summaryrefslogtreecommitdiffstats
path: root/module/zfs/arc.c
blob: 79b04f45564f6111acb2d2818b881ca71b38e621 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
5322
5323
5324
5325
5326
5327
5328
5329
5330
5331
5332
5333
5334
5335
5336
5337
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
5349
5350
5351
5352
5353
5354
5355
5356
5357
5358
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
5377
5378
5379
5380
5381
5382
5383
5384
5385
5386
5387
5388
5389
5390
5391
5392
5393
5394
5395
5396
5397
5398
5399
5400
5401
5402
5403
5404
5405
5406
5407
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
5538
5539
5540
5541
5542
5543
5544
5545
5546
5547
5548
5549
5550
5551
5552
5553
5554
5555
5556
5557
5558
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
5599
5600
5601
5602
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
5636
5637
5638
5639
5640
5641
5642
5643
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
5659
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
5682
5683
5684
5685
5686
5687
5688
5689
5690
5691
5692
5693
5694
5695
5696
5697
5698
5699
5700
5701
5702
5703
5704
5705
5706
5707
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
5723
5724
5725
5726
5727
5728
5729
5730
5731
5732
5733
5734
5735
5736
5737
5738
5739
5740
5741
5742
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
5758
5759
5760
5761
5762
5763
5764
5765
5766
5767
5768
5769
5770
5771
5772
5773
5774
5775
5776
5777
5778
5779
5780
5781
5782
5783
5784
5785
5786
5787
5788
5789
5790
5791
5792
5793
5794
5795
5796
5797
5798
5799
5800
5801
5802
5803
5804
5805
5806
5807
5808
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
5834
5835
5836
5837
5838
5839
5840
5841
5842
5843
5844
5845
5846
5847
5848
5849
5850
5851
5852
5853
5854
5855
5856
5857
5858
5859
5860
5861
5862
5863
5864
5865
5866
5867
5868
5869
5870
5871
5872
5873
5874
5875
5876
5877
5878
5879
5880
5881
5882
5883
5884
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
5900
5901
5902
5903
5904
5905
5906
5907
5908
5909
5910
5911
5912
5913
5914
5915
5916
5917
5918
5919
5920
5921
5922
5923
5924
5925
5926
5927
5928
5929
5930
5931
5932
5933
5934
5935
5936
5937
5938
5939
5940
5941
5942
5943
5944
5945
5946
5947
5948
5949
5950
5951
5952
5953
5954
5955
5956
5957
5958
5959
5960
5961
5962
5963
5964
5965
5966
5967
5968
5969
5970
5971
5972
5973
5974
5975
5976
5977
5978
5979
5980
5981
5982
5983
5984
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
5996
5997
5998
5999
6000
6001
6002
6003
6004
6005
6006
6007
6008
6009
6010
6011
6012
6013
6014
6015
6016
6017
6018
6019
6020
6021
6022
6023
6024
6025
6026
6027
6028
6029
6030
6031
6032
6033
6034
6035
6036
6037
6038
6039
6040
6041
6042
6043
6044
6045
6046
6047
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
6060
6061
6062
6063
6064
6065
6066
6067
6068
6069
6070
6071
6072
6073
6074
6075
6076
6077
6078
6079
6080
6081
6082
6083
6084
6085
6086
6087
6088
6089
6090
6091
6092
6093
6094
6095
6096
6097
6098
6099
6100
6101
6102
6103
6104
6105
6106
6107
6108
6109
6110
6111
6112
6113
6114
6115
6116
6117
6118
6119
6120
6121
6122
6123
6124
6125
6126
6127
6128
6129
6130
6131
6132
6133
6134
6135
6136
6137
6138
6139
6140
6141
6142
6143
6144
6145
6146
6147
6148
6149
6150
6151
6152
6153
6154
6155
6156
6157
6158
6159
6160
6161
6162
6163
6164
6165
6166
6167
6168
6169
6170
6171
6172
6173
6174
6175
6176
6177
6178
6179
6180
6181
6182
6183
6184
6185
6186
6187
6188
6189
6190
6191
6192
6193
6194
6195
6196
6197
6198
6199
6200
6201
6202
6203
6204
6205
6206
6207
6208
6209
6210
6211
6212
6213
6214
6215
6216
6217
6218
6219
6220
6221
6222
6223
6224
6225
6226
6227
6228
6229
6230
6231
6232
6233
6234
6235
6236
6237
6238
6239
6240
6241
6242
6243
6244
6245
6246
6247
6248
6249
6250
6251
6252
6253
6254
6255
6256
6257
6258
6259
6260
6261
6262
6263
6264
6265
6266
6267
6268
6269
6270
6271
6272
6273
6274
6275
6276
6277
6278
6279
6280
6281
6282
6283
6284
6285
6286
6287
6288
6289
6290
6291
6292
6293
6294
6295
6296
6297
6298
6299
6300
6301
6302
6303
6304
6305
6306
6307
6308
6309
6310
6311
6312
6313
6314
6315
6316
6317
6318
6319
6320
6321
6322
6323
6324
6325
6326
6327
6328
6329
6330
6331
6332
6333
6334
6335
6336
6337
6338
6339
6340
6341
6342
6343
6344
6345
6346
6347
6348
6349
6350
6351
6352
6353
6354
6355
6356
6357
6358
6359
6360
6361
6362
6363
6364
6365
6366
6367
6368
6369
6370
6371
6372
6373
6374
6375
6376
6377
6378
6379
6380
6381
6382
6383
6384
6385
6386
6387
6388
6389
6390
6391
6392
6393
6394
6395
6396
6397
6398
6399
6400
6401
6402
6403
6404
6405
6406
6407
6408
6409
6410
6411
6412
6413
6414
6415
6416
6417
6418
6419
6420
6421
6422
6423
6424
6425
6426
6427
6428
6429
6430
6431
6432
6433
6434
6435
6436
6437
6438
6439
6440
6441
6442
6443
6444
6445
6446
6447
6448
6449
6450
6451
6452
6453
6454
6455
6456
6457
6458
6459
6460
6461
6462
6463
6464
6465
6466
6467
6468
6469
6470
6471
6472
6473
6474
6475
6476
6477
6478
6479
6480
6481
6482
6483
6484
6485
6486
6487
6488
6489
6490
6491
6492
6493
6494
6495
6496
6497
6498
6499
6500
6501
6502
6503
6504
6505
6506
6507
6508
6509
6510
6511
6512
6513
6514
6515
6516
6517
6518
6519
6520
6521
6522
6523
6524
6525
6526
6527
6528
6529
6530
6531
6532
6533
6534
6535
6536
6537
6538
6539
6540
6541
6542
6543
6544
6545
6546
6547
6548
6549
6550
6551
6552
6553
6554
6555
6556
6557
6558
6559
6560
6561
6562
6563
6564
6565
6566
6567
6568
6569
6570
6571
6572
6573
6574
6575
6576
6577
6578
6579
6580
6581
6582
6583
6584
6585
6586
6587
6588
6589
6590
6591
6592
6593
6594
6595
6596
6597
6598
6599
6600
6601
6602
6603
6604
6605
6606
6607
6608
6609
6610
6611
6612
6613
6614
6615
6616
6617
6618
6619
6620
6621
6622
6623
6624
6625
6626
6627
6628
6629
6630
6631
6632
6633
6634
6635
6636
6637
6638
6639
6640
6641
6642
6643
6644
6645
6646
6647
6648
6649
6650
6651
6652
6653
6654
6655
6656
6657
6658
6659
6660
6661
6662
6663
6664
6665
6666
6667
6668
6669
6670
6671
6672
6673
6674
6675
6676
6677
6678
6679
6680
6681
6682
6683
6684
6685
6686
6687
6688
6689
6690
6691
6692
6693
6694
6695
6696
6697
6698
6699
6700
6701
6702
6703
6704
6705
6706
6707
6708
6709
6710
6711
6712
6713
6714
6715
6716
6717
6718
6719
6720
6721
6722
6723
6724
6725
6726
6727
6728
6729
6730
6731
6732
6733
6734
6735
6736
6737
6738
6739
6740
6741
6742
6743
6744
6745
6746
6747
6748
6749
6750
6751
6752
6753
6754
6755
6756
6757
6758
6759
6760
6761
6762
6763
6764
6765
6766
6767
6768
6769
6770
6771
6772
6773
6774
6775
6776
6777
6778
6779
6780
6781
6782
6783
6784
6785
6786
6787
6788
6789
6790
6791
6792
6793
6794
6795
6796
6797
6798
6799
6800
6801
6802
6803
6804
6805
6806
6807
6808
6809
6810
6811
6812
6813
6814
6815
6816
6817
6818
6819
6820
6821
6822
6823
6824
6825
6826
6827
6828
6829
6830
6831
6832
6833
6834
6835
6836
6837
6838
6839
6840
6841
6842
6843
6844
6845
6846
6847
6848
6849
6850
6851
6852
6853
6854
6855
6856
6857
6858
6859
6860
6861
6862
6863
6864
6865
6866
6867
6868
6869
6870
6871
6872
6873
6874
6875
6876
6877
6878
6879
6880
6881
6882
6883
6884
6885
6886
6887
6888
6889
6890
6891
6892
6893
6894
6895
6896
6897
6898
6899
6900
6901
6902
6903
6904
6905
6906
6907
6908
6909
6910
6911
6912
6913
6914
6915
6916
6917
6918
6919
6920
6921
6922
6923
6924
6925
6926
6927
6928
6929
6930
6931
6932
6933
6934
6935
6936
6937
6938
6939
6940
6941
6942
6943
6944
6945
6946
6947
6948
6949
6950
6951
6952
6953
6954
6955
6956
6957
6958
6959
6960
6961
6962
6963
6964
6965
6966
6967
6968
6969
6970
6971
6972
6973
6974
6975
6976
6977
6978
6979
6980
6981
6982
6983
6984
6985
6986
6987
6988
6989
6990
6991
6992
6993
6994
6995
6996
6997
6998
6999
7000
7001
7002
7003
7004
7005
7006
7007
7008
7009
7010
7011
7012
7013
7014
7015
7016
7017
7018
7019
7020
7021
7022
7023
7024
7025
7026
7027
7028
7029
7030
7031
7032
7033
7034
7035
7036
7037
7038
7039
7040
7041
7042
7043
7044
7045
7046
7047
7048
7049
7050
7051
7052
7053
7054
7055
7056
7057
7058
7059
7060
7061
7062
7063
7064
7065
7066
7067
7068
7069
7070
7071
7072
7073
7074
7075
7076
7077
7078
7079
7080
7081
7082
7083
7084
7085
7086
7087
7088
7089
7090
7091
7092
7093
7094
7095
7096
7097
7098
7099
7100
7101
7102
7103
7104
7105
7106
7107
7108
7109
7110
7111
7112
7113
7114
7115
7116
7117
7118
7119
7120
7121
7122
7123
7124
7125
7126
7127
7128
7129
7130
7131
7132
7133
7134
7135
7136
7137
7138
7139
7140
7141
7142
7143
7144
7145
7146
7147
7148
7149
7150
7151
7152
7153
7154
7155
7156
7157
7158
7159
7160
7161
7162
7163
7164
7165
7166
7167
7168
7169
7170
7171
7172
7173
7174
7175
7176
7177
7178
7179
7180
7181
7182
7183
7184
7185
7186
7187
7188
7189
7190
7191
7192
7193
7194
7195
7196
7197
7198
7199
7200
7201
7202
7203
7204
7205
7206
7207
7208
7209
7210
7211
7212
7213
7214
7215
7216
7217
7218
7219
7220
7221
7222
7223
7224
7225
7226
7227
7228
7229
7230
7231
7232
7233
7234
7235
7236
7237
7238
7239
7240
7241
7242
7243
7244
7245
7246
7247
7248
7249
7250
7251
7252
7253
7254
7255
7256
7257
7258
7259
7260
7261
7262
7263
7264
7265
7266
7267
7268
7269
7270
7271
7272
7273
7274
7275
7276
7277
7278
7279
7280
7281
7282
7283
7284
7285
7286
7287
7288
7289
7290
7291
7292
7293
7294
7295
7296
7297
7298
7299
7300
7301
7302
7303
7304
7305
7306
7307
7308
7309
7310
7311
7312
7313
7314
7315
7316
7317
7318
7319
7320
7321
7322
7323
7324
7325
7326
7327
7328
7329
7330
7331
7332
7333
7334
7335
7336
7337
7338
7339
7340
7341
7342
7343
7344
7345
7346
7347
7348
7349
7350
7351
7352
7353
7354
7355
7356
7357
7358
7359
7360
7361
7362
7363
7364
7365
7366
7367
7368
7369
7370
7371
7372
7373
7374
7375
7376
7377
7378
7379
7380
7381
7382
7383
7384
7385
7386
7387
7388
7389
7390
7391
7392
7393
7394
7395
7396
7397
7398
7399
7400
7401
7402
7403
7404
7405
7406
7407
7408
7409
7410
7411
7412
7413
7414
7415
7416
7417
7418
7419
7420
7421
7422
7423
7424
7425
7426
7427
7428
7429
7430
7431
7432
7433
7434
7435
7436
7437
7438
7439
7440
7441
7442
7443
7444
7445
7446
7447
7448
7449
7450
7451
7452
7453
7454
7455
7456
7457
7458
7459
7460
7461
7462
7463
7464
7465
7466
7467
7468
7469
7470
7471
7472
7473
7474
7475
7476
7477
7478
7479
7480
7481
7482
7483
7484
7485
7486
7487
7488
7489
7490
7491
7492
7493
7494
7495
7496
7497
7498
7499
7500
7501
7502
7503
7504
7505
7506
7507
7508
7509
7510
7511
7512
7513
7514
7515
7516
7517
7518
7519
7520
7521
7522
7523
7524
7525
7526
7527
7528
7529
7530
7531
7532
7533
7534
7535
7536
7537
7538
7539
7540
7541
7542
7543
7544
7545
7546
7547
7548
7549
7550
7551
7552
7553
7554
7555
7556
7557
7558
7559
7560
7561
7562
7563
7564
7565
7566
7567
7568
7569
7570
7571
7572
7573
7574
7575
7576
7577
7578
7579
7580
7581
7582
7583
7584
7585
7586
7587
7588
7589
7590
7591
7592
7593
7594
7595
7596
7597
7598
7599
7600
7601
7602
7603
7604
7605
7606
7607
7608
7609
7610
7611
7612
7613
7614
7615
7616
7617
7618
7619
7620
7621
7622
7623
7624
7625
7626
7627
7628
7629
7630
7631
7632
7633
7634
7635
7636
7637
7638
7639
7640
7641
7642
7643
7644
7645
7646
7647
7648
7649
7650
7651
7652
7653
7654
7655
7656
7657
7658
7659
7660
7661
7662
7663
7664
7665
7666
7667
7668
7669
7670
7671
7672
7673
7674
7675
7676
7677
7678
7679
7680
7681
7682
7683
7684
7685
7686
7687
7688
7689
7690
7691
7692
7693
7694
7695
7696
7697
7698
7699
7700
7701
7702
7703
7704
7705
7706
7707
7708
7709
7710
7711
7712
7713
7714
7715
7716
7717
7718
7719
7720
7721
7722
7723
7724
7725
7726
7727
7728
7729
7730
7731
7732
7733
7734
7735
7736
7737
7738
7739
7740
7741
7742
7743
7744
7745
7746
7747
7748
7749
7750
7751
7752
7753
7754
7755
7756
7757
7758
7759
7760
7761
7762
7763
7764
7765
7766
7767
7768
7769
7770
7771
7772
7773
7774
7775
7776
7777
7778
7779
7780
7781
7782
7783
7784
7785
7786
7787
7788
7789
7790
7791
7792
7793
7794
7795
7796
7797
7798
7799
7800
7801
7802
7803
7804
7805
7806
7807
7808
7809
7810
7811
7812
7813
7814
7815
/*
 * CDDL HEADER START
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License (the "License").
 * You may not use this file except in compliance with the License.
 *
 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
 * or http://www.opensolaris.org/os/licensing.
 * See the License for the specific language governing permissions
 * and limitations under the License.
 *
 * When distributing Covered Code, include this CDDL HEADER in each
 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
 * If applicable, add the following below this CDDL HEADER, with the
 * fields enclosed by brackets "[]" replaced with your own identifying
 * information: Portions Copyright [yyyy] [name of copyright owner]
 *
 * CDDL HEADER END
 */
/*
 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
 * Copyright (c) 2012, Joyent, Inc. All rights reserved.
 * Copyright (c) 2011, 2017 by Delphix. All rights reserved.
 * Copyright (c) 2014 by Saso Kiselkov. All rights reserved.
 * Copyright 2015 Nexenta Systems, Inc.  All rights reserved.
 */

/*
 * DVA-based Adjustable Replacement Cache
 *
 * While much of the theory of operation used here is
 * based on the self-tuning, low overhead replacement cache
 * presented by Megiddo and Modha at FAST 2003, there are some
 * significant differences:
 *
 * 1. The Megiddo and Modha model assumes any page is evictable.
 * Pages in its cache cannot be "locked" into memory.  This makes
 * the eviction algorithm simple: evict the last page in the list.
 * This also make the performance characteristics easy to reason
 * about.  Our cache is not so simple.  At any given moment, some
 * subset of the blocks in the cache are un-evictable because we
 * have handed out a reference to them.  Blocks are only evictable
 * when there are no external references active.  This makes
 * eviction far more problematic:  we choose to evict the evictable
 * blocks that are the "lowest" in the list.
 *
 * There are times when it is not possible to evict the requested
 * space.  In these circumstances we are unable to adjust the cache
 * size.  To prevent the cache growing unbounded at these times we
 * implement a "cache throttle" that slows the flow of new data
 * into the cache until we can make space available.
 *
 * 2. The Megiddo and Modha model assumes a fixed cache size.
 * Pages are evicted when the cache is full and there is a cache
 * miss.  Our model has a variable sized cache.  It grows with
 * high use, but also tries to react to memory pressure from the
 * operating system: decreasing its size when system memory is
 * tight.
 *
 * 3. The Megiddo and Modha model assumes a fixed page size. All
 * elements of the cache are therefore exactly the same size.  So
 * when adjusting the cache size following a cache miss, its simply
 * a matter of choosing a single page to evict.  In our model, we
 * have variable sized cache blocks (rangeing from 512 bytes to
 * 128K bytes).  We therefore choose a set of blocks to evict to make
 * space for a cache miss that approximates as closely as possible
 * the space used by the new block.
 *
 * See also:  "ARC: A Self-Tuning, Low Overhead Replacement Cache"
 * by N. Megiddo & D. Modha, FAST 2003
 */

/*
 * The locking model:
 *
 * A new reference to a cache buffer can be obtained in two
 * ways: 1) via a hash table lookup using the DVA as a key,
 * or 2) via one of the ARC lists.  The arc_read() interface
 * uses method 1, while the internal ARC algorithms for
 * adjusting the cache use method 2.  We therefore provide two
 * types of locks: 1) the hash table lock array, and 2) the
 * ARC list locks.
 *
 * Buffers do not have their own mutexes, rather they rely on the
 * hash table mutexes for the bulk of their protection (i.e. most
 * fields in the arc_buf_hdr_t are protected by these mutexes).
 *
 * buf_hash_find() returns the appropriate mutex (held) when it
 * locates the requested buffer in the hash table.  It returns
 * NULL for the mutex if the buffer was not in the table.
 *
 * buf_hash_remove() expects the appropriate hash mutex to be
 * already held before it is invoked.
 *
 * Each ARC state also has a mutex which is used to protect the
 * buffer list associated with the state.  When attempting to
 * obtain a hash table lock while holding an ARC list lock you
 * must use: mutex_tryenter() to avoid deadlock.  Also note that
 * the active state mutex must be held before the ghost state mutex.
 *
 * It as also possible to register a callback which is run when the
 * arc_meta_limit is reached and no buffers can be safely evicted.  In
 * this case the arc user should drop a reference on some arc buffers so
 * they can be reclaimed and the arc_meta_limit honored.  For example,
 * when using the ZPL each dentry holds a references on a znode.  These
 * dentries must be pruned before the arc buffer holding the znode can
 * be safely evicted.
 *
 * Note that the majority of the performance stats are manipulated
 * with atomic operations.
 *
 * The L2ARC uses the l2ad_mtx on each vdev for the following:
 *
 *	- L2ARC buflist creation
 *	- L2ARC buflist eviction
 *	- L2ARC write completion, which walks L2ARC buflists
 *	- ARC header destruction, as it removes from L2ARC buflists
 *	- ARC header release, as it removes from L2ARC buflists
 */

/*
 * ARC operation:
 *
 * Every block that is in the ARC is tracked by an arc_buf_hdr_t structure.
 * This structure can point either to a block that is still in the cache or to
 * one that is only accessible in an L2 ARC device, or it can provide
 * information about a block that was recently evicted. If a block is
 * only accessible in the L2ARC, then the arc_buf_hdr_t only has enough
 * information to retrieve it from the L2ARC device. This information is
 * stored in the l2arc_buf_hdr_t sub-structure of the arc_buf_hdr_t. A block
 * that is in this state cannot access the data directly.
 *
 * Blocks that are actively being referenced or have not been evicted
 * are cached in the L1ARC. The L1ARC (l1arc_buf_hdr_t) is a structure within
 * the arc_buf_hdr_t that will point to the data block in memory. A block can
 * only be read by a consumer if it has an l1arc_buf_hdr_t. The L1ARC
 * caches data in two ways -- in a list of ARC buffers (arc_buf_t) and
 * also in the arc_buf_hdr_t's private physical data block pointer (b_pabd).
 *
 * The L1ARC's data pointer may or may not be uncompressed. The ARC has the
 * ability to store the physical data (b_pabd) associated with the DVA of the
 * arc_buf_hdr_t. Since the b_pabd is a copy of the on-disk physical block,
 * it will match its on-disk compression characteristics. This behavior can be
 * disabled by setting 'zfs_compressed_arc_enabled' to B_FALSE. When the
 * compressed ARC functionality is disabled, the b_pabd will point to an
 * uncompressed version of the on-disk data.
 *
 * Data in the L1ARC is not accessed by consumers of the ARC directly. Each
 * arc_buf_hdr_t can have multiple ARC buffers (arc_buf_t) which reference it.
 * Each ARC buffer (arc_buf_t) is being actively accessed by a specific ARC
 * consumer. The ARC will provide references to this data and will keep it
 * cached until it is no longer in use. The ARC caches only the L1ARC's physical
 * data block and will evict any arc_buf_t that is no longer referenced. The
 * amount of memory consumed by the arc_buf_ts' data buffers can be seen via the
 * "overhead_size" kstat.
 *
 * Depending on the consumer, an arc_buf_t can be requested in uncompressed or
 * compressed form. The typical case is that consumers will want uncompressed
 * data, and when that happens a new data buffer is allocated where the data is
 * decompressed for them to use. Currently the only consumer who wants
 * compressed arc_buf_t's is "zfs send", when it streams data exactly as it
 * exists on disk. When this happens, the arc_buf_t's data buffer is shared
 * with the arc_buf_hdr_t.
 *
 * Here is a diagram showing an arc_buf_hdr_t referenced by two arc_buf_t's. The
 * first one is owned by a compressed send consumer (and therefore references
 * the same compressed data buffer as the arc_buf_hdr_t) and the second could be
 * used by any other consumer (and has its own uncompressed copy of the data
 * buffer).
 *
 *   arc_buf_hdr_t
 *   +-----------+
 *   | fields    |
 *   | common to |
 *   | L1- and   |
 *   | L2ARC     |
 *   +-----------+
 *   | l2arc_buf_hdr_t
 *   |           |
 *   +-----------+
 *   | l1arc_buf_hdr_t
 *   |           |              arc_buf_t
 *   | b_buf     +------------>+-----------+      arc_buf_t
 *   | b_pabd    +-+           |b_next     +---->+-----------+
 *   +-----------+ |           |-----------|     |b_next     +-->NULL
 *                 |           |b_comp = T |     +-----------+
 *                 |           |b_data     +-+   |b_comp = F |
 *                 |           +-----------+ |   |b_data     +-+
 *                 +->+------+               |   +-----------+ |
 *        compressed  |      |               |                 |
 *           data     |      |<--------------+                 | uncompressed
 *                    +------+          compressed,            |     data
 *                                        shared               +-->+------+
 *                                         data                    |      |
 *                                                                 |      |
 *                                                                 +------+
 *
 * When a consumer reads a block, the ARC must first look to see if the
 * arc_buf_hdr_t is cached. If the hdr is cached then the ARC allocates a new
 * arc_buf_t and either copies uncompressed data into a new data buffer from an
 * existing uncompressed arc_buf_t, decompresses the hdr's b_pabd buffer into a
 * new data buffer, or shares the hdr's b_pabd buffer, depending on whether the
 * hdr is compressed and the desired compression characteristics of the
 * arc_buf_t consumer. If the arc_buf_t ends up sharing data with the
 * arc_buf_hdr_t and both of them are uncompressed then the arc_buf_t must be
 * the last buffer in the hdr's b_buf list, however a shared compressed buf can
 * be anywhere in the hdr's list.
 *
 * The diagram below shows an example of an uncompressed ARC hdr that is
 * sharing its data with an arc_buf_t (note that the shared uncompressed buf is
 * the last element in the buf list):
 *
 *                arc_buf_hdr_t
 *                +-----------+
 *                |           |
 *                |           |
 *                |           |
 *                +-----------+
 * l2arc_buf_hdr_t|           |
 *                |           |
 *                +-----------+
 * l1arc_buf_hdr_t|           |
 *                |           |                 arc_buf_t    (shared)
 *                |    b_buf  +------------>+---------+      arc_buf_t
 *                |           |             |b_next   +---->+---------+
 *                |  b_pabd   +-+           |---------|     |b_next   +-->NULL
 *                +-----------+ |           |         |     +---------+
 *                              |           |b_data   +-+   |         |
 *                              |           +---------+ |   |b_data   +-+
 *                              +->+------+             |   +---------+ |
 *                                 |      |             |               |
 *                   uncompressed  |      |             |               |
 *                        data     +------+             |               |
 *                                    ^                 +->+------+     |
 *                                    |       uncompressed |      |     |
 *                                    |           data     |      |     |
 *                                    |                    +------+     |
 *                                    +---------------------------------+
 *
 * Writing to the ARC requires that the ARC first discard the hdr's b_pabd
 * since the physical block is about to be rewritten. The new data contents
 * will be contained in the arc_buf_t. As the I/O pipeline performs the write,
 * it may compress the data before writing it to disk. The ARC will be called
 * with the transformed data and will bcopy the transformed on-disk block into
 * a newly allocated b_pabd. Writes are always done into buffers which have
 * either been loaned (and hence are new and don't have other readers) or
 * buffers which have been released (and hence have their own hdr, if there
 * were originally other readers of the buf's original hdr). This ensures that
 * the ARC only needs to update a single buf and its hdr after a write occurs.
 *
 * When the L2ARC is in use, it will also take advantage of the b_pabd. The
 * L2ARC will always write the contents of b_pabd to the L2ARC. This means
 * that when compressed ARC is enabled that the L2ARC blocks are identical
 * to the on-disk block in the main data pool. This provides a significant
 * advantage since the ARC can leverage the bp's checksum when reading from the
 * L2ARC to determine if the contents are valid. However, if the compressed
 * ARC is disabled, then the L2ARC's block must be transformed to look
 * like the physical block in the main data pool before comparing the
 * checksum and determining its validity.
 */

#include <sys/spa.h>
#include <sys/zio.h>
#include <sys/spa_impl.h>
#include <sys/zio_compress.h>
#include <sys/zio_checksum.h>
#include <sys/zfs_context.h>
#include <sys/arc.h>
#include <sys/refcount.h>
#include <sys/vdev.h>
#include <sys/vdev_impl.h>
#include <sys/dsl_pool.h>
#include <sys/zio_checksum.h>
#include <sys/multilist.h>
#include <sys/abd.h>
#ifdef _KERNEL
#include <sys/vmsystm.h>
#include <vm/anon.h>
#include <sys/fs/swapnode.h>
#include <sys/zpl.h>
#include <linux/mm_compat.h>
#endif
#include <sys/callb.h>
#include <sys/kstat.h>
#include <sys/dmu_tx.h>
#include <zfs_fletcher.h>
#include <sys/arc_impl.h>
#include <sys/trace_arc.h>

#ifndef _KERNEL
/* set with ZFS_DEBUG=watch, to enable watchpoints on frozen buffers */
boolean_t arc_watch = B_FALSE;
#endif

static kmutex_t		arc_reclaim_lock;
static kcondvar_t	arc_reclaim_thread_cv;
static boolean_t	arc_reclaim_thread_exit;
static kcondvar_t	arc_reclaim_waiters_cv;

/*
 * The number of headers to evict in arc_evict_state_impl() before
 * dropping the sublist lock and evicting from another sublist. A lower
 * value means we're more likely to evict the "correct" header (i.e. the
 * oldest header in the arc state), but comes with higher overhead
 * (i.e. more invocations of arc_evict_state_impl()).
 */
int zfs_arc_evict_batch_limit = 10;

/* number of seconds before growing cache again */
static int		arc_grow_retry = 5;

/* shift of arc_c for calculating overflow limit in arc_get_data_impl */
int		zfs_arc_overflow_shift = 8;

/* shift of arc_c for calculating both min and max arc_p */
static int		arc_p_min_shift = 4;

/* log2(fraction of arc to reclaim) */
static int		arc_shrink_shift = 7;

/*
 * log2(fraction of ARC which must be free to allow growing).
 * I.e. If there is less than arc_c >> arc_no_grow_shift free memory,
 * when reading a new block into the ARC, we will evict an equal-sized block
 * from the ARC.
 *
 * This must be less than arc_shrink_shift, so that when we shrink the ARC,
 * we will still not allow it to grow.
 */
int			arc_no_grow_shift = 5;


/*
 * minimum lifespan of a prefetch block in clock ticks
 * (initialized in arc_init())
 */
static int		arc_min_prefetch_lifespan;

/*
 * If this percent of memory is free, don't throttle.
 */
int arc_lotsfree_percent = 10;

static int arc_dead;

/*
 * The arc has filled available memory and has now warmed up.
 */
static boolean_t arc_warm;

/*
 * log2 fraction of the zio arena to keep free.
 */
int arc_zio_arena_free_shift = 2;

/*
 * These tunables are for performance analysis.
 */
unsigned long zfs_arc_max = 0;
unsigned long zfs_arc_min = 0;
unsigned long zfs_arc_meta_limit = 0;
unsigned long zfs_arc_meta_min = 0;
unsigned long zfs_arc_dnode_limit = 0;
unsigned long zfs_arc_dnode_reduce_percent = 10;
int zfs_arc_grow_retry = 0;
int zfs_arc_shrink_shift = 0;
int zfs_arc_p_min_shift = 0;
int zfs_arc_average_blocksize = 8 * 1024; /* 8KB */

int zfs_compressed_arc_enabled = B_TRUE;

/*
 * ARC will evict meta buffers that exceed arc_meta_limit. This
 * tunable make arc_meta_limit adjustable for different workloads.
 */
unsigned long zfs_arc_meta_limit_percent = 75;

/*
 * Percentage that can be consumed by dnodes of ARC meta buffers.
 */
unsigned long zfs_arc_dnode_limit_percent = 10;

/*
 * These tunables are Linux specific
 */
unsigned long zfs_arc_sys_free = 0;
int zfs_arc_min_prefetch_lifespan = 0;
int zfs_arc_p_aggressive_disable = 1;
int zfs_arc_p_dampener_disable = 1;
int zfs_arc_meta_prune = 10000;
int zfs_arc_meta_strategy = ARC_STRATEGY_META_BALANCED;
int zfs_arc_meta_adjust_restarts = 4096;
int zfs_arc_lotsfree_percent = 10;

/* The 6 states: */
static arc_state_t ARC_anon;
static arc_state_t ARC_mru;
static arc_state_t ARC_mru_ghost;
static arc_state_t ARC_mfu;
static arc_state_t ARC_mfu_ghost;
static arc_state_t ARC_l2c_only;

typedef struct arc_stats {
	kstat_named_t arcstat_hits;
	kstat_named_t arcstat_misses;
	kstat_named_t arcstat_demand_data_hits;
	kstat_named_t arcstat_demand_data_misses;
	kstat_named_t arcstat_demand_metadata_hits;
	kstat_named_t arcstat_demand_metadata_misses;
	kstat_named_t arcstat_prefetch_data_hits;
	kstat_named_t arcstat_prefetch_data_misses;
	kstat_named_t arcstat_prefetch_metadata_hits;
	kstat_named_t arcstat_prefetch_metadata_misses;
	kstat_named_t arcstat_mru_hits;
	kstat_named_t arcstat_mru_ghost_hits;
	kstat_named_t arcstat_mfu_hits;
	kstat_named_t arcstat_mfu_ghost_hits;
	kstat_named_t arcstat_deleted;
	/*
	 * Number of buffers that could not be evicted because the hash lock
	 * was held by another thread.  The lock may not necessarily be held
	 * by something using the same buffer, since hash locks are shared
	 * by multiple buffers.
	 */
	kstat_named_t arcstat_mutex_miss;
	/*
	 * Number of buffers skipped because they have I/O in progress, are
	 * indrect prefetch buffers that have not lived long enough, or are
	 * not from the spa we're trying to evict from.
	 */
	kstat_named_t arcstat_evict_skip;
	/*
	 * Number of times arc_evict_state() was unable to evict enough
	 * buffers to reach its target amount.
	 */
	kstat_named_t arcstat_evict_not_enough;
	kstat_named_t arcstat_evict_l2_cached;
	kstat_named_t arcstat_evict_l2_eligible;
	kstat_named_t arcstat_evict_l2_ineligible;
	kstat_named_t arcstat_evict_l2_skip;
	kstat_named_t arcstat_hash_elements;
	kstat_named_t arcstat_hash_elements_max;
	kstat_named_t arcstat_hash_collisions;
	kstat_named_t arcstat_hash_chains;
	kstat_named_t arcstat_hash_chain_max;
	kstat_named_t arcstat_p;
	kstat_named_t arcstat_c;
	kstat_named_t arcstat_c_min;
	kstat_named_t arcstat_c_max;
	kstat_named_t arcstat_size;
	/*
	 * Number of compressed bytes stored in the arc_buf_hdr_t's b_pabd.
	 * Note that the compressed bytes may match the uncompressed bytes
	 * if the block is either not compressed or compressed arc is disabled.
	 */
	kstat_named_t arcstat_compressed_size;
	/*
	 * Uncompressed size of the data stored in b_pabd. If compressed
	 * arc is disabled then this value will be identical to the stat
	 * above.
	 */
	kstat_named_t arcstat_uncompressed_size;
	/*
	 * Number of bytes stored in all the arc_buf_t's. This is classified
	 * as "overhead" since this data is typically short-lived and will
	 * be evicted from the arc when it becomes unreferenced unless the
	 * zfs_keep_uncompressed_metadata or zfs_keep_uncompressed_level
	 * values have been set (see comment in dbuf.c for more information).
	 */
	kstat_named_t arcstat_overhead_size;
	/*
	 * Number of bytes consumed by internal ARC structures necessary
	 * for tracking purposes; these structures are not actually
	 * backed by ARC buffers. This includes arc_buf_hdr_t structures
	 * (allocated via arc_buf_hdr_t_full and arc_buf_hdr_t_l2only
	 * caches), and arc_buf_t structures (allocated via arc_buf_t
	 * cache).
	 */
	kstat_named_t arcstat_hdr_size;
	/*
	 * Number of bytes consumed by ARC buffers of type equal to
	 * ARC_BUFC_DATA. This is generally consumed by buffers backing
	 * on disk user data (e.g. plain file contents).
	 */
	kstat_named_t arcstat_data_size;
	/*
	 * Number of bytes consumed by ARC buffers of type equal to
	 * ARC_BUFC_METADATA. This is generally consumed by buffers
	 * backing on disk data that is used for internal ZFS
	 * structures (e.g. ZAP, dnode, indirect blocks, etc).
	 */
	kstat_named_t arcstat_metadata_size;
	/*
	 * Number of bytes consumed by dmu_buf_impl_t objects.
	 */
	kstat_named_t arcstat_dbuf_size;
	/*
	 * Number of bytes consumed by dnode_t objects.
	 */
	kstat_named_t arcstat_dnode_size;
	/*
	 * Number of bytes consumed by bonus buffers.
	 */
	kstat_named_t arcstat_bonus_size;
	/*
	 * Total number of bytes consumed by ARC buffers residing in the
	 * arc_anon state. This includes *all* buffers in the arc_anon
	 * state; e.g. data, metadata, evictable, and unevictable buffers
	 * are all included in this value.
	 */
	kstat_named_t arcstat_anon_size;
	/*
	 * Number of bytes consumed by ARC buffers that meet the
	 * following criteria: backing buffers of type ARC_BUFC_DATA,
	 * residing in the arc_anon state, and are eligible for eviction
	 * (e.g. have no outstanding holds on the buffer).
	 */
	kstat_named_t arcstat_anon_evictable_data;
	/*
	 * Number of bytes consumed by ARC buffers that meet the
	 * following criteria: backing buffers of type ARC_BUFC_METADATA,
	 * residing in the arc_anon state, and are eligible for eviction
	 * (e.g. have no outstanding holds on the buffer).
	 */
	kstat_named_t arcstat_anon_evictable_metadata;
	/*
	 * Total number of bytes consumed by ARC buffers residing in the
	 * arc_mru state. This includes *all* buffers in the arc_mru
	 * state; e.g. data, metadata, evictable, and unevictable buffers
	 * are all included in this value.
	 */
	kstat_named_t arcstat_mru_size;
	/*
	 * Number of bytes consumed by ARC buffers that meet the
	 * following criteria: backing buffers of type ARC_BUFC_DATA,
	 * residing in the arc_mru state, and are eligible for eviction
	 * (e.g. have no outstanding holds on the buffer).
	 */
	kstat_named_t arcstat_mru_evictable_data;
	/*
	 * Number of bytes consumed by ARC buffers that meet the
	 * following criteria: backing buffers of type ARC_BUFC_METADATA,
	 * residing in the arc_mru state, and are eligible for eviction
	 * (e.g. have no outstanding holds on the buffer).
	 */
	kstat_named_t arcstat_mru_evictable_metadata;
	/*
	 * Total number of bytes that *would have been* consumed by ARC
	 * buffers in the arc_mru_ghost state. The key thing to note
	 * here, is the fact that this size doesn't actually indicate
	 * RAM consumption. The ghost lists only consist of headers and
	 * don't actually have ARC buffers linked off of these headers.
	 * Thus, *if* the headers had associated ARC buffers, these
	 * buffers *would have* consumed this number of bytes.
	 */
	kstat_named_t arcstat_mru_ghost_size;
	/*
	 * Number of bytes that *would have been* consumed by ARC
	 * buffers that are eligible for eviction, of type
	 * ARC_BUFC_DATA, and linked off the arc_mru_ghost state.
	 */
	kstat_named_t arcstat_mru_ghost_evictable_data;
	/*
	 * Number of bytes that *would have been* consumed by ARC
	 * buffers that are eligible for eviction, of type
	 * ARC_BUFC_METADATA, and linked off the arc_mru_ghost state.
	 */
	kstat_named_t arcstat_mru_ghost_evictable_metadata;
	/*
	 * Total number of bytes consumed by ARC buffers residing in the
	 * arc_mfu state. This includes *all* buffers in the arc_mfu
	 * state; e.g. data, metadata, evictable, and unevictable buffers
	 * are all included in this value.
	 */
	kstat_named_t arcstat_mfu_size;
	/*
	 * Number of bytes consumed by ARC buffers that are eligible for
	 * eviction, of type ARC_BUFC_DATA, and reside in the arc_mfu
	 * state.
	 */
	kstat_named_t arcstat_mfu_evictable_data;
	/*
	 * Number of bytes consumed by ARC buffers that are eligible for
	 * eviction, of type ARC_BUFC_METADATA, and reside in the
	 * arc_mfu state.
	 */
	kstat_named_t arcstat_mfu_evictable_metadata;
	/*
	 * Total number of bytes that *would have been* consumed by ARC
	 * buffers in the arc_mfu_ghost state. See the comment above
	 * arcstat_mru_ghost_size for more details.
	 */
	kstat_named_t arcstat_mfu_ghost_size;
	/*
	 * Number of bytes that *would have been* consumed by ARC
	 * buffers that are eligible for eviction, of type
	 * ARC_BUFC_DATA, and linked off the arc_mfu_ghost state.
	 */
	kstat_named_t arcstat_mfu_ghost_evictable_data;
	/*
	 * Number of bytes that *would have been* consumed by ARC
	 * buffers that are eligible for eviction, of type
	 * ARC_BUFC_METADATA, and linked off the arc_mru_ghost state.
	 */
	kstat_named_t arcstat_mfu_ghost_evictable_metadata;
	kstat_named_t arcstat_l2_hits;
	kstat_named_t arcstat_l2_misses;
	kstat_named_t arcstat_l2_feeds;
	kstat_named_t arcstat_l2_rw_clash;
	kstat_named_t arcstat_l2_read_bytes;
	kstat_named_t arcstat_l2_write_bytes;
	kstat_named_t arcstat_l2_writes_sent;
	kstat_named_t arcstat_l2_writes_done;
	kstat_named_t arcstat_l2_writes_error;
	kstat_named_t arcstat_l2_writes_lock_retry;
	kstat_named_t arcstat_l2_evict_lock_retry;
	kstat_named_t arcstat_l2_evict_reading;
	kstat_named_t arcstat_l2_evict_l1cached;
	kstat_named_t arcstat_l2_free_on_write;
	kstat_named_t arcstat_l2_abort_lowmem;
	kstat_named_t arcstat_l2_cksum_bad;
	kstat_named_t arcstat_l2_io_error;
	kstat_named_t arcstat_l2_size;
	kstat_named_t arcstat_l2_asize;
	kstat_named_t arcstat_l2_hdr_size;
	kstat_named_t arcstat_memory_throttle_count;
	kstat_named_t arcstat_memory_direct_count;
	kstat_named_t arcstat_memory_indirect_count;
	kstat_named_t arcstat_no_grow;
	kstat_named_t arcstat_tempreserve;
	kstat_named_t arcstat_loaned_bytes;
	kstat_named_t arcstat_prune;
	kstat_named_t arcstat_meta_used;
	kstat_named_t arcstat_meta_limit;
	kstat_named_t arcstat_dnode_limit;
	kstat_named_t arcstat_meta_max;
	kstat_named_t arcstat_meta_min;
	kstat_named_t arcstat_sync_wait_for_async;
	kstat_named_t arcstat_demand_hit_predictive_prefetch;
	kstat_named_t arcstat_need_free;
	kstat_named_t arcstat_sys_free;
} arc_stats_t;

static arc_stats_t arc_stats = {
	{ "hits",			KSTAT_DATA_UINT64 },
	{ "misses",			KSTAT_DATA_UINT64 },
	{ "demand_data_hits",		KSTAT_DATA_UINT64 },
	{ "demand_data_misses",		KSTAT_DATA_UINT64 },
	{ "demand_metadata_hits",	KSTAT_DATA_UINT64 },
	{ "demand_metadata_misses",	KSTAT_DATA_UINT64 },
	{ "prefetch_data_hits",		KSTAT_DATA_UINT64 },
	{ "prefetch_data_misses",	KSTAT_DATA_UINT64 },
	{ "prefetch_metadata_hits",	KSTAT_DATA_UINT64 },
	{ "prefetch_metadata_misses",	KSTAT_DATA_UINT64 },
	{ "mru_hits",			KSTAT_DATA_UINT64 },
	{ "mru_ghost_hits",		KSTAT_DATA_UINT64 },
	{ "mfu_hits",			KSTAT_DATA_UINT64 },
	{ "mfu_ghost_hits",		KSTAT_DATA_UINT64 },
	{ "deleted",			KSTAT_DATA_UINT64 },
	{ "mutex_miss",			KSTAT_DATA_UINT64 },
	{ "evict_skip",			KSTAT_DATA_UINT64 },
	{ "evict_not_enough",		KSTAT_DATA_UINT64 },
	{ "evict_l2_cached",		KSTAT_DATA_UINT64 },
	{ "evict_l2_eligible",		KSTAT_DATA_UINT64 },
	{ "evict_l2_ineligible",	KSTAT_DATA_UINT64 },
	{ "evict_l2_skip",		KSTAT_DATA_UINT64 },
	{ "hash_elements",		KSTAT_DATA_UINT64 },
	{ "hash_elements_max",		KSTAT_DATA_UINT64 },
	{ "hash_collisions",		KSTAT_DATA_UINT64 },
	{ "hash_chains",		KSTAT_DATA_UINT64 },
	{ "hash_chain_max",		KSTAT_DATA_UINT64 },
	{ "p",				KSTAT_DATA_UINT64 },
	{ "c",				KSTAT_DATA_UINT64 },
	{ "c_min",			KSTAT_DATA_UINT64 },
	{ "c_max",			KSTAT_DATA_UINT64 },
	{ "size",			KSTAT_DATA_UINT64 },
	{ "compressed_size",		KSTAT_DATA_UINT64 },
	{ "uncompressed_size",		KSTAT_DATA_UINT64 },
	{ "overhead_size",		KSTAT_DATA_UINT64 },
	{ "hdr_size",			KSTAT_DATA_UINT64 },
	{ "data_size",			KSTAT_DATA_UINT64 },
	{ "metadata_size",		KSTAT_DATA_UINT64 },
	{ "dbuf_size",			KSTAT_DATA_UINT64 },
	{ "dnode_size",			KSTAT_DATA_UINT64 },
	{ "bonus_size",			KSTAT_DATA_UINT64 },
	{ "anon_size",			KSTAT_DATA_UINT64 },
	{ "anon_evictable_data",	KSTAT_DATA_UINT64 },
	{ "anon_evictable_metadata",	KSTAT_DATA_UINT64 },
	{ "mru_size",			KSTAT_DATA_UINT64 },
	{ "mru_evictable_data",		KSTAT_DATA_UINT64 },
	{ "mru_evictable_metadata",	KSTAT_DATA_UINT64 },
	{ "mru_ghost_size",		KSTAT_DATA_UINT64 },
	{ "mru_ghost_evictable_data",	KSTAT_DATA_UINT64 },
	{ "mru_ghost_evictable_metadata", KSTAT_DATA_UINT64 },
	{ "mfu_size",			KSTAT_DATA_UINT64 },
	{ "mfu_evictable_data",		KSTAT_DATA_UINT64 },
	{ "mfu_evictable_metadata",	KSTAT_DATA_UINT64 },
	{ "mfu_ghost_size",		KSTAT_DATA_UINT64 },
	{ "mfu_ghost_evictable_data",	KSTAT_DATA_UINT64 },
	{ "mfu_ghost_evictable_metadata", KSTAT_DATA_UINT64 },
	{ "l2_hits",			KSTAT_DATA_UINT64 },
	{ "l2_misses",			KSTAT_DATA_UINT64 },
	{ "l2_feeds",			KSTAT_DATA_UINT64 },
	{ "l2_rw_clash",		KSTAT_DATA_UINT64 },
	{ "l2_read_bytes",		KSTAT_DATA_UINT64 },
	{ "l2_write_bytes",		KSTAT_DATA_UINT64 },
	{ "l2_writes_sent",		KSTAT_DATA_UINT64 },
	{ "l2_writes_done",		KSTAT_DATA_UINT64 },
	{ "l2_writes_error",		KSTAT_DATA_UINT64 },
	{ "l2_writes_lock_retry",	KSTAT_DATA_UINT64 },
	{ "l2_evict_lock_retry",	KSTAT_DATA_UINT64 },
	{ "l2_evict_reading",		KSTAT_DATA_UINT64 },
	{ "l2_evict_l1cached",		KSTAT_DATA_UINT64 },
	{ "l2_free_on_write",		KSTAT_DATA_UINT64 },
	{ "l2_abort_lowmem",		KSTAT_DATA_UINT64 },
	{ "l2_cksum_bad",		KSTAT_DATA_UINT64 },
	{ "l2_io_error",		KSTAT_DATA_UINT64 },
	{ "l2_size",			KSTAT_DATA_UINT64 },
	{ "l2_asize",			KSTAT_DATA_UINT64 },
	{ "l2_hdr_size",		KSTAT_DATA_UINT64 },
	{ "memory_throttle_count",	KSTAT_DATA_UINT64 },
	{ "memory_direct_count",	KSTAT_DATA_UINT64 },
	{ "memory_indirect_count",	KSTAT_DATA_UINT64 },
	{ "arc_no_grow",		KSTAT_DATA_UINT64 },
	{ "arc_tempreserve",		KSTAT_DATA_UINT64 },
	{ "arc_loaned_bytes",		KSTAT_DATA_UINT64 },
	{ "arc_prune",			KSTAT_DATA_UINT64 },
	{ "arc_meta_used",		KSTAT_DATA_UINT64 },
	{ "arc_meta_limit",		KSTAT_DATA_UINT64 },
	{ "arc_dnode_limit",		KSTAT_DATA_UINT64 },
	{ "arc_meta_max",		KSTAT_DATA_UINT64 },
	{ "arc_meta_min",		KSTAT_DATA_UINT64 },
	{ "sync_wait_for_async",	KSTAT_DATA_UINT64 },
	{ "demand_hit_predictive_prefetch", KSTAT_DATA_UINT64 },
	{ "arc_need_free",		KSTAT_DATA_UINT64 },
	{ "arc_sys_free",		KSTAT_DATA_UINT64 }
};

#define	ARCSTAT(stat)	(arc_stats.stat.value.ui64)

#define	ARCSTAT_INCR(stat, val) \
	atomic_add_64(&arc_stats.stat.value.ui64, (val))

#define	ARCSTAT_BUMP(stat)	ARCSTAT_INCR(stat, 1)
#define	ARCSTAT_BUMPDOWN(stat)	ARCSTAT_INCR(stat, -1)

#define	ARCSTAT_MAX(stat, val) {					\
	uint64_t m;							\
	while ((val) > (m = arc_stats.stat.value.ui64) &&		\
	    (m != atomic_cas_64(&arc_stats.stat.value.ui64, m, (val))))	\
		continue;						\
}

#define	ARCSTAT_MAXSTAT(stat) \
	ARCSTAT_MAX(stat##_max, arc_stats.stat.value.ui64)

/*
 * We define a macro to allow ARC hits/misses to be easily broken down by
 * two separate conditions, giving a total of four different subtypes for
 * each of hits and misses (so eight statistics total).
 */
#define	ARCSTAT_CONDSTAT(cond1, stat1, notstat1, cond2, stat2, notstat2, stat) \
	if (cond1) {							\
		if (cond2) {						\
			ARCSTAT_BUMP(arcstat_##stat1##_##stat2##_##stat); \
		} else {						\
			ARCSTAT_BUMP(arcstat_##stat1##_##notstat2##_##stat); \
		}							\
	} else {							\
		if (cond2) {						\
			ARCSTAT_BUMP(arcstat_##notstat1##_##stat2##_##stat); \
		} else {						\
			ARCSTAT_BUMP(arcstat_##notstat1##_##notstat2##_##stat);\
		}							\
	}

kstat_t			*arc_ksp;
static arc_state_t	*arc_anon;
static arc_state_t	*arc_mru;
static arc_state_t	*arc_mru_ghost;
static arc_state_t	*arc_mfu;
static arc_state_t	*arc_mfu_ghost;
static arc_state_t	*arc_l2c_only;

/*
 * There are several ARC variables that are critical to export as kstats --
 * but we don't want to have to grovel around in the kstat whenever we wish to
 * manipulate them.  For these variables, we therefore define them to be in
 * terms of the statistic variable.  This assures that we are not introducing
 * the possibility of inconsistency by having shadow copies of the variables,
 * while still allowing the code to be readable.
 */
#define	arc_size	ARCSTAT(arcstat_size)	/* actual total arc size */
#define	arc_p		ARCSTAT(arcstat_p)	/* target size of MRU */
#define	arc_c		ARCSTAT(arcstat_c)	/* target size of cache */
#define	arc_c_min	ARCSTAT(arcstat_c_min)	/* min target cache size */
#define	arc_c_max	ARCSTAT(arcstat_c_max)	/* max target cache size */
#define	arc_no_grow	ARCSTAT(arcstat_no_grow) /* do not grow cache size */
#define	arc_tempreserve	ARCSTAT(arcstat_tempreserve)
#define	arc_loaned_bytes	ARCSTAT(arcstat_loaned_bytes)
#define	arc_meta_limit	ARCSTAT(arcstat_meta_limit) /* max size for metadata */
#define	arc_dnode_limit	ARCSTAT(arcstat_dnode_limit) /* max size for dnodes */
#define	arc_meta_min	ARCSTAT(arcstat_meta_min) /* min size for metadata */
#define	arc_meta_used	ARCSTAT(arcstat_meta_used) /* size of metadata */
#define	arc_meta_max	ARCSTAT(arcstat_meta_max) /* max size of metadata */
#define	arc_dbuf_size	ARCSTAT(arcstat_dbuf_size) /* dbuf metadata */
#define	arc_dnode_size	ARCSTAT(arcstat_dnode_size) /* dnode metadata */
#define	arc_bonus_size	ARCSTAT(arcstat_bonus_size) /* bonus buffer metadata */
#define	arc_need_free	ARCSTAT(arcstat_need_free) /* bytes to be freed */
#define	arc_sys_free	ARCSTAT(arcstat_sys_free) /* target system free bytes */

/* compressed size of entire arc */
#define	arc_compressed_size	ARCSTAT(arcstat_compressed_size)
/* uncompressed size of entire arc */
#define	arc_uncompressed_size	ARCSTAT(arcstat_uncompressed_size)
/* number of bytes in the arc from arc_buf_t's */
#define	arc_overhead_size	ARCSTAT(arcstat_overhead_size)

static list_t arc_prune_list;
static kmutex_t arc_prune_mtx;
static taskq_t *arc_prune_taskq;

#define	GHOST_STATE(state)	\
	((state) == arc_mru_ghost || (state) == arc_mfu_ghost ||	\
	(state) == arc_l2c_only)

#define	HDR_IN_HASH_TABLE(hdr)	((hdr)->b_flags & ARC_FLAG_IN_HASH_TABLE)
#define	HDR_IO_IN_PROGRESS(hdr)	((hdr)->b_flags & ARC_FLAG_IO_IN_PROGRESS)
#define	HDR_IO_ERROR(hdr)	((hdr)->b_flags & ARC_FLAG_IO_ERROR)
#define	HDR_PREFETCH(hdr)	((hdr)->b_flags & ARC_FLAG_PREFETCH)
#define	HDR_COMPRESSION_ENABLED(hdr)	\
	((hdr)->b_flags & ARC_FLAG_COMPRESSED_ARC)

#define	HDR_L2CACHE(hdr)	((hdr)->b_flags & ARC_FLAG_L2CACHE)
#define	HDR_L2_READING(hdr)	\
	(((hdr)->b_flags & ARC_FLAG_IO_IN_PROGRESS) &&	\
	((hdr)->b_flags & ARC_FLAG_HAS_L2HDR))
#define	HDR_L2_WRITING(hdr)	((hdr)->b_flags & ARC_FLAG_L2_WRITING)
#define	HDR_L2_EVICTED(hdr)	((hdr)->b_flags & ARC_FLAG_L2_EVICTED)
#define	HDR_L2_WRITE_HEAD(hdr)	((hdr)->b_flags & ARC_FLAG_L2_WRITE_HEAD)
#define	HDR_SHARED_DATA(hdr)	((hdr)->b_flags & ARC_FLAG_SHARED_DATA)

#define	HDR_ISTYPE_METADATA(hdr)	\
	((hdr)->b_flags & ARC_FLAG_BUFC_METADATA)
#define	HDR_ISTYPE_DATA(hdr)	(!HDR_ISTYPE_METADATA(hdr))

#define	HDR_HAS_L1HDR(hdr)	((hdr)->b_flags & ARC_FLAG_HAS_L1HDR)
#define	HDR_HAS_L2HDR(hdr)	((hdr)->b_flags & ARC_FLAG_HAS_L2HDR)

/* For storing compression mode in b_flags */
#define	HDR_COMPRESS_OFFSET	(highbit64(ARC_FLAG_COMPRESS_0) - 1)

#define	HDR_GET_COMPRESS(hdr)	((enum zio_compress)BF32_GET((hdr)->b_flags, \
	HDR_COMPRESS_OFFSET, SPA_COMPRESSBITS))
#define	HDR_SET_COMPRESS(hdr, cmp) BF32_SET((hdr)->b_flags, \
	HDR_COMPRESS_OFFSET, SPA_COMPRESSBITS, (cmp));

#define	ARC_BUF_LAST(buf)	((buf)->b_next == NULL)
#define	ARC_BUF_SHARED(buf)	((buf)->b_flags & ARC_BUF_FLAG_SHARED)
#define	ARC_BUF_COMPRESSED(buf)	((buf)->b_flags & ARC_BUF_FLAG_COMPRESSED)

/*
 * Other sizes
 */

#define	HDR_FULL_SIZE ((int64_t)sizeof (arc_buf_hdr_t))
#define	HDR_L2ONLY_SIZE ((int64_t)offsetof(arc_buf_hdr_t, b_l1hdr))

/*
 * Hash table routines
 */

#define	HT_LOCK_ALIGN	64
#define	HT_LOCK_PAD	(P2NPHASE(sizeof (kmutex_t), (HT_LOCK_ALIGN)))

struct ht_lock {
	kmutex_t	ht_lock;
#ifdef _KERNEL
	unsigned char	pad[HT_LOCK_PAD];
#endif
};

#define	BUF_LOCKS 8192
typedef struct buf_hash_table {
	uint64_t ht_mask;
	arc_buf_hdr_t **ht_table;
	struct ht_lock ht_locks[BUF_LOCKS];
} buf_hash_table_t;

static buf_hash_table_t buf_hash_table;

#define	BUF_HASH_INDEX(spa, dva, birth) \
	(buf_hash(spa, dva, birth) & buf_hash_table.ht_mask)
#define	BUF_HASH_LOCK_NTRY(idx) (buf_hash_table.ht_locks[idx & (BUF_LOCKS-1)])
#define	BUF_HASH_LOCK(idx)	(&(BUF_HASH_LOCK_NTRY(idx).ht_lock))
#define	HDR_LOCK(hdr) \
	(BUF_HASH_LOCK(BUF_HASH_INDEX(hdr->b_spa, &hdr->b_dva, hdr->b_birth)))

uint64_t zfs_crc64_table[256];

/*
 * Level 2 ARC
 */

#define	L2ARC_WRITE_SIZE	(8 * 1024 * 1024)	/* initial write max */
#define	L2ARC_HEADROOM		2			/* num of writes */

/*
 * If we discover during ARC scan any buffers to be compressed, we boost
 * our headroom for the next scanning cycle by this percentage multiple.
 */
#define	L2ARC_HEADROOM_BOOST	200
#define	L2ARC_FEED_SECS		1		/* caching interval secs */
#define	L2ARC_FEED_MIN_MS	200		/* min caching interval ms */

/*
 * We can feed L2ARC from two states of ARC buffers, mru and mfu,
 * and each of the state has two types: data and metadata.
 */
#define	L2ARC_FEED_TYPES	4

#define	l2arc_writes_sent	ARCSTAT(arcstat_l2_writes_sent)
#define	l2arc_writes_done	ARCSTAT(arcstat_l2_writes_done)

/* L2ARC Performance Tunables */
unsigned long l2arc_write_max = L2ARC_WRITE_SIZE;	/* def max write size */
unsigned long l2arc_write_boost = L2ARC_WRITE_SIZE;	/* extra warmup write */
unsigned long l2arc_headroom = L2ARC_HEADROOM;		/* # of dev writes */
unsigned long l2arc_headroom_boost = L2ARC_HEADROOM_BOOST;
unsigned long l2arc_feed_secs = L2ARC_FEED_SECS;	/* interval seconds */
unsigned long l2arc_feed_min_ms = L2ARC_FEED_MIN_MS;	/* min interval msecs */
int l2arc_noprefetch = B_TRUE;			/* don't cache prefetch bufs */
int l2arc_feed_again = B_TRUE;			/* turbo warmup */
int l2arc_norw = B_FALSE;			/* no reads during writes */

/*
 * L2ARC Internals
 */
static list_t L2ARC_dev_list;			/* device list */
static list_t *l2arc_dev_list;			/* device list pointer */
static kmutex_t l2arc_dev_mtx;			/* device list mutex */
static l2arc_dev_t *l2arc_dev_last;		/* last device used */
static list_t L2ARC_free_on_write;		/* free after write buf list */
static list_t *l2arc_free_on_write;		/* free after write list ptr */
static kmutex_t l2arc_free_on_write_mtx;	/* mutex for list */
static uint64_t l2arc_ndev;			/* number of devices */

typedef struct l2arc_read_callback {
	arc_buf_hdr_t		*l2rcb_hdr;		/* read header */
	blkptr_t		l2rcb_bp;		/* original blkptr */
	zbookmark_phys_t	l2rcb_zb;		/* original bookmark */
	int			l2rcb_flags;		/* original flags */
} l2arc_read_callback_t;

typedef struct l2arc_data_free {
	/* protected by l2arc_free_on_write_mtx */
	abd_t		*l2df_abd;
	size_t		l2df_size;
	arc_buf_contents_t l2df_type;
	list_node_t	l2df_list_node;
} l2arc_data_free_t;

static kmutex_t l2arc_feed_thr_lock;
static kcondvar_t l2arc_feed_thr_cv;
static uint8_t l2arc_thread_exit;

static abd_t *arc_get_data_abd(arc_buf_hdr_t *, uint64_t, void *);
static void *arc_get_data_buf(arc_buf_hdr_t *, uint64_t, void *);
static void arc_get_data_impl(arc_buf_hdr_t *, uint64_t, void *);
static void arc_free_data_abd(arc_buf_hdr_t *, abd_t *, uint64_t, void *);
static void arc_free_data_buf(arc_buf_hdr_t *, void *, uint64_t, void *);
static void arc_free_data_impl(arc_buf_hdr_t *hdr, uint64_t size, void *tag);
static void arc_hdr_free_pabd(arc_buf_hdr_t *);
static void arc_hdr_alloc_pabd(arc_buf_hdr_t *);
static void arc_access(arc_buf_hdr_t *, kmutex_t *);
static boolean_t arc_is_overflowing(void);
static void arc_buf_watch(arc_buf_t *);
static void arc_tuning_update(void);
static void arc_prune_async(int64_t);
static uint64_t arc_all_memory(void);

static arc_buf_contents_t arc_buf_type(arc_buf_hdr_t *);
static uint32_t arc_bufc_to_flags(arc_buf_contents_t);
static inline void arc_hdr_set_flags(arc_buf_hdr_t *hdr, arc_flags_t flags);
static inline void arc_hdr_clear_flags(arc_buf_hdr_t *hdr, arc_flags_t flags);

static boolean_t l2arc_write_eligible(uint64_t, arc_buf_hdr_t *);
static void l2arc_read_done(zio_t *);

static uint64_t
buf_hash(uint64_t spa, const dva_t *dva, uint64_t birth)
{
	uint8_t *vdva = (uint8_t *)dva;
	uint64_t crc = -1ULL;
	int i;

	ASSERT(zfs_crc64_table[128] == ZFS_CRC64_POLY);

	for (i = 0; i < sizeof (dva_t); i++)
		crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ vdva[i]) & 0xFF];

	crc ^= (spa>>8) ^ birth;

	return (crc);
}

#define	HDR_EMPTY(hdr)						\
	((hdr)->b_dva.dva_word[0] == 0 &&			\
	(hdr)->b_dva.dva_word[1] == 0)

#define	HDR_EQUAL(spa, dva, birth, hdr)				\
	((hdr)->b_dva.dva_word[0] == (dva)->dva_word[0]) &&	\
	((hdr)->b_dva.dva_word[1] == (dva)->dva_word[1]) &&	\
	((hdr)->b_birth == birth) && ((hdr)->b_spa == spa)

static void
buf_discard_identity(arc_buf_hdr_t *hdr)
{
	hdr->b_dva.dva_word[0] = 0;
	hdr->b_dva.dva_word[1] = 0;
	hdr->b_birth = 0;
}

static arc_buf_hdr_t *
buf_hash_find(uint64_t spa, const blkptr_t *bp, kmutex_t **lockp)
{
	const dva_t *dva = BP_IDENTITY(bp);
	uint64_t birth = BP_PHYSICAL_BIRTH(bp);
	uint64_t idx = BUF_HASH_INDEX(spa, dva, birth);
	kmutex_t *hash_lock = BUF_HASH_LOCK(idx);
	arc_buf_hdr_t *hdr;

	mutex_enter(hash_lock);
	for (hdr = buf_hash_table.ht_table[idx]; hdr != NULL;
	    hdr = hdr->b_hash_next) {
		if (HDR_EQUAL(spa, dva, birth, hdr)) {
			*lockp = hash_lock;
			return (hdr);
		}
	}
	mutex_exit(hash_lock);
	*lockp = NULL;
	return (NULL);
}

/*
 * Insert an entry into the hash table.  If there is already an element
 * equal to elem in the hash table, then the already existing element
 * will be returned and the new element will not be inserted.
 * Otherwise returns NULL.
 * If lockp == NULL, the caller is assumed to already hold the hash lock.
 */
static arc_buf_hdr_t *
buf_hash_insert(arc_buf_hdr_t *hdr, kmutex_t **lockp)
{
	uint64_t idx = BUF_HASH_INDEX(hdr->b_spa, &hdr->b_dva, hdr->b_birth);
	kmutex_t *hash_lock = BUF_HASH_LOCK(idx);
	arc_buf_hdr_t *fhdr;
	uint32_t i;

	ASSERT(!DVA_IS_EMPTY(&hdr->b_dva));
	ASSERT(hdr->b_birth != 0);
	ASSERT(!HDR_IN_HASH_TABLE(hdr));

	if (lockp != NULL) {
		*lockp = hash_lock;
		mutex_enter(hash_lock);
	} else {
		ASSERT(MUTEX_HELD(hash_lock));
	}

	for (fhdr = buf_hash_table.ht_table[idx], i = 0; fhdr != NULL;
	    fhdr = fhdr->b_hash_next, i++) {
		if (HDR_EQUAL(hdr->b_spa, &hdr->b_dva, hdr->b_birth, fhdr))
			return (fhdr);
	}

	hdr->b_hash_next = buf_hash_table.ht_table[idx];
	buf_hash_table.ht_table[idx] = hdr;
	arc_hdr_set_flags(hdr, ARC_FLAG_IN_HASH_TABLE);

	/* collect some hash table performance data */
	if (i > 0) {
		ARCSTAT_BUMP(arcstat_hash_collisions);
		if (i == 1)
			ARCSTAT_BUMP(arcstat_hash_chains);

		ARCSTAT_MAX(arcstat_hash_chain_max, i);
	}

	ARCSTAT_BUMP(arcstat_hash_elements);
	ARCSTAT_MAXSTAT(arcstat_hash_elements);

	return (NULL);
}

static void
buf_hash_remove(arc_buf_hdr_t *hdr)
{
	arc_buf_hdr_t *fhdr, **hdrp;
	uint64_t idx = BUF_HASH_INDEX(hdr->b_spa, &hdr->b_dva, hdr->b_birth);

	ASSERT(MUTEX_HELD(BUF_HASH_LOCK(idx)));
	ASSERT(HDR_IN_HASH_TABLE(hdr));

	hdrp = &buf_hash_table.ht_table[idx];
	while ((fhdr = *hdrp) != hdr) {
		ASSERT3P(fhdr, !=, NULL);
		hdrp = &fhdr->b_hash_next;
	}
	*hdrp = hdr->b_hash_next;
	hdr->b_hash_next = NULL;
	arc_hdr_clear_flags(hdr, ARC_FLAG_IN_HASH_TABLE);

	/* collect some hash table performance data */
	ARCSTAT_BUMPDOWN(arcstat_hash_elements);

	if (buf_hash_table.ht_table[idx] &&
	    buf_hash_table.ht_table[idx]->b_hash_next == NULL)
		ARCSTAT_BUMPDOWN(arcstat_hash_chains);
}

/*
 * Global data structures and functions for the buf kmem cache.
 */
static kmem_cache_t *hdr_full_cache;
static kmem_cache_t *hdr_l2only_cache;
static kmem_cache_t *buf_cache;

static void
buf_fini(void)
{
	int i;

#if defined(_KERNEL) && defined(HAVE_SPL)
	/*
	 * Large allocations which do not require contiguous pages
	 * should be using vmem_free() in the linux kernel\
	 */
	vmem_free(buf_hash_table.ht_table,
	    (buf_hash_table.ht_mask + 1) * sizeof (void *));
#else
	kmem_free(buf_hash_table.ht_table,
	    (buf_hash_table.ht_mask + 1) * sizeof (void *));
#endif
	for (i = 0; i < BUF_LOCKS; i++)
		mutex_destroy(&buf_hash_table.ht_locks[i].ht_lock);
	kmem_cache_destroy(hdr_full_cache);
	kmem_cache_destroy(hdr_l2only_cache);
	kmem_cache_destroy(buf_cache);
}

/*
 * Constructor callback - called when the cache is empty
 * and a new buf is requested.
 */
/* ARGSUSED */
static int
hdr_full_cons(void *vbuf, void *unused, int kmflag)
{
	arc_buf_hdr_t *hdr = vbuf;

	bzero(hdr, HDR_FULL_SIZE);
	cv_init(&hdr->b_l1hdr.b_cv, NULL, CV_DEFAULT, NULL);
	refcount_create(&hdr->b_l1hdr.b_refcnt);
	mutex_init(&hdr->b_l1hdr.b_freeze_lock, NULL, MUTEX_DEFAULT, NULL);
	list_link_init(&hdr->b_l1hdr.b_arc_node);
	list_link_init(&hdr->b_l2hdr.b_l2node);
	multilist_link_init(&hdr->b_l1hdr.b_arc_node);
	arc_space_consume(HDR_FULL_SIZE, ARC_SPACE_HDRS);

	return (0);
}

/* ARGSUSED */
static int
hdr_l2only_cons(void *vbuf, void *unused, int kmflag)
{
	arc_buf_hdr_t *hdr = vbuf;

	bzero(hdr, HDR_L2ONLY_SIZE);
	arc_space_consume(HDR_L2ONLY_SIZE, ARC_SPACE_L2HDRS);

	return (0);
}

/* ARGSUSED */
static int
buf_cons(void *vbuf, void *unused, int kmflag)
{
	arc_buf_t *buf = vbuf;

	bzero(buf, sizeof (arc_buf_t));
	mutex_init(&buf->b_evict_lock, NULL, MUTEX_DEFAULT, NULL);
	arc_space_consume(sizeof (arc_buf_t), ARC_SPACE_HDRS);

	return (0);
}

/*
 * Destructor callback - called when a cached buf is
 * no longer required.
 */
/* ARGSUSED */
static void
hdr_full_dest(void *vbuf, void *unused)
{
	arc_buf_hdr_t *hdr = vbuf;

	ASSERT(HDR_EMPTY(hdr));
	cv_destroy(&hdr->b_l1hdr.b_cv);
	refcount_destroy(&hdr->b_l1hdr.b_refcnt);
	mutex_destroy(&hdr->b_l1hdr.b_freeze_lock);
	ASSERT(!multilist_link_active(&hdr->b_l1hdr.b_arc_node));
	arc_space_return(HDR_FULL_SIZE, ARC_SPACE_HDRS);
}

/* ARGSUSED */
static void
hdr_l2only_dest(void *vbuf, void *unused)
{
	ASSERTV(arc_buf_hdr_t *hdr = vbuf);

	ASSERT(HDR_EMPTY(hdr));
	arc_space_return(HDR_L2ONLY_SIZE, ARC_SPACE_L2HDRS);
}

/* ARGSUSED */
static void
buf_dest(void *vbuf, void *unused)
{
	arc_buf_t *buf = vbuf;

	mutex_destroy(&buf->b_evict_lock);
	arc_space_return(sizeof (arc_buf_t), ARC_SPACE_HDRS);
}

/*
 * Reclaim callback -- invoked when memory is low.
 */
/* ARGSUSED */
static void
hdr_recl(void *unused)
{
	dprintf("hdr_recl called\n");
	/*
	 * umem calls the reclaim func when we destroy the buf cache,
	 * which is after we do arc_fini().
	 */
	if (!arc_dead)
		cv_signal(&arc_reclaim_thread_cv);
}

static void
buf_init(void)
{
	uint64_t *ct = NULL;
	uint64_t hsize = 1ULL << 12;
	int i, j;

	/*
	 * The hash table is big enough to fill all of physical memory
	 * with an average block size of zfs_arc_average_blocksize (default 8K).
	 * By default, the table will take up
	 * totalmem * sizeof(void*) / 8K (1MB per GB with 8-byte pointers).
	 */
	while (hsize * zfs_arc_average_blocksize < arc_all_memory())
		hsize <<= 1;
retry:
	buf_hash_table.ht_mask = hsize - 1;
#if defined(_KERNEL) && defined(HAVE_SPL)
	/*
	 * Large allocations which do not require contiguous pages
	 * should be using vmem_alloc() in the linux kernel
	 */
	buf_hash_table.ht_table =
	    vmem_zalloc(hsize * sizeof (void*), KM_SLEEP);
#else
	buf_hash_table.ht_table =
	    kmem_zalloc(hsize * sizeof (void*), KM_NOSLEEP);
#endif
	if (buf_hash_table.ht_table == NULL) {
		ASSERT(hsize > (1ULL << 8));
		hsize >>= 1;
		goto retry;
	}

	hdr_full_cache = kmem_cache_create("arc_buf_hdr_t_full", HDR_FULL_SIZE,
	    0, hdr_full_cons, hdr_full_dest, hdr_recl, NULL, NULL, 0);
	hdr_l2only_cache = kmem_cache_create("arc_buf_hdr_t_l2only",
	    HDR_L2ONLY_SIZE, 0, hdr_l2only_cons, hdr_l2only_dest, hdr_recl,
	    NULL, NULL, 0);
	buf_cache = kmem_cache_create("arc_buf_t", sizeof (arc_buf_t),
	    0, buf_cons, buf_dest, NULL, NULL, NULL, 0);

	for (i = 0; i < 256; i++)
		for (ct = zfs_crc64_table + i, *ct = i, j = 8; j > 0; j--)
			*ct = (*ct >> 1) ^ (-(*ct & 1) & ZFS_CRC64_POLY);

	for (i = 0; i < BUF_LOCKS; i++) {
		mutex_init(&buf_hash_table.ht_locks[i].ht_lock,
		    NULL, MUTEX_DEFAULT, NULL);
	}
}

#define	ARC_MINTIME	(hz>>4) /* 62 ms */

/*
 * This is the size that the buf occupies in memory. If the buf is compressed,
 * it will correspond to the compressed size. You should use this method of
 * getting the buf size unless you explicitly need the logical size.
 */
uint64_t
arc_buf_size(arc_buf_t *buf)
{
	return (ARC_BUF_COMPRESSED(buf) ?
	    HDR_GET_PSIZE(buf->b_hdr) : HDR_GET_LSIZE(buf->b_hdr));
}

uint64_t
arc_buf_lsize(arc_buf_t *buf)
{
	return (HDR_GET_LSIZE(buf->b_hdr));
}

enum zio_compress
arc_get_compression(arc_buf_t *buf)
{
	return (ARC_BUF_COMPRESSED(buf) ?
	    HDR_GET_COMPRESS(buf->b_hdr) : ZIO_COMPRESS_OFF);
}

static inline boolean_t
arc_buf_is_shared(arc_buf_t *buf)
{
	boolean_t shared = (buf->b_data != NULL &&
	    buf->b_hdr->b_l1hdr.b_pabd != NULL &&
	    abd_is_linear(buf->b_hdr->b_l1hdr.b_pabd) &&
	    buf->b_data == abd_to_buf(buf->b_hdr->b_l1hdr.b_pabd));
	IMPLY(shared, HDR_SHARED_DATA(buf->b_hdr));
	IMPLY(shared, ARC_BUF_SHARED(buf));
	IMPLY(shared, ARC_BUF_COMPRESSED(buf) || ARC_BUF_LAST(buf));

	/*
	 * It would be nice to assert arc_can_share() too, but the "hdr isn't
	 * already being shared" requirement prevents us from doing that.
	 */

	return (shared);
}

/*
 * Free the checksum associated with this header. If there is no checksum, this
 * is a no-op.
 */
static inline void
arc_cksum_free(arc_buf_hdr_t *hdr)
{
	ASSERT(HDR_HAS_L1HDR(hdr));
	mutex_enter(&hdr->b_l1hdr.b_freeze_lock);
	if (hdr->b_l1hdr.b_freeze_cksum != NULL) {
		kmem_free(hdr->b_l1hdr.b_freeze_cksum, sizeof (zio_cksum_t));
		hdr->b_l1hdr.b_freeze_cksum = NULL;
	}
	mutex_exit(&hdr->b_l1hdr.b_freeze_lock);
}

/*
 * Return true iff at least one of the bufs on hdr is not compressed.
 */
static boolean_t
arc_hdr_has_uncompressed_buf(arc_buf_hdr_t *hdr)
{
	for (arc_buf_t *b = hdr->b_l1hdr.b_buf; b != NULL; b = b->b_next) {
		if (!ARC_BUF_COMPRESSED(b)) {
			return (B_TRUE);
		}
	}
	return (B_FALSE);
}


/*
 * If we've turned on the ZFS_DEBUG_MODIFY flag, verify that the buf's data
 * matches the checksum that is stored in the hdr. If there is no checksum,
 * or if the buf is compressed, this is a no-op.
 */
static void
arc_cksum_verify(arc_buf_t *buf)
{
	arc_buf_hdr_t *hdr = buf->b_hdr;
	zio_cksum_t zc;

	if (!(zfs_flags & ZFS_DEBUG_MODIFY))
		return;

	if (ARC_BUF_COMPRESSED(buf)) {
		ASSERT(hdr->b_l1hdr.b_freeze_cksum == NULL ||
		    arc_hdr_has_uncompressed_buf(hdr));
		return;
	}

	ASSERT(HDR_HAS_L1HDR(hdr));

	mutex_enter(&hdr->b_l1hdr.b_freeze_lock);
	if (hdr->b_l1hdr.b_freeze_cksum == NULL || HDR_IO_ERROR(hdr)) {
		mutex_exit(&hdr->b_l1hdr.b_freeze_lock);
		return;
	}

	fletcher_2_native(buf->b_data, arc_buf_size(buf), NULL, &zc);
	if (!ZIO_CHECKSUM_EQUAL(*hdr->b_l1hdr.b_freeze_cksum, zc))
		panic("buffer modified while frozen!");
	mutex_exit(&hdr->b_l1hdr.b_freeze_lock);
}

static boolean_t
arc_cksum_is_equal(arc_buf_hdr_t *hdr, zio_t *zio)
{
	enum zio_compress compress = BP_GET_COMPRESS(zio->io_bp);
	boolean_t valid_cksum;

	ASSERT(!BP_IS_EMBEDDED(zio->io_bp));
	VERIFY3U(BP_GET_PSIZE(zio->io_bp), ==, HDR_GET_PSIZE(hdr));

	/*
	 * We rely on the blkptr's checksum to determine if the block
	 * is valid or not. When compressed arc is enabled, the l2arc
	 * writes the block to the l2arc just as it appears in the pool.
	 * This allows us to use the blkptr's checksum to validate the
	 * data that we just read off of the l2arc without having to store
	 * a separate checksum in the arc_buf_hdr_t. However, if compressed
	 * arc is disabled, then the data written to the l2arc is always
	 * uncompressed and won't match the block as it exists in the main
	 * pool. When this is the case, we must first compress it if it is
	 * compressed on the main pool before we can validate the checksum.
	 */
	if (!HDR_COMPRESSION_ENABLED(hdr) && compress != ZIO_COMPRESS_OFF) {
		uint64_t lsize;
		uint64_t csize;
		void *cbuf;
		ASSERT3U(HDR_GET_COMPRESS(hdr), ==, ZIO_COMPRESS_OFF);

		cbuf = zio_buf_alloc(HDR_GET_PSIZE(hdr));
		lsize = HDR_GET_LSIZE(hdr);
		csize = zio_compress_data(compress, zio->io_abd, cbuf, lsize);

		ASSERT3U(csize, <=, HDR_GET_PSIZE(hdr));
		if (csize < HDR_GET_PSIZE(hdr)) {
			/*
			 * Compressed blocks are always a multiple of the
			 * smallest ashift in the pool. Ideally, we would
			 * like to round up the csize to the next
			 * spa_min_ashift but that value may have changed
			 * since the block was last written. Instead,
			 * we rely on the fact that the hdr's psize
			 * was set to the psize of the block when it was
			 * last written. We set the csize to that value
			 * and zero out any part that should not contain
			 * data.
			 */
			bzero((char *)cbuf + csize, HDR_GET_PSIZE(hdr) - csize);
			csize = HDR_GET_PSIZE(hdr);
		}
		zio_push_transform(zio, cbuf, csize, HDR_GET_PSIZE(hdr), NULL);
	}

	/*
	 * Block pointers always store the checksum for the logical data.
	 * If the block pointer has the gang bit set, then the checksum
	 * it represents is for the reconstituted data and not for an
	 * individual gang member. The zio pipeline, however, must be able to
	 * determine the checksum of each of the gang constituents so it
	 * treats the checksum comparison differently than what we need
	 * for l2arc blocks. This prevents us from using the
	 * zio_checksum_error() interface directly. Instead we must call the
	 * zio_checksum_error_impl() so that we can ensure the checksum is
	 * generated using the correct checksum algorithm and accounts for the
	 * logical I/O size and not just a gang fragment.
	 */
	valid_cksum = (zio_checksum_error_impl(zio->io_spa, zio->io_bp,
	    BP_GET_CHECKSUM(zio->io_bp), zio->io_abd, zio->io_size,
	    zio->io_offset, NULL) == 0);
	zio_pop_transforms(zio);
	return (valid_cksum);
}

/*
 * Given a buf full of data, if ZFS_DEBUG_MODIFY is enabled this computes a
 * checksum and attaches it to the buf's hdr so that we can ensure that the buf
 * isn't modified later on. If buf is compressed or there is already a checksum
 * on the hdr, this is a no-op (we only checksum uncompressed bufs).
 */
static void
arc_cksum_compute(arc_buf_t *buf)
{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	if (!(zfs_flags & ZFS_DEBUG_MODIFY))
		return;

	ASSERT(HDR_HAS_L1HDR(hdr));

	mutex_enter(&buf->b_hdr->b_l1hdr.b_freeze_lock);
	if (hdr->b_l1hdr.b_freeze_cksum != NULL) {
		ASSERT(arc_hdr_has_uncompressed_buf(hdr));
		mutex_exit(&hdr->b_l1hdr.b_freeze_lock);
		return;
	} else if (ARC_BUF_COMPRESSED(buf)) {
		mutex_exit(&hdr->b_l1hdr.b_freeze_lock);
		return;
	}

	ASSERT(!ARC_BUF_COMPRESSED(buf));
	hdr->b_l1hdr.b_freeze_cksum = kmem_alloc(sizeof (zio_cksum_t),
	    KM_SLEEP);
	fletcher_2_native(buf->b_data, arc_buf_size(buf), NULL,
	    hdr->b_l1hdr.b_freeze_cksum);
	mutex_exit(&hdr->b_l1hdr.b_freeze_lock);
	arc_buf_watch(buf);
}

#ifndef _KERNEL
void
arc_buf_sigsegv(int sig, siginfo_t *si, void *unused)
{
	panic("Got SIGSEGV at address: 0x%lx\n", (long)si->si_addr);
}
#endif

/* ARGSUSED */
static void
arc_buf_unwatch(arc_buf_t *buf)
{
#ifndef _KERNEL
	if (arc_watch) {
		ASSERT0(mprotect(buf->b_data, arc_buf_size(buf),
		    PROT_READ | PROT_WRITE));
	}
#endif
}

/* ARGSUSED */
static void
arc_buf_watch(arc_buf_t *buf)
{
#ifndef _KERNEL
	if (arc_watch)
		ASSERT0(mprotect(buf->b_data, arc_buf_size(buf),
		    PROT_READ));
#endif
}

static arc_buf_contents_t
arc_buf_type(arc_buf_hdr_t *hdr)
{
	arc_buf_contents_t type;
	if (HDR_ISTYPE_METADATA(hdr)) {
		type = ARC_BUFC_METADATA;
	} else {
		type = ARC_BUFC_DATA;
	}
	VERIFY3U(hdr->b_type, ==, type);
	return (type);
}

boolean_t
arc_is_metadata(arc_buf_t *buf)
{
	return (HDR_ISTYPE_METADATA(buf->b_hdr) != 0);
}

static uint32_t
arc_bufc_to_flags(arc_buf_contents_t type)
{
	switch (type) {
	case ARC_BUFC_DATA:
		/* metadata field is 0 if buffer contains normal data */
		return (0);
	case ARC_BUFC_METADATA:
		return (ARC_FLAG_BUFC_METADATA);
	default:
		break;
	}
	panic("undefined ARC buffer type!");
	return ((uint32_t)-1);
}

void
arc_buf_thaw(arc_buf_t *buf)
{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	ASSERT3P(hdr->b_l1hdr.b_state, ==, arc_anon);
	ASSERT(!HDR_IO_IN_PROGRESS(hdr));

	arc_cksum_verify(buf);

	/*
	 * Compressed buffers do not manipulate the b_freeze_cksum or
	 * allocate b_thawed.
	 */
	if (ARC_BUF_COMPRESSED(buf)) {
		ASSERT(hdr->b_l1hdr.b_freeze_cksum == NULL ||
		    arc_hdr_has_uncompressed_buf(hdr));
		return;
	}

	ASSERT(HDR_HAS_L1HDR(hdr));
	arc_cksum_free(hdr);
	arc_buf_unwatch(buf);
}

void
arc_buf_freeze(arc_buf_t *buf)
{
	arc_buf_hdr_t *hdr = buf->b_hdr;
	kmutex_t *hash_lock;

	if (!(zfs_flags & ZFS_DEBUG_MODIFY))
		return;

	if (ARC_BUF_COMPRESSED(buf)) {
		ASSERT(hdr->b_l1hdr.b_freeze_cksum == NULL ||
		    arc_hdr_has_uncompressed_buf(hdr));
		return;
	}

	hash_lock = HDR_LOCK(hdr);
	mutex_enter(hash_lock);

	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT(hdr->b_l1hdr.b_freeze_cksum != NULL ||
	    hdr->b_l1hdr.b_state == arc_anon);
	arc_cksum_compute(buf);
	mutex_exit(hash_lock);
}

/*
 * The arc_buf_hdr_t's b_flags should never be modified directly. Instead,
 * the following functions should be used to ensure that the flags are
 * updated in a thread-safe way. When manipulating the flags either
 * the hash_lock must be held or the hdr must be undiscoverable. This
 * ensures that we're not racing with any other threads when updating
 * the flags.
 */
static inline void
arc_hdr_set_flags(arc_buf_hdr_t *hdr, arc_flags_t flags)
{
	ASSERT(MUTEX_HELD(HDR_LOCK(hdr)) || HDR_EMPTY(hdr));
	hdr->b_flags |= flags;
}

static inline void
arc_hdr_clear_flags(arc_buf_hdr_t *hdr, arc_flags_t flags)
{
	ASSERT(MUTEX_HELD(HDR_LOCK(hdr)) || HDR_EMPTY(hdr));
	hdr->b_flags &= ~flags;
}

/*
 * Setting the compression bits in the arc_buf_hdr_t's b_flags is
 * done in a special way since we have to clear and set bits
 * at the same time. Consumers that wish to set the compression bits
 * must use this function to ensure that the flags are updated in
 * thread-safe manner.
 */
static void
arc_hdr_set_compress(arc_buf_hdr_t *hdr, enum zio_compress cmp)
{
	ASSERT(MUTEX_HELD(HDR_LOCK(hdr)) || HDR_EMPTY(hdr));

	/*
	 * Holes and embedded blocks will always have a psize = 0 so
	 * we ignore the compression of the blkptr and set the
	 * want to uncompress them. Mark them as uncompressed.
	 */
	if (!zfs_compressed_arc_enabled || HDR_GET_PSIZE(hdr) == 0) {
		arc_hdr_clear_flags(hdr, ARC_FLAG_COMPRESSED_ARC);
		HDR_SET_COMPRESS(hdr, ZIO_COMPRESS_OFF);
		ASSERT(!HDR_COMPRESSION_ENABLED(hdr));
		ASSERT3U(HDR_GET_COMPRESS(hdr), ==, ZIO_COMPRESS_OFF);
	} else {
		arc_hdr_set_flags(hdr, ARC_FLAG_COMPRESSED_ARC);
		HDR_SET_COMPRESS(hdr, cmp);
		ASSERT3U(HDR_GET_COMPRESS(hdr), ==, cmp);
		ASSERT(HDR_COMPRESSION_ENABLED(hdr));
	}
}

/*
 * Looks for another buf on the same hdr which has the data decompressed, copies
 * from it, and returns true. If no such buf exists, returns false.
 */
static boolean_t
arc_buf_try_copy_decompressed_data(arc_buf_t *buf)
{
	arc_buf_hdr_t *hdr = buf->b_hdr;
	boolean_t copied = B_FALSE;

	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT3P(buf->b_data, !=, NULL);
	ASSERT(!ARC_BUF_COMPRESSED(buf));

	for (arc_buf_t *from = hdr->b_l1hdr.b_buf; from != NULL;
	    from = from->b_next) {
		/* can't use our own data buffer */
		if (from == buf) {
			continue;
		}

		if (!ARC_BUF_COMPRESSED(from)) {
			bcopy(from->b_data, buf->b_data, arc_buf_size(buf));
			copied = B_TRUE;
			break;
		}
	}

	/*
	 * There were no decompressed bufs, so there should not be a
	 * checksum on the hdr either.
	 */
	EQUIV(!copied, hdr->b_l1hdr.b_freeze_cksum == NULL);

	return (copied);
}

/*
 * Given a buf that has a data buffer attached to it, this function will
 * efficiently fill the buf with data of the specified compression setting from
 * the hdr and update the hdr's b_freeze_cksum if necessary. If the buf and hdr
 * are already sharing a data buf, no copy is performed.
 *
 * If the buf is marked as compressed but uncompressed data was requested, this
 * will allocate a new data buffer for the buf, remove that flag, and fill the
 * buf with uncompressed data. You can't request a compressed buf on a hdr with
 * uncompressed data, and (since we haven't added support for it yet) if you
 * want compressed data your buf must already be marked as compressed and have
 * the correct-sized data buffer.
 */
static int
arc_buf_fill(arc_buf_t *buf, boolean_t compressed)
{
	arc_buf_hdr_t *hdr = buf->b_hdr;
	boolean_t hdr_compressed = (HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF);
	dmu_object_byteswap_t bswap = hdr->b_l1hdr.b_byteswap;

	ASSERT3P(buf->b_data, !=, NULL);
	IMPLY(compressed, hdr_compressed);
	IMPLY(compressed, ARC_BUF_COMPRESSED(buf));

	if (hdr_compressed == compressed) {
		if (!arc_buf_is_shared(buf)) {
			abd_copy_to_buf(buf->b_data, hdr->b_l1hdr.b_pabd,
			    arc_buf_size(buf));
		}
	} else {
		ASSERT(hdr_compressed);
		ASSERT(!compressed);
		ASSERT3U(HDR_GET_LSIZE(hdr), !=, HDR_GET_PSIZE(hdr));

		/*
		 * If the buf is sharing its data with the hdr, unlink it and
		 * allocate a new data buffer for the buf.
		 */
		if (arc_buf_is_shared(buf)) {
			ASSERT(ARC_BUF_COMPRESSED(buf));

			/* We need to give the buf it's own b_data */
			buf->b_flags &= ~ARC_BUF_FLAG_SHARED;
			buf->b_data =
			    arc_get_data_buf(hdr, HDR_GET_LSIZE(hdr), buf);
			arc_hdr_clear_flags(hdr, ARC_FLAG_SHARED_DATA);

			/* Previously overhead was 0; just add new overhead */
			ARCSTAT_INCR(arcstat_overhead_size, HDR_GET_LSIZE(hdr));
		} else if (ARC_BUF_COMPRESSED(buf)) {
			/* We need to reallocate the buf's b_data */
			arc_free_data_buf(hdr, buf->b_data, HDR_GET_PSIZE(hdr),
			    buf);
			buf->b_data =
			    arc_get_data_buf(hdr, HDR_GET_LSIZE(hdr), buf);

			/* We increased the size of b_data; update overhead */
			ARCSTAT_INCR(arcstat_overhead_size,
			    HDR_GET_LSIZE(hdr) - HDR_GET_PSIZE(hdr));
		}

		/*
		 * Regardless of the buf's previous compression settings, it
		 * should not be compressed at the end of this function.
		 */
		buf->b_flags &= ~ARC_BUF_FLAG_COMPRESSED;

		/*
		 * Try copying the data from another buf which already has a
		 * decompressed version. If that's not possible, it's time to
		 * bite the bullet and decompress the data from the hdr.
		 */
		if (arc_buf_try_copy_decompressed_data(buf)) {
			/* Skip byteswapping and checksumming (already done) */
			ASSERT3P(hdr->b_l1hdr.b_freeze_cksum, !=, NULL);
			return (0);
		} else {
			int error = zio_decompress_data(HDR_GET_COMPRESS(hdr),
			    hdr->b_l1hdr.b_pabd, buf->b_data,
			    HDR_GET_PSIZE(hdr), HDR_GET_LSIZE(hdr));

			/*
			 * Absent hardware errors or software bugs, this should
			 * be impossible, but log it anyway so we can debug it.
			 */
			if (error != 0) {
				zfs_dbgmsg(
				    "hdr %p, compress %d, psize %d, lsize %d",
				    hdr, HDR_GET_COMPRESS(hdr),
				    HDR_GET_PSIZE(hdr), HDR_GET_LSIZE(hdr));
				return (SET_ERROR(EIO));
			}
		}
	}

	/* Byteswap the buf's data if necessary */
	if (bswap != DMU_BSWAP_NUMFUNCS) {
		ASSERT(!HDR_SHARED_DATA(hdr));
		ASSERT3U(bswap, <, DMU_BSWAP_NUMFUNCS);
		dmu_ot_byteswap[bswap].ob_func(buf->b_data, HDR_GET_LSIZE(hdr));
	}

	/* Compute the hdr's checksum if necessary */
	arc_cksum_compute(buf);

	return (0);
}

int
arc_decompress(arc_buf_t *buf)
{
	return (arc_buf_fill(buf, B_FALSE));
}

/*
 * Return the size of the block, b_pabd, that is stored in the arc_buf_hdr_t.
 */
static uint64_t
arc_hdr_size(arc_buf_hdr_t *hdr)
{
	uint64_t size;

	if (HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF &&
	    HDR_GET_PSIZE(hdr) > 0) {
		size = HDR_GET_PSIZE(hdr);
	} else {
		ASSERT3U(HDR_GET_LSIZE(hdr), !=, 0);
		size = HDR_GET_LSIZE(hdr);
	}
	return (size);
}

/*
 * Increment the amount of evictable space in the arc_state_t's refcount.
 * We account for the space used by the hdr and the arc buf individually
 * so that we can add and remove them from the refcount individually.
 */
static void
arc_evictable_space_increment(arc_buf_hdr_t *hdr, arc_state_t *state)
{
	arc_buf_contents_t type = arc_buf_type(hdr);
	arc_buf_t *buf;

	ASSERT(HDR_HAS_L1HDR(hdr));

	if (GHOST_STATE(state)) {
		ASSERT0(hdr->b_l1hdr.b_bufcnt);
		ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
		ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
		(void) refcount_add_many(&state->arcs_esize[type],
		    HDR_GET_LSIZE(hdr), hdr);
		return;
	}

	ASSERT(!GHOST_STATE(state));
	if (hdr->b_l1hdr.b_pabd != NULL) {
		(void) refcount_add_many(&state->arcs_esize[type],
		    arc_hdr_size(hdr), hdr);
	}
	for (buf = hdr->b_l1hdr.b_buf; buf != NULL; buf = buf->b_next) {
		if (arc_buf_is_shared(buf))
			continue;
		(void) refcount_add_many(&state->arcs_esize[type],
		    arc_buf_size(buf), buf);
	}
}

/*
 * Decrement the amount of evictable space in the arc_state_t's refcount.
 * We account for the space used by the hdr and the arc buf individually
 * so that we can add and remove them from the refcount individually.
 */
static void
arc_evictable_space_decrement(arc_buf_hdr_t *hdr, arc_state_t *state)
{
	arc_buf_contents_t type = arc_buf_type(hdr);
	arc_buf_t *buf;

	ASSERT(HDR_HAS_L1HDR(hdr));

	if (GHOST_STATE(state)) {
		ASSERT0(hdr->b_l1hdr.b_bufcnt);
		ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
		ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
		(void) refcount_remove_many(&state->arcs_esize[type],
		    HDR_GET_LSIZE(hdr), hdr);
		return;
	}

	ASSERT(!GHOST_STATE(state));
	if (hdr->b_l1hdr.b_pabd != NULL) {
		(void) refcount_remove_many(&state->arcs_esize[type],
		    arc_hdr_size(hdr), hdr);
	}
	for (buf = hdr->b_l1hdr.b_buf; buf != NULL; buf = buf->b_next) {
		if (arc_buf_is_shared(buf))
			continue;
		(void) refcount_remove_many(&state->arcs_esize[type],
		    arc_buf_size(buf), buf);
	}
}

/*
 * Add a reference to this hdr indicating that someone is actively
 * referencing that memory. When the refcount transitions from 0 to 1,
 * we remove it from the respective arc_state_t list to indicate that
 * it is not evictable.
 */
static void
add_reference(arc_buf_hdr_t *hdr, void *tag)
{
	arc_state_t *state;

	ASSERT(HDR_HAS_L1HDR(hdr));
	if (!MUTEX_HELD(HDR_LOCK(hdr))) {
		ASSERT(hdr->b_l1hdr.b_state == arc_anon);
		ASSERT(refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
		ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
	}

	state = hdr->b_l1hdr.b_state;

	if ((refcount_add(&hdr->b_l1hdr.b_refcnt, tag) == 1) &&
	    (state != arc_anon)) {
		/* We don't use the L2-only state list. */
		if (state != arc_l2c_only) {
			multilist_remove(state->arcs_list[arc_buf_type(hdr)],
			    hdr);
			arc_evictable_space_decrement(hdr, state);
		}
		/* remove the prefetch flag if we get a reference */
		arc_hdr_clear_flags(hdr, ARC_FLAG_PREFETCH);
	}
}

/*
 * Remove a reference from this hdr. When the reference transitions from
 * 1 to 0 and we're not anonymous, then we add this hdr to the arc_state_t's
 * list making it eligible for eviction.
 */
static int
remove_reference(arc_buf_hdr_t *hdr, kmutex_t *hash_lock, void *tag)
{
	int cnt;
	arc_state_t *state = hdr->b_l1hdr.b_state;

	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT(state == arc_anon || MUTEX_HELD(hash_lock));
	ASSERT(!GHOST_STATE(state));

	/*
	 * arc_l2c_only counts as a ghost state so we don't need to explicitly
	 * check to prevent usage of the arc_l2c_only list.
	 */
	if (((cnt = refcount_remove(&hdr->b_l1hdr.b_refcnt, tag)) == 0) &&
	    (state != arc_anon)) {
		multilist_insert(state->arcs_list[arc_buf_type(hdr)], hdr);
		ASSERT3U(hdr->b_l1hdr.b_bufcnt, >, 0);
		arc_evictable_space_increment(hdr, state);
	}
	return (cnt);
}

/*
 * Returns detailed information about a specific arc buffer.  When the
 * state_index argument is set the function will calculate the arc header
 * list position for its arc state.  Since this requires a linear traversal
 * callers are strongly encourage not to do this.  However, it can be helpful
 * for targeted analysis so the functionality is provided.
 */
void
arc_buf_info(arc_buf_t *ab, arc_buf_info_t *abi, int state_index)
{
	arc_buf_hdr_t *hdr = ab->b_hdr;
	l1arc_buf_hdr_t *l1hdr = NULL;
	l2arc_buf_hdr_t *l2hdr = NULL;
	arc_state_t *state = NULL;

	memset(abi, 0, sizeof (arc_buf_info_t));

	if (hdr == NULL)
		return;

	abi->abi_flags = hdr->b_flags;

	if (HDR_HAS_L1HDR(hdr)) {
		l1hdr = &hdr->b_l1hdr;
		state = l1hdr->b_state;
	}
	if (HDR_HAS_L2HDR(hdr))
		l2hdr = &hdr->b_l2hdr;

	if (l1hdr) {
		abi->abi_bufcnt = l1hdr->b_bufcnt;
		abi->abi_access = l1hdr->b_arc_access;
		abi->abi_mru_hits = l1hdr->b_mru_hits;
		abi->abi_mru_ghost_hits = l1hdr->b_mru_ghost_hits;
		abi->abi_mfu_hits = l1hdr->b_mfu_hits;
		abi->abi_mfu_ghost_hits = l1hdr->b_mfu_ghost_hits;
		abi->abi_holds = refcount_count(&l1hdr->b_refcnt);
	}

	if (l2hdr) {
		abi->abi_l2arc_dattr = l2hdr->b_daddr;
		abi->abi_l2arc_hits = l2hdr->b_hits;
	}

	abi->abi_state_type = state ? state->arcs_state : ARC_STATE_ANON;
	abi->abi_state_contents = arc_buf_type(hdr);
	abi->abi_size = arc_hdr_size(hdr);
}

/*
 * Move the supplied buffer to the indicated state. The hash lock
 * for the buffer must be held by the caller.
 */
static void
arc_change_state(arc_state_t *new_state, arc_buf_hdr_t *hdr,
    kmutex_t *hash_lock)
{
	arc_state_t *old_state;
	int64_t refcnt;
	uint32_t bufcnt;
	boolean_t update_old, update_new;
	arc_buf_contents_t buftype = arc_buf_type(hdr);

	/*
	 * We almost always have an L1 hdr here, since we call arc_hdr_realloc()
	 * in arc_read() when bringing a buffer out of the L2ARC.  However, the
	 * L1 hdr doesn't always exist when we change state to arc_anon before
	 * destroying a header, in which case reallocating to add the L1 hdr is
	 * pointless.
	 */
	if (HDR_HAS_L1HDR(hdr)) {
		old_state = hdr->b_l1hdr.b_state;
		refcnt = refcount_count(&hdr->b_l1hdr.b_refcnt);
		bufcnt = hdr->b_l1hdr.b_bufcnt;
		update_old = (bufcnt > 0 || hdr->b_l1hdr.b_pabd != NULL);
	} else {
		old_state = arc_l2c_only;
		refcnt = 0;
		bufcnt = 0;
		update_old = B_FALSE;
	}
	update_new = update_old;

	ASSERT(MUTEX_HELD(hash_lock));
	ASSERT3P(new_state, !=, old_state);
	ASSERT(!GHOST_STATE(new_state) || bufcnt == 0);
	ASSERT(old_state != arc_anon || bufcnt <= 1);

	/*
	 * If this buffer is evictable, transfer it from the
	 * old state list to the new state list.
	 */
	if (refcnt == 0) {
		if (old_state != arc_anon && old_state != arc_l2c_only) {
			ASSERT(HDR_HAS_L1HDR(hdr));
			multilist_remove(old_state->arcs_list[buftype], hdr);

			if (GHOST_STATE(old_state)) {
				ASSERT0(bufcnt);
				ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
				update_old = B_TRUE;
			}
			arc_evictable_space_decrement(hdr, old_state);
		}
		if (new_state != arc_anon && new_state != arc_l2c_only) {
			/*
			 * An L1 header always exists here, since if we're
			 * moving to some L1-cached state (i.e. not l2c_only or
			 * anonymous), we realloc the header to add an L1hdr
			 * beforehand.
			 */
			ASSERT(HDR_HAS_L1HDR(hdr));
			multilist_insert(new_state->arcs_list[buftype], hdr);

			if (GHOST_STATE(new_state)) {
				ASSERT0(bufcnt);
				ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
				update_new = B_TRUE;
			}
			arc_evictable_space_increment(hdr, new_state);
		}
	}

	ASSERT(!HDR_EMPTY(hdr));
	if (new_state == arc_anon && HDR_IN_HASH_TABLE(hdr))
		buf_hash_remove(hdr);

	/* adjust state sizes (ignore arc_l2c_only) */

	if (update_new && new_state != arc_l2c_only) {
		ASSERT(HDR_HAS_L1HDR(hdr));
		if (GHOST_STATE(new_state)) {
			ASSERT0(bufcnt);

			/*
			 * When moving a header to a ghost state, we first
			 * remove all arc buffers. Thus, we'll have a
			 * bufcnt of zero, and no arc buffer to use for
			 * the reference. As a result, we use the arc
			 * header pointer for the reference.
			 */
			(void) refcount_add_many(&new_state->arcs_size,
			    HDR_GET_LSIZE(hdr), hdr);
			ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
		} else {
			arc_buf_t *buf;
			uint32_t buffers = 0;

			/*
			 * Each individual buffer holds a unique reference,
			 * thus we must remove each of these references one
			 * at a time.
			 */
			for (buf = hdr->b_l1hdr.b_buf; buf != NULL;
			    buf = buf->b_next) {
				ASSERT3U(bufcnt, !=, 0);
				buffers++;

				/*
				 * When the arc_buf_t is sharing the data
				 * block with the hdr, the owner of the
				 * reference belongs to the hdr. Only
				 * add to the refcount if the arc_buf_t is
				 * not shared.
				 */
				if (arc_buf_is_shared(buf))
					continue;

				(void) refcount_add_many(&new_state->arcs_size,
				    arc_buf_size(buf), buf);
			}
			ASSERT3U(bufcnt, ==, buffers);

			if (hdr->b_l1hdr.b_pabd != NULL) {
				(void) refcount_add_many(&new_state->arcs_size,
				    arc_hdr_size(hdr), hdr);
			} else {
				ASSERT(GHOST_STATE(old_state));
			}
		}
	}

	if (update_old && old_state != arc_l2c_only) {
		ASSERT(HDR_HAS_L1HDR(hdr));
		if (GHOST_STATE(old_state)) {
			ASSERT0(bufcnt);
			ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);

			/*
			 * When moving a header off of a ghost state,
			 * the header will not contain any arc buffers.
			 * We use the arc header pointer for the reference
			 * which is exactly what we did when we put the
			 * header on the ghost state.
			 */

			(void) refcount_remove_many(&old_state->arcs_size,
			    HDR_GET_LSIZE(hdr), hdr);
		} else {
			arc_buf_t *buf;
			uint32_t buffers = 0;

			/*
			 * Each individual buffer holds a unique reference,
			 * thus we must remove each of these references one
			 * at a time.
			 */
			for (buf = hdr->b_l1hdr.b_buf; buf != NULL;
			    buf = buf->b_next) {
				ASSERT3U(bufcnt, !=, 0);
				buffers++;

				/*
				 * When the arc_buf_t is sharing the data
				 * block with the hdr, the owner of the
				 * reference belongs to the hdr. Only
				 * add to the refcount if the arc_buf_t is
				 * not shared.
				 */
				if (arc_buf_is_shared(buf))
					continue;

				(void) refcount_remove_many(
				    &old_state->arcs_size, arc_buf_size(buf),
				    buf);
			}
			ASSERT3U(bufcnt, ==, buffers);
			ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
			(void) refcount_remove_many(
			    &old_state->arcs_size, arc_hdr_size(hdr), hdr);
		}
	}

	if (HDR_HAS_L1HDR(hdr))
		hdr->b_l1hdr.b_state = new_state;

	/*
	 * L2 headers should never be on the L2 state list since they don't
	 * have L1 headers allocated.
	 */
	ASSERT(multilist_is_empty(arc_l2c_only->arcs_list[ARC_BUFC_DATA]) &&
	    multilist_is_empty(arc_l2c_only->arcs_list[ARC_BUFC_METADATA]));
}

void
arc_space_consume(uint64_t space, arc_space_type_t type)
{
	ASSERT(type >= 0 && type < ARC_SPACE_NUMTYPES);

	switch (type) {
	default:
		break;
	case ARC_SPACE_DATA:
		ARCSTAT_INCR(arcstat_data_size, space);
		break;
	case ARC_SPACE_META:
		ARCSTAT_INCR(arcstat_metadata_size, space);
		break;
	case ARC_SPACE_BONUS:
		ARCSTAT_INCR(arcstat_bonus_size, space);
		break;
	case ARC_SPACE_DNODE:
		ARCSTAT_INCR(arcstat_dnode_size, space);
		break;
	case ARC_SPACE_DBUF:
		ARCSTAT_INCR(arcstat_dbuf_size, space);
		break;
	case ARC_SPACE_HDRS:
		ARCSTAT_INCR(arcstat_hdr_size, space);
		break;
	case ARC_SPACE_L2HDRS:
		ARCSTAT_INCR(arcstat_l2_hdr_size, space);
		break;
	}

	if (type != ARC_SPACE_DATA)
		ARCSTAT_INCR(arcstat_meta_used, space);

	atomic_add_64(&arc_size, space);
}

void
arc_space_return(uint64_t space, arc_space_type_t type)
{
	ASSERT(type >= 0 && type < ARC_SPACE_NUMTYPES);

	switch (type) {
	default:
		break;
	case ARC_SPACE_DATA:
		ARCSTAT_INCR(arcstat_data_size, -space);
		break;
	case ARC_SPACE_META:
		ARCSTAT_INCR(arcstat_metadata_size, -space);
		break;
	case ARC_SPACE_BONUS:
		ARCSTAT_INCR(arcstat_bonus_size, -space);
		break;
	case ARC_SPACE_DNODE:
		ARCSTAT_INCR(arcstat_dnode_size, -space);
		break;
	case ARC_SPACE_DBUF:
		ARCSTAT_INCR(arcstat_dbuf_size, -space);
		break;
	case ARC_SPACE_HDRS:
		ARCSTAT_INCR(arcstat_hdr_size, -space);
		break;
	case ARC_SPACE_L2HDRS:
		ARCSTAT_INCR(arcstat_l2_hdr_size, -space);
		break;
	}

	if (type != ARC_SPACE_DATA) {
		ASSERT(arc_meta_used >= space);
		if (arc_meta_max < arc_meta_used)
			arc_meta_max = arc_meta_used;
		ARCSTAT_INCR(arcstat_meta_used, -space);
	}

	ASSERT(arc_size >= space);
	atomic_add_64(&arc_size, -space);
}

/*
 * Given a hdr and a buf, returns whether that buf can share its b_data buffer
 * with the hdr's b_pabd.
 */
static boolean_t
arc_can_share(arc_buf_hdr_t *hdr, arc_buf_t *buf)
{
	/*
	 * The criteria for sharing a hdr's data are:
	 * 1. the hdr's compression matches the buf's compression
	 * 2. the hdr doesn't need to be byteswapped
	 * 3. the hdr isn't already being shared
	 * 4. the buf is either compressed or it is the last buf in the hdr list
	 *
	 * Criterion #4 maintains the invariant that shared uncompressed
	 * bufs must be the final buf in the hdr's b_buf list. Reading this, you
	 * might ask, "if a compressed buf is allocated first, won't that be the
	 * last thing in the list?", but in that case it's impossible to create
	 * a shared uncompressed buf anyway (because the hdr must be compressed
	 * to have the compressed buf). You might also think that #3 is
	 * sufficient to make this guarantee, however it's possible
	 * (specifically in the rare L2ARC write race mentioned in
	 * arc_buf_alloc_impl()) there will be an existing uncompressed buf that
	 * is sharable, but wasn't at the time of its allocation. Rather than
	 * allow a new shared uncompressed buf to be created and then shuffle
	 * the list around to make it the last element, this simply disallows
	 * sharing if the new buf isn't the first to be added.
	 */
	ASSERT3P(buf->b_hdr, ==, hdr);
	boolean_t hdr_compressed = HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF;
	boolean_t buf_compressed = ARC_BUF_COMPRESSED(buf) != 0;
	return (buf_compressed == hdr_compressed &&
	    hdr->b_l1hdr.b_byteswap == DMU_BSWAP_NUMFUNCS &&
	    !HDR_SHARED_DATA(hdr) &&
	    (ARC_BUF_LAST(buf) || ARC_BUF_COMPRESSED(buf)));
}

/*
 * Allocate a buf for this hdr. If you care about the data that's in the hdr,
 * or if you want a compressed buffer, pass those flags in. Returns 0 if the
 * copy was made successfully, or an error code otherwise.
 */
static int
arc_buf_alloc_impl(arc_buf_hdr_t *hdr, void *tag, boolean_t compressed,
    boolean_t fill, arc_buf_t **ret)
{
	arc_buf_t *buf;

	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT3U(HDR_GET_LSIZE(hdr), >, 0);
	VERIFY(hdr->b_type == ARC_BUFC_DATA ||
	    hdr->b_type == ARC_BUFC_METADATA);
	ASSERT3P(ret, !=, NULL);
	ASSERT3P(*ret, ==, NULL);

	hdr->b_l1hdr.b_mru_hits = 0;
	hdr->b_l1hdr.b_mru_ghost_hits = 0;
	hdr->b_l1hdr.b_mfu_hits = 0;
	hdr->b_l1hdr.b_mfu_ghost_hits = 0;
	hdr->b_l1hdr.b_l2_hits = 0;

	buf = *ret = kmem_cache_alloc(buf_cache, KM_PUSHPAGE);
	buf->b_hdr = hdr;
	buf->b_data = NULL;
	buf->b_next = hdr->b_l1hdr.b_buf;
	buf->b_flags = 0;

	add_reference(hdr, tag);

	/*
	 * We're about to change the hdr's b_flags. We must either
	 * hold the hash_lock or be undiscoverable.
	 */
	ASSERT(MUTEX_HELD(HDR_LOCK(hdr)) || HDR_EMPTY(hdr));

	/*
	 * Only honor requests for compressed bufs if the hdr is actually
	 * compressed.
	 */
	if (compressed && HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF)
		buf->b_flags |= ARC_BUF_FLAG_COMPRESSED;

	/*
	 * If the hdr's data can be shared then we share the data buffer and
	 * set the appropriate bit in the hdr's b_flags to indicate the hdr is
	 * allocate a new buffer to store the buf's data.
	 *
	 * There are two additional restrictions here because we're sharing
	 * hdr -> buf instead of the usual buf -> hdr. First, the hdr can't be
	 * actively involved in an L2ARC write, because if this buf is used by
	 * an arc_write() then the hdr's data buffer will be released when the
	 * write completes, even though the L2ARC write might still be using it.
	 * Second, the hdr's ABD must be linear so that the buf's user doesn't
	 * need to be ABD-aware.
	 */
	boolean_t can_share = arc_can_share(hdr, buf) && !HDR_L2_WRITING(hdr) &&
	    abd_is_linear(hdr->b_l1hdr.b_pabd);

	/* Set up b_data and sharing */
	if (can_share) {
		buf->b_data = abd_to_buf(hdr->b_l1hdr.b_pabd);
		buf->b_flags |= ARC_BUF_FLAG_SHARED;
		arc_hdr_set_flags(hdr, ARC_FLAG_SHARED_DATA);
	} else {
		buf->b_data =
		    arc_get_data_buf(hdr, arc_buf_size(buf), buf);
		ARCSTAT_INCR(arcstat_overhead_size, arc_buf_size(buf));
	}
	VERIFY3P(buf->b_data, !=, NULL);

	hdr->b_l1hdr.b_buf = buf;
	hdr->b_l1hdr.b_bufcnt += 1;

	/*
	 * If the user wants the data from the hdr, we need to either copy or
	 * decompress the data.
	 */
	if (fill) {
		return (arc_buf_fill(buf, ARC_BUF_COMPRESSED(buf) != 0));
	}

	return (0);
}

static char *arc_onloan_tag = "onloan";

static inline void
arc_loaned_bytes_update(int64_t delta)
{
	atomic_add_64(&arc_loaned_bytes, delta);

	/* assert that it did not wrap around */
	ASSERT3S(atomic_add_64_nv(&arc_loaned_bytes, 0), >=, 0);
}

/*
 * Loan out an anonymous arc buffer. Loaned buffers are not counted as in
 * flight data by arc_tempreserve_space() until they are "returned". Loaned
 * buffers must be returned to the arc before they can be used by the DMU or
 * freed.
 */
arc_buf_t *
arc_loan_buf(spa_t *spa, boolean_t is_metadata, int size)
{
	arc_buf_t *buf = arc_alloc_buf(spa, arc_onloan_tag,
	    is_metadata ? ARC_BUFC_METADATA : ARC_BUFC_DATA, size);

	arc_loaned_bytes_update(size);

	return (buf);
}

arc_buf_t *
arc_loan_compressed_buf(spa_t *spa, uint64_t psize, uint64_t lsize,
    enum zio_compress compression_type)
{
	arc_buf_t *buf = arc_alloc_compressed_buf(spa, arc_onloan_tag,
	    psize, lsize, compression_type);

	arc_loaned_bytes_update(psize);

	return (buf);
}


/*
 * Return a loaned arc buffer to the arc.
 */
void
arc_return_buf(arc_buf_t *buf, void *tag)
{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	ASSERT3P(buf->b_data, !=, NULL);
	ASSERT(HDR_HAS_L1HDR(hdr));
	(void) refcount_add(&hdr->b_l1hdr.b_refcnt, tag);
	(void) refcount_remove(&hdr->b_l1hdr.b_refcnt, arc_onloan_tag);

	arc_loaned_bytes_update(-arc_buf_size(buf));
}

/* Detach an arc_buf from a dbuf (tag) */
void
arc_loan_inuse_buf(arc_buf_t *buf, void *tag)
{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	ASSERT3P(buf->b_data, !=, NULL);
	ASSERT(HDR_HAS_L1HDR(hdr));
	(void) refcount_add(&hdr->b_l1hdr.b_refcnt, arc_onloan_tag);
	(void) refcount_remove(&hdr->b_l1hdr.b_refcnt, tag);

	arc_loaned_bytes_update(arc_buf_size(buf));
}

static void
l2arc_free_abd_on_write(abd_t *abd, size_t size, arc_buf_contents_t type)
{
	l2arc_data_free_t *df = kmem_alloc(sizeof (*df), KM_SLEEP);

	df->l2df_abd = abd;
	df->l2df_size = size;
	df->l2df_type = type;
	mutex_enter(&l2arc_free_on_write_mtx);
	list_insert_head(l2arc_free_on_write, df);
	mutex_exit(&l2arc_free_on_write_mtx);
}

static void
arc_hdr_free_on_write(arc_buf_hdr_t *hdr)
{
	arc_state_t *state = hdr->b_l1hdr.b_state;
	arc_buf_contents_t type = arc_buf_type(hdr);
	uint64_t size = arc_hdr_size(hdr);

	/* protected by hash lock, if in the hash table */
	if (multilist_link_active(&hdr->b_l1hdr.b_arc_node)) {
		ASSERT(refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
		ASSERT(state != arc_anon && state != arc_l2c_only);

		(void) refcount_remove_many(&state->arcs_esize[type],
		    size, hdr);
	}
	(void) refcount_remove_many(&state->arcs_size, size, hdr);
	if (type == ARC_BUFC_METADATA) {
		arc_space_return(size, ARC_SPACE_META);
	} else {
		ASSERT(type == ARC_BUFC_DATA);
		arc_space_return(size, ARC_SPACE_DATA);
	}

	l2arc_free_abd_on_write(hdr->b_l1hdr.b_pabd, size, type);
}

/*
 * Share the arc_buf_t's data with the hdr. Whenever we are sharing the
 * data buffer, we transfer the refcount ownership to the hdr and update
 * the appropriate kstats.
 */
static void
arc_share_buf(arc_buf_hdr_t *hdr, arc_buf_t *buf)
{
	ASSERT(arc_can_share(hdr, buf));
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	ASSERT(MUTEX_HELD(HDR_LOCK(hdr)) || HDR_EMPTY(hdr));

	/*
	 * Start sharing the data buffer. We transfer the
	 * refcount ownership to the hdr since it always owns
	 * the refcount whenever an arc_buf_t is shared.
	 */
	refcount_transfer_ownership(&hdr->b_l1hdr.b_state->arcs_size, buf, hdr);
	hdr->b_l1hdr.b_pabd = abd_get_from_buf(buf->b_data, arc_buf_size(buf));
	abd_take_ownership_of_buf(hdr->b_l1hdr.b_pabd,
	    HDR_ISTYPE_METADATA(hdr));
	arc_hdr_set_flags(hdr, ARC_FLAG_SHARED_DATA);
	buf->b_flags |= ARC_BUF_FLAG_SHARED;

	/*
	 * Since we've transferred ownership to the hdr we need
	 * to increment its compressed and uncompressed kstats and
	 * decrement the overhead size.
	 */
	ARCSTAT_INCR(arcstat_compressed_size, arc_hdr_size(hdr));
	ARCSTAT_INCR(arcstat_uncompressed_size, HDR_GET_LSIZE(hdr));
	ARCSTAT_INCR(arcstat_overhead_size, -arc_buf_size(buf));
}

static void
arc_unshare_buf(arc_buf_hdr_t *hdr, arc_buf_t *buf)
{
	ASSERT(arc_buf_is_shared(buf));
	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
	ASSERT(MUTEX_HELD(HDR_LOCK(hdr)) || HDR_EMPTY(hdr));

	/*
	 * We are no longer sharing this buffer so we need
	 * to transfer its ownership to the rightful owner.
	 */
	refcount_transfer_ownership(&hdr->b_l1hdr.b_state->arcs_size, hdr, buf);
	arc_hdr_clear_flags(hdr, ARC_FLAG_SHARED_DATA);
	abd_release_ownership_of_buf(hdr->b_l1hdr.b_pabd);
	abd_put(hdr->b_l1hdr.b_pabd);
	hdr->b_l1hdr.b_pabd = NULL;
	buf->b_flags &= ~ARC_BUF_FLAG_SHARED;

	/*
	 * Since the buffer is no longer shared between
	 * the arc buf and the hdr, count it as overhead.
	 */
	ARCSTAT_INCR(arcstat_compressed_size, -arc_hdr_size(hdr));
	ARCSTAT_INCR(arcstat_uncompressed_size, -HDR_GET_LSIZE(hdr));
	ARCSTAT_INCR(arcstat_overhead_size, arc_buf_size(buf));
}

/*
 * Remove an arc_buf_t from the hdr's buf list and return the last
 * arc_buf_t on the list. If no buffers remain on the list then return
 * NULL.
 */
static arc_buf_t *
arc_buf_remove(arc_buf_hdr_t *hdr, arc_buf_t *buf)
{
	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT(MUTEX_HELD(HDR_LOCK(hdr)) || HDR_EMPTY(hdr));

	arc_buf_t **bufp = &hdr->b_l1hdr.b_buf;
	arc_buf_t *lastbuf = NULL;

	/*
	 * Remove the buf from the hdr list and locate the last
	 * remaining buffer on the list.
	 */
	while (*bufp != NULL) {
		if (*bufp == buf)
			*bufp = buf->b_next;

		/*
		 * If we've removed a buffer in the middle of
		 * the list then update the lastbuf and update
		 * bufp.
		 */
		if (*bufp != NULL) {
			lastbuf = *bufp;
			bufp = &(*bufp)->b_next;
		}
	}
	buf->b_next = NULL;
	ASSERT3P(lastbuf, !=, buf);
	IMPLY(hdr->b_l1hdr.b_bufcnt > 0, lastbuf != NULL);
	IMPLY(hdr->b_l1hdr.b_bufcnt > 0, hdr->b_l1hdr.b_buf != NULL);
	IMPLY(lastbuf != NULL, ARC_BUF_LAST(lastbuf));

	return (lastbuf);
}

/*
 * Free up buf->b_data and pull the arc_buf_t off of the the arc_buf_hdr_t's
 * list and free it.
 */
static void
arc_buf_destroy_impl(arc_buf_t *buf)
{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	/*
	 * Free up the data associated with the buf but only if we're not
	 * sharing this with the hdr. If we are sharing it with the hdr, the
	 * hdr is responsible for doing the free.
	 */
	if (buf->b_data != NULL) {
		/*
		 * We're about to change the hdr's b_flags. We must either
		 * hold the hash_lock or be undiscoverable.
		 */
		ASSERT(MUTEX_HELD(HDR_LOCK(hdr)) || HDR_EMPTY(hdr));

		arc_cksum_verify(buf);
		arc_buf_unwatch(buf);

		if (arc_buf_is_shared(buf)) {
			arc_hdr_clear_flags(hdr, ARC_FLAG_SHARED_DATA);
		} else {
			uint64_t size = arc_buf_size(buf);
			arc_free_data_buf(hdr, buf->b_data, size, buf);
			ARCSTAT_INCR(arcstat_overhead_size, -size);
		}
		buf->b_data = NULL;

		ASSERT(hdr->b_l1hdr.b_bufcnt > 0);
		hdr->b_l1hdr.b_bufcnt -= 1;
	}

	arc_buf_t *lastbuf = arc_buf_remove(hdr, buf);

	if (ARC_BUF_SHARED(buf) && !ARC_BUF_COMPRESSED(buf)) {
		/*
		 * If the current arc_buf_t is sharing its data buffer with the
		 * hdr, then reassign the hdr's b_pabd to share it with the new
		 * buffer at the end of the list. The shared buffer is always
		 * the last one on the hdr's buffer list.
		 *
		 * There is an equivalent case for compressed bufs, but since
		 * they aren't guaranteed to be the last buf in the list and
		 * that is an exceedingly rare case, we just allow that space be
		 * wasted temporarily.
		 */
		if (lastbuf != NULL) {
			/* Only one buf can be shared at once */
			VERIFY(!arc_buf_is_shared(lastbuf));
			/* hdr is uncompressed so can't have compressed buf */
			VERIFY(!ARC_BUF_COMPRESSED(lastbuf));

			ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
			arc_hdr_free_pabd(hdr);

			/*
			 * We must setup a new shared block between the
			 * last buffer and the hdr. The data would have
			 * been allocated by the arc buf so we need to transfer
			 * ownership to the hdr since it's now being shared.
			 */
			arc_share_buf(hdr, lastbuf);
		}
	} else if (HDR_SHARED_DATA(hdr)) {
		/*
		 * Uncompressed shared buffers are always at the end
		 * of the list. Compressed buffers don't have the
		 * same requirements. This makes it hard to
		 * simply assert that the lastbuf is shared so
		 * we rely on the hdr's compression flags to determine
		 * if we have a compressed, shared buffer.
		 */
		ASSERT3P(lastbuf, !=, NULL);
		ASSERT(arc_buf_is_shared(lastbuf) ||
		    HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF);
	}

	/*
	 * Free the checksum if we're removing the last uncompressed buf from
	 * this hdr.
	 */
	if (!arc_hdr_has_uncompressed_buf(hdr)) {
		arc_cksum_free(hdr);
	}

	/* clean up the buf */
	buf->b_hdr = NULL;
	kmem_cache_free(buf_cache, buf);
}

static void
arc_hdr_alloc_pabd(arc_buf_hdr_t *hdr)
{
	ASSERT3U(HDR_GET_LSIZE(hdr), >, 0);
	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT(!HDR_SHARED_DATA(hdr));

	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	hdr->b_l1hdr.b_pabd = arc_get_data_abd(hdr, arc_hdr_size(hdr), hdr);
	hdr->b_l1hdr.b_byteswap = DMU_BSWAP_NUMFUNCS;
	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);

	ARCSTAT_INCR(arcstat_compressed_size, arc_hdr_size(hdr));
	ARCSTAT_INCR(arcstat_uncompressed_size, HDR_GET_LSIZE(hdr));
}

static void
arc_hdr_free_pabd(arc_buf_hdr_t *hdr)
{
	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);

	/*
	 * If the hdr is currently being written to the l2arc then
	 * we defer freeing the data by adding it to the l2arc_free_on_write
	 * list. The l2arc will free the data once it's finished
	 * writing it to the l2arc device.
	 */
	if (HDR_L2_WRITING(hdr)) {
		arc_hdr_free_on_write(hdr);
		ARCSTAT_BUMP(arcstat_l2_free_on_write);
	} else {
		arc_free_data_abd(hdr, hdr->b_l1hdr.b_pabd,
		    arc_hdr_size(hdr), hdr);
	}
	hdr->b_l1hdr.b_pabd = NULL;
	hdr->b_l1hdr.b_byteswap = DMU_BSWAP_NUMFUNCS;

	ARCSTAT_INCR(arcstat_compressed_size, -arc_hdr_size(hdr));
	ARCSTAT_INCR(arcstat_uncompressed_size, -HDR_GET_LSIZE(hdr));
}

static arc_buf_hdr_t *
arc_hdr_alloc(uint64_t spa, int32_t psize, int32_t lsize,
    enum zio_compress compression_type, arc_buf_contents_t type)
{
	arc_buf_hdr_t *hdr;

	VERIFY(type == ARC_BUFC_DATA || type == ARC_BUFC_METADATA);

	hdr = kmem_cache_alloc(hdr_full_cache, KM_PUSHPAGE);
	ASSERT(HDR_EMPTY(hdr));
	ASSERT3P(hdr->b_l1hdr.b_freeze_cksum, ==, NULL);
	HDR_SET_PSIZE(hdr, psize);
	HDR_SET_LSIZE(hdr, lsize);
	hdr->b_spa = spa;
	hdr->b_type = type;
	hdr->b_flags = 0;
	arc_hdr_set_flags(hdr, arc_bufc_to_flags(type) | ARC_FLAG_HAS_L1HDR);
	arc_hdr_set_compress(hdr, compression_type);

	hdr->b_l1hdr.b_state = arc_anon;
	hdr->b_l1hdr.b_arc_access = 0;
	hdr->b_l1hdr.b_bufcnt = 0;
	hdr->b_l1hdr.b_buf = NULL;

	/*
	 * Allocate the hdr's buffer. This will contain either
	 * the compressed or uncompressed data depending on the block
	 * it references and compressed arc enablement.
	 */
	arc_hdr_alloc_pabd(hdr);
	ASSERT(refcount_is_zero(&hdr->b_l1hdr.b_refcnt));

	return (hdr);
}

/*
 * Transition between the two allocation states for the arc_buf_hdr struct.
 * The arc_buf_hdr struct can be allocated with (hdr_full_cache) or without
 * (hdr_l2only_cache) the fields necessary for the L1 cache - the smaller
 * version is used when a cache buffer is only in the L2ARC in order to reduce
 * memory usage.
 */
static arc_buf_hdr_t *
arc_hdr_realloc(arc_buf_hdr_t *hdr, kmem_cache_t *old, kmem_cache_t *new)
{
	arc_buf_hdr_t *nhdr;
	l2arc_dev_t *dev = hdr->b_l2hdr.b_dev;

	ASSERT(HDR_HAS_L2HDR(hdr));
	ASSERT((old == hdr_full_cache && new == hdr_l2only_cache) ||
	    (old == hdr_l2only_cache && new == hdr_full_cache));

	nhdr = kmem_cache_alloc(new, KM_PUSHPAGE);

	ASSERT(MUTEX_HELD(HDR_LOCK(hdr)));
	buf_hash_remove(hdr);

	bcopy(hdr, nhdr, HDR_L2ONLY_SIZE);

	if (new == hdr_full_cache) {
		arc_hdr_set_flags(nhdr, ARC_FLAG_HAS_L1HDR);
		/*
		 * arc_access and arc_change_state need to be aware that a
		 * header has just come out of L2ARC, so we set its state to
		 * l2c_only even though it's about to change.
		 */
		nhdr->b_l1hdr.b_state = arc_l2c_only;

		/* Verify previous threads set to NULL before freeing */
		ASSERT3P(nhdr->b_l1hdr.b_pabd, ==, NULL);
	} else {
		ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
		ASSERT0(hdr->b_l1hdr.b_bufcnt);
		ASSERT3P(hdr->b_l1hdr.b_freeze_cksum, ==, NULL);

		/*
		 * If we've reached here, We must have been called from
		 * arc_evict_hdr(), as such we should have already been
		 * removed from any ghost list we were previously on
		 * (which protects us from racing with arc_evict_state),
		 * thus no locking is needed during this check.
		 */
		ASSERT(!multilist_link_active(&hdr->b_l1hdr.b_arc_node));

		/*
		 * A buffer must not be moved into the arc_l2c_only
		 * state if it's not finished being written out to the
		 * l2arc device. Otherwise, the b_l1hdr.b_pabd field
		 * might try to be accessed, even though it was removed.
		 */
		VERIFY(!HDR_L2_WRITING(hdr));
		VERIFY3P(hdr->b_l1hdr.b_pabd, ==, NULL);

		arc_hdr_clear_flags(nhdr, ARC_FLAG_HAS_L1HDR);
	}
	/*
	 * The header has been reallocated so we need to re-insert it into any
	 * lists it was on.
	 */
	(void) buf_hash_insert(nhdr, NULL);

	ASSERT(list_link_active(&hdr->b_l2hdr.b_l2node));

	mutex_enter(&dev->l2ad_mtx);

	/*
	 * We must place the realloc'ed header back into the list at
	 * the same spot. Otherwise, if it's placed earlier in the list,
	 * l2arc_write_buffers() could find it during the function's
	 * write phase, and try to write it out to the l2arc.
	 */
	list_insert_after(&dev->l2ad_buflist, hdr, nhdr);
	list_remove(&dev->l2ad_buflist, hdr);

	mutex_exit(&dev->l2ad_mtx);

	/*
	 * Since we're using the pointer address as the tag when
	 * incrementing and decrementing the l2ad_alloc refcount, we
	 * must remove the old pointer (that we're about to destroy) and
	 * add the new pointer to the refcount. Otherwise we'd remove
	 * the wrong pointer address when calling arc_hdr_destroy() later.
	 */

	(void) refcount_remove_many(&dev->l2ad_alloc, arc_hdr_size(hdr), hdr);
	(void) refcount_add_many(&dev->l2ad_alloc, arc_hdr_size(nhdr), nhdr);

	buf_discard_identity(hdr);
	kmem_cache_free(old, hdr);

	return (nhdr);
}

/*
 * Allocate a new arc_buf_hdr_t and arc_buf_t and return the buf to the caller.
 * The buf is returned thawed since we expect the consumer to modify it.
 */
arc_buf_t *
arc_alloc_buf(spa_t *spa, void *tag, arc_buf_contents_t type, int32_t size)
{
	arc_buf_hdr_t *hdr = arc_hdr_alloc(spa_load_guid(spa), size, size,
	    ZIO_COMPRESS_OFF, type);
	ASSERT(!MUTEX_HELD(HDR_LOCK(hdr)));

	arc_buf_t *buf = NULL;
	VERIFY0(arc_buf_alloc_impl(hdr, tag, B_FALSE, B_FALSE, &buf));
	arc_buf_thaw(buf);

	return (buf);
}

/*
 * Allocate a compressed buf in the same manner as arc_alloc_buf. Don't use this
 * for bufs containing metadata.
 */
arc_buf_t *
arc_alloc_compressed_buf(spa_t *spa, void *tag, uint64_t psize, uint64_t lsize,
    enum zio_compress compression_type)
{
	ASSERT3U(lsize, >, 0);
	ASSERT3U(lsize, >=, psize);
	ASSERT(compression_type > ZIO_COMPRESS_OFF);
	ASSERT(compression_type < ZIO_COMPRESS_FUNCTIONS);

	arc_buf_hdr_t *hdr = arc_hdr_alloc(spa_load_guid(spa), psize, lsize,
	    compression_type, ARC_BUFC_DATA);
	ASSERT(!MUTEX_HELD(HDR_LOCK(hdr)));

	arc_buf_t *buf = NULL;
	VERIFY0(arc_buf_alloc_impl(hdr, tag, B_TRUE, B_FALSE, &buf));
	arc_buf_thaw(buf);
	ASSERT3P(hdr->b_l1hdr.b_freeze_cksum, ==, NULL);

	if (!arc_buf_is_shared(buf)) {
		/*
		 * To ensure that the hdr has the correct data in it if we call
		 * arc_decompress() on this buf before it's been written to
		 * disk, it's easiest if we just set up sharing between the
		 * buf and the hdr.
		 */
		ASSERT(!abd_is_linear(hdr->b_l1hdr.b_pabd));
		arc_hdr_free_pabd(hdr);
		arc_share_buf(hdr, buf);
	}

	return (buf);
}

static void
arc_hdr_l2hdr_destroy(arc_buf_hdr_t *hdr)
{
	l2arc_buf_hdr_t *l2hdr = &hdr->b_l2hdr;
	l2arc_dev_t *dev = l2hdr->b_dev;
	uint64_t asize = arc_hdr_size(hdr);

	ASSERT(MUTEX_HELD(&dev->l2ad_mtx));
	ASSERT(HDR_HAS_L2HDR(hdr));

	list_remove(&dev->l2ad_buflist, hdr);

	ARCSTAT_INCR(arcstat_l2_asize, -asize);
	ARCSTAT_INCR(arcstat_l2_size, -HDR_GET_LSIZE(hdr));

	vdev_space_update(dev->l2ad_vdev, -asize, 0, 0);

	(void) refcount_remove_many(&dev->l2ad_alloc, asize, hdr);
	arc_hdr_clear_flags(hdr, ARC_FLAG_HAS_L2HDR);
}

static void
arc_hdr_destroy(arc_buf_hdr_t *hdr)
{
	if (HDR_HAS_L1HDR(hdr)) {
		ASSERT(hdr->b_l1hdr.b_buf == NULL ||
		    hdr->b_l1hdr.b_bufcnt > 0);
		ASSERT(refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
		ASSERT3P(hdr->b_l1hdr.b_state, ==, arc_anon);
	}
	ASSERT(!HDR_IO_IN_PROGRESS(hdr));
	ASSERT(!HDR_IN_HASH_TABLE(hdr));

	if (!HDR_EMPTY(hdr))
		buf_discard_identity(hdr);

	if (HDR_HAS_L2HDR(hdr)) {
		l2arc_dev_t *dev = hdr->b_l2hdr.b_dev;
		boolean_t buflist_held = MUTEX_HELD(&dev->l2ad_mtx);

		if (!buflist_held)
			mutex_enter(&dev->l2ad_mtx);

		/*
		 * Even though we checked this conditional above, we
		 * need to check this again now that we have the
		 * l2ad_mtx. This is because we could be racing with
		 * another thread calling l2arc_evict() which might have
		 * destroyed this header's L2 portion as we were waiting
		 * to acquire the l2ad_mtx. If that happens, we don't
		 * want to re-destroy the header's L2 portion.
		 */
		if (HDR_HAS_L2HDR(hdr))
			arc_hdr_l2hdr_destroy(hdr);

		if (!buflist_held)
			mutex_exit(&dev->l2ad_mtx);
	}

	if (HDR_HAS_L1HDR(hdr)) {
		arc_cksum_free(hdr);

		while (hdr->b_l1hdr.b_buf != NULL)
			arc_buf_destroy_impl(hdr->b_l1hdr.b_buf);

		if (hdr->b_l1hdr.b_pabd != NULL)
			arc_hdr_free_pabd(hdr);
	}

	ASSERT3P(hdr->b_hash_next, ==, NULL);
	if (HDR_HAS_L1HDR(hdr)) {
		ASSERT(!multilist_link_active(&hdr->b_l1hdr.b_arc_node));
		ASSERT3P(hdr->b_l1hdr.b_acb, ==, NULL);
		kmem_cache_free(hdr_full_cache, hdr);
	} else {
		kmem_cache_free(hdr_l2only_cache, hdr);
	}
}

void
arc_buf_destroy(arc_buf_t *buf, void* tag)
{
	arc_buf_hdr_t *hdr = buf->b_hdr;
	kmutex_t *hash_lock = HDR_LOCK(hdr);

	if (hdr->b_l1hdr.b_state == arc_anon) {
		ASSERT3U(hdr->b_l1hdr.b_bufcnt, ==, 1);
		ASSERT(!HDR_IO_IN_PROGRESS(hdr));
		VERIFY0(remove_reference(hdr, NULL, tag));
		arc_hdr_destroy(hdr);
		return;
	}

	mutex_enter(hash_lock);
	ASSERT3P(hdr, ==, buf->b_hdr);
	ASSERT(hdr->b_l1hdr.b_bufcnt > 0);
	ASSERT3P(hash_lock, ==, HDR_LOCK(hdr));
	ASSERT3P(hdr->b_l1hdr.b_state, !=, arc_anon);
	ASSERT3P(buf->b_data, !=, NULL);

	(void) remove_reference(hdr, hash_lock, tag);
	arc_buf_destroy_impl(buf);
	mutex_exit(hash_lock);
}

/*
 * Evict the arc_buf_hdr that is provided as a parameter. The resultant
 * state of the header is dependent on its state prior to entering this
 * function. The following transitions are possible:
 *
 *    - arc_mru -> arc_mru_ghost
 *    - arc_mfu -> arc_mfu_ghost
 *    - arc_mru_ghost -> arc_l2c_only
 *    - arc_mru_ghost -> deleted
 *    - arc_mfu_ghost -> arc_l2c_only
 *    - arc_mfu_ghost -> deleted
 */
static int64_t
arc_evict_hdr(arc_buf_hdr_t *hdr, kmutex_t *hash_lock)
{
	arc_state_t *evicted_state, *state;
	int64_t bytes_evicted = 0;

	ASSERT(MUTEX_HELD(hash_lock));
	ASSERT(HDR_HAS_L1HDR(hdr));

	state = hdr->b_l1hdr.b_state;
	if (GHOST_STATE(state)) {
		ASSERT(!HDR_IO_IN_PROGRESS(hdr));
		ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);

		/*
		 * l2arc_write_buffers() relies on a header's L1 portion
		 * (i.e. its b_pabd field) during it's write phase.
		 * Thus, we cannot push a header onto the arc_l2c_only
		 * state (removing its L1 piece) until the header is
		 * done being written to the l2arc.
		 */
		if (HDR_HAS_L2HDR(hdr) && HDR_L2_WRITING(hdr)) {
			ARCSTAT_BUMP(arcstat_evict_l2_skip);
			return (bytes_evicted);
		}

		ARCSTAT_BUMP(arcstat_deleted);
		bytes_evicted += HDR_GET_LSIZE(hdr);

		DTRACE_PROBE1(arc__delete, arc_buf_hdr_t *, hdr);

		if (HDR_HAS_L2HDR(hdr)) {
			ASSERT(hdr->b_l1hdr.b_pabd == NULL);
			/*
			 * This buffer is cached on the 2nd Level ARC;
			 * don't destroy the header.
			 */
			arc_change_state(arc_l2c_only, hdr, hash_lock);
			/*
			 * dropping from L1+L2 cached to L2-only,
			 * realloc to remove the L1 header.
			 */
			hdr = arc_hdr_realloc(hdr, hdr_full_cache,
			    hdr_l2only_cache);
		} else {
			arc_change_state(arc_anon, hdr, hash_lock);
			arc_hdr_destroy(hdr);
		}
		return (bytes_evicted);
	}

	ASSERT(state == arc_mru || state == arc_mfu);
	evicted_state = (state == arc_mru) ? arc_mru_ghost : arc_mfu_ghost;

	/* prefetch buffers have a minimum lifespan */
	if (HDR_IO_IN_PROGRESS(hdr) ||
	    ((hdr->b_flags & (ARC_FLAG_PREFETCH | ARC_FLAG_INDIRECT)) &&
	    ddi_get_lbolt() - hdr->b_l1hdr.b_arc_access <
	    arc_min_prefetch_lifespan)) {
		ARCSTAT_BUMP(arcstat_evict_skip);
		return (bytes_evicted);
	}

	ASSERT0(refcount_count(&hdr->b_l1hdr.b_refcnt));
	while (hdr->b_l1hdr.b_buf) {
		arc_buf_t *buf = hdr->b_l1hdr.b_buf;
		if (!mutex_tryenter(&buf->b_evict_lock)) {
			ARCSTAT_BUMP(arcstat_mutex_miss);
			break;
		}
		if (buf->b_data != NULL)
			bytes_evicted += HDR_GET_LSIZE(hdr);
		mutex_exit(&buf->b_evict_lock);
		arc_buf_destroy_impl(buf);
	}

	if (HDR_HAS_L2HDR(hdr)) {
		ARCSTAT_INCR(arcstat_evict_l2_cached, HDR_GET_LSIZE(hdr));
	} else {
		if (l2arc_write_eligible(hdr->b_spa, hdr)) {
			ARCSTAT_INCR(arcstat_evict_l2_eligible,
			    HDR_GET_LSIZE(hdr));
		} else {
			ARCSTAT_INCR(arcstat_evict_l2_ineligible,
			    HDR_GET_LSIZE(hdr));
		}
	}

	if (hdr->b_l1hdr.b_bufcnt == 0) {
		arc_cksum_free(hdr);

		bytes_evicted += arc_hdr_size(hdr);

		/*
		 * If this hdr is being evicted and has a compressed
		 * buffer then we discard it here before we change states.
		 * This ensures that the accounting is updated correctly
		 * in arc_free_data_impl().
		 */
		arc_hdr_free_pabd(hdr);

		arc_change_state(evicted_state, hdr, hash_lock);
		ASSERT(HDR_IN_HASH_TABLE(hdr));
		arc_hdr_set_flags(hdr, ARC_FLAG_IN_HASH_TABLE);
		DTRACE_PROBE1(arc__evict, arc_buf_hdr_t *, hdr);
	}

	return (bytes_evicted);
}

static uint64_t
arc_evict_state_impl(multilist_t *ml, int idx, arc_buf_hdr_t *marker,
    uint64_t spa, int64_t bytes)
{
	multilist_sublist_t *mls;
	uint64_t bytes_evicted = 0;
	arc_buf_hdr_t *hdr;
	kmutex_t *hash_lock;
	int evict_count = 0;

	ASSERT3P(marker, !=, NULL);
	IMPLY(bytes < 0, bytes == ARC_EVICT_ALL);

	mls = multilist_sublist_lock(ml, idx);

	for (hdr = multilist_sublist_prev(mls, marker); hdr != NULL;
	    hdr = multilist_sublist_prev(mls, marker)) {
		if ((bytes != ARC_EVICT_ALL && bytes_evicted >= bytes) ||
		    (evict_count >= zfs_arc_evict_batch_limit))
			break;

		/*
		 * To keep our iteration location, move the marker
		 * forward. Since we're not holding hdr's hash lock, we
		 * must be very careful and not remove 'hdr' from the
		 * sublist. Otherwise, other consumers might mistake the
		 * 'hdr' as not being on a sublist when they call the
		 * multilist_link_active() function (they all rely on
		 * the hash lock protecting concurrent insertions and
		 * removals). multilist_sublist_move_forward() was
		 * specifically implemented to ensure this is the case
		 * (only 'marker' will be removed and re-inserted).
		 */
		multilist_sublist_move_forward(mls, marker);

		/*
		 * The only case where the b_spa field should ever be
		 * zero, is the marker headers inserted by
		 * arc_evict_state(). It's possible for multiple threads
		 * to be calling arc_evict_state() concurrently (e.g.
		 * dsl_pool_close() and zio_inject_fault()), so we must
		 * skip any markers we see from these other threads.
		 */
		if (hdr->b_spa == 0)
			continue;

		/* we're only interested in evicting buffers of a certain spa */
		if (spa != 0 && hdr->b_spa != spa) {
			ARCSTAT_BUMP(arcstat_evict_skip);
			continue;
		}

		hash_lock = HDR_LOCK(hdr);

		/*
		 * We aren't calling this function from any code path
		 * that would already be holding a hash lock, so we're
		 * asserting on this assumption to be defensive in case
		 * this ever changes. Without this check, it would be
		 * possible to incorrectly increment arcstat_mutex_miss
		 * below (e.g. if the code changed such that we called
		 * this function with a hash lock held).
		 */
		ASSERT(!MUTEX_HELD(hash_lock));

		if (mutex_tryenter(hash_lock)) {
			uint64_t evicted = arc_evict_hdr(hdr, hash_lock);
			mutex_exit(hash_lock);

			bytes_evicted += evicted;

			/*
			 * If evicted is zero, arc_evict_hdr() must have
			 * decided to skip this header, don't increment
			 * evict_count in this case.
			 */
			if (evicted != 0)
				evict_count++;

			/*
			 * If arc_size isn't overflowing, signal any
			 * threads that might happen to be waiting.
			 *
			 * For each header evicted, we wake up a single
			 * thread. If we used cv_broadcast, we could
			 * wake up "too many" threads causing arc_size
			 * to significantly overflow arc_c; since
			 * arc_get_data_impl() doesn't check for overflow
			 * when it's woken up (it doesn't because it's
			 * possible for the ARC to be overflowing while
			 * full of un-evictable buffers, and the
			 * function should proceed in this case).
			 *
			 * If threads are left sleeping, due to not
			 * using cv_broadcast, they will be woken up
			 * just before arc_reclaim_thread() sleeps.
			 */
			mutex_enter(&arc_reclaim_lock);
			if (!arc_is_overflowing())
				cv_signal(&arc_reclaim_waiters_cv);
			mutex_exit(&arc_reclaim_lock);
		} else {
			ARCSTAT_BUMP(arcstat_mutex_miss);
		}
	}

	multilist_sublist_unlock(mls);

	return (bytes_evicted);
}

/*
 * Evict buffers from the given arc state, until we've removed the
 * specified number of bytes. Move the removed buffers to the
 * appropriate evict state.
 *
 * This function makes a "best effort". It skips over any buffers
 * it can't get a hash_lock on, and so, may not catch all candidates.
 * It may also return without evicting as much space as requested.
 *
 * If bytes is specified using the special value ARC_EVICT_ALL, this
 * will evict all available (i.e. unlocked and evictable) buffers from
 * the given arc state; which is used by arc_flush().
 */
static uint64_t
arc_evict_state(arc_state_t *state, uint64_t spa, int64_t bytes,
    arc_buf_contents_t type)
{
	uint64_t total_evicted = 0;
	multilist_t *ml = state->arcs_list[type];
	int num_sublists;
	arc_buf_hdr_t **markers;
	int i;

	IMPLY(bytes < 0, bytes == ARC_EVICT_ALL);

	num_sublists = multilist_get_num_sublists(ml);

	/*
	 * If we've tried to evict from each sublist, made some
	 * progress, but still have not hit the target number of bytes
	 * to evict, we want to keep trying. The markers allow us to
	 * pick up where we left off for each individual sublist, rather
	 * than starting from the tail each time.
	 */
	markers = kmem_zalloc(sizeof (*markers) * num_sublists, KM_SLEEP);
	for (i = 0; i < num_sublists; i++) {
		multilist_sublist_t *mls;

		markers[i] = kmem_cache_alloc(hdr_full_cache, KM_SLEEP);

		/*
		 * A b_spa of 0 is used to indicate that this header is
		 * a marker. This fact is used in arc_adjust_type() and
		 * arc_evict_state_impl().
		 */
		markers[i]->b_spa = 0;

		mls = multilist_sublist_lock(ml, i);
		multilist_sublist_insert_tail(mls, markers[i]);
		multilist_sublist_unlock(mls);
	}

	/*
	 * While we haven't hit our target number of bytes to evict, or
	 * we're evicting all available buffers.
	 */
	while (total_evicted < bytes || bytes == ARC_EVICT_ALL) {
		int sublist_idx = multilist_get_random_index(ml);
		uint64_t scan_evicted = 0;

		/*
		 * Try to reduce pinned dnodes with a floor of arc_dnode_limit.
		 * Request that 10% of the LRUs be scanned by the superblock
		 * shrinker.
		 */
		if (type == ARC_BUFC_DATA && arc_dnode_size > arc_dnode_limit)
			arc_prune_async((arc_dnode_size - arc_dnode_limit) /
			    sizeof (dnode_t) / zfs_arc_dnode_reduce_percent);

		/*
		 * Start eviction using a randomly selected sublist,
		 * this is to try and evenly balance eviction across all
		 * sublists. Always starting at the same sublist
		 * (e.g. index 0) would cause evictions to favor certain
		 * sublists over others.
		 */
		for (i = 0; i < num_sublists; i++) {
			uint64_t bytes_remaining;
			uint64_t bytes_evicted;

			if (bytes == ARC_EVICT_ALL)
				bytes_remaining = ARC_EVICT_ALL;
			else if (total_evicted < bytes)
				bytes_remaining = bytes - total_evicted;
			else
				break;

			bytes_evicted = arc_evict_state_impl(ml, sublist_idx,
			    markers[sublist_idx], spa, bytes_remaining);

			scan_evicted += bytes_evicted;
			total_evicted += bytes_evicted;

			/* we've reached the end, wrap to the beginning */
			if (++sublist_idx >= num_sublists)
				sublist_idx = 0;
		}

		/*
		 * If we didn't evict anything during this scan, we have
		 * no reason to believe we'll evict more during another
		 * scan, so break the loop.
		 */
		if (scan_evicted == 0) {
			/* This isn't possible, let's make that obvious */
			ASSERT3S(bytes, !=, 0);

			/*
			 * When bytes is ARC_EVICT_ALL, the only way to
			 * break the loop is when scan_evicted is zero.
			 * In that case, we actually have evicted enough,
			 * so we don't want to increment the kstat.
			 */
			if (bytes != ARC_EVICT_ALL) {
				ASSERT3S(total_evicted, <, bytes);
				ARCSTAT_BUMP(arcstat_evict_not_enough);
			}

			break;
		}
	}

	for (i = 0; i < num_sublists; i++) {
		multilist_sublist_t *mls = multilist_sublist_lock(ml, i);
		multilist_sublist_remove(mls, markers[i]);
		multilist_sublist_unlock(mls);

		kmem_cache_free(hdr_full_cache, markers[i]);
	}
	kmem_free(markers, sizeof (*markers) * num_sublists);

	return (total_evicted);
}

/*
 * Flush all "evictable" data of the given type from the arc state
 * specified. This will not evict any "active" buffers (i.e. referenced).
 *
 * When 'retry' is set to B_FALSE, the function will make a single pass
 * over the state and evict any buffers that it can. Since it doesn't
 * continually retry the eviction, it might end up leaving some buffers
 * in the ARC due to lock misses.
 *
 * When 'retry' is set to B_TRUE, the function will continually retry the
 * eviction until *all* evictable buffers have been removed from the
 * state. As a result, if concurrent insertions into the state are
 * allowed (e.g. if the ARC isn't shutting down), this function might
 * wind up in an infinite loop, continually trying to evict buffers.
 */
static uint64_t
arc_flush_state(arc_state_t *state, uint64_t spa, arc_buf_contents_t type,
    boolean_t retry)
{
	uint64_t evicted = 0;

	while (refcount_count(&state->arcs_esize[type]) != 0) {
		evicted += arc_evict_state(state, spa, ARC_EVICT_ALL, type);

		if (!retry)
			break;
	}

	return (evicted);
}

/*
 * Helper function for arc_prune_async() it is responsible for safely
 * handling the execution of a registered arc_prune_func_t.
 */
static void
arc_prune_task(void *ptr)
{
	arc_prune_t *ap = (arc_prune_t *)ptr;
	arc_prune_func_t *func = ap->p_pfunc;

	if (func != NULL)
		func(ap->p_adjust, ap->p_private);

	refcount_remove(&ap->p_refcnt, func);
}

/*
 * Notify registered consumers they must drop holds on a portion of the ARC
 * buffered they reference.  This provides a mechanism to ensure the ARC can
 * honor the arc_meta_limit and reclaim otherwise pinned ARC buffers.  This
 * is analogous to dnlc_reduce_cache() but more generic.
 *
 * This operation is performed asynchronously so it may be safely called
 * in the context of the arc_reclaim_thread().  A reference is taken here
 * for each registered arc_prune_t and the arc_prune_task() is responsible
 * for releasing it once the registered arc_prune_func_t has completed.
 */
static void
arc_prune_async(int64_t adjust)
{
	arc_prune_t *ap;

	mutex_enter(&arc_prune_mtx);
	for (ap = list_head(&arc_prune_list); ap != NULL;
	    ap = list_next(&arc_prune_list, ap)) {

		if (refcount_count(&ap->p_refcnt) >= 2)
			continue;

		refcount_add(&ap->p_refcnt, ap->p_pfunc);
		ap->p_adjust = adjust;
		if (taskq_dispatch(arc_prune_taskq, arc_prune_task,
		    ap, TQ_SLEEP) == TASKQID_INVALID) {
			refcount_remove(&ap->p_refcnt, ap->p_pfunc);
			continue;
		}
		ARCSTAT_BUMP(arcstat_prune);
	}
	mutex_exit(&arc_prune_mtx);
}

/*
 * Evict the specified number of bytes from the state specified,
 * restricting eviction to the spa and type given. This function
 * prevents us from trying to evict more from a state's list than
 * is "evictable", and to skip evicting altogether when passed a
 * negative value for "bytes". In contrast, arc_evict_state() will
 * evict everything it can, when passed a negative value for "bytes".
 */
static uint64_t
arc_adjust_impl(arc_state_t *state, uint64_t spa, int64_t bytes,
    arc_buf_contents_t type)
{
	int64_t delta;

	if (bytes > 0 && refcount_count(&state->arcs_esize[type]) > 0) {
		delta = MIN(refcount_count(&state->arcs_esize[type]), bytes);
		return (arc_evict_state(state, spa, delta, type));
	}

	return (0);
}

/*
 * The goal of this function is to evict enough meta data buffers from the
 * ARC in order to enforce the arc_meta_limit.  Achieving this is slightly
 * more complicated than it appears because it is common for data buffers
 * to have holds on meta data buffers.  In addition, dnode meta data buffers
 * will be held by the dnodes in the block preventing them from being freed.
 * This means we can't simply traverse the ARC and expect to always find
 * enough unheld meta data buffer to release.
 *
 * Therefore, this function has been updated to make alternating passes
 * over the ARC releasing data buffers and then newly unheld meta data
 * buffers.  This ensures forward progress is maintained and arc_meta_used
 * will decrease.  Normally this is sufficient, but if required the ARC
 * will call the registered prune callbacks causing dentry and inodes to
 * be dropped from the VFS cache.  This will make dnode meta data buffers
 * available for reclaim.
 */
static uint64_t
arc_adjust_meta_balanced(void)
{
	int64_t delta, prune = 0, adjustmnt;
	uint64_t total_evicted = 0;
	arc_buf_contents_t type = ARC_BUFC_DATA;
	int restarts = MAX(zfs_arc_meta_adjust_restarts, 0);

restart:
	/*
	 * This slightly differs than the way we evict from the mru in
	 * arc_adjust because we don't have a "target" value (i.e. no
	 * "meta" arc_p). As a result, I think we can completely
	 * cannibalize the metadata in the MRU before we evict the
	 * metadata from the MFU. I think we probably need to implement a
	 * "metadata arc_p" value to do this properly.
	 */
	adjustmnt = arc_meta_used - arc_meta_limit;

	if (adjustmnt > 0 && refcount_count(&arc_mru->arcs_esize[type]) > 0) {
		delta = MIN(refcount_count(&arc_mru->arcs_esize[type]),
		    adjustmnt);
		total_evicted += arc_adjust_impl(arc_mru, 0, delta, type);
		adjustmnt -= delta;
	}

	/*
	 * We can't afford to recalculate adjustmnt here. If we do,
	 * new metadata buffers can sneak into the MRU or ANON lists,
	 * thus penalize the MFU metadata. Although the fudge factor is
	 * small, it has been empirically shown to be significant for
	 * certain workloads (e.g. creating many empty directories). As
	 * such, we use the original calculation for adjustmnt, and
	 * simply decrement the amount of data evicted from the MRU.
	 */

	if (adjustmnt > 0 && refcount_count(&arc_mfu->arcs_esize[type]) > 0) {
		delta = MIN(refcount_count(&arc_mfu->arcs_esize[type]),
		    adjustmnt);
		total_evicted += arc_adjust_impl(arc_mfu, 0, delta, type);
	}

	adjustmnt = arc_meta_used - arc_meta_limit;

	if (adjustmnt > 0 &&
	    refcount_count(&arc_mru_ghost->arcs_esize[type]) > 0) {
		delta = MIN(adjustmnt,
		    refcount_count(&arc_mru_ghost->arcs_esize[type]));
		total_evicted += arc_adjust_impl(arc_mru_ghost, 0, delta, type);
		adjustmnt -= delta;
	}

	if (adjustmnt > 0 &&
	    refcount_count(&arc_mfu_ghost->arcs_esize[type]) > 0) {
		delta = MIN(adjustmnt,
		    refcount_count(&arc_mfu_ghost->arcs_esize[type]));
		total_evicted += arc_adjust_impl(arc_mfu_ghost, 0, delta, type);
	}

	/*
	 * If after attempting to make the requested adjustment to the ARC
	 * the meta limit is still being exceeded then request that the
	 * higher layers drop some cached objects which have holds on ARC
	 * meta buffers.  Requests to the upper layers will be made with
	 * increasingly large scan sizes until the ARC is below the limit.
	 */
	if (arc_meta_used > arc_meta_limit) {
		if (type == ARC_BUFC_DATA) {
			type = ARC_BUFC_METADATA;
		} else {
			type = ARC_BUFC_DATA;

			if (zfs_arc_meta_prune) {
				prune += zfs_arc_meta_prune;
				arc_prune_async(prune);
			}
		}

		if (restarts > 0) {
			restarts--;
			goto restart;
		}
	}
	return (total_evicted);
}

/*
 * Evict metadata buffers from the cache, such that arc_meta_used is
 * capped by the arc_meta_limit tunable.
 */
static uint64_t
arc_adjust_meta_only(void)
{
	uint64_t total_evicted = 0;
	int64_t target;

	/*
	 * If we're over the meta limit, we want to evict enough
	 * metadata to get back under the meta limit. We don't want to
	 * evict so much that we drop the MRU below arc_p, though. If
	 * we're over the meta limit more than we're over arc_p, we
	 * evict some from the MRU here, and some from the MFU below.
	 */
	target = MIN((int64_t)(arc_meta_used - arc_meta_limit),
	    (int64_t)(refcount_count(&arc_anon->arcs_size) +
	    refcount_count(&arc_mru->arcs_size) - arc_p));

	total_evicted += arc_adjust_impl(arc_mru, 0, target, ARC_BUFC_METADATA);

	/*
	 * Similar to the above, we want to evict enough bytes to get us
	 * below the meta limit, but not so much as to drop us below the
	 * space allotted to the MFU (which is defined as arc_c - arc_p).
	 */
	target = MIN((int64_t)(arc_meta_used - arc_meta_limit),
	    (int64_t)(refcount_count(&arc_mfu->arcs_size) - (arc_c - arc_p)));

	total_evicted += arc_adjust_impl(arc_mfu, 0, target, ARC_BUFC_METADATA);

	return (total_evicted);
}

static uint64_t
arc_adjust_meta(void)
{
	if (zfs_arc_meta_strategy == ARC_STRATEGY_META_ONLY)
		return (arc_adjust_meta_only());
	else
		return (arc_adjust_meta_balanced());
}

/*
 * Return the type of the oldest buffer in the given arc state
 *
 * This function will select a random sublist of type ARC_BUFC_DATA and
 * a random sublist of type ARC_BUFC_METADATA. The tail of each sublist
 * is compared, and the type which contains the "older" buffer will be
 * returned.
 */
static arc_buf_contents_t
arc_adjust_type(arc_state_t *state)
{
	multilist_t *data_ml = state->arcs_list[ARC_BUFC_DATA];
	multilist_t *meta_ml = state->arcs_list[ARC_BUFC_METADATA];
	int data_idx = multilist_get_random_index(data_ml);
	int meta_idx = multilist_get_random_index(meta_ml);
	multilist_sublist_t *data_mls;
	multilist_sublist_t *meta_mls;
	arc_buf_contents_t type;
	arc_buf_hdr_t *data_hdr;
	arc_buf_hdr_t *meta_hdr;

	/*
	 * We keep the sublist lock until we're finished, to prevent
	 * the headers from being destroyed via arc_evict_state().
	 */
	data_mls = multilist_sublist_lock(data_ml, data_idx);
	meta_mls = multilist_sublist_lock(meta_ml, meta_idx);

	/*
	 * These two loops are to ensure we skip any markers that
	 * might be at the tail of the lists due to arc_evict_state().
	 */

	for (data_hdr = multilist_sublist_tail(data_mls); data_hdr != NULL;
	    data_hdr = multilist_sublist_prev(data_mls, data_hdr)) {
		if (data_hdr->b_spa != 0)
			break;
	}

	for (meta_hdr = multilist_sublist_tail(meta_mls); meta_hdr != NULL;
	    meta_hdr = multilist_sublist_prev(meta_mls, meta_hdr)) {
		if (meta_hdr->b_spa != 0)
			break;
	}

	if (data_hdr == NULL && meta_hdr == NULL) {
		type = ARC_BUFC_DATA;
	} else if (data_hdr == NULL) {
		ASSERT3P(meta_hdr, !=, NULL);
		type = ARC_BUFC_METADATA;
	} else if (meta_hdr == NULL) {
		ASSERT3P(data_hdr, !=, NULL);
		type = ARC_BUFC_DATA;
	} else {
		ASSERT3P(data_hdr, !=, NULL);
		ASSERT3P(meta_hdr, !=, NULL);

		/* The headers can't be on the sublist without an L1 header */
		ASSERT(HDR_HAS_L1HDR(data_hdr));
		ASSERT(HDR_HAS_L1HDR(meta_hdr));

		if (data_hdr->b_l1hdr.b_arc_access <
		    meta_hdr->b_l1hdr.b_arc_access) {
			type = ARC_BUFC_DATA;
		} else {
			type = ARC_BUFC_METADATA;
		}
	}

	multilist_sublist_unlock(meta_mls);
	multilist_sublist_unlock(data_mls);

	return (type);
}

/*
 * Evict buffers from the cache, such that arc_size is capped by arc_c.
 */
static uint64_t
arc_adjust(void)
{
	uint64_t total_evicted = 0;
	uint64_t bytes;
	int64_t target;

	/*
	 * If we're over arc_meta_limit, we want to correct that before
	 * potentially evicting data buffers below.
	 */
	total_evicted += arc_adjust_meta();

	/*
	 * Adjust MRU size
	 *
	 * If we're over the target cache size, we want to evict enough
	 * from the list to get back to our target size. We don't want
	 * to evict too much from the MRU, such that it drops below
	 * arc_p. So, if we're over our target cache size more than
	 * the MRU is over arc_p, we'll evict enough to get back to
	 * arc_p here, and then evict more from the MFU below.
	 */
	target = MIN((int64_t)(arc_size - arc_c),
	    (int64_t)(refcount_count(&arc_anon->arcs_size) +
	    refcount_count(&arc_mru->arcs_size) + arc_meta_used - arc_p));

	/*
	 * If we're below arc_meta_min, always prefer to evict data.
	 * Otherwise, try to satisfy the requested number of bytes to
	 * evict from the type which contains older buffers; in an
	 * effort to keep newer buffers in the cache regardless of their
	 * type. If we cannot satisfy the number of bytes from this
	 * type, spill over into the next type.
	 */
	if (arc_adjust_type(arc_mru) == ARC_BUFC_METADATA &&
	    arc_meta_used > arc_meta_min) {
		bytes = arc_adjust_impl(arc_mru, 0, target, ARC_BUFC_METADATA);
		total_evicted += bytes;

		/*
		 * If we couldn't evict our target number of bytes from
		 * metadata, we try to get the rest from data.
		 */
		target -= bytes;

		total_evicted +=
		    arc_adjust_impl(arc_mru, 0, target, ARC_BUFC_DATA);
	} else {
		bytes = arc_adjust_impl(arc_mru, 0, target, ARC_BUFC_DATA);
		total_evicted += bytes;

		/*
		 * If we couldn't evict our target number of bytes from
		 * data, we try to get the rest from metadata.
		 */
		target -= bytes;

		total_evicted +=
		    arc_adjust_impl(arc_mru, 0, target, ARC_BUFC_METADATA);
	}

	/*
	 * Adjust MFU size
	 *
	 * Now that we've tried to evict enough from the MRU to get its
	 * size back to arc_p, if we're still above the target cache
	 * size, we evict the rest from the MFU.
	 */
	target = arc_size - arc_c;

	if (arc_adjust_type(arc_mfu) == ARC_BUFC_METADATA &&
	    arc_meta_used > arc_meta_min) {
		bytes = arc_adjust_impl(arc_mfu, 0, target, ARC_BUFC_METADATA);
		total_evicted += bytes;

		/*
		 * If we couldn't evict our target number of bytes from
		 * metadata, we try to get the rest from data.
		 */
		target -= bytes;

		total_evicted +=
		    arc_adjust_impl(arc_mfu, 0, target, ARC_BUFC_DATA);
	} else {
		bytes = arc_adjust_impl(arc_mfu, 0, target, ARC_BUFC_DATA);
		total_evicted += bytes;

		/*
		 * If we couldn't evict our target number of bytes from
		 * data, we try to get the rest from data.
		 */
		target -= bytes;

		total_evicted +=
		    arc_adjust_impl(arc_mfu, 0, target, ARC_BUFC_METADATA);
	}

	/*
	 * Adjust ghost lists
	 *
	 * In addition to the above, the ARC also defines target values
	 * for the ghost lists. The sum of the mru list and mru ghost
	 * list should never exceed the target size of the cache, and
	 * the sum of the mru list, mfu list, mru ghost list, and mfu
	 * ghost list should never exceed twice the target size of the
	 * cache. The following logic enforces these limits on the ghost
	 * caches, and evicts from them as needed.
	 */
	target = refcount_count(&arc_mru->arcs_size) +
	    refcount_count(&arc_mru_ghost->arcs_size) - arc_c;

	bytes = arc_adjust_impl(arc_mru_ghost, 0, target, ARC_BUFC_DATA);
	total_evicted += bytes;

	target -= bytes;

	total_evicted +=
	    arc_adjust_impl(arc_mru_ghost, 0, target, ARC_BUFC_METADATA);

	/*
	 * We assume the sum of the mru list and mfu list is less than
	 * or equal to arc_c (we enforced this above), which means we
	 * can use the simpler of the two equations below:
	 *
	 *	mru + mfu + mru ghost + mfu ghost <= 2 * arc_c
	 *		    mru ghost + mfu ghost <= arc_c
	 */
	target = refcount_count(&arc_mru_ghost->arcs_size) +
	    refcount_count(&arc_mfu_ghost->arcs_size) - arc_c;

	bytes = arc_adjust_impl(arc_mfu_ghost, 0, target, ARC_BUFC_DATA);
	total_evicted += bytes;

	target -= bytes;

	total_evicted +=
	    arc_adjust_impl(arc_mfu_ghost, 0, target, ARC_BUFC_METADATA);

	return (total_evicted);
}

void
arc_flush(spa_t *spa, boolean_t retry)
{
	uint64_t guid = 0;

	/*
	 * If retry is B_TRUE, a spa must not be specified since we have
	 * no good way to determine if all of a spa's buffers have been
	 * evicted from an arc state.
	 */
	ASSERT(!retry || spa == 0);

	if (spa != NULL)
		guid = spa_load_guid(spa);

	(void) arc_flush_state(arc_mru, guid, ARC_BUFC_DATA, retry);
	(void) arc_flush_state(arc_mru, guid, ARC_BUFC_METADATA, retry);

	(void) arc_flush_state(arc_mfu, guid, ARC_BUFC_DATA, retry);
	(void) arc_flush_state(arc_mfu, guid, ARC_BUFC_METADATA, retry);

	(void) arc_flush_state(arc_mru_ghost, guid, ARC_BUFC_DATA, retry);
	(void) arc_flush_state(arc_mru_ghost, guid, ARC_BUFC_METADATA, retry);

	(void) arc_flush_state(arc_mfu_ghost, guid, ARC_BUFC_DATA, retry);
	(void) arc_flush_state(arc_mfu_ghost, guid, ARC_BUFC_METADATA, retry);
}

void
arc_shrink(int64_t to_free)
{
	uint64_t c = arc_c;

	if (c > to_free && c - to_free > arc_c_min) {
		arc_c = c - to_free;
		atomic_add_64(&arc_p, -(arc_p >> arc_shrink_shift));
		if (arc_c > arc_size)
			arc_c = MAX(arc_size, arc_c_min);
		if (arc_p > arc_c)
			arc_p = (arc_c >> 1);
		ASSERT(arc_c >= arc_c_min);
		ASSERT((int64_t)arc_p >= 0);
	} else {
		arc_c = arc_c_min;
	}

	if (arc_size > arc_c)
		(void) arc_adjust();
}

/*
 * Return maximum amount of memory that we could possibly use.  Reduced
 * to half of all memory in user space which is primarily used for testing.
 */
static uint64_t
arc_all_memory(void)
{
#ifdef _KERNEL
	return (MIN(ptob(physmem),
	    vmem_size(heap_arena, VMEM_FREE | VMEM_ALLOC)));
#else
	return (ptob(physmem) / 2);
#endif
}

typedef enum free_memory_reason_t {
	FMR_UNKNOWN,
	FMR_NEEDFREE,
	FMR_LOTSFREE,
	FMR_SWAPFS_MINFREE,
	FMR_PAGES_PP_MAXIMUM,
	FMR_HEAP_ARENA,
	FMR_ZIO_ARENA,
} free_memory_reason_t;

int64_t last_free_memory;
free_memory_reason_t last_free_reason;

#ifdef _KERNEL
/*
 * Additional reserve of pages for pp_reserve.
 */
int64_t arc_pages_pp_reserve = 64;

/*
 * Additional reserve of pages for swapfs.
 */
int64_t arc_swapfs_reserve = 64;
#endif /* _KERNEL */

/*
 * Return the amount of memory that can be consumed before reclaim will be
 * needed.  Positive if there is sufficient free memory, negative indicates
 * the amount of memory that needs to be freed up.
 */
static int64_t
arc_available_memory(void)
{
	int64_t lowest = INT64_MAX;
	free_memory_reason_t r = FMR_UNKNOWN;
#ifdef _KERNEL
	uint64_t available_memory = ptob(freemem);
	int64_t n;
#ifdef __linux__
	pgcnt_t needfree = btop(arc_need_free);
	pgcnt_t lotsfree = btop(arc_sys_free);
	pgcnt_t desfree = 0;
#endif

#if defined(__i386)
	available_memory =
	    MIN(available_memory, vmem_size(heap_arena, VMEM_FREE));
#endif

	if (needfree > 0) {
		n = PAGESIZE * (-needfree);
		if (n < lowest) {
			lowest = n;
			r = FMR_NEEDFREE;
		}
	}

	/*
	 * check that we're out of range of the pageout scanner.  It starts to
	 * schedule paging if freemem is less than lotsfree and needfree.
	 * lotsfree is the high-water mark for pageout, and needfree is the
	 * number of needed free pages.  We add extra pages here to make sure
	 * the scanner doesn't start up while we're freeing memory.
	 */
	n = PAGESIZE * (btop(available_memory) - lotsfree - needfree - desfree);
	if (n < lowest) {
		lowest = n;
		r = FMR_LOTSFREE;
	}

#ifndef __linux__
	/*
	 * check to make sure that swapfs has enough space so that anon
	 * reservations can still succeed. anon_resvmem() checks that the
	 * availrmem is greater than swapfs_minfree, and the number of reserved
	 * swap pages.  We also add a bit of extra here just to prevent
	 * circumstances from getting really dire.
	 */
	n = PAGESIZE * (availrmem - swapfs_minfree - swapfs_reserve -
	    desfree - arc_swapfs_reserve);
	if (n < lowest) {
		lowest = n;
		r = FMR_SWAPFS_MINFREE;
	}


	/*
	 * Check that we have enough availrmem that memory locking (e.g., via
	 * mlock(3C) or memcntl(2)) can still succeed.  (pages_pp_maximum
	 * stores the number of pages that cannot be locked; when availrmem
	 * drops below pages_pp_maximum, page locking mechanisms such as
	 * page_pp_lock() will fail.)
	 */
	n = PAGESIZE * (availrmem - pages_pp_maximum -
	    arc_pages_pp_reserve);
	if (n < lowest) {
		lowest = n;
		r = FMR_PAGES_PP_MAXIMUM;
	}
#endif

#if defined(__i386)
	/*
	 * If we're on an i386 platform, it's possible that we'll exhaust the
	 * kernel heap space before we ever run out of available physical
	 * memory.  Most checks of the size of the heap_area compare against
	 * tune.t_minarmem, which is the minimum available real memory that we
	 * can have in the system.  However, this is generally fixed at 25 pages
	 * which is so low that it's useless.  In this comparison, we seek to
	 * calculate the total heap-size, and reclaim if more than 3/4ths of the
	 * heap is allocated.  (Or, in the calculation, if less than 1/4th is
	 * free)
	 */
	n = vmem_size(heap_arena, VMEM_FREE) -
	    (vmem_size(heap_arena, VMEM_FREE | VMEM_ALLOC) >> 2);
	if (n < lowest) {
		lowest = n;
		r = FMR_HEAP_ARENA;
	}
#endif

	/*
	 * If zio data pages are being allocated out of a separate heap segment,
	 * then enforce that the size of available vmem for this arena remains
	 * above about 1/4th (1/(2^arc_zio_arena_free_shift)) free.
	 *
	 * Note that reducing the arc_zio_arena_free_shift keeps more virtual
	 * memory (in the zio_arena) free, which can avoid memory
	 * fragmentation issues.
	 */
	if (zio_arena != NULL) {
		n = (int64_t)vmem_size(zio_arena, VMEM_FREE) -
		    (vmem_size(zio_arena, VMEM_ALLOC) >>
		    arc_zio_arena_free_shift);
		if (n < lowest) {
			lowest = n;
			r = FMR_ZIO_ARENA;
		}
	}
#else /* _KERNEL */
	/* Every 100 calls, free a small amount */
	if (spa_get_random(100) == 0)
		lowest = -1024;
#endif /* _KERNEL */

	last_free_memory = lowest;
	last_free_reason = r;

	return (lowest);
}

/*
 * Determine if the system is under memory pressure and is asking
 * to reclaim memory. A return value of B_TRUE indicates that the system
 * is under memory pressure and that the arc should adjust accordingly.
 */
static boolean_t
arc_reclaim_needed(void)
{
	return (arc_available_memory() < 0);
}

static void
arc_kmem_reap_now(void)
{
	size_t			i;
	kmem_cache_t		*prev_cache = NULL;
	kmem_cache_t		*prev_data_cache = NULL;
	extern kmem_cache_t	*zio_buf_cache[];
	extern kmem_cache_t	*zio_data_buf_cache[];
	extern kmem_cache_t	*range_seg_cache;

	if ((arc_meta_used >= arc_meta_limit) && zfs_arc_meta_prune) {
		/*
		 * We are exceeding our meta-data cache limit.
		 * Prune some entries to release holds on meta-data.
		 */
		arc_prune_async(zfs_arc_meta_prune);
	}

	for (i = 0; i < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; i++) {
#ifdef _ILP32
		/* reach upper limit of cache size on 32-bit */
		if (zio_buf_cache[i] == NULL)
			break;
#endif
		if (zio_buf_cache[i] != prev_cache) {
			prev_cache = zio_buf_cache[i];
			kmem_cache_reap_now(zio_buf_cache[i]);
		}
		if (zio_data_buf_cache[i] != prev_data_cache) {
			prev_data_cache = zio_data_buf_cache[i];
			kmem_cache_reap_now(zio_data_buf_cache[i]);
		}
	}
	kmem_cache_reap_now(buf_cache);
	kmem_cache_reap_now(hdr_full_cache);
	kmem_cache_reap_now(hdr_l2only_cache);
	kmem_cache_reap_now(range_seg_cache);

	if (zio_arena != NULL) {
		/*
		 * Ask the vmem arena to reclaim unused memory from its
		 * quantum caches.
		 */
		vmem_qcache_reap(zio_arena);
	}
}

/*
 * Threads can block in arc_get_data_impl() waiting for this thread to evict
 * enough data and signal them to proceed. When this happens, the threads in
 * arc_get_data_impl() are sleeping while holding the hash lock for their
 * particular arc header. Thus, we must be careful to never sleep on a
 * hash lock in this thread. This is to prevent the following deadlock:
 *
 *  - Thread A sleeps on CV in arc_get_data_impl() holding hash lock "L",
 *    waiting for the reclaim thread to signal it.
 *
 *  - arc_reclaim_thread() tries to acquire hash lock "L" using mutex_enter,
 *    fails, and goes to sleep forever.
 *
 * This possible deadlock is avoided by always acquiring a hash lock
 * using mutex_tryenter() from arc_reclaim_thread().
 */
static void
arc_reclaim_thread(void)
{
	fstrans_cookie_t	cookie = spl_fstrans_mark();
	hrtime_t		growtime = 0;
	callb_cpr_t		cpr;

	CALLB_CPR_INIT(&cpr, &arc_reclaim_lock, callb_generic_cpr, FTAG);

	mutex_enter(&arc_reclaim_lock);
	while (!arc_reclaim_thread_exit) {
		int64_t to_free;
		uint64_t evicted = 0;
		uint64_t need_free = arc_need_free;
		arc_tuning_update();

		/*
		 * This is necessary in order for the mdb ::arc dcmd to
		 * show up to date information. Since the ::arc command
		 * does not call the kstat's update function, without
		 * this call, the command may show stale stats for the
		 * anon, mru, mru_ghost, mfu, and mfu_ghost lists. Even
		 * with this change, the data might be up to 1 second
		 * out of date; but that should suffice. The arc_state_t
		 * structures can be queried directly if more accurate
		 * information is needed.
		 */
#ifndef __linux__
		if (arc_ksp != NULL)
			arc_ksp->ks_update(arc_ksp, KSTAT_READ);
#endif
		mutex_exit(&arc_reclaim_lock);

		/*
		 * We call arc_adjust() before (possibly) calling
		 * arc_kmem_reap_now(), so that we can wake up
		 * arc_get_data_buf() sooner.
		 */
		evicted = arc_adjust();

		int64_t free_memory = arc_available_memory();
		if (free_memory < 0) {

			arc_no_grow = B_TRUE;
			arc_warm = B_TRUE;

			/*
			 * Wait at least zfs_grow_retry (default 5) seconds
			 * before considering growing.
			 */
			growtime = gethrtime() + SEC2NSEC(arc_grow_retry);

			arc_kmem_reap_now();

			/*
			 * If we are still low on memory, shrink the ARC
			 * so that we have arc_shrink_min free space.
			 */
			free_memory = arc_available_memory();

			to_free = (arc_c >> arc_shrink_shift) - free_memory;
			if (to_free > 0) {
#ifdef _KERNEL
				to_free = MAX(to_free, need_free);
#endif
				arc_shrink(to_free);
			}
		} else if (free_memory < arc_c >> arc_no_grow_shift) {
			arc_no_grow = B_TRUE;
		} else if (gethrtime() >= growtime) {
			arc_no_grow = B_FALSE;
		}

		mutex_enter(&arc_reclaim_lock);

		/*
		 * If evicted is zero, we couldn't evict anything via
		 * arc_adjust(). This could be due to hash lock
		 * collisions, but more likely due to the majority of
		 * arc buffers being unevictable. Therefore, even if
		 * arc_size is above arc_c, another pass is unlikely to
		 * be helpful and could potentially cause us to enter an
		 * infinite loop.
		 */
		if (arc_size <= arc_c || evicted == 0) {
			/*
			 * We're either no longer overflowing, or we
			 * can't evict anything more, so we should wake
			 * up any threads before we go to sleep and remove
			 * the bytes we were working on from arc_need_free
			 * since nothing more will be done here.
			 */
			cv_broadcast(&arc_reclaim_waiters_cv);
			ARCSTAT_INCR(arcstat_need_free, -need_free);

			/*
			 * Block until signaled, or after one second (we
			 * might need to perform arc_kmem_reap_now()
			 * even if we aren't being signalled)
			 */
			CALLB_CPR_SAFE_BEGIN(&cpr);
			(void) cv_timedwait_sig_hires(&arc_reclaim_thread_cv,
			    &arc_reclaim_lock, SEC2NSEC(1), MSEC2NSEC(1), 0);
			CALLB_CPR_SAFE_END(&cpr, &arc_reclaim_lock);
		}
	}

	arc_reclaim_thread_exit = B_FALSE;
	cv_broadcast(&arc_reclaim_thread_cv);
	CALLB_CPR_EXIT(&cpr);		/* drops arc_reclaim_lock */
	spl_fstrans_unmark(cookie);
	thread_exit();
}

#ifdef _KERNEL
/*
 * Determine the amount of memory eligible for eviction contained in the
 * ARC. All clean data reported by the ghost lists can always be safely
 * evicted. Due to arc_c_min, the same does not hold for all clean data
 * contained by the regular mru and mfu lists.
 *
 * In the case of the regular mru and mfu lists, we need to report as
 * much clean data as possible, such that evicting that same reported
 * data will not bring arc_size below arc_c_min. Thus, in certain
 * circumstances, the total amount of clean data in the mru and mfu
 * lists might not actually be evictable.
 *
 * The following two distinct cases are accounted for:
 *
 * 1. The sum of the amount of dirty data contained by both the mru and
 *    mfu lists, plus the ARC's other accounting (e.g. the anon list),
 *    is greater than or equal to arc_c_min.
 *    (i.e. amount of dirty data >= arc_c_min)
 *
 *    This is the easy case; all clean data contained by the mru and mfu
 *    lists is evictable. Evicting all clean data can only drop arc_size
 *    to the amount of dirty data, which is greater than arc_c_min.
 *
 * 2. The sum of the amount of dirty data contained by both the mru and
 *    mfu lists, plus the ARC's other accounting (e.g. the anon list),
 *    is less than arc_c_min.
 *    (i.e. arc_c_min > amount of dirty data)
 *
 *    2.1. arc_size is greater than or equal arc_c_min.
 *         (i.e. arc_size >= arc_c_min > amount of dirty data)
 *
 *         In this case, not all clean data from the regular mru and mfu
 *         lists is actually evictable; we must leave enough clean data
 *         to keep arc_size above arc_c_min. Thus, the maximum amount of
 *         evictable data from the two lists combined, is exactly the
 *         difference between arc_size and arc_c_min.
 *
 *    2.2. arc_size is less than arc_c_min
 *         (i.e. arc_c_min > arc_size > amount of dirty data)
 *
 *         In this case, none of the data contained in the mru and mfu
 *         lists is evictable, even if it's clean. Since arc_size is
 *         already below arc_c_min, evicting any more would only
 *         increase this negative difference.
 */
static uint64_t
arc_evictable_memory(void)
{
	uint64_t arc_clean =
	    refcount_count(&arc_mru->arcs_esize[ARC_BUFC_DATA]) +
	    refcount_count(&arc_mru->arcs_esize[ARC_BUFC_METADATA]) +
	    refcount_count(&arc_mfu->arcs_esize[ARC_BUFC_DATA]) +
	    refcount_count(&arc_mfu->arcs_esize[ARC_BUFC_METADATA]);
	uint64_t arc_dirty = MAX((int64_t)arc_size - (int64_t)arc_clean, 0);

	if (arc_dirty >= arc_c_min)
		return (arc_clean);

	return (MAX((int64_t)arc_size - (int64_t)arc_c_min, 0));
}

/*
 * If sc->nr_to_scan is zero, the caller is requesting a query of the
 * number of objects which can potentially be freed.  If it is nonzero,
 * the request is to free that many objects.
 *
 * Linux kernels >= 3.12 have the count_objects and scan_objects callbacks
 * in struct shrinker and also require the shrinker to return the number
 * of objects freed.
 *
 * Older kernels require the shrinker to return the number of freeable
 * objects following the freeing of nr_to_free.
 */
static spl_shrinker_t
__arc_shrinker_func(struct shrinker *shrink, struct shrink_control *sc)
{
	int64_t pages;

	/* The arc is considered warm once reclaim has occurred */
	if (unlikely(arc_warm == B_FALSE))
		arc_warm = B_TRUE;

	/* Return the potential number of reclaimable pages */
	pages = btop((int64_t)arc_evictable_memory());
	if (sc->nr_to_scan == 0)
		return (pages);

	/* Not allowed to perform filesystem reclaim */
	if (!(sc->gfp_mask & __GFP_FS))
		return (SHRINK_STOP);

	/* Reclaim in progress */
	if (mutex_tryenter(&arc_reclaim_lock) == 0) {
		ARCSTAT_INCR(arcstat_need_free, ptob(sc->nr_to_scan));
		return (0);
	}

	mutex_exit(&arc_reclaim_lock);

	/*
	 * Evict the requested number of pages by shrinking arc_c the
	 * requested amount.  If there is nothing left to evict just
	 * reap whatever we can from the various arc slabs.
	 */
	if (pages > 0) {
		arc_shrink(ptob(sc->nr_to_scan));
		arc_kmem_reap_now();
#ifdef HAVE_SPLIT_SHRINKER_CALLBACK
		pages = MAX(pages - btop(arc_evictable_memory()), 0);
#else
		pages = btop(arc_evictable_memory());
#endif
		/*
		 * We've shrunk what we can, wake up threads.
		 */
		cv_broadcast(&arc_reclaim_waiters_cv);

	} else {
		arc_kmem_reap_now();
		pages = SHRINK_STOP;
	}

	/*
	 * When direct reclaim is observed it usually indicates a rapid
	 * increase in memory pressure.  This occurs because the kswapd
	 * threads were unable to asynchronously keep enough free memory
	 * available.  In this case set arc_no_grow to briefly pause arc
	 * growth to avoid compounding the memory pressure.
	 */
	if (current_is_kswapd()) {
		ARCSTAT_BUMP(arcstat_memory_indirect_count);
	} else {
		arc_no_grow = B_TRUE;
		ARCSTAT_BUMP(arcstat_memory_direct_count);
	}

	return (pages);
}
SPL_SHRINKER_CALLBACK_WRAPPER(arc_shrinker_func);

SPL_SHRINKER_DECLARE(arc_shrinker, arc_shrinker_func, DEFAULT_SEEKS);
#endif /* _KERNEL */

/*
 * Adapt arc info given the number of bytes we are trying to add and
 * the state that we are coming from.  This function is only called
 * when we are adding new content to the cache.
 */
static void
arc_adapt(int bytes, arc_state_t *state)
{
	int mult;
	uint64_t arc_p_min = (arc_c >> arc_p_min_shift);
	int64_t mrug_size = refcount_count(&arc_mru_ghost->arcs_size);
	int64_t mfug_size = refcount_count(&arc_mfu_ghost->arcs_size);

	if (state == arc_l2c_only)
		return;

	ASSERT(bytes > 0);
	/*
	 * Adapt the target size of the MRU list:
	 *	- if we just hit in the MRU ghost list, then increase
	 *	  the target size of the MRU list.
	 *	- if we just hit in the MFU ghost list, then increase
	 *	  the target size of the MFU list by decreasing the
	 *	  target size of the MRU list.
	 */
	if (state == arc_mru_ghost) {
		mult = (mrug_size >= mfug_size) ? 1 : (mfug_size / mrug_size);
		if (!zfs_arc_p_dampener_disable)
			mult = MIN(mult, 10); /* avoid wild arc_p adjustment */

		arc_p = MIN(arc_c - arc_p_min, arc_p + bytes * mult);
	} else if (state == arc_mfu_ghost) {
		uint64_t delta;

		mult = (mfug_size >= mrug_size) ? 1 : (mrug_size / mfug_size);
		if (!zfs_arc_p_dampener_disable)
			mult = MIN(mult, 10);

		delta = MIN(bytes * mult, arc_p);
		arc_p = MAX(arc_p_min, arc_p - delta);
	}
	ASSERT((int64_t)arc_p >= 0);

	if (arc_reclaim_needed()) {
		cv_signal(&arc_reclaim_thread_cv);
		return;
	}

	if (arc_no_grow)
		return;

	if (arc_c >= arc_c_max)
		return;

	/*
	 * If we're within (2 * maxblocksize) bytes of the target
	 * cache size, increment the target cache size
	 */
	ASSERT3U(arc_c, >=, 2ULL << SPA_MAXBLOCKSHIFT);
	if (arc_size >= arc_c - (2ULL << SPA_MAXBLOCKSHIFT)) {
		atomic_add_64(&arc_c, (int64_t)bytes);
		if (arc_c > arc_c_max)
			arc_c = arc_c_max;
		else if (state == arc_anon)
			atomic_add_64(&arc_p, (int64_t)bytes);
		if (arc_p > arc_c)
			arc_p = arc_c;
	}
	ASSERT((int64_t)arc_p >= 0);
}

/*
 * Check if arc_size has grown past our upper threshold, determined by
 * zfs_arc_overflow_shift.
 */
static boolean_t
arc_is_overflowing(void)
{
	/* Always allow at least one block of overflow */
	uint64_t overflow = MAX(SPA_MAXBLOCKSIZE,
	    arc_c >> zfs_arc_overflow_shift);

	return (arc_size >= arc_c + overflow);
}

static abd_t *
arc_get_data_abd(arc_buf_hdr_t *hdr, uint64_t size, void *tag)
{
	arc_buf_contents_t type = arc_buf_type(hdr);

	arc_get_data_impl(hdr, size, tag);
	if (type == ARC_BUFC_METADATA) {
		return (abd_alloc(size, B_TRUE));
	} else {
		ASSERT(type == ARC_BUFC_DATA);
		return (abd_alloc(size, B_FALSE));
	}
}

static void *
arc_get_data_buf(arc_buf_hdr_t *hdr, uint64_t size, void *tag)
{
	arc_buf_contents_t type = arc_buf_type(hdr);

	arc_get_data_impl(hdr, size, tag);
	if (type == ARC_BUFC_METADATA) {
		return (zio_buf_alloc(size));
	} else {
		ASSERT(type == ARC_BUFC_DATA);
		return (zio_data_buf_alloc(size));
	}
}

/*
 * Allocate a block and return it to the caller. If we are hitting the
 * hard limit for the cache size, we must sleep, waiting for the eviction
 * thread to catch up. If we're past the target size but below the hard
 * limit, we'll only signal the reclaim thread and continue on.
 */
static void
arc_get_data_impl(arc_buf_hdr_t *hdr, uint64_t size, void *tag)
{
	arc_state_t *state = hdr->b_l1hdr.b_state;
	arc_buf_contents_t type = arc_buf_type(hdr);

	arc_adapt(size, state);

	/*
	 * If arc_size is currently overflowing, and has grown past our
	 * upper limit, we must be adding data faster than the evict
	 * thread can evict. Thus, to ensure we don't compound the
	 * problem by adding more data and forcing arc_size to grow even
	 * further past it's target size, we halt and wait for the
	 * eviction thread to catch up.
	 *
	 * It's also possible that the reclaim thread is unable to evict
	 * enough buffers to get arc_size below the overflow limit (e.g.
	 * due to buffers being un-evictable, or hash lock collisions).
	 * In this case, we want to proceed regardless if we're
	 * overflowing; thus we don't use a while loop here.
	 */
	if (arc_is_overflowing()) {
		mutex_enter(&arc_reclaim_lock);

		/*
		 * Now that we've acquired the lock, we may no longer be
		 * over the overflow limit, lets check.
		 *
		 * We're ignoring the case of spurious wake ups. If that
		 * were to happen, it'd let this thread consume an ARC
		 * buffer before it should have (i.e. before we're under
		 * the overflow limit and were signalled by the reclaim
		 * thread). As long as that is a rare occurrence, it
		 * shouldn't cause any harm.
		 */
		if (arc_is_overflowing()) {
			cv_signal(&arc_reclaim_thread_cv);
			cv_wait(&arc_reclaim_waiters_cv, &arc_reclaim_lock);
		}

		mutex_exit(&arc_reclaim_lock);
	}

	VERIFY3U(hdr->b_type, ==, type);
	if (type == ARC_BUFC_METADATA) {
		arc_space_consume(size, ARC_SPACE_META);
	} else {
		arc_space_consume(size, ARC_SPACE_DATA);
	}

	/*
	 * Update the state size.  Note that ghost states have a
	 * "ghost size" and so don't need to be updated.
	 */
	if (!GHOST_STATE(state)) {

		(void) refcount_add_many(&state->arcs_size, size, tag);

		/*
		 * If this is reached via arc_read, the link is
		 * protected by the hash lock. If reached via
		 * arc_buf_alloc, the header should not be accessed by
		 * any other thread. And, if reached via arc_read_done,
		 * the hash lock will protect it if it's found in the
		 * hash table; otherwise no other thread should be
		 * trying to [add|remove]_reference it.
		 */
		if (multilist_link_active(&hdr->b_l1hdr.b_arc_node)) {
			ASSERT(refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
			(void) refcount_add_many(&state->arcs_esize[type],
			    size, tag);
		}

		/*
		 * If we are growing the cache, and we are adding anonymous
		 * data, and we have outgrown arc_p, update arc_p
		 */
		if (arc_size < arc_c && hdr->b_l1hdr.b_state == arc_anon &&
		    (refcount_count(&arc_anon->arcs_size) +
		    refcount_count(&arc_mru->arcs_size) > arc_p))
			arc_p = MIN(arc_c, arc_p + size);
	}
}

static void
arc_free_data_abd(arc_buf_hdr_t *hdr, abd_t *abd, uint64_t size, void *tag)
{
	arc_free_data_impl(hdr, size, tag);
	abd_free(abd);
}

static void
arc_free_data_buf(arc_buf_hdr_t *hdr, void *buf, uint64_t size, void *tag)
{
	arc_buf_contents_t type = arc_buf_type(hdr);

	arc_free_data_impl(hdr, size, tag);
	if (type == ARC_BUFC_METADATA) {
		zio_buf_free(buf, size);
	} else {
		ASSERT(type == ARC_BUFC_DATA);
		zio_data_buf_free(buf, size);
	}
}

/*
 * Free the arc data buffer.
 */
static void
arc_free_data_impl(arc_buf_hdr_t *hdr, uint64_t size, void *tag)
{
	arc_state_t *state = hdr->b_l1hdr.b_state;
	arc_buf_contents_t type = arc_buf_type(hdr);

	/* protected by hash lock, if in the hash table */
	if (multilist_link_active(&hdr->b_l1hdr.b_arc_node)) {
		ASSERT(refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
		ASSERT(state != arc_anon && state != arc_l2c_only);

		(void) refcount_remove_many(&state->arcs_esize[type],
		    size, tag);
	}
	(void) refcount_remove_many(&state->arcs_size, size, tag);

	VERIFY3U(hdr->b_type, ==, type);
	if (type == ARC_BUFC_METADATA) {
		arc_space_return(size, ARC_SPACE_META);
	} else {
		ASSERT(type == ARC_BUFC_DATA);
		arc_space_return(size, ARC_SPACE_DATA);
	}
}

/*
 * This routine is called whenever a buffer is accessed.
 * NOTE: the hash lock is dropped in this function.
 */
static void
arc_access(arc_buf_hdr_t *hdr, kmutex_t *hash_lock)
{
	clock_t now;

	ASSERT(MUTEX_HELD(hash_lock));
	ASSERT(HDR_HAS_L1HDR(hdr));

	if (hdr->b_l1hdr.b_state == arc_anon) {
		/*
		 * This buffer is not in the cache, and does not
		 * appear in our "ghost" list.  Add the new buffer
		 * to the MRU state.
		 */

		ASSERT0(hdr->b_l1hdr.b_arc_access);
		hdr->b_l1hdr.b_arc_access = ddi_get_lbolt();
		DTRACE_PROBE1(new_state__mru, arc_buf_hdr_t *, hdr);
		arc_change_state(arc_mru, hdr, hash_lock);

	} else if (hdr->b_l1hdr.b_state == arc_mru) {
		now = ddi_get_lbolt();

		/*
		 * If this buffer is here because of a prefetch, then either:
		 * - clear the flag if this is a "referencing" read
		 *   (any subsequent access will bump this into the MFU state).
		 * or
		 * - move the buffer to the head of the list if this is
		 *   another prefetch (to make it less likely to be evicted).
		 */
		if (HDR_PREFETCH(hdr)) {
			if (refcount_count(&hdr->b_l1hdr.b_refcnt) == 0) {
				/* link protected by hash lock */
				ASSERT(multilist_link_active(
				    &hdr->b_l1hdr.b_arc_node));
			} else {
				arc_hdr_clear_flags(hdr, ARC_FLAG_PREFETCH);
				atomic_inc_32(&hdr->b_l1hdr.b_mru_hits);
				ARCSTAT_BUMP(arcstat_mru_hits);
			}
			hdr->b_l1hdr.b_arc_access = now;
			return;
		}

		/*
		 * This buffer has been "accessed" only once so far,
		 * but it is still in the cache. Move it to the MFU
		 * state.
		 */
		if (ddi_time_after(now, hdr->b_l1hdr.b_arc_access +
		    ARC_MINTIME)) {
			/*
			 * More than 125ms have passed since we
			 * instantiated this buffer.  Move it to the
			 * most frequently used state.
			 */
			hdr->b_l1hdr.b_arc_access = now;
			DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, hdr);
			arc_change_state(arc_mfu, hdr, hash_lock);
		}
		atomic_inc_32(&hdr->b_l1hdr.b_mru_hits);
		ARCSTAT_BUMP(arcstat_mru_hits);
	} else if (hdr->b_l1hdr.b_state == arc_mru_ghost) {
		arc_state_t	*new_state;
		/*
		 * This buffer has been "accessed" recently, but
		 * was evicted from the cache.  Move it to the
		 * MFU state.
		 */

		if (HDR_PREFETCH(hdr)) {
			new_state = arc_mru;
			if (refcount_count(&hdr->b_l1hdr.b_refcnt) > 0)
				arc_hdr_clear_flags(hdr, ARC_FLAG_PREFETCH);
			DTRACE_PROBE1(new_state__mru, arc_buf_hdr_t *, hdr);
		} else {
			new_state = arc_mfu;
			DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, hdr);
		}

		hdr->b_l1hdr.b_arc_access = ddi_get_lbolt();
		arc_change_state(new_state, hdr, hash_lock);

		atomic_inc_32(&hdr->b_l1hdr.b_mru_ghost_hits);
		ARCSTAT_BUMP(arcstat_mru_ghost_hits);
	} else if (hdr->b_l1hdr.b_state == arc_mfu) {
		/*
		 * This buffer has been accessed more than once and is
		 * still in the cache.  Keep it in the MFU state.
		 *
		 * NOTE: an add_reference() that occurred when we did
		 * the arc_read() will have kicked this off the list.
		 * If it was a prefetch, we will explicitly move it to
		 * the head of the list now.
		 */
		if ((HDR_PREFETCH(hdr)) != 0) {
			ASSERT(refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
			/* link protected by hash_lock */
			ASSERT(multilist_link_active(&hdr->b_l1hdr.b_arc_node));
		}
		atomic_inc_32(&hdr->b_l1hdr.b_mfu_hits);
		ARCSTAT_BUMP(arcstat_mfu_hits);
		hdr->b_l1hdr.b_arc_access = ddi_get_lbolt();
	} else if (hdr->b_l1hdr.b_state == arc_mfu_ghost) {
		arc_state_t	*new_state = arc_mfu;
		/*
		 * This buffer has been accessed more than once but has
		 * been evicted from the cache.  Move it back to the
		 * MFU state.
		 */

		if (HDR_PREFETCH(hdr)) {
			/*
			 * This is a prefetch access...
			 * move this block back to the MRU state.
			 */
			ASSERT0(refcount_count(&hdr->b_l1hdr.b_refcnt));
			new_state = arc_mru;
		}

		hdr->b_l1hdr.b_arc_access = ddi_get_lbolt();
		DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, hdr);
		arc_change_state(new_state, hdr, hash_lock);

		atomic_inc_32(&hdr->b_l1hdr.b_mfu_ghost_hits);
		ARCSTAT_BUMP(arcstat_mfu_ghost_hits);
	} else if (hdr->b_l1hdr.b_state == arc_l2c_only) {
		/*
		 * This buffer is on the 2nd Level ARC.
		 */

		hdr->b_l1hdr.b_arc_access = ddi_get_lbolt();
		DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, hdr);
		arc_change_state(arc_mfu, hdr, hash_lock);
	} else {
		cmn_err(CE_PANIC, "invalid arc state 0x%p",
		    hdr->b_l1hdr.b_state);
	}
}

/* a generic arc_done_func_t which you can use */
/* ARGSUSED */
void
arc_bcopy_func(zio_t *zio, arc_buf_t *buf, void *arg)
{
	if (zio == NULL || zio->io_error == 0)
		bcopy(buf->b_data, arg, arc_buf_size(buf));
	arc_buf_destroy(buf, arg);
}

/* a generic arc_done_func_t */
void
arc_getbuf_func(zio_t *zio, arc_buf_t *buf, void *arg)
{
	arc_buf_t **bufp = arg;
	if (zio && zio->io_error) {
		arc_buf_destroy(buf, arg);
		*bufp = NULL;
	} else {
		*bufp = buf;
		ASSERT(buf->b_data);
	}
}

static void
arc_hdr_verify(arc_buf_hdr_t *hdr, blkptr_t *bp)
{
	if (BP_IS_HOLE(bp) || BP_IS_EMBEDDED(bp)) {
		ASSERT3U(HDR_GET_PSIZE(hdr), ==, 0);
		ASSERT3U(HDR_GET_COMPRESS(hdr), ==, ZIO_COMPRESS_OFF);
	} else {
		if (HDR_COMPRESSION_ENABLED(hdr)) {
			ASSERT3U(HDR_GET_COMPRESS(hdr), ==,
			    BP_GET_COMPRESS(bp));
		}
		ASSERT3U(HDR_GET_LSIZE(hdr), ==, BP_GET_LSIZE(bp));
		ASSERT3U(HDR_GET_PSIZE(hdr), ==, BP_GET_PSIZE(bp));
	}
}

static void
arc_read_done(zio_t *zio)
{
	arc_buf_hdr_t	*hdr = zio->io_private;
	kmutex_t	*hash_lock = NULL;
	arc_callback_t	*callback_list;
	arc_callback_t	*acb;
	boolean_t	freeable = B_FALSE;
	boolean_t	no_zio_error = (zio->io_error == 0);

	/*
	 * The hdr was inserted into hash-table and removed from lists
	 * prior to starting I/O.  We should find this header, since
	 * it's in the hash table, and it should be legit since it's
	 * not possible to evict it during the I/O.  The only possible
	 * reason for it not to be found is if we were freed during the
	 * read.
	 */
	if (HDR_IN_HASH_TABLE(hdr)) {
		arc_buf_hdr_t *found;

		ASSERT3U(hdr->b_birth, ==, BP_PHYSICAL_BIRTH(zio->io_bp));
		ASSERT3U(hdr->b_dva.dva_word[0], ==,
		    BP_IDENTITY(zio->io_bp)->dva_word[0]);
		ASSERT3U(hdr->b_dva.dva_word[1], ==,
		    BP_IDENTITY(zio->io_bp)->dva_word[1]);

		found = buf_hash_find(hdr->b_spa, zio->io_bp, &hash_lock);

		ASSERT((found == hdr &&
		    DVA_EQUAL(&hdr->b_dva, BP_IDENTITY(zio->io_bp))) ||
		    (found == hdr && HDR_L2_READING(hdr)));
		ASSERT3P(hash_lock, !=, NULL);
	}

	if (no_zio_error) {
		/* byteswap if necessary */
		if (BP_SHOULD_BYTESWAP(zio->io_bp)) {
			if (BP_GET_LEVEL(zio->io_bp) > 0) {
				hdr->b_l1hdr.b_byteswap = DMU_BSWAP_UINT64;
			} else {
				hdr->b_l1hdr.b_byteswap =
				    DMU_OT_BYTESWAP(BP_GET_TYPE(zio->io_bp));
			}
		} else {
			hdr->b_l1hdr.b_byteswap = DMU_BSWAP_NUMFUNCS;
		}
	}

	arc_hdr_clear_flags(hdr, ARC_FLAG_L2_EVICTED);
	if (l2arc_noprefetch && HDR_PREFETCH(hdr))
		arc_hdr_clear_flags(hdr, ARC_FLAG_L2CACHE);

	callback_list = hdr->b_l1hdr.b_acb;
	ASSERT3P(callback_list, !=, NULL);

	if (hash_lock && no_zio_error && hdr->b_l1hdr.b_state == arc_anon) {
		/*
		 * Only call arc_access on anonymous buffers.  This is because
		 * if we've issued an I/O for an evicted buffer, we've already
		 * called arc_access (to prevent any simultaneous readers from
		 * getting confused).
		 */
		arc_access(hdr, hash_lock);
	}

	/*
	 * If a read request has a callback (i.e. acb_done is not NULL), then we
	 * make a buf containing the data according to the parameters which were
	 * passed in. The implementation of arc_buf_alloc_impl() ensures that we
	 * aren't needlessly decompressing the data multiple times.
	 */
	int callback_cnt = 0;
	for (acb = callback_list; acb != NULL; acb = acb->acb_next) {
		if (!acb->acb_done)
			continue;

		/* This is a demand read since prefetches don't use callbacks */
		callback_cnt++;

		int error = arc_buf_alloc_impl(hdr, acb->acb_private,
		    acb->acb_compressed, no_zio_error, &acb->acb_buf);
		if (no_zio_error) {
			zio->io_error = error;
		}
	}
	hdr->b_l1hdr.b_acb = NULL;
	arc_hdr_clear_flags(hdr, ARC_FLAG_IO_IN_PROGRESS);
	if (callback_cnt == 0) {
		ASSERT(HDR_PREFETCH(hdr));
		ASSERT0(hdr->b_l1hdr.b_bufcnt);
		ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
	}

	ASSERT(refcount_is_zero(&hdr->b_l1hdr.b_refcnt) ||
	    callback_list != NULL);

	if (no_zio_error) {
		arc_hdr_verify(hdr, zio->io_bp);
	} else {
		arc_hdr_set_flags(hdr, ARC_FLAG_IO_ERROR);
		if (hdr->b_l1hdr.b_state != arc_anon)
			arc_change_state(arc_anon, hdr, hash_lock);
		if (HDR_IN_HASH_TABLE(hdr))
			buf_hash_remove(hdr);
		freeable = refcount_is_zero(&hdr->b_l1hdr.b_refcnt);
	}

	/*
	 * Broadcast before we drop the hash_lock to avoid the possibility
	 * that the hdr (and hence the cv) might be freed before we get to
	 * the cv_broadcast().
	 */
	cv_broadcast(&hdr->b_l1hdr.b_cv);

	if (hash_lock != NULL) {
		mutex_exit(hash_lock);
	} else {
		/*
		 * This block was freed while we waited for the read to
		 * complete.  It has been removed from the hash table and
		 * moved to the anonymous state (so that it won't show up
		 * in the cache).
		 */
		ASSERT3P(hdr->b_l1hdr.b_state, ==, arc_anon);
		freeable = refcount_is_zero(&hdr->b_l1hdr.b_refcnt);
	}

	/* execute each callback and free its structure */
	while ((acb = callback_list) != NULL) {
		if (acb->acb_done)
			acb->acb_done(zio, acb->acb_buf, acb->acb_private);

		if (acb->acb_zio_dummy != NULL) {
			acb->acb_zio_dummy->io_error = zio->io_error;
			zio_nowait(acb->acb_zio_dummy);
		}

		callback_list = acb->acb_next;
		kmem_free(acb, sizeof (arc_callback_t));
	}

	if (freeable)
		arc_hdr_destroy(hdr);
}

/*
 * "Read" the block at the specified DVA (in bp) via the
 * cache.  If the block is found in the cache, invoke the provided
 * callback immediately and return.  Note that the `zio' parameter
 * in the callback will be NULL in this case, since no IO was
 * required.  If the block is not in the cache pass the read request
 * on to the spa with a substitute callback function, so that the
 * requested block will be added to the cache.
 *
 * If a read request arrives for a block that has a read in-progress,
 * either wait for the in-progress read to complete (and return the
 * results); or, if this is a read with a "done" func, add a record
 * to the read to invoke the "done" func when the read completes,
 * and return; or just return.
 *
 * arc_read_done() will invoke all the requested "done" functions
 * for readers of this block.
 */
int
arc_read(zio_t *pio, spa_t *spa, const blkptr_t *bp, arc_done_func_t *done,
    void *private, zio_priority_t priority, int zio_flags,
    arc_flags_t *arc_flags, const zbookmark_phys_t *zb)
{
	arc_buf_hdr_t *hdr = NULL;
	kmutex_t *hash_lock = NULL;
	zio_t *rzio;
	uint64_t guid = spa_load_guid(spa);
	boolean_t compressed_read = (zio_flags & ZIO_FLAG_RAW) != 0;
	int rc = 0;

	ASSERT(!BP_IS_EMBEDDED(bp) ||
	    BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA);

top:
	if (!BP_IS_EMBEDDED(bp)) {
		/*
		 * Embedded BP's have no DVA and require no I/O to "read".
		 * Create an anonymous arc buf to back it.
		 */
		hdr = buf_hash_find(guid, bp, &hash_lock);
	}

	if (hdr != NULL && HDR_HAS_L1HDR(hdr) && hdr->b_l1hdr.b_pabd != NULL) {
		arc_buf_t *buf = NULL;
		*arc_flags |= ARC_FLAG_CACHED;

		if (HDR_IO_IN_PROGRESS(hdr)) {

			if ((hdr->b_flags & ARC_FLAG_PRIO_ASYNC_READ) &&
			    priority == ZIO_PRIORITY_SYNC_READ) {
				/*
				 * This sync read must wait for an
				 * in-progress async read (e.g. a predictive
				 * prefetch).  Async reads are queued
				 * separately at the vdev_queue layer, so
				 * this is a form of priority inversion.
				 * Ideally, we would "inherit" the demand
				 * i/o's priority by moving the i/o from
				 * the async queue to the synchronous queue,
				 * but there is currently no mechanism to do
				 * so.  Track this so that we can evaluate
				 * the magnitude of this potential performance
				 * problem.
				 *
				 * Note that if the prefetch i/o is already
				 * active (has been issued to the device),
				 * the prefetch improved performance, because
				 * we issued it sooner than we would have
				 * without the prefetch.
				 */
				DTRACE_PROBE1(arc__sync__wait__for__async,
				    arc_buf_hdr_t *, hdr);
				ARCSTAT_BUMP(arcstat_sync_wait_for_async);
			}
			if (hdr->b_flags & ARC_FLAG_PREDICTIVE_PREFETCH) {
				arc_hdr_clear_flags(hdr,
				    ARC_FLAG_PREDICTIVE_PREFETCH);
			}

			if (*arc_flags & ARC_FLAG_WAIT) {
				cv_wait(&hdr->b_l1hdr.b_cv, hash_lock);
				mutex_exit(hash_lock);
				goto top;
			}
			ASSERT(*arc_flags & ARC_FLAG_NOWAIT);

			if (done) {
				arc_callback_t *acb = NULL;

				acb = kmem_zalloc(sizeof (arc_callback_t),
				    KM_SLEEP);
				acb->acb_done = done;
				acb->acb_private = private;
				acb->acb_compressed = compressed_read;
				if (pio != NULL)
					acb->acb_zio_dummy = zio_null(pio,
					    spa, NULL, NULL, NULL, zio_flags);

				ASSERT3P(acb->acb_done, !=, NULL);
				acb->acb_next = hdr->b_l1hdr.b_acb;
				hdr->b_l1hdr.b_acb = acb;
				mutex_exit(hash_lock);
				goto out;
			}
			mutex_exit(hash_lock);
			goto out;
		}

		ASSERT(hdr->b_l1hdr.b_state == arc_mru ||
		    hdr->b_l1hdr.b_state == arc_mfu);

		if (done) {
			if (hdr->b_flags & ARC_FLAG_PREDICTIVE_PREFETCH) {
				/*
				 * This is a demand read which does not have to
				 * wait for i/o because we did a predictive
				 * prefetch i/o for it, which has completed.
				 */
				DTRACE_PROBE1(
				    arc__demand__hit__predictive__prefetch,
				    arc_buf_hdr_t *, hdr);
				ARCSTAT_BUMP(
				    arcstat_demand_hit_predictive_prefetch);
				arc_hdr_clear_flags(hdr,
				    ARC_FLAG_PREDICTIVE_PREFETCH);
			}
			ASSERT(!BP_IS_EMBEDDED(bp) || !BP_IS_HOLE(bp));

			/* Get a buf with the desired data in it. */
			VERIFY0(arc_buf_alloc_impl(hdr, private,
			    compressed_read, B_TRUE, &buf));
		} else if (*arc_flags & ARC_FLAG_PREFETCH &&
		    refcount_count(&hdr->b_l1hdr.b_refcnt) == 0) {
			arc_hdr_set_flags(hdr, ARC_FLAG_PREFETCH);
		}
		DTRACE_PROBE1(arc__hit, arc_buf_hdr_t *, hdr);
		arc_access(hdr, hash_lock);
		if (*arc_flags & ARC_FLAG_L2CACHE)
			arc_hdr_set_flags(hdr, ARC_FLAG_L2CACHE);
		mutex_exit(hash_lock);
		ARCSTAT_BUMP(arcstat_hits);
		ARCSTAT_CONDSTAT(!HDR_PREFETCH(hdr),
		    demand, prefetch, !HDR_ISTYPE_METADATA(hdr),
		    data, metadata, hits);

		if (done)
			done(NULL, buf, private);
	} else {
		uint64_t lsize = BP_GET_LSIZE(bp);
		uint64_t psize = BP_GET_PSIZE(bp);
		arc_callback_t *acb;
		vdev_t *vd = NULL;
		uint64_t addr = 0;
		boolean_t devw = B_FALSE;
		uint64_t size;

		/*
		 * Gracefully handle a damaged logical block size as a
		 * checksum error.
		 */
		if (lsize > spa_maxblocksize(spa)) {
			rc = SET_ERROR(ECKSUM);
			goto out;
		}

		if (hdr == NULL) {
			/* this block is not in the cache */
			arc_buf_hdr_t *exists = NULL;
			arc_buf_contents_t type = BP_GET_BUFC_TYPE(bp);
			hdr = arc_hdr_alloc(spa_load_guid(spa), psize, lsize,
			    BP_GET_COMPRESS(bp), type);

			if (!BP_IS_EMBEDDED(bp)) {
				hdr->b_dva = *BP_IDENTITY(bp);
				hdr->b_birth = BP_PHYSICAL_BIRTH(bp);
				exists = buf_hash_insert(hdr, &hash_lock);
			}
			if (exists != NULL) {
				/* somebody beat us to the hash insert */
				mutex_exit(hash_lock);
				buf_discard_identity(hdr);
				arc_hdr_destroy(hdr);
				goto top; /* restart the IO request */
			}
		} else {
			/*
			 * This block is in the ghost cache. If it was L2-only
			 * (and thus didn't have an L1 hdr), we realloc the
			 * header to add an L1 hdr.
			 */
			if (!HDR_HAS_L1HDR(hdr)) {
				hdr = arc_hdr_realloc(hdr, hdr_l2only_cache,
				    hdr_full_cache);
			}

			ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
			ASSERT(GHOST_STATE(hdr->b_l1hdr.b_state));
			ASSERT(!HDR_IO_IN_PROGRESS(hdr));
			ASSERT(refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
			ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
			ASSERT3P(hdr->b_l1hdr.b_freeze_cksum, ==, NULL);

			/*
			 * This is a delicate dance that we play here.
			 * This hdr is in the ghost list so we access it
			 * to move it out of the ghost list before we
			 * initiate the read. If it's a prefetch then
			 * it won't have a callback so we'll remove the
			 * reference that arc_buf_alloc_impl() created. We
			 * do this after we've called arc_access() to
			 * avoid hitting an assert in remove_reference().
			 */
			arc_access(hdr, hash_lock);
			arc_hdr_alloc_pabd(hdr);
		}
		ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
		size = arc_hdr_size(hdr);

		/*
		 * If compression is enabled on the hdr, then will do
		 * RAW I/O and will store the compressed data in the hdr's
		 * data block. Otherwise, the hdr's data block will contain
		 * the uncompressed data.
		 */
		if (HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF) {
			zio_flags |= ZIO_FLAG_RAW;
		}

		if (*arc_flags & ARC_FLAG_PREFETCH)
			arc_hdr_set_flags(hdr, ARC_FLAG_PREFETCH);
		if (*arc_flags & ARC_FLAG_L2CACHE)
			arc_hdr_set_flags(hdr, ARC_FLAG_L2CACHE);
		if (BP_GET_LEVEL(bp) > 0)
			arc_hdr_set_flags(hdr, ARC_FLAG_INDIRECT);
		if (*arc_flags & ARC_FLAG_PREDICTIVE_PREFETCH)
			arc_hdr_set_flags(hdr, ARC_FLAG_PREDICTIVE_PREFETCH);
		ASSERT(!GHOST_STATE(hdr->b_l1hdr.b_state));

		acb = kmem_zalloc(sizeof (arc_callback_t), KM_SLEEP);
		acb->acb_done = done;
		acb->acb_private = private;
		acb->acb_compressed = compressed_read;

		ASSERT3P(hdr->b_l1hdr.b_acb, ==, NULL);
		hdr->b_l1hdr.b_acb = acb;
		arc_hdr_set_flags(hdr, ARC_FLAG_IO_IN_PROGRESS);

		if (HDR_HAS_L2HDR(hdr) &&
		    (vd = hdr->b_l2hdr.b_dev->l2ad_vdev) != NULL) {
			devw = hdr->b_l2hdr.b_dev->l2ad_writing;
			addr = hdr->b_l2hdr.b_daddr;
			/*
			 * Lock out device removal.
			 */
			if (vdev_is_dead(vd) ||
			    !spa_config_tryenter(spa, SCL_L2ARC, vd, RW_READER))
				vd = NULL;
		}

		if (priority == ZIO_PRIORITY_ASYNC_READ)
			arc_hdr_set_flags(hdr, ARC_FLAG_PRIO_ASYNC_READ);
		else
			arc_hdr_clear_flags(hdr, ARC_FLAG_PRIO_ASYNC_READ);

		if (hash_lock != NULL)
			mutex_exit(hash_lock);

		/*
		 * At this point, we have a level 1 cache miss.  Try again in
		 * L2ARC if possible.
		 */
		ASSERT3U(HDR_GET_LSIZE(hdr), ==, lsize);

		DTRACE_PROBE4(arc__miss, arc_buf_hdr_t *, hdr, blkptr_t *, bp,
		    uint64_t, lsize, zbookmark_phys_t *, zb);
		ARCSTAT_BUMP(arcstat_misses);
		ARCSTAT_CONDSTAT(!HDR_PREFETCH(hdr),
		    demand, prefetch, !HDR_ISTYPE_METADATA(hdr),
		    data, metadata, misses);

		if (vd != NULL && l2arc_ndev != 0 && !(l2arc_norw && devw)) {
			/*
			 * Read from the L2ARC if the following are true:
			 * 1. The L2ARC vdev was previously cached.
			 * 2. This buffer still has L2ARC metadata.
			 * 3. This buffer isn't currently writing to the L2ARC.
			 * 4. The L2ARC entry wasn't evicted, which may
			 *    also have invalidated the vdev.
			 * 5. This isn't prefetch and l2arc_noprefetch is set.
			 */
			if (HDR_HAS_L2HDR(hdr) &&
			    !HDR_L2_WRITING(hdr) && !HDR_L2_EVICTED(hdr) &&
			    !(l2arc_noprefetch && HDR_PREFETCH(hdr))) {
				l2arc_read_callback_t *cb;

				DTRACE_PROBE1(l2arc__hit, arc_buf_hdr_t *, hdr);
				ARCSTAT_BUMP(arcstat_l2_hits);
				atomic_inc_32(&hdr->b_l2hdr.b_hits);

				cb = kmem_zalloc(sizeof (l2arc_read_callback_t),
				    KM_SLEEP);
				cb->l2rcb_hdr = hdr;
				cb->l2rcb_bp = *bp;
				cb->l2rcb_zb = *zb;
				cb->l2rcb_flags = zio_flags;

				ASSERT(addr >= VDEV_LABEL_START_SIZE &&
				    addr + lsize < vd->vdev_psize -
				    VDEV_LABEL_END_SIZE);

				/*
				 * l2arc read.  The SCL_L2ARC lock will be
				 * released by l2arc_read_done().
				 * Issue a null zio if the underlying buffer
				 * was squashed to zero size by compression.
				 */
				ASSERT3U(HDR_GET_COMPRESS(hdr), !=,
				    ZIO_COMPRESS_EMPTY);
				rzio = zio_read_phys(pio, vd, addr,
				    size, hdr->b_l1hdr.b_pabd,
				    ZIO_CHECKSUM_OFF,
				    l2arc_read_done, cb, priority,
				    zio_flags | ZIO_FLAG_DONT_CACHE |
				    ZIO_FLAG_CANFAIL |
				    ZIO_FLAG_DONT_PROPAGATE |
				    ZIO_FLAG_DONT_RETRY, B_FALSE);

				DTRACE_PROBE2(l2arc__read, vdev_t *, vd,
				    zio_t *, rzio);
				ARCSTAT_INCR(arcstat_l2_read_bytes, size);

				if (*arc_flags & ARC_FLAG_NOWAIT) {
					zio_nowait(rzio);
					goto out;
				}

				ASSERT(*arc_flags & ARC_FLAG_WAIT);
				if (zio_wait(rzio) == 0)
					goto out;

				/* l2arc read error; goto zio_read() */
			} else {
				DTRACE_PROBE1(l2arc__miss,
				    arc_buf_hdr_t *, hdr);
				ARCSTAT_BUMP(arcstat_l2_misses);
				if (HDR_L2_WRITING(hdr))
					ARCSTAT_BUMP(arcstat_l2_rw_clash);
				spa_config_exit(spa, SCL_L2ARC, vd);
			}
		} else {
			if (vd != NULL)
				spa_config_exit(spa, SCL_L2ARC, vd);
			if (l2arc_ndev != 0) {
				DTRACE_PROBE1(l2arc__miss,
				    arc_buf_hdr_t *, hdr);
				ARCSTAT_BUMP(arcstat_l2_misses);
			}
		}

		rzio = zio_read(pio, spa, bp, hdr->b_l1hdr.b_pabd, size,
		    arc_read_done, hdr, priority, zio_flags, zb);

		if (*arc_flags & ARC_FLAG_WAIT) {
			rc = zio_wait(rzio);
			goto out;
		}

		ASSERT(*arc_flags & ARC_FLAG_NOWAIT);
		zio_nowait(rzio);
	}

out:
	spa_read_history_add(spa, zb, *arc_flags);
	return (rc);
}

arc_prune_t *
arc_add_prune_callback(arc_prune_func_t *func, void *private)
{
	arc_prune_t *p;

	p = kmem_alloc(sizeof (*p), KM_SLEEP);
	p->p_pfunc = func;
	p->p_private = private;
	list_link_init(&p->p_node);
	refcount_create(&p->p_refcnt);

	mutex_enter(&arc_prune_mtx);
	refcount_add(&p->p_refcnt, &arc_prune_list);
	list_insert_head(&arc_prune_list, p);
	mutex_exit(&arc_prune_mtx);

	return (p);
}

void
arc_remove_prune_callback(arc_prune_t *p)
{
	boolean_t wait = B_FALSE;
	mutex_enter(&arc_prune_mtx);
	list_remove(&arc_prune_list, p);
	if (refcount_remove(&p->p_refcnt, &arc_prune_list) > 0)
		wait = B_TRUE;
	mutex_exit(&arc_prune_mtx);

	/* wait for arc_prune_task to finish */
	if (wait)
		taskq_wait_outstanding(arc_prune_taskq, 0);
	ASSERT0(refcount_count(&p->p_refcnt));
	refcount_destroy(&p->p_refcnt);
	kmem_free(p, sizeof (*p));
}

/*
 * Notify the arc that a block was freed, and thus will never be used again.
 */
void
arc_freed(spa_t *spa, const blkptr_t *bp)
{
	arc_buf_hdr_t *hdr;
	kmutex_t *hash_lock;
	uint64_t guid = spa_load_guid(spa);

	ASSERT(!BP_IS_EMBEDDED(bp));

	hdr = buf_hash_find(guid, bp, &hash_lock);
	if (hdr == NULL)
		return;

	/*
	 * We might be trying to free a block that is still doing I/O
	 * (i.e. prefetch) or has a reference (i.e. a dedup-ed,
	 * dmu_sync-ed block). If this block is being prefetched, then it
	 * would still have the ARC_FLAG_IO_IN_PROGRESS flag set on the hdr
	 * until the I/O completes. A block may also have a reference if it is
	 * part of a dedup-ed, dmu_synced write. The dmu_sync() function would
	 * have written the new block to its final resting place on disk but
	 * without the dedup flag set. This would have left the hdr in the MRU
	 * state and discoverable. When the txg finally syncs it detects that
	 * the block was overridden in open context and issues an override I/O.
	 * Since this is a dedup block, the override I/O will determine if the
	 * block is already in the DDT. If so, then it will replace the io_bp
	 * with the bp from the DDT and allow the I/O to finish. When the I/O
	 * reaches the done callback, dbuf_write_override_done, it will
	 * check to see if the io_bp and io_bp_override are identical.
	 * If they are not, then it indicates that the bp was replaced with
	 * the bp in the DDT and the override bp is freed. This allows
	 * us to arrive here with a reference on a block that is being
	 * freed. So if we have an I/O in progress, or a reference to
	 * this hdr, then we don't destroy the hdr.
	 */
	if (!HDR_HAS_L1HDR(hdr) || (!HDR_IO_IN_PROGRESS(hdr) &&
	    refcount_is_zero(&hdr->b_l1hdr.b_refcnt))) {
		arc_change_state(arc_anon, hdr, hash_lock);
		arc_hdr_destroy(hdr);
		mutex_exit(hash_lock);
	} else {
		mutex_exit(hash_lock);
	}

}

/*
 * Release this buffer from the cache, making it an anonymous buffer.  This
 * must be done after a read and prior to modifying the buffer contents.
 * If the buffer has more than one reference, we must make
 * a new hdr for the buffer.
 */
void
arc_release(arc_buf_t *buf, void *tag)
{
	kmutex_t *hash_lock;
	arc_state_t *state;
	arc_buf_hdr_t *hdr = buf->b_hdr;

	/*
	 * It would be nice to assert that if its DMU metadata (level >
	 * 0 || it's the dnode file), then it must be syncing context.
	 * But we don't know that information at this level.
	 */

	mutex_enter(&buf->b_evict_lock);

	ASSERT(HDR_HAS_L1HDR(hdr));

	/*
	 * We don't grab the hash lock prior to this check, because if
	 * the buffer's header is in the arc_anon state, it won't be
	 * linked into the hash table.
	 */
	if (hdr->b_l1hdr.b_state == arc_anon) {
		mutex_exit(&buf->b_evict_lock);
		ASSERT(!HDR_IO_IN_PROGRESS(hdr));
		ASSERT(!HDR_IN_HASH_TABLE(hdr));
		ASSERT(!HDR_HAS_L2HDR(hdr));
		ASSERT(HDR_EMPTY(hdr));

		ASSERT3U(hdr->b_l1hdr.b_bufcnt, ==, 1);
		ASSERT3S(refcount_count(&hdr->b_l1hdr.b_refcnt), ==, 1);
		ASSERT(!list_link_active(&hdr->b_l1hdr.b_arc_node));

		hdr->b_l1hdr.b_arc_access = 0;

		/*
		 * If the buf is being overridden then it may already
		 * have a hdr that is not empty.
		 */
		buf_discard_identity(hdr);
		arc_buf_thaw(buf);

		return;
	}

	hash_lock = HDR_LOCK(hdr);
	mutex_enter(hash_lock);

	/*
	 * This assignment is only valid as long as the hash_lock is
	 * held, we must be careful not to reference state or the
	 * b_state field after dropping the lock.
	 */
	state = hdr->b_l1hdr.b_state;
	ASSERT3P(hash_lock, ==, HDR_LOCK(hdr));
	ASSERT3P(state, !=, arc_anon);

	/* this buffer is not on any list */
	ASSERT3S(refcount_count(&hdr->b_l1hdr.b_refcnt), >, 0);

	if (HDR_HAS_L2HDR(hdr)) {
		mutex_enter(&hdr->b_l2hdr.b_dev->l2ad_mtx);

		/*
		 * We have to recheck this conditional again now that
		 * we're holding the l2ad_mtx to prevent a race with
		 * another thread which might be concurrently calling
		 * l2arc_evict(). In that case, l2arc_evict() might have
		 * destroyed the header's L2 portion as we were waiting
		 * to acquire the l2ad_mtx.
		 */
		if (HDR_HAS_L2HDR(hdr))
			arc_hdr_l2hdr_destroy(hdr);

		mutex_exit(&hdr->b_l2hdr.b_dev->l2ad_mtx);
	}

	/*
	 * Do we have more than one buf?
	 */
	if (hdr->b_l1hdr.b_bufcnt > 1) {
		arc_buf_hdr_t *nhdr;
		uint64_t spa = hdr->b_spa;
		uint64_t psize = HDR_GET_PSIZE(hdr);
		uint64_t lsize = HDR_GET_LSIZE(hdr);
		enum zio_compress compress = HDR_GET_COMPRESS(hdr);
		arc_buf_contents_t type = arc_buf_type(hdr);
		VERIFY3U(hdr->b_type, ==, type);

		ASSERT(hdr->b_l1hdr.b_buf != buf || buf->b_next != NULL);
		(void) remove_reference(hdr, hash_lock, tag);

		if (arc_buf_is_shared(buf) && !ARC_BUF_COMPRESSED(buf)) {
			ASSERT3P(hdr->b_l1hdr.b_buf, !=, buf);
			ASSERT(ARC_BUF_LAST(buf));
		}

		/*
		 * Pull the data off of this hdr and attach it to
		 * a new anonymous hdr. Also find the last buffer
		 * in the hdr's buffer list.
		 */
		arc_buf_t *lastbuf = arc_buf_remove(hdr, buf);
		ASSERT3P(lastbuf, !=, NULL);

		/*
		 * If the current arc_buf_t and the hdr are sharing their data
		 * buffer, then we must stop sharing that block.
		 */
		if (arc_buf_is_shared(buf)) {
			ASSERT3P(hdr->b_l1hdr.b_buf, !=, buf);
			VERIFY(!arc_buf_is_shared(lastbuf));

			/*
			 * First, sever the block sharing relationship between
			 * buf and the arc_buf_hdr_t.
			 */
			arc_unshare_buf(hdr, buf);

			/*
			 * Now we need to recreate the hdr's b_pabd. Since we
			 * have lastbuf handy, we try to share with it, but if
			 * we can't then we allocate a new b_pabd and copy the
			 * data from buf into it.
			 */
			if (arc_can_share(hdr, lastbuf)) {
				arc_share_buf(hdr, lastbuf);
			} else {
				arc_hdr_alloc_pabd(hdr);
				abd_copy_from_buf(hdr->b_l1hdr.b_pabd,
				    buf->b_data, psize);
			}
			VERIFY3P(lastbuf->b_data, !=, NULL);
		} else if (HDR_SHARED_DATA(hdr)) {
			/*
			 * Uncompressed shared buffers are always at the end
			 * of the list. Compressed buffers don't have the
			 * same requirements. This makes it hard to
			 * simply assert that the lastbuf is shared so
			 * we rely on the hdr's compression flags to determine
			 * if we have a compressed, shared buffer.
			 */
			ASSERT(arc_buf_is_shared(lastbuf) ||
			    HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF);
			ASSERT(!ARC_BUF_SHARED(buf));
		}
		ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
		ASSERT3P(state, !=, arc_l2c_only);

		(void) refcount_remove_many(&state->arcs_size,
		    arc_buf_size(buf), buf);

		if (refcount_is_zero(&hdr->b_l1hdr.b_refcnt)) {
			ASSERT3P(state, !=, arc_l2c_only);
			(void) refcount_remove_many(&state->arcs_esize[type],
			    arc_buf_size(buf), buf);
		}

		hdr->b_l1hdr.b_bufcnt -= 1;
		arc_cksum_verify(buf);
		arc_buf_unwatch(buf);

		mutex_exit(hash_lock);

		/*
		 * Allocate a new hdr. The new hdr will contain a b_pabd
		 * buffer which will be freed in arc_write().
		 */
		nhdr = arc_hdr_alloc(spa, psize, lsize, compress, type);
		ASSERT3P(nhdr->b_l1hdr.b_buf, ==, NULL);
		ASSERT0(nhdr->b_l1hdr.b_bufcnt);
		ASSERT0(refcount_count(&nhdr->b_l1hdr.b_refcnt));
		VERIFY3U(nhdr->b_type, ==, type);
		ASSERT(!HDR_SHARED_DATA(nhdr));

		nhdr->b_l1hdr.b_buf = buf;
		nhdr->b_l1hdr.b_bufcnt = 1;
		nhdr->b_l1hdr.b_mru_hits = 0;
		nhdr->b_l1hdr.b_mru_ghost_hits = 0;
		nhdr->b_l1hdr.b_mfu_hits = 0;
		nhdr->b_l1hdr.b_mfu_ghost_hits = 0;
		nhdr->b_l1hdr.b_l2_hits = 0;
		(void) refcount_add(&nhdr->b_l1hdr.b_refcnt, tag);
		buf->b_hdr = nhdr;

		mutex_exit(&buf->b_evict_lock);
		(void) refcount_add_many(&arc_anon->arcs_size,
		    HDR_GET_LSIZE(nhdr), buf);
	} else {
		mutex_exit(&buf->b_evict_lock);
		ASSERT(refcount_count(&hdr->b_l1hdr.b_refcnt) == 1);
		/* protected by hash lock, or hdr is on arc_anon */
		ASSERT(!multilist_link_active(&hdr->b_l1hdr.b_arc_node));
		ASSERT(!HDR_IO_IN_PROGRESS(hdr));
		hdr->b_l1hdr.b_mru_hits = 0;
		hdr->b_l1hdr.b_mru_ghost_hits = 0;
		hdr->b_l1hdr.b_mfu_hits = 0;
		hdr->b_l1hdr.b_mfu_ghost_hits = 0;
		hdr->b_l1hdr.b_l2_hits = 0;
		arc_change_state(arc_anon, hdr, hash_lock);
		hdr->b_l1hdr.b_arc_access = 0;
		mutex_exit(hash_lock);

		buf_discard_identity(hdr);
		arc_buf_thaw(buf);
	}
}

int
arc_released(arc_buf_t *buf)
{
	int released;

	mutex_enter(&buf->b_evict_lock);
	released = (buf->b_data != NULL &&
	    buf->b_hdr->b_l1hdr.b_state == arc_anon);
	mutex_exit(&buf->b_evict_lock);
	return (released);
}

#ifdef ZFS_DEBUG
int
arc_referenced(arc_buf_t *buf)
{
	int referenced;

	mutex_enter(&buf->b_evict_lock);
	referenced = (refcount_count(&buf->b_hdr->b_l1hdr.b_refcnt));
	mutex_exit(&buf->b_evict_lock);
	return (referenced);
}
#endif

static void
arc_write_ready(zio_t *zio)
{
	arc_write_callback_t *callback = zio->io_private;
	arc_buf_t *buf = callback->awcb_buf;
	arc_buf_hdr_t *hdr = buf->b_hdr;
	uint64_t psize = BP_IS_HOLE(zio->io_bp) ? 0 : BP_GET_PSIZE(zio->io_bp);
	enum zio_compress compress;
	fstrans_cookie_t cookie = spl_fstrans_mark();

	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT(!refcount_is_zero(&buf->b_hdr->b_l1hdr.b_refcnt));
	ASSERT(hdr->b_l1hdr.b_bufcnt > 0);

	/*
	 * If we're reexecuting this zio because the pool suspended, then
	 * cleanup any state that was previously set the first time the
	 * callback was invoked.
	 */
	if (zio->io_flags & ZIO_FLAG_REEXECUTED) {
		arc_cksum_free(hdr);
		arc_buf_unwatch(buf);
		if (hdr->b_l1hdr.b_pabd != NULL) {
			if (arc_buf_is_shared(buf)) {
				arc_unshare_buf(hdr, buf);
			} else {
				arc_hdr_free_pabd(hdr);
			}
		}
	}
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	ASSERT(!HDR_SHARED_DATA(hdr));
	ASSERT(!arc_buf_is_shared(buf));

	callback->awcb_ready(zio, buf, callback->awcb_private);

	if (HDR_IO_IN_PROGRESS(hdr))
		ASSERT(zio->io_flags & ZIO_FLAG_REEXECUTED);

	arc_cksum_compute(buf);
	arc_hdr_set_flags(hdr, ARC_FLAG_IO_IN_PROGRESS);

	if (BP_IS_HOLE(zio->io_bp) || BP_IS_EMBEDDED(zio->io_bp)) {
		compress = ZIO_COMPRESS_OFF;
	} else {
		ASSERT3U(HDR_GET_LSIZE(hdr), ==, BP_GET_LSIZE(zio->io_bp));
		compress = BP_GET_COMPRESS(zio->io_bp);
	}
	HDR_SET_PSIZE(hdr, psize);
	arc_hdr_set_compress(hdr, compress);

	/*
	 * Fill the hdr with data. If the hdr is compressed, the data we want
	 * is available from the zio, otherwise we can take it from the buf.
	 *
	 * We might be able to share the buf's data with the hdr here. However,
	 * doing so would cause the ARC to be full of linear ABDs if we write a
	 * lot of shareable data. As a compromise, we check whether scattered
	 * ABDs are allowed, and assume that if they are then the user wants
	 * the ARC to be primarily filled with them regardless of the data being
	 * written. Therefore, if they're allowed then we allocate one and copy
	 * the data into it; otherwise, we share the data directly if we can.
	 */
	if (zfs_abd_scatter_enabled || !arc_can_share(hdr, buf)) {
		arc_hdr_alloc_pabd(hdr);

		/*
		 * Ideally, we would always copy the io_abd into b_pabd, but the
		 * user may have disabled compressed ARC, thus we must check the
		 * hdr's compression setting rather than the io_bp's.
		 */
		if (HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF) {
			ASSERT3U(BP_GET_COMPRESS(zio->io_bp), !=,
			    ZIO_COMPRESS_OFF);
			ASSERT3U(psize, >, 0);

			abd_copy(hdr->b_l1hdr.b_pabd, zio->io_abd, psize);
		} else {
			ASSERT3U(zio->io_orig_size, ==, arc_hdr_size(hdr));

			abd_copy_from_buf(hdr->b_l1hdr.b_pabd, buf->b_data,
			    arc_buf_size(buf));
		}
	} else {
		ASSERT3P(buf->b_data, ==, abd_to_buf(zio->io_orig_abd));
		ASSERT3U(zio->io_orig_size, ==, arc_buf_size(buf));
		ASSERT3U(hdr->b_l1hdr.b_bufcnt, ==, 1);

		arc_share_buf(hdr, buf);
	}

	arc_hdr_verify(hdr, zio->io_bp);
	spl_fstrans_unmark(cookie);
}

static void
arc_write_children_ready(zio_t *zio)
{
	arc_write_callback_t *callback = zio->io_private;
	arc_buf_t *buf = callback->awcb_buf;

	callback->awcb_children_ready(zio, buf, callback->awcb_private);
}

/*
 * The SPA calls this callback for each physical write that happens on behalf
 * of a logical write.  See the comment in dbuf_write_physdone() for details.
 */
static void
arc_write_physdone(zio_t *zio)
{
	arc_write_callback_t *cb = zio->io_private;
	if (cb->awcb_physdone != NULL)
		cb->awcb_physdone(zio, cb->awcb_buf, cb->awcb_private);
}

static void
arc_write_done(zio_t *zio)
{
	arc_write_callback_t *callback = zio->io_private;
	arc_buf_t *buf = callback->awcb_buf;
	arc_buf_hdr_t *hdr = buf->b_hdr;

	ASSERT3P(hdr->b_l1hdr.b_acb, ==, NULL);

	if (zio->io_error == 0) {
		arc_hdr_verify(hdr, zio->io_bp);

		if (BP_IS_HOLE(zio->io_bp) || BP_IS_EMBEDDED(zio->io_bp)) {
			buf_discard_identity(hdr);
		} else {
			hdr->b_dva = *BP_IDENTITY(zio->io_bp);
			hdr->b_birth = BP_PHYSICAL_BIRTH(zio->io_bp);
		}
	} else {
		ASSERT(HDR_EMPTY(hdr));
	}

	/*
	 * If the block to be written was all-zero or compressed enough to be
	 * embedded in the BP, no write was performed so there will be no
	 * dva/birth/checksum.  The buffer must therefore remain anonymous
	 * (and uncached).
	 */
	if (!HDR_EMPTY(hdr)) {
		arc_buf_hdr_t *exists;
		kmutex_t *hash_lock;

		ASSERT3U(zio->io_error, ==, 0);

		arc_cksum_verify(buf);

		exists = buf_hash_insert(hdr, &hash_lock);
		if (exists != NULL) {
			/*
			 * This can only happen if we overwrite for
			 * sync-to-convergence, because we remove
			 * buffers from the hash table when we arc_free().
			 */
			if (zio->io_flags & ZIO_FLAG_IO_REWRITE) {
				if (!BP_EQUAL(&zio->io_bp_orig, zio->io_bp))
					panic("bad overwrite, hdr=%p exists=%p",
					    (void *)hdr, (void *)exists);
				ASSERT(refcount_is_zero(
				    &exists->b_l1hdr.b_refcnt));
				arc_change_state(arc_anon, exists, hash_lock);
				mutex_exit(hash_lock);
				arc_hdr_destroy(exists);
				exists = buf_hash_insert(hdr, &hash_lock);
				ASSERT3P(exists, ==, NULL);
			} else if (zio->io_flags & ZIO_FLAG_NOPWRITE) {
				/* nopwrite */
				ASSERT(zio->io_prop.zp_nopwrite);
				if (!BP_EQUAL(&zio->io_bp_orig, zio->io_bp))
					panic("bad nopwrite, hdr=%p exists=%p",
					    (void *)hdr, (void *)exists);
			} else {
				/* Dedup */
				ASSERT(hdr->b_l1hdr.b_bufcnt == 1);
				ASSERT(hdr->b_l1hdr.b_state == arc_anon);
				ASSERT(BP_GET_DEDUP(zio->io_bp));
				ASSERT(BP_GET_LEVEL(zio->io_bp) == 0);
			}
		}
		arc_hdr_clear_flags(hdr, ARC_FLAG_IO_IN_PROGRESS);
		/* if it's not anon, we are doing a scrub */
		if (exists == NULL && hdr->b_l1hdr.b_state == arc_anon)
			arc_access(hdr, hash_lock);
		mutex_exit(hash_lock);
	} else {
		arc_hdr_clear_flags(hdr, ARC_FLAG_IO_IN_PROGRESS);
	}

	ASSERT(!refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
	callback->awcb_done(zio, buf, callback->awcb_private);

	abd_put(zio->io_abd);
	kmem_free(callback, sizeof (arc_write_callback_t));
}

zio_t *
arc_write(zio_t *pio, spa_t *spa, uint64_t txg,
    blkptr_t *bp, arc_buf_t *buf, boolean_t l2arc,
    const zio_prop_t *zp, arc_done_func_t *ready,
    arc_done_func_t *children_ready, arc_done_func_t *physdone,
    arc_done_func_t *done, void *private, zio_priority_t priority,
    int zio_flags, const zbookmark_phys_t *zb)
{
	arc_buf_hdr_t *hdr = buf->b_hdr;
	arc_write_callback_t *callback;
	zio_t *zio;

	ASSERT3P(ready, !=, NULL);
	ASSERT3P(done, !=, NULL);
	ASSERT(!HDR_IO_ERROR(hdr));
	ASSERT(!HDR_IO_IN_PROGRESS(hdr));
	ASSERT3P(hdr->b_l1hdr.b_acb, ==, NULL);
	ASSERT3U(hdr->b_l1hdr.b_bufcnt, >, 0);
	if (l2arc)
		arc_hdr_set_flags(hdr, ARC_FLAG_L2CACHE);
	if (ARC_BUF_COMPRESSED(buf)) {
		ASSERT3U(zp->zp_compress, !=, ZIO_COMPRESS_OFF);
		ASSERT3U(HDR_GET_LSIZE(hdr), !=, arc_buf_size(buf));
		zio_flags |= ZIO_FLAG_RAW;
	}
	callback = kmem_zalloc(sizeof (arc_write_callback_t), KM_SLEEP);
	callback->awcb_ready = ready;
	callback->awcb_children_ready = children_ready;
	callback->awcb_physdone = physdone;
	callback->awcb_done = done;
	callback->awcb_private = private;
	callback->awcb_buf = buf;

	/*
	 * The hdr's b_pabd is now stale, free it now. A new data block
	 * will be allocated when the zio pipeline calls arc_write_ready().
	 */
	if (hdr->b_l1hdr.b_pabd != NULL) {
		/*
		 * If the buf is currently sharing the data block with
		 * the hdr then we need to break that relationship here.
		 * The hdr will remain with a NULL data pointer and the
		 * buf will take sole ownership of the block.
		 */
		if (arc_buf_is_shared(buf)) {
			arc_unshare_buf(hdr, buf);
		} else {
			arc_hdr_free_pabd(hdr);
		}
		VERIFY3P(buf->b_data, !=, NULL);
		arc_hdr_set_compress(hdr, ZIO_COMPRESS_OFF);
	}
	ASSERT(!arc_buf_is_shared(buf));
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);

	zio = zio_write(pio, spa, txg, bp,
	    abd_get_from_buf(buf->b_data, HDR_GET_LSIZE(hdr)),
	    HDR_GET_LSIZE(hdr), arc_buf_size(buf), zp,
	    arc_write_ready,
	    (children_ready != NULL) ? arc_write_children_ready : NULL,
	    arc_write_physdone, arc_write_done, callback,
	    priority, zio_flags, zb);

	return (zio);
}

static int
arc_memory_throttle(uint64_t reserve, uint64_t txg)
{
#ifdef _KERNEL
	uint64_t available_memory = ptob(freemem);
	static uint64_t page_load = 0;
	static uint64_t last_txg = 0;
#ifdef __linux__
	pgcnt_t minfree = btop(arc_sys_free / 4);
#endif

#if defined(__i386)
	available_memory =
	    MIN(available_memory, vmem_size(heap_arena, VMEM_FREE));
#endif

	if (available_memory > arc_all_memory() * arc_lotsfree_percent / 100)
		return (0);

	if (txg > last_txg) {
		last_txg = txg;
		page_load = 0;
	}
	/*
	 * If we are in pageout, we know that memory is already tight,
	 * the arc is already going to be evicting, so we just want to
	 * continue to let page writes occur as quickly as possible.
	 */
	if (current_is_kswapd()) {
		if (page_load > MAX(ptob(minfree), available_memory) / 4) {
			DMU_TX_STAT_BUMP(dmu_tx_memory_reclaim);
			return (SET_ERROR(ERESTART));
		}
		/* Note: reserve is inflated, so we deflate */
		page_load += reserve / 8;
		return (0);
	} else if (page_load > 0 && arc_reclaim_needed()) {
		/* memory is low, delay before restarting */
		ARCSTAT_INCR(arcstat_memory_throttle_count, 1);
		DMU_TX_STAT_BUMP(dmu_tx_memory_reclaim);
		return (SET_ERROR(EAGAIN));
	}
	page_load = 0;
#endif
	return (0);
}

void
arc_tempreserve_clear(uint64_t reserve)
{
	atomic_add_64(&arc_tempreserve, -reserve);
	ASSERT((int64_t)arc_tempreserve >= 0);
}

int
arc_tempreserve_space(uint64_t reserve, uint64_t txg)
{
	int error;
	uint64_t anon_size;

	if (!arc_no_grow &&
	    reserve > arc_c/4 &&
	    reserve * 4 > (2ULL << SPA_MAXBLOCKSHIFT))
		arc_c = MIN(arc_c_max, reserve * 4);

	/*
	 * Throttle when the calculated memory footprint for the TXG
	 * exceeds the target ARC size.
	 */
	if (reserve > arc_c) {
		DMU_TX_STAT_BUMP(dmu_tx_memory_reserve);
		return (SET_ERROR(ERESTART));
	}

	/*
	 * Don't count loaned bufs as in flight dirty data to prevent long
	 * network delays from blocking transactions that are ready to be
	 * assigned to a txg.
	 */

	/* assert that it has not wrapped around */
	ASSERT3S(atomic_add_64_nv(&arc_loaned_bytes, 0), >=, 0);

	anon_size = MAX((int64_t)(refcount_count(&arc_anon->arcs_size) -
	    arc_loaned_bytes), 0);

	/*
	 * Writes will, almost always, require additional memory allocations
	 * in order to compress/encrypt/etc the data.  We therefore need to
	 * make sure that there is sufficient available memory for this.
	 */
	error = arc_memory_throttle(reserve, txg);
	if (error != 0)
		return (error);

	/*
	 * Throttle writes when the amount of dirty data in the cache
	 * gets too large.  We try to keep the cache less than half full
	 * of dirty blocks so that our sync times don't grow too large.
	 * Note: if two requests come in concurrently, we might let them
	 * both succeed, when one of them should fail.  Not a huge deal.
	 */

	if (reserve + arc_tempreserve + anon_size > arc_c / 2 &&
	    anon_size > arc_c / 4) {
		uint64_t meta_esize =
		    refcount_count(&arc_anon->arcs_esize[ARC_BUFC_METADATA]);
		uint64_t data_esize =
		    refcount_count(&arc_anon->arcs_esize[ARC_BUFC_DATA]);
		dprintf("failing, arc_tempreserve=%lluK anon_meta=%lluK "
		    "anon_data=%lluK tempreserve=%lluK arc_c=%lluK\n",
		    arc_tempreserve >> 10, meta_esize >> 10,
		    data_esize >> 10, reserve >> 10, arc_c >> 10);
		DMU_TX_STAT_BUMP(dmu_tx_dirty_throttle);
		return (SET_ERROR(ERESTART));
	}
	atomic_add_64(&arc_tempreserve, reserve);
	return (0);
}

static void
arc_kstat_update_state(arc_state_t *state, kstat_named_t *size,
    kstat_named_t *evict_data, kstat_named_t *evict_metadata)
{
	size->value.ui64 = refcount_count(&state->arcs_size);
	evict_data->value.ui64 =
	    refcount_count(&state->arcs_esize[ARC_BUFC_DATA]);
	evict_metadata->value.ui64 =
	    refcount_count(&state->arcs_esize[ARC_BUFC_METADATA]);
}

static int
arc_kstat_update(kstat_t *ksp, int rw)
{
	arc_stats_t *as = ksp->ks_data;

	if (rw == KSTAT_WRITE) {
		return (EACCES);
	} else {
		arc_kstat_update_state(arc_anon,
		    &as->arcstat_anon_size,
		    &as->arcstat_anon_evictable_data,
		    &as->arcstat_anon_evictable_metadata);
		arc_kstat_update_state(arc_mru,
		    &as->arcstat_mru_size,
		    &as->arcstat_mru_evictable_data,
		    &as->arcstat_mru_evictable_metadata);
		arc_kstat_update_state(arc_mru_ghost,
		    &as->arcstat_mru_ghost_size,
		    &as->arcstat_mru_ghost_evictable_data,
		    &as->arcstat_mru_ghost_evictable_metadata);
		arc_kstat_update_state(arc_mfu,
		    &as->arcstat_mfu_size,
		    &as->arcstat_mfu_evictable_data,
		    &as->arcstat_mfu_evictable_metadata);
		arc_kstat_update_state(arc_mfu_ghost,
		    &as->arcstat_mfu_ghost_size,
		    &as->arcstat_mfu_ghost_evictable_data,
		    &as->arcstat_mfu_ghost_evictable_metadata);
	}

	return (0);
}

/*
 * This function *must* return indices evenly distributed between all
 * sublists of the multilist. This is needed due to how the ARC eviction
 * code is laid out; arc_evict_state() assumes ARC buffers are evenly
 * distributed between all sublists and uses this assumption when
 * deciding which sublist to evict from and how much to evict from it.
 */
unsigned int
arc_state_multilist_index_func(multilist_t *ml, void *obj)
{
	arc_buf_hdr_t *hdr = obj;

	/*
	 * We rely on b_dva to generate evenly distributed index
	 * numbers using buf_hash below. So, as an added precaution,
	 * let's make sure we never add empty buffers to the arc lists.
	 */
	ASSERT(!HDR_EMPTY(hdr));

	/*
	 * The assumption here, is the hash value for a given
	 * arc_buf_hdr_t will remain constant throughout its lifetime
	 * (i.e. its b_spa, b_dva, and b_birth fields don't change).
	 * Thus, we don't need to store the header's sublist index
	 * on insertion, as this index can be recalculated on removal.
	 *
	 * Also, the low order bits of the hash value are thought to be
	 * distributed evenly. Otherwise, in the case that the multilist
	 * has a power of two number of sublists, each sublists' usage
	 * would not be evenly distributed.
	 */
	return (buf_hash(hdr->b_spa, &hdr->b_dva, hdr->b_birth) %
	    multilist_get_num_sublists(ml));
}

/*
 * Called during module initialization and periodically thereafter to
 * apply reasonable changes to the exposed performance tunings.  Non-zero
 * zfs_* values which differ from the currently set values will be applied.
 */
static void
arc_tuning_update(void)
{
	uint64_t percent, allmem = arc_all_memory();

	/* Valid range: 64M - <all physical memory> */
	if ((zfs_arc_max) && (zfs_arc_max != arc_c_max) &&
	    (zfs_arc_max > 64 << 20) && (zfs_arc_max < allmem) &&
	    (zfs_arc_max > arc_c_min)) {
		arc_c_max = zfs_arc_max;
		arc_c = arc_c_max;
		arc_p = (arc_c >> 1);
		/* Valid range of arc_meta_limit: arc_meta_min - arc_c_max */
		percent = MIN(zfs_arc_meta_limit_percent, 100);
		arc_meta_limit = MAX(arc_meta_min, (percent * arc_c_max) / 100);
		percent = MIN(zfs_arc_dnode_limit_percent, 100);
		arc_dnode_limit = (percent * arc_meta_limit) / 100;
	}

	/* Valid range: 32M - <arc_c_max> */
	if ((zfs_arc_min) && (zfs_arc_min != arc_c_min) &&
	    (zfs_arc_min >= 2ULL << SPA_MAXBLOCKSHIFT) &&
	    (zfs_arc_min <= arc_c_max)) {
		arc_c_min = zfs_arc_min;
		arc_c = MAX(arc_c, arc_c_min);
	}

	/* Valid range: 16M - <arc_c_max> */
	if ((zfs_arc_meta_min) && (zfs_arc_meta_min != arc_meta_min) &&
	    (zfs_arc_meta_min >= 1ULL << SPA_MAXBLOCKSHIFT) &&
	    (zfs_arc_meta_min <= arc_c_max)) {
		arc_meta_min = zfs_arc_meta_min;
		arc_meta_limit = MAX(arc_meta_limit, arc_meta_min);
		arc_dnode_limit = arc_meta_limit / 10;
	}

	/* Valid range: <arc_meta_min> - <arc_c_max> */
	if ((zfs_arc_meta_limit) && (zfs_arc_meta_limit != arc_meta_limit) &&
	    (zfs_arc_meta_limit >= zfs_arc_meta_min) &&
	    (zfs_arc_meta_limit <= arc_c_max))
		arc_meta_limit = zfs_arc_meta_limit;

	/* Valid range: <arc_meta_min> - <arc_c_max> */
	if ((zfs_arc_dnode_limit) && (zfs_arc_dnode_limit != arc_dnode_limit) &&
	    (zfs_arc_dnode_limit >= zfs_arc_meta_min) &&
	    (zfs_arc_dnode_limit <= arc_c_max))
		arc_dnode_limit = zfs_arc_dnode_limit;

	/* Valid range: 1 - N */
	if (zfs_arc_grow_retry)
		arc_grow_retry = zfs_arc_grow_retry;

	/* Valid range: 1 - N */
	if (zfs_arc_shrink_shift) {
		arc_shrink_shift = zfs_arc_shrink_shift;
		arc_no_grow_shift = MIN(arc_no_grow_shift, arc_shrink_shift -1);
	}

	/* Valid range: 1 - N */
	if (zfs_arc_p_min_shift)
		arc_p_min_shift = zfs_arc_p_min_shift;

	/* Valid range: 1 - N ticks */
	if (zfs_arc_min_prefetch_lifespan)
		arc_min_prefetch_lifespan = zfs_arc_min_prefetch_lifespan;

	/* Valid range: 0 - 100 */
	if ((zfs_arc_lotsfree_percent >= 0) &&
	    (zfs_arc_lotsfree_percent <= 100))
		arc_lotsfree_percent = zfs_arc_lotsfree_percent;

	/* Valid range: 0 - <all physical memory> */
	if ((zfs_arc_sys_free) && (zfs_arc_sys_free != arc_sys_free))
		arc_sys_free = MIN(MAX(zfs_arc_sys_free, 0), allmem);

}

static void
arc_state_init(void)
{
	arc_anon = &ARC_anon;
	arc_mru = &ARC_mru;
	arc_mru_ghost = &ARC_mru_ghost;
	arc_mfu = &ARC_mfu;
	arc_mfu_ghost = &ARC_mfu_ghost;
	arc_l2c_only = &ARC_l2c_only;

	arc_mru->arcs_list[ARC_BUFC_METADATA] =
	    multilist_create(sizeof (arc_buf_hdr_t),
	    offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	    arc_state_multilist_index_func);
	arc_mru->arcs_list[ARC_BUFC_DATA] =
	    multilist_create(sizeof (arc_buf_hdr_t),
	    offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	    arc_state_multilist_index_func);
	arc_mru_ghost->arcs_list[ARC_BUFC_METADATA] =
	    multilist_create(sizeof (arc_buf_hdr_t),
	    offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	    arc_state_multilist_index_func);
	arc_mru_ghost->arcs_list[ARC_BUFC_DATA] =
	    multilist_create(sizeof (arc_buf_hdr_t),
	    offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	    arc_state_multilist_index_func);
	arc_mfu->arcs_list[ARC_BUFC_METADATA] =
	    multilist_create(sizeof (arc_buf_hdr_t),
	    offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	    arc_state_multilist_index_func);
	arc_mfu->arcs_list[ARC_BUFC_DATA] =
	    multilist_create(sizeof (arc_buf_hdr_t),
	    offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	    arc_state_multilist_index_func);
	arc_mfu_ghost->arcs_list[ARC_BUFC_METADATA] =
	    multilist_create(sizeof (arc_buf_hdr_t),
	    offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	    arc_state_multilist_index_func);
	arc_mfu_ghost->arcs_list[ARC_BUFC_DATA] =
	    multilist_create(sizeof (arc_buf_hdr_t),
	    offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	    arc_state_multilist_index_func);
	arc_l2c_only->arcs_list[ARC_BUFC_METADATA] =
	    multilist_create(sizeof (arc_buf_hdr_t),
	    offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	    arc_state_multilist_index_func);
	arc_l2c_only->arcs_list[ARC_BUFC_DATA] =
	    multilist_create(sizeof (arc_buf_hdr_t),
	    offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	    arc_state_multilist_index_func);

	refcount_create(&arc_anon->arcs_esize[ARC_BUFC_METADATA]);
	refcount_create(&arc_anon->arcs_esize[ARC_BUFC_DATA]);
	refcount_create(&arc_mru->arcs_esize[ARC_BUFC_METADATA]);
	refcount_create(&arc_mru->arcs_esize[ARC_BUFC_DATA]);
	refcount_create(&arc_mru_ghost->arcs_esize[ARC_BUFC_METADATA]);
	refcount_create(&arc_mru_ghost->arcs_esize[ARC_BUFC_DATA]);
	refcount_create(&arc_mfu->arcs_esize[ARC_BUFC_METADATA]);
	refcount_create(&arc_mfu->arcs_esize[ARC_BUFC_DATA]);
	refcount_create(&arc_mfu_ghost->arcs_esize[ARC_BUFC_METADATA]);
	refcount_create(&arc_mfu_ghost->arcs_esize[ARC_BUFC_DATA]);
	refcount_create(&arc_l2c_only->arcs_esize[ARC_BUFC_METADATA]);
	refcount_create(&arc_l2c_only->arcs_esize[ARC_BUFC_DATA]);

	refcount_create(&arc_anon->arcs_size);
	refcount_create(&arc_mru->arcs_size);
	refcount_create(&arc_mru_ghost->arcs_size);
	refcount_create(&arc_mfu->arcs_size);
	refcount_create(&arc_mfu_ghost->arcs_size);
	refcount_create(&arc_l2c_only->arcs_size);

	arc_anon->arcs_state = ARC_STATE_ANON;
	arc_mru->arcs_state = ARC_STATE_MRU;
	arc_mru_ghost->arcs_state = ARC_STATE_MRU_GHOST;
	arc_mfu->arcs_state = ARC_STATE_MFU;
	arc_mfu_ghost->arcs_state = ARC_STATE_MFU_GHOST;
	arc_l2c_only->arcs_state = ARC_STATE_L2C_ONLY;
}

static void
arc_state_fini(void)
{
	refcount_destroy(&arc_anon->arcs_esize[ARC_BUFC_METADATA]);
	refcount_destroy(&arc_anon->arcs_esize[ARC_BUFC_DATA]);
	refcount_destroy(&arc_mru->arcs_esize[ARC_BUFC_METADATA]);
	refcount_destroy(&arc_mru->arcs_esize[ARC_BUFC_DATA]);
	refcount_destroy(&arc_mru_ghost->arcs_esize[ARC_BUFC_METADATA]);
	refcount_destroy(&arc_mru_ghost->arcs_esize[ARC_BUFC_DATA]);
	refcount_destroy(&arc_mfu->arcs_esize[ARC_BUFC_METADATA]);
	refcount_destroy(&arc_mfu->arcs_esize[ARC_BUFC_DATA]);
	refcount_destroy(&arc_mfu_ghost->arcs_esize[ARC_BUFC_METADATA]);
	refcount_destroy(&arc_mfu_ghost->arcs_esize[ARC_BUFC_DATA]);
	refcount_destroy(&arc_l2c_only->arcs_esize[ARC_BUFC_METADATA]);
	refcount_destroy(&arc_l2c_only->arcs_esize[ARC_BUFC_DATA]);

	refcount_destroy(&arc_anon->arcs_size);
	refcount_destroy(&arc_mru->arcs_size);
	refcount_destroy(&arc_mru_ghost->arcs_size);
	refcount_destroy(&arc_mfu->arcs_size);
	refcount_destroy(&arc_mfu_ghost->arcs_size);
	refcount_destroy(&arc_l2c_only->arcs_size);

	multilist_destroy(arc_mru->arcs_list[ARC_BUFC_METADATA]);
	multilist_destroy(arc_mru_ghost->arcs_list[ARC_BUFC_METADATA]);
	multilist_destroy(arc_mfu->arcs_list[ARC_BUFC_METADATA]);
	multilist_destroy(arc_mfu_ghost->arcs_list[ARC_BUFC_METADATA]);
	multilist_destroy(arc_mru->arcs_list[ARC_BUFC_DATA]);
	multilist_destroy(arc_mru_ghost->arcs_list[ARC_BUFC_DATA]);
	multilist_destroy(arc_mfu->arcs_list[ARC_BUFC_DATA]);
	multilist_destroy(arc_mfu_ghost->arcs_list[ARC_BUFC_DATA]);
	multilist_destroy(arc_l2c_only->arcs_list[ARC_BUFC_METADATA]);
	multilist_destroy(arc_l2c_only->arcs_list[ARC_BUFC_DATA]);
}

uint64_t
arc_max_bytes(void)
{
	return (arc_c_max);
}

void
arc_init(void)
{
	uint64_t percent, allmem = arc_all_memory();

	mutex_init(&arc_reclaim_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&arc_reclaim_thread_cv, NULL, CV_DEFAULT, NULL);
	cv_init(&arc_reclaim_waiters_cv, NULL, CV_DEFAULT, NULL);

	/* Convert seconds to clock ticks */
	arc_min_prefetch_lifespan = 1 * hz;

#ifdef _KERNEL
	/*
	 * Register a shrinker to support synchronous (direct) memory
	 * reclaim from the arc.  This is done to prevent kswapd from
	 * swapping out pages when it is preferable to shrink the arc.
	 */
	spl_register_shrinker(&arc_shrinker);

	/* Set to 1/64 of all memory or a minimum of 512K */
	arc_sys_free = MAX(allmem / 64, (512 * 1024));
	arc_need_free = 0;
#endif

	/* Set max to 1/2 of all memory */
	arc_c_max = allmem / 2;

	/*
	 * In userland, there's only the memory pressure that we artificially
	 * create (see arc_available_memory()).  Don't let arc_c get too
	 * small, because it can cause transactions to be larger than
	 * arc_c, causing arc_tempreserve_space() to fail.
	 */
#ifndef	_KERNEL
	arc_c_min = MAX(arc_c_max / 2, 2ULL << SPA_MAXBLOCKSHIFT);
#else
	arc_c_min = 2ULL << SPA_MAXBLOCKSHIFT;
#endif

	arc_c = arc_c_max;
	arc_p = (arc_c >> 1);
	arc_size = 0;

	/* Set min to 1/2 of arc_c_min */
	arc_meta_min = 1ULL << SPA_MAXBLOCKSHIFT;
	/* Initialize maximum observed usage to zero */
	arc_meta_max = 0;
	/*
	 * Set arc_meta_limit to a percent of arc_c_max with a floor of
	 * arc_meta_min, and a ceiling of arc_c_max.
	 */
	percent = MIN(zfs_arc_meta_limit_percent, 100);
	arc_meta_limit = MAX(arc_meta_min, (percent * arc_c_max) / 100);
	percent = MIN(zfs_arc_dnode_limit_percent, 100);
	arc_dnode_limit = (percent * arc_meta_limit) / 100;

	/* Apply user specified tunings */
	arc_tuning_update();

	/* if kmem_flags are set, lets try to use less memory */
	if (kmem_debugging())
		arc_c = arc_c / 2;
	if (arc_c < arc_c_min)
		arc_c = arc_c_min;

	arc_state_init();
	buf_init();

	list_create(&arc_prune_list, sizeof (arc_prune_t),
	    offsetof(arc_prune_t, p_node));
	mutex_init(&arc_prune_mtx, NULL, MUTEX_DEFAULT, NULL);

	arc_prune_taskq = taskq_create("arc_prune", max_ncpus, defclsyspri,
	    max_ncpus, INT_MAX, TASKQ_PREPOPULATE | TASKQ_DYNAMIC);

	arc_reclaim_thread_exit = B_FALSE;

	arc_ksp = kstat_create("zfs", 0, "arcstats", "misc", KSTAT_TYPE_NAMED,
	    sizeof (arc_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL);

	if (arc_ksp != NULL) {
		arc_ksp->ks_data = &arc_stats;
		arc_ksp->ks_update = arc_kstat_update;
		kstat_install(arc_ksp);
	}

	(void) thread_create(NULL, 0, arc_reclaim_thread, NULL, 0, &p0,
	    TS_RUN, defclsyspri);

	arc_dead = B_FALSE;
	arc_warm = B_FALSE;

	/*
	 * Calculate maximum amount of dirty data per pool.
	 *
	 * If it has been set by a module parameter, take that.
	 * Otherwise, use a percentage of physical memory defined by
	 * zfs_dirty_data_max_percent (default 10%) with a cap at
	 * zfs_dirty_data_max_max (default 4G or 25% of physical memory).
	 */
	if (zfs_dirty_data_max_max == 0)
		zfs_dirty_data_max_max = MIN(4ULL * 1024 * 1024 * 1024,
		    allmem * zfs_dirty_data_max_max_percent / 100);

	if (zfs_dirty_data_max == 0) {
		zfs_dirty_data_max = allmem *
		    zfs_dirty_data_max_percent / 100;
		zfs_dirty_data_max = MIN(zfs_dirty_data_max,
		    zfs_dirty_data_max_max);
	}
}

void
arc_fini(void)
{
	arc_prune_t *p;

#ifdef _KERNEL
	spl_unregister_shrinker(&arc_shrinker);
#endif /* _KERNEL */

	mutex_enter(&arc_reclaim_lock);
	arc_reclaim_thread_exit = B_TRUE;
	/*
	 * The reclaim thread will set arc_reclaim_thread_exit back to
	 * B_FALSE when it is finished exiting; we're waiting for that.
	 */
	while (arc_reclaim_thread_exit) {
		cv_signal(&arc_reclaim_thread_cv);
		cv_wait(&arc_reclaim_thread_cv, &arc_reclaim_lock);
	}
	mutex_exit(&arc_reclaim_lock);

	/* Use B_TRUE to ensure *all* buffers are evicted */
	arc_flush(NULL, B_TRUE);

	arc_dead = B_TRUE;

	if (arc_ksp != NULL) {
		kstat_delete(arc_ksp);
		arc_ksp = NULL;
	}

	taskq_wait(arc_prune_taskq);
	taskq_destroy(arc_prune_taskq);

	mutex_enter(&arc_prune_mtx);
	while ((p = list_head(&arc_prune_list)) != NULL) {
		list_remove(&arc_prune_list, p);
		refcount_remove(&p->p_refcnt, &arc_prune_list);
		refcount_destroy(&p->p_refcnt);
		kmem_free(p, sizeof (*p));
	}
	mutex_exit(&arc_prune_mtx);

	list_destroy(&arc_prune_list);
	mutex_destroy(&arc_prune_mtx);
	mutex_destroy(&arc_reclaim_lock);
	cv_destroy(&arc_reclaim_thread_cv);
	cv_destroy(&arc_reclaim_waiters_cv);

	arc_state_fini();
	buf_fini();

	ASSERT0(arc_loaned_bytes);
}

/*
 * Level 2 ARC
 *
 * The level 2 ARC (L2ARC) is a cache layer in-between main memory and disk.
 * It uses dedicated storage devices to hold cached data, which are populated
 * using large infrequent writes.  The main role of this cache is to boost
 * the performance of random read workloads.  The intended L2ARC devices
 * include short-stroked disks, solid state disks, and other media with
 * substantially faster read latency than disk.
 *
 *                 +-----------------------+
 *                 |         ARC           |
 *                 +-----------------------+
 *                    |         ^     ^
 *                    |         |     |
 *      l2arc_feed_thread()    arc_read()
 *                    |         |     |
 *                    |  l2arc read   |
 *                    V         |     |
 *               +---------------+    |
 *               |     L2ARC     |    |
 *               +---------------+    |
 *                   |    ^           |
 *          l2arc_write() |           |
 *                   |    |           |
 *                   V    |           |
 *                 +-------+      +-------+
 *                 | vdev  |      | vdev  |
 *                 | cache |      | cache |
 *                 +-------+      +-------+
 *                 +=========+     .-----.
 *                 :  L2ARC  :    |-_____-|
 *                 : devices :    | Disks |
 *                 +=========+    `-_____-'
 *
 * Read requests are satisfied from the following sources, in order:
 *
 *	1) ARC
 *	2) vdev cache of L2ARC devices
 *	3) L2ARC devices
 *	4) vdev cache of disks
 *	5) disks
 *
 * Some L2ARC device types exhibit extremely slow write performance.
 * To accommodate for this there are some significant differences between
 * the L2ARC and traditional cache design:
 *
 * 1. There is no eviction path from the ARC to the L2ARC.  Evictions from
 * the ARC behave as usual, freeing buffers and placing headers on ghost
 * lists.  The ARC does not send buffers to the L2ARC during eviction as
 * this would add inflated write latencies for all ARC memory pressure.
 *
 * 2. The L2ARC attempts to cache data from the ARC before it is evicted.
 * It does this by periodically scanning buffers from the eviction-end of
 * the MFU and MRU ARC lists, copying them to the L2ARC devices if they are
 * not already there. It scans until a headroom of buffers is satisfied,
 * which itself is a buffer for ARC eviction. If a compressible buffer is
 * found during scanning and selected for writing to an L2ARC device, we
 * temporarily boost scanning headroom during the next scan cycle to make
 * sure we adapt to compression effects (which might significantly reduce
 * the data volume we write to L2ARC). The thread that does this is
 * l2arc_feed_thread(), illustrated below; example sizes are included to
 * provide a better sense of ratio than this diagram:
 *
 *	       head -->                        tail
 *	        +---------------------+----------+
 *	ARC_mfu |:::::#:::::::::::::::|o#o###o###|-->.   # already on L2ARC
 *	        +---------------------+----------+   |   o L2ARC eligible
 *	ARC_mru |:#:::::::::::::::::::|#o#ooo####|-->|   : ARC buffer
 *	        +---------------------+----------+   |
 *	             15.9 Gbytes      ^ 32 Mbytes    |
 *	                           headroom          |
 *	                                      l2arc_feed_thread()
 *	                                             |
 *	                 l2arc write hand <--[oooo]--'
 *	                         |           8 Mbyte
 *	                         |          write max
 *	                         V
 *		  +==============================+
 *	L2ARC dev |####|#|###|###|    |####| ... |
 *	          +==============================+
 *	                     32 Gbytes
 *
 * 3. If an ARC buffer is copied to the L2ARC but then hit instead of
 * evicted, then the L2ARC has cached a buffer much sooner than it probably
 * needed to, potentially wasting L2ARC device bandwidth and storage.  It is
 * safe to say that this is an uncommon case, since buffers at the end of
 * the ARC lists have moved there due to inactivity.
 *
 * 4. If the ARC evicts faster than the L2ARC can maintain a headroom,
 * then the L2ARC simply misses copying some buffers.  This serves as a
 * pressure valve to prevent heavy read workloads from both stalling the ARC
 * with waits and clogging the L2ARC with writes.  This also helps prevent
 * the potential for the L2ARC to churn if it attempts to cache content too
 * quickly, such as during backups of the entire pool.
 *
 * 5. After system boot and before the ARC has filled main memory, there are
 * no evictions from the ARC and so the tails of the ARC_mfu and ARC_mru
 * lists can remain mostly static.  Instead of searching from tail of these
 * lists as pictured, the l2arc_feed_thread() will search from the list heads
 * for eligible buffers, greatly increasing its chance of finding them.
 *
 * The L2ARC device write speed is also boosted during this time so that
 * the L2ARC warms up faster.  Since there have been no ARC evictions yet,
 * there are no L2ARC reads, and no fear of degrading read performance
 * through increased writes.
 *
 * 6. Writes to the L2ARC devices are grouped and sent in-sequence, so that
 * the vdev queue can aggregate them into larger and fewer writes.  Each
 * device is written to in a rotor fashion, sweeping writes through
 * available space then repeating.
 *
 * 7. The L2ARC does not store dirty content.  It never needs to flush
 * write buffers back to disk based storage.
 *
 * 8. If an ARC buffer is written (and dirtied) which also exists in the
 * L2ARC, the now stale L2ARC buffer is immediately dropped.
 *
 * The performance of the L2ARC can be tweaked by a number of tunables, which
 * may be necessary for different workloads:
 *
 *	l2arc_write_max		max write bytes per interval
 *	l2arc_write_boost	extra write bytes during device warmup
 *	l2arc_noprefetch	skip caching prefetched buffers
 *	l2arc_headroom		number of max device writes to precache
 *	l2arc_headroom_boost	when we find compressed buffers during ARC
 *				scanning, we multiply headroom by this
 *				percentage factor for the next scan cycle,
 *				since more compressed buffers are likely to
 *				be present
 *	l2arc_feed_secs		seconds between L2ARC writing
 *
 * Tunables may be removed or added as future performance improvements are
 * integrated, and also may become zpool properties.
 *
 * There are three key functions that control how the L2ARC warms up:
 *
 *	l2arc_write_eligible()	check if a buffer is eligible to cache
 *	l2arc_write_size()	calculate how much to write
 *	l2arc_write_interval()	calculate sleep delay between writes
 *
 * These three functions determine what to write, how much, and how quickly
 * to send writes.
 */

static boolean_t
l2arc_write_eligible(uint64_t spa_guid, arc_buf_hdr_t *hdr)
{
	/*
	 * A buffer is *not* eligible for the L2ARC if it:
	 * 1. belongs to a different spa.
	 * 2. is already cached on the L2ARC.
	 * 3. has an I/O in progress (it may be an incomplete read).
	 * 4. is flagged not eligible (zfs property).
	 */
	if (hdr->b_spa != spa_guid || HDR_HAS_L2HDR(hdr) ||
	    HDR_IO_IN_PROGRESS(hdr) || !HDR_L2CACHE(hdr))
		return (B_FALSE);

	return (B_TRUE);
}

static uint64_t
l2arc_write_size(void)
{
	uint64_t size;

	/*
	 * Make sure our globals have meaningful values in case the user
	 * altered them.
	 */
	size = l2arc_write_max;
	if (size == 0) {
		cmn_err(CE_NOTE, "Bad value for l2arc_write_max, value must "
		    "be greater than zero, resetting it to the default (%d)",
		    L2ARC_WRITE_SIZE);
		size = l2arc_write_max = L2ARC_WRITE_SIZE;
	}

	if (arc_warm == B_FALSE)
		size += l2arc_write_boost;

	return (size);

}

static clock_t
l2arc_write_interval(clock_t began, uint64_t wanted, uint64_t wrote)
{
	clock_t interval, next, now;

	/*
	 * If the ARC lists are busy, increase our write rate; if the
	 * lists are stale, idle back.  This is achieved by checking
	 * how much we previously wrote - if it was more than half of
	 * what we wanted, schedule the next write much sooner.
	 */
	if (l2arc_feed_again && wrote > (wanted / 2))
		interval = (hz * l2arc_feed_min_ms) / 1000;
	else
		interval = hz * l2arc_feed_secs;

	now = ddi_get_lbolt();
	next = MAX(now, MIN(now + interval, began + interval));

	return (next);
}

/*
 * Cycle through L2ARC devices.  This is how L2ARC load balances.
 * If a device is returned, this also returns holding the spa config lock.
 */
static l2arc_dev_t *
l2arc_dev_get_next(void)
{
	l2arc_dev_t *first, *next = NULL;

	/*
	 * Lock out the removal of spas (spa_namespace_lock), then removal
	 * of cache devices (l2arc_dev_mtx).  Once a device has been selected,
	 * both locks will be dropped and a spa config lock held instead.
	 */
	mutex_enter(&spa_namespace_lock);
	mutex_enter(&l2arc_dev_mtx);

	/* if there are no vdevs, there is nothing to do */
	if (l2arc_ndev == 0)
		goto out;

	first = NULL;
	next = l2arc_dev_last;
	do {
		/* loop around the list looking for a non-faulted vdev */
		if (next == NULL) {
			next = list_head(l2arc_dev_list);
		} else {
			next = list_next(l2arc_dev_list, next);
			if (next == NULL)
				next = list_head(l2arc_dev_list);
		}

		/* if we have come back to the start, bail out */
		if (first == NULL)
			first = next;
		else if (next == first)
			break;

	} while (vdev_is_dead(next->l2ad_vdev));

	/* if we were unable to find any usable vdevs, return NULL */
	if (vdev_is_dead(next->l2ad_vdev))
		next = NULL;

	l2arc_dev_last = next;

out:
	mutex_exit(&l2arc_dev_mtx);

	/*
	 * Grab the config lock to prevent the 'next' device from being
	 * removed while we are writing to it.
	 */
	if (next != NULL)
		spa_config_enter(next->l2ad_spa, SCL_L2ARC, next, RW_READER);
	mutex_exit(&spa_namespace_lock);

	return (next);
}

/*
 * Free buffers that were tagged for destruction.
 */
static void
l2arc_do_free_on_write(void)
{
	list_t *buflist;
	l2arc_data_free_t *df, *df_prev;

	mutex_enter(&l2arc_free_on_write_mtx);
	buflist = l2arc_free_on_write;

	for (df = list_tail(buflist); df; df = df_prev) {
		df_prev = list_prev(buflist, df);
		ASSERT3P(df->l2df_abd, !=, NULL);
		abd_free(df->l2df_abd);
		list_remove(buflist, df);
		kmem_free(df, sizeof (l2arc_data_free_t));
	}

	mutex_exit(&l2arc_free_on_write_mtx);
}

/*
 * A write to a cache device has completed.  Update all headers to allow
 * reads from these buffers to begin.
 */
static void
l2arc_write_done(zio_t *zio)
{
	l2arc_write_callback_t *cb;
	l2arc_dev_t *dev;
	list_t *buflist;
	arc_buf_hdr_t *head, *hdr, *hdr_prev;
	kmutex_t *hash_lock;
	int64_t bytes_dropped = 0;

	cb = zio->io_private;
	ASSERT3P(cb, !=, NULL);
	dev = cb->l2wcb_dev;
	ASSERT3P(dev, !=, NULL);
	head = cb->l2wcb_head;
	ASSERT3P(head, !=, NULL);
	buflist = &dev->l2ad_buflist;
	ASSERT3P(buflist, !=, NULL);
	DTRACE_PROBE2(l2arc__iodone, zio_t *, zio,
	    l2arc_write_callback_t *, cb);

	if (zio->io_error != 0)
		ARCSTAT_BUMP(arcstat_l2_writes_error);

	/*
	 * All writes completed, or an error was hit.
	 */
top:
	mutex_enter(&dev->l2ad_mtx);
	for (hdr = list_prev(buflist, head); hdr; hdr = hdr_prev) {
		hdr_prev = list_prev(buflist, hdr);

		hash_lock = HDR_LOCK(hdr);

		/*
		 * We cannot use mutex_enter or else we can deadlock
		 * with l2arc_write_buffers (due to swapping the order
		 * the hash lock and l2ad_mtx are taken).
		 */
		if (!mutex_tryenter(hash_lock)) {
			/*
			 * Missed the hash lock. We must retry so we
			 * don't leave the ARC_FLAG_L2_WRITING bit set.
			 */
			ARCSTAT_BUMP(arcstat_l2_writes_lock_retry);

			/*
			 * We don't want to rescan the headers we've
			 * already marked as having been written out, so
			 * we reinsert the head node so we can pick up
			 * where we left off.
			 */
			list_remove(buflist, head);
			list_insert_after(buflist, hdr, head);

			mutex_exit(&dev->l2ad_mtx);

			/*
			 * We wait for the hash lock to become available
			 * to try and prevent busy waiting, and increase
			 * the chance we'll be able to acquire the lock
			 * the next time around.
			 */
			mutex_enter(hash_lock);
			mutex_exit(hash_lock);
			goto top;
		}

		/*
		 * We could not have been moved into the arc_l2c_only
		 * state while in-flight due to our ARC_FLAG_L2_WRITING
		 * bit being set. Let's just ensure that's being enforced.
		 */
		ASSERT(HDR_HAS_L1HDR(hdr));

		/*
		 * Skipped - drop L2ARC entry and mark the header as no
		 * longer L2 eligibile.
		 */
		if (zio->io_error != 0) {
			/*
			 * Error - drop L2ARC entry.
			 */
			list_remove(buflist, hdr);
			arc_hdr_clear_flags(hdr, ARC_FLAG_HAS_L2HDR);

			ARCSTAT_INCR(arcstat_l2_asize, -arc_hdr_size(hdr));
			ARCSTAT_INCR(arcstat_l2_size, -HDR_GET_LSIZE(hdr));

			bytes_dropped += arc_hdr_size(hdr);
			(void) refcount_remove_many(&dev->l2ad_alloc,
			    arc_hdr_size(hdr), hdr);
		}

		/*
		 * Allow ARC to begin reads and ghost list evictions to
		 * this L2ARC entry.
		 */
		arc_hdr_clear_flags(hdr, ARC_FLAG_L2_WRITING);

		mutex_exit(hash_lock);
	}

	atomic_inc_64(&l2arc_writes_done);
	list_remove(buflist, head);
	ASSERT(!HDR_HAS_L1HDR(head));
	kmem_cache_free(hdr_l2only_cache, head);
	mutex_exit(&dev->l2ad_mtx);

	vdev_space_update(dev->l2ad_vdev, -bytes_dropped, 0, 0);

	l2arc_do_free_on_write();

	kmem_free(cb, sizeof (l2arc_write_callback_t));
}

/*
 * A read to a cache device completed.  Validate buffer contents before
 * handing over to the regular ARC routines.
 */
static void
l2arc_read_done(zio_t *zio)
{
	l2arc_read_callback_t *cb;
	arc_buf_hdr_t *hdr;
	kmutex_t *hash_lock;
	boolean_t valid_cksum;

	ASSERT3P(zio->io_vd, !=, NULL);
	ASSERT(zio->io_flags & ZIO_FLAG_DONT_PROPAGATE);

	spa_config_exit(zio->io_spa, SCL_L2ARC, zio->io_vd);

	cb = zio->io_private;
	ASSERT3P(cb, !=, NULL);
	hdr = cb->l2rcb_hdr;
	ASSERT3P(hdr, !=, NULL);

	hash_lock = HDR_LOCK(hdr);
	mutex_enter(hash_lock);
	ASSERT3P(hash_lock, ==, HDR_LOCK(hdr));

	ASSERT3P(zio->io_abd, !=, NULL);

	/*
	 * Check this survived the L2ARC journey.
	 */
	ASSERT3P(zio->io_abd, ==, hdr->b_l1hdr.b_pabd);
	zio->io_bp_copy = cb->l2rcb_bp;	/* XXX fix in L2ARC 2.0	*/
	zio->io_bp = &zio->io_bp_copy;	/* XXX fix in L2ARC 2.0	*/

	valid_cksum = arc_cksum_is_equal(hdr, zio);
	if (valid_cksum && zio->io_error == 0 && !HDR_L2_EVICTED(hdr)) {
		mutex_exit(hash_lock);
		zio->io_private = hdr;
		arc_read_done(zio);
	} else {
		mutex_exit(hash_lock);
		/*
		 * Buffer didn't survive caching.  Increment stats and
		 * reissue to the original storage device.
		 */
		if (zio->io_error != 0) {
			ARCSTAT_BUMP(arcstat_l2_io_error);
		} else {
			zio->io_error = SET_ERROR(EIO);
		}
		if (!valid_cksum)
			ARCSTAT_BUMP(arcstat_l2_cksum_bad);

		/*
		 * If there's no waiter, issue an async i/o to the primary
		 * storage now.  If there *is* a waiter, the caller must
		 * issue the i/o in a context where it's OK to block.
		 */
		if (zio->io_waiter == NULL) {
			zio_t *pio = zio_unique_parent(zio);

			ASSERT(!pio || pio->io_child_type == ZIO_CHILD_LOGICAL);

			zio_nowait(zio_read(pio, zio->io_spa, zio->io_bp,
			    hdr->b_l1hdr.b_pabd, zio->io_size, arc_read_done,
			    hdr, zio->io_priority, cb->l2rcb_flags,
			    &cb->l2rcb_zb));
		}
	}

	kmem_free(cb, sizeof (l2arc_read_callback_t));
}

/*
 * This is the list priority from which the L2ARC will search for pages to
 * cache.  This is used within loops (0..3) to cycle through lists in the
 * desired order.  This order can have a significant effect on cache
 * performance.
 *
 * Currently the metadata lists are hit first, MFU then MRU, followed by
 * the data lists.  This function returns a locked list, and also returns
 * the lock pointer.
 */
static multilist_sublist_t *
l2arc_sublist_lock(int list_num)
{
	multilist_t *ml = NULL;
	unsigned int idx;

	ASSERT(list_num >= 0 && list_num < L2ARC_FEED_TYPES);

	switch (list_num) {
	case 0:
		ml = arc_mfu->arcs_list[ARC_BUFC_METADATA];
		break;
	case 1:
		ml = arc_mru->arcs_list[ARC_BUFC_METADATA];
		break;
	case 2:
		ml = arc_mfu->arcs_list[ARC_BUFC_DATA];
		break;
	case 3:
		ml = arc_mru->arcs_list[ARC_BUFC_DATA];
		break;
	default:
		return (NULL);
	}

	/*
	 * Return a randomly-selected sublist. This is acceptable
	 * because the caller feeds only a little bit of data for each
	 * call (8MB). Subsequent calls will result in different
	 * sublists being selected.
	 */
	idx = multilist_get_random_index(ml);
	return (multilist_sublist_lock(ml, idx));
}

/*
 * Evict buffers from the device write hand to the distance specified in
 * bytes.  This distance may span populated buffers, it may span nothing.
 * This is clearing a region on the L2ARC device ready for writing.
 * If the 'all' boolean is set, every buffer is evicted.
 */
static void
l2arc_evict(l2arc_dev_t *dev, uint64_t distance, boolean_t all)
{
	list_t *buflist;
	arc_buf_hdr_t *hdr, *hdr_prev;
	kmutex_t *hash_lock;
	uint64_t taddr;

	buflist = &dev->l2ad_buflist;

	if (!all && dev->l2ad_first) {
		/*
		 * This is the first sweep through the device.  There is
		 * nothing to evict.
		 */
		return;
	}

	if (dev->l2ad_hand >= (dev->l2ad_end - (2 * distance))) {
		/*
		 * When nearing the end of the device, evict to the end
		 * before the device write hand jumps to the start.
		 */
		taddr = dev->l2ad_end;
	} else {
		taddr = dev->l2ad_hand + distance;
	}
	DTRACE_PROBE4(l2arc__evict, l2arc_dev_t *, dev, list_t *, buflist,
	    uint64_t, taddr, boolean_t, all);

top:
	mutex_enter(&dev->l2ad_mtx);
	for (hdr = list_tail(buflist); hdr; hdr = hdr_prev) {
		hdr_prev = list_prev(buflist, hdr);

		hash_lock = HDR_LOCK(hdr);

		/*
		 * We cannot use mutex_enter or else we can deadlock
		 * with l2arc_write_buffers (due to swapping the order
		 * the hash lock and l2ad_mtx are taken).
		 */
		if (!mutex_tryenter(hash_lock)) {
			/*
			 * Missed the hash lock.  Retry.
			 */
			ARCSTAT_BUMP(arcstat_l2_evict_lock_retry);
			mutex_exit(&dev->l2ad_mtx);
			mutex_enter(hash_lock);
			mutex_exit(hash_lock);
			goto top;
		}

		if (HDR_L2_WRITE_HEAD(hdr)) {
			/*
			 * We hit a write head node.  Leave it for
			 * l2arc_write_done().
			 */
			list_remove(buflist, hdr);
			mutex_exit(hash_lock);
			continue;
		}

		if (!all && HDR_HAS_L2HDR(hdr) &&
		    (hdr->b_l2hdr.b_daddr > taddr ||
		    hdr->b_l2hdr.b_daddr < dev->l2ad_hand)) {
			/*
			 * We've evicted to the target address,
			 * or the end of the device.
			 */
			mutex_exit(hash_lock);
			break;
		}

		ASSERT(HDR_HAS_L2HDR(hdr));
		if (!HDR_HAS_L1HDR(hdr)) {
			ASSERT(!HDR_L2_READING(hdr));
			/*
			 * This doesn't exist in the ARC.  Destroy.
			 * arc_hdr_destroy() will call list_remove()
			 * and decrement arcstat_l2_size.
			 */
			arc_change_state(arc_anon, hdr, hash_lock);
			arc_hdr_destroy(hdr);
		} else {
			ASSERT(hdr->b_l1hdr.b_state != arc_l2c_only);
			ARCSTAT_BUMP(arcstat_l2_evict_l1cached);
			/*
			 * Invalidate issued or about to be issued
			 * reads, since we may be about to write
			 * over this location.
			 */
			if (HDR_L2_READING(hdr)) {
				ARCSTAT_BUMP(arcstat_l2_evict_reading);
				arc_hdr_set_flags(hdr, ARC_FLAG_L2_EVICTED);
			}

			/* Ensure this header has finished being written */
			ASSERT(!HDR_L2_WRITING(hdr));

			arc_hdr_l2hdr_destroy(hdr);
		}
		mutex_exit(hash_lock);
	}
	mutex_exit(&dev->l2ad_mtx);
}

/*
 * Find and write ARC buffers to the L2ARC device.
 *
 * An ARC_FLAG_L2_WRITING flag is set so that the L2ARC buffers are not valid
 * for reading until they have completed writing.
 * The headroom_boost is an in-out parameter used to maintain headroom boost
 * state between calls to this function.
 *
 * Returns the number of bytes actually written (which may be smaller than
 * the delta by which the device hand has changed due to alignment).
 */
static uint64_t
l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz)
{
	arc_buf_hdr_t *hdr, *hdr_prev, *head;
	uint64_t write_asize, write_psize, write_sz, headroom;
	boolean_t full;
	l2arc_write_callback_t *cb;
	zio_t *pio, *wzio;
	uint64_t guid = spa_load_guid(spa);
	int try;

	ASSERT3P(dev->l2ad_vdev, !=, NULL);

	pio = NULL;
	write_sz = write_asize = write_psize = 0;
	full = B_FALSE;
	head = kmem_cache_alloc(hdr_l2only_cache, KM_PUSHPAGE);
	arc_hdr_set_flags(head, ARC_FLAG_L2_WRITE_HEAD | ARC_FLAG_HAS_L2HDR);

	/*
	 * Copy buffers for L2ARC writing.
	 */
	for (try = 0; try < L2ARC_FEED_TYPES; try++) {
		multilist_sublist_t *mls = l2arc_sublist_lock(try);
		uint64_t passed_sz = 0;

		VERIFY3P(mls, !=, NULL);

		/*
		 * L2ARC fast warmup.
		 *
		 * Until the ARC is warm and starts to evict, read from the
		 * head of the ARC lists rather than the tail.
		 */
		if (arc_warm == B_FALSE)
			hdr = multilist_sublist_head(mls);
		else
			hdr = multilist_sublist_tail(mls);

		headroom = target_sz * l2arc_headroom;
		if (zfs_compressed_arc_enabled)
			headroom = (headroom * l2arc_headroom_boost) / 100;

		for (; hdr; hdr = hdr_prev) {
			kmutex_t *hash_lock;
			uint64_t asize, size;
			abd_t *to_write;

			if (arc_warm == B_FALSE)
				hdr_prev = multilist_sublist_next(mls, hdr);
			else
				hdr_prev = multilist_sublist_prev(mls, hdr);

			hash_lock = HDR_LOCK(hdr);
			if (!mutex_tryenter(hash_lock)) {
				/*
				 * Skip this buffer rather than waiting.
				 */
				continue;
			}

			passed_sz += HDR_GET_LSIZE(hdr);
			if (passed_sz > headroom) {
				/*
				 * Searched too far.
				 */
				mutex_exit(hash_lock);
				break;
			}

			if (!l2arc_write_eligible(guid, hdr)) {
				mutex_exit(hash_lock);
				continue;
			}

			if ((write_asize + HDR_GET_LSIZE(hdr)) > target_sz) {
				full = B_TRUE;
				mutex_exit(hash_lock);
				break;
			}

			if (pio == NULL) {
				/*
				 * Insert a dummy header on the buflist so
				 * l2arc_write_done() can find where the
				 * write buffers begin without searching.
				 */
				mutex_enter(&dev->l2ad_mtx);
				list_insert_head(&dev->l2ad_buflist, head);
				mutex_exit(&dev->l2ad_mtx);

				cb = kmem_alloc(
				    sizeof (l2arc_write_callback_t), KM_SLEEP);
				cb->l2wcb_dev = dev;
				cb->l2wcb_head = head;
				pio = zio_root(spa, l2arc_write_done, cb,
				    ZIO_FLAG_CANFAIL);
			}

			hdr->b_l2hdr.b_dev = dev;
			hdr->b_l2hdr.b_hits = 0;

			hdr->b_l2hdr.b_daddr = dev->l2ad_hand;
			arc_hdr_set_flags(hdr,
			    ARC_FLAG_L2_WRITING | ARC_FLAG_HAS_L2HDR);

			mutex_enter(&dev->l2ad_mtx);
			list_insert_head(&dev->l2ad_buflist, hdr);
			mutex_exit(&dev->l2ad_mtx);

			/*
			 * We rely on the L1 portion of the header below, so
			 * it's invalid for this header to have been evicted out
			 * of the ghost cache, prior to being written out. The
			 * ARC_FLAG_L2_WRITING bit ensures this won't happen.
			 */
			ASSERT(HDR_HAS_L1HDR(hdr));

			ASSERT3U(HDR_GET_PSIZE(hdr), >, 0);
			ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
			ASSERT3U(arc_hdr_size(hdr), >, 0);
			size = arc_hdr_size(hdr);

			(void) refcount_add_many(&dev->l2ad_alloc, size, hdr);

			/*
			 * Normally the L2ARC can use the hdr's data, but if
			 * we're sharing data between the hdr and one of its
			 * bufs, L2ARC needs its own copy of the data so that
			 * the ZIO below can't race with the buf consumer. To
			 * ensure that this copy will be available for the
			 * lifetime of the ZIO and be cleaned up afterwards, we
			 * add it to the l2arc_free_on_write queue.
			 */
			if (!HDR_SHARED_DATA(hdr)) {
				to_write = hdr->b_l1hdr.b_pabd;
			} else {
				to_write = abd_alloc_for_io(size,
				    HDR_ISTYPE_METADATA(hdr));
				abd_copy(to_write, hdr->b_l1hdr.b_pabd, size);
				l2arc_free_abd_on_write(to_write, size,
				    arc_buf_type(hdr));
			}
			wzio = zio_write_phys(pio, dev->l2ad_vdev,
			    hdr->b_l2hdr.b_daddr, size, to_write,
			    ZIO_CHECKSUM_OFF, NULL, hdr,
			    ZIO_PRIORITY_ASYNC_WRITE,
			    ZIO_FLAG_CANFAIL, B_FALSE);

			write_sz += HDR_GET_LSIZE(hdr);
			DTRACE_PROBE2(l2arc__write, vdev_t *, dev->l2ad_vdev,
			    zio_t *, wzio);

			write_asize += size;
			/*
			 * Keep the clock hand suitably device-aligned.
			 */
			asize = vdev_psize_to_asize(dev->l2ad_vdev, size);
			write_psize += asize;
			dev->l2ad_hand += asize;

			mutex_exit(hash_lock);

			(void) zio_nowait(wzio);
		}

		multilist_sublist_unlock(mls);

		if (full == B_TRUE)
			break;
	}

	/* No buffers selected for writing? */
	if (pio == NULL) {
		ASSERT0(write_sz);
		ASSERT(!HDR_HAS_L1HDR(head));
		kmem_cache_free(hdr_l2only_cache, head);
		return (0);
	}

	ASSERT3U(write_asize, <=, target_sz);
	ARCSTAT_BUMP(arcstat_l2_writes_sent);
	ARCSTAT_INCR(arcstat_l2_write_bytes, write_asize);
	ARCSTAT_INCR(arcstat_l2_size, write_sz);
	ARCSTAT_INCR(arcstat_l2_asize, write_asize);
	vdev_space_update(dev->l2ad_vdev, write_asize, 0, 0);

	/*
	 * Bump device hand to the device start if it is approaching the end.
	 * l2arc_evict() will already have evicted ahead for this case.
	 */
	if (dev->l2ad_hand >= (dev->l2ad_end - target_sz)) {
		dev->l2ad_hand = dev->l2ad_start;
		dev->l2ad_first = B_FALSE;
	}

	dev->l2ad_writing = B_TRUE;
	(void) zio_wait(pio);
	dev->l2ad_writing = B_FALSE;

	return (write_asize);
}

/*
 * This thread feeds the L2ARC at regular intervals.  This is the beating
 * heart of the L2ARC.
 */
static void
l2arc_feed_thread(void)
{
	callb_cpr_t cpr;
	l2arc_dev_t *dev;
	spa_t *spa;
	uint64_t size, wrote;
	clock_t begin, next = ddi_get_lbolt();
	fstrans_cookie_t cookie;

	CALLB_CPR_INIT(&cpr, &l2arc_feed_thr_lock, callb_generic_cpr, FTAG);

	mutex_enter(&l2arc_feed_thr_lock);

	cookie = spl_fstrans_mark();
	while (l2arc_thread_exit == 0) {
		CALLB_CPR_SAFE_BEGIN(&cpr);
		(void) cv_timedwait_sig(&l2arc_feed_thr_cv,
		    &l2arc_feed_thr_lock, next);
		CALLB_CPR_SAFE_END(&cpr, &l2arc_feed_thr_lock);
		next = ddi_get_lbolt() + hz;

		/*
		 * Quick check for L2ARC devices.
		 */
		mutex_enter(&l2arc_dev_mtx);
		if (l2arc_ndev == 0) {
			mutex_exit(&l2arc_dev_mtx);
			continue;
		}
		mutex_exit(&l2arc_dev_mtx);
		begin = ddi_get_lbolt();

		/*
		 * This selects the next l2arc device to write to, and in
		 * doing so the next spa to feed from: dev->l2ad_spa.   This
		 * will return NULL if there are now no l2arc devices or if
		 * they are all faulted.
		 *
		 * If a device is returned, its spa's config lock is also
		 * held to prevent device removal.  l2arc_dev_get_next()
		 * will grab and release l2arc_dev_mtx.
		 */
		if ((dev = l2arc_dev_get_next()) == NULL)
			continue;

		spa = dev->l2ad_spa;
		ASSERT3P(spa, !=, NULL);

		/*
		 * If the pool is read-only then force the feed thread to
		 * sleep a little longer.
		 */
		if (!spa_writeable(spa)) {
			next = ddi_get_lbolt() + 5 * l2arc_feed_secs * hz;
			spa_config_exit(spa, SCL_L2ARC, dev);
			continue;
		}

		/*
		 * Avoid contributing to memory pressure.
		 */
		if (arc_reclaim_needed()) {
			ARCSTAT_BUMP(arcstat_l2_abort_lowmem);
			spa_config_exit(spa, SCL_L2ARC, dev);
			continue;
		}

		ARCSTAT_BUMP(arcstat_l2_feeds);

		size = l2arc_write_size();

		/*
		 * Evict L2ARC buffers that will be overwritten.
		 */
		l2arc_evict(dev, size, B_FALSE);

		/*
		 * Write ARC buffers.
		 */
		wrote = l2arc_write_buffers(spa, dev, size);

		/*
		 * Calculate interval between writes.
		 */
		next = l2arc_write_interval(begin, size, wrote);
		spa_config_exit(spa, SCL_L2ARC, dev);
	}
	spl_fstrans_unmark(cookie);

	l2arc_thread_exit = 0;
	cv_broadcast(&l2arc_feed_thr_cv);
	CALLB_CPR_EXIT(&cpr);		/* drops l2arc_feed_thr_lock */
	thread_exit();
}

boolean_t
l2arc_vdev_present(vdev_t *vd)
{
	l2arc_dev_t *dev;

	mutex_enter(&l2arc_dev_mtx);
	for (dev = list_head(l2arc_dev_list); dev != NULL;
	    dev = list_next(l2arc_dev_list, dev)) {
		if (dev->l2ad_vdev == vd)
			break;
	}
	mutex_exit(&l2arc_dev_mtx);

	return (dev != NULL);
}

/*
 * Add a vdev for use by the L2ARC.  By this point the spa has already
 * validated the vdev and opened it.
 */
void
l2arc_add_vdev(spa_t *spa, vdev_t *vd)
{
	l2arc_dev_t *adddev;

	ASSERT(!l2arc_vdev_present(vd));

	/*
	 * Create a new l2arc device entry.
	 */
	adddev = kmem_zalloc(sizeof (l2arc_dev_t), KM_SLEEP);
	adddev->l2ad_spa = spa;
	adddev->l2ad_vdev = vd;
	adddev->l2ad_start = VDEV_LABEL_START_SIZE;
	adddev->l2ad_end = VDEV_LABEL_START_SIZE + vdev_get_min_asize(vd);
	adddev->l2ad_hand = adddev->l2ad_start;
	adddev->l2ad_first = B_TRUE;
	adddev->l2ad_writing = B_FALSE;
	list_link_init(&adddev->l2ad_node);

	mutex_init(&adddev->l2ad_mtx, NULL, MUTEX_DEFAULT, NULL);
	/*
	 * This is a list of all ARC buffers that are still valid on the
	 * device.
	 */
	list_create(&adddev->l2ad_buflist, sizeof (arc_buf_hdr_t),
	    offsetof(arc_buf_hdr_t, b_l2hdr.b_l2node));

	vdev_space_update(vd, 0, 0, adddev->l2ad_end - adddev->l2ad_hand);
	refcount_create(&adddev->l2ad_alloc);

	/*
	 * Add device to global list
	 */
	mutex_enter(&l2arc_dev_mtx);
	list_insert_head(l2arc_dev_list, adddev);
	atomic_inc_64(&l2arc_ndev);
	mutex_exit(&l2arc_dev_mtx);
}

/*
 * Remove a vdev from the L2ARC.
 */
void
l2arc_remove_vdev(vdev_t *vd)
{
	l2arc_dev_t *dev, *nextdev, *remdev = NULL;

	/*
	 * Find the device by vdev
	 */
	mutex_enter(&l2arc_dev_mtx);
	for (dev = list_head(l2arc_dev_list); dev; dev = nextdev) {
		nextdev = list_next(l2arc_dev_list, dev);
		if (vd == dev->l2ad_vdev) {
			remdev = dev;
			break;
		}
	}
	ASSERT3P(remdev, !=, NULL);

	/*
	 * Remove device from global list
	 */
	list_remove(l2arc_dev_list, remdev);
	l2arc_dev_last = NULL;		/* may have been invalidated */
	atomic_dec_64(&l2arc_ndev);
	mutex_exit(&l2arc_dev_mtx);

	/*
	 * Clear all buflists and ARC references.  L2ARC device flush.
	 */
	l2arc_evict(remdev, 0, B_TRUE);
	list_destroy(&remdev->l2ad_buflist);
	mutex_destroy(&remdev->l2ad_mtx);
	refcount_destroy(&remdev->l2ad_alloc);
	kmem_free(remdev, sizeof (l2arc_dev_t));
}

void
l2arc_init(void)
{
	l2arc_thread_exit = 0;
	l2arc_ndev = 0;
	l2arc_writes_sent = 0;
	l2arc_writes_done = 0;

	mutex_init(&l2arc_feed_thr_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&l2arc_feed_thr_cv, NULL, CV_DEFAULT, NULL);
	mutex_init(&l2arc_dev_mtx, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&l2arc_free_on_write_mtx, NULL, MUTEX_DEFAULT, NULL);

	l2arc_dev_list = &L2ARC_dev_list;
	l2arc_free_on_write = &L2ARC_free_on_write;
	list_create(l2arc_dev_list, sizeof (l2arc_dev_t),
	    offsetof(l2arc_dev_t, l2ad_node));
	list_create(l2arc_free_on_write, sizeof (l2arc_data_free_t),
	    offsetof(l2arc_data_free_t, l2df_list_node));
}

void
l2arc_fini(void)
{
	/*
	 * This is called from dmu_fini(), which is called from spa_fini();
	 * Because of this, we can assume that all l2arc devices have
	 * already been removed when the pools themselves were removed.
	 */

	l2arc_do_free_on_write();

	mutex_destroy(&l2arc_feed_thr_lock);
	cv_destroy(&l2arc_feed_thr_cv);
	mutex_destroy(&l2arc_dev_mtx);
	mutex_destroy(&l2arc_free_on_write_mtx);

	list_destroy(l2arc_dev_list);
	list_destroy(l2arc_free_on_write);
}

void
l2arc_start(void)
{
	if (!(spa_mode_global & FWRITE))
		return;

	(void) thread_create(NULL, 0, l2arc_feed_thread, NULL, 0, &p0,
	    TS_RUN, defclsyspri);
}

void
l2arc_stop(void)
{
	if (!(spa_mode_global & FWRITE))
		return;

	mutex_enter(&l2arc_feed_thr_lock);
	cv_signal(&l2arc_feed_thr_cv);	/* kick thread out of startup */
	l2arc_thread_exit = 1;
	while (l2arc_thread_exit != 0)
		cv_wait(&l2arc_feed_thr_cv, &l2arc_feed_thr_lock);
	mutex_exit(&l2arc_feed_thr_lock);
}

#if defined(_KERNEL) && defined(HAVE_SPL)
EXPORT_SYMBOL(arc_buf_size);
EXPORT_SYMBOL(arc_write);
EXPORT_SYMBOL(arc_read);
EXPORT_SYMBOL(arc_buf_info);
EXPORT_SYMBOL(arc_getbuf_func);
EXPORT_SYMBOL(arc_add_prune_callback);
EXPORT_SYMBOL(arc_remove_prune_callback);

/* BEGIN CSTYLED */
module_param(zfs_arc_min, ulong, 0644);
MODULE_PARM_DESC(zfs_arc_min, "Min arc size");

module_param(zfs_arc_max, ulong, 0644);
MODULE_PARM_DESC(zfs_arc_max, "Max arc size");

module_param(zfs_arc_meta_limit, ulong, 0644);
MODULE_PARM_DESC(zfs_arc_meta_limit, "Meta limit for arc size");

module_param(zfs_arc_meta_limit_percent, ulong, 0644);
MODULE_PARM_DESC(zfs_arc_meta_limit_percent,
	"Percent of arc size for arc meta limit");

module_param(zfs_arc_meta_min, ulong, 0644);
MODULE_PARM_DESC(zfs_arc_meta_min, "Min arc metadata");

module_param(zfs_arc_meta_prune, int, 0644);
MODULE_PARM_DESC(zfs_arc_meta_prune, "Meta objects to scan for prune");

module_param(zfs_arc_meta_adjust_restarts, int, 0644);
MODULE_PARM_DESC(zfs_arc_meta_adjust_restarts,
	"Limit number of restarts in arc_adjust_meta");

module_param(zfs_arc_meta_strategy, int, 0644);
MODULE_PARM_DESC(zfs_arc_meta_strategy, "Meta reclaim strategy");

module_param(zfs_arc_grow_retry, int, 0644);
MODULE_PARM_DESC(zfs_arc_grow_retry, "Seconds before growing arc size");

module_param(zfs_arc_p_aggressive_disable, int, 0644);
MODULE_PARM_DESC(zfs_arc_p_aggressive_disable, "disable aggressive arc_p grow");

module_param(zfs_arc_p_dampener_disable, int, 0644);
MODULE_PARM_DESC(zfs_arc_p_dampener_disable, "disable arc_p adapt dampener");

module_param(zfs_arc_shrink_shift, int, 0644);
MODULE_PARM_DESC(zfs_arc_shrink_shift, "log2(fraction of arc to reclaim)");

module_param(zfs_arc_p_min_shift, int, 0644);
MODULE_PARM_DESC(zfs_arc_p_min_shift, "arc_c shift to calc min/max arc_p");

module_param(zfs_arc_average_blocksize, int, 0444);
MODULE_PARM_DESC(zfs_arc_average_blocksize, "Target average block size");

module_param(zfs_compressed_arc_enabled, int, 0644);
MODULE_PARM_DESC(zfs_compressed_arc_enabled, "Disable compressed arc buffers");

module_param(zfs_arc_min_prefetch_lifespan, int, 0644);
MODULE_PARM_DESC(zfs_arc_min_prefetch_lifespan, "Min life of prefetch block");

module_param(l2arc_write_max, ulong, 0644);
MODULE_PARM_DESC(l2arc_write_max, "Max write bytes per interval");

module_param(l2arc_write_boost, ulong, 0644);
MODULE_PARM_DESC(l2arc_write_boost, "Extra write bytes during device warmup");

module_param(l2arc_headroom, ulong, 0644);
MODULE_PARM_DESC(l2arc_headroom, "Number of max device writes to precache");

module_param(l2arc_headroom_boost, ulong, 0644);
MODULE_PARM_DESC(l2arc_headroom_boost, "Compressed l2arc_headroom multiplier");

module_param(l2arc_feed_secs, ulong, 0644);
MODULE_PARM_DESC(l2arc_feed_secs, "Seconds between L2ARC writing");

module_param(l2arc_feed_min_ms, ulong, 0644);
MODULE_PARM_DESC(l2arc_feed_min_ms, "Min feed interval in milliseconds");

module_param(l2arc_noprefetch, int, 0644);
MODULE_PARM_DESC(l2arc_noprefetch, "Skip caching prefetched buffers");

module_param(l2arc_feed_again, int, 0644);
MODULE_PARM_DESC(l2arc_feed_again, "Turbo L2ARC warmup");

module_param(l2arc_norw, int, 0644);
MODULE_PARM_DESC(l2arc_norw, "No reads during writes");

module_param(zfs_arc_lotsfree_percent, int, 0644);
MODULE_PARM_DESC(zfs_arc_lotsfree_percent,
	"System free memory I/O throttle in bytes");

module_param(zfs_arc_sys_free, ulong, 0644);
MODULE_PARM_DESC(zfs_arc_sys_free, "System free memory target size in bytes");

module_param(zfs_arc_dnode_limit, ulong, 0644);
MODULE_PARM_DESC(zfs_arc_dnode_limit, "Minimum bytes of dnodes in arc");

module_param(zfs_arc_dnode_limit_percent, ulong, 0644);
MODULE_PARM_DESC(zfs_arc_dnode_limit_percent,
	"Percent of ARC meta buffers for dnodes");

module_param(zfs_arc_dnode_reduce_percent, ulong, 0644);
MODULE_PARM_DESC(zfs_arc_dnode_reduce_percent,
	"Percentage of excess dnodes to try to unpin");
/* END CSTYLED */
#endif