aboutsummaryrefslogtreecommitdiff
path: root/sys/vm/uma_core.c
blob: c44b10cd3f1452e904e3c88af5cc1b3832cbf230 (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
/*-
 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
 *
 * Copyright (c) 2002-2005, 2009, 2013 Jeffrey Roberson <jeff@FreeBSD.org>
 * Copyright (c) 2004, 2005 Bosko Milekic <bmilekic@FreeBSD.org>
 * Copyright (c) 2004-2006 Robert N. M. Watson
 * All rights reserved.
 *
 * 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 unmodified, 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 ``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 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.
 */

/*
 * uma_core.c  Implementation of the Universal Memory allocator
 *
 * This allocator is intended to replace the multitude of similar object caches
 * in the standard FreeBSD kernel.  The intent is to be flexible as well as
 * efficient.  A primary design goal is to return unused memory to the rest of
 * the system.  This will make the system as a whole more flexible due to the
 * ability to move memory to subsystems which most need it instead of leaving
 * pools of reserved memory unused.
 *
 * The basic ideas stem from similar slab/zone based allocators whose algorithms
 * are well known.
 *
 */

/*
 * TODO:
 *	- Improve memory usage for large allocations
 *	- Investigate cache size adjustments
 */

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

#include "opt_ddb.h"
#include "opt_param.h"
#include "opt_vm.h"

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/bitset.h>
#include <sys/eventhandler.h>
#include <sys/kernel.h>
#include <sys/types.h>
#include <sys/limits.h>
#include <sys/queue.h>
#include <sys/malloc.h>
#include <sys/ktr.h>
#include <sys/lock.h>
#include <sys/sysctl.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/random.h>
#include <sys/rwlock.h>
#include <sys/sbuf.h>
#include <sys/sched.h>
#include <sys/smp.h>
#include <sys/taskqueue.h>
#include <sys/vmmeter.h>

#include <vm/vm.h>
#include <vm/vm_object.h>
#include <vm/vm_page.h>
#include <vm/vm_pageout.h>
#include <vm/vm_param.h>
#include <vm/vm_phys.h>
#include <vm/vm_map.h>
#include <vm/vm_kern.h>
#include <vm/vm_extern.h>
#include <vm/uma.h>
#include <vm/uma_int.h>
#include <vm/uma_dbg.h>

#include <ddb/ddb.h>

#ifdef DEBUG_MEMGUARD
#include <vm/memguard.h>
#endif

/*
 * This is the zone and keg from which all zones are spawned.
 */
static uma_zone_t kegs;
static uma_zone_t zones;

/* This is the zone from which all offpage uma_slab_ts are allocated. */
static uma_zone_t slabzone;

/*
 * The initial hash tables come out of this zone so they can be allocated
 * prior to malloc coming up.
 */
static uma_zone_t hashzone;

/* The boot-time adjusted value for cache line alignment. */
int uma_align_cache = 64 - 1;

static MALLOC_DEFINE(M_UMAHASH, "UMAHash", "UMA Hash Buckets");

/*
 * Are we allowed to allocate buckets?
 */
static int bucketdisable = 1;

/* Linked list of all kegs in the system */
static LIST_HEAD(,uma_keg) uma_kegs = LIST_HEAD_INITIALIZER(uma_kegs);

/* Linked list of all cache-only zones in the system */
static LIST_HEAD(,uma_zone) uma_cachezones =
    LIST_HEAD_INITIALIZER(uma_cachezones);

/* This RW lock protects the keg list */
static struct rwlock_padalign __exclusive_cache_line uma_rwlock;

/*
 * Pointer and counter to pool of pages, that is preallocated at
 * startup to bootstrap UMA.
 */
static char *bootmem;
static int boot_pages;

static struct sx uma_drain_lock;

/* kmem soft limit. */
static unsigned long uma_kmem_limit = LONG_MAX;
static volatile unsigned long uma_kmem_total;

/* Is the VM done starting up? */
static enum { BOOT_COLD = 0, BOOT_STRAPPED, BOOT_PAGEALLOC, BOOT_BUCKETS,
    BOOT_RUNNING } booted = BOOT_COLD;

/*
 * This is the handle used to schedule events that need to happen
 * outside of the allocation fast path.
 */
static struct callout uma_callout;
#define	UMA_TIMEOUT	20		/* Seconds for callout interval. */

/*
 * This structure is passed as the zone ctor arg so that I don't have to create
 * a special allocation function just for zones.
 */
struct uma_zctor_args {
	const char *name;
	size_t size;
	uma_ctor ctor;
	uma_dtor dtor;
	uma_init uminit;
	uma_fini fini;
	uma_import import;
	uma_release release;
	void *arg;
	uma_keg_t keg;
	int align;
	uint32_t flags;
};

struct uma_kctor_args {
	uma_zone_t zone;
	size_t size;
	uma_init uminit;
	uma_fini fini;
	int align;
	uint32_t flags;
};

struct uma_bucket_zone {
	uma_zone_t	ubz_zone;
	char		*ubz_name;
	int		ubz_entries;	/* Number of items it can hold. */
	int		ubz_maxsize;	/* Maximum allocation size per-item. */
};

/*
 * Compute the actual number of bucket entries to pack them in power
 * of two sizes for more efficient space utilization.
 */
#define	BUCKET_SIZE(n)						\
    (((sizeof(void *) * (n)) - sizeof(struct uma_bucket)) / sizeof(void *))

#define	BUCKET_MAX	BUCKET_SIZE(256)

struct uma_bucket_zone bucket_zones[] = {
	{ NULL, "4 Bucket", BUCKET_SIZE(4), 4096 },
	{ NULL, "6 Bucket", BUCKET_SIZE(6), 3072 },
	{ NULL, "8 Bucket", BUCKET_SIZE(8), 2048 },
	{ NULL, "12 Bucket", BUCKET_SIZE(12), 1536 },
	{ NULL, "16 Bucket", BUCKET_SIZE(16), 1024 },
	{ NULL, "32 Bucket", BUCKET_SIZE(32), 512 },
	{ NULL, "64 Bucket", BUCKET_SIZE(64), 256 },
	{ NULL, "128 Bucket", BUCKET_SIZE(128), 128 },
	{ NULL, "256 Bucket", BUCKET_SIZE(256), 64 },
	{ NULL, NULL, 0}
};

/*
 * Flags and enumerations to be passed to internal functions.
 */
enum zfreeskip { SKIP_NONE = 0, SKIP_DTOR, SKIP_FINI };

#define	UMA_ANYDOMAIN	-1	/* Special value for domain search. */

/* Prototypes.. */

int	uma_startup_count(int);
void	uma_startup(void *, int);
void	uma_startup1(void);
void	uma_startup2(void);

static void *noobj_alloc(uma_zone_t, vm_size_t, int, uint8_t *, int);
static void *page_alloc(uma_zone_t, vm_size_t, int, uint8_t *, int);
static void *startup_alloc(uma_zone_t, vm_size_t, int, uint8_t *, int);
static void page_free(void *, vm_size_t, uint8_t);
static uma_slab_t keg_alloc_slab(uma_keg_t, uma_zone_t, int, int);
static void cache_drain(uma_zone_t);
static void bucket_drain(uma_zone_t, uma_bucket_t);
static void bucket_cache_drain(uma_zone_t zone);
static int keg_ctor(void *, int, void *, int);
static void keg_dtor(void *, int, void *);
static int zone_ctor(void *, int, void *, int);
static void zone_dtor(void *, int, void *);
static int zero_init(void *, int, int);
static void keg_small_init(uma_keg_t keg);
static void keg_large_init(uma_keg_t keg);
static void zone_foreach(void (*zfunc)(uma_zone_t));
static void zone_timeout(uma_zone_t zone);
static int hash_alloc(struct uma_hash *);
static int hash_expand(struct uma_hash *, struct uma_hash *);
static void hash_free(struct uma_hash *hash);
static void uma_timeout(void *);
static void uma_startup3(void);
static void *zone_alloc_item(uma_zone_t, void *, int, int);
static void zone_free_item(uma_zone_t, void *, void *, enum zfreeskip);
static void bucket_enable(void);
static void bucket_init(void);
static uma_bucket_t bucket_alloc(uma_zone_t zone, void *, int);
static void bucket_free(uma_zone_t zone, uma_bucket_t, void *);
static void bucket_zone_drain(void);
static uma_bucket_t zone_alloc_bucket(uma_zone_t, void *, int, int);
static uma_slab_t zone_fetch_slab(uma_zone_t, uma_keg_t, int, int);
static uma_slab_t zone_fetch_slab_multi(uma_zone_t, uma_keg_t, int, int);
static void *slab_alloc_item(uma_keg_t keg, uma_slab_t slab);
static void slab_free_item(uma_keg_t keg, uma_slab_t slab, void *item);
static uma_keg_t uma_kcreate(uma_zone_t zone, size_t size, uma_init uminit,
    uma_fini fini, int align, uint32_t flags);
static int zone_import(uma_zone_t, void **, int, int, int);
static void zone_release(uma_zone_t, void **, int);
static void uma_zero_item(void *, uma_zone_t);

void uma_print_zone(uma_zone_t);
void uma_print_stats(void);
static int sysctl_vm_zone_count(SYSCTL_HANDLER_ARGS);
static int sysctl_vm_zone_stats(SYSCTL_HANDLER_ARGS);

#ifdef INVARIANTS
static bool uma_dbg_kskip(uma_keg_t keg, void *mem);
static bool uma_dbg_zskip(uma_zone_t zone, void *mem);
static void uma_dbg_free(uma_zone_t zone, uma_slab_t slab, void *item);
static void uma_dbg_alloc(uma_zone_t zone, uma_slab_t slab, void *item);

static SYSCTL_NODE(_vm, OID_AUTO, debug, CTLFLAG_RD, 0,
    "Memory allocation debugging");

static u_int dbg_divisor = 1;
SYSCTL_UINT(_vm_debug, OID_AUTO, divisor,
    CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &dbg_divisor, 0,
    "Debug & thrash every this item in memory allocator");

static counter_u64_t uma_dbg_cnt = EARLY_COUNTER;
static counter_u64_t uma_skip_cnt = EARLY_COUNTER;
SYSCTL_COUNTER_U64(_vm_debug, OID_AUTO, trashed, CTLFLAG_RD,
    &uma_dbg_cnt, "memory items debugged");
SYSCTL_COUNTER_U64(_vm_debug, OID_AUTO, skipped, CTLFLAG_RD,
    &uma_skip_cnt, "memory items skipped, not debugged");
#endif

SYSINIT(uma_startup3, SI_SUB_VM_CONF, SI_ORDER_SECOND, uma_startup3, NULL);

SYSCTL_PROC(_vm, OID_AUTO, zone_count, CTLFLAG_RD|CTLTYPE_INT,
    0, 0, sysctl_vm_zone_count, "I", "Number of UMA zones");

SYSCTL_PROC(_vm, OID_AUTO, zone_stats, CTLFLAG_RD|CTLTYPE_STRUCT,
    0, 0, sysctl_vm_zone_stats, "s,struct uma_type_header", "Zone Stats");

static int zone_warnings = 1;
SYSCTL_INT(_vm, OID_AUTO, zone_warnings, CTLFLAG_RWTUN, &zone_warnings, 0,
    "Warn when UMA zones becomes full");

/* Adjust bytes under management by UMA. */
static inline void
uma_total_dec(unsigned long size)
{

	atomic_subtract_long(&uma_kmem_total, size);
}

static inline void
uma_total_inc(unsigned long size)
{

	if (atomic_fetchadd_long(&uma_kmem_total, size) > uma_kmem_limit)
		uma_reclaim_wakeup();
}

/*
 * This routine checks to see whether or not it's safe to enable buckets.
 */
static void
bucket_enable(void)
{
	bucketdisable = vm_page_count_min();
}

/*
 * Initialize bucket_zones, the array of zones of buckets of various sizes.
 *
 * For each zone, calculate the memory required for each bucket, consisting
 * of the header and an array of pointers.
 */
static void
bucket_init(void)
{
	struct uma_bucket_zone *ubz;
	int size;

	for (ubz = &bucket_zones[0]; ubz->ubz_entries != 0; ubz++) {
		size = roundup(sizeof(struct uma_bucket), sizeof(void *));
		size += sizeof(void *) * ubz->ubz_entries;
		ubz->ubz_zone = uma_zcreate(ubz->ubz_name, size,
		    NULL, NULL, NULL, NULL, UMA_ALIGN_PTR,
		    UMA_ZONE_MTXCLASS | UMA_ZFLAG_BUCKET | UMA_ZONE_NUMA);
	}
}

/*
 * Given a desired number of entries for a bucket, return the zone from which
 * to allocate the bucket.
 */
static struct uma_bucket_zone *
bucket_zone_lookup(int entries)
{
	struct uma_bucket_zone *ubz;

	for (ubz = &bucket_zones[0]; ubz->ubz_entries != 0; ubz++)
		if (ubz->ubz_entries >= entries)
			return (ubz);
	ubz--;
	return (ubz);
}

static int
bucket_select(int size)
{
	struct uma_bucket_zone *ubz;

	ubz = &bucket_zones[0];
	if (size > ubz->ubz_maxsize)
		return MAX((ubz->ubz_maxsize * ubz->ubz_entries) / size, 1);

	for (; ubz->ubz_entries != 0; ubz++)
		if (ubz->ubz_maxsize < size)
			break;
	ubz--;
	return (ubz->ubz_entries);
}

static uma_bucket_t
bucket_alloc(uma_zone_t zone, void *udata, int flags)
{
	struct uma_bucket_zone *ubz;
	uma_bucket_t bucket;

	/*
	 * This is to stop us from allocating per cpu buckets while we're
	 * running out of vm.boot_pages.  Otherwise, we would exhaust the
	 * boot pages.  This also prevents us from allocating buckets in
	 * low memory situations.
	 */
	if (bucketdisable)
		return (NULL);
	/*
	 * To limit bucket recursion we store the original zone flags
	 * in a cookie passed via zalloc_arg/zfree_arg.  This allows the
	 * NOVM flag to persist even through deep recursions.  We also
	 * store ZFLAG_BUCKET once we have recursed attempting to allocate
	 * a bucket for a bucket zone so we do not allow infinite bucket
	 * recursion.  This cookie will even persist to frees of unused
	 * buckets via the allocation path or bucket allocations in the
	 * free path.
	 */
	if ((zone->uz_flags & UMA_ZFLAG_BUCKET) == 0)
		udata = (void *)(uintptr_t)zone->uz_flags;
	else {
		if ((uintptr_t)udata & UMA_ZFLAG_BUCKET)
			return (NULL);
		udata = (void *)((uintptr_t)udata | UMA_ZFLAG_BUCKET);
	}
	if ((uintptr_t)udata & UMA_ZFLAG_CACHEONLY)
		flags |= M_NOVM;
	ubz = bucket_zone_lookup(zone->uz_count);
	if (ubz->ubz_zone == zone && (ubz + 1)->ubz_entries != 0)
		ubz++;
	bucket = uma_zalloc_arg(ubz->ubz_zone, udata, flags);
	if (bucket) {
#ifdef INVARIANTS
		bzero(bucket->ub_bucket, sizeof(void *) * ubz->ubz_entries);
#endif
		bucket->ub_cnt = 0;
		bucket->ub_entries = ubz->ubz_entries;
	}

	return (bucket);
}

static void
bucket_free(uma_zone_t zone, uma_bucket_t bucket, void *udata)
{
	struct uma_bucket_zone *ubz;

	KASSERT(bucket->ub_cnt == 0,
	    ("bucket_free: Freeing a non free bucket."));
	if ((zone->uz_flags & UMA_ZFLAG_BUCKET) == 0)
		udata = (void *)(uintptr_t)zone->uz_flags;
	ubz = bucket_zone_lookup(bucket->ub_entries);
	uma_zfree_arg(ubz->ubz_zone, bucket, udata);
}

static void
bucket_zone_drain(void)
{
	struct uma_bucket_zone *ubz;

	for (ubz = &bucket_zones[0]; ubz->ubz_entries != 0; ubz++)
		zone_drain(ubz->ubz_zone);
}

static void
zone_log_warning(uma_zone_t zone)
{
	static const struct timeval warninterval = { 300, 0 };

	if (!zone_warnings || zone->uz_warning == NULL)
		return;

	if (ratecheck(&zone->uz_ratecheck, &warninterval))
		printf("[zone: %s] %s\n", zone->uz_name, zone->uz_warning);
}

static inline void
zone_maxaction(uma_zone_t zone)
{

	if (zone->uz_maxaction.ta_func != NULL)
		taskqueue_enqueue(taskqueue_thread, &zone->uz_maxaction);
}

static void
zone_foreach_keg(uma_zone_t zone, void (*kegfn)(uma_keg_t))
{
	uma_klink_t klink;

	LIST_FOREACH(klink, &zone->uz_kegs, kl_link)
		kegfn(klink->kl_keg);
}

/*
 * Routine called by timeout which is used to fire off some time interval
 * based calculations.  (stats, hash size, etc.)
 *
 * Arguments:
 *	arg   Unused
 *
 * Returns:
 *	Nothing
 */
static void
uma_timeout(void *unused)
{
	bucket_enable();
	zone_foreach(zone_timeout);

	/* Reschedule this event */
	callout_reset(&uma_callout, UMA_TIMEOUT * hz, uma_timeout, NULL);
}

/*
 * Routine to perform timeout driven calculations.  This expands the
 * hashes and does per cpu statistics aggregation.
 *
 *  Returns nothing.
 */
static void
keg_timeout(uma_keg_t keg)
{

	KEG_LOCK(keg);
	/*
	 * Expand the keg hash table.
	 *
	 * This is done if the number of slabs is larger than the hash size.
	 * What I'm trying to do here is completely reduce collisions.  This
	 * may be a little aggressive.  Should I allow for two collisions max?
	 */
	if (keg->uk_flags & UMA_ZONE_HASH &&
	    keg->uk_pages / keg->uk_ppera >= keg->uk_hash.uh_hashsize) {
		struct uma_hash newhash;
		struct uma_hash oldhash;
		int ret;

		/*
		 * This is so involved because allocating and freeing
		 * while the keg lock is held will lead to deadlock.
		 * I have to do everything in stages and check for
		 * races.
		 */
		newhash = keg->uk_hash;
		KEG_UNLOCK(keg);
		ret = hash_alloc(&newhash);
		KEG_LOCK(keg);
		if (ret) {
			if (hash_expand(&keg->uk_hash, &newhash)) {
				oldhash = keg->uk_hash;
				keg->uk_hash = newhash;
			} else
				oldhash = newhash;

			KEG_UNLOCK(keg);
			hash_free(&oldhash);
			return;
		}
	}
	KEG_UNLOCK(keg);
}

static void
zone_timeout(uma_zone_t zone)
{

	zone_foreach_keg(zone, &keg_timeout);
}

/*
 * Allocate and zero fill the next sized hash table from the appropriate
 * backing store.
 *
 * Arguments:
 *	hash  A new hash structure with the old hash size in uh_hashsize
 *
 * Returns:
 *	1 on success and 0 on failure.
 */
static int
hash_alloc(struct uma_hash *hash)
{
	int oldsize;
	int alloc;

	oldsize = hash->uh_hashsize;

	/* We're just going to go to a power of two greater */
	if (oldsize)  {
		hash->uh_hashsize = oldsize * 2;
		alloc = sizeof(hash->uh_slab_hash[0]) * hash->uh_hashsize;
		hash->uh_slab_hash = (struct slabhead *)malloc(alloc,
		    M_UMAHASH, M_NOWAIT);
	} else {
		alloc = sizeof(hash->uh_slab_hash[0]) * UMA_HASH_SIZE_INIT;
		hash->uh_slab_hash = zone_alloc_item(hashzone, NULL,
		    UMA_ANYDOMAIN, M_WAITOK);
		hash->uh_hashsize = UMA_HASH_SIZE_INIT;
	}
	if (hash->uh_slab_hash) {
		bzero(hash->uh_slab_hash, alloc);
		hash->uh_hashmask = hash->uh_hashsize - 1;
		return (1);
	}

	return (0);
}

/*
 * Expands the hash table for HASH zones.  This is done from zone_timeout
 * to reduce collisions.  This must not be done in the regular allocation
 * path, otherwise, we can recurse on the vm while allocating pages.
 *
 * Arguments:
 *	oldhash  The hash you want to expand
 *	newhash  The hash structure for the new table
 *
 * Returns:
 *	Nothing
 *
 * Discussion:
 */
static int
hash_expand(struct uma_hash *oldhash, struct uma_hash *newhash)
{
	uma_slab_t slab;
	int hval;
	int i;

	if (!newhash->uh_slab_hash)
		return (0);

	if (oldhash->uh_hashsize >= newhash->uh_hashsize)
		return (0);

	/*
	 * I need to investigate hash algorithms for resizing without a
	 * full rehash.
	 */

	for (i = 0; i < oldhash->uh_hashsize; i++)
		while (!SLIST_EMPTY(&oldhash->uh_slab_hash[i])) {
			slab = SLIST_FIRST(&oldhash->uh_slab_hash[i]);
			SLIST_REMOVE_HEAD(&oldhash->uh_slab_hash[i], us_hlink);
			hval = UMA_HASH(newhash, slab->us_data);
			SLIST_INSERT_HEAD(&newhash->uh_slab_hash[hval],
			    slab, us_hlink);
		}

	return (1);
}

/*
 * Free the hash bucket to the appropriate backing store.
 *
 * Arguments:
 *	slab_hash  The hash bucket we're freeing
 *	hashsize   The number of entries in that hash bucket
 *
 * Returns:
 *	Nothing
 */
static void
hash_free(struct uma_hash *hash)
{
	if (hash->uh_slab_hash == NULL)
		return;
	if (hash->uh_hashsize == UMA_HASH_SIZE_INIT)
		zone_free_item(hashzone, hash->uh_slab_hash, NULL, SKIP_NONE);
	else
		free(hash->uh_slab_hash, M_UMAHASH);
}

/*
 * Frees all outstanding items in a bucket
 *
 * Arguments:
 *	zone   The zone to free to, must be unlocked.
 *	bucket The free/alloc bucket with items, cpu queue must be locked.
 *
 * Returns:
 *	Nothing
 */

static void
bucket_drain(uma_zone_t zone, uma_bucket_t bucket)
{
	int i;

	if (bucket == NULL)
		return;

	if (zone->uz_fini)
		for (i = 0; i < bucket->ub_cnt; i++) 
			zone->uz_fini(bucket->ub_bucket[i], zone->uz_size);
	zone->uz_release(zone->uz_arg, bucket->ub_bucket, bucket->ub_cnt);
	bucket->ub_cnt = 0;
}

/*
 * Drains the per cpu caches for a zone.
 *
 * NOTE: This may only be called while the zone is being turn down, and not
 * during normal operation.  This is necessary in order that we do not have
 * to migrate CPUs to drain the per-CPU caches.
 *
 * Arguments:
 *	zone     The zone to drain, must be unlocked.
 *
 * Returns:
 *	Nothing
 */
static void
cache_drain(uma_zone_t zone)
{
	uma_cache_t cache;
	int cpu;

	/*
	 * XXX: It is safe to not lock the per-CPU caches, because we're
	 * tearing down the zone anyway.  I.e., there will be no further use
	 * of the caches at this point.
	 *
	 * XXX: It would good to be able to assert that the zone is being
	 * torn down to prevent improper use of cache_drain().
	 *
	 * XXX: We lock the zone before passing into bucket_cache_drain() as
	 * it is used elsewhere.  Should the tear-down path be made special
	 * there in some form?
	 */
	CPU_FOREACH(cpu) {
		cache = &zone->uz_cpu[cpu];
		bucket_drain(zone, cache->uc_allocbucket);
		bucket_drain(zone, cache->uc_freebucket);
		if (cache->uc_allocbucket != NULL)
			bucket_free(zone, cache->uc_allocbucket, NULL);
		if (cache->uc_freebucket != NULL)
			bucket_free(zone, cache->uc_freebucket, NULL);
		cache->uc_allocbucket = cache->uc_freebucket = NULL;
	}
	ZONE_LOCK(zone);
	bucket_cache_drain(zone);
	ZONE_UNLOCK(zone);
}

static void
cache_shrink(uma_zone_t zone)
{

	if (zone->uz_flags & UMA_ZFLAG_INTERNAL)
		return;

	ZONE_LOCK(zone);
	zone->uz_count = (zone->uz_count_min + zone->uz_count) / 2;
	ZONE_UNLOCK(zone);
}

static void
cache_drain_safe_cpu(uma_zone_t zone)
{
	uma_cache_t cache;
	uma_bucket_t b1, b2;
	int domain;

	if (zone->uz_flags & UMA_ZFLAG_INTERNAL)
		return;

	b1 = b2 = NULL;
	ZONE_LOCK(zone);
	critical_enter();
	if (zone->uz_flags & UMA_ZONE_NUMA)
		domain = PCPU_GET(domain);
	else
		domain = 0;
	cache = &zone->uz_cpu[curcpu];
	if (cache->uc_allocbucket) {
		if (cache->uc_allocbucket->ub_cnt != 0)
			LIST_INSERT_HEAD(&zone->uz_domain[domain].uzd_buckets,
			    cache->uc_allocbucket, ub_link);
		else
			b1 = cache->uc_allocbucket;
		cache->uc_allocbucket = NULL;
	}
	if (cache->uc_freebucket) {
		if (cache->uc_freebucket->ub_cnt != 0)
			LIST_INSERT_HEAD(&zone->uz_domain[domain].uzd_buckets,
			    cache->uc_freebucket, ub_link);
		else
			b2 = cache->uc_freebucket;
		cache->uc_freebucket = NULL;
	}
	critical_exit();
	ZONE_UNLOCK(zone);
	if (b1)
		bucket_free(zone, b1, NULL);
	if (b2)
		bucket_free(zone, b2, NULL);
}

/*
 * Safely drain per-CPU caches of a zone(s) to alloc bucket.
 * This is an expensive call because it needs to bind to all CPUs
 * one by one and enter a critical section on each of them in order
 * to safely access their cache buckets.
 * Zone lock must not be held on call this function.
 */
static void
cache_drain_safe(uma_zone_t zone)
{
	int cpu;

	/*
	 * Polite bucket sizes shrinking was not enouth, shrink aggressively.
	 */
	if (zone)
		cache_shrink(zone);
	else
		zone_foreach(cache_shrink);

	CPU_FOREACH(cpu) {
		thread_lock(curthread);
		sched_bind(curthread, cpu);
		thread_unlock(curthread);

		if (zone)
			cache_drain_safe_cpu(zone);
		else
			zone_foreach(cache_drain_safe_cpu);
	}
	thread_lock(curthread);
	sched_unbind(curthread);
	thread_unlock(curthread);
}

/*
 * Drain the cached buckets from a zone.  Expects a locked zone on entry.
 */
static void
bucket_cache_drain(uma_zone_t zone)
{
	uma_zone_domain_t zdom;
	uma_bucket_t bucket;
	int i;

	/*
	 * Drain the bucket queues and free the buckets.
	 */
	for (i = 0; i < vm_ndomains; i++) {
		zdom = &zone->uz_domain[i];
		while ((bucket = LIST_FIRST(&zdom->uzd_buckets)) != NULL) {
			LIST_REMOVE(bucket, ub_link);
			ZONE_UNLOCK(zone);
			bucket_drain(zone, bucket);
			bucket_free(zone, bucket, NULL);
			ZONE_LOCK(zone);
		}
	}

	/*
	 * Shrink further bucket sizes.  Price of single zone lock collision
	 * is probably lower then price of global cache drain.
	 */
	if (zone->uz_count > zone->uz_count_min)
		zone->uz_count--;
}

static void
keg_free_slab(uma_keg_t keg, uma_slab_t slab, int start)
{
	uint8_t *mem;
	int i;
	uint8_t flags;

	CTR4(KTR_UMA, "keg_free_slab keg %s(%p) slab %p, returning %d bytes",
	    keg->uk_name, keg, slab, PAGE_SIZE * keg->uk_ppera);

	mem = slab->us_data;
	flags = slab->us_flags;
	i = start;
	if (keg->uk_fini != NULL) {
		for (i--; i > -1; i--)
#ifdef INVARIANTS
		/*
		 * trash_fini implies that dtor was trash_dtor. trash_fini
		 * would check that memory hasn't been modified since free,
		 * which executed trash_dtor.
		 * That's why we need to run uma_dbg_kskip() check here,
		 * albeit we don't make skip check for other init/fini
		 * invocations.
		 */
		if (!uma_dbg_kskip(keg, slab->us_data + (keg->uk_rsize * i)) ||
		    keg->uk_fini != trash_fini)
#endif
			keg->uk_fini(slab->us_data + (keg->uk_rsize * i),
			    keg->uk_size);
	}
	if (keg->uk_flags & UMA_ZONE_OFFPAGE)
		zone_free_item(keg->uk_slabzone, slab, NULL, SKIP_NONE);
	keg->uk_freef(mem, PAGE_SIZE * keg->uk_ppera, flags);
	uma_total_dec(PAGE_SIZE * keg->uk_ppera);
}

/*
 * Frees pages from a keg back to the system.  This is done on demand from
 * the pageout daemon.
 *
 * Returns nothing.
 */
static void
keg_drain(uma_keg_t keg)
{
	struct slabhead freeslabs = { 0 };
	uma_domain_t dom;
	uma_slab_t slab, tmp;
	int i;

	/*
	 * We don't want to take pages from statically allocated kegs at this
	 * time
	 */
	if (keg->uk_flags & UMA_ZONE_NOFREE || keg->uk_freef == NULL)
		return;

	CTR3(KTR_UMA, "keg_drain %s(%p) free items: %u",
	    keg->uk_name, keg, keg->uk_free);
	KEG_LOCK(keg);
	if (keg->uk_free == 0)
		goto finished;

	for (i = 0; i < vm_ndomains; i++) {
		dom = &keg->uk_domain[i];
		LIST_FOREACH_SAFE(slab, &dom->ud_free_slab, us_link, tmp) {
			/* We have nowhere to free these to. */
			if (slab->us_flags & UMA_SLAB_BOOT)
				continue;

			LIST_REMOVE(slab, us_link);
			keg->uk_pages -= keg->uk_ppera;
			keg->uk_free -= keg->uk_ipers;

			if (keg->uk_flags & UMA_ZONE_HASH)
				UMA_HASH_REMOVE(&keg->uk_hash, slab,
				    slab->us_data);

			SLIST_INSERT_HEAD(&freeslabs, slab, us_hlink);
		}
	}

finished:
	KEG_UNLOCK(keg);

	while ((slab = SLIST_FIRST(&freeslabs)) != NULL) {
		SLIST_REMOVE(&freeslabs, slab, uma_slab, us_hlink);
		keg_free_slab(keg, slab, keg->uk_ipers);
	}
}

static void
zone_drain_wait(uma_zone_t zone, int waitok)
{

	/*
	 * Set draining to interlock with zone_dtor() so we can release our
	 * locks as we go.  Only dtor() should do a WAITOK call since it
	 * is the only call that knows the structure will still be available
	 * when it wakes up.
	 */
	ZONE_LOCK(zone);
	while (zone->uz_flags & UMA_ZFLAG_DRAINING) {
		if (waitok == M_NOWAIT)
			goto out;
		msleep(zone, zone->uz_lockptr, PVM, "zonedrain", 1);
	}
	zone->uz_flags |= UMA_ZFLAG_DRAINING;
	bucket_cache_drain(zone);
	ZONE_UNLOCK(zone);
	/*
	 * The DRAINING flag protects us from being freed while
	 * we're running.  Normally the uma_rwlock would protect us but we
	 * must be able to release and acquire the right lock for each keg.
	 */
	zone_foreach_keg(zone, &keg_drain);
	ZONE_LOCK(zone);
	zone->uz_flags &= ~UMA_ZFLAG_DRAINING;
	wakeup(zone);
out:
	ZONE_UNLOCK(zone);
}

void
zone_drain(uma_zone_t zone)
{

	zone_drain_wait(zone, M_NOWAIT);
}

/*
 * Allocate a new slab for a keg.  This does not insert the slab onto a list.
 *
 * Arguments:
 *	wait  Shall we wait?
 *
 * Returns:
 *	The slab that was allocated or NULL if there is no memory and the
 *	caller specified M_NOWAIT.
 */
static uma_slab_t
keg_alloc_slab(uma_keg_t keg, uma_zone_t zone, int domain, int wait)
{
	uma_alloc allocf;
	uma_slab_t slab;
	unsigned long size;
	uint8_t *mem;
	uint8_t flags;
	int i;

	KASSERT(domain >= 0 && domain < vm_ndomains,
	    ("keg_alloc_slab: domain %d out of range", domain));
	mtx_assert(&keg->uk_lock, MA_OWNED);
	slab = NULL;
	mem = NULL;

	allocf = keg->uk_allocf;
	KEG_UNLOCK(keg);
	size = keg->uk_ppera * PAGE_SIZE;

	if (keg->uk_flags & UMA_ZONE_OFFPAGE) {
		slab = zone_alloc_item(keg->uk_slabzone, NULL, domain, wait);
		if (slab == NULL)
			goto out;
	}

	/*
	 * This reproduces the old vm_zone behavior of zero filling pages the
	 * first time they are added to a zone.
	 *
	 * Malloced items are zeroed in uma_zalloc.
	 */

	if ((keg->uk_flags & UMA_ZONE_MALLOC) == 0)
		wait |= M_ZERO;
	else
		wait &= ~M_ZERO;

	if (keg->uk_flags & UMA_ZONE_NODUMP)
		wait |= M_NODUMP;

	/* zone is passed for legacy reasons. */
	mem = allocf(zone, size, domain, &flags, wait);
	if (mem == NULL) {
		if (keg->uk_flags & UMA_ZONE_OFFPAGE)
			zone_free_item(keg->uk_slabzone, slab, NULL, SKIP_NONE);
		slab = NULL;
		goto out;
	}
	uma_total_inc(size);

	/* Point the slab into the allocated memory */
	if (!(keg->uk_flags & UMA_ZONE_OFFPAGE))
		slab = (uma_slab_t )(mem + keg->uk_pgoff);

	if (keg->uk_flags & UMA_ZONE_VTOSLAB)
		for (i = 0; i < keg->uk_ppera; i++)
			vsetslab((vm_offset_t)mem + (i * PAGE_SIZE), slab);

	slab->us_keg = keg;
	slab->us_data = mem;
	slab->us_freecount = keg->uk_ipers;
	slab->us_flags = flags;
	slab->us_domain = domain;
	BIT_FILL(SLAB_SETSIZE, &slab->us_free);
#ifdef INVARIANTS
	BIT_ZERO(SLAB_SETSIZE, &slab->us_debugfree);
#endif

	if (keg->uk_init != NULL) {
		for (i = 0; i < keg->uk_ipers; i++)
			if (keg->uk_init(slab->us_data + (keg->uk_rsize * i),
			    keg->uk_size, wait) != 0)
				break;
		if (i != keg->uk_ipers) {
			keg_free_slab(keg, slab, i);
			slab = NULL;
			goto out;
		}
	}
out:
	KEG_LOCK(keg);

	CTR3(KTR_UMA, "keg_alloc_slab: allocated slab %p for %s(%p)",
	    slab, keg->uk_name, keg);

	if (slab != NULL) {
		if (keg->uk_flags & UMA_ZONE_HASH)
			UMA_HASH_INSERT(&keg->uk_hash, slab, mem);

		keg->uk_pages += keg->uk_ppera;
		keg->uk_free += keg->uk_ipers;
	}

	return (slab);
}

/*
 * This function is intended to be used early on in place of page_alloc() so
 * that we may use the boot time page cache to satisfy allocations before
 * the VM is ready.
 */
static void *
startup_alloc(uma_zone_t zone, vm_size_t bytes, int domain, uint8_t *pflag,
    int wait)
{
	uma_keg_t keg;
	void *mem;
	int pages;

	keg = zone_first_keg(zone);

	/*
	 * If we are in BOOT_BUCKETS or higher, than switch to real
	 * allocator.  Zones with page sized slabs switch at BOOT_PAGEALLOC.
	 */
	switch (booted) {
		case BOOT_COLD:
		case BOOT_STRAPPED:
			break;
		case BOOT_PAGEALLOC:
			if (keg->uk_ppera > 1)
				break;
		case BOOT_BUCKETS:
		case BOOT_RUNNING:
#ifdef UMA_MD_SMALL_ALLOC
			keg->uk_allocf = (keg->uk_ppera > 1) ?
			    page_alloc : uma_small_alloc;
#else
			keg->uk_allocf = page_alloc;
#endif
			return keg->uk_allocf(zone, bytes, domain, pflag, wait);
	}

	/*
	 * Check our small startup cache to see if it has pages remaining.
	 */
	pages = howmany(bytes, PAGE_SIZE);
	KASSERT(pages > 0, ("%s can't reserve 0 pages", __func__));
	if (pages > boot_pages)
		panic("UMA zone \"%s\": Increase vm.boot_pages", zone->uz_name);
#ifdef DIAGNOSTIC
	printf("%s from \"%s\", %d boot pages left\n", __func__, zone->uz_name,
	    boot_pages);
#endif
	mem = bootmem;
	boot_pages -= pages;
	bootmem += pages * PAGE_SIZE;
	*pflag = UMA_SLAB_BOOT;

	return (mem);
}

/*
 * Allocates a number of pages from the system
 *
 * Arguments:
 *	bytes  The number of bytes requested
 *	wait  Shall we wait?
 *
 * Returns:
 *	A pointer to the alloced memory or possibly
 *	NULL if M_NOWAIT is set.
 */
static void *
page_alloc(uma_zone_t zone, vm_size_t bytes, int domain, uint8_t *pflag,
    int wait)
{
	void *p;	/* Returned page */

	*pflag = UMA_SLAB_KERNEL;
	p = (void *) kmem_malloc_domain(domain, bytes, wait);

	return (p);
}

/*
 * Allocates a number of pages from within an object
 *
 * Arguments:
 *	bytes  The number of bytes requested
 *	wait   Shall we wait?
 *
 * Returns:
 *	A pointer to the alloced memory or possibly
 *	NULL if M_NOWAIT is set.
 */
static void *
noobj_alloc(uma_zone_t zone, vm_size_t bytes, int domain, uint8_t *flags,
    int wait)
{
	TAILQ_HEAD(, vm_page) alloctail;
	u_long npages;
	vm_offset_t retkva, zkva;
	vm_page_t p, p_next;
	uma_keg_t keg;

	TAILQ_INIT(&alloctail);
	keg = zone_first_keg(zone);

	npages = howmany(bytes, PAGE_SIZE);
	while (npages > 0) {
		p = vm_page_alloc_domain(NULL, 0, domain, VM_ALLOC_INTERRUPT |
		    VM_ALLOC_WIRED | VM_ALLOC_NOOBJ |
		    ((wait & M_WAITOK) != 0 ? VM_ALLOC_WAITOK :
		    VM_ALLOC_NOWAIT));
		if (p != NULL) {
			/*
			 * Since the page does not belong to an object, its
			 * listq is unused.
			 */
			TAILQ_INSERT_TAIL(&alloctail, p, listq);
			npages--;
			continue;
		}
		/*
		 * Page allocation failed, free intermediate pages and
		 * exit.
		 */
		TAILQ_FOREACH_SAFE(p, &alloctail, listq, p_next) {
			vm_page_unwire(p, PQ_NONE);
			vm_page_free(p); 
		}
		return (NULL);
	}
	*flags = UMA_SLAB_PRIV;
	zkva = keg->uk_kva +
	    atomic_fetchadd_long(&keg->uk_offset, round_page(bytes));
	retkva = zkva;
	TAILQ_FOREACH(p, &alloctail, listq) {
		pmap_qenter(zkva, &p, 1);
		zkva += PAGE_SIZE;
	}

	return ((void *)retkva);
}

/*
 * Frees a number of pages to the system
 *
 * Arguments:
 *	mem   A pointer to the memory to be freed
 *	size  The size of the memory being freed
 *	flags The original p->us_flags field
 *
 * Returns:
 *	Nothing
 */
static void
page_free(void *mem, vm_size_t size, uint8_t flags)
{
	struct vmem *vmem;

	if (flags & UMA_SLAB_KERNEL)
		vmem = kernel_arena;
	else
		panic("UMA: page_free used with invalid flags %x", flags);

	kmem_free(vmem, (vm_offset_t)mem, size);
}

/*
 * Zero fill initializer
 *
 * Arguments/Returns follow uma_init specifications
 */
static int
zero_init(void *mem, int size, int flags)
{
	bzero(mem, size);
	return (0);
}

/*
 * Finish creating a small uma keg.  This calculates ipers, and the keg size.
 *
 * Arguments
 *	keg  The zone we should initialize
 *
 * Returns
 *	Nothing
 */
static void
keg_small_init(uma_keg_t keg)
{
	u_int rsize;
	u_int memused;
	u_int wastedspace;
	u_int shsize;
	u_int slabsize;

	if (keg->uk_flags & UMA_ZONE_PCPU) {
		u_int ncpus = (mp_maxid + 1) ? (mp_maxid + 1) : MAXCPU;

		slabsize = sizeof(struct pcpu);
		keg->uk_ppera = howmany(ncpus * sizeof(struct pcpu),
		    PAGE_SIZE);
	} else {
		slabsize = UMA_SLAB_SIZE;
		keg->uk_ppera = 1;
	}

	/*
	 * Calculate the size of each allocation (rsize) according to
	 * alignment.  If the requested size is smaller than we have
	 * allocation bits for we round it up.
	 */
	rsize = keg->uk_size;
	if (rsize < slabsize / SLAB_SETSIZE)
		rsize = slabsize / SLAB_SETSIZE;
	if (rsize & keg->uk_align)
		rsize = (rsize & ~keg->uk_align) + (keg->uk_align + 1);
	keg->uk_rsize = rsize;

	KASSERT((keg->uk_flags & UMA_ZONE_PCPU) == 0 ||
	    keg->uk_rsize < sizeof(struct pcpu),
	    ("%s: size %u too large", __func__, keg->uk_rsize));

	if (keg->uk_flags & UMA_ZONE_OFFPAGE)
		shsize = 0;
	else 
		shsize = sizeof(struct uma_slab);

	if (rsize <= slabsize - shsize)
		keg->uk_ipers = (slabsize - shsize) / rsize;
	else {
		/* Handle special case when we have 1 item per slab, so
		 * alignment requirement can be relaxed. */
		KASSERT(keg->uk_size <= slabsize - shsize,
		    ("%s: size %u greater than slab", __func__, keg->uk_size));
		keg->uk_ipers = 1;
	}
	KASSERT(keg->uk_ipers > 0 && keg->uk_ipers <= SLAB_SETSIZE,
	    ("%s: keg->uk_ipers %u", __func__, keg->uk_ipers));

	memused = keg->uk_ipers * rsize + shsize;
	wastedspace = slabsize - memused;

	/*
	 * We can't do OFFPAGE if we're internal or if we've been
	 * asked to not go to the VM for buckets.  If we do this we
	 * may end up going to the VM  for slabs which we do not
	 * want to do if we're UMA_ZFLAG_CACHEONLY as a result
	 * of UMA_ZONE_VM, which clearly forbids it.
	 */
	if ((keg->uk_flags & UMA_ZFLAG_INTERNAL) ||
	    (keg->uk_flags & UMA_ZFLAG_CACHEONLY))
		return;

	/*
	 * See if using an OFFPAGE slab will limit our waste.  Only do
	 * this if it permits more items per-slab.
	 *
	 * XXX We could try growing slabsize to limit max waste as well.
	 * Historically this was not done because the VM could not
	 * efficiently handle contiguous allocations.
	 */
	if ((wastedspace >= slabsize / UMA_MAX_WASTE) &&
	    (keg->uk_ipers < (slabsize / keg->uk_rsize))) {
		keg->uk_ipers = slabsize / keg->uk_rsize;
		KASSERT(keg->uk_ipers > 0 && keg->uk_ipers <= SLAB_SETSIZE,
		    ("%s: keg->uk_ipers %u", __func__, keg->uk_ipers));
		CTR6(KTR_UMA, "UMA decided we need offpage slab headers for "
		    "keg: %s(%p), calculated wastedspace = %d, "
		    "maximum wasted space allowed = %d, "
		    "calculated ipers = %d, "
		    "new wasted space = %d\n", keg->uk_name, keg, wastedspace,
		    slabsize / UMA_MAX_WASTE, keg->uk_ipers,
		    slabsize - keg->uk_ipers * keg->uk_rsize);
		keg->uk_flags |= UMA_ZONE_OFFPAGE;
	}

	if ((keg->uk_flags & UMA_ZONE_OFFPAGE) &&
	    (keg->uk_flags & UMA_ZONE_VTOSLAB) == 0)
		keg->uk_flags |= UMA_ZONE_HASH;
}

/*
 * Finish creating a large (> UMA_SLAB_SIZE) uma kegs.  Just give in and do
 * OFFPAGE for now.  When I can allow for more dynamic slab sizes this will be
 * more complicated.
 *
 * Arguments
 *	keg  The keg we should initialize
 *
 * Returns
 *	Nothing
 */
static void
keg_large_init(uma_keg_t keg)
{
	u_int shsize;

	KASSERT(keg != NULL, ("Keg is null in keg_large_init"));
	KASSERT((keg->uk_flags & UMA_ZFLAG_CACHEONLY) == 0,
	    ("keg_large_init: Cannot large-init a UMA_ZFLAG_CACHEONLY keg"));
	KASSERT((keg->uk_flags & UMA_ZONE_PCPU) == 0,
	    ("%s: Cannot large-init a UMA_ZONE_PCPU keg", __func__));

	keg->uk_ppera = howmany(keg->uk_size, PAGE_SIZE);
	keg->uk_ipers = 1;
	keg->uk_rsize = keg->uk_size;

	/* Check whether we have enough space to not do OFFPAGE. */
	if ((keg->uk_flags & UMA_ZONE_OFFPAGE) == 0) {
		shsize = sizeof(struct uma_slab);
		if (shsize & UMA_ALIGN_PTR)
			shsize = (shsize & ~UMA_ALIGN_PTR) +
			    (UMA_ALIGN_PTR + 1);

		if (PAGE_SIZE * keg->uk_ppera - keg->uk_rsize < shsize) {
			/*
			 * We can't do OFFPAGE if we're internal, in which case
			 * we need an extra page per allocation to contain the
			 * slab header.
			 */
			if ((keg->uk_flags & UMA_ZFLAG_INTERNAL) == 0)
				keg->uk_flags |= UMA_ZONE_OFFPAGE;
			else
				keg->uk_ppera++;
		}
	}

	if ((keg->uk_flags & UMA_ZONE_OFFPAGE) &&
	    (keg->uk_flags & UMA_ZONE_VTOSLAB) == 0)
		keg->uk_flags |= UMA_ZONE_HASH;
}

static void
keg_cachespread_init(uma_keg_t keg)
{
	int alignsize;
	int trailer;
	int pages;
	int rsize;

	KASSERT((keg->uk_flags & UMA_ZONE_PCPU) == 0,
	    ("%s: Cannot cachespread-init a UMA_ZONE_PCPU keg", __func__));

	alignsize = keg->uk_align + 1;
	rsize = keg->uk_size;
	/*
	 * We want one item to start on every align boundary in a page.  To
	 * do this we will span pages.  We will also extend the item by the
	 * size of align if it is an even multiple of align.  Otherwise, it
	 * would fall on the same boundary every time.
	 */
	if (rsize & keg->uk_align)
		rsize = (rsize & ~keg->uk_align) + alignsize;
	if ((rsize & alignsize) == 0)
		rsize += alignsize;
	trailer = rsize - keg->uk_size;
	pages = (rsize * (PAGE_SIZE / alignsize)) / PAGE_SIZE;
	pages = MIN(pages, (128 * 1024) / PAGE_SIZE);
	keg->uk_rsize = rsize;
	keg->uk_ppera = pages;
	keg->uk_ipers = ((pages * PAGE_SIZE) + trailer) / rsize;
	keg->uk_flags |= UMA_ZONE_OFFPAGE | UMA_ZONE_VTOSLAB;
	KASSERT(keg->uk_ipers <= SLAB_SETSIZE,
	    ("%s: keg->uk_ipers too high(%d) increase max_ipers", __func__,
	    keg->uk_ipers));
}

/*
 * Keg header ctor.  This initializes all fields, locks, etc.  And inserts
 * the keg onto the global keg list.
 *
 * Arguments/Returns follow uma_ctor specifications
 *	udata  Actually uma_kctor_args
 */
static int
keg_ctor(void *mem, int size, void *udata, int flags)
{
	struct uma_kctor_args *arg = udata;
	uma_keg_t keg = mem;
	uma_zone_t zone;

	bzero(keg, size);
	keg->uk_size = arg->size;
	keg->uk_init = arg->uminit;
	keg->uk_fini = arg->fini;
	keg->uk_align = arg->align;
	keg->uk_cursor = 0;
	keg->uk_free = 0;
	keg->uk_reserve = 0;
	keg->uk_pages = 0;
	keg->uk_flags = arg->flags;
	keg->uk_slabzone = NULL;

	/*
	 * The master zone is passed to us at keg-creation time.
	 */
	zone = arg->zone;
	keg->uk_name = zone->uz_name;

	if (arg->flags & UMA_ZONE_VM)
		keg->uk_flags |= UMA_ZFLAG_CACHEONLY;

	if (arg->flags & UMA_ZONE_ZINIT)
		keg->uk_init = zero_init;

	if (arg->flags & UMA_ZONE_MALLOC)
		keg->uk_flags |= UMA_ZONE_VTOSLAB;

	if (arg->flags & UMA_ZONE_PCPU)
#ifdef SMP
		keg->uk_flags |= UMA_ZONE_OFFPAGE;
#else
		keg->uk_flags &= ~UMA_ZONE_PCPU;
#endif

	if (keg->uk_flags & UMA_ZONE_CACHESPREAD) {
		keg_cachespread_init(keg);
	} else {
		if (keg->uk_size > UMA_SLAB_SPACE)
			keg_large_init(keg);
		else
			keg_small_init(keg);
	}

	if (keg->uk_flags & UMA_ZONE_OFFPAGE)
		keg->uk_slabzone = slabzone;

	/*
	 * If we haven't booted yet we need allocations to go through the
	 * startup cache until the vm is ready.
	 */
	if (booted < BOOT_PAGEALLOC)
		keg->uk_allocf = startup_alloc;
#ifdef UMA_MD_SMALL_ALLOC
	else if (keg->uk_ppera == 1)
		keg->uk_allocf = uma_small_alloc;
#endif
	else
		keg->uk_allocf = page_alloc;
#ifdef UMA_MD_SMALL_ALLOC
	if (keg->uk_ppera == 1)
		keg->uk_freef = uma_small_free;
	else
#endif
		keg->uk_freef = page_free;

	/*
	 * Initialize keg's lock
	 */
	KEG_LOCK_INIT(keg, (arg->flags & UMA_ZONE_MTXCLASS));

	/*
	 * If we're putting the slab header in the actual page we need to
	 * figure out where in each page it goes.  This calculates a right
	 * justified offset into the memory on an ALIGN_PTR boundary.
	 */
	if (!(keg->uk_flags & UMA_ZONE_OFFPAGE)) {
		u_int totsize;

		/* Size of the slab struct and free list */
		totsize = sizeof(struct uma_slab);

		if (totsize & UMA_ALIGN_PTR)
			totsize = (totsize & ~UMA_ALIGN_PTR) +
			    (UMA_ALIGN_PTR + 1);
		keg->uk_pgoff = (PAGE_SIZE * keg->uk_ppera) - totsize;

		/*
		 * The only way the following is possible is if with our
		 * UMA_ALIGN_PTR adjustments we are now bigger than
		 * UMA_SLAB_SIZE.  I haven't checked whether this is
		 * mathematically possible for all cases, so we make
		 * sure here anyway.
		 */
		totsize = keg->uk_pgoff + sizeof(struct uma_slab);
		if (totsize > PAGE_SIZE * keg->uk_ppera) {
			printf("zone %s ipers %d rsize %d size %d\n",
			    zone->uz_name, keg->uk_ipers, keg->uk_rsize,
			    keg->uk_size);
			panic("UMA slab won't fit.");
		}
	}

	if (keg->uk_flags & UMA_ZONE_HASH)
		hash_alloc(&keg->uk_hash);

	CTR5(KTR_UMA, "keg_ctor %p zone %s(%p) out %d free %d\n",
	    keg, zone->uz_name, zone,
	    (keg->uk_pages / keg->uk_ppera) * keg->uk_ipers - keg->uk_free,
	    keg->uk_free);

	LIST_INSERT_HEAD(&keg->uk_zones, zone, uz_link);

	rw_wlock(&uma_rwlock);
	LIST_INSERT_HEAD(&uma_kegs, keg, uk_link);
	rw_wunlock(&uma_rwlock);
	return (0);
}

/*
 * Zone header ctor.  This initializes all fields, locks, etc.
 *
 * Arguments/Returns follow uma_ctor specifications
 *	udata  Actually uma_zctor_args
 */
static int
zone_ctor(void *mem, int size, void *udata, int flags)
{
	struct uma_zctor_args *arg = udata;
	uma_zone_t zone = mem;
	uma_zone_t z;
	uma_keg_t keg;

	bzero(zone, size);
	zone->uz_name = arg->name;
	zone->uz_ctor = arg->ctor;
	zone->uz_dtor = arg->dtor;
	zone->uz_slab = zone_fetch_slab;
	zone->uz_init = NULL;
	zone->uz_fini = NULL;
	zone->uz_allocs = 0;
	zone->uz_frees = 0;
	zone->uz_fails = 0;
	zone->uz_sleeps = 0;
	zone->uz_count = 0;
	zone->uz_count_min = 0;
	zone->uz_flags = 0;
	zone->uz_warning = NULL;
	/* The domain structures follow the cpu structures. */
	zone->uz_domain = (struct uma_zone_domain *)&zone->uz_cpu[mp_ncpus];
	timevalclear(&zone->uz_ratecheck);
	keg = arg->keg;

	ZONE_LOCK_INIT(zone, (arg->flags & UMA_ZONE_MTXCLASS));

	/*
	 * This is a pure cache zone, no kegs.
	 */
	if (arg->import) {
		if (arg->flags & UMA_ZONE_VM)
			arg->flags |= UMA_ZFLAG_CACHEONLY;
		zone->uz_flags = arg->flags;
		zone->uz_size = arg->size;
		zone->uz_import = arg->import;
		zone->uz_release = arg->release;
		zone->uz_arg = arg->arg;
		zone->uz_lockptr = &zone->uz_lock;
		rw_wlock(&uma_rwlock);
		LIST_INSERT_HEAD(&uma_cachezones, zone, uz_link);
		rw_wunlock(&uma_rwlock);
		goto out;
	}

	/*
	 * Use the regular zone/keg/slab allocator.
	 */
	zone->uz_import = (uma_import)zone_import;
	zone->uz_release = (uma_release)zone_release;
	zone->uz_arg = zone; 

	if (arg->flags & UMA_ZONE_SECONDARY) {
		KASSERT(arg->keg != NULL, ("Secondary zone on zero'd keg"));
		zone->uz_init = arg->uminit;
		zone->uz_fini = arg->fini;
		zone->uz_lockptr = &keg->uk_lock;
		zone->uz_flags |= UMA_ZONE_SECONDARY;
		rw_wlock(&uma_rwlock);
		ZONE_LOCK(zone);
		LIST_FOREACH(z, &keg->uk_zones, uz_link) {
			if (LIST_NEXT(z, uz_link) == NULL) {
				LIST_INSERT_AFTER(z, zone, uz_link);
				break;
			}
		}
		ZONE_UNLOCK(zone);
		rw_wunlock(&uma_rwlock);
	} else if (keg == NULL) {
		if ((keg = uma_kcreate(zone, arg->size, arg->uminit, arg->fini,
		    arg->align, arg->flags)) == NULL)
			return (ENOMEM);
	} else {
		struct uma_kctor_args karg;
		int error;

		/* We should only be here from uma_startup() */
		karg.size = arg->size;
		karg.uminit = arg->uminit;
		karg.fini = arg->fini;
		karg.align = arg->align;
		karg.flags = arg->flags;
		karg.zone = zone;
		error = keg_ctor(arg->keg, sizeof(struct uma_keg), &karg,
		    flags);
		if (error)
			return (error);
	}

	/*
	 * Link in the first keg.
	 */
	zone->uz_klink.kl_keg = keg;
	LIST_INSERT_HEAD(&zone->uz_kegs, &zone->uz_klink, kl_link);
	zone->uz_lockptr = &keg->uk_lock;
	zone->uz_size = keg->uk_size;
	zone->uz_flags |= (keg->uk_flags &
	    (UMA_ZONE_INHERIT | UMA_ZFLAG_INHERIT));

	/*
	 * Some internal zones don't have room allocated for the per cpu
	 * caches.  If we're internal, bail out here.
	 */
	if (keg->uk_flags & UMA_ZFLAG_INTERNAL) {
		KASSERT((zone->uz_flags & UMA_ZONE_SECONDARY) == 0,
		    ("Secondary zone requested UMA_ZFLAG_INTERNAL"));
		return (0);
	}

out:
	KASSERT((arg->flags & (UMA_ZONE_MAXBUCKET | UMA_ZONE_NOBUCKET)) !=
	    (UMA_ZONE_MAXBUCKET | UMA_ZONE_NOBUCKET),
	    ("Invalid zone flag combination"));
	if ((arg->flags & UMA_ZONE_MAXBUCKET) != 0)
		zone->uz_count = BUCKET_MAX;
	else if ((arg->flags & UMA_ZONE_NOBUCKET) != 0)
		zone->uz_count = 0;
	else
		zone->uz_count = bucket_select(zone->uz_size);
	zone->uz_count_min = zone->uz_count;

	return (0);
}

/*
 * Keg header dtor.  This frees all data, destroys locks, frees the hash
 * table and removes the keg from the global list.
 *
 * Arguments/Returns follow uma_dtor specifications
 *	udata  unused
 */
static void
keg_dtor(void *arg, int size, void *udata)
{
	uma_keg_t keg;

	keg = (uma_keg_t)arg;
	KEG_LOCK(keg);
	if (keg->uk_free != 0) {
		printf("Freed UMA keg (%s) was not empty (%d items). "
		    " Lost %d pages of memory.\n",
		    keg->uk_name ? keg->uk_name : "",
		    keg->uk_free, keg->uk_pages);
	}
	KEG_UNLOCK(keg);

	hash_free(&keg->uk_hash);

	KEG_LOCK_FINI(keg);
}

/*
 * Zone header dtor.
 *
 * Arguments/Returns follow uma_dtor specifications
 *	udata  unused
 */
static void
zone_dtor(void *arg, int size, void *udata)
{
	uma_klink_t klink;
	uma_zone_t zone;
	uma_keg_t keg;

	zone = (uma_zone_t)arg;
	keg = zone_first_keg(zone);

	if (!(zone->uz_flags & UMA_ZFLAG_INTERNAL))
		cache_drain(zone);

	rw_wlock(&uma_rwlock);
	LIST_REMOVE(zone, uz_link);
	rw_wunlock(&uma_rwlock);
	/*
	 * XXX there are some races here where
	 * the zone can be drained but zone lock
	 * released and then refilled before we
	 * remove it... we dont care for now
	 */
	zone_drain_wait(zone, M_WAITOK);
	/*
	 * Unlink all of our kegs.
	 */
	while ((klink = LIST_FIRST(&zone->uz_kegs)) != NULL) {
		klink->kl_keg = NULL;
		LIST_REMOVE(klink, kl_link);
		if (klink == &zone->uz_klink)
			continue;
		free(klink, M_TEMP);
	}
	/*
	 * We only destroy kegs from non secondary zones.
	 */
	if (keg != NULL && (zone->uz_flags & UMA_ZONE_SECONDARY) == 0)  {
		rw_wlock(&uma_rwlock);
		LIST_REMOVE(keg, uk_link);
		rw_wunlock(&uma_rwlock);
		zone_free_item(kegs, keg, NULL, SKIP_NONE);
	}
	ZONE_LOCK_FINI(zone);
}

/*
 * Traverses every zone in the system and calls a callback
 *
 * Arguments:
 *	zfunc  A pointer to a function which accepts a zone
 *		as an argument.
 *
 * Returns:
 *	Nothing
 */
static void
zone_foreach(void (*zfunc)(uma_zone_t))
{
	uma_keg_t keg;
	uma_zone_t zone;

	rw_rlock(&uma_rwlock);
	LIST_FOREACH(keg, &uma_kegs, uk_link) {
		LIST_FOREACH(zone, &keg->uk_zones, uz_link)
			zfunc(zone);
	}
	rw_runlock(&uma_rwlock);
}

/*
 * Count how many pages do we need to bootstrap.  VM supplies
 * its need in early zones in the argument, we add up our zones,
 * which consist of: UMA Slabs, UMA Hash and 9 Bucket zones. The
 * zone of zones and zone of kegs are accounted separately.
 */
#define	UMA_BOOT_ZONES	11
/* Zone of zones and zone of kegs have arbitrary alignment. */
#define	UMA_BOOT_ALIGN	32
static int zsize, ksize;
int
uma_startup_count(int vm_zones)
{
	int zones, pages;

	ksize = sizeof(struct uma_keg) +
	    (sizeof(struct uma_domain) * vm_ndomains);
	zsize = sizeof(struct uma_zone) +
	    (sizeof(struct uma_cache) * (mp_maxid + 1)) +
	    (sizeof(struct uma_zone_domain) * vm_ndomains);

	/*
	 * Memory for the zone of kegs and its keg,
	 * and for zone of zones.
	 */
	pages = howmany(roundup(zsize, CACHE_LINE_SIZE) * 2 +
	    roundup(ksize, CACHE_LINE_SIZE), PAGE_SIZE);

#ifdef	UMA_MD_SMALL_ALLOC
	zones = UMA_BOOT_ZONES;
#else
	zones = UMA_BOOT_ZONES + vm_zones;
	vm_zones = 0;
#endif

	/* Memory for the rest of startup zones, UMA and VM, ... */
	if (zsize > UMA_SLAB_SPACE)
		pages += (zones + vm_zones) *
		    howmany(roundup2(zsize, UMA_BOOT_ALIGN), UMA_SLAB_SIZE);
	else if (roundup2(zsize, UMA_BOOT_ALIGN) > UMA_SLAB_SPACE)
		pages += zones;
	else
		pages += howmany(zones,
		    UMA_SLAB_SPACE / roundup2(zsize, UMA_BOOT_ALIGN));

	/* ... and their kegs. Note that zone of zones allocates a keg! */
	pages += howmany(zones + 1,
	    UMA_SLAB_SPACE / roundup2(ksize, UMA_BOOT_ALIGN));

	/*
	 * Most of startup zones are not going to be offpages, that's
	 * why we use UMA_SLAB_SPACE instead of UMA_SLAB_SIZE in all
	 * calculations.  Some large bucket zones will be offpage, and
	 * thus will allocate hashes.  We take conservative approach
	 * and assume that all zones may allocate hash.  This may give
	 * us some positive inaccuracy, usually an extra single page.
	 */
	pages += howmany(zones, UMA_SLAB_SPACE /
	    (sizeof(struct slabhead *) * UMA_HASH_SIZE_INIT));

	return (pages);
}

void
uma_startup(void *mem, int npages)
{
	struct uma_zctor_args args;
	uma_keg_t masterkeg;
	uintptr_t m;

#ifdef DIAGNOSTIC
	printf("Entering %s with %d boot pages configured\n", __func__, npages);
#endif

	rw_init(&uma_rwlock, "UMA lock");

	/* Use bootpages memory for the zone of zones and zone of kegs. */
	m = (uintptr_t)mem;
	zones = (uma_zone_t)m;
	m += roundup(zsize, CACHE_LINE_SIZE);
	kegs = (uma_zone_t)m;
	m += roundup(zsize, CACHE_LINE_SIZE);
	masterkeg = (uma_keg_t)m;
	m += roundup(ksize, CACHE_LINE_SIZE);
	m = roundup(m, PAGE_SIZE);
	npages -= (m - (uintptr_t)mem) / PAGE_SIZE;
	mem = (void *)m;

	/* "manually" create the initial zone */
	memset(&args, 0, sizeof(args));
	args.name = "UMA Kegs";
	args.size = ksize;
	args.ctor = keg_ctor;
	args.dtor = keg_dtor;
	args.uminit = zero_init;
	args.fini = NULL;
	args.keg = masterkeg;
	args.align = UMA_BOOT_ALIGN - 1;
	args.flags = UMA_ZFLAG_INTERNAL;
	zone_ctor(kegs, zsize, &args, M_WAITOK);

	bootmem = mem;
	boot_pages = npages;

	args.name = "UMA Zones";
	args.size = zsize;
	args.ctor = zone_ctor;
	args.dtor = zone_dtor;
	args.uminit = zero_init;
	args.fini = NULL;
	args.keg = NULL;
	args.align = UMA_BOOT_ALIGN - 1;
	args.flags = UMA_ZFLAG_INTERNAL;
	zone_ctor(zones, zsize, &args, M_WAITOK);

	/* Now make a zone for slab headers */
	slabzone = uma_zcreate("UMA Slabs",
				sizeof(struct uma_slab),
				NULL, NULL, NULL, NULL,
				UMA_ALIGN_PTR, UMA_ZFLAG_INTERNAL);

	hashzone = uma_zcreate("UMA Hash",
	    sizeof(struct slabhead *) * UMA_HASH_SIZE_INIT,
	    NULL, NULL, NULL, NULL,
	    UMA_ALIGN_PTR, UMA_ZFLAG_INTERNAL);

	bucket_init();

	booted = BOOT_STRAPPED;
}

void
uma_startup1(void)
{

#ifdef DIAGNOSTIC
	printf("Entering %s with %d boot pages left\n", __func__, boot_pages);
#endif
	booted = BOOT_PAGEALLOC;
}

void
uma_startup2(void)
{

#ifdef DIAGNOSTIC
	printf("Entering %s with %d boot pages left\n", __func__, boot_pages);
#endif
	booted = BOOT_BUCKETS;
	sx_init(&uma_drain_lock, "umadrain");
	bucket_enable();
}

/*
 * Initialize our callout handle
 *
 */
static void
uma_startup3(void)
{

#ifdef INVARIANTS
	TUNABLE_INT_FETCH("vm.debug.divisor", &dbg_divisor);
	uma_dbg_cnt = counter_u64_alloc(M_WAITOK);
	uma_skip_cnt = counter_u64_alloc(M_WAITOK);
#endif
	callout_init(&uma_callout, 1);
	callout_reset(&uma_callout, UMA_TIMEOUT * hz, uma_timeout, NULL);
	booted = BOOT_RUNNING;
}

static uma_keg_t
uma_kcreate(uma_zone_t zone, size_t size, uma_init uminit, uma_fini fini,
		int align, uint32_t flags)
{
	struct uma_kctor_args args;

	args.size = size;
	args.uminit = uminit;
	args.fini = fini;
	args.align = (align == UMA_ALIGN_CACHE) ? uma_align_cache : align;
	args.flags = flags;
	args.zone = zone;
	return (zone_alloc_item(kegs, &args, UMA_ANYDOMAIN, M_WAITOK));
}

/* Public functions */
/* See uma.h */
void
uma_set_align(int align)
{

	if (align != UMA_ALIGN_CACHE)
		uma_align_cache = align;
}

/* See uma.h */
uma_zone_t
uma_zcreate(const char *name, size_t size, uma_ctor ctor, uma_dtor dtor,
		uma_init uminit, uma_fini fini, int align, uint32_t flags)

{
	struct uma_zctor_args args;
	uma_zone_t res;
	bool locked;

	KASSERT(powerof2(align + 1), ("invalid zone alignment %d for \"%s\"",
	    align, name));

	/* This stuff is essential for the zone ctor */
	memset(&args, 0, sizeof(args));
	args.name = name;
	args.size = size;
	args.ctor = ctor;
	args.dtor = dtor;
	args.uminit = uminit;
	args.fini = fini;
#ifdef  INVARIANTS
	/*
	 * If a zone is being created with an empty constructor and
	 * destructor, pass UMA constructor/destructor which checks for
	 * memory use after free.
	 */
	if ((!(flags & (UMA_ZONE_ZINIT | UMA_ZONE_NOFREE))) &&
	    ctor == NULL && dtor == NULL && uminit == NULL && fini == NULL) {
		args.ctor = trash_ctor;
		args.dtor = trash_dtor;
		args.uminit = trash_init;
		args.fini = trash_fini;
	}
#endif
	args.align = align;
	args.flags = flags;
	args.keg = NULL;

	if (booted < BOOT_BUCKETS) {
		locked = false;
	} else {
		sx_slock(&uma_drain_lock);
		locked = true;
	}
	res = zone_alloc_item(zones, &args, UMA_ANYDOMAIN, M_WAITOK);
	if (locked)
		sx_sunlock(&uma_drain_lock);
	return (res);
}

/* See uma.h */
uma_zone_t
uma_zsecond_create(char *name, uma_ctor ctor, uma_dtor dtor,
		    uma_init zinit, uma_fini zfini, uma_zone_t master)
{
	struct uma_zctor_args args;
	uma_keg_t keg;
	uma_zone_t res;
	bool locked;

	keg = zone_first_keg(master);
	memset(&args, 0, sizeof(args));
	args.name = name;
	args.size = keg->uk_size;
	args.ctor = ctor;
	args.dtor = dtor;
	args.uminit = zinit;
	args.fini = zfini;
	args.align = keg->uk_align;
	args.flags = keg->uk_flags | UMA_ZONE_SECONDARY;
	args.keg = keg;

	if (booted < BOOT_BUCKETS) {
		locked = false;
	} else {
		sx_slock(&uma_drain_lock);
		locked = true;
	}
	/* XXX Attaches only one keg of potentially many. */
	res = zone_alloc_item(zones, &args, UMA_ANYDOMAIN, M_WAITOK);
	if (locked)
		sx_sunlock(&uma_drain_lock);
	return (res);
}

/* See uma.h */
uma_zone_t
uma_zcache_create(char *name, int size, uma_ctor ctor, uma_dtor dtor,
		    uma_init zinit, uma_fini zfini, uma_import zimport,
		    uma_release zrelease, void *arg, int flags)
{
	struct uma_zctor_args args;

	memset(&args, 0, sizeof(args));
	args.name = name;
	args.size = size;
	args.ctor = ctor;
	args.dtor = dtor;
	args.uminit = zinit;
	args.fini = zfini;
	args.import = zimport;
	args.release = zrelease;
	args.arg = arg;
	args.align = 0;
	args.flags = flags;

	return (zone_alloc_item(zones, &args, UMA_ANYDOMAIN, M_WAITOK));
}

static void
zone_lock_pair(uma_zone_t a, uma_zone_t b)
{
	if (a < b) {
		ZONE_LOCK(a);
		mtx_lock_flags(b->uz_lockptr, MTX_DUPOK);
	} else {
		ZONE_LOCK(b);
		mtx_lock_flags(a->uz_lockptr, MTX_DUPOK);
	}
}

static void
zone_unlock_pair(uma_zone_t a, uma_zone_t b)
{

	ZONE_UNLOCK(a);
	ZONE_UNLOCK(b);
}

int
uma_zsecond_add(uma_zone_t zone, uma_zone_t master)
{
	uma_klink_t klink;
	uma_klink_t kl;
	int error;

	error = 0;
	klink = malloc(sizeof(*klink), M_TEMP, M_WAITOK | M_ZERO);

	zone_lock_pair(zone, master);
	/*
	 * zone must use vtoslab() to resolve objects and must already be
	 * a secondary.
	 */
	if ((zone->uz_flags & (UMA_ZONE_VTOSLAB | UMA_ZONE_SECONDARY))
	    != (UMA_ZONE_VTOSLAB | UMA_ZONE_SECONDARY)) {
		error = EINVAL;
		goto out;
	}
	/*
	 * The new master must also use vtoslab().
	 */
	if ((zone->uz_flags & UMA_ZONE_VTOSLAB) != UMA_ZONE_VTOSLAB) {
		error = EINVAL;
		goto out;
	}

	/*
	 * The underlying object must be the same size.  rsize
	 * may be different.
	 */
	if (master->uz_size != zone->uz_size) {
		error = E2BIG;
		goto out;
	}
	/*
	 * Put it at the end of the list.
	 */
	klink->kl_keg = zone_first_keg(master);
	LIST_FOREACH(kl, &zone->uz_kegs, kl_link) {
		if (LIST_NEXT(kl, kl_link) == NULL) {
			LIST_INSERT_AFTER(kl, klink, kl_link);
			break;
		}
	}
	klink = NULL;
	zone->uz_flags |= UMA_ZFLAG_MULTI;
	zone->uz_slab = zone_fetch_slab_multi;

out:
	zone_unlock_pair(zone, master);
	if (klink != NULL)
		free(klink, M_TEMP);

	return (error);
}


/* See uma.h */
void
uma_zdestroy(uma_zone_t zone)
{

	sx_slock(&uma_drain_lock);
	zone_free_item(zones, zone, NULL, SKIP_NONE);
	sx_sunlock(&uma_drain_lock);
}

void
uma_zwait(uma_zone_t zone)
{
	void *item;

	item = uma_zalloc_arg(zone, NULL, M_WAITOK);
	uma_zfree(zone, item);
}

void *
uma_zalloc_pcpu_arg(uma_zone_t zone, void *udata, int flags)
{
	void *item;
	int i;

	MPASS(zone->uz_flags & UMA_ZONE_PCPU);
	item = uma_zalloc_arg(zone, udata, flags &~ M_ZERO);
	if (item != NULL && (flags & M_ZERO)) {
		CPU_FOREACH(i)
			bzero(zpcpu_get_cpu(item, i), zone->uz_size);
	}
	return (item);
}

/*
 * A stub while both regular and pcpu cases are identical.
 */
void
uma_zfree_pcpu_arg(uma_zone_t zone, void *item, void *udata)
{

	MPASS(zone->uz_flags & UMA_ZONE_PCPU);
	uma_zfree_arg(zone, item, udata);
}

/* See uma.h */
void *
uma_zalloc_arg(uma_zone_t zone, void *udata, int flags)
{
	uma_zone_domain_t zdom;
	uma_bucket_t bucket;
	uma_cache_t cache;
	void *item;
	int cpu, domain, lockfail;
#ifdef INVARIANTS
	bool skipdbg;
#endif

	/* Enable entropy collection for RANDOM_ENABLE_UMA kernel option */
	random_harvest_fast_uma(&zone, sizeof(zone), 1, RANDOM_UMA);

	/* This is the fast path allocation */
	CTR4(KTR_UMA, "uma_zalloc_arg thread %x zone %s(%p) flags %d",
	    curthread, zone->uz_name, zone, flags);

	if (flags & M_WAITOK) {
		WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
		    "uma_zalloc_arg: zone \"%s\"", zone->uz_name);
	}
	KASSERT(curthread->td_critnest == 0 || SCHEDULER_STOPPED(),
	    ("uma_zalloc_arg: called with spinlock or critical section held"));
	if (zone->uz_flags & UMA_ZONE_PCPU)
		KASSERT((flags & M_ZERO) == 0, ("allocating from a pcpu zone "
		    "with M_ZERO passed"));

#ifdef DEBUG_MEMGUARD
	if (memguard_cmp_zone(zone)) {
		item = memguard_alloc(zone->uz_size, flags);
		if (item != NULL) {
			if (zone->uz_init != NULL &&
			    zone->uz_init(item, zone->uz_size, flags) != 0)
				return (NULL);
			if (zone->uz_ctor != NULL &&
			    zone->uz_ctor(item, zone->uz_size, udata,
			    flags) != 0) {
			    	zone->uz_fini(item, zone->uz_size);
				return (NULL);
			}
			return (item);
		}
		/* This is unfortunate but should not be fatal. */
	}
#endif
	/*
	 * If possible, allocate from the per-CPU cache.  There are two
	 * requirements for safe access to the per-CPU cache: (1) the thread
	 * accessing the cache must not be preempted or yield during access,
	 * and (2) the thread must not migrate CPUs without switching which
	 * cache it accesses.  We rely on a critical section to prevent
	 * preemption and migration.  We release the critical section in
	 * order to acquire the zone mutex if we are unable to allocate from
	 * the current cache; when we re-acquire the critical section, we
	 * must detect and handle migration if it has occurred.
	 */
	critical_enter();
	cpu = curcpu;
	cache = &zone->uz_cpu[cpu];

zalloc_start:
	bucket = cache->uc_allocbucket;
	if (bucket != NULL && bucket->ub_cnt > 0) {
		bucket->ub_cnt--;
		item = bucket->ub_bucket[bucket->ub_cnt];
#ifdef INVARIANTS
		bucket->ub_bucket[bucket->ub_cnt] = NULL;
#endif
		KASSERT(item != NULL, ("uma_zalloc: Bucket pointer mangled."));
		cache->uc_allocs++;
		critical_exit();
#ifdef INVARIANTS
		skipdbg = uma_dbg_zskip(zone, item);
#endif
		if (zone->uz_ctor != NULL &&
#ifdef INVARIANTS
		    (!skipdbg || zone->uz_ctor != trash_ctor ||
		    zone->uz_dtor != trash_dtor) &&
#endif
		    zone->uz_ctor(item, zone->uz_size, udata, flags) != 0) {
			atomic_add_long(&zone->uz_fails, 1);
			zone_free_item(zone, item, udata, SKIP_DTOR);
			return (NULL);
		}
#ifdef INVARIANTS
		if (!skipdbg)
			uma_dbg_alloc(zone, NULL, item);
#endif
		if (flags & M_ZERO)
			uma_zero_item(item, zone);
		return (item);
	}

	/*
	 * We have run out of items in our alloc bucket.
	 * See if we can switch with our free bucket.
	 */
	bucket = cache->uc_freebucket;
	if (bucket != NULL && bucket->ub_cnt > 0) {
		CTR2(KTR_UMA,
		    "uma_zalloc: zone %s(%p) swapping empty with alloc",
		    zone->uz_name, zone);
		cache->uc_freebucket = cache->uc_allocbucket;
		cache->uc_allocbucket = bucket;
		goto zalloc_start;
	}

	/*
	 * Discard any empty allocation bucket while we hold no locks.
	 */
	bucket = cache->uc_allocbucket;
	cache->uc_allocbucket = NULL;
	critical_exit();
	if (bucket != NULL)
		bucket_free(zone, bucket, udata);

	if (zone->uz_flags & UMA_ZONE_NUMA)
		domain = PCPU_GET(domain);
	else
		domain = UMA_ANYDOMAIN;

	/* Short-circuit for zones without buckets and low memory. */
	if (zone->uz_count == 0 || bucketdisable)
		goto zalloc_item;

	/*
	 * Attempt to retrieve the item from the per-CPU cache has failed, so
	 * we must go back to the zone.  This requires the zone lock, so we
	 * must drop the critical section, then re-acquire it when we go back
	 * to the cache.  Since the critical section is released, we may be
	 * preempted or migrate.  As such, make sure not to maintain any
	 * thread-local state specific to the cache from prior to releasing
	 * the critical section.
	 */
	lockfail = 0;
	if (ZONE_TRYLOCK(zone) == 0) {
		/* Record contention to size the buckets. */
		ZONE_LOCK(zone);
		lockfail = 1;
	}
	critical_enter();
	cpu = curcpu;
	cache = &zone->uz_cpu[cpu];

	/* See if we lost the race to fill the cache. */
	if (cache->uc_allocbucket != NULL) {
		ZONE_UNLOCK(zone);
		goto zalloc_start;
	}

	/*
	 * Check the zone's cache of buckets.
	 */
	if (domain == UMA_ANYDOMAIN)
		zdom = &zone->uz_domain[0];
	else
		zdom = &zone->uz_domain[domain];
	if ((bucket = LIST_FIRST(&zdom->uzd_buckets)) != NULL) {
		KASSERT(bucket->ub_cnt != 0,
		    ("uma_zalloc_arg: Returning an empty bucket."));

		LIST_REMOVE(bucket, ub_link);
		cache->uc_allocbucket = bucket;
		ZONE_UNLOCK(zone);
		goto zalloc_start;
	}
	/* We are no longer associated with this CPU. */
	critical_exit();

	/*
	 * We bump the uz count when the cache size is insufficient to
	 * handle the working set.
	 */
	if (lockfail && zone->uz_count < BUCKET_MAX)
		zone->uz_count++;
	ZONE_UNLOCK(zone);

	/*
	 * Now lets just fill a bucket and put it on the free list.  If that
	 * works we'll restart the allocation from the beginning and it
	 * will use the just filled bucket.
	 */
	bucket = zone_alloc_bucket(zone, udata, domain, flags);
	CTR3(KTR_UMA, "uma_zalloc: zone %s(%p) bucket zone returned %p",
	    zone->uz_name, zone, bucket);
	if (bucket != NULL) {
		ZONE_LOCK(zone);
		critical_enter();
		cpu = curcpu;
		cache = &zone->uz_cpu[cpu];
		/*
		 * See if we lost the race or were migrated.  Cache the
		 * initialized bucket to make this less likely or claim
		 * the memory directly.
		 */
		if (cache->uc_allocbucket != NULL ||
		    (zone->uz_flags & UMA_ZONE_NUMA &&
		    domain != PCPU_GET(domain)))
			LIST_INSERT_HEAD(&zdom->uzd_buckets, bucket, ub_link);
		else
			cache->uc_allocbucket = bucket;
		ZONE_UNLOCK(zone);
		goto zalloc_start;
	}

	/*
	 * We may not be able to get a bucket so return an actual item.
	 */
zalloc_item:
	item = zone_alloc_item(zone, udata, domain, flags);

	return (item);
}

void *
uma_zalloc_domain(uma_zone_t zone, void *udata, int domain, int flags)
{

	/* Enable entropy collection for RANDOM_ENABLE_UMA kernel option */
	random_harvest_fast_uma(&zone, sizeof(zone), 1, RANDOM_UMA);

	/* This is the fast path allocation */
	CTR5(KTR_UMA,
	    "uma_zalloc_domain thread %x zone %s(%p) domain %d flags %d",
	    curthread, zone->uz_name, zone, domain, flags);

	if (flags & M_WAITOK) {
		WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
		    "uma_zalloc_domain: zone \"%s\"", zone->uz_name);
	}
	KASSERT(curthread->td_critnest == 0 || SCHEDULER_STOPPED(),
	    ("uma_zalloc_domain: called with spinlock or critical section held"));

	return (zone_alloc_item(zone, udata, domain, flags));
}

/*
 * Find a slab with some space.  Prefer slabs that are partially used over those
 * that are totally full.  This helps to reduce fragmentation.
 *
 * If 'rr' is 1, search all domains starting from 'domain'.  Otherwise check
 * only 'domain'.
 */
static uma_slab_t
keg_first_slab(uma_keg_t keg, int domain, int rr)
{
	uma_domain_t dom;
	uma_slab_t slab;
	int start;

	KASSERT(domain >= 0 && domain < vm_ndomains,
	    ("keg_first_slab: domain %d out of range", domain));

	slab = NULL;
	start = domain;
	do {
		dom = &keg->uk_domain[domain];
		if (!LIST_EMPTY(&dom->ud_part_slab))
			return (LIST_FIRST(&dom->ud_part_slab));
		if (!LIST_EMPTY(&dom->ud_free_slab)) {
			slab = LIST_FIRST(&dom->ud_free_slab);
			LIST_REMOVE(slab, us_link);
			LIST_INSERT_HEAD(&dom->ud_part_slab, slab, us_link);
			return (slab);
		}
		if (rr)
			domain = (domain + 1) % vm_ndomains;
	} while (domain != start);

	return (NULL);
}

static uma_slab_t
keg_fetch_slab(uma_keg_t keg, uma_zone_t zone, int rdomain, int flags)
{
	uma_domain_t dom;
	uma_slab_t slab;
	int allocflags, domain, reserve, rr, start;

	mtx_assert(&keg->uk_lock, MA_OWNED);
	slab = NULL;
	reserve = 0;
	allocflags = flags;
	if ((flags & M_USE_RESERVE) == 0)
		reserve = keg->uk_reserve;

	/*
	 * Round-robin for non first-touch zones when there is more than one
	 * domain.
	 */
	if (vm_ndomains == 1)
		rdomain = 0;
	rr = rdomain == UMA_ANYDOMAIN;
	if (rr) {
		keg->uk_cursor = (keg->uk_cursor + 1) % vm_ndomains;
		domain = start = keg->uk_cursor;
		/* Only block on the second pass. */
		if ((flags & (M_WAITOK | M_NOVM)) == M_WAITOK)
			allocflags = (allocflags & ~M_WAITOK) | M_NOWAIT;
	} else
		domain = start = rdomain;

again:
	do {
		if (keg->uk_free > reserve &&
		    (slab = keg_first_slab(keg, domain, rr)) != NULL) {
			MPASS(slab->us_keg == keg);
			return (slab);
		}

		/*
		 * M_NOVM means don't ask at all!
		 */
		if (flags & M_NOVM)
			break;

		if (keg->uk_maxpages && keg->uk_pages >= keg->uk_maxpages) {
			keg->uk_flags |= UMA_ZFLAG_FULL;
			/*
			 * If this is not a multi-zone, set the FULL bit.
			 * Otherwise slab_multi() takes care of it.
			 */
			if ((zone->uz_flags & UMA_ZFLAG_MULTI) == 0) {
				zone->uz_flags |= UMA_ZFLAG_FULL;
				zone_log_warning(zone);
				zone_maxaction(zone);
			}
			if (flags & M_NOWAIT)
				return (NULL);
			zone->uz_sleeps++;
			msleep(keg, &keg->uk_lock, PVM, "keglimit", 0);
			continue;
		}
		slab = keg_alloc_slab(keg, zone, domain, allocflags);
		/*
		 * If we got a slab here it's safe to mark it partially used
		 * and return.  We assume that the caller is going to remove
		 * at least one item.
		 */
		if (slab) {
			MPASS(slab->us_keg == keg);
			dom = &keg->uk_domain[slab->us_domain];
			LIST_INSERT_HEAD(&dom->ud_part_slab, slab, us_link);
			return (slab);
		}
		if (rr) {
			keg->uk_cursor = (keg->uk_cursor + 1) % vm_ndomains;
			domain = keg->uk_cursor;
		}
	} while (domain != start);

	/* Retry domain scan with blocking. */
	if (allocflags != flags) {
		allocflags = flags;
		goto again;
	}

	/*
	 * We might not have been able to get a slab but another cpu
	 * could have while we were unlocked.  Check again before we
	 * fail.
	 */
	if (keg->uk_free > reserve &&
	    (slab = keg_first_slab(keg, domain, rr)) != NULL) {
		MPASS(slab->us_keg == keg);
		return (slab);
	}
	return (NULL);
}

static uma_slab_t
zone_fetch_slab(uma_zone_t zone, uma_keg_t keg, int domain, int flags)
{
	uma_slab_t slab;

	if (keg == NULL) {
		keg = zone_first_keg(zone);
		KEG_LOCK(keg);
	}

	for (;;) {
		slab = keg_fetch_slab(keg, zone, domain, flags);
		if (slab)
			return (slab);
		if (flags & (M_NOWAIT | M_NOVM))
			break;
	}
	KEG_UNLOCK(keg);
	return (NULL);
}

/*
 * uma_zone_fetch_slab_multi:  Fetches a slab from one available keg.  Returns
 * with the keg locked.  On NULL no lock is held.
 *
 * The last pointer is used to seed the search.  It is not required.
 */
static uma_slab_t
zone_fetch_slab_multi(uma_zone_t zone, uma_keg_t last, int domain, int rflags)
{
	uma_klink_t klink;
	uma_slab_t slab;
	uma_keg_t keg;
	int flags;
	int empty;
	int full;

	/*
	 * Don't wait on the first pass.  This will skip limit tests
	 * as well.  We don't want to block if we can find a provider
	 * without blocking.
	 */
	flags = (rflags & ~M_WAITOK) | M_NOWAIT;
	/*
	 * Use the last slab allocated as a hint for where to start
	 * the search.
	 */
	if (last != NULL) {
		slab = keg_fetch_slab(last, zone, domain, flags);
		if (slab)
			return (slab);
		KEG_UNLOCK(last);
	}
	/*
	 * Loop until we have a slab incase of transient failures
	 * while M_WAITOK is specified.  I'm not sure this is 100%
	 * required but we've done it for so long now.
	 */
	for (;;) {
		empty = 0;
		full = 0;
		/*
		 * Search the available kegs for slabs.  Be careful to hold the
		 * correct lock while calling into the keg layer.
		 */
		LIST_FOREACH(klink, &zone->uz_kegs, kl_link) {
			keg = klink->kl_keg;
			KEG_LOCK(keg);
			if ((keg->uk_flags & UMA_ZFLAG_FULL) == 0) {
				slab = keg_fetch_slab(keg, zone, domain, flags);
				if (slab)
					return (slab);
			}
			if (keg->uk_flags & UMA_ZFLAG_FULL)
				full++;
			else
				empty++;
			KEG_UNLOCK(keg);
		}
		if (rflags & (M_NOWAIT | M_NOVM))
			break;
		flags = rflags;
		/*
		 * All kegs are full.  XXX We can't atomically check all kegs
		 * and sleep so just sleep for a short period and retry.
		 */
		if (full && !empty) {
			ZONE_LOCK(zone);
			zone->uz_flags |= UMA_ZFLAG_FULL;
			zone->uz_sleeps++;
			zone_log_warning(zone);
			zone_maxaction(zone);
			msleep(zone, zone->uz_lockptr, PVM,
			    "zonelimit", hz/100);
			zone->uz_flags &= ~UMA_ZFLAG_FULL;
			ZONE_UNLOCK(zone);
			continue;
		}
	}
	return (NULL);
}

static void *
slab_alloc_item(uma_keg_t keg, uma_slab_t slab)
{
	uma_domain_t dom;
	void *item;
	uint8_t freei;

	MPASS(keg == slab->us_keg);
	mtx_assert(&keg->uk_lock, MA_OWNED);

	freei = BIT_FFS(SLAB_SETSIZE, &slab->us_free) - 1;
	BIT_CLR(SLAB_SETSIZE, freei, &slab->us_free);
	item = slab->us_data + (keg->uk_rsize * freei);
	slab->us_freecount--;
	keg->uk_free--;

	/* Move this slab to the full list */
	if (slab->us_freecount == 0) {
		LIST_REMOVE(slab, us_link);
		dom = &keg->uk_domain[slab->us_domain];
		LIST_INSERT_HEAD(&dom->ud_full_slab, slab, us_link);
	}

	return (item);
}

static int
zone_import(uma_zone_t zone, void **bucket, int max, int domain, int flags)
{
	uma_slab_t slab;
	uma_keg_t keg;
	int stripe;
	int i;

	slab = NULL;
	keg = NULL;
	/* Try to keep the buckets totally full */
	for (i = 0; i < max; ) {
		if ((slab = zone->uz_slab(zone, keg, domain, flags)) == NULL)
			break;
		keg = slab->us_keg;
		stripe = howmany(max, vm_ndomains);
		while (slab->us_freecount && i < max) { 
			bucket[i++] = slab_alloc_item(keg, slab);
			if (keg->uk_free <= keg->uk_reserve)
				break;
#ifdef NUMA
			/*
			 * If the zone is striped we pick a new slab for every
			 * N allocations.  Eliminating this conditional will
			 * instead pick a new domain for each bucket rather
			 * than stripe within each bucket.  The current option
			 * produces more fragmentation and requires more cpu
			 * time but yields better distribution.
			 */
			if ((zone->uz_flags & UMA_ZONE_NUMA) == 0 &&
			    vm_ndomains > 1 && --stripe == 0)
				break;
#endif
		}
		/* Don't block if we allocated any successfully. */
		flags &= ~M_WAITOK;
		flags |= M_NOWAIT;
	}
	if (slab != NULL)
		KEG_UNLOCK(keg);

	return i;
}

static uma_bucket_t
zone_alloc_bucket(uma_zone_t zone, void *udata, int domain, int flags)
{
	uma_bucket_t bucket;
	int max;

	/* Don't wait for buckets, preserve caller's NOVM setting. */
	bucket = bucket_alloc(zone, udata, M_NOWAIT | (flags & M_NOVM));
	if (bucket == NULL)
		return (NULL);

	max = MIN(bucket->ub_entries, zone->uz_count);
	bucket->ub_cnt = zone->uz_import(zone->uz_arg, bucket->ub_bucket,
	    max, domain, flags);

	/*
	 * Initialize the memory if necessary.
	 */
	if (bucket->ub_cnt != 0 && zone->uz_init != NULL) {
		int i;

		for (i = 0; i < bucket->ub_cnt; i++)
			if (zone->uz_init(bucket->ub_bucket[i], zone->uz_size,
			    flags) != 0)
				break;
		/*
		 * If we couldn't initialize the whole bucket, put the
		 * rest back onto the freelist.
		 */
		if (i != bucket->ub_cnt) {
			zone->uz_release(zone->uz_arg, &bucket->ub_bucket[i],
			    bucket->ub_cnt - i);
#ifdef INVARIANTS
			bzero(&bucket->ub_bucket[i],
			    sizeof(void *) * (bucket->ub_cnt - i));
#endif
			bucket->ub_cnt = i;
		}
	}

	if (bucket->ub_cnt == 0) {
		bucket_free(zone, bucket, udata);
		atomic_add_long(&zone->uz_fails, 1);
		return (NULL);
	}

	return (bucket);
}

/*
 * Allocates a single item from a zone.
 *
 * Arguments
 *	zone   The zone to alloc for.
 *	udata  The data to be passed to the constructor.
 *	domain The domain to allocate from or UMA_ANYDOMAIN.
 *	flags  M_WAITOK, M_NOWAIT, M_ZERO.
 *
 * Returns
 *	NULL if there is no memory and M_NOWAIT is set
 *	An item if successful
 */

static void *
zone_alloc_item(uma_zone_t zone, void *udata, int domain, int flags)
{
	void *item;
#ifdef INVARIANTS
	bool skipdbg;
#endif

	item = NULL;

	if (zone->uz_import(zone->uz_arg, &item, 1, domain, flags) != 1)
		goto fail;
	atomic_add_long(&zone->uz_allocs, 1);

#ifdef INVARIANTS
	skipdbg = uma_dbg_zskip(zone, item);
#endif
	/*
	 * We have to call both the zone's init (not the keg's init)
	 * and the zone's ctor.  This is because the item is going from
	 * a keg slab directly to the user, and the user is expecting it
	 * to be both zone-init'd as well as zone-ctor'd.
	 */
	if (zone->uz_init != NULL) {
		if (zone->uz_init(item, zone->uz_size, flags) != 0) {
			zone_free_item(zone, item, udata, SKIP_FINI);
			goto fail;
		}
	}
	if (zone->uz_ctor != NULL &&
#ifdef INVARIANTS
	    (!skipdbg || zone->uz_ctor != trash_ctor ||
	    zone->uz_dtor != trash_dtor) &&
#endif
	    zone->uz_ctor(item, zone->uz_size, udata, flags) != 0) {
		zone_free_item(zone, item, udata, SKIP_DTOR);
		goto fail;
	}
#ifdef INVARIANTS
	if (!skipdbg)
		uma_dbg_alloc(zone, NULL, item);
#endif
	if (flags & M_ZERO)
		uma_zero_item(item, zone);

	CTR3(KTR_UMA, "zone_alloc_item item %p from %s(%p)", item,
	    zone->uz_name, zone);

	return (item);

fail:
	CTR2(KTR_UMA, "zone_alloc_item failed from %s(%p)",
	    zone->uz_name, zone);
	atomic_add_long(&zone->uz_fails, 1);
	return (NULL);
}

/* See uma.h */
void
uma_zfree_arg(uma_zone_t zone, void *item, void *udata)
{
	uma_cache_t cache;
	uma_bucket_t bucket;
	uma_zone_domain_t zdom;
	int cpu, domain, lockfail;
#ifdef INVARIANTS
	bool skipdbg;
#endif

	/* Enable entropy collection for RANDOM_ENABLE_UMA kernel option */
	random_harvest_fast_uma(&zone, sizeof(zone), 1, RANDOM_UMA);

	CTR2(KTR_UMA, "uma_zfree_arg thread %x zone %s", curthread,
	    zone->uz_name);

	KASSERT(curthread->td_critnest == 0 || SCHEDULER_STOPPED(),
	    ("uma_zfree_arg: called with spinlock or critical section held"));

        /* uma_zfree(..., NULL) does nothing, to match free(9). */
        if (item == NULL)
                return;
#ifdef DEBUG_MEMGUARD
	if (is_memguard_addr(item)) {
		if (zone->uz_dtor != NULL)
			zone->uz_dtor(item, zone->uz_size, udata);
		if (zone->uz_fini != NULL)
			zone->uz_fini(item, zone->uz_size);
		memguard_free(item);
		return;
	}
#endif
#ifdef INVARIANTS
	skipdbg = uma_dbg_zskip(zone, item);
	if (skipdbg == false) {
		if (zone->uz_flags & UMA_ZONE_MALLOC)
			uma_dbg_free(zone, udata, item);
		else
			uma_dbg_free(zone, NULL, item);
	}
	if (zone->uz_dtor != NULL && (!skipdbg ||
	    zone->uz_dtor != trash_dtor || zone->uz_ctor != trash_ctor))
#else
	if (zone->uz_dtor != NULL)
#endif
		zone->uz_dtor(item, zone->uz_size, udata);

	/*
	 * The race here is acceptable.  If we miss it we'll just have to wait
	 * a little longer for the limits to be reset.
	 */
	if (zone->uz_flags & UMA_ZFLAG_FULL)
		goto zfree_item;

	/*
	 * If possible, free to the per-CPU cache.  There are two
	 * requirements for safe access to the per-CPU cache: (1) the thread
	 * accessing the cache must not be preempted or yield during access,
	 * and (2) the thread must not migrate CPUs without switching which
	 * cache it accesses.  We rely on a critical section to prevent
	 * preemption and migration.  We release the critical section in
	 * order to acquire the zone mutex if we are unable to free to the
	 * current cache; when we re-acquire the critical section, we must
	 * detect and handle migration if it has occurred.
	 */
zfree_restart:
	critical_enter();
	cpu = curcpu;
	cache = &zone->uz_cpu[cpu];

zfree_start:
	/*
	 * Try to free into the allocbucket first to give LIFO ordering
	 * for cache-hot datastructures.  Spill over into the freebucket
	 * if necessary.  Alloc will swap them if one runs dry.
	 */
	bucket = cache->uc_allocbucket;
	if (bucket == NULL || bucket->ub_cnt >= bucket->ub_entries)
		bucket = cache->uc_freebucket;
	if (bucket != NULL && bucket->ub_cnt < bucket->ub_entries) {
		KASSERT(bucket->ub_bucket[bucket->ub_cnt] == NULL,
		    ("uma_zfree: Freeing to non free bucket index."));
		bucket->ub_bucket[bucket->ub_cnt] = item;
		bucket->ub_cnt++;
		cache->uc_frees++;
		critical_exit();
		return;
	}

	/*
	 * We must go back the zone, which requires acquiring the zone lock,
	 * which in turn means we must release and re-acquire the critical
	 * section.  Since the critical section is released, we may be
	 * preempted or migrate.  As such, make sure not to maintain any
	 * thread-local state specific to the cache from prior to releasing
	 * the critical section.
	 */
	critical_exit();
	if (zone->uz_count == 0 || bucketdisable)
		goto zfree_item;

	lockfail = 0;
	if (ZONE_TRYLOCK(zone) == 0) {
		/* Record contention to size the buckets. */
		ZONE_LOCK(zone);
		lockfail = 1;
	}
	critical_enter();
	cpu = curcpu;
	cache = &zone->uz_cpu[cpu];

	/*
	 * Since we have locked the zone we may as well send back our stats.
	 */
	atomic_add_long(&zone->uz_allocs, cache->uc_allocs);
	atomic_add_long(&zone->uz_frees, cache->uc_frees);
	cache->uc_allocs = 0;
	cache->uc_frees = 0;

	bucket = cache->uc_freebucket;
	if (bucket != NULL && bucket->ub_cnt < bucket->ub_entries) {
		ZONE_UNLOCK(zone);
		goto zfree_start;
	}
	cache->uc_freebucket = NULL;
	/* We are no longer associated with this CPU. */
	critical_exit();

	if ((zone->uz_flags & UMA_ZONE_NUMA) != 0)
		domain = PCPU_GET(domain);
	else 
		domain = 0;
	zdom = &zone->uz_domain[0];

	/* Can we throw this on the zone full list? */
	if (bucket != NULL) {
		CTR3(KTR_UMA,
		    "uma_zfree: zone %s(%p) putting bucket %p on free list",
		    zone->uz_name, zone, bucket);
		/* ub_cnt is pointing to the last free item */
		KASSERT(bucket->ub_cnt != 0,
		    ("uma_zfree: Attempting to insert an empty bucket onto the full list.\n"));
		if ((zone->uz_flags & UMA_ZONE_NOBUCKETCACHE) != 0) {
			ZONE_UNLOCK(zone);
			bucket_drain(zone, bucket);
			bucket_free(zone, bucket, udata);
			goto zfree_restart;
		} else
			LIST_INSERT_HEAD(&zdom->uzd_buckets, bucket, ub_link);
	}

	/*
	 * We bump the uz count when the cache size is insufficient to
	 * handle the working set.
	 */
	if (lockfail && zone->uz_count < BUCKET_MAX)
		zone->uz_count++;
	ZONE_UNLOCK(zone);

	bucket = bucket_alloc(zone, udata, M_NOWAIT);
	CTR3(KTR_UMA, "uma_zfree: zone %s(%p) allocated bucket %p",
	    zone->uz_name, zone, bucket);
	if (bucket) {
		critical_enter();
		cpu = curcpu;
		cache = &zone->uz_cpu[cpu];
		if (cache->uc_freebucket == NULL &&
		    ((zone->uz_flags & UMA_ZONE_NUMA) == 0 ||
		    domain == PCPU_GET(domain))) {
			cache->uc_freebucket = bucket;
			goto zfree_start;
		}
		/*
		 * We lost the race, start over.  We have to drop our
		 * critical section to free the bucket.
		 */
		critical_exit();
		bucket_free(zone, bucket, udata);
		goto zfree_restart;
	}

	/*
	 * If nothing else caught this, we'll just do an internal free.
	 */
zfree_item:
	zone_free_item(zone, item, udata, SKIP_DTOR);

	return;
}

void
uma_zfree_domain(uma_zone_t zone, void *item, void *udata)
{

	/* Enable entropy collection for RANDOM_ENABLE_UMA kernel option */
	random_harvest_fast_uma(&zone, sizeof(zone), 1, RANDOM_UMA);

	CTR2(KTR_UMA, "uma_zfree_domain thread %x zone %s", curthread,
	    zone->uz_name);

	KASSERT(curthread->td_critnest == 0 || SCHEDULER_STOPPED(),
	    ("uma_zfree_domain: called with spinlock or critical section held"));

        /* uma_zfree(..., NULL) does nothing, to match free(9). */
        if (item == NULL)
                return;
	zone_free_item(zone, item, udata, SKIP_NONE);
}

static void
slab_free_item(uma_keg_t keg, uma_slab_t slab, void *item)
{
	uma_domain_t dom;
	uint8_t freei;

	mtx_assert(&keg->uk_lock, MA_OWNED);
	MPASS(keg == slab->us_keg);

	dom = &keg->uk_domain[slab->us_domain];

	/* Do we need to remove from any lists? */
	if (slab->us_freecount+1 == keg->uk_ipers) {
		LIST_REMOVE(slab, us_link);
		LIST_INSERT_HEAD(&dom->ud_free_slab, slab, us_link);
	} else if (slab->us_freecount == 0) {
		LIST_REMOVE(slab, us_link);
		LIST_INSERT_HEAD(&dom->ud_part_slab, slab, us_link);
	}

	/* Slab management. */
	freei = ((uintptr_t)item - (uintptr_t)slab->us_data) / keg->uk_rsize;
	BIT_SET(SLAB_SETSIZE, freei, &slab->us_free);
	slab->us_freecount++;

	/* Keg statistics. */
	keg->uk_free++;
}

static void
zone_release(uma_zone_t zone, void **bucket, int cnt)
{
	void *item;
	uma_slab_t slab;
	uma_keg_t keg;
	uint8_t *mem;
	int clearfull;
	int i;

	clearfull = 0;
	keg = zone_first_keg(zone);
	KEG_LOCK(keg);
	for (i = 0; i < cnt; i++) {
		item = bucket[i];
		if (!(zone->uz_flags & UMA_ZONE_VTOSLAB)) {
			mem = (uint8_t *)((uintptr_t)item & (~UMA_SLAB_MASK));
			if (zone->uz_flags & UMA_ZONE_HASH) {
				slab = hash_sfind(&keg->uk_hash, mem);
			} else {
				mem += keg->uk_pgoff;
				slab = (uma_slab_t)mem;
			}
		} else {
			slab = vtoslab((vm_offset_t)item);
			if (slab->us_keg != keg) {
				KEG_UNLOCK(keg);
				keg = slab->us_keg;
				KEG_LOCK(keg);
			}
		}
		slab_free_item(keg, slab, item);
		if (keg->uk_flags & UMA_ZFLAG_FULL) {
			if (keg->uk_pages < keg->uk_maxpages) {
				keg->uk_flags &= ~UMA_ZFLAG_FULL;
				clearfull = 1;
			}

			/* 
			 * We can handle one more allocation. Since we're
			 * clearing ZFLAG_FULL, wake up all procs blocked
			 * on pages. This should be uncommon, so keeping this
			 * simple for now (rather than adding count of blocked 
			 * threads etc).
			 */
			wakeup(keg);
		}
	}
	KEG_UNLOCK(keg);
	if (clearfull) {
		ZONE_LOCK(zone);
		zone->uz_flags &= ~UMA_ZFLAG_FULL;
		wakeup(zone);
		ZONE_UNLOCK(zone);
	}

}

/*
 * Frees a single item to any zone.
 *
 * Arguments:
 *	zone   The zone to free to
 *	item   The item we're freeing
 *	udata  User supplied data for the dtor
 *	skip   Skip dtors and finis
 */
static void
zone_free_item(uma_zone_t zone, void *item, void *udata, enum zfreeskip skip)
{
#ifdef INVARIANTS
	bool skipdbg;

	skipdbg = uma_dbg_zskip(zone, item);
	if (skip == SKIP_NONE && !skipdbg) {
		if (zone->uz_flags & UMA_ZONE_MALLOC)
			uma_dbg_free(zone, udata, item);
		else
			uma_dbg_free(zone, NULL, item);
	}

	if (skip < SKIP_DTOR && zone->uz_dtor != NULL &&
	    (!skipdbg || zone->uz_dtor != trash_dtor ||
	    zone->uz_ctor != trash_ctor))
#else
	if (skip < SKIP_DTOR && zone->uz_dtor != NULL)
#endif
		zone->uz_dtor(item, zone->uz_size, udata);

	if (skip < SKIP_FINI && zone->uz_fini)
		zone->uz_fini(item, zone->uz_size);

	atomic_add_long(&zone->uz_frees, 1);
	zone->uz_release(zone->uz_arg, &item, 1);
}

/* See uma.h */
int
uma_zone_set_max(uma_zone_t zone, int nitems)
{
	uma_keg_t keg;

	keg = zone_first_keg(zone);
	if (keg == NULL)
		return (0);
	KEG_LOCK(keg);
	keg->uk_maxpages = (nitems / keg->uk_ipers) * keg->uk_ppera;
	if (keg->uk_maxpages * keg->uk_ipers < nitems)
		keg->uk_maxpages += keg->uk_ppera;
	nitems = (keg->uk_maxpages / keg->uk_ppera) * keg->uk_ipers;
	KEG_UNLOCK(keg);

	return (nitems);
}

/* See uma.h */
int
uma_zone_get_max(uma_zone_t zone)
{
	int nitems;
	uma_keg_t keg;

	keg = zone_first_keg(zone);
	if (keg == NULL)
		return (0);
	KEG_LOCK(keg);
	nitems = (keg->uk_maxpages / keg->uk_ppera) * keg->uk_ipers;
	KEG_UNLOCK(keg);

	return (nitems);
}

/* See uma.h */
void
uma_zone_set_warning(uma_zone_t zone, const char *warning)
{

	ZONE_LOCK(zone);
	zone->uz_warning = warning;
	ZONE_UNLOCK(zone);
}

/* See uma.h */
void
uma_zone_set_maxaction(uma_zone_t zone, uma_maxaction_t maxaction)
{

	ZONE_LOCK(zone);
	TASK_INIT(&zone->uz_maxaction, 0, (task_fn_t *)maxaction, zone);
	ZONE_UNLOCK(zone);
}

/* See uma.h */
int
uma_zone_get_cur(uma_zone_t zone)
{
	int64_t nitems;
	u_int i;

	ZONE_LOCK(zone);
	nitems = zone->uz_allocs - zone->uz_frees;
	CPU_FOREACH(i) {
		/*
		 * See the comment in sysctl_vm_zone_stats() regarding the
		 * safety of accessing the per-cpu caches. With the zone lock
		 * held, it is safe, but can potentially result in stale data.
		 */
		nitems += zone->uz_cpu[i].uc_allocs -
		    zone->uz_cpu[i].uc_frees;
	}
	ZONE_UNLOCK(zone);

	return (nitems < 0 ? 0 : nitems);
}

/* See uma.h */
void
uma_zone_set_init(uma_zone_t zone, uma_init uminit)
{
	uma_keg_t keg;

	keg = zone_first_keg(zone);
	KASSERT(keg != NULL, ("uma_zone_set_init: Invalid zone type"));
	KEG_LOCK(keg);
	KASSERT(keg->uk_pages == 0,
	    ("uma_zone_set_init on non-empty keg"));
	keg->uk_init = uminit;
	KEG_UNLOCK(keg);
}

/* See uma.h */
void
uma_zone_set_fini(uma_zone_t zone, uma_fini fini)
{
	uma_keg_t keg;

	keg = zone_first_keg(zone);
	KASSERT(keg != NULL, ("uma_zone_set_fini: Invalid zone type"));
	KEG_LOCK(keg);
	KASSERT(keg->uk_pages == 0,
	    ("uma_zone_set_fini on non-empty keg"));
	keg->uk_fini = fini;
	KEG_UNLOCK(keg);
}

/* See uma.h */
void
uma_zone_set_zinit(uma_zone_t zone, uma_init zinit)
{

	ZONE_LOCK(zone);
	KASSERT(zone_first_keg(zone)->uk_pages == 0,
	    ("uma_zone_set_zinit on non-empty keg"));
	zone->uz_init = zinit;
	ZONE_UNLOCK(zone);
}

/* See uma.h */
void
uma_zone_set_zfini(uma_zone_t zone, uma_fini zfini)
{

	ZONE_LOCK(zone);
	KASSERT(zone_first_keg(zone)->uk_pages == 0,
	    ("uma_zone_set_zfini on non-empty keg"));
	zone->uz_fini = zfini;
	ZONE_UNLOCK(zone);
}

/* See uma.h */
/* XXX uk_freef is not actually used with the zone locked */
void
uma_zone_set_freef(uma_zone_t zone, uma_free freef)
{
	uma_keg_t keg;

	keg = zone_first_keg(zone);
	KASSERT(keg != NULL, ("uma_zone_set_freef: Invalid zone type"));
	KEG_LOCK(keg);
	keg->uk_freef = freef;
	KEG_UNLOCK(keg);
}

/* See uma.h */
/* XXX uk_allocf is not actually used with the zone locked */
void
uma_zone_set_allocf(uma_zone_t zone, uma_alloc allocf)
{
	uma_keg_t keg;

	keg = zone_first_keg(zone);
	KEG_LOCK(keg);
	keg->uk_allocf = allocf;
	KEG_UNLOCK(keg);
}

/* See uma.h */
void
uma_zone_reserve(uma_zone_t zone, int items)
{
	uma_keg_t keg;

	keg = zone_first_keg(zone);
	if (keg == NULL)
		return;
	KEG_LOCK(keg);
	keg->uk_reserve = items;
	KEG_UNLOCK(keg);

	return;
}

/* See uma.h */
int
uma_zone_reserve_kva(uma_zone_t zone, int count)
{
	uma_keg_t keg;
	vm_offset_t kva;
	u_int pages;

	keg = zone_first_keg(zone);
	if (keg == NULL)
		return (0);
	pages = count / keg->uk_ipers;

	if (pages * keg->uk_ipers < count)
		pages++;
	pages *= keg->uk_ppera;

#ifdef UMA_MD_SMALL_ALLOC
	if (keg->uk_ppera > 1) {
#else
	if (1) {
#endif
		kva = kva_alloc((vm_size_t)pages * PAGE_SIZE);
		if (kva == 0)
			return (0);
	} else
		kva = 0;
	KEG_LOCK(keg);
	keg->uk_kva = kva;
	keg->uk_offset = 0;
	keg->uk_maxpages = pages;
#ifdef UMA_MD_SMALL_ALLOC
	keg->uk_allocf = (keg->uk_ppera > 1) ? noobj_alloc : uma_small_alloc;
#else
	keg->uk_allocf = noobj_alloc;
#endif
	keg->uk_flags |= UMA_ZONE_NOFREE;
	KEG_UNLOCK(keg);

	return (1);
}

/* See uma.h */
void
uma_prealloc(uma_zone_t zone, int items)
{
	uma_domain_t dom;
	uma_slab_t slab;
	uma_keg_t keg;
	int domain, slabs;

	keg = zone_first_keg(zone);
	if (keg == NULL)
		return;
	KEG_LOCK(keg);
	slabs = items / keg->uk_ipers;
	domain = 0;
	if (slabs * keg->uk_ipers < items)
		slabs++;
	while (slabs > 0) {
		slab = keg_alloc_slab(keg, zone, domain, M_WAITOK);
		if (slab == NULL)
			break;
		MPASS(slab->us_keg == keg);
		dom = &keg->uk_domain[slab->us_domain];
		LIST_INSERT_HEAD(&dom->ud_free_slab, slab, us_link);
		slabs--;
		domain = (domain + 1) % vm_ndomains;
	}
	KEG_UNLOCK(keg);
}

/* See uma.h */
static void
uma_reclaim_locked(bool kmem_danger)
{

	CTR0(KTR_UMA, "UMA: vm asked us to release pages!");
	sx_assert(&uma_drain_lock, SA_XLOCKED);
	bucket_enable();
	zone_foreach(zone_drain);
	if (vm_page_count_min() || kmem_danger) {
		cache_drain_safe(NULL);
		zone_foreach(zone_drain);
	}
	/*
	 * Some slabs may have been freed but this zone will be visited early
	 * we visit again so that we can free pages that are empty once other
	 * zones are drained.  We have to do the same for buckets.
	 */
	zone_drain(slabzone);
	bucket_zone_drain();
}

void
uma_reclaim(void)
{

	sx_xlock(&uma_drain_lock);
	uma_reclaim_locked(false);
	sx_xunlock(&uma_drain_lock);
}

static volatile int uma_reclaim_needed;

void
uma_reclaim_wakeup(void)
{

	if (atomic_fetchadd_int(&uma_reclaim_needed, 1) == 0)
		wakeup(uma_reclaim);
}

void
uma_reclaim_worker(void *arg __unused)
{

	for (;;) {
		sx_xlock(&uma_drain_lock);
		while (atomic_load_int(&uma_reclaim_needed) == 0)
			sx_sleep(uma_reclaim, &uma_drain_lock, PVM, "umarcl",
			    hz);
		sx_xunlock(&uma_drain_lock);
		EVENTHANDLER_INVOKE(vm_lowmem, VM_LOW_KMEM);
		sx_xlock(&uma_drain_lock);
		uma_reclaim_locked(true);
		atomic_store_int(&uma_reclaim_needed, 0);
		sx_xunlock(&uma_drain_lock);
		/* Don't fire more than once per-second. */
		pause("umarclslp", hz);
	}
}

/* See uma.h */
int
uma_zone_exhausted(uma_zone_t zone)
{
	int full;

	ZONE_LOCK(zone);
	full = (zone->uz_flags & UMA_ZFLAG_FULL);
	ZONE_UNLOCK(zone);
	return (full);	
}

int
uma_zone_exhausted_nolock(uma_zone_t zone)
{
	return (zone->uz_flags & UMA_ZFLAG_FULL);
}

void *
uma_large_malloc_domain(vm_size_t size, int domain, int wait)
{
	vm_offset_t addr;
	uma_slab_t slab;

	slab = zone_alloc_item(slabzone, NULL, domain, wait);
	if (slab == NULL)
		return (NULL);
	if (domain == UMA_ANYDOMAIN)
		addr = kmem_malloc(kernel_arena, size, wait);
	else
		addr = kmem_malloc_domain(domain, size, wait);
	if (addr != 0) {
		vsetslab(addr, slab);
		slab->us_data = (void *)addr;
		slab->us_flags = UMA_SLAB_KERNEL | UMA_SLAB_MALLOC;
		slab->us_size = size;
		slab->us_domain = vm_phys_domain(PHYS_TO_VM_PAGE(
		    pmap_kextract(addr)));
		uma_total_inc(size);
	} else {
		zone_free_item(slabzone, slab, NULL, SKIP_NONE);
	}

	return ((void *)addr);
}

void *
uma_large_malloc(vm_size_t size, int wait)
{

	return uma_large_malloc_domain(size, UMA_ANYDOMAIN, wait);
}

void
uma_large_free(uma_slab_t slab)
{

	KASSERT((slab->us_flags & UMA_SLAB_KERNEL) != 0,
	    ("uma_large_free:  Memory not allocated with uma_large_malloc."));
	kmem_free(kernel_arena, (vm_offset_t)slab->us_data, slab->us_size);
	uma_total_dec(slab->us_size);
	zone_free_item(slabzone, slab, NULL, SKIP_NONE);
}

static void
uma_zero_item(void *item, uma_zone_t zone)
{

	bzero(item, zone->uz_size);
}

unsigned long
uma_limit(void)
{

	return (uma_kmem_limit);
}

void
uma_set_limit(unsigned long limit)
{

	uma_kmem_limit = limit;
}

unsigned long
uma_size(void)
{

	return (uma_kmem_total);
}

long
uma_avail(void)
{

	return (uma_kmem_limit - uma_kmem_total);
}

void
uma_print_stats(void)
{
	zone_foreach(uma_print_zone);
}

static void
slab_print(uma_slab_t slab)
{
	printf("slab: keg %p, data %p, freecount %d\n",
		slab->us_keg, slab->us_data, slab->us_freecount);
}

static void
cache_print(uma_cache_t cache)
{
	printf("alloc: %p(%d), free: %p(%d)\n",
		cache->uc_allocbucket,
		cache->uc_allocbucket?cache->uc_allocbucket->ub_cnt:0,
		cache->uc_freebucket,
		cache->uc_freebucket?cache->uc_freebucket->ub_cnt:0);
}

static void
uma_print_keg(uma_keg_t keg)
{
	uma_domain_t dom;
	uma_slab_t slab;
	int i;

	printf("keg: %s(%p) size %d(%d) flags %#x ipers %d ppera %d "
	    "out %d free %d limit %d\n",
	    keg->uk_name, keg, keg->uk_size, keg->uk_rsize, keg->uk_flags,
	    keg->uk_ipers, keg->uk_ppera,
	    (keg->uk_pages / keg->uk_ppera) * keg->uk_ipers - keg->uk_free,
	    keg->uk_free, (keg->uk_maxpages / keg->uk_ppera) * keg->uk_ipers);
	for (i = 0; i < vm_ndomains; i++) {
		dom = &keg->uk_domain[i];
		printf("Part slabs:\n");
		LIST_FOREACH(slab, &dom->ud_part_slab, us_link)
			slab_print(slab);
		printf("Free slabs:\n");
		LIST_FOREACH(slab, &dom->ud_free_slab, us_link)
			slab_print(slab);
		printf("Full slabs:\n");
		LIST_FOREACH(slab, &dom->ud_full_slab, us_link)
			slab_print(slab);
	}
}

void
uma_print_zone(uma_zone_t zone)
{
	uma_cache_t cache;
	uma_klink_t kl;
	int i;

	printf("zone: %s(%p) size %d flags %#x\n",
	    zone->uz_name, zone, zone->uz_size, zone->uz_flags);
	LIST_FOREACH(kl, &zone->uz_kegs, kl_link)
		uma_print_keg(kl->kl_keg);
	CPU_FOREACH(i) {
		cache = &zone->uz_cpu[i];
		printf("CPU %d Cache:\n", i);
		cache_print(cache);
	}
}

#ifdef DDB
/*
 * Generate statistics across both the zone and its per-cpu cache's.  Return
 * desired statistics if the pointer is non-NULL for that statistic.
 *
 * Note: does not update the zone statistics, as it can't safely clear the
 * per-CPU cache statistic.
 *
 * XXXRW: Following the uc_allocbucket and uc_freebucket pointers here isn't
 * safe from off-CPU; we should modify the caches to track this information
 * directly so that we don't have to.
 */
static void
uma_zone_sumstat(uma_zone_t z, int *cachefreep, uint64_t *allocsp,
    uint64_t *freesp, uint64_t *sleepsp)
{
	uma_cache_t cache;
	uint64_t allocs, frees, sleeps;
	int cachefree, cpu;

	allocs = frees = sleeps = 0;
	cachefree = 0;
	CPU_FOREACH(cpu) {
		cache = &z->uz_cpu[cpu];
		if (cache->uc_allocbucket != NULL)
			cachefree += cache->uc_allocbucket->ub_cnt;
		if (cache->uc_freebucket != NULL)
			cachefree += cache->uc_freebucket->ub_cnt;
		allocs += cache->uc_allocs;
		frees += cache->uc_frees;
	}
	allocs += z->uz_allocs;
	frees += z->uz_frees;
	sleeps += z->uz_sleeps;
	if (cachefreep != NULL)
		*cachefreep = cachefree;
	if (allocsp != NULL)
		*allocsp = allocs;
	if (freesp != NULL)
		*freesp = frees;
	if (sleepsp != NULL)
		*sleepsp = sleeps;
}
#endif /* DDB */

static int
sysctl_vm_zone_count(SYSCTL_HANDLER_ARGS)
{
	uma_keg_t kz;
	uma_zone_t z;
	int count;

	count = 0;
	rw_rlock(&uma_rwlock);
	LIST_FOREACH(kz, &uma_kegs, uk_link) {
		LIST_FOREACH(z, &kz->uk_zones, uz_link)
			count++;
	}
	rw_runlock(&uma_rwlock);
	return (sysctl_handle_int(oidp, &count, 0, req));
}

static int
sysctl_vm_zone_stats(SYSCTL_HANDLER_ARGS)
{
	struct uma_stream_header ush;
	struct uma_type_header uth;
	struct uma_percpu_stat *ups;
	uma_bucket_t bucket;
	uma_zone_domain_t zdom;
	struct sbuf sbuf;
	uma_cache_t cache;
	uma_klink_t kl;
	uma_keg_t kz;
	uma_zone_t z;
	uma_keg_t k;
	int count, error, i;

	error = sysctl_wire_old_buffer(req, 0);
	if (error != 0)
		return (error);
	sbuf_new_for_sysctl(&sbuf, NULL, 128, req);
	sbuf_clear_flags(&sbuf, SBUF_INCLUDENUL);
	ups = malloc((mp_maxid + 1) * sizeof(*ups), M_TEMP, M_WAITOK);

	count = 0;
	rw_rlock(&uma_rwlock);
	LIST_FOREACH(kz, &uma_kegs, uk_link) {
		LIST_FOREACH(z, &kz->uk_zones, uz_link)
			count++;
	}

	/*
	 * Insert stream header.
	 */
	bzero(&ush, sizeof(ush));
	ush.ush_version = UMA_STREAM_VERSION;
	ush.ush_maxcpus = (mp_maxid + 1);
	ush.ush_count = count;
	(void)sbuf_bcat(&sbuf, &ush, sizeof(ush));

	LIST_FOREACH(kz, &uma_kegs, uk_link) {
		LIST_FOREACH(z, &kz->uk_zones, uz_link) {
			bzero(&uth, sizeof(uth));
			ZONE_LOCK(z);
			strlcpy(uth.uth_name, z->uz_name, UTH_MAX_NAME);
			uth.uth_align = kz->uk_align;
			uth.uth_size = kz->uk_size;
			uth.uth_rsize = kz->uk_rsize;
			LIST_FOREACH(kl, &z->uz_kegs, kl_link) {
				k = kl->kl_keg;
				uth.uth_maxpages += k->uk_maxpages;
				uth.uth_pages += k->uk_pages;
				uth.uth_keg_free += k->uk_free;
				uth.uth_limit = (k->uk_maxpages / k->uk_ppera)
				    * k->uk_ipers;
			}

			/*
			 * A zone is secondary is it is not the first entry
			 * on the keg's zone list.
			 */
			if ((z->uz_flags & UMA_ZONE_SECONDARY) &&
			    (LIST_FIRST(&kz->uk_zones) != z))
				uth.uth_zone_flags = UTH_ZONE_SECONDARY;

			for (i = 0; i < vm_ndomains; i++) {
				zdom = &z->uz_domain[i];
				LIST_FOREACH(bucket, &zdom->uzd_buckets,
				    ub_link)
					uth.uth_zone_free += bucket->ub_cnt;
			}
			uth.uth_allocs = z->uz_allocs;
			uth.uth_frees = z->uz_frees;
			uth.uth_fails = z->uz_fails;
			uth.uth_sleeps = z->uz_sleeps;
			/*
			 * While it is not normally safe to access the cache
			 * bucket pointers while not on the CPU that owns the
			 * cache, we only allow the pointers to be exchanged
			 * without the zone lock held, not invalidated, so
			 * accept the possible race associated with bucket
			 * exchange during monitoring.
			 */
			for (i = 0; i < mp_maxid + 1; i++) {
				bzero(&ups[i], sizeof(*ups));
				if (kz->uk_flags & UMA_ZFLAG_INTERNAL ||
				    CPU_ABSENT(i))
					continue;
				cache = &z->uz_cpu[i];
				if (cache->uc_allocbucket != NULL)
					ups[i].ups_cache_free +=
					    cache->uc_allocbucket->ub_cnt;
				if (cache->uc_freebucket != NULL)
					ups[i].ups_cache_free +=
					    cache->uc_freebucket->ub_cnt;
				ups[i].ups_allocs = cache->uc_allocs;
				ups[i].ups_frees = cache->uc_frees;
			}
			ZONE_UNLOCK(z);
			(void)sbuf_bcat(&sbuf, &uth, sizeof(uth));
			for (i = 0; i < mp_maxid + 1; i++)
				(void)sbuf_bcat(&sbuf, &ups[i], sizeof(ups[i]));
		}
	}
	rw_runlock(&uma_rwlock);
	error = sbuf_finish(&sbuf);
	sbuf_delete(&sbuf);
	free(ups, M_TEMP);
	return (error);
}

int
sysctl_handle_uma_zone_max(SYSCTL_HANDLER_ARGS)
{
	uma_zone_t zone = *(uma_zone_t *)arg1;
	int error, max;

	max = uma_zone_get_max(zone);
	error = sysctl_handle_int(oidp, &max, 0, req);
	if (error || !req->newptr)
		return (error);

	uma_zone_set_max(zone, max);

	return (0);
}

int
sysctl_handle_uma_zone_cur(SYSCTL_HANDLER_ARGS)
{
	uma_zone_t zone = *(uma_zone_t *)arg1;
	int cur;

	cur = uma_zone_get_cur(zone);
	return (sysctl_handle_int(oidp, &cur, 0, req));
}

#ifdef INVARIANTS
static uma_slab_t
uma_dbg_getslab(uma_zone_t zone, void *item)
{
	uma_slab_t slab;
	uma_keg_t keg;
	uint8_t *mem;

	mem = (uint8_t *)((uintptr_t)item & (~UMA_SLAB_MASK));
	if (zone->uz_flags & UMA_ZONE_VTOSLAB) {
		slab = vtoslab((vm_offset_t)mem);
	} else {
		/*
		 * It is safe to return the slab here even though the
		 * zone is unlocked because the item's allocation state
		 * essentially holds a reference.
		 */
		ZONE_LOCK(zone);
		keg = LIST_FIRST(&zone->uz_kegs)->kl_keg;
		if (keg->uk_flags & UMA_ZONE_HASH)
			slab = hash_sfind(&keg->uk_hash, mem);
		else
			slab = (uma_slab_t)(mem + keg->uk_pgoff);
		ZONE_UNLOCK(zone);
	}

	return (slab);
}

static bool
uma_dbg_zskip(uma_zone_t zone, void *mem)
{
	uma_keg_t keg;

	if ((keg = zone_first_keg(zone)) == NULL)
		return (true);

	return (uma_dbg_kskip(keg, mem));
}

static bool
uma_dbg_kskip(uma_keg_t keg, void *mem)
{
	uintptr_t idx;

	if (dbg_divisor == 0)
		return (true);

	if (dbg_divisor == 1)
		return (false);

	idx = (uintptr_t)mem >> PAGE_SHIFT;
	if (keg->uk_ipers > 1) {
		idx *= keg->uk_ipers;
		idx += ((uintptr_t)mem & PAGE_MASK) / keg->uk_rsize;
	}

	if ((idx / dbg_divisor) * dbg_divisor != idx) {
		counter_u64_add(uma_skip_cnt, 1);
		return (true);
	}
	counter_u64_add(uma_dbg_cnt, 1);

	return (false);
}

/*
 * Set up the slab's freei data such that uma_dbg_free can function.
 *
 */
static void
uma_dbg_alloc(uma_zone_t zone, uma_slab_t slab, void *item)
{
	uma_keg_t keg;
	int freei;

	if (slab == NULL) {
		slab = uma_dbg_getslab(zone, item);
		if (slab == NULL) 
			panic("uma: item %p did not belong to zone %s\n",
			    item, zone->uz_name);
	}
	keg = slab->us_keg;
	freei = ((uintptr_t)item - (uintptr_t)slab->us_data) / keg->uk_rsize;

	if (BIT_ISSET(SLAB_SETSIZE, freei, &slab->us_debugfree))
		panic("Duplicate alloc of %p from zone %p(%s) slab %p(%d)\n",
		    item, zone, zone->uz_name, slab, freei);
	BIT_SET_ATOMIC(SLAB_SETSIZE, freei, &slab->us_debugfree);

	return;
}

/*
 * Verifies freed addresses.  Checks for alignment, valid slab membership
 * and duplicate frees.
 *
 */
static void
uma_dbg_free(uma_zone_t zone, uma_slab_t slab, void *item)
{
	uma_keg_t keg;
	int freei;

	if (slab == NULL) {
		slab = uma_dbg_getslab(zone, item);
		if (slab == NULL) 
			panic("uma: Freed item %p did not belong to zone %s\n",
			    item, zone->uz_name);
	}
	keg = slab->us_keg;
	freei = ((uintptr_t)item - (uintptr_t)slab->us_data) / keg->uk_rsize;

	if (freei >= keg->uk_ipers)
		panic("Invalid free of %p from zone %p(%s) slab %p(%d)\n",
		    item, zone, zone->uz_name, slab, freei);

	if (((freei * keg->uk_rsize) + slab->us_data) != item) 
		panic("Unaligned free of %p from zone %p(%s) slab %p(%d)\n",
		    item, zone, zone->uz_name, slab, freei);

	if (!BIT_ISSET(SLAB_SETSIZE, freei, &slab->us_debugfree))
		panic("Duplicate free of %p from zone %p(%s) slab %p(%d)\n",
		    item, zone, zone->uz_name, slab, freei);

	BIT_CLR_ATOMIC(SLAB_SETSIZE, freei, &slab->us_debugfree);
}
#endif /* INVARIANTS */

#ifdef DDB
DB_SHOW_COMMAND(uma, db_show_uma)
{
	uma_bucket_t bucket;
	uma_keg_t kz;
	uma_zone_t z;
	uma_zone_domain_t zdom;
	uint64_t allocs, frees, sleeps;
	int cachefree, i;

	db_printf("%18s %8s %8s %8s %12s %8s %8s\n", "Zone", "Size", "Used",
	    "Free", "Requests", "Sleeps", "Bucket");
	LIST_FOREACH(kz, &uma_kegs, uk_link) {
		LIST_FOREACH(z, &kz->uk_zones, uz_link) {
			if (kz->uk_flags & UMA_ZFLAG_INTERNAL) {
				allocs = z->uz_allocs;
				frees = z->uz_frees;
				sleeps = z->uz_sleeps;
				cachefree = 0;
			} else
				uma_zone_sumstat(z, &cachefree, &allocs,
				    &frees, &sleeps);
			if (!((z->uz_flags & UMA_ZONE_SECONDARY) &&
			    (LIST_FIRST(&kz->uk_zones) != z)))
				cachefree += kz->uk_free;
			for (i = 0; i < vm_ndomains; i++) {
				zdom = &z->uz_domain[i];
				LIST_FOREACH(bucket, &zdom->uzd_buckets,
				    ub_link)
					cachefree += bucket->ub_cnt;
			}
			db_printf("%18s %8ju %8jd %8d %12ju %8ju %8u\n",
			    z->uz_name, (uintmax_t)kz->uk_size,
			    (intmax_t)(allocs - frees), cachefree,
			    (uintmax_t)allocs, sleeps, z->uz_count);
			if (db_pager_quit)
				return;
		}
	}
}

DB_SHOW_COMMAND(umacache, db_show_umacache)
{
	uma_bucket_t bucket;
	uma_zone_t z;
	uma_zone_domain_t zdom;
	uint64_t allocs, frees;
	int cachefree, i;

	db_printf("%18s %8s %8s %8s %12s %8s\n", "Zone", "Size", "Used", "Free",
	    "Requests", "Bucket");
	LIST_FOREACH(z, &uma_cachezones, uz_link) {
		uma_zone_sumstat(z, &cachefree, &allocs, &frees, NULL);
		for (i = 0; i < vm_ndomains; i++) {
			zdom = &z->uz_domain[i];
			LIST_FOREACH(bucket, &zdom->uzd_buckets, ub_link)
				cachefree += bucket->ub_cnt;
		}
		db_printf("%18s %8ju %8jd %8d %12ju %8u\n",
		    z->uz_name, (uintmax_t)z->uz_size,
		    (intmax_t)(allocs - frees), cachefree,
		    (uintmax_t)allocs, z->uz_count);
		if (db_pager_quit)
			return;
	}
}
#endif	/* DDB */