aboutsummaryrefslogtreecommitdiffstats
path: root/uts/common/fs/zfs/dbuf.c
blob: 9c4e0296db2bdab35fca237db669679c66cebae7 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
/*
 * 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.
 */

#include <sys/zfs_context.h>
#include <sys/dmu.h>
#include <sys/dmu_impl.h>
#include <sys/dbuf.h>
#include <sys/dmu_objset.h>
#include <sys/dsl_dataset.h>
#include <sys/dsl_dir.h>
#include <sys/dmu_tx.h>
#include <sys/spa.h>
#include <sys/zio.h>
#include <sys/dmu_zfetch.h>
#include <sys/sa.h>
#include <sys/sa_impl.h>

static void dbuf_destroy(dmu_buf_impl_t *db);
static int dbuf_undirty(dmu_buf_impl_t *db, dmu_tx_t *tx);
static void dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx);

/*
 * Global data structures and functions for the dbuf cache.
 */
static kmem_cache_t *dbuf_cache;

/* ARGSUSED */
static int
dbuf_cons(void *vdb, void *unused, int kmflag)
{
	dmu_buf_impl_t *db = vdb;
	bzero(db, sizeof (dmu_buf_impl_t));

	mutex_init(&db->db_mtx, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&db->db_changed, NULL, CV_DEFAULT, NULL);
	refcount_create(&db->db_holds);
	return (0);
}

/* ARGSUSED */
static void
dbuf_dest(void *vdb, void *unused)
{
	dmu_buf_impl_t *db = vdb;
	mutex_destroy(&db->db_mtx);
	cv_destroy(&db->db_changed);
	refcount_destroy(&db->db_holds);
}

/*
 * dbuf hash table routines
 */
static dbuf_hash_table_t dbuf_hash_table;

static uint64_t dbuf_hash_count;

static uint64_t
dbuf_hash(void *os, uint64_t obj, uint8_t lvl, uint64_t blkid)
{
	uintptr_t osv = (uintptr_t)os;
	uint64_t crc = -1ULL;

	ASSERT(zfs_crc64_table[128] == ZFS_CRC64_POLY);
	crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (lvl)) & 0xFF];
	crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (osv >> 6)) & 0xFF];
	crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (obj >> 0)) & 0xFF];
	crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (obj >> 8)) & 0xFF];
	crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (blkid >> 0)) & 0xFF];
	crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (blkid >> 8)) & 0xFF];

	crc ^= (osv>>14) ^ (obj>>16) ^ (blkid>>16);

	return (crc);
}

#define	DBUF_HASH(os, obj, level, blkid) dbuf_hash(os, obj, level, blkid);

#define	DBUF_EQUAL(dbuf, os, obj, level, blkid)		\
	((dbuf)->db.db_object == (obj) &&		\
	(dbuf)->db_objset == (os) &&			\
	(dbuf)->db_level == (level) &&			\
	(dbuf)->db_blkid == (blkid))

dmu_buf_impl_t *
dbuf_find(dnode_t *dn, uint8_t level, uint64_t blkid)
{
	dbuf_hash_table_t *h = &dbuf_hash_table;
	objset_t *os = dn->dn_objset;
	uint64_t obj = dn->dn_object;
	uint64_t hv = DBUF_HASH(os, obj, level, blkid);
	uint64_t idx = hv & h->hash_table_mask;
	dmu_buf_impl_t *db;

	mutex_enter(DBUF_HASH_MUTEX(h, idx));
	for (db = h->hash_table[idx]; db != NULL; db = db->db_hash_next) {
		if (DBUF_EQUAL(db, os, obj, level, blkid)) {
			mutex_enter(&db->db_mtx);
			if (db->db_state != DB_EVICTING) {
				mutex_exit(DBUF_HASH_MUTEX(h, idx));
				return (db);
			}
			mutex_exit(&db->db_mtx);
		}
	}
	mutex_exit(DBUF_HASH_MUTEX(h, idx));
	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.
 */
static dmu_buf_impl_t *
dbuf_hash_insert(dmu_buf_impl_t *db)
{
	dbuf_hash_table_t *h = &dbuf_hash_table;
	objset_t *os = db->db_objset;
	uint64_t obj = db->db.db_object;
	int level = db->db_level;
	uint64_t blkid = db->db_blkid;
	uint64_t hv = DBUF_HASH(os, obj, level, blkid);
	uint64_t idx = hv & h->hash_table_mask;
	dmu_buf_impl_t *dbf;

	mutex_enter(DBUF_HASH_MUTEX(h, idx));
	for (dbf = h->hash_table[idx]; dbf != NULL; dbf = dbf->db_hash_next) {
		if (DBUF_EQUAL(dbf, os, obj, level, blkid)) {
			mutex_enter(&dbf->db_mtx);
			if (dbf->db_state != DB_EVICTING) {
				mutex_exit(DBUF_HASH_MUTEX(h, idx));
				return (dbf);
			}
			mutex_exit(&dbf->db_mtx);
		}
	}

	mutex_enter(&db->db_mtx);
	db->db_hash_next = h->hash_table[idx];
	h->hash_table[idx] = db;
	mutex_exit(DBUF_HASH_MUTEX(h, idx));
	atomic_add_64(&dbuf_hash_count, 1);

	return (NULL);
}

/*
 * Remove an entry from the hash table.  This operation will
 * fail if there are any existing holds on the db.
 */
static void
dbuf_hash_remove(dmu_buf_impl_t *db)
{
	dbuf_hash_table_t *h = &dbuf_hash_table;
	uint64_t hv = DBUF_HASH(db->db_objset, db->db.db_object,
	    db->db_level, db->db_blkid);
	uint64_t idx = hv & h->hash_table_mask;
	dmu_buf_impl_t *dbf, **dbp;

	/*
	 * We musn't hold db_mtx to maintin lock ordering:
	 * DBUF_HASH_MUTEX > db_mtx.
	 */
	ASSERT(refcount_is_zero(&db->db_holds));
	ASSERT(db->db_state == DB_EVICTING);
	ASSERT(!MUTEX_HELD(&db->db_mtx));

	mutex_enter(DBUF_HASH_MUTEX(h, idx));
	dbp = &h->hash_table[idx];
	while ((dbf = *dbp) != db) {
		dbp = &dbf->db_hash_next;
		ASSERT(dbf != NULL);
	}
	*dbp = db->db_hash_next;
	db->db_hash_next = NULL;
	mutex_exit(DBUF_HASH_MUTEX(h, idx));
	atomic_add_64(&dbuf_hash_count, -1);
}

static arc_evict_func_t dbuf_do_evict;

static void
dbuf_evict_user(dmu_buf_impl_t *db)
{
	ASSERT(MUTEX_HELD(&db->db_mtx));

	if (db->db_level != 0 || db->db_evict_func == NULL)
		return;

	if (db->db_user_data_ptr_ptr)
		*db->db_user_data_ptr_ptr = db->db.db_data;
	db->db_evict_func(&db->db, db->db_user_ptr);
	db->db_user_ptr = NULL;
	db->db_user_data_ptr_ptr = NULL;
	db->db_evict_func = NULL;
}

boolean_t
dbuf_is_metadata(dmu_buf_impl_t *db)
{
	if (db->db_level > 0) {
		return (B_TRUE);
	} else {
		boolean_t is_metadata;

		DB_DNODE_ENTER(db);
		is_metadata = dmu_ot[DB_DNODE(db)->dn_type].ot_metadata;
		DB_DNODE_EXIT(db);

		return (is_metadata);
	}
}

void
dbuf_evict(dmu_buf_impl_t *db)
{
	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(db->db_buf == NULL);
	ASSERT(db->db_data_pending == NULL);

	dbuf_clear(db);
	dbuf_destroy(db);
}

void
dbuf_init(void)
{
	uint64_t hsize = 1ULL << 16;
	dbuf_hash_table_t *h = &dbuf_hash_table;
	int i;

	/*
	 * The hash table is big enough to fill all of physical memory
	 * with an average 4K block size.  The table will take up
	 * totalmem*sizeof(void*)/4K (i.e. 2MB/GB with 8-byte pointers).
	 */
	while (hsize * 4096 < physmem * PAGESIZE)
		hsize <<= 1;

retry:
	h->hash_table_mask = hsize - 1;
	h->hash_table = kmem_zalloc(hsize * sizeof (void *), KM_NOSLEEP);
	if (h->hash_table == NULL) {
		/* XXX - we should really return an error instead of assert */
		ASSERT(hsize > (1ULL << 10));
		hsize >>= 1;
		goto retry;
	}

	dbuf_cache = kmem_cache_create("dmu_buf_impl_t",
	    sizeof (dmu_buf_impl_t),
	    0, dbuf_cons, dbuf_dest, NULL, NULL, NULL, 0);

	for (i = 0; i < DBUF_MUTEXES; i++)
		mutex_init(&h->hash_mutexes[i], NULL, MUTEX_DEFAULT, NULL);
}

void
dbuf_fini(void)
{
	dbuf_hash_table_t *h = &dbuf_hash_table;
	int i;

	for (i = 0; i < DBUF_MUTEXES; i++)
		mutex_destroy(&h->hash_mutexes[i]);
	kmem_free(h->hash_table, (h->hash_table_mask + 1) * sizeof (void *));
	kmem_cache_destroy(dbuf_cache);
}

/*
 * Other stuff.
 */

#ifdef ZFS_DEBUG
static void
dbuf_verify(dmu_buf_impl_t *db)
{
	dnode_t *dn;
	dbuf_dirty_record_t *dr;

	ASSERT(MUTEX_HELD(&db->db_mtx));

	if (!(zfs_flags & ZFS_DEBUG_DBUF_VERIFY))
		return;

	ASSERT(db->db_objset != NULL);
	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	if (dn == NULL) {
		ASSERT(db->db_parent == NULL);
		ASSERT(db->db_blkptr == NULL);
	} else {
		ASSERT3U(db->db.db_object, ==, dn->dn_object);
		ASSERT3P(db->db_objset, ==, dn->dn_objset);
		ASSERT3U(db->db_level, <, dn->dn_nlevels);
		ASSERT(db->db_blkid == DMU_BONUS_BLKID ||
		    db->db_blkid == DMU_SPILL_BLKID ||
		    !list_is_empty(&dn->dn_dbufs));
	}
	if (db->db_blkid == DMU_BONUS_BLKID) {
		ASSERT(dn != NULL);
		ASSERT3U(db->db.db_size, >=, dn->dn_bonuslen);
		ASSERT3U(db->db.db_offset, ==, DMU_BONUS_BLKID);
	} else if (db->db_blkid == DMU_SPILL_BLKID) {
		ASSERT(dn != NULL);
		ASSERT3U(db->db.db_size, >=, dn->dn_bonuslen);
		ASSERT3U(db->db.db_offset, ==, 0);
	} else {
		ASSERT3U(db->db.db_offset, ==, db->db_blkid * db->db.db_size);
	}

	for (dr = db->db_data_pending; dr != NULL; dr = dr->dr_next)
		ASSERT(dr->dr_dbuf == db);

	for (dr = db->db_last_dirty; dr != NULL; dr = dr->dr_next)
		ASSERT(dr->dr_dbuf == db);

	/*
	 * We can't assert that db_size matches dn_datablksz because it
	 * can be momentarily different when another thread is doing
	 * dnode_set_blksz().
	 */
	if (db->db_level == 0 && db->db.db_object == DMU_META_DNODE_OBJECT) {
		dr = db->db_data_pending;
		/*
		 * It should only be modified in syncing context, so
		 * make sure we only have one copy of the data.
		 */
		ASSERT(dr == NULL || dr->dt.dl.dr_data == db->db_buf);
	}

	/* verify db->db_blkptr */
	if (db->db_blkptr) {
		if (db->db_parent == dn->dn_dbuf) {
			/* db is pointed to by the dnode */
			/* ASSERT3U(db->db_blkid, <, dn->dn_nblkptr); */
			if (DMU_OBJECT_IS_SPECIAL(db->db.db_object))
				ASSERT(db->db_parent == NULL);
			else
				ASSERT(db->db_parent != NULL);
			if (db->db_blkid != DMU_SPILL_BLKID)
				ASSERT3P(db->db_blkptr, ==,
				    &dn->dn_phys->dn_blkptr[db->db_blkid]);
		} else {
			/* db is pointed to by an indirect block */
			int epb = db->db_parent->db.db_size >> SPA_BLKPTRSHIFT;
			ASSERT3U(db->db_parent->db_level, ==, db->db_level+1);
			ASSERT3U(db->db_parent->db.db_object, ==,
			    db->db.db_object);
			/*
			 * dnode_grow_indblksz() can make this fail if we don't
			 * have the struct_rwlock.  XXX indblksz no longer
			 * grows.  safe to do this now?
			 */
			if (RW_WRITE_HELD(&dn->dn_struct_rwlock)) {
				ASSERT3P(db->db_blkptr, ==,
				    ((blkptr_t *)db->db_parent->db.db_data +
				    db->db_blkid % epb));
			}
		}
	}
	if ((db->db_blkptr == NULL || BP_IS_HOLE(db->db_blkptr)) &&
	    (db->db_buf == NULL || db->db_buf->b_data) &&
	    db->db.db_data && db->db_blkid != DMU_BONUS_BLKID &&
	    db->db_state != DB_FILL && !dn->dn_free_txg) {
		/*
		 * If the blkptr isn't set but they have nonzero data,
		 * it had better be dirty, otherwise we'll lose that
		 * data when we evict this buffer.
		 */
		if (db->db_dirtycnt == 0) {
			uint64_t *buf = db->db.db_data;
			int i;

			for (i = 0; i < db->db.db_size >> 3; i++) {
				ASSERT(buf[i] == 0);
			}
		}
	}
	DB_DNODE_EXIT(db);
}
#endif

static void
dbuf_update_data(dmu_buf_impl_t *db)
{
	ASSERT(MUTEX_HELD(&db->db_mtx));
	if (db->db_level == 0 && db->db_user_data_ptr_ptr) {
		ASSERT(!refcount_is_zero(&db->db_holds));
		*db->db_user_data_ptr_ptr = db->db.db_data;
	}
}

static void
dbuf_set_data(dmu_buf_impl_t *db, arc_buf_t *buf)
{
	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(db->db_buf == NULL || !arc_has_callback(db->db_buf));
	db->db_buf = buf;
	if (buf != NULL) {
		ASSERT(buf->b_data != NULL);
		db->db.db_data = buf->b_data;
		if (!arc_released(buf))
			arc_set_callback(buf, dbuf_do_evict, db);
		dbuf_update_data(db);
	} else {
		dbuf_evict_user(db);
		db->db.db_data = NULL;
		if (db->db_state != DB_NOFILL)
			db->db_state = DB_UNCACHED;
	}
}

/*
 * Loan out an arc_buf for read.  Return the loaned arc_buf.
 */
arc_buf_t *
dbuf_loan_arcbuf(dmu_buf_impl_t *db)
{
	arc_buf_t *abuf;

	mutex_enter(&db->db_mtx);
	if (arc_released(db->db_buf) || refcount_count(&db->db_holds) > 1) {
		int blksz = db->db.db_size;
		spa_t *spa;

		mutex_exit(&db->db_mtx);
		DB_GET_SPA(&spa, db);
		abuf = arc_loan_buf(spa, blksz);
		bcopy(db->db.db_data, abuf->b_data, blksz);
	} else {
		abuf = db->db_buf;
		arc_loan_inuse_buf(abuf, db);
		dbuf_set_data(db, NULL);
		mutex_exit(&db->db_mtx);
	}
	return (abuf);
}

uint64_t
dbuf_whichblock(dnode_t *dn, uint64_t offset)
{
	if (dn->dn_datablkshift) {
		return (offset >> dn->dn_datablkshift);
	} else {
		ASSERT3U(offset, <, dn->dn_datablksz);
		return (0);
	}
}

static void
dbuf_read_done(zio_t *zio, arc_buf_t *buf, void *vdb)
{
	dmu_buf_impl_t *db = vdb;

	mutex_enter(&db->db_mtx);
	ASSERT3U(db->db_state, ==, DB_READ);
	/*
	 * All reads are synchronous, so we must have a hold on the dbuf
	 */
	ASSERT(refcount_count(&db->db_holds) > 0);
	ASSERT(db->db_buf == NULL);
	ASSERT(db->db.db_data == NULL);
	if (db->db_level == 0 && db->db_freed_in_flight) {
		/* we were freed in flight; disregard any error */
		arc_release(buf, db);
		bzero(buf->b_data, db->db.db_size);
		arc_buf_freeze(buf);
		db->db_freed_in_flight = FALSE;
		dbuf_set_data(db, buf);
		db->db_state = DB_CACHED;
	} else if (zio == NULL || zio->io_error == 0) {
		dbuf_set_data(db, buf);
		db->db_state = DB_CACHED;
	} else {
		ASSERT(db->db_blkid != DMU_BONUS_BLKID);
		ASSERT3P(db->db_buf, ==, NULL);
		VERIFY(arc_buf_remove_ref(buf, db) == 1);
		db->db_state = DB_UNCACHED;
	}
	cv_broadcast(&db->db_changed);
	dbuf_rele_and_unlock(db, NULL);
}

static void
dbuf_read_impl(dmu_buf_impl_t *db, zio_t *zio, uint32_t *flags)
{
	dnode_t *dn;
	spa_t *spa;
	zbookmark_t zb;
	uint32_t aflags = ARC_NOWAIT;
	arc_buf_t *pbuf;

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	ASSERT(!refcount_is_zero(&db->db_holds));
	/* We need the struct_rwlock to prevent db_blkptr from changing. */
	ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));
	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(db->db_state == DB_UNCACHED);
	ASSERT(db->db_buf == NULL);

	if (db->db_blkid == DMU_BONUS_BLKID) {
		int bonuslen = MIN(dn->dn_bonuslen, dn->dn_phys->dn_bonuslen);

		ASSERT3U(bonuslen, <=, db->db.db_size);
		db->db.db_data = zio_buf_alloc(DN_MAX_BONUSLEN);
		arc_space_consume(DN_MAX_BONUSLEN, ARC_SPACE_OTHER);
		if (bonuslen < DN_MAX_BONUSLEN)
			bzero(db->db.db_data, DN_MAX_BONUSLEN);
		if (bonuslen)
			bcopy(DN_BONUS(dn->dn_phys), db->db.db_data, bonuslen);
		DB_DNODE_EXIT(db);
		dbuf_update_data(db);
		db->db_state = DB_CACHED;
		mutex_exit(&db->db_mtx);
		return;
	}

	/*
	 * Recheck BP_IS_HOLE() after dnode_block_freed() in case dnode_sync()
	 * processes the delete record and clears the bp while we are waiting
	 * for the dn_mtx (resulting in a "no" from block_freed).
	 */
	if (db->db_blkptr == NULL || BP_IS_HOLE(db->db_blkptr) ||
	    (db->db_level == 0 && (dnode_block_freed(dn, db->db_blkid) ||
	    BP_IS_HOLE(db->db_blkptr)))) {
		arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);

		dbuf_set_data(db, arc_buf_alloc(dn->dn_objset->os_spa,
		    db->db.db_size, db, type));
		DB_DNODE_EXIT(db);
		bzero(db->db.db_data, db->db.db_size);
		db->db_state = DB_CACHED;
		*flags |= DB_RF_CACHED;
		mutex_exit(&db->db_mtx);
		return;
	}

	spa = dn->dn_objset->os_spa;
	DB_DNODE_EXIT(db);

	db->db_state = DB_READ;
	mutex_exit(&db->db_mtx);

	if (DBUF_IS_L2CACHEABLE(db))
		aflags |= ARC_L2CACHE;

	SET_BOOKMARK(&zb, db->db_objset->os_dsl_dataset ?
	    db->db_objset->os_dsl_dataset->ds_object : DMU_META_OBJSET,
	    db->db.db_object, db->db_level, db->db_blkid);

	dbuf_add_ref(db, NULL);
	/* ZIO_FLAG_CANFAIL callers have to check the parent zio's error */

	if (db->db_parent)
		pbuf = db->db_parent->db_buf;
	else
		pbuf = db->db_objset->os_phys_buf;

	(void) dsl_read(zio, spa, db->db_blkptr, pbuf,
	    dbuf_read_done, db, ZIO_PRIORITY_SYNC_READ,
	    (*flags & DB_RF_CANFAIL) ? ZIO_FLAG_CANFAIL : ZIO_FLAG_MUSTSUCCEED,
	    &aflags, &zb);
	if (aflags & ARC_CACHED)
		*flags |= DB_RF_CACHED;
}

int
dbuf_read(dmu_buf_impl_t *db, zio_t *zio, uint32_t flags)
{
	int err = 0;
	int havepzio = (zio != NULL);
	int prefetch;
	dnode_t *dn;

	/*
	 * We don't have to hold the mutex to check db_state because it
	 * can't be freed while we have a hold on the buffer.
	 */
	ASSERT(!refcount_is_zero(&db->db_holds));

	if (db->db_state == DB_NOFILL)
		return (EIO);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	if ((flags & DB_RF_HAVESTRUCT) == 0)
		rw_enter(&dn->dn_struct_rwlock, RW_READER);

	prefetch = db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID &&
	    (flags & DB_RF_NOPREFETCH) == 0 && dn != NULL &&
	    DBUF_IS_CACHEABLE(db);

	mutex_enter(&db->db_mtx);
	if (db->db_state == DB_CACHED) {
		mutex_exit(&db->db_mtx);
		if (prefetch)
			dmu_zfetch(&dn->dn_zfetch, db->db.db_offset,
			    db->db.db_size, TRUE);
		if ((flags & DB_RF_HAVESTRUCT) == 0)
			rw_exit(&dn->dn_struct_rwlock);
		DB_DNODE_EXIT(db);
	} else if (db->db_state == DB_UNCACHED) {
		spa_t *spa = dn->dn_objset->os_spa;

		if (zio == NULL)
			zio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL);
		dbuf_read_impl(db, zio, &flags);

		/* dbuf_read_impl has dropped db_mtx for us */

		if (prefetch)
			dmu_zfetch(&dn->dn_zfetch, db->db.db_offset,
			    db->db.db_size, flags & DB_RF_CACHED);

		if ((flags & DB_RF_HAVESTRUCT) == 0)
			rw_exit(&dn->dn_struct_rwlock);
		DB_DNODE_EXIT(db);

		if (!havepzio)
			err = zio_wait(zio);
	} else {
		mutex_exit(&db->db_mtx);
		if (prefetch)
			dmu_zfetch(&dn->dn_zfetch, db->db.db_offset,
			    db->db.db_size, TRUE);
		if ((flags & DB_RF_HAVESTRUCT) == 0)
			rw_exit(&dn->dn_struct_rwlock);
		DB_DNODE_EXIT(db);

		mutex_enter(&db->db_mtx);
		if ((flags & DB_RF_NEVERWAIT) == 0) {
			while (db->db_state == DB_READ ||
			    db->db_state == DB_FILL) {
				ASSERT(db->db_state == DB_READ ||
				    (flags & DB_RF_HAVESTRUCT) == 0);
				cv_wait(&db->db_changed, &db->db_mtx);
			}
			if (db->db_state == DB_UNCACHED)
				err = EIO;
		}
		mutex_exit(&db->db_mtx);
	}

	ASSERT(err || havepzio || db->db_state == DB_CACHED);
	return (err);
}

static void
dbuf_noread(dmu_buf_impl_t *db)
{
	ASSERT(!refcount_is_zero(&db->db_holds));
	ASSERT(db->db_blkid != DMU_BONUS_BLKID);
	mutex_enter(&db->db_mtx);
	while (db->db_state == DB_READ || db->db_state == DB_FILL)
		cv_wait(&db->db_changed, &db->db_mtx);
	if (db->db_state == DB_UNCACHED) {
		arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
		spa_t *spa;

		ASSERT(db->db_buf == NULL);
		ASSERT(db->db.db_data == NULL);
		DB_GET_SPA(&spa, db);
		dbuf_set_data(db, arc_buf_alloc(spa, db->db.db_size, db, type));
		db->db_state = DB_FILL;
	} else if (db->db_state == DB_NOFILL) {
		dbuf_set_data(db, NULL);
	} else {
		ASSERT3U(db->db_state, ==, DB_CACHED);
	}
	mutex_exit(&db->db_mtx);
}

/*
 * This is our just-in-time copy function.  It makes a copy of
 * buffers, that have been modified in a previous transaction
 * group, before we modify them in the current active group.
 *
 * This function is used in two places: when we are dirtying a
 * buffer for the first time in a txg, and when we are freeing
 * a range in a dnode that includes this buffer.
 *
 * Note that when we are called from dbuf_free_range() we do
 * not put a hold on the buffer, we just traverse the active
 * dbuf list for the dnode.
 */
static void
dbuf_fix_old_data(dmu_buf_impl_t *db, uint64_t txg)
{
	dbuf_dirty_record_t *dr = db->db_last_dirty;

	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(db->db.db_data != NULL);
	ASSERT(db->db_level == 0);
	ASSERT(db->db.db_object != DMU_META_DNODE_OBJECT);

	if (dr == NULL ||
	    (dr->dt.dl.dr_data !=
	    ((db->db_blkid  == DMU_BONUS_BLKID) ? db->db.db_data : db->db_buf)))
		return;

	/*
	 * If the last dirty record for this dbuf has not yet synced
	 * and its referencing the dbuf data, either:
	 *	reset the reference to point to a new copy,
	 * or (if there a no active holders)
	 *	just null out the current db_data pointer.
	 */
	ASSERT(dr->dr_txg >= txg - 2);
	if (db->db_blkid == DMU_BONUS_BLKID) {
		/* Note that the data bufs here are zio_bufs */
		dr->dt.dl.dr_data = zio_buf_alloc(DN_MAX_BONUSLEN);
		arc_space_consume(DN_MAX_BONUSLEN, ARC_SPACE_OTHER);
		bcopy(db->db.db_data, dr->dt.dl.dr_data, DN_MAX_BONUSLEN);
	} else if (refcount_count(&db->db_holds) > db->db_dirtycnt) {
		int size = db->db.db_size;
		arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
		spa_t *spa;

		DB_GET_SPA(&spa, db);
		dr->dt.dl.dr_data = arc_buf_alloc(spa, size, db, type);
		bcopy(db->db.db_data, dr->dt.dl.dr_data->b_data, size);
	} else {
		dbuf_set_data(db, NULL);
	}
}

void
dbuf_unoverride(dbuf_dirty_record_t *dr)
{
	dmu_buf_impl_t *db = dr->dr_dbuf;
	blkptr_t *bp = &dr->dt.dl.dr_overridden_by;
	uint64_t txg = dr->dr_txg;

	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(dr->dt.dl.dr_override_state != DR_IN_DMU_SYNC);
	ASSERT(db->db_level == 0);

	if (db->db_blkid == DMU_BONUS_BLKID ||
	    dr->dt.dl.dr_override_state == DR_NOT_OVERRIDDEN)
		return;

	ASSERT(db->db_data_pending != dr);

	/* free this block */
	if (!BP_IS_HOLE(bp)) {
		spa_t *spa;

		DB_GET_SPA(&spa, db);
		zio_free(spa, txg, bp);
	}
	dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN;
	/*
	 * Release the already-written buffer, so we leave it in
	 * a consistent dirty state.  Note that all callers are
	 * modifying the buffer, so they will immediately do
	 * another (redundant) arc_release().  Therefore, leave
	 * the buf thawed to save the effort of freezing &
	 * immediately re-thawing it.
	 */
	arc_release(dr->dt.dl.dr_data, db);
}

/*
 * Evict (if its unreferenced) or clear (if its referenced) any level-0
 * data blocks in the free range, so that any future readers will find
 * empty blocks.  Also, if we happen accross any level-1 dbufs in the
 * range that have not already been marked dirty, mark them dirty so
 * they stay in memory.
 */
void
dbuf_free_range(dnode_t *dn, uint64_t start, uint64_t end, dmu_tx_t *tx)
{
	dmu_buf_impl_t *db, *db_next;
	uint64_t txg = tx->tx_txg;
	int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT;
	uint64_t first_l1 = start >> epbs;
	uint64_t last_l1 = end >> epbs;

	if (end > dn->dn_maxblkid && (end != DMU_SPILL_BLKID)) {
		end = dn->dn_maxblkid;
		last_l1 = end >> epbs;
	}
	dprintf_dnode(dn, "start=%llu end=%llu\n", start, end);
	mutex_enter(&dn->dn_dbufs_mtx);
	for (db = list_head(&dn->dn_dbufs); db; db = db_next) {
		db_next = list_next(&dn->dn_dbufs, db);
		ASSERT(db->db_blkid != DMU_BONUS_BLKID);

		if (db->db_level == 1 &&
		    db->db_blkid >= first_l1 && db->db_blkid <= last_l1) {
			mutex_enter(&db->db_mtx);
			if (db->db_last_dirty &&
			    db->db_last_dirty->dr_txg < txg) {
				dbuf_add_ref(db, FTAG);
				mutex_exit(&db->db_mtx);
				dbuf_will_dirty(db, tx);
				dbuf_rele(db, FTAG);
			} else {
				mutex_exit(&db->db_mtx);
			}
		}

		if (db->db_level != 0)
			continue;
		dprintf_dbuf(db, "found buf %s\n", "");
		if (db->db_blkid < start || db->db_blkid > end)
			continue;

		/* found a level 0 buffer in the range */
		if (dbuf_undirty(db, tx))
			continue;

		mutex_enter(&db->db_mtx);
		if (db->db_state == DB_UNCACHED ||
		    db->db_state == DB_NOFILL ||
		    db->db_state == DB_EVICTING) {
			ASSERT(db->db.db_data == NULL);
			mutex_exit(&db->db_mtx);
			continue;
		}
		if (db->db_state == DB_READ || db->db_state == DB_FILL) {
			/* will be handled in dbuf_read_done or dbuf_rele */
			db->db_freed_in_flight = TRUE;
			mutex_exit(&db->db_mtx);
			continue;
		}
		if (refcount_count(&db->db_holds) == 0) {
			ASSERT(db->db_buf);
			dbuf_clear(db);
			continue;
		}
		/* The dbuf is referenced */

		if (db->db_last_dirty != NULL) {
			dbuf_dirty_record_t *dr = db->db_last_dirty;

			if (dr->dr_txg == txg) {
				/*
				 * This buffer is "in-use", re-adjust the file
				 * size to reflect that this buffer may
				 * contain new data when we sync.
				 */
				if (db->db_blkid != DMU_SPILL_BLKID &&
				    db->db_blkid > dn->dn_maxblkid)
					dn->dn_maxblkid = db->db_blkid;
				dbuf_unoverride(dr);
			} else {
				/*
				 * This dbuf is not dirty in the open context.
				 * Either uncache it (if its not referenced in
				 * the open context) or reset its contents to
				 * empty.
				 */
				dbuf_fix_old_data(db, txg);
			}
		}
		/* clear the contents if its cached */
		if (db->db_state == DB_CACHED) {
			ASSERT(db->db.db_data != NULL);
			arc_release(db->db_buf, db);
			bzero(db->db.db_data, db->db.db_size);
			arc_buf_freeze(db->db_buf);
		}

		mutex_exit(&db->db_mtx);
	}
	mutex_exit(&dn->dn_dbufs_mtx);
}

static int
dbuf_block_freeable(dmu_buf_impl_t *db)
{
	dsl_dataset_t *ds = db->db_objset->os_dsl_dataset;
	uint64_t birth_txg = 0;

	/*
	 * We don't need any locking to protect db_blkptr:
	 * If it's syncing, then db_last_dirty will be set
	 * so we'll ignore db_blkptr.
	 */
	ASSERT(MUTEX_HELD(&db->db_mtx));
	if (db->db_last_dirty)
		birth_txg = db->db_last_dirty->dr_txg;
	else if (db->db_blkptr)
		birth_txg = db->db_blkptr->blk_birth;

	/*
	 * If we don't exist or are in a snapshot, we can't be freed.
	 * Don't pass the bp to dsl_dataset_block_freeable() since we
	 * are holding the db_mtx lock and might deadlock if we are
	 * prefetching a dedup-ed block.
	 */
	if (birth_txg)
		return (ds == NULL ||
		    dsl_dataset_block_freeable(ds, NULL, birth_txg));
	else
		return (FALSE);
}

void
dbuf_new_size(dmu_buf_impl_t *db, int size, dmu_tx_t *tx)
{
	arc_buf_t *buf, *obuf;
	int osize = db->db.db_size;
	arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
	dnode_t *dn;

	ASSERT(db->db_blkid != DMU_BONUS_BLKID);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);

	/* XXX does *this* func really need the lock? */
	ASSERT(RW_WRITE_HELD(&dn->dn_struct_rwlock));

	/*
	 * This call to dbuf_will_dirty() with the dn_struct_rwlock held
	 * is OK, because there can be no other references to the db
	 * when we are changing its size, so no concurrent DB_FILL can
	 * be happening.
	 */
	/*
	 * XXX we should be doing a dbuf_read, checking the return
	 * value and returning that up to our callers
	 */
	dbuf_will_dirty(db, tx);

	/* create the data buffer for the new block */
	buf = arc_buf_alloc(dn->dn_objset->os_spa, size, db, type);

	/* copy old block data to the new block */
	obuf = db->db_buf;
	bcopy(obuf->b_data, buf->b_data, MIN(osize, size));
	/* zero the remainder */
	if (size > osize)
		bzero((uint8_t *)buf->b_data + osize, size - osize);

	mutex_enter(&db->db_mtx);
	dbuf_set_data(db, buf);
	VERIFY(arc_buf_remove_ref(obuf, db) == 1);
	db->db.db_size = size;

	if (db->db_level == 0) {
		ASSERT3U(db->db_last_dirty->dr_txg, ==, tx->tx_txg);
		db->db_last_dirty->dt.dl.dr_data = buf;
	}
	mutex_exit(&db->db_mtx);

	dnode_willuse_space(dn, size-osize, tx);
	DB_DNODE_EXIT(db);
}

void
dbuf_release_bp(dmu_buf_impl_t *db)
{
	objset_t *os;
	zbookmark_t zb;

	DB_GET_OBJSET(&os, db);
	ASSERT(dsl_pool_sync_context(dmu_objset_pool(os)));
	ASSERT(arc_released(os->os_phys_buf) ||
	    list_link_active(&os->os_dsl_dataset->ds_synced_link));
	ASSERT(db->db_parent == NULL || arc_released(db->db_parent->db_buf));

	zb.zb_objset = os->os_dsl_dataset ?
	    os->os_dsl_dataset->ds_object : 0;
	zb.zb_object = db->db.db_object;
	zb.zb_level = db->db_level;
	zb.zb_blkid = db->db_blkid;
	(void) arc_release_bp(db->db_buf, db,
	    db->db_blkptr, os->os_spa, &zb);
}

dbuf_dirty_record_t *
dbuf_dirty(dmu_buf_impl_t *db, dmu_tx_t *tx)
{
	dnode_t *dn;
	objset_t *os;
	dbuf_dirty_record_t **drp, *dr;
	int drop_struct_lock = FALSE;
	boolean_t do_free_accounting = B_FALSE;
	int txgoff = tx->tx_txg & TXG_MASK;

	ASSERT(tx->tx_txg != 0);
	ASSERT(!refcount_is_zero(&db->db_holds));
	DMU_TX_DIRTY_BUF(tx, db);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	/*
	 * Shouldn't dirty a regular buffer in syncing context.  Private
	 * objects may be dirtied in syncing context, but only if they
	 * were already pre-dirtied in open context.
	 */
	ASSERT(!dmu_tx_is_syncing(tx) ||
	    BP_IS_HOLE(dn->dn_objset->os_rootbp) ||
	    DMU_OBJECT_IS_SPECIAL(dn->dn_object) ||
	    dn->dn_objset->os_dsl_dataset == NULL);
	/*
	 * We make this assert for private objects as well, but after we
	 * check if we're already dirty.  They are allowed to re-dirty
	 * in syncing context.
	 */
	ASSERT(dn->dn_object == DMU_META_DNODE_OBJECT ||
	    dn->dn_dirtyctx == DN_UNDIRTIED || dn->dn_dirtyctx ==
	    (dmu_tx_is_syncing(tx) ? DN_DIRTY_SYNC : DN_DIRTY_OPEN));

	mutex_enter(&db->db_mtx);
	/*
	 * XXX make this true for indirects too?  The problem is that
	 * transactions created with dmu_tx_create_assigned() from
	 * syncing context don't bother holding ahead.
	 */
	ASSERT(db->db_level != 0 ||
	    db->db_state == DB_CACHED || db->db_state == DB_FILL ||
	    db->db_state == DB_NOFILL);

	mutex_enter(&dn->dn_mtx);
	/*
	 * Don't set dirtyctx to SYNC if we're just modifying this as we
	 * initialize the objset.
	 */
	if (dn->dn_dirtyctx == DN_UNDIRTIED &&
	    !BP_IS_HOLE(dn->dn_objset->os_rootbp)) {
		dn->dn_dirtyctx =
		    (dmu_tx_is_syncing(tx) ? DN_DIRTY_SYNC : DN_DIRTY_OPEN);
		ASSERT(dn->dn_dirtyctx_firstset == NULL);
		dn->dn_dirtyctx_firstset = kmem_alloc(1, KM_SLEEP);
	}
	mutex_exit(&dn->dn_mtx);

	if (db->db_blkid == DMU_SPILL_BLKID)
		dn->dn_have_spill = B_TRUE;

	/*
	 * If this buffer is already dirty, we're done.
	 */
	drp = &db->db_last_dirty;
	ASSERT(*drp == NULL || (*drp)->dr_txg <= tx->tx_txg ||
	    db->db.db_object == DMU_META_DNODE_OBJECT);
	while ((dr = *drp) != NULL && dr->dr_txg > tx->tx_txg)
		drp = &dr->dr_next;
	if (dr && dr->dr_txg == tx->tx_txg) {
		DB_DNODE_EXIT(db);

		if (db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID) {
			/*
			 * If this buffer has already been written out,
			 * we now need to reset its state.
			 */
			dbuf_unoverride(dr);
			if (db->db.db_object != DMU_META_DNODE_OBJECT &&
			    db->db_state != DB_NOFILL)
				arc_buf_thaw(db->db_buf);
		}
		mutex_exit(&db->db_mtx);
		return (dr);
	}

	/*
	 * Only valid if not already dirty.
	 */
	ASSERT(dn->dn_object == 0 ||
	    dn->dn_dirtyctx == DN_UNDIRTIED || dn->dn_dirtyctx ==
	    (dmu_tx_is_syncing(tx) ? DN_DIRTY_SYNC : DN_DIRTY_OPEN));

	ASSERT3U(dn->dn_nlevels, >, db->db_level);
	ASSERT((dn->dn_phys->dn_nlevels == 0 && db->db_level == 0) ||
	    dn->dn_phys->dn_nlevels > db->db_level ||
	    dn->dn_next_nlevels[txgoff] > db->db_level ||
	    dn->dn_next_nlevels[(tx->tx_txg-1) & TXG_MASK] > db->db_level ||
	    dn->dn_next_nlevels[(tx->tx_txg-2) & TXG_MASK] > db->db_level);

	/*
	 * We should only be dirtying in syncing context if it's the
	 * mos or we're initializing the os or it's a special object.
	 * However, we are allowed to dirty in syncing context provided
	 * we already dirtied it in open context.  Hence we must make
	 * this assertion only if we're not already dirty.
	 */
	os = dn->dn_objset;
	ASSERT(!dmu_tx_is_syncing(tx) || DMU_OBJECT_IS_SPECIAL(dn->dn_object) ||
	    os->os_dsl_dataset == NULL || BP_IS_HOLE(os->os_rootbp));
	ASSERT(db->db.db_size != 0);

	dprintf_dbuf(db, "size=%llx\n", (u_longlong_t)db->db.db_size);

	if (db->db_blkid != DMU_BONUS_BLKID) {
		/*
		 * Update the accounting.
		 * Note: we delay "free accounting" until after we drop
		 * the db_mtx.  This keeps us from grabbing other locks
		 * (and possibly deadlocking) in bp_get_dsize() while
		 * also holding the db_mtx.
		 */
		dnode_willuse_space(dn, db->db.db_size, tx);
		do_free_accounting = dbuf_block_freeable(db);
	}

	/*
	 * If this buffer is dirty in an old transaction group we need
	 * to make a copy of it so that the changes we make in this
	 * transaction group won't leak out when we sync the older txg.
	 */
	dr = kmem_zalloc(sizeof (dbuf_dirty_record_t), KM_SLEEP);
	if (db->db_level == 0) {
		void *data_old = db->db_buf;

		if (db->db_state != DB_NOFILL) {
			if (db->db_blkid == DMU_BONUS_BLKID) {
				dbuf_fix_old_data(db, tx->tx_txg);
				data_old = db->db.db_data;
			} else if (db->db.db_object != DMU_META_DNODE_OBJECT) {
				/*
				 * Release the data buffer from the cache so
				 * that we can modify it without impacting
				 * possible other users of this cached data
				 * block.  Note that indirect blocks and
				 * private objects are not released until the
				 * syncing state (since they are only modified
				 * then).
				 */
				arc_release(db->db_buf, db);
				dbuf_fix_old_data(db, tx->tx_txg);
				data_old = db->db_buf;
			}
			ASSERT(data_old != NULL);
		}
		dr->dt.dl.dr_data = data_old;
	} else {
		mutex_init(&dr->dt.di.dr_mtx, NULL, MUTEX_DEFAULT, NULL);
		list_create(&dr->dt.di.dr_children,
		    sizeof (dbuf_dirty_record_t),
		    offsetof(dbuf_dirty_record_t, dr_dirty_node));
	}
	dr->dr_dbuf = db;
	dr->dr_txg = tx->tx_txg;
	dr->dr_next = *drp;
	*drp = dr;

	/*
	 * We could have been freed_in_flight between the dbuf_noread
	 * and dbuf_dirty.  We win, as though the dbuf_noread() had
	 * happened after the free.
	 */
	if (db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID &&
	    db->db_blkid != DMU_SPILL_BLKID) {
		mutex_enter(&dn->dn_mtx);
		dnode_clear_range(dn, db->db_blkid, 1, tx);
		mutex_exit(&dn->dn_mtx);
		db->db_freed_in_flight = FALSE;
	}

	/*
	 * This buffer is now part of this txg
	 */
	dbuf_add_ref(db, (void *)(uintptr_t)tx->tx_txg);
	db->db_dirtycnt += 1;
	ASSERT3U(db->db_dirtycnt, <=, 3);

	mutex_exit(&db->db_mtx);

	if (db->db_blkid == DMU_BONUS_BLKID ||
	    db->db_blkid == DMU_SPILL_BLKID) {
		mutex_enter(&dn->dn_mtx);
		ASSERT(!list_link_active(&dr->dr_dirty_node));
		list_insert_tail(&dn->dn_dirty_records[txgoff], dr);
		mutex_exit(&dn->dn_mtx);
		dnode_setdirty(dn, tx);
		DB_DNODE_EXIT(db);
		return (dr);
	} else if (do_free_accounting) {
		blkptr_t *bp = db->db_blkptr;
		int64_t willfree = (bp && !BP_IS_HOLE(bp)) ?
		    bp_get_dsize(os->os_spa, bp) : db->db.db_size;
		/*
		 * This is only a guess -- if the dbuf is dirty
		 * in a previous txg, we don't know how much
		 * space it will use on disk yet.  We should
		 * really have the struct_rwlock to access
		 * db_blkptr, but since this is just a guess,
		 * it's OK if we get an odd answer.
		 */
		ddt_prefetch(os->os_spa, bp);
		dnode_willuse_space(dn, -willfree, tx);
	}

	if (!RW_WRITE_HELD(&dn->dn_struct_rwlock)) {
		rw_enter(&dn->dn_struct_rwlock, RW_READER);
		drop_struct_lock = TRUE;
	}

	if (db->db_level == 0) {
		dnode_new_blkid(dn, db->db_blkid, tx, drop_struct_lock);
		ASSERT(dn->dn_maxblkid >= db->db_blkid);
	}

	if (db->db_level+1 < dn->dn_nlevels) {
		dmu_buf_impl_t *parent = db->db_parent;
		dbuf_dirty_record_t *di;
		int parent_held = FALSE;

		if (db->db_parent == NULL || db->db_parent == dn->dn_dbuf) {
			int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT;

			parent = dbuf_hold_level(dn, db->db_level+1,
			    db->db_blkid >> epbs, FTAG);
			ASSERT(parent != NULL);
			parent_held = TRUE;
		}
		if (drop_struct_lock)
			rw_exit(&dn->dn_struct_rwlock);
		ASSERT3U(db->db_level+1, ==, parent->db_level);
		di = dbuf_dirty(parent, tx);
		if (parent_held)
			dbuf_rele(parent, FTAG);

		mutex_enter(&db->db_mtx);
		/*  possible race with dbuf_undirty() */
		if (db->db_last_dirty == dr ||
		    dn->dn_object == DMU_META_DNODE_OBJECT) {
			mutex_enter(&di->dt.di.dr_mtx);
			ASSERT3U(di->dr_txg, ==, tx->tx_txg);
			ASSERT(!list_link_active(&dr->dr_dirty_node));
			list_insert_tail(&di->dt.di.dr_children, dr);
			mutex_exit(&di->dt.di.dr_mtx);
			dr->dr_parent = di;
		}
		mutex_exit(&db->db_mtx);
	} else {
		ASSERT(db->db_level+1 == dn->dn_nlevels);
		ASSERT(db->db_blkid < dn->dn_nblkptr);
		ASSERT(db->db_parent == NULL || db->db_parent == dn->dn_dbuf);
		mutex_enter(&dn->dn_mtx);
		ASSERT(!list_link_active(&dr->dr_dirty_node));
		list_insert_tail(&dn->dn_dirty_records[txgoff], dr);
		mutex_exit(&dn->dn_mtx);
		if (drop_struct_lock)
			rw_exit(&dn->dn_struct_rwlock);
	}

	dnode_setdirty(dn, tx);
	DB_DNODE_EXIT(db);
	return (dr);
}

static int
dbuf_undirty(dmu_buf_impl_t *db, dmu_tx_t *tx)
{
	dnode_t *dn;
	uint64_t txg = tx->tx_txg;
	dbuf_dirty_record_t *dr, **drp;

	ASSERT(txg != 0);
	ASSERT(db->db_blkid != DMU_BONUS_BLKID);

	mutex_enter(&db->db_mtx);
	/*
	 * If this buffer is not dirty, we're done.
	 */
	for (drp = &db->db_last_dirty; (dr = *drp) != NULL; drp = &dr->dr_next)
		if (dr->dr_txg <= txg)
			break;
	if (dr == NULL || dr->dr_txg < txg) {
		mutex_exit(&db->db_mtx);
		return (0);
	}
	ASSERT(dr->dr_txg == txg);
	ASSERT(dr->dr_dbuf == db);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);

	/*
	 * If this buffer is currently held, we cannot undirty
	 * it, since one of the current holders may be in the
	 * middle of an update.  Note that users of dbuf_undirty()
	 * should not place a hold on the dbuf before the call.
	 */
	if (refcount_count(&db->db_holds) > db->db_dirtycnt) {
		mutex_exit(&db->db_mtx);
		/* Make sure we don't toss this buffer at sync phase */
		mutex_enter(&dn->dn_mtx);
		dnode_clear_range(dn, db->db_blkid, 1, tx);
		mutex_exit(&dn->dn_mtx);
		DB_DNODE_EXIT(db);
		return (0);
	}

	dprintf_dbuf(db, "size=%llx\n", (u_longlong_t)db->db.db_size);

	ASSERT(db->db.db_size != 0);

	/* XXX would be nice to fix up dn_towrite_space[] */

	*drp = dr->dr_next;

	if (dr->dr_parent) {
		mutex_enter(&dr->dr_parent->dt.di.dr_mtx);
		list_remove(&dr->dr_parent->dt.di.dr_children, dr);
		mutex_exit(&dr->dr_parent->dt.di.dr_mtx);
	} else if (db->db_level+1 == dn->dn_nlevels) {
		ASSERT(db->db_blkptr == NULL || db->db_parent == dn->dn_dbuf);
		mutex_enter(&dn->dn_mtx);
		list_remove(&dn->dn_dirty_records[txg & TXG_MASK], dr);
		mutex_exit(&dn->dn_mtx);
	}
	DB_DNODE_EXIT(db);

	if (db->db_level == 0) {
		if (db->db_state != DB_NOFILL) {
			dbuf_unoverride(dr);

			ASSERT(db->db_buf != NULL);
			ASSERT(dr->dt.dl.dr_data != NULL);
			if (dr->dt.dl.dr_data != db->db_buf)
				VERIFY(arc_buf_remove_ref(dr->dt.dl.dr_data,
				    db) == 1);
		}
	} else {
		ASSERT(db->db_buf != NULL);
		ASSERT(list_head(&dr->dt.di.dr_children) == NULL);
		mutex_destroy(&dr->dt.di.dr_mtx);
		list_destroy(&dr->dt.di.dr_children);
	}
	kmem_free(dr, sizeof (dbuf_dirty_record_t));

	ASSERT(db->db_dirtycnt > 0);
	db->db_dirtycnt -= 1;

	if (refcount_remove(&db->db_holds, (void *)(uintptr_t)txg) == 0) {
		arc_buf_t *buf = db->db_buf;

		ASSERT(db->db_state == DB_NOFILL || arc_released(buf));
		dbuf_set_data(db, NULL);
		VERIFY(arc_buf_remove_ref(buf, db) == 1);
		dbuf_evict(db);
		return (1);
	}

	mutex_exit(&db->db_mtx);
	return (0);
}

#pragma weak dmu_buf_will_dirty = dbuf_will_dirty
void
dbuf_will_dirty(dmu_buf_impl_t *db, dmu_tx_t *tx)
{
	int rf = DB_RF_MUST_SUCCEED | DB_RF_NOPREFETCH;

	ASSERT(tx->tx_txg != 0);
	ASSERT(!refcount_is_zero(&db->db_holds));

	DB_DNODE_ENTER(db);
	if (RW_WRITE_HELD(&DB_DNODE(db)->dn_struct_rwlock))
		rf |= DB_RF_HAVESTRUCT;
	DB_DNODE_EXIT(db);
	(void) dbuf_read(db, NULL, rf);
	(void) dbuf_dirty(db, tx);
}

void
dmu_buf_will_not_fill(dmu_buf_t *db_fake, dmu_tx_t *tx)
{
	dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;

	db->db_state = DB_NOFILL;

	dmu_buf_will_fill(db_fake, tx);
}

void
dmu_buf_will_fill(dmu_buf_t *db_fake, dmu_tx_t *tx)
{
	dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;

	ASSERT(db->db_blkid != DMU_BONUS_BLKID);
	ASSERT(tx->tx_txg != 0);
	ASSERT(db->db_level == 0);
	ASSERT(!refcount_is_zero(&db->db_holds));

	ASSERT(db->db.db_object != DMU_META_DNODE_OBJECT ||
	    dmu_tx_private_ok(tx));

	dbuf_noread(db);
	(void) dbuf_dirty(db, tx);
}

#pragma weak dmu_buf_fill_done = dbuf_fill_done
/* ARGSUSED */
void
dbuf_fill_done(dmu_buf_impl_t *db, dmu_tx_t *tx)
{
	mutex_enter(&db->db_mtx);
	DBUF_VERIFY(db);

	if (db->db_state == DB_FILL) {
		if (db->db_level == 0 && db->db_freed_in_flight) {
			ASSERT(db->db_blkid != DMU_BONUS_BLKID);
			/* we were freed while filling */
			/* XXX dbuf_undirty? */
			bzero(db->db.db_data, db->db.db_size);
			db->db_freed_in_flight = FALSE;
		}
		db->db_state = DB_CACHED;
		cv_broadcast(&db->db_changed);
	}
	mutex_exit(&db->db_mtx);
}

/*
 * Directly assign a provided arc buf to a given dbuf if it's not referenced
 * by anybody except our caller. Otherwise copy arcbuf's contents to dbuf.
 */
void
dbuf_assign_arcbuf(dmu_buf_impl_t *db, arc_buf_t *buf, dmu_tx_t *tx)
{
	ASSERT(!refcount_is_zero(&db->db_holds));
	ASSERT(db->db_blkid != DMU_BONUS_BLKID);
	ASSERT(db->db_level == 0);
	ASSERT(DBUF_GET_BUFC_TYPE(db) == ARC_BUFC_DATA);
	ASSERT(buf != NULL);
	ASSERT(arc_buf_size(buf) == db->db.db_size);
	ASSERT(tx->tx_txg != 0);

	arc_return_buf(buf, db);
	ASSERT(arc_released(buf));

	mutex_enter(&db->db_mtx);

	while (db->db_state == DB_READ || db->db_state == DB_FILL)
		cv_wait(&db->db_changed, &db->db_mtx);

	ASSERT(db->db_state == DB_CACHED || db->db_state == DB_UNCACHED);

	if (db->db_state == DB_CACHED &&
	    refcount_count(&db->db_holds) - 1 > db->db_dirtycnt) {
		mutex_exit(&db->db_mtx);
		(void) dbuf_dirty(db, tx);
		bcopy(buf->b_data, db->db.db_data, db->db.db_size);
		VERIFY(arc_buf_remove_ref(buf, db) == 1);
		xuio_stat_wbuf_copied();
		return;
	}

	xuio_stat_wbuf_nocopy();
	if (db->db_state == DB_CACHED) {
		dbuf_dirty_record_t *dr = db->db_last_dirty;

		ASSERT(db->db_buf != NULL);
		if (dr != NULL && dr->dr_txg == tx->tx_txg) {
			ASSERT(dr->dt.dl.dr_data == db->db_buf);
			if (!arc_released(db->db_buf)) {
				ASSERT(dr->dt.dl.dr_override_state ==
				    DR_OVERRIDDEN);
				arc_release(db->db_buf, db);
			}
			dr->dt.dl.dr_data = buf;
			VERIFY(arc_buf_remove_ref(db->db_buf, db) == 1);
		} else if (dr == NULL || dr->dt.dl.dr_data != db->db_buf) {
			arc_release(db->db_buf, db);
			VERIFY(arc_buf_remove_ref(db->db_buf, db) == 1);
		}
		db->db_buf = NULL;
	}
	ASSERT(db->db_buf == NULL);
	dbuf_set_data(db, buf);
	db->db_state = DB_FILL;
	mutex_exit(&db->db_mtx);
	(void) dbuf_dirty(db, tx);
	dbuf_fill_done(db, tx);
}

/*
 * "Clear" the contents of this dbuf.  This will mark the dbuf
 * EVICTING and clear *most* of its references.  Unfortunetely,
 * when we are not holding the dn_dbufs_mtx, we can't clear the
 * entry in the dn_dbufs list.  We have to wait until dbuf_destroy()
 * in this case.  For callers from the DMU we will usually see:
 *	dbuf_clear()->arc_buf_evict()->dbuf_do_evict()->dbuf_destroy()
 * For the arc callback, we will usually see:
 *	dbuf_do_evict()->dbuf_clear();dbuf_destroy()
 * Sometimes, though, we will get a mix of these two:
 *	DMU: dbuf_clear()->arc_buf_evict()
 *	ARC: dbuf_do_evict()->dbuf_destroy()
 */
void
dbuf_clear(dmu_buf_impl_t *db)
{
	dnode_t *dn;
	dmu_buf_impl_t *parent = db->db_parent;
	dmu_buf_impl_t *dndb;
	int dbuf_gone = FALSE;

	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(refcount_is_zero(&db->db_holds));

	dbuf_evict_user(db);

	if (db->db_state == DB_CACHED) {
		ASSERT(db->db.db_data != NULL);
		if (db->db_blkid == DMU_BONUS_BLKID) {
			zio_buf_free(db->db.db_data, DN_MAX_BONUSLEN);
			arc_space_return(DN_MAX_BONUSLEN, ARC_SPACE_OTHER);
		}
		db->db.db_data = NULL;
		db->db_state = DB_UNCACHED;
	}

	ASSERT(db->db_state == DB_UNCACHED || db->db_state == DB_NOFILL);
	ASSERT(db->db_data_pending == NULL);

	db->db_state = DB_EVICTING;
	db->db_blkptr = NULL;

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	dndb = dn->dn_dbuf;
	if (db->db_blkid != DMU_BONUS_BLKID && MUTEX_HELD(&dn->dn_dbufs_mtx)) {
		list_remove(&dn->dn_dbufs, db);
		(void) atomic_dec_32_nv(&dn->dn_dbufs_count);
		membar_producer();
		DB_DNODE_EXIT(db);
		/*
		 * Decrementing the dbuf count means that the hold corresponding
		 * to the removed dbuf is no longer discounted in dnode_move(),
		 * so the dnode cannot be moved until after we release the hold.
		 * The membar_producer() ensures visibility of the decremented
		 * value in dnode_move(), since DB_DNODE_EXIT doesn't actually
		 * release any lock.
		 */
		dnode_rele(dn, db);
		db->db_dnode_handle = NULL;
	} else {
		DB_DNODE_EXIT(db);
	}

	if (db->db_buf)
		dbuf_gone = arc_buf_evict(db->db_buf);

	if (!dbuf_gone)
		mutex_exit(&db->db_mtx);

	/*
	 * If this dbuf is referenced from an indirect dbuf,
	 * decrement the ref count on the indirect dbuf.
	 */
	if (parent && parent != dndb)
		dbuf_rele(parent, db);
}

static int
dbuf_findbp(dnode_t *dn, int level, uint64_t blkid, int fail_sparse,
    dmu_buf_impl_t **parentp, blkptr_t **bpp)
{
	int nlevels, epbs;

	*parentp = NULL;
	*bpp = NULL;

	ASSERT(blkid != DMU_BONUS_BLKID);

	if (blkid == DMU_SPILL_BLKID) {
		mutex_enter(&dn->dn_mtx);
		if (dn->dn_have_spill &&
		    (dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR))
			*bpp = &dn->dn_phys->dn_spill;
		else
			*bpp = NULL;
		dbuf_add_ref(dn->dn_dbuf, NULL);
		*parentp = dn->dn_dbuf;
		mutex_exit(&dn->dn_mtx);
		return (0);
	}

	if (dn->dn_phys->dn_nlevels == 0)
		nlevels = 1;
	else
		nlevels = dn->dn_phys->dn_nlevels;

	epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT;

	ASSERT3U(level * epbs, <, 64);
	ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));
	if (level >= nlevels ||
	    (blkid > (dn->dn_phys->dn_maxblkid >> (level * epbs)))) {
		/* the buffer has no parent yet */
		return (ENOENT);
	} else if (level < nlevels-1) {
		/* this block is referenced from an indirect block */
		int err = dbuf_hold_impl(dn, level+1,
		    blkid >> epbs, fail_sparse, NULL, parentp);
		if (err)
			return (err);
		err = dbuf_read(*parentp, NULL,
		    (DB_RF_HAVESTRUCT | DB_RF_NOPREFETCH | DB_RF_CANFAIL));
		if (err) {
			dbuf_rele(*parentp, NULL);
			*parentp = NULL;
			return (err);
		}
		*bpp = ((blkptr_t *)(*parentp)->db.db_data) +
		    (blkid & ((1ULL << epbs) - 1));
		return (0);
	} else {
		/* the block is referenced from the dnode */
		ASSERT3U(level, ==, nlevels-1);
		ASSERT(dn->dn_phys->dn_nblkptr == 0 ||
		    blkid < dn->dn_phys->dn_nblkptr);
		if (dn->dn_dbuf) {
			dbuf_add_ref(dn->dn_dbuf, NULL);
			*parentp = dn->dn_dbuf;
		}
		*bpp = &dn->dn_phys->dn_blkptr[blkid];
		return (0);
	}
}

static dmu_buf_impl_t *
dbuf_create(dnode_t *dn, uint8_t level, uint64_t blkid,
    dmu_buf_impl_t *parent, blkptr_t *blkptr)
{
	objset_t *os = dn->dn_objset;
	dmu_buf_impl_t *db, *odb;

	ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));
	ASSERT(dn->dn_type != DMU_OT_NONE);

	db = kmem_cache_alloc(dbuf_cache, KM_SLEEP);

	db->db_objset = os;
	db->db.db_object = dn->dn_object;
	db->db_level = level;
	db->db_blkid = blkid;
	db->db_last_dirty = NULL;
	db->db_dirtycnt = 0;
	db->db_dnode_handle = dn->dn_handle;
	db->db_parent = parent;
	db->db_blkptr = blkptr;

	db->db_user_ptr = NULL;
	db->db_user_data_ptr_ptr = NULL;
	db->db_evict_func = NULL;
	db->db_immediate_evict = 0;
	db->db_freed_in_flight = 0;

	if (blkid == DMU_BONUS_BLKID) {
		ASSERT3P(parent, ==, dn->dn_dbuf);
		db->db.db_size = DN_MAX_BONUSLEN -
		    (dn->dn_nblkptr-1) * sizeof (blkptr_t);
		ASSERT3U(db->db.db_size, >=, dn->dn_bonuslen);
		db->db.db_offset = DMU_BONUS_BLKID;
		db->db_state = DB_UNCACHED;
		/* the bonus dbuf is not placed in the hash table */
		arc_space_consume(sizeof (dmu_buf_impl_t), ARC_SPACE_OTHER);
		return (db);
	} else if (blkid == DMU_SPILL_BLKID) {
		db->db.db_size = (blkptr != NULL) ?
		    BP_GET_LSIZE(blkptr) : SPA_MINBLOCKSIZE;
		db->db.db_offset = 0;
	} else {
		int blocksize =
		    db->db_level ? 1<<dn->dn_indblkshift :  dn->dn_datablksz;
		db->db.db_size = blocksize;
		db->db.db_offset = db->db_blkid * blocksize;
	}

	/*
	 * Hold the dn_dbufs_mtx while we get the new dbuf
	 * in the hash table *and* added to the dbufs list.
	 * This prevents a possible deadlock with someone
	 * trying to look up this dbuf before its added to the
	 * dn_dbufs list.
	 */
	mutex_enter(&dn->dn_dbufs_mtx);
	db->db_state = DB_EVICTING;
	if ((odb = dbuf_hash_insert(db)) != NULL) {
		/* someone else inserted it first */
		kmem_cache_free(dbuf_cache, db);
		mutex_exit(&dn->dn_dbufs_mtx);
		return (odb);
	}
	list_insert_head(&dn->dn_dbufs, db);
	db->db_state = DB_UNCACHED;
	mutex_exit(&dn->dn_dbufs_mtx);
	arc_space_consume(sizeof (dmu_buf_impl_t), ARC_SPACE_OTHER);

	if (parent && parent != dn->dn_dbuf)
		dbuf_add_ref(parent, db);

	ASSERT(dn->dn_object == DMU_META_DNODE_OBJECT ||
	    refcount_count(&dn->dn_holds) > 0);
	(void) refcount_add(&dn->dn_holds, db);
	(void) atomic_inc_32_nv(&dn->dn_dbufs_count);

	dprintf_dbuf(db, "db=%p\n", db);

	return (db);
}

static int
dbuf_do_evict(void *private)
{
	arc_buf_t *buf = private;
	dmu_buf_impl_t *db = buf->b_private;

	if (!MUTEX_HELD(&db->db_mtx))
		mutex_enter(&db->db_mtx);

	ASSERT(refcount_is_zero(&db->db_holds));

	if (db->db_state != DB_EVICTING) {
		ASSERT(db->db_state == DB_CACHED);
		DBUF_VERIFY(db);
		db->db_buf = NULL;
		dbuf_evict(db);
	} else {
		mutex_exit(&db->db_mtx);
		dbuf_destroy(db);
	}
	return (0);
}

static void
dbuf_destroy(dmu_buf_impl_t *db)
{
	ASSERT(refcount_is_zero(&db->db_holds));

	if (db->db_blkid != DMU_BONUS_BLKID) {
		/*
		 * If this dbuf is still on the dn_dbufs list,
		 * remove it from that list.
		 */
		if (db->db_dnode_handle != NULL) {
			dnode_t *dn;

			DB_DNODE_ENTER(db);
			dn = DB_DNODE(db);
			mutex_enter(&dn->dn_dbufs_mtx);
			list_remove(&dn->dn_dbufs, db);
			(void) atomic_dec_32_nv(&dn->dn_dbufs_count);
			mutex_exit(&dn->dn_dbufs_mtx);
			DB_DNODE_EXIT(db);
			/*
			 * Decrementing the dbuf count means that the hold
			 * corresponding to the removed dbuf is no longer
			 * discounted in dnode_move(), so the dnode cannot be
			 * moved until after we release the hold.
			 */
			dnode_rele(dn, db);
			db->db_dnode_handle = NULL;
		}
		dbuf_hash_remove(db);
	}
	db->db_parent = NULL;
	db->db_buf = NULL;

	ASSERT(!list_link_active(&db->db_link));
	ASSERT(db->db.db_data == NULL);
	ASSERT(db->db_hash_next == NULL);
	ASSERT(db->db_blkptr == NULL);
	ASSERT(db->db_data_pending == NULL);

	kmem_cache_free(dbuf_cache, db);
	arc_space_return(sizeof (dmu_buf_impl_t), ARC_SPACE_OTHER);
}

void
dbuf_prefetch(dnode_t *dn, uint64_t blkid)
{
	dmu_buf_impl_t *db = NULL;
	blkptr_t *bp = NULL;

	ASSERT(blkid != DMU_BONUS_BLKID);
	ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));

	if (dnode_block_freed(dn, blkid))
		return;

	/* dbuf_find() returns with db_mtx held */
	if (db = dbuf_find(dn, 0, blkid)) {
		/*
		 * This dbuf is already in the cache.  We assume that
		 * it is already CACHED, or else about to be either
		 * read or filled.
		 */
		mutex_exit(&db->db_mtx);
		return;
	}

	if (dbuf_findbp(dn, 0, blkid, TRUE, &db, &bp) == 0) {
		if (bp && !BP_IS_HOLE(bp)) {
			int priority = dn->dn_type == DMU_OT_DDT_ZAP ?
			    ZIO_PRIORITY_DDT_PREFETCH : ZIO_PRIORITY_ASYNC_READ;
			arc_buf_t *pbuf;
			dsl_dataset_t *ds = dn->dn_objset->os_dsl_dataset;
			uint32_t aflags = ARC_NOWAIT | ARC_PREFETCH;
			zbookmark_t zb;

			SET_BOOKMARK(&zb, ds ? ds->ds_object : DMU_META_OBJSET,
			    dn->dn_object, 0, blkid);

			if (db)
				pbuf = db->db_buf;
			else
				pbuf = dn->dn_objset->os_phys_buf;

			(void) dsl_read(NULL, dn->dn_objset->os_spa,
			    bp, pbuf, NULL, NULL, priority,
			    ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE,
			    &aflags, &zb);
		}
		if (db)
			dbuf_rele(db, NULL);
	}
}

/*
 * Returns with db_holds incremented, and db_mtx not held.
 * Note: dn_struct_rwlock must be held.
 */
int
dbuf_hold_impl(dnode_t *dn, uint8_t level, uint64_t blkid, int fail_sparse,
    void *tag, dmu_buf_impl_t **dbp)
{
	dmu_buf_impl_t *db, *parent = NULL;

	ASSERT(blkid != DMU_BONUS_BLKID);
	ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));
	ASSERT3U(dn->dn_nlevels, >, level);

	*dbp = NULL;
top:
	/* dbuf_find() returns with db_mtx held */
	db = dbuf_find(dn, level, blkid);

	if (db == NULL) {
		blkptr_t *bp = NULL;
		int err;

		ASSERT3P(parent, ==, NULL);
		err = dbuf_findbp(dn, level, blkid, fail_sparse, &parent, &bp);
		if (fail_sparse) {
			if (err == 0 && bp && BP_IS_HOLE(bp))
				err = ENOENT;
			if (err) {
				if (parent)
					dbuf_rele(parent, NULL);
				return (err);
			}
		}
		if (err && err != ENOENT)
			return (err);
		db = dbuf_create(dn, level, blkid, parent, bp);
	}

	if (db->db_buf && refcount_is_zero(&db->db_holds)) {
		arc_buf_add_ref(db->db_buf, db);
		if (db->db_buf->b_data == NULL) {
			dbuf_clear(db);
			if (parent) {
				dbuf_rele(parent, NULL);
				parent = NULL;
			}
			goto top;
		}
		ASSERT3P(db->db.db_data, ==, db->db_buf->b_data);
	}

	ASSERT(db->db_buf == NULL || arc_referenced(db->db_buf));

	/*
	 * If this buffer is currently syncing out, and we are are
	 * still referencing it from db_data, we need to make a copy
	 * of it in case we decide we want to dirty it again in this txg.
	 */
	if (db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID &&
	    dn->dn_object != DMU_META_DNODE_OBJECT &&
	    db->db_state == DB_CACHED && db->db_data_pending) {
		dbuf_dirty_record_t *dr = db->db_data_pending;

		if (dr->dt.dl.dr_data == db->db_buf) {
			arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);

			dbuf_set_data(db,
			    arc_buf_alloc(dn->dn_objset->os_spa,
			    db->db.db_size, db, type));
			bcopy(dr->dt.dl.dr_data->b_data, db->db.db_data,
			    db->db.db_size);
		}
	}

	(void) refcount_add(&db->db_holds, tag);
	dbuf_update_data(db);
	DBUF_VERIFY(db);
	mutex_exit(&db->db_mtx);

	/* NOTE: we can't rele the parent until after we drop the db_mtx */
	if (parent)
		dbuf_rele(parent, NULL);

	ASSERT3P(DB_DNODE(db), ==, dn);
	ASSERT3U(db->db_blkid, ==, blkid);
	ASSERT3U(db->db_level, ==, level);
	*dbp = db;

	return (0);
}

dmu_buf_impl_t *
dbuf_hold(dnode_t *dn, uint64_t blkid, void *tag)
{
	dmu_buf_impl_t *db;
	int err = dbuf_hold_impl(dn, 0, blkid, FALSE, tag, &db);
	return (err ? NULL : db);
}

dmu_buf_impl_t *
dbuf_hold_level(dnode_t *dn, int level, uint64_t blkid, void *tag)
{
	dmu_buf_impl_t *db;
	int err = dbuf_hold_impl(dn, level, blkid, FALSE, tag, &db);
	return (err ? NULL : db);
}

void
dbuf_create_bonus(dnode_t *dn)
{
	ASSERT(RW_WRITE_HELD(&dn->dn_struct_rwlock));

	ASSERT(dn->dn_bonus == NULL);
	dn->dn_bonus = dbuf_create(dn, 0, DMU_BONUS_BLKID, dn->dn_dbuf, NULL);
}

int
dbuf_spill_set_blksz(dmu_buf_t *db_fake, uint64_t blksz, dmu_tx_t *tx)
{
	dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;
	dnode_t *dn;

	if (db->db_blkid != DMU_SPILL_BLKID)
		return (ENOTSUP);
	if (blksz == 0)
		blksz = SPA_MINBLOCKSIZE;
	if (blksz > SPA_MAXBLOCKSIZE)
		blksz = SPA_MAXBLOCKSIZE;
	else
		blksz = P2ROUNDUP(blksz, SPA_MINBLOCKSIZE);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
	dbuf_new_size(db, blksz, tx);
	rw_exit(&dn->dn_struct_rwlock);
	DB_DNODE_EXIT(db);

	return (0);
}

void
dbuf_rm_spill(dnode_t *dn, dmu_tx_t *tx)
{
	dbuf_free_range(dn, DMU_SPILL_BLKID, DMU_SPILL_BLKID, tx);
}

#pragma weak dmu_buf_add_ref = dbuf_add_ref
void
dbuf_add_ref(dmu_buf_impl_t *db, void *tag)
{
	int64_t holds = refcount_add(&db->db_holds, tag);
	ASSERT(holds > 1);
}

/*
 * If you call dbuf_rele() you had better not be referencing the dnode handle
 * unless you have some other direct or indirect hold on the dnode. (An indirect
 * hold is a hold on one of the dnode's dbufs, including the bonus buffer.)
 * Without that, the dbuf_rele() could lead to a dnode_rele() followed by the
 * dnode's parent dbuf evicting its dnode handles.
 */
#pragma weak dmu_buf_rele = dbuf_rele
void
dbuf_rele(dmu_buf_impl_t *db, void *tag)
{
	mutex_enter(&db->db_mtx);
	dbuf_rele_and_unlock(db, tag);
}

/*
 * dbuf_rele() for an already-locked dbuf.  This is necessary to allow
 * db_dirtycnt and db_holds to be updated atomically.
 */
void
dbuf_rele_and_unlock(dmu_buf_impl_t *db, void *tag)
{
	int64_t holds;

	ASSERT(MUTEX_HELD(&db->db_mtx));
	DBUF_VERIFY(db);

	/*
	 * Remove the reference to the dbuf before removing its hold on the
	 * dnode so we can guarantee in dnode_move() that a referenced bonus
	 * buffer has a corresponding dnode hold.
	 */
	holds = refcount_remove(&db->db_holds, tag);
	ASSERT(holds >= 0);

	/*
	 * We can't freeze indirects if there is a possibility that they
	 * may be modified in the current syncing context.
	 */
	if (db->db_buf && holds == (db->db_level == 0 ? db->db_dirtycnt : 0))
		arc_buf_freeze(db->db_buf);

	if (holds == db->db_dirtycnt &&
	    db->db_level == 0 && db->db_immediate_evict)
		dbuf_evict_user(db);

	if (holds == 0) {
		if (db->db_blkid == DMU_BONUS_BLKID) {
			mutex_exit(&db->db_mtx);

			/*
			 * If the dnode moves here, we cannot cross this barrier
			 * until the move completes.
			 */
			DB_DNODE_ENTER(db);
			(void) atomic_dec_32_nv(&DB_DNODE(db)->dn_dbufs_count);
			DB_DNODE_EXIT(db);
			/*
			 * The bonus buffer's dnode hold is no longer discounted
			 * in dnode_move(). The dnode cannot move until after
			 * the dnode_rele().
			 */
			dnode_rele(DB_DNODE(db), db);
		} else if (db->db_buf == NULL) {
			/*
			 * This is a special case: we never associated this
			 * dbuf with any data allocated from the ARC.
			 */
			ASSERT(db->db_state == DB_UNCACHED ||
			    db->db_state == DB_NOFILL);
			dbuf_evict(db);
		} else if (arc_released(db->db_buf)) {
			arc_buf_t *buf = db->db_buf;
			/*
			 * This dbuf has anonymous data associated with it.
			 */
			dbuf_set_data(db, NULL);
			VERIFY(arc_buf_remove_ref(buf, db) == 1);
			dbuf_evict(db);
		} else {
			VERIFY(arc_buf_remove_ref(db->db_buf, db) == 0);
			if (!DBUF_IS_CACHEABLE(db))
				dbuf_clear(db);
			else
				mutex_exit(&db->db_mtx);
		}
	} else {
		mutex_exit(&db->db_mtx);
	}
}

#pragma weak dmu_buf_refcount = dbuf_refcount
uint64_t
dbuf_refcount(dmu_buf_impl_t *db)
{
	return (refcount_count(&db->db_holds));
}

void *
dmu_buf_set_user(dmu_buf_t *db_fake, void *user_ptr, void *user_data_ptr_ptr,
    dmu_buf_evict_func_t *evict_func)
{
	return (dmu_buf_update_user(db_fake, NULL, user_ptr,
	    user_data_ptr_ptr, evict_func));
}

void *
dmu_buf_set_user_ie(dmu_buf_t *db_fake, void *user_ptr, void *user_data_ptr_ptr,
    dmu_buf_evict_func_t *evict_func)
{
	dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;

	db->db_immediate_evict = TRUE;
	return (dmu_buf_update_user(db_fake, NULL, user_ptr,
	    user_data_ptr_ptr, evict_func));
}

void *
dmu_buf_update_user(dmu_buf_t *db_fake, void *old_user_ptr, void *user_ptr,
    void *user_data_ptr_ptr, dmu_buf_evict_func_t *evict_func)
{
	dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;
	ASSERT(db->db_level == 0);

	ASSERT((user_ptr == NULL) == (evict_func == NULL));

	mutex_enter(&db->db_mtx);

	if (db->db_user_ptr == old_user_ptr) {
		db->db_user_ptr = user_ptr;
		db->db_user_data_ptr_ptr = user_data_ptr_ptr;
		db->db_evict_func = evict_func;

		dbuf_update_data(db);
	} else {
		old_user_ptr = db->db_user_ptr;
	}

	mutex_exit(&db->db_mtx);
	return (old_user_ptr);
}

void *
dmu_buf_get_user(dmu_buf_t *db_fake)
{
	dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;
	ASSERT(!refcount_is_zero(&db->db_holds));

	return (db->db_user_ptr);
}

boolean_t
dmu_buf_freeable(dmu_buf_t *dbuf)
{
	boolean_t res = B_FALSE;
	dmu_buf_impl_t *db = (dmu_buf_impl_t *)dbuf;

	if (db->db_blkptr)
		res = dsl_dataset_block_freeable(db->db_objset->os_dsl_dataset,
		    db->db_blkptr, db->db_blkptr->blk_birth);

	return (res);
}

static void
dbuf_check_blkptr(dnode_t *dn, dmu_buf_impl_t *db)
{
	/* ASSERT(dmu_tx_is_syncing(tx) */
	ASSERT(MUTEX_HELD(&db->db_mtx));

	if (db->db_blkptr != NULL)
		return;

	if (db->db_blkid == DMU_SPILL_BLKID) {
		db->db_blkptr = &dn->dn_phys->dn_spill;
		BP_ZERO(db->db_blkptr);
		return;
	}
	if (db->db_level == dn->dn_phys->dn_nlevels-1) {
		/*
		 * This buffer was allocated at a time when there was
		 * no available blkptrs from the dnode, or it was
		 * inappropriate to hook it in (i.e., nlevels mis-match).
		 */
		ASSERT(db->db_blkid < dn->dn_phys->dn_nblkptr);
		ASSERT(db->db_parent == NULL);
		db->db_parent = dn->dn_dbuf;
		db->db_blkptr = &dn->dn_phys->dn_blkptr[db->db_blkid];
		DBUF_VERIFY(db);
	} else {
		dmu_buf_impl_t *parent = db->db_parent;
		int epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT;

		ASSERT(dn->dn_phys->dn_nlevels > 1);
		if (parent == NULL) {
			mutex_exit(&db->db_mtx);
			rw_enter(&dn->dn_struct_rwlock, RW_READER);
			(void) dbuf_hold_impl(dn, db->db_level+1,
			    db->db_blkid >> epbs, FALSE, db, &parent);
			rw_exit(&dn->dn_struct_rwlock);
			mutex_enter(&db->db_mtx);
			db->db_parent = parent;
		}
		db->db_blkptr = (blkptr_t *)parent->db.db_data +
		    (db->db_blkid & ((1ULL << epbs) - 1));
		DBUF_VERIFY(db);
	}
}

static void
dbuf_sync_indirect(dbuf_dirty_record_t *dr, dmu_tx_t *tx)
{
	dmu_buf_impl_t *db = dr->dr_dbuf;
	dnode_t *dn;
	zio_t *zio;

	ASSERT(dmu_tx_is_syncing(tx));

	dprintf_dbuf_bp(db, db->db_blkptr, "blkptr=%p", db->db_blkptr);

	mutex_enter(&db->db_mtx);

	ASSERT(db->db_level > 0);
	DBUF_VERIFY(db);

	if (db->db_buf == NULL) {
		mutex_exit(&db->db_mtx);
		(void) dbuf_read(db, NULL, DB_RF_MUST_SUCCEED);
		mutex_enter(&db->db_mtx);
	}
	ASSERT3U(db->db_state, ==, DB_CACHED);
	ASSERT(db->db_buf != NULL);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	ASSERT3U(db->db.db_size, ==, 1<<dn->dn_phys->dn_indblkshift);
	dbuf_check_blkptr(dn, db);
	DB_DNODE_EXIT(db);

	db->db_data_pending = dr;

	mutex_exit(&db->db_mtx);
	dbuf_write(dr, db->db_buf, tx);

	zio = dr->dr_zio;
	mutex_enter(&dr->dt.di.dr_mtx);
	dbuf_sync_list(&dr->dt.di.dr_children, tx);
	ASSERT(list_head(&dr->dt.di.dr_children) == NULL);
	mutex_exit(&dr->dt.di.dr_mtx);
	zio_nowait(zio);
}

static void
dbuf_sync_leaf(dbuf_dirty_record_t *dr, dmu_tx_t *tx)
{
	arc_buf_t **datap = &dr->dt.dl.dr_data;
	dmu_buf_impl_t *db = dr->dr_dbuf;
	dnode_t *dn;
	objset_t *os;
	uint64_t txg = tx->tx_txg;

	ASSERT(dmu_tx_is_syncing(tx));

	dprintf_dbuf_bp(db, db->db_blkptr, "blkptr=%p", db->db_blkptr);

	mutex_enter(&db->db_mtx);
	/*
	 * To be synced, we must be dirtied.  But we
	 * might have been freed after the dirty.
	 */
	if (db->db_state == DB_UNCACHED) {
		/* This buffer has been freed since it was dirtied */
		ASSERT(db->db.db_data == NULL);
	} else if (db->db_state == DB_FILL) {
		/* This buffer was freed and is now being re-filled */
		ASSERT(db->db.db_data != dr->dt.dl.dr_data);
	} else {
		ASSERT(db->db_state == DB_CACHED || db->db_state == DB_NOFILL);
	}
	DBUF_VERIFY(db);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);

	if (db->db_blkid == DMU_SPILL_BLKID) {
		mutex_enter(&dn->dn_mtx);
		dn->dn_phys->dn_flags |= DNODE_FLAG_SPILL_BLKPTR;
		mutex_exit(&dn->dn_mtx);
	}

	/*
	 * If this is a bonus buffer, simply copy the bonus data into the
	 * dnode.  It will be written out when the dnode is synced (and it
	 * will be synced, since it must have been dirty for dbuf_sync to
	 * be called).
	 */
	if (db->db_blkid == DMU_BONUS_BLKID) {
		dbuf_dirty_record_t **drp;

		ASSERT(*datap != NULL);
		ASSERT3U(db->db_level, ==, 0);
		ASSERT3U(dn->dn_phys->dn_bonuslen, <=, DN_MAX_BONUSLEN);
		bcopy(*datap, DN_BONUS(dn->dn_phys), dn->dn_phys->dn_bonuslen);
		DB_DNODE_EXIT(db);

		if (*datap != db->db.db_data) {
			zio_buf_free(*datap, DN_MAX_BONUSLEN);
			arc_space_return(DN_MAX_BONUSLEN, ARC_SPACE_OTHER);
		}
		db->db_data_pending = NULL;
		drp = &db->db_last_dirty;
		while (*drp != dr)
			drp = &(*drp)->dr_next;
		ASSERT(dr->dr_next == NULL);
		ASSERT(dr->dr_dbuf == db);
		*drp = dr->dr_next;
		kmem_free(dr, sizeof (dbuf_dirty_record_t));
		ASSERT(db->db_dirtycnt > 0);
		db->db_dirtycnt -= 1;
		dbuf_rele_and_unlock(db, (void *)(uintptr_t)txg);
		return;
	}

	os = dn->dn_objset;

	/*
	 * This function may have dropped the db_mtx lock allowing a dmu_sync
	 * operation to sneak in. As a result, we need to ensure that we
	 * don't check the dr_override_state until we have returned from
	 * dbuf_check_blkptr.
	 */
	dbuf_check_blkptr(dn, db);

	/*
	 * If this buffer is in the middle of an immediate write,
	 * wait for the synchronous IO to complete.
	 */
	while (dr->dt.dl.dr_override_state == DR_IN_DMU_SYNC) {
		ASSERT(dn->dn_object != DMU_META_DNODE_OBJECT);
		cv_wait(&db->db_changed, &db->db_mtx);
		ASSERT(dr->dt.dl.dr_override_state != DR_NOT_OVERRIDDEN);
	}

	if (db->db_state != DB_NOFILL &&
	    dn->dn_object != DMU_META_DNODE_OBJECT &&
	    refcount_count(&db->db_holds) > 1 &&
	    dr->dt.dl.dr_override_state != DR_OVERRIDDEN &&
	    *datap == db->db_buf) {
		/*
		 * If this buffer is currently "in use" (i.e., there
		 * are active holds and db_data still references it),
		 * then make a copy before we start the write so that
		 * any modifications from the open txg will not leak
		 * into this write.
		 *
		 * NOTE: this copy does not need to be made for
		 * objects only modified in the syncing context (e.g.
		 * DNONE_DNODE blocks).
		 */
		int blksz = arc_buf_size(*datap);
		arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
		*datap = arc_buf_alloc(os->os_spa, blksz, db, type);
		bcopy(db->db.db_data, (*datap)->b_data, blksz);
	}
	db->db_data_pending = dr;

	mutex_exit(&db->db_mtx);

	dbuf_write(dr, *datap, tx);

	ASSERT(!list_link_active(&dr->dr_dirty_node));
	if (dn->dn_object == DMU_META_DNODE_OBJECT) {
		list_insert_tail(&dn->dn_dirty_records[txg&TXG_MASK], dr);
		DB_DNODE_EXIT(db);
	} else {
		/*
		 * Although zio_nowait() does not "wait for an IO", it does
		 * initiate the IO. If this is an empty write it seems plausible
		 * that the IO could actually be completed before the nowait
		 * returns. We need to DB_DNODE_EXIT() first in case
		 * zio_nowait() invalidates the dbuf.
		 */
		DB_DNODE_EXIT(db);
		zio_nowait(dr->dr_zio);
	}
}

void
dbuf_sync_list(list_t *list, dmu_tx_t *tx)
{
	dbuf_dirty_record_t *dr;

	while (dr = list_head(list)) {
		if (dr->dr_zio != NULL) {
			/*
			 * If we find an already initialized zio then we
			 * are processing the meta-dnode, and we have finished.
			 * The dbufs for all dnodes are put back on the list
			 * during processing, so that we can zio_wait()
			 * these IOs after initiating all child IOs.
			 */
			ASSERT3U(dr->dr_dbuf->db.db_object, ==,
			    DMU_META_DNODE_OBJECT);
			break;
		}
		list_remove(list, dr);
		if (dr->dr_dbuf->db_level > 0)
			dbuf_sync_indirect(dr, tx);
		else
			dbuf_sync_leaf(dr, tx);
	}
}

/* ARGSUSED */
static void
dbuf_write_ready(zio_t *zio, arc_buf_t *buf, void *vdb)
{
	dmu_buf_impl_t *db = vdb;
	dnode_t *dn;
	blkptr_t *bp = zio->io_bp;
	blkptr_t *bp_orig = &zio->io_bp_orig;
	spa_t *spa = zio->io_spa;
	int64_t delta;
	uint64_t fill = 0;
	int i;

	ASSERT(db->db_blkptr == bp);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	delta = bp_get_dsize_sync(spa, bp) - bp_get_dsize_sync(spa, bp_orig);
	dnode_diduse_space(dn, delta - zio->io_prev_space_delta);
	zio->io_prev_space_delta = delta;

	if (BP_IS_HOLE(bp)) {
		ASSERT(bp->blk_fill == 0);
		DB_DNODE_EXIT(db);
		return;
	}

	ASSERT((db->db_blkid != DMU_SPILL_BLKID &&
	    BP_GET_TYPE(bp) == dn->dn_type) ||
	    (db->db_blkid == DMU_SPILL_BLKID &&
	    BP_GET_TYPE(bp) == dn->dn_bonustype));
	ASSERT(BP_GET_LEVEL(bp) == db->db_level);

	mutex_enter(&db->db_mtx);

#ifdef ZFS_DEBUG
	if (db->db_blkid == DMU_SPILL_BLKID) {
		ASSERT(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR);
		ASSERT(!(BP_IS_HOLE(db->db_blkptr)) &&
		    db->db_blkptr == &dn->dn_phys->dn_spill);
	}
#endif

	if (db->db_level == 0) {
		mutex_enter(&dn->dn_mtx);
		if (db->db_blkid > dn->dn_phys->dn_maxblkid &&
		    db->db_blkid != DMU_SPILL_BLKID)
			dn->dn_phys->dn_maxblkid = db->db_blkid;
		mutex_exit(&dn->dn_mtx);

		if (dn->dn_type == DMU_OT_DNODE) {
			dnode_phys_t *dnp = db->db.db_data;
			for (i = db->db.db_size >> DNODE_SHIFT; i > 0;
			    i--, dnp++) {
				if (dnp->dn_type != DMU_OT_NONE)
					fill++;
			}
		} else {
			fill = 1;
		}
	} else {
		blkptr_t *ibp = db->db.db_data;
		ASSERT3U(db->db.db_size, ==, 1<<dn->dn_phys->dn_indblkshift);
		for (i = db->db.db_size >> SPA_BLKPTRSHIFT; i > 0; i--, ibp++) {
			if (BP_IS_HOLE(ibp))
				continue;
			fill += ibp->blk_fill;
		}
	}
	DB_DNODE_EXIT(db);

	bp->blk_fill = fill;

	mutex_exit(&db->db_mtx);
}

/* ARGSUSED */
static void
dbuf_write_done(zio_t *zio, arc_buf_t *buf, void *vdb)
{
	dmu_buf_impl_t *db = vdb;
	blkptr_t *bp = zio->io_bp;
	blkptr_t *bp_orig = &zio->io_bp_orig;
	uint64_t txg = zio->io_txg;
	dbuf_dirty_record_t **drp, *dr;

	ASSERT3U(zio->io_error, ==, 0);
	ASSERT(db->db_blkptr == bp);

	if (zio->io_flags & ZIO_FLAG_IO_REWRITE) {
		ASSERT(BP_EQUAL(bp, bp_orig));
	} else {
		objset_t *os;
		dsl_dataset_t *ds;
		dmu_tx_t *tx;

		DB_GET_OBJSET(&os, db);
		ds = os->os_dsl_dataset;
		tx = os->os_synctx;

		(void) dsl_dataset_block_kill(ds, bp_orig, tx, B_TRUE);
		dsl_dataset_block_born(ds, bp, tx);
	}

	mutex_enter(&db->db_mtx);

	DBUF_VERIFY(db);

	drp = &db->db_last_dirty;
	while ((dr = *drp) != db->db_data_pending)
		drp = &dr->dr_next;
	ASSERT(!list_link_active(&dr->dr_dirty_node));
	ASSERT(dr->dr_txg == txg);
	ASSERT(dr->dr_dbuf == db);
	ASSERT(dr->dr_next == NULL);
	*drp = dr->dr_next;

#ifdef ZFS_DEBUG
	if (db->db_blkid == DMU_SPILL_BLKID) {
		dnode_t *dn;

		DB_DNODE_ENTER(db);
		dn = DB_DNODE(db);
		ASSERT(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR);
		ASSERT(!(BP_IS_HOLE(db->db_blkptr)) &&
		    db->db_blkptr == &dn->dn_phys->dn_spill);
		DB_DNODE_EXIT(db);
	}
#endif

	if (db->db_level == 0) {
		ASSERT(db->db_blkid != DMU_BONUS_BLKID);
		ASSERT(dr->dt.dl.dr_override_state == DR_NOT_OVERRIDDEN);
		if (db->db_state != DB_NOFILL) {
			if (dr->dt.dl.dr_data != db->db_buf)
				VERIFY(arc_buf_remove_ref(dr->dt.dl.dr_data,
				    db) == 1);
			else if (!arc_released(db->db_buf))
				arc_set_callback(db->db_buf, dbuf_do_evict, db);
		}
	} else {
		dnode_t *dn;

		DB_DNODE_ENTER(db);
		dn = DB_DNODE(db);
		ASSERT(list_head(&dr->dt.di.dr_children) == NULL);
		ASSERT3U(db->db.db_size, ==, 1<<dn->dn_phys->dn_indblkshift);
		if (!BP_IS_HOLE(db->db_blkptr)) {
			int epbs =
			    dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT;
			ASSERT3U(BP_GET_LSIZE(db->db_blkptr), ==,
			    db->db.db_size);
			ASSERT3U(dn->dn_phys->dn_maxblkid
			    >> (db->db_level * epbs), >=, db->db_blkid);
			arc_set_callback(db->db_buf, dbuf_do_evict, db);
		}
		DB_DNODE_EXIT(db);
		mutex_destroy(&dr->dt.di.dr_mtx);
		list_destroy(&dr->dt.di.dr_children);
	}
	kmem_free(dr, sizeof (dbuf_dirty_record_t));

	cv_broadcast(&db->db_changed);
	ASSERT(db->db_dirtycnt > 0);
	db->db_dirtycnt -= 1;
	db->db_data_pending = NULL;
	dbuf_rele_and_unlock(db, (void *)(uintptr_t)txg);
}

static void
dbuf_write_nofill_ready(zio_t *zio)
{
	dbuf_write_ready(zio, NULL, zio->io_private);
}

static void
dbuf_write_nofill_done(zio_t *zio)
{
	dbuf_write_done(zio, NULL, zio->io_private);
}

static void
dbuf_write_override_ready(zio_t *zio)
{
	dbuf_dirty_record_t *dr = zio->io_private;
	dmu_buf_impl_t *db = dr->dr_dbuf;

	dbuf_write_ready(zio, NULL, db);
}

static void
dbuf_write_override_done(zio_t *zio)
{
	dbuf_dirty_record_t *dr = zio->io_private;
	dmu_buf_impl_t *db = dr->dr_dbuf;
	blkptr_t *obp = &dr->dt.dl.dr_overridden_by;

	mutex_enter(&db->db_mtx);
	if (!BP_EQUAL(zio->io_bp, obp)) {
		if (!BP_IS_HOLE(obp))
			dsl_free(spa_get_dsl(zio->io_spa), zio->io_txg, obp);
		arc_release(dr->dt.dl.dr_data, db);
	}
	mutex_exit(&db->db_mtx);

	dbuf_write_done(zio, NULL, db);
}

static void
dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx)
{
	dmu_buf_impl_t *db = dr->dr_dbuf;
	dnode_t *dn;
	objset_t *os;
	dmu_buf_impl_t *parent = db->db_parent;
	uint64_t txg = tx->tx_txg;
	zbookmark_t zb;
	zio_prop_t zp;
	zio_t *zio;
	int wp_flag = 0;

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	os = dn->dn_objset;

	if (db->db_state != DB_NOFILL) {
		if (db->db_level > 0 || dn->dn_type == DMU_OT_DNODE) {
			/*
			 * Private object buffers are released here rather
			 * than in dbuf_dirty() since they are only modified
			 * in the syncing context and we don't want the
			 * overhead of making multiple copies of the data.
			 */
			if (BP_IS_HOLE(db->db_blkptr)) {
				arc_buf_thaw(data);
			} else {
				dbuf_release_bp(db);
			}
		}
	}

	if (parent != dn->dn_dbuf) {
		ASSERT(parent && parent->db_data_pending);
		ASSERT(db->db_level == parent->db_level-1);
		ASSERT(arc_released(parent->db_buf));
		zio = parent->db_data_pending->dr_zio;
	} else {
		ASSERT((db->db_level == dn->dn_phys->dn_nlevels-1 &&
		    db->db_blkid != DMU_SPILL_BLKID) ||
		    (db->db_blkid == DMU_SPILL_BLKID && db->db_level == 0));
		if (db->db_blkid != DMU_SPILL_BLKID)
			ASSERT3P(db->db_blkptr, ==,
			    &dn->dn_phys->dn_blkptr[db->db_blkid]);
		zio = dn->dn_zio;
	}

	ASSERT(db->db_level == 0 || data == db->db_buf);
	ASSERT3U(db->db_blkptr->blk_birth, <=, txg);
	ASSERT(zio);

	SET_BOOKMARK(&zb, os->os_dsl_dataset ?
	    os->os_dsl_dataset->ds_object : DMU_META_OBJSET,
	    db->db.db_object, db->db_level, db->db_blkid);

	if (db->db_blkid == DMU_SPILL_BLKID)
		wp_flag = WP_SPILL;
	wp_flag |= (db->db_state == DB_NOFILL) ? WP_NOFILL : 0;

	dmu_write_policy(os, dn, db->db_level, wp_flag, &zp);
	DB_DNODE_EXIT(db);

	if (db->db_level == 0 && dr->dt.dl.dr_override_state == DR_OVERRIDDEN) {
		ASSERT(db->db_state != DB_NOFILL);
		dr->dr_zio = zio_write(zio, os->os_spa, txg,
		    db->db_blkptr, data->b_data, arc_buf_size(data), &zp,
		    dbuf_write_override_ready, dbuf_write_override_done, dr,
		    ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb);
		mutex_enter(&db->db_mtx);
		dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN;
		zio_write_override(dr->dr_zio, &dr->dt.dl.dr_overridden_by,
		    dr->dt.dl.dr_copies);
		mutex_exit(&db->db_mtx);
	} else if (db->db_state == DB_NOFILL) {
		ASSERT(zp.zp_checksum == ZIO_CHECKSUM_OFF);
		dr->dr_zio = zio_write(zio, os->os_spa, txg,
		    db->db_blkptr, NULL, db->db.db_size, &zp,
		    dbuf_write_nofill_ready, dbuf_write_nofill_done, db,
		    ZIO_PRIORITY_ASYNC_WRITE,
		    ZIO_FLAG_MUSTSUCCEED | ZIO_FLAG_NODATA, &zb);
	} else {
		ASSERT(arc_released(data));
		dr->dr_zio = arc_write(zio, os->os_spa, txg,
		    db->db_blkptr, data, DBUF_IS_L2CACHEABLE(db), &zp,
		    dbuf_write_ready, dbuf_write_done, db,
		    ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb);
	}
}