc(addr + PTE1_SIZE);
		if (nextva < addr)
			nextva = end_addr;

		pte1_idx = pte1_index(addr);
		src_pte1 = src_pmap->pm_pt1[pte1_idx];
		if (pte1_is_section(src_pte1)) {
			if ((addr & PTE1_OFFSET) != 0 ||
			    (addr + PTE1_SIZE) > end_addr)
				continue;
			if (dst_pmap->pm_pt1[pte1_idx] == 0 &&
			    (!pte1_is_managed(src_pte1) ||
			    pmap_pv_insert_pte1(dst_pmap, addr, src_pte1,
			    PMAP_ENTER_NORECLAIM))) {
				dst_pmap->pm_pt1[pte1_idx] = src_pte1 &
				    ~PTE1_W;
				dst_pmap->pm_stats.resident_count +=
				    PTE1_SIZE / PAGE_SIZE;
				pmap_pte1_mappings++;
			}
			continue;
		} else if (!pte1_is_link(src_pte1))
			continue;

		src_mpt2pg = PHYS_TO_VM_PAGE(pte1_link_pa(src_pte1));

		/*
		 * We leave PT2s to be linked from PT1 even if they are not
		 * referenced until all PT2s in a page are without reference.
		 *
		 * QQQ: It could be changed ...
		 */
#if 0 /* single_pt2_link_is_cleared */
		KASSERT(pt2_wirecount_get(src_mpt2pg, pte1_idx) > 0,
		    ("%s: source page table page is unused", __func__));
#else
		if (pt2_wirecount_get(src_mpt2pg, pte1_idx) == 0)
			continue;
#endif
		if (nextva > end_addr)
			nextva = end_addr;

		src_pte2p = pt2map_entry(addr);
		while (addr < nextva) {
			pt2_entry_t temp_pte2;
			temp_pte2 = pte2_load(src_pte2p);
			/*
			 * we only virtual copy managed pages
			 */
			if (pte2_is_managed(temp_pte2)) {
				dst_mpt2pg = pmap_allocpte2(dst_pmap, addr,
				    PMAP_ENTER_NOSLEEP);
				if (dst_mpt2pg == NULL)
					goto out;
				dst_pte2p = pmap_pte2_quick(dst_pmap, addr);
				if (!pte2_is_valid(pte2_load(dst_pte2p)) &&
				    pmap_try_insert_pv_entry(dst_pmap, addr,
				    PHYS_TO_VM_PAGE(pte2_pa(temp_pte2)))) {
					/*
					 * Clear the wired, modified, and
					 * accessed (referenced) bits
					 * during the copy.
					 */
					temp_pte2 &=  ~(PTE2_W | PTE2_A);
					temp_pte2 |= PTE2_NM;
					pte2_store(dst_pte2p, temp_pte2);
					dst_pmap->pm_stats.resident_count++;
				} else {
					SLIST_INIT(&free);
					if (pmap_unwire_pt2(dst_pmap, addr,
					    dst_mpt2pg, &free)) {
						pmap_tlb_flush(dst_pmap, addr);
						vm_page_free_pages_toq(&free,
						    false);
					}
					goto out;
				}
				if (pt2_wirecount_get(dst_mpt2pg, pte1_idx) >=
				    pt2_wirecount_get(src_mpt2pg, pte1_idx))
					break;
			}
			addr += PAGE_SIZE;
			src_pte2p++;
		}
	}
out:
	sched_unpin();
	rw_wunlock(&pvh_global_lock);
	PMAP_UNLOCK(src_pmap);
	PMAP_UNLOCK(dst_pmap);
}

/*
 *	Increase the starting virtual address of the given mapping if a
 *	different alignment might result in more section mappings.
 */
void
pmap_align_superpage(vm_object_t object, vm_ooffset_t offset,
    vm_offset_t *addr, vm_size_t size)
{
	vm_offset_t pte1_offset;

	if (size < PTE1_SIZE)
		return;
	if (object != NULL && (object->flags & OBJ_COLORED) != 0)
		offset += ptoa(object->pg_color);
	pte1_offset = offset & PTE1_OFFSET;
	if (size - ((PTE1_SIZE - pte1_offset) & PTE1_OFFSET) < PTE1_SIZE ||
	    (*addr & PTE1_OFFSET) == pte1_offset)
		return;
	if ((*addr & PTE1_OFFSET) < pte1_offset)
		*addr = pte1_trunc(*addr) + pte1_offset;
	else
		*addr = pte1_roundup(*addr) + pte1_offset;
}

void
pmap_activate(struct thread *td)
{
	pmap_t pmap, oldpmap;
	u_int cpuid, ttb;

	PDEBUG(9, printf("%s: td = %08x\n",<