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
|
/*
* 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 2011 Nexenta Systems, Inc. All rights reserved.
* Copyright (c) 2012, 2015 by Delphix. All rights reserved.
* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
*/
#include <sys/zfs_context.h>
#include <sys/arc.h>
#include <sys/dmu.h>
#include <sys/dmu_send.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>
#include <sys/zfeature.h>
#include <sys/blkptr.h>
#include <sys/range_tree.h>
#include <sys/trace_dbuf.h>
#include <sys/callb.h>
#include <sys/abd.h>
struct dbuf_hold_impl_data {
/* Function arguments */
dnode_t *dh_dn;
uint8_t dh_level;
uint64_t dh_blkid;
boolean_t dh_fail_sparse;
boolean_t dh_fail_uncached;
void *dh_tag;
dmu_buf_impl_t **dh_dbp;
/* Local variables */
dmu_buf_impl_t *dh_db;
dmu_buf_impl_t *dh_parent;
blkptr_t *dh_bp;
int dh_err;
dbuf_dirty_record_t *dh_dr;
arc_buf_contents_t dh_type;
int dh_depth;
};
static void __dbuf_hold_impl_init(struct dbuf_hold_impl_data *dh,
dnode_t *dn, uint8_t level, uint64_t blkid, boolean_t fail_sparse,
boolean_t fail_uncached,
void *tag, dmu_buf_impl_t **dbp, int depth);
static int __dbuf_hold_impl(struct dbuf_hold_impl_data *dh);
uint_t zfs_dbuf_evict_key;
/*
* Number of times that zfs_free_range() took the slow path while doing
* a zfs receive. A nonzero value indicates a potential performance problem.
*/
uint64_t zfs_free_range_recv_miss;
static boolean_t 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);
#ifndef __lint
extern inline void dmu_buf_init_user(dmu_buf_user_t *dbu,
dmu_buf_evict_func_t *evict_func, dmu_buf_t **clear_on_evict_dbufp);
#endif /* ! __lint */
/*
* Global data structures and functions for the dbuf cache.
*/
static kmem_cache_t *dbuf_kmem_cache;
static taskq_t *dbu_evict_taskq;
static kthread_t *dbuf_cache_evict_thread;
static kmutex_t dbuf_evict_lock;
static kcondvar_t dbuf_evict_cv;
static boolean_t dbuf_evict_thread_exit;
/*
* LRU cache of dbufs. The dbuf cache maintains a list of dbufs that
* are not currently held but have been recently released. These dbufs
* are not eligible for arc eviction until they are aged out of the cache.
* Dbufs are added to the dbuf cache once the last hold is released. If a
* dbuf is later accessed and still exists in the dbuf cache, then it will
* be removed from the cache and later re-added to the head of the cache.
* Dbufs that are aged out of the cache will be immediately destroyed and
* become eligible for arc eviction.
*/
static multilist_t dbuf_cache;
static refcount_t dbuf_cache_size;
unsigned long dbuf_cache_max_bytes = 100 * 1024 * 1024;
/* Cap the size of the dbuf cache to log2 fraction of arc size. */
int dbuf_cache_max_shift = 5;
/*
* The dbuf cache uses a three-stage eviction policy:
* - A low water marker designates when the dbuf eviction thread
* should stop evicting from the dbuf cache.
* - When we reach the maximum size (aka mid water mark), we
* signal the eviction thread to run.
* - The high water mark indicates when the eviction thread
* is unable to keep up with the incoming load and eviction must
* happen in the context of the calling thread.
*
* The dbuf cache:
* (max size)
* low water mid water hi water
* +----------------------------------------+----------+----------+
* | | | |
* | | | |
* | | | |
* | | | |
* +----------------------------------------+----------+----------+
* stop signal evict
* evicting eviction directly
* thread
*
* The high and low water marks indicate the operating range for the eviction
* thread. The low water mark is, by default, 90% of the total size of the
* cache and the high water mark is at 110% (both of these percentages can be
* changed by setting dbuf_cache_lowater_pct and dbuf_cache_hiwater_pct,
* respectively). The eviction thread will try to ensure that the cache remains
* within this range by waking up every second and checking if the cache is
* above the low water mark. The thread can also be woken up by callers adding
* elements into the cache if the cache is larger than the mid water (i.e max
* cache size). Once the eviction thread is woken up and eviction is required,
* it will continue evicting buffers until it's able to reduce the cache size
* to the low water mark. If the cache size continues to grow and hits the high
* water mark, then callers adding elements to the cache will begin to evict
* directly from the cache until the cache is no longer above the high water
* mark.
*/
/*
* The percentage above and below the maximum cache size.
*/
uint_t dbuf_cache_hiwater_pct = 10;
uint_t dbuf_cache_lowater_pct = 10;
/* 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);
multilist_link_init(&db->db_cache_link);
refcount_create(&db->db_holds);
multilist_link_init(&db->db_cache_link);
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);
ASSERT(!multilist_link_active(&db->db_cache_link));
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_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(objset_t *os, uint64_t obj, uint8_t level, uint64_t blkid)
{
dbuf_hash_table_t *h = &dbuf_hash_table;
uint64_t hv;
uint64_t idx;
dmu_buf_impl_t *db;
hv = dbuf_hash(os, obj, level, blkid);
idx = hv & h->hash_table_mask;
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);
}
static dmu_buf_impl_t *
dbuf_find_bonus(objset_t *os, uint64_t object)
{
dnode_t *dn;
dmu_buf_impl_t *db = NULL;
if (dnode_hold(os, object, FTAG, &dn) == 0) {
rw_enter(&dn->dn_struct_rwlock, RW_READER);
if (dn->dn_bonus != NULL) {
db = dn->dn_bonus;
mutex_enter(&db->db_mtx);
}
rw_exit(&dn->dn_struct_rwlock);
dnode_rele(dn, FTAG);
}
return (db);
}
/*
* 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, hv, idx;
dmu_buf_impl_t *dbf;
blkid = db->db_blkid;
hv = dbuf_hash(os, obj, level, blkid);
idx = hv & h->hash_table_mask;
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_inc_64(&dbuf_hash_count);
return (NULL);
}
/*
* Remove an entry from the hash table. It must be in the EVICTING state.
*/
static void
dbuf_hash_remove(dmu_buf_impl_t *db)
{
dbuf_hash_table_t *h = &dbuf_hash_table;
uint64_t hv, idx;
dmu_buf_impl_t *dbf, **dbp;
hv = dbuf_hash(db->db_objset, db->db.db_object,
db->db_level, db->db_blkid);
idx = hv & h->hash_table_mask;
/*
* We mustn't hold db_mtx to maintain 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_dec_64(&dbuf_hash_count);
}
typedef enum {
DBVU_EVICTING,
DBVU_NOT_EVICTING
} dbvu_verify_type_t;
static void
dbuf_verify_user(dmu_buf_impl_t *db, dbvu_verify_type_t verify_type)
{
#ifdef ZFS_DEBUG
int64_t holds;
if (db->db_user == NULL)
return;
/* Only data blocks support the attachment of user data. */
ASSERT(db->db_level == 0);
/* Clients must resolve a dbuf before attaching user data. */
ASSERT(db->db.db_data != NULL);
ASSERT3U(db->db_state, ==, DB_CACHED);
holds = refcount_count(&db->db_holds);
if (verify_type == DBVU_EVICTING) {
/*
* Immediate eviction occurs when holds == dirtycnt.
* For normal eviction buffers, holds is zero on
* eviction, except when dbuf_fix_old_data() calls
* dbuf_clear_data(). However, the hold count can grow
* during eviction even though db_mtx is held (see
* dmu_bonus_hold() for an example), so we can only
* test the generic invariant that holds >= dirtycnt.
*/
ASSERT3U(holds, >=, db->db_dirtycnt);
} else {
if (db->db_user_immediate_evict == TRUE)
ASSERT3U(holds, >=, db->db_dirtycnt);
else
ASSERT3U(holds, >, 0);
}
#endif
}
static void
dbuf_evict_user(dmu_buf_impl_t *db)
{
dmu_buf_user_t *dbu = db->db_user;
ASSERT(MUTEX_HELD(&db->db_mtx));
if (dbu == NULL)
return;
dbuf_verify_user(db, DBVU_EVICTING);
db->db_user = NULL;
#ifdef ZFS_DEBUG
if (dbu->dbu_clear_on_evict_dbufp != NULL)
*dbu->dbu_clear_on_evict_dbufp = NULL;
#endif
/*
* Invoke the callback from a taskq to avoid lock order reversals
* and limit stack depth.
*/
taskq_dispatch_ent(dbu_evict_taskq, dbu->dbu_evict_func, dbu, 0,
&dbu->dbu_tqent);
}
boolean_t
dbuf_is_metadata(dmu_buf_impl_t *db)
{
/*
* Consider indirect blocks and spill blocks to be meta data.
*/
if (db->db_level > 0 || db->db_blkid == DMU_SPILL_BLKID) {
return (B_TRUE);
} else {
boolean_t is_metadata;
DB_DNODE_ENTER(db);
is_metadata = DMU_OT_IS_METADATA(DB_DNODE(db)->dn_type);
DB_DNODE_EXIT(db);
return (is_metadata);
}
}
/*
* This function *must* return indices evenly distributed between all
* sublists of the multilist. This is needed due to how the dbuf eviction
* code is laid out; dbuf_evict_thread() assumes dbufs 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
dbuf_cache_multilist_index_func(multilist_t *ml, void *obj)
{
dmu_buf_impl_t *db = obj;
/*
* The assumption here, is the hash value for a given
* dmu_buf_impl_t will remain constant throughout it's lifetime
* (i.e. it's objset, object, level and blkid fields don't change).
* Thus, we don't need to store the dbuf'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 (dbuf_hash(db->db_objset, db->db.db_object,
db->db_level, db->db_blkid) %
multilist_get_num_sublists(ml));
}
static inline boolean_t
dbuf_cache_above_hiwater(void)
{
uint64_t dbuf_cache_hiwater_bytes =
(dbuf_cache_max_bytes * dbuf_cache_hiwater_pct) / 100;
return (refcount_count(&dbuf_cache_size) >
dbuf_cache_max_bytes + dbuf_cache_hiwater_bytes);
}
static inline boolean_t
dbuf_cache_above_lowater(void)
{
uint64_t dbuf_cache_lowater_bytes =
(dbuf_cache_max_bytes * dbuf_cache_lowater_pct) / 100;
return (refcount_count(&dbuf_cache_size) >
dbuf_cache_max_bytes - dbuf_cache_lowater_bytes);
}
/*
* Evict the oldest eligible dbuf from the dbuf cache.
*/
static void
dbuf_evict_one(void)
{
int idx = multilist_get_random_index(&dbuf_cache);
multilist_sublist_t *mls = multilist_sublist_lock(&dbuf_cache, idx);
dmu_buf_impl_t *db;
ASSERT(!MUTEX_HELD(&dbuf_evict_lock));
/*
* Set the thread's tsd to indicate that it's processing evictions.
* Once a thread stops evicting from the dbuf cache it will
* reset its tsd to NULL.
*/
ASSERT3P(tsd_get(zfs_dbuf_evict_key), ==, NULL);
(void) tsd_set(zfs_dbuf_evict_key, (void *)B_TRUE);
db = multilist_sublist_tail(mls);
while (db != NULL && mutex_tryenter(&db->db_mtx) == 0) {
db = multilist_sublist_prev(mls, db);
}
DTRACE_PROBE2(dbuf__evict__one, dmu_buf_impl_t *, db,
multilist_sublist_t *, mls);
if (db != NULL) {
multilist_sublist_remove(mls, db);
multilist_sublist_unlock(mls);
(void) refcount_remove_many(&dbuf_cache_size,
db->db.db_size, db);
dbuf_destroy(db);
} else {
multilist_sublist_unlock(mls);
}
(void) tsd_set(zfs_dbuf_evict_key, NULL);
}
/*
* The dbuf evict thread is responsible for aging out dbufs from the
* cache. Once the cache has reached it's maximum size, dbufs are removed
* and destroyed. The eviction thread will continue running until the size
* of the dbuf cache is at or below the maximum size. Once the dbuf is aged
* out of the cache it is destroyed and becomes eligible for arc eviction.
*/
static void
dbuf_evict_thread(void)
{
callb_cpr_t cpr;
CALLB_CPR_INIT(&cpr, &dbuf_evict_lock, callb_generic_cpr, FTAG);
mutex_enter(&dbuf_evict_lock);
while (!dbuf_evict_thread_exit) {
while (!dbuf_cache_above_lowater() && !dbuf_evict_thread_exit) {
CALLB_CPR_SAFE_BEGIN(&cpr);
(void) cv_timedwait_sig_hires(&dbuf_evict_cv,
&dbuf_evict_lock, SEC2NSEC(1), MSEC2NSEC(1), 0);
CALLB_CPR_SAFE_END(&cpr, &dbuf_evict_lock);
}
mutex_exit(&dbuf_evict_lock);
/*
* Keep evicting as long as we're above the low water mark
* for the cache. We do this without holding the locks to
* minimize lock contention.
*/
while (dbuf_cache_above_lowater() && !dbuf_evict_thread_exit) {
dbuf_evict_one();
}
mutex_enter(&dbuf_evict_lock);
}
dbuf_evict_thread_exit = B_FALSE;
cv_broadcast(&dbuf_evict_cv);
CALLB_CPR_EXIT(&cpr); /* drops dbuf_evict_lock */
thread_exit();
}
/*
* Wake up the dbuf eviction thread if the dbuf cache is at its max size.
* If the dbuf cache is at its high water mark, then evict a dbuf from the
* dbuf cache using the callers context.
*/
static void
dbuf_evict_notify(void)
{
/*
* We use thread specific data to track when a thread has
* started processing evictions. This allows us to avoid deeply
* nested stacks that would have a call flow similar to this:
*
* dbuf_rele()-->dbuf_rele_and_unlock()-->dbuf_evict_notify()
* ^ |
* | |
* +-----dbuf_destroy()<--dbuf_evict_one()<--------+
*
* The dbuf_eviction_thread will always have its tsd set until
* that thread exits. All other threads will only set their tsd
* if they are participating in the eviction process. This only
* happens if the eviction thread is unable to process evictions
* fast enough. To keep the dbuf cache size in check, other threads
* can evict from the dbuf cache directly. Those threads will set
* their tsd values so that we ensure that they only evict one dbuf
* from the dbuf cache.
*/
if (tsd_get(zfs_dbuf_evict_key) != NULL)
return;
if (refcount_count(&dbuf_cache_size) > dbuf_cache_max_bytes) {
boolean_t evict_now = B_FALSE;
mutex_enter(&dbuf_evict_lock);
if (refcount_count(&dbuf_cache_size) > dbuf_cache_max_bytes) {
evict_now = dbuf_cache_above_hiwater();
cv_signal(&dbuf_evict_cv);
}
mutex_exit(&dbuf_evict_lock);
if (evict_now) {
dbuf_evict_one();
}
}
}
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 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 < physmem * PAGESIZE)
hsize <<= 1;
retry:
h->hash_table_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
*/
h->hash_table = vmem_zalloc(hsize * sizeof (void *), KM_SLEEP);
#else
h->hash_table = kmem_zalloc(hsize * sizeof (void *), KM_NOSLEEP);
#endif
if (h->hash_table == NULL) {
/* XXX - we should really return an error instead of assert */
ASSERT(hsize > (1ULL << 10));
hsize >>= 1;
goto retry;
}
dbuf_kmem_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);
dbuf_stats_init(h);
/*
* Setup the parameters for the dbuf cache. We cap the size of the
* dbuf cache to 1/32nd (default) of the size of the ARC.
*/
dbuf_cache_max_bytes = MIN(dbuf_cache_max_bytes,
arc_max_bytes() >> dbuf_cache_max_shift);
/*
* All entries are queued via taskq_dispatch_ent(), so min/maxalloc
* configuration is not required.
*/
dbu_evict_taskq = taskq_create("dbu_evict", 1, defclsyspri, 0, 0, 0);
multilist_create(&dbuf_cache, sizeof (dmu_buf_impl_t),
offsetof(dmu_buf_impl_t, db_cache_link),
zfs_arc_num_sublists_per_state,
dbuf_cache_multilist_index_func);
refcount_create(&dbuf_cache_size);
tsd_create(&zfs_dbuf_evict_key, NULL);
dbuf_evict_thread_exit = B_FALSE;
mutex_init(&dbuf_evict_lock, NULL, MUTEX_DEFAULT, NULL);
cv_init(&dbuf_evict_cv, NULL, CV_DEFAULT, NULL);
dbuf_cache_evict_thread = thread_create(NULL, 0, dbuf_evict_thread,
NULL, 0, &p0, TS_RUN, minclsyspri);
}
void
dbuf_fini(void)
{
dbuf_hash_table_t *h = &dbuf_hash_table;
int i;
dbuf_stats_destroy();
for (i = 0; i < DBUF_MUTEXES; i++)
mutex_destroy(&h->hash_mutexes[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(h->hash_table, (h->hash_table_mask + 1) * sizeof (void *));
#else
kmem_free(h->hash_table, (h->hash_table_mask + 1) * sizeof (void *));
#endif
kmem_cache_destroy(dbuf_kmem_cache);
taskq_destroy(dbu_evict_taskq);
mutex_enter(&dbuf_evict_lock);
dbuf_evict_thread_exit = B_TRUE;
while (dbuf_evict_thread_exit) {
cv_signal(&dbuf_evict_cv);
cv_wait(&dbuf_evict_cv, &dbuf_evict_lock);
}
mutex_exit(&dbuf_evict_lock);
tsd_destroy(&zfs_dbuf_evict_key);
mutex_destroy(&dbuf_evict_lock);
cv_destroy(&dbuf_evict_cv);
refcount_destroy(&dbuf_cache_size);
multilist_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 ||
!avl_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);
ASSERT0(db->db.db_offset);
} 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 */
ASSERTV(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.
*
* There is an exception to this rule for indirect blocks; in
* this case, if the indirect block is a hole, we fill in a few
* fields on each of the child blocks (importantly, birth time)
* to prevent hole birth times from being lost when you
* partially fill in a hole.
*/
if (db->db_dirtycnt == 0) {
if (db->db_level == 0) {
uint64_t *buf = db->db.db_data;
int i;
for (i = 0; i < db->db.db_size >> 3; i++) {
ASSERT(buf[i] == 0);
}
} else {
int i;
blkptr_t *bps = db->db.db_data;
ASSERT3U(1 << DB_DNODE(db)->dn_indblkshift, ==,
db->db.db_size);
/*
* We want to verify that all the blkptrs in the
* indirect block are holes, but we may have
* automatically set up a few fields for them.
* We iterate through each blkptr and verify
* they only have those fields set.
*/
for (i = 0;
i < db->db.db_size / sizeof (blkptr_t);
i++) {
blkptr_t *bp = &bps[i];
ASSERT(ZIO_CHECKSUM_IS_ZERO(
&bp->blk_cksum));
ASSERT(
DVA_IS_EMPTY(&bp->blk_dva[0]) &&
DVA_IS_EMPTY(&bp->blk_dva[1]) &&
DVA_IS_EMPTY(&bp->blk_dva[2]));
ASSERT0(bp->blk_fill);
ASSERT0(bp->blk_pad[0]);
ASSERT0(bp->blk_pad[1]);
ASSERT(!BP_IS_EMBEDDED(bp));
ASSERT(BP_IS_HOLE(bp));
ASSERT0(bp->blk_phys_birth);
}
}
}
}
DB_DNODE_EXIT(db);
}
#endif
static void
dbuf_clear_data(dmu_buf_impl_t *db)
{
ASSERT(MUTEX_HELD(&db->db_mtx));
dbuf_evict_user(db);
ASSERT3P(db->db_buf, ==, NULL);
db->db.db_data = NULL;
if (db->db_state != DB_NOFILL)
db->db_state = DB_UNCACHED;
}
static void
dbuf_set_data(dmu_buf_impl_t *db, arc_buf_t *buf)
{
ASSERT(MUTEX_HELD(&db->db_mtx));
ASSERT(buf != NULL);
db->db_buf = buf;
ASSERT(buf->b_data != NULL);
db->db.db_data = buf->b_data;
}
/*
* 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;
ASSERT(db->db_blkid != DMU_BONUS_BLKID);
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 = db->db_objset->os_spa;
mutex_exit(&db->db_mtx);
abuf = arc_loan_buf(spa, B_FALSE, blksz);
bcopy(db->db.db_data, abuf->b_data, blksz);
} else {
abuf = db->db_buf;
arc_loan_inuse_buf(abuf, db);
db->db_buf = NULL;
dbuf_clear_data(db);
mutex_exit(&db->db_mtx);
}
return (abuf);
}
/*
* Calculate which level n block references the data at the level 0 offset
* provided.
*/
uint64_t
dbuf_whichblock(const dnode_t *dn, const int64_t level, const uint64_t offset)
{
if (dn->dn_datablkshift != 0 && dn->dn_indblkshift != 0) {
/*
* The level n blkid is equal to the level 0 blkid divided by
* the number of level 0s in a level n block.
*
* The level 0 blkid is offset >> datablkshift =
* offset / 2^datablkshift.
*
* The number of level 0s in a level n is the number of block
* pointers in an indirect block, raised to the power of level.
* This is 2^(indblkshift - SPA_BLKPTRSHIFT)^level =
* 2^(level*(indblkshift - SPA_BLKPTRSHIFT)).
*
* Thus, the level n blkid is: offset /
* ((2^datablkshift)*(2^(level*(indblkshift - SPA_BLKPTRSHIFT)))
* = offset / 2^(datablkshift + level *
* (indblkshift - SPA_BLKPTRSHIFT))
* = offset >> (datablkshift + level *
* (indblkshift - SPA_BLKPTRSHIFT))
*/
const unsigned exp = dn->dn_datablkshift +
level * (dn->dn_indblkshift - SPA_BLKPTRSHIFT);
if (exp >= 8 * sizeof (offset)) {
/* This only happens on the highest indirection level */
ASSERT3U(level, ==, dn->dn_nlevels - 1);
return (0);
}
ASSERT3U(exp, <, 8 * sizeof (offset));
return (offset >> exp);
} 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);
arc_buf_destroy(buf, db);
db->db_state = DB_UNCACHED;
}
cv_broadcast(&db->db_changed);
dbuf_rele_and_unlock(db, NULL);
}
static int
dbuf_read_impl(dmu_buf_impl_t *db, zio_t *zio, uint32_t flags)
{
dnode_t *dn;
zbookmark_phys_t zb;
uint32_t aflags = ARC_FLAG_NOWAIT;
int err;
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) {
/*
* The bonus length stored in the dnode may be less than
* the maximum available space in the bonus buffer.
*/
int bonuslen = MIN(dn->dn_bonuslen, dn->dn_phys->dn_bonuslen);
int max_bonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots);
ASSERT3U(bonuslen, <=, db->db.db_size);
db->db.db_data = kmem_alloc(max_bonuslen, KM_SLEEP);
arc_space_consume(max_bonuslen, ARC_SPACE_BONUS);
if (bonuslen < max_bonuslen)
bzero(db->db.db_data, max_bonuslen);
if (bonuslen)
bcopy(DN_BONUS(dn->dn_phys), db->db.db_data, bonuslen);
DB_DNODE_EXIT(db);
db->db_state = DB_CACHED;
mutex_exit(&db->db_mtx);
return (0);
}
/*
* 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_alloc_buf(db->db_objset->os_spa, db, type,
db->db.db_size));
bzero(db->db.db_data, db->db.db_size);
if (db->db_blkptr != NULL && db->db_level > 0 &&
BP_IS_HOLE(db->db_blkptr) &&
db->db_blkptr->blk_birth != 0) {
blkptr_t *bps = db->db.db_data;
int i;
for (i = 0; i < ((1 <<
DB_DNODE(db)->dn_indblkshift) / sizeof (blkptr_t));
i++) {
blkptr_t *bp = &bps[i];
ASSERT3U(BP_GET_LSIZE(db->db_blkptr), ==,
1 << dn->dn_indblkshift);
BP_SET_LSIZE(bp,
BP_GET_LEVEL(db->db_blkptr) == 1 ?
dn->dn_datablksz :
BP_GET_LSIZE(db->db_blkptr));
BP_SET_TYPE(bp, BP_GET_TYPE(db->db_blkptr));
BP_SET_LEVEL(bp,
BP_GET_LEVEL(db->db_blkptr) - 1);
BP_SET_BIRTH(bp, db->db_blkptr->blk_birth, 0);
}
}
DB_DNODE_EXIT(db);
db->db_state = DB_CACHED;
mutex_exit(&db->db_mtx);
return (0);
}
DB_DNODE_EXIT(db);
db->db_state = DB_READ;
mutex_exit(&db->db_mtx);
if (DBUF_IS_L2CACHEABLE(db))
aflags |= ARC_FLAG_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);
err = arc_read(zio, db->db_objset->os_spa, db->db_blkptr,
dbuf_read_done, db, ZIO_PRIORITY_SYNC_READ,
(flags & DB_RF_CANFAIL) ? ZIO_FLAG_CANFAIL : ZIO_FLAG_MUSTSUCCEED,
&aflags, &zb);
return (err);
}
/*
* 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 access them in
* the current active group.
*
* This function is used in three places: when we are dirtying a buffer for the
* first time in a txg, when we are freeing a range in a dnode that includes
* this buffer, and when we are accessing a buffer which was received compressed
* and later referenced in a WRITE_BYREF record.
*
* 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) {
dnode_t *dn = DB_DNODE(db);
int bonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots);
dr->dt.dl.dr_data = kmem_alloc(bonuslen, KM_SLEEP);
arc_space_consume(bonuslen, ARC_SPACE_BONUS);
bcopy(db->db.db_data, dr->dt.dl.dr_data, bonuslen);
} else if (refcount_count(&db->db_holds) > db->db_dirtycnt) {
int size = arc_buf_size(db->db_buf);
arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
spa_t *spa = db->db_objset->os_spa;
enum zio_compress compress_type =
arc_get_compression(db->db_buf);
if (compress_type == ZIO_COMPRESS_OFF) {
dr->dt.dl.dr_data = arc_alloc_buf(spa, db, type, size);
} else {
ASSERT3U(type, ==, ARC_BUFC_DATA);
dr->dt.dl.dr_data = arc_alloc_compressed_buf(spa, db,
size, arc_buf_lsize(db->db_buf), compress_type);
}
bcopy(db->db.db_data, dr->dt.dl.dr_data->b_data, size);
} else {
db->db_buf = NULL;
dbuf_clear_data(db);
}
}
int
dbuf_read(dmu_buf_impl_t *db, zio_t *zio, uint32_t flags)
{
int err = 0;
boolean_t havepzio = (zio != NULL);
boolean_t 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 (SET_ERROR(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) {
/*
* If the arc buf is compressed, we need to decompress it to
* read the data. This could happen during the "zfs receive" of
* a stream which is compressed and deduplicated.
*/
if (db->db_buf != NULL &&
arc_get_compression(db->db_buf) != ZIO_COMPRESS_OFF) {
dbuf_fix_old_data(db,
spa_syncing_txg(dmu_objset_spa(db->db_objset)));
err = arc_decompress(db->db_buf);
dbuf_set_data(db, db->db_buf);
}
mutex_exit(&db->db_mtx);
if (prefetch)
dmu_zfetch(&dn->dn_zfetch, db->db_blkid, 1, B_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 &&
db->db_blkptr != NULL && !BP_IS_HOLE(db->db_blkptr))
zio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL);
err = dbuf_read_impl(db, zio, flags);
/* dbuf_read_impl has dropped db_mtx for us */
if (!err && prefetch)
dmu_zfetch(&dn->dn_zfetch, db->db_blkid, 1, B_TRUE);
if ((flags & DB_RF_HAVESTRUCT) == 0)
rw_exit(&dn->dn_struct_rwlock);
DB_DNODE_EXIT(db);
if (!err && !havepzio && zio != NULL)
err = zio_wait(zio);
} else {
/*
* Another reader came in while the dbuf was in flight
* between UNCACHED and CACHED. Either a writer will finish
* writing the buffer (sending the dbuf to CACHED) or the
* first reader's request will reach the read_done callback
* and send the dbuf to CACHED. Otherwise, a failure
* occurred and the dbuf went to UNCACHED.
*/
mutex_exit(&db->db_mtx);
if (prefetch)
dmu_zfetch(&dn->dn_zfetch, db->db_blkid, 1, B_TRUE);
if ((flags & DB_RF_HAVESTRUCT) == 0)
rw_exit(&dn->dn_struct_rwlock);
DB_DNODE_EXIT(db);
/* Skip the wait per the caller's request. */
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);
DTRACE_PROBE2(blocked__read, dmu_buf_impl_t *,
db, zio_t *, zio);
cv_wait(&db->db_changed, &db->db_mtx);
}
if (db->db_state == DB_UNCACHED)
err = SET_ERROR(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 = db->db_objset->os_spa;
ASSERT(db->db_buf == NULL);
ASSERT(db->db.db_data == NULL);
dbuf_set_data(db, arc_alloc_buf(spa, db, type, db->db.db_size));
db->db_state = DB_FILL;
} else if (db->db_state == DB_NOFILL) {
dbuf_clear_data(db);
} else {
ASSERT3U(db->db_state, ==, DB_CACHED);
}
mutex_exit(&db->db_mtx);
}
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) && !dr->dt.dl.dr_nopwrite)
zio_free(db->db_objset->os_spa, txg, bp);
dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN;
dr->dt.dl.dr_nopwrite = B_FALSE;
/*
* 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.
*
* This is a no-op if the dataset is in the middle of an incremental
* receive; see comment below for details.
*/
void
dbuf_free_range(dnode_t *dn, uint64_t start_blkid, uint64_t end_blkid,
dmu_tx_t *tx)
{
dmu_buf_impl_t *db_search;
dmu_buf_impl_t *db, *db_next;
uint64_t txg = tx->tx_txg;
avl_index_t where;
boolean_t freespill =
(start_blkid == DMU_SPILL_BLKID || end_blkid == DMU_SPILL_BLKID);
if (end_blkid > dn->dn_maxblkid && !freespill)
end_blkid = dn->dn_maxblkid;
dprintf_dnode(dn, "start=%llu end=%llu\n", start_blkid, end_blkid);
db_search = kmem_alloc(sizeof (dmu_buf_impl_t), KM_SLEEP);
db_search->db_level = 0;
db_search->db_blkid = start_blkid;
db_search->db_state = DB_SEARCH;
mutex_enter(&dn->dn_dbufs_mtx);
if (start_blkid >= dn->dn_unlisted_l0_blkid && !freespill) {
/* There can't be any dbufs in this range; no need to search. */
#ifdef DEBUG
db = avl_find(&dn->dn_dbufs, db_search, &where);
ASSERT3P(db, ==, NULL);
db = avl_nearest(&dn->dn_dbufs, where, AVL_AFTER);
ASSERT(db == NULL || db->db_level > 0);
#endif
goto out;
} else if (dmu_objset_is_receiving(dn->dn_objset)) {
/*
* If we are receiving, we expect there to be no dbufs in
* the range to be freed, because receive modifies each
* block at most once, and in offset order. If this is
* not the case, it can lead to performance problems,
* so note that we unexpectedly took the slow path.
*/
atomic_inc_64(&zfs_free_range_recv_miss);
}
db = avl_find(&dn->dn_dbufs, db_search, &where);
ASSERT3P(db, ==, NULL);
db = avl_nearest(&dn->dn_dbufs, where, AVL_AFTER);
for (; db != NULL; db = db_next) {
db_next = AVL_NEXT(&dn->dn_dbufs, db);
ASSERT(db->db_blkid != DMU_BONUS_BLKID);
if (db->db_level != 0 || db->db_blkid > end_blkid) {
break;
}
ASSERT3U(db->db_blkid, >=, start_blkid);
/* found a level 0 buffer in the range */
mutex_enter(&db->db_mtx);
if (dbuf_undirty(db, tx)) {
/* mutex has been dropped and dbuf destroyed */
continue;
}
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_destroy(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);
}
out:
kmem_free(db_search, sizeof (dmu_buf_impl_t));
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.
*
* This logic ensures that only block births for
* filled blocks are considered.
*/
ASSERT(MUTEX_HELD(&db->db_mtx));
if (db->db_last_dirty && (db->db_blkptr == NULL ||
!BP_IS_HOLE(db->db_blkptr))) {
birth_txg = db->db_last_dirty->dr_txg;
} else if (db->db_blkptr != NULL && !BP_IS_HOLE(db->db_blkptr)) {
birth_txg = db->db_blkptr->blk_birth;
}
/*
* If this block don't exist or is in a snapshot, it 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 != 0)
return (ds == NULL ||
dsl_dataset_block_freeable(ds, NULL, birth_txg));
else
return (B_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 dmu_buf_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
*/
dmu_buf_will_dirty(&db->db, tx);
/* create the data buffer for the new block */
buf = arc_alloc_buf(dn->dn_objset->os_spa, db, type, size);
/* 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);
arc_buf_destroy(obuf, db);
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)
{
ASSERTV(objset_t *os = db->db_objset);
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));
(void) arc_release(db->db_buf, db);
}
/*
* We already have a dirty record for this TXG, and we are being
* dirtied again.
*/
static void
dbuf_redirty(dbuf_dirty_record_t *dr)
{
dmu_buf_impl_t *db = dr->dr_dbuf;
ASSERT(MUTEX_HELD(&db->db_mtx));
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) {
/* Already released on initial dirty, so just thaw. */
ASSERT(arc_released(db->db_buf));
arc_buf_thaw(db->db_buf);
}
}
}
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);
dbuf_redirty(dr);
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);
list_link_init(&dr->dr_dirty_node);
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_NOLOCKDEP, NULL);
list_create(&dr->dt.di.dr_children,
sizeof (dbuf_dirty_record_t),
offsetof(dbuf_dirty_record_t, dr_dirty_node));
}
if (db->db_blkid != DMU_BONUS_BLKID && os->os_dsl_dataset != NULL)
dr->dr_accounted = db->db.db_size;
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);
if (dn->dn_free_ranges[txgoff] != NULL) {
range_tree_clear(dn->dn_free_ranges[txgoff],
db->db_blkid, 1);
}
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);
}
/*
* The dn_struct_rwlock prevents db_blkptr from changing
* due to a write from syncing context completing
* while we are running, so we want to acquire it before
* looking at db_blkptr.
*/
if (!RW_WRITE_HELD(&dn->dn_struct_rwlock)) {
rw_enter(&dn->dn_struct_rwlock, RW_READER);
drop_struct_lock = TRUE;
}
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 (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);
/*
* Since we've dropped the mutex, it's possible that
* dbuf_undirty() might have changed this out from under us.
*/
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);
}
/*
* Undirty a buffer in the transaction group referenced by the given
* transaction. Return whether this evicted the dbuf.
*/
static boolean_t
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);
/*
* Due to our use of dn_nlevels below, this can only be called
* in open context, unless we are operating on the MOS.
* From syncing context, dn_nlevels may be different from the
* dn_nlevels used when dbuf was dirtied.
*/
ASSERT(db->db_objset ==
dmu_objset_pool(db->db_objset)->dp_meta_objset ||
txg != spa_syncing_txg(dmu_objset_spa(db->db_objset)));
ASSERT(db->db_blkid != DMU_BONUS_BLKID);
ASSERT0(db->db_level);
ASSERT(MUTEX_HELD(&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)
return (B_FALSE);
ASSERT(dr->dr_txg == txg);
ASSERT(dr->dr_dbuf == db);
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
dprintf_dbuf(db, "size=%llx\n", (u_longlong_t)db->db.db_size);
ASSERT(db->db.db_size != 0);
dsl_pool_undirty_space(dmu_objset_pool(dn->dn_objset),
dr->dr_accounted, txg);
*drp = dr->dr_next;
/*
* Note that there are three places in dbuf_dirty()
* where this dirty record may be put on a list.
* Make sure to do a list_remove corresponding to
* every one of those list_insert calls.
*/
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_blkid == DMU_SPILL_BLKID ||
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_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)
arc_buf_destroy(dr->dt.dl.dr_data, db);
}
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) {
ASSERT(db->db_state == DB_NOFILL || arc_released(db->db_buf));
dbuf_destroy(db);
return (B_TRUE);
}
return (B_FALSE);
}
void
dmu_buf_will_dirty(dmu_buf_t *db_fake, dmu_tx_t *tx)
{
dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;
int rf = DB_RF_MUST_SUCCEED | DB_RF_NOPREFETCH;
dbuf_dirty_record_t *dr;
ASSERT(tx->tx_txg != 0);
ASSERT(!refcount_is_zero(&db->db_holds));
/*
* Quick check for dirtyness. For already dirty blocks, this
* reduces runtime of this function by >90%, and overall performance
* by 50% for some workloads (e.g. file deletion with indirect blocks
* cached).
*/
mutex_enter(&db->db_mtx);
for (dr = db->db_last_dirty;
dr != NULL && dr->dr_txg >= tx->tx_txg; dr = dr->dr_next) {
/*
* It's possible that it is already dirty but not cached,
* because there are some calls to dbuf_dirty() that don't
* go through dmu_buf_will_dirty().
*/
if (dr->dr_txg == tx->tx_txg && db->db_state == DB_CACHED) {
/* This dbuf is already dirty and cached. */
dbuf_redirty(dr);
mutex_exit(&db->db_mtx);
return;
}
}
mutex_exit(&db->db_mtx);
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);
}
void
dmu_buf_write_embedded(dmu_buf_t *dbuf, void *data,
bp_embedded_type_t etype, enum zio_compress comp,
int uncompressed_size, int compressed_size, int byteorder,
dmu_tx_t *tx)
{
dmu_buf_impl_t *db = (dmu_buf_impl_t *)dbuf;
struct dirty_leaf *dl;
dmu_object_type_t type;
if (etype == BP_EMBEDDED_TYPE_DATA) {
ASSERT(spa_feature_is_active(dmu_objset_spa(db->db_objset),
SPA_FEATURE_EMBEDDED_DATA));
}
DB_DNODE_ENTER(db);
type = DB_DNODE(db)->dn_type;
DB_DNODE_EXIT(db);
ASSERT0(db->db_level);
ASSERT(db->db_blkid != DMU_BONUS_BLKID);
dmu_buf_will_not_fill(dbuf, tx);
ASSERT3U(db->db_last_dirty->dr_txg, ==, tx->tx_txg);
dl = &db->db_last_dirty->dt.dl;
encode_embedded_bp_compressed(&dl->dr_overridden_by,
data, comp, uncompressed_size, compressed_size);
BPE_SET_ETYPE(&dl->dr_overridden_by, etype);
BP_SET_TYPE(&dl->dr_overridden_by, type);
BP_SET_LEVEL(&dl->dr_overridden_by, 0);
BP_SET_BYTEORDER(&dl->dr_overridden_by, byteorder);
dl->dr_override_state = DR_OVERRIDDEN;
dl->dr_overridden_by.blk_birth = db->db_last_dirty->dr_txg;
}
/*
* 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);
ASSERT3U(dbuf_is_metadata(db), ==, arc_is_metadata(buf));
ASSERT(buf != NULL);
ASSERT(arc_buf_lsize(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);
arc_buf_destroy(buf, db);
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;
arc_buf_destroy(db->db_buf, db);
} else if (dr == NULL || dr->dt.dl.dr_data != db->db_buf) {
arc_release(db->db_buf, db);
arc_buf_destroy(db->db_buf, db);
}
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);
dmu_buf_fill_done(&db->db, tx);
}
void
dbuf_destroy(dmu_buf_impl_t *db)
{
dnode_t *dn;
dmu_buf_impl_t *parent = db->db_parent;
dmu_buf_impl_t *dndb;
ASSERT(MUTEX_HELD(&db->db_mtx));
ASSERT(refcount_is_zero(&db->db_holds));
if (db->db_buf != NULL) {
arc_buf_destroy(db->db_buf, db);
db->db_buf = NULL;
}
if (db->db_blkid == DMU_BONUS_BLKID) {
int slots = DB_DNODE(db)->dn_num_slots;
int bonuslen = DN_SLOTS_TO_BONUSLEN(slots);
ASSERT(db->db.db_data != NULL);
kmem_free(db->db.db_data, bonuslen);
arc_space_return(bonuslen, ARC_SPACE_BONUS);
db->db_state = DB_UNCACHED;
}
dbuf_clear_data(db);
if (multilist_link_active(&db->db_cache_link)) {
multilist_remove(&dbuf_cache, db);
(void) refcount_remove_many(&dbuf_cache_size,
db->db.db_size, db);
}
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;
/*
* Now that db_state is DB_EVICTING, nobody else can find this via
* the hash table. We can now drop db_mtx, which allows us to
* acquire the dn_dbufs_mtx.
*/
mutex_exit(&db->db_mtx);
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
dndb = dn->dn_dbuf;
if (db->db_blkid != DMU_BONUS_BLKID) {
boolean_t needlock = !MUTEX_HELD(&dn->dn_dbufs_mtx);
if (needlock)
mutex_enter(&dn->dn_dbufs_mtx);
avl_remove(&dn->dn_dbufs, db);
atomic_dec_32(&dn->dn_dbufs_count);
membar_producer();
DB_DNODE_EXIT(db);
if (needlock)
mutex_exit(&dn->dn_dbufs_mtx);
/*
* 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;
dbuf_hash_remove(db);
} else {
DB_DNODE_EXIT(db);
}
ASSERT(refcount_is_zero(&db->db_holds));
db->db_parent = NULL;
ASSERT(db->db_buf == NULL);
ASSERT(db->db.db_data == NULL);
ASSERT(db->db_hash_next == NULL);
ASSERT(db->db_blkptr == NULL);
ASSERT(db->db_data_pending == NULL);
ASSERT(!multilist_link_active(&db->db_cache_link));
kmem_cache_free(dbuf_kmem_cache, db);
arc_space_return(sizeof (dmu_buf_impl_t), ARC_SPACE_DBUF);
/*
* 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);
}
/*
* Note: While bpp will always be updated if the function returns success,
* parentp will not be updated if the dnode does not have dn_dbuf filled in;
* this happens when the dnode is the meta-dnode, or a userused or groupused
* object.
*/
__attribute__((always_inline))
static inline int
dbuf_findbp(dnode_t *dn, int level, uint64_t blkid, int fail_sparse,
dmu_buf_impl_t **parentp, blkptr_t **bpp, struct dbuf_hold_impl_data *dh)
{
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_SPILL_BLKPTR(dn->dn_phys);
else
*bpp = NULL;
dbuf_add_ref(dn->dn_dbuf, NULL);
*parentp = dn->dn_dbuf;
mutex_exit(&dn->dn_mtx);
return (0);
}
nlevels =
(dn->dn_phys->dn_nlevels == 0) ? 1 : dn->dn_phys->dn_nlevels;
epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT;
ASSERT3U(level * epbs, <, 64);
ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));
/*
* This assertion shouldn't trip as long as the max indirect block size
* is less than 1M. The reason for this is that up to that point,
* the number of levels required to address an entire object with blocks
* of size SPA_MINBLOCKSIZE satisfies nlevels * epbs + 1 <= 64. In
* other words, if N * epbs + 1 > 64, then if (N-1) * epbs + 1 > 55
* (i.e. we can address the entire object), objects will all use at most
* N-1 levels and the assertion won't overflow. However, once epbs is
* 13, 4 * 13 + 1 = 53, but 5 * 13 + 1 = 66. Then, 4 levels will not be
* enough to address an entire object, so objects will have 5 levels,
* but then this assertion will overflow.
*
* All this is to say that if we ever increase DN_MAX_INDBLKSHIFT, we
* need to redo this logic to handle overflows.
*/
ASSERT(level >= nlevels ||
((nlevels - level - 1) * epbs) +
highbit64(dn->dn_phys->dn_nblkptr) <= 64);
if (level >= nlevels ||
blkid >= ((uint64_t)dn->dn_phys->dn_nblkptr <<
((nlevels - level - 1) * epbs)) ||
(fail_sparse &&
blkid > (dn->dn_phys->dn_maxblkid >> (level * epbs)))) {
/* the buffer has no parent yet */
return (SET_ERROR(ENOENT));
} else if (level < nlevels-1) {
/* this block is referenced from an indirect block */
int err;
if (dh == NULL) {
err = dbuf_hold_impl(dn, level+1,
blkid >> epbs, fail_sparse, FALSE, NULL, parentp);
} else {
__dbuf_hold_impl_init(dh + 1, dn, dh->dh_level + 1,
blkid >> epbs, fail_sparse, FALSE, NULL,
parentp, dh->dh_depth + 1);
err = __dbuf_hold_impl(dh + 1);
}
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));
if (blkid > (dn->dn_phys->dn_maxblkid >> (level * epbs)))
ASSERT(BP_IS_HOLE(*bpp));
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_kmem_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 = NULL;
db->db_user_immediate_evict = FALSE;
db->db_freed_in_flight = FALSE;
db->db_pending_evict = FALSE;
if (blkid == DMU_BONUS_BLKID) {
ASSERT3P(parent, ==, dn->dn_dbuf);
db->db.db_size = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots) -
(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_DBUF);
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_kmem_cache, db);
mutex_exit(&dn->dn_dbufs_mtx);
return (odb);
}
avl_add(&dn->dn_dbufs, db);
if (db->db_level == 0 && db->db_blkid >=
dn->dn_unlisted_l0_blkid)
dn->dn_unlisted_l0_blkid = db->db_blkid + 1;
db->db_state = DB_UNCACHED;
mutex_exit(&dn->dn_dbufs_mtx);
arc_space_consume(sizeof (dmu_buf_impl_t), ARC_SPACE_DBUF);
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);
atomic_inc_32(&dn->dn_dbufs_count);
dprintf_dbuf(db, "db=%p\n", db);
return (db);
}
typedef struct dbuf_prefetch_arg {
spa_t *dpa_spa; /* The spa to issue the prefetch in. */
zbookmark_phys_t dpa_zb; /* The target block to prefetch. */
int dpa_epbs; /* Entries (blkptr_t's) Per Block Shift. */
int dpa_curlevel; /* The current level that we're reading */
dnode_t *dpa_dnode; /* The dnode associated with the prefetch */
zio_priority_t dpa_prio; /* The priority I/Os should be issued at. */
zio_t *dpa_zio; /* The parent zio_t for all prefetches. */
arc_flags_t dpa_aflags; /* Flags to pass to the final prefetch. */
} dbuf_prefetch_arg_t;
/*
* Actually issue the prefetch read for the block given.
*/
static void
dbuf_issue_final_prefetch(dbuf_prefetch_arg_t *dpa, blkptr_t *bp)
{
arc_flags_t aflags;
if (BP_IS_HOLE(bp) || BP_IS_EMBEDDED(bp))
return;
aflags = dpa->dpa_aflags | ARC_FLAG_NOWAIT | ARC_FLAG_PREFETCH;
ASSERT3U(dpa->dpa_curlevel, ==, BP_GET_LEVEL(bp));
ASSERT3U(dpa->dpa_curlevel, ==, dpa->dpa_zb.zb_level);
ASSERT(dpa->dpa_zio != NULL);
(void) arc_read(dpa->dpa_zio, dpa->dpa_spa, bp, NULL, NULL,
dpa->dpa_prio, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE,
&aflags, &dpa->dpa_zb);
}
/*
* Called when an indirect block above our prefetch target is read in. This
* will either read in the next indirect block down the tree or issue the actual
* prefetch if the next block down is our target.
*/
static void
dbuf_prefetch_indirect_done(zio_t *zio, arc_buf_t *abuf, void *private)
{
dbuf_prefetch_arg_t *dpa = private;
uint64_t nextblkid;
blkptr_t *bp;
ASSERT3S(dpa->dpa_zb.zb_level, <, dpa->dpa_curlevel);
ASSERT3S(dpa->dpa_curlevel, >, 0);
/*
* The dpa_dnode is only valid if we are called with a NULL
* zio. This indicates that the arc_read() returned without
* first calling zio_read() to issue a physical read. Once
* a physical read is made the dpa_dnode must be invalidated
* as the locks guarding it may have been dropped. If the
* dpa_dnode is still valid, then we want to add it to the dbuf
* cache. To do so, we must hold the dbuf associated with the block
* we just prefetched, read its contents so that we associate it
* with an arc_buf_t, and then release it.
*/
if (zio != NULL) {
ASSERT3S(BP_GET_LEVEL(zio->io_bp), ==, dpa->dpa_curlevel);
if (zio->io_flags & ZIO_FLAG_RAW) {
ASSERT3U(BP_GET_PSIZE(zio->io_bp), ==, zio->io_size);
} else {
ASSERT3U(BP_GET_LSIZE(zio->io_bp), ==, zio->io_size);
}
ASSERT3P(zio->io_spa, ==, dpa->dpa_spa);
dpa->dpa_dnode = NULL;
} else if (dpa->dpa_dnode != NULL) {
uint64_t curblkid = dpa->dpa_zb.zb_blkid >>
(dpa->dpa_epbs * (dpa->dpa_curlevel -
dpa->dpa_zb.zb_level));
dmu_buf_impl_t *db = dbuf_hold_level(dpa->dpa_dnode,
dpa->dpa_curlevel, curblkid, FTAG);
(void) dbuf_read(db, NULL,
DB_RF_MUST_SUCCEED | DB_RF_NOPREFETCH | DB_RF_HAVESTRUCT);
dbuf_rele(db, FTAG);
}
dpa->dpa_curlevel--;
nextblkid = dpa->dpa_zb.zb_blkid >>
(dpa->dpa_epbs * (dpa->dpa_curlevel - dpa->dpa_zb.zb_level));
bp = ((blkptr_t *)abuf->b_data) +
P2PHASE(nextblkid, 1ULL << dpa->dpa_epbs);
if (BP_IS_HOLE(bp) || (zio != NULL && zio->io_error != 0)) {
kmem_free(dpa, sizeof (*dpa));
} else if (dpa->dpa_curlevel == dpa->dpa_zb.zb_level) {
ASSERT3U(nextblkid, ==, dpa->dpa_zb.zb_blkid);
dbuf_issue_final_prefetch(dpa, bp);
kmem_free(dpa, sizeof (*dpa));
} else {
arc_flags_t iter_aflags = ARC_FLAG_NOWAIT;
zbookmark_phys_t zb;
ASSERT3U(dpa->dpa_curlevel, ==, BP_GET_LEVEL(bp));
SET_BOOKMARK(&zb, dpa->dpa_zb.zb_objset,
dpa->dpa_zb.zb_object, dpa->dpa_curlevel, nextblkid);
(void) arc_read(dpa->dpa_zio, dpa->dpa_spa,
bp, dbuf_prefetch_indirect_done, dpa, dpa->dpa_prio,
ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE,
&iter_aflags, &zb);
}
arc_buf_destroy(abuf, private);
}
/*
* Issue prefetch reads for the given block on the given level. If the indirect
* blocks above that block are not in memory, we will read them in
* asynchronously. As a result, this call never blocks waiting for a read to
* complete.
*/
void
dbuf_prefetch(dnode_t *dn, int64_t level, uint64_t blkid, zio_priority_t prio,
arc_flags_t aflags)
{
blkptr_t bp;
int epbs, nlevels, curlevel;
uint64_t curblkid;
dmu_buf_impl_t *db;
zio_t *pio;
dbuf_prefetch_arg_t *dpa;
dsl_dataset_t *ds;
ASSERT(blkid != DMU_BONUS_BLKID);
ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));
if (blkid > dn->dn_maxblkid)
return;
if (dnode_block_freed(dn, blkid))
return;
/*
* This dnode hasn't been written to disk yet, so there's nothing to
* prefetch.
*/
nlevels = dn->dn_phys->dn_nlevels;
if (level >= nlevels || dn->dn_phys->dn_nblkptr == 0)
return;
epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT;
if (dn->dn_phys->dn_maxblkid < blkid << (epbs * level))
return;
db = dbuf_find(dn->dn_objset, dn->dn_object,
level, blkid);
if (db != NULL) {
mutex_exit(&db->db_mtx);
/*
* This dbuf already exists. It is either CACHED, or
* (we assume) about to be read or filled.
*/
return;
}
/*
* Find the closest ancestor (indirect block) of the target block
* that is present in the cache. In this indirect block, we will
* find the bp that is at curlevel, curblkid.
*/
curlevel = level;
curblkid = blkid;
while (curlevel < nlevels - 1) {
int parent_level = curlevel + 1;
uint64_t parent_blkid = curblkid >> epbs;
dmu_buf_impl_t *db;
if (dbuf_hold_impl(dn, parent_level, parent_blkid,
FALSE, TRUE, FTAG, &db) == 0) {
blkptr_t *bpp = db->db_buf->b_data;
bp = bpp[P2PHASE(curblkid, 1 << epbs)];
dbuf_rele(db, FTAG);
break;
}
curlevel = parent_level;
curblkid = parent_blkid;
}
if (curlevel == nlevels - 1) {
/* No cached indirect blocks found. */
ASSERT3U(curblkid, <, dn->dn_phys->dn_nblkptr);
bp = dn->dn_phys->dn_blkptr[curblkid];
}
if (BP_IS_HOLE(&bp))
return;
ASSERT3U(curlevel, ==, BP_GET_LEVEL(&bp));
pio = zio_root(dmu_objset_spa(dn->dn_objset), NULL, NULL,
ZIO_FLAG_CANFAIL);
dpa = kmem_zalloc(sizeof (*dpa), KM_SLEEP);
ds = dn->dn_objset->os_dsl_dataset;
SET_BOOKMARK(&dpa->dpa_zb, ds != NULL ? ds->ds_object : DMU_META_OBJSET,
dn->dn_object, level, blkid);
dpa->dpa_curlevel = curlevel;
dpa->dpa_prio = prio;
dpa->dpa_aflags = aflags;
dpa->dpa_spa = dn->dn_objset->os_spa;
dpa->dpa_dnode = dn;
dpa->dpa_epbs = epbs;
dpa->dpa_zio = pio;
/*
* If we have the indirect just above us, no need to do the asynchronous
* prefetch chain; we'll just run the last step ourselves. If we're at
* a higher level, though, we want to issue the prefetches for all the
* indirect blocks asynchronously, so we can go on with whatever we were
* doing.
*/
if (curlevel == level) {
ASSERT3U(curblkid, ==, blkid);
dbuf_issue_final_prefetch(dpa, &bp);
kmem_free(dpa, sizeof (*dpa));
} else {
arc_flags_t iter_aflags = ARC_FLAG_NOWAIT;
zbookmark_phys_t zb;
SET_BOOKMARK(&zb, ds != NULL ? ds->ds_object : DMU_META_OBJSET,
dn->dn_object, curlevel, curblkid);
(void) arc_read(dpa->dpa_zio, dpa->dpa_spa,
&bp, dbuf_prefetch_indirect_done, dpa, prio,
ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE,
&iter_aflags, &zb);
}
/*
* We use pio here instead of dpa_zio since it's possible that
* dpa may have already been freed.
*/
zio_nowait(pio);
}
#define DBUF_HOLD_IMPL_MAX_DEPTH 20
/*
* Returns with db_holds incremented, and db_mtx not held.
* Note: dn_struct_rwlock must be held.
*/
static int
__dbuf_hold_impl(struct dbuf_hold_impl_data *dh)
{
ASSERT3S(dh->dh_depth, <, DBUF_HOLD_IMPL_MAX_DEPTH);
dh->dh_parent = NULL;
ASSERT(dh->dh_blkid != DMU_BONUS_BLKID);
ASSERT(RW_LOCK_HELD(&dh->dh_dn->dn_struct_rwlock));
ASSERT3U(dh->dh_dn->dn_nlevels, >, dh->dh_level);
*(dh->dh_dbp) = NULL;
/* dbuf_find() returns with db_mtx held */
dh->dh_db = dbuf_find(dh->dh_dn->dn_objset, dh->dh_dn->dn_object,
dh->dh_level, dh->dh_blkid);
if (dh->dh_db == NULL) {
dh->dh_bp = NULL;
if (dh->dh_fail_uncached)
return (SET_ERROR(ENOENT));
ASSERT3P(dh->dh_parent, ==, NULL);
dh->dh_err = dbuf_findbp(dh->dh_dn, dh->dh_level, dh->dh_blkid,
dh->dh_fail_sparse, &dh->dh_parent, &dh->dh_bp, dh);
if (dh->dh_fail_sparse) {
if (dh->dh_err == 0 &&
dh->dh_bp && BP_IS_HOLE(dh->dh_bp))
dh->dh_err = SET_ERROR(ENOENT);
if (dh->dh_err) {
if (dh->dh_parent)
dbuf_rele(dh->dh_parent, NULL);
return (dh->dh_err);
}
}
if (dh->dh_err && dh->dh_err != ENOENT)
return (dh->dh_err);
dh->dh_db = dbuf_create(dh->dh_dn, dh->dh_level, dh->dh_blkid,
dh->dh_parent, dh->dh_bp);
}
if (dh->dh_fail_uncached && dh->dh_db->db_state != DB_CACHED) {
mutex_exit(&dh->dh_db->db_mtx);
return (SET_ERROR(ENOENT));
}
if (dh->dh_db->db_buf != NULL)
ASSERT3P(dh->dh_db->db.db_data, ==, dh->dh_db->db_buf->b_data);
ASSERT(dh->dh_db->db_buf == NULL || arc_referenced(dh->dh_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 (dh->dh_db->db_level == 0 &&
dh->dh_db->db_blkid != DMU_BONUS_BLKID &&
dh->dh_dn->dn_object != DMU_META_DNODE_OBJECT &&
dh->dh_db->db_state == DB_CACHED && dh->dh_db->db_data_pending) {
dh->dh_dr = dh->dh_db->db_data_pending;
if (dh->dh_dr->dt.dl.dr_data == dh->dh_db->db_buf) {
dh->dh_type = DBUF_GET_BUFC_TYPE(dh->dh_db);
dbuf_set_data(dh->dh_db,
arc_alloc_buf(dh->dh_dn->dn_objset->os_spa,
dh->dh_db, dh->dh_type, dh->dh_db->db.db_size));
bcopy(dh->dh_dr->dt.dl.dr_data->b_data,
dh->dh_db->db.db_data, dh->dh_db->db.db_size);
}
}
if (multilist_link_active(&dh->dh_db->db_cache_link)) {
ASSERT(refcount_is_zero(&dh->dh_db->db_holds));
multilist_remove(&dbuf_cache, dh->dh_db);
(void) refcount_remove_many(&dbuf_cache_size,
dh->dh_db->db.db_size, dh->dh_db);
}
(void) refcount_add(&dh->dh_db->db_holds, dh->dh_tag);
DBUF_VERIFY(dh->dh_db);
mutex_exit(&dh->dh_db->db_mtx);
/* NOTE: we can't rele the parent until after we drop the db_mtx */
if (dh->dh_parent)
dbuf_rele(dh->dh_parent, NULL);
ASSERT3P(DB_DNODE(dh->dh_db), ==, dh->dh_dn);
ASSERT3U(dh->dh_db->db_blkid, ==, dh->dh_blkid);
ASSERT3U(dh->dh_db->db_level, ==, dh->dh_level);
*(dh->dh_dbp) = dh->dh_db;
return (0);
}
/*
* The following code preserves the recursive function dbuf_hold_impl()
* but moves the local variables AND function arguments to the heap to
* minimize the stack frame size. Enough space is initially allocated
* on the stack for 20 levels of recursion.
*/
int
dbuf_hold_impl(dnode_t *dn, uint8_t level, uint64_t blkid,
boolean_t fail_sparse, boolean_t fail_uncached,
void *tag, dmu_buf_impl_t **dbp)
{
struct dbuf_hold_impl_data *dh;
int error;
dh = kmem_alloc(sizeof (struct dbuf_hold_impl_data) *
DBUF_HOLD_IMPL_MAX_DEPTH, KM_SLEEP);
__dbuf_hold_impl_init(dh, dn, level, blkid, fail_sparse,
fail_uncached, tag, dbp, 0);
error = __dbuf_hold_impl(dh);
kmem_free(dh, sizeof (struct dbuf_hold_impl_data) *
DBUF_HOLD_IMPL_MAX_DEPTH);
return (error);
}
static void
__dbuf_hold_impl_init(struct dbuf_hold_impl_data *dh,
dnode_t *dn, uint8_t level, uint64_t blkid,
boolean_t fail_sparse, boolean_t fail_uncached,
void *tag, dmu_buf_impl_t **dbp, int depth)
{
dh->dh_dn = dn;
dh->dh_level = level;
dh->dh_blkid = blkid;
dh->dh_fail_sparse = fail_sparse;
dh->dh_fail_uncached = fail_uncached;
dh->dh_tag = tag;
dh->dh_dbp = dbp;
dh->dh_db = NULL;
dh->dh_parent = NULL;
dh->dh_bp = NULL;
dh->dh_err = 0;
dh->dh_dr = NULL;
dh->dh_type = 0;
dh->dh_depth = depth;
}
dmu_buf_impl_t *
dbuf_hold(dnode_t *dn, uint64_t blkid, void *tag)
{
return (dbuf_hold_level(dn, 0, blkid, tag));
}
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, 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 (SET_ERROR(ENOTSUP));
if (blksz == 0)
blksz = SPA_MINBLOCKSIZE;
ASSERT3U(blksz, <=, spa_maxblocksize(dmu_objset_spa(db->db_objset)));
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);
VERIFY3S(holds, >, 1);
}
#pragma weak dmu_buf_try_add_ref = dbuf_try_add_ref
boolean_t
dbuf_try_add_ref(dmu_buf_t *db_fake, objset_t *os, uint64_t obj, uint64_t blkid,
void *tag)
{
dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;
dmu_buf_impl_t *found_db;
boolean_t result = B_FALSE;
if (blkid == DMU_BONUS_BLKID)
found_db = dbuf_find_bonus(os, obj);
else
found_db = dbuf_find(os, obj, 0, blkid);
if (found_db != NULL) {
if (db == found_db && dbuf_refcount(db) > db->db_dirtycnt) {
(void) refcount_add(&db->db_holds, tag);
result = B_TRUE;
}
mutex_exit(&found_db->db_mtx);
}
return (result);
}
/*
* 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.
*/
void
dbuf_rele(dmu_buf_impl_t *db, void *tag)
{
mutex_enter(&db->db_mtx);
dbuf_rele_and_unlock(db, tag);
}
void
dmu_buf_rele(dmu_buf_t *db, void *tag)
{
dbuf_rele((dmu_buf_impl_t *)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 != NULL &&
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_user_immediate_evict)
dbuf_evict_user(db);
if (holds == 0) {
if (db->db_blkid == DMU_BONUS_BLKID) {
dnode_t *dn;
boolean_t evict_dbuf = db->db_pending_evict;
/*
* If the dnode moves here, we cannot cross this
* barrier until the move completes.
*/
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
atomic_dec_32(&dn->dn_dbufs_count);
/*
* Decrementing the dbuf count means that the bonus
* buffer's dnode hold is no longer discounted in
* dnode_move(). The dnode cannot move until after
* the dnode_rele() below.
*/
DB_DNODE_EXIT(db);
/*
* Do not reference db after its lock is dropped.
* Another thread may evict it.
*/
mutex_exit(&db->db_mtx);
if (evict_dbuf)
dnode_evict_bonus(dn);
dnode_rele(dn, 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_destroy(db);
} else if (arc_released(db->db_buf)) {
/*
* This dbuf has anonymous data associated with it.
*/
dbuf_destroy(db);
} else {
boolean_t do_arc_evict = B_FALSE;
blkptr_t bp;
spa_t *spa = dmu_objset_spa(db->db_objset);
if (!DBUF_IS_CACHEABLE(db) &&
db->db_blkptr != NULL &&
!BP_IS_HOLE(db->db_blkptr) &&
!BP_IS_EMBEDDED(db->db_blkptr)) {
do_arc_evict = B_TRUE;
bp = *db->db_blkptr;
}
if (!DBUF_IS_CACHEABLE(db) ||
db->db_pending_evict) {
dbuf_destroy(db);
} else if (!multilist_link_active(&db->db_cache_link)) {
multilist_insert(&dbuf_cache, db);
(void) refcount_add_many(&dbuf_cache_size,
db->db.db_size, db);
mutex_exit(&db->db_mtx);
dbuf_evict_notify();
}
if (do_arc_evict)
arc_freed(spa, &bp);
}
} 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_replace_user(dmu_buf_t *db_fake, dmu_buf_user_t *old_user,
dmu_buf_user_t *new_user)
{
dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;
mutex_enter(&db->db_mtx);
dbuf_verify_user(db, DBVU_NOT_EVICTING);
if (db->db_user == old_user)
db->db_user = new_user;
else
old_user = db->db_user;
dbuf_verify_user(db, DBVU_NOT_EVICTING);
mutex_exit(&db->db_mtx);
return (old_user);
}
void *
dmu_buf_set_user(dmu_buf_t *db_fake, dmu_buf_user_t *user)
{
return (dmu_buf_replace_user(db_fake, NULL, user));
}
void *
dmu_buf_set_user_ie(dmu_buf_t *db_fake, dmu_buf_user_t *user)
{
dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;
db->db_user_immediate_evict = TRUE;
return (dmu_buf_set_user(db_fake, user));
}
void *
dmu_buf_remove_user(dmu_buf_t *db_fake, dmu_buf_user_t *user)
{
return (dmu_buf_replace_user(db_fake, user, NULL));
}
void *
dmu_buf_get_user(dmu_buf_t *db_fake)
{
dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake;
dbuf_verify_user(db, DBVU_NOT_EVICTING);
return (db->db_user);
}
void
dmu_buf_user_evict_wait()
{
taskq_wait(dbu_evict_taskq);
}
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);
}
blkptr_t *
dmu_buf_get_blkptr(dmu_buf_t *db)
{
dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db;
return (dbi->db_blkptr);
}
objset_t *
dmu_buf_get_objset(dmu_buf_t *db)
{
dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db;
return (dbi->db_objset);
}
dnode_t *
dmu_buf_dnode_enter(dmu_buf_t *db)
{
dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db;
DB_DNODE_ENTER(dbi);
return (DB_DNODE(dbi));
}
void
dmu_buf_dnode_exit(dmu_buf_t *db)
{
dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db;
DB_DNODE_EXIT(dbi);
}
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_SPILL_BLKPTR(dn->dn_phys);
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);
parent = dbuf_hold_level(dn, db->db_level + 1,
db->db_blkid >> epbs, db);
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);
}
}
/*
* dbuf_sync_indirect() is called recursively from dbuf_sync_list() so it
* is critical the we not allow the compiler to inline this function in to
* dbuf_sync_list() thereby drastically bloating the stack usage.
*/
noinline 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);
/* Read the block if it hasn't been read yet. */
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);
/* Indirect block size must match what the dnode thinks it is. */
ASSERT3U(db->db.db_size, ==, 1<<dn->dn_phys->dn_indblkshift);
dbuf_check_blkptr(dn, db);
DB_DNODE_EXIT(db);
/* Provide the pending dirty record to child dbufs */
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, db->db_level - 1, tx);
ASSERT(list_head(&dr->dt.di.dr_children) == NULL);
mutex_exit(&dr->dt.di.dr_mtx);
zio_nowait(zio);
}
/*
* dbuf_sync_leaf() is called recursively from dbuf_sync_list() so it is
* critical the we not allow the compiler to inline this function in to
* dbuf_sync_list() thereby drastically bloating the stack usage.
*/
noinline 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);
if (!(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR)) {
/*
* In the previous transaction group, the bonus buffer
* was entirely used to store the attributes for the
* dnode which overrode the dn_spill field. However,
* when adding more attributes to the file a spill
* block was required to hold the extra attributes.
*
* Make sure to clear the garbage left in the dn_spill
* field from the previous attributes in the bonus
* buffer. Otherwise, after writing out the spill
* block to the new allocated dva, it will free
* the old block pointed to by the invalid dn_spill.
*/
db->db_blkptr = NULL;
}
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);
ASSERT0(db->db_level);
ASSERT3U(dn->dn_phys->dn_bonuslen, <=,
DN_SLOTS_TO_BONUSLEN(dn->dn_phys->dn_extra_slots + 1));
bcopy(*datap, DN_BONUS(dn->dn_phys), dn->dn_phys->dn_bonuslen);
DB_DNODE_EXIT(db);
if (*datap != db->db.db_data) {
int slots = DB_DNODE(db)->dn_num_slots;
int bonuslen = DN_SLOTS_TO_BONUSLEN(slots);
kmem_free(*datap, bonuslen);
arc_space_return(bonuslen, ARC_SPACE_BONUS);
}
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;
if (dr->dr_dbuf->db_level != 0) {
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;
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 psize = arc_buf_size(*datap);
arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
enum zio_compress compress_type = arc_get_compression(*datap);
if (compress_type == ZIO_COMPRESS_OFF) {
*datap = arc_alloc_buf(os->os_spa, db, type, psize);
} else {
int lsize = arc_buf_lsize(*datap);
ASSERT3U(type, ==, ARC_BUFC_DATA);
*datap = arc_alloc_compressed_buf(os->os_spa, db,
psize, lsize, compress_type);
}
bcopy(db->db.db_data, (*datap)->b_data, psize);
}
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, int level, 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;
}
if (dr->dr_dbuf->db_blkid != DMU_BONUS_BLKID &&
dr->dr_dbuf->db_blkid != DMU_SPILL_BLKID) {
VERIFY3U(dr->dr_dbuf->db_level, ==, level);
}
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;
ASSERT3P(db->db_blkptr, !=, NULL);
ASSERT3P(&db->db_data_pending->dr_bp_copy, ==, 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->blk_birth != 0) {
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) ||
BP_IS_EMBEDDED(bp));
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(bp)) &&
db->db_blkptr == DN_SPILL_BLKPTR(dn->dn_phys));
}
#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) {
i = 0;
while (i < db->db.db_size) {
dnode_phys_t *dnp = db->db.db_data + i;
i += DNODE_MIN_SIZE;
if (dnp->dn_type != DMU_OT_NONE) {
fill++;
i += dnp->dn_extra_slots *
DNODE_MIN_SIZE;
}
}
} else {
if (BP_IS_HOLE(bp)) {
fill = 0;
} 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 += BP_GET_FILL(ibp);
}
}
DB_DNODE_EXIT(db);
if (!BP_IS_EMBEDDED(bp))
bp->blk_fill = fill;
mutex_exit(&db->db_mtx);
rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
*db->db_blkptr = *bp;
rw_exit(&dn->dn_struct_rwlock);
}
/* ARGSUSED */
/*
* This function gets called just prior to running through the compression
* stage of the zio pipeline. If we're an indirect block comprised of only
* holes, then we want this indirect to be compressed away to a hole. In
* order to do that we must zero out any information about the holes that
* this indirect points to prior to before we try to compress it.
*/
static void
dbuf_write_children_ready(zio_t *zio, arc_buf_t *buf, void *vdb)
{
dmu_buf_impl_t *db = vdb;
dnode_t *dn;
blkptr_t *bp;
uint64_t i;
int epbs;
ASSERT3U(db->db_level, >, 0);
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT;
/* Determine if all our children are holes */
for (i = 0, bp = db->db.db_data; i < 1ULL << epbs; i++, bp++) {
if (!BP_IS_HOLE(bp))
break;
}
/*
* If all the children are holes, then zero them all out so that
* we may get compressed away.
*/
if (i == 1ULL << epbs) {
/* didn't find any non-holes */
bzero(db->db.db_data, db->db.db_size);
}
DB_DNODE_EXIT(db);
}
/*
* The SPA will call this callback several times for each zio - once
* for every physical child i/o (zio->io_phys_children times). This
* allows the DMU to monitor the progress of each logical i/o. For example,
* there may be 2 copies of an indirect block, or many fragments of a RAID-Z
* block. There may be a long delay before all copies/fragments are completed,
* so this callback allows us to retire dirty space gradually, as the physical
* i/os complete.
*/
/* ARGSUSED */
static void
dbuf_write_physdone(zio_t *zio, arc_buf_t *buf, void *arg)
{
dmu_buf_impl_t *db = arg;
objset_t *os = db->db_objset;
dsl_pool_t *dp = dmu_objset_pool(os);
dbuf_dirty_record_t *dr;
int delta = 0;
dr = db->db_data_pending;
ASSERT3U(dr->dr_txg, ==, zio->io_txg);
/*
* The callback will be called io_phys_children times. Retire one
* portion of our dirty space each time we are called. Any rounding
* error will be cleaned up by dsl_pool_sync()'s call to
* dsl_pool_undirty_space().
*/
delta = dr->dr_accounted / zio->io_phys_children;
dsl_pool_undirty_space(dp, delta, zio->io_txg);
}
/* ARGSUSED */
static void
dbuf_write_done(zio_t *zio, arc_buf_t *buf, void *vdb)
{
dmu_buf_impl_t *db = vdb;
blkptr_t *bp_orig = &zio->io_bp_orig;
blkptr_t *bp = db->db_blkptr;
objset_t *os = db->db_objset;
dmu_tx_t *tx = os->os_synctx;
dbuf_dirty_record_t **drp, *dr;
ASSERT0(zio->io_error);
ASSERT(db->db_blkptr == bp);
/*
* For nopwrites and rewrites we ensure that the bp matches our
* original and bypass all the accounting.
*/
if (zio->io_flags & (ZIO_FLAG_IO_REWRITE | ZIO_FLAG_NOPWRITE)) {
ASSERT(BP_EQUAL(bp, bp_orig));
} else {
dsl_dataset_t *ds = os->os_dsl_dataset;
(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_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_SPILL_BLKPTR(dn->dn_phys));
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)
arc_buf_destroy(dr->dt.dl.dr_data, 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)) {
ASSERTV(int epbs = dn->dn_phys->dn_indblkshift -
SPA_BLKPTRSHIFT);
ASSERT3U(db->db_blkid, <=,
dn->dn_phys->dn_maxblkid >> (db->db_level * epbs));
ASSERT3U(BP_GET_LSIZE(db->db_blkptr), ==,
db->db.db_size);
}
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)tx->tx_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);
if (zio->io_abd != NULL)
abd_put(zio->io_abd);
}
/* Issue I/O to commit a dirty buffer to disk. */
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_phys_t zb;
zio_prop_t zp;
zio_t *zio;
int wp_flag = 0;
ASSERT(dmu_tx_is_syncing(tx));
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) {
/* Our parent is an indirect block. */
/* We have a dirty parent that has been scheduled for write. */
ASSERT(parent && parent->db_data_pending);
/* Our parent's buffer is one level closer to the dnode. */
ASSERT(db->db_level == parent->db_level-1);
/*
* We're about to modify our parent's db_data by modifying
* our block pointer, so the parent must be released.
*/
ASSERT(arc_released(parent->db_buf));
zio = parent->db_data_pending->dr_zio;
} else {
/* Our parent is the dnode itself. */
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,
(data != NULL && arc_get_compression(data) != ZIO_COMPRESS_OFF) ?
arc_get_compression(data) : ZIO_COMPRESS_INHERIT, &zp);
DB_DNODE_EXIT(db);
/*
* We copy the blkptr now (rather than when we instantiate the dirty
* record), because its value can change between open context and
* syncing context. We do not need to hold dn_struct_rwlock to read
* db_blkptr because we are in syncing context.
*/
dr->dr_bp_copy = *db->db_blkptr;
if (db->db_level == 0 &&
dr->dt.dl.dr_override_state == DR_OVERRIDDEN) {
/*
* The BP for this block has been provided by open context
* (by dmu_sync() or dmu_buf_write_embedded()).
*/
abd_t *contents = (data != NULL) ?
abd_get_from_buf(data->b_data, arc_buf_size(data)) : NULL;
dr->dr_zio = zio_write(zio, os->os_spa, txg,
&dr->dr_bp_copy, contents, db->db.db_size, db->db.db_size,
&zp, dbuf_write_override_ready, NULL, NULL,
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, dr->dt.dl.dr_nopwrite);
mutex_exit(&db->db_mtx);
} else if (db->db_state == DB_NOFILL) {
ASSERT(zp.zp_checksum == ZIO_CHECKSUM_OFF ||
zp.zp_checksum == ZIO_CHECKSUM_NOPARITY);
dr->dr_zio = zio_write(zio, os->os_spa, txg,
&dr->dr_bp_copy, NULL, db->db.db_size, db->db.db_size, &zp,
dbuf_write_nofill_ready, NULL, NULL,
dbuf_write_nofill_done, db,
ZIO_PRIORITY_ASYNC_WRITE,
ZIO_FLAG_MUSTSUCCEED | ZIO_FLAG_NODATA, &zb);
} else {
arc_done_func_t *children_ready_cb = NULL;
ASSERT(arc_released(data));
/*
* For indirect blocks, we want to setup the children
* ready callback so that we can properly handle an indirect
* block that only contains holes.
*/
if (db->db_level != 0)
children_ready_cb = dbuf_write_children_ready;
dr->dr_zio = arc_write(zio, os->os_spa, txg,
&dr->dr_bp_copy, data, DBUF_IS_L2CACHEABLE(db),
&zp, dbuf_write_ready,
children_ready_cb, dbuf_write_physdone,
dbuf_write_done, db, ZIO_PRIORITY_ASYNC_WRITE,
ZIO_FLAG_MUSTSUCCEED, &zb);
}
}
#if defined(_KERNEL) && defined(HAVE_SPL)
EXPORT_SYMBOL(dbuf_find);
EXPORT_SYMBOL(dbuf_is_metadata);
EXPORT_SYMBOL(dbuf_destroy);
EXPORT_SYMBOL(dbuf_loan_arcbuf);
EXPORT_SYMBOL(dbuf_whichblock);
EXPORT_SYMBOL(dbuf_read);
EXPORT_SYMBOL(dbuf_unoverride);
EXPORT_SYMBOL(dbuf_free_range);
EXPORT_SYMBOL(dbuf_new_size);
EXPORT_SYMBOL(dbuf_release_bp);
EXPORT_SYMBOL(dbuf_dirty);
EXPORT_SYMBOL(dmu_buf_will_dirty);
EXPORT_SYMBOL(dmu_buf_will_not_fill);
EXPORT_SYMBOL(dmu_buf_will_fill);
EXPORT_SYMBOL(dmu_buf_fill_done);
EXPORT_SYMBOL(dmu_buf_rele);
EXPORT_SYMBOL(dbuf_assign_arcbuf);
EXPORT_SYMBOL(dbuf_prefetch);
EXPORT_SYMBOL(dbuf_hold_impl);
EXPORT_SYMBOL(dbuf_hold);
EXPORT_SYMBOL(dbuf_hold_level);
EXPORT_SYMBOL(dbuf_create_bonus);
EXPORT_SYMBOL(dbuf_spill_set_blksz);
EXPORT_SYMBOL(dbuf_rm_spill);
EXPORT_SYMBOL(dbuf_add_ref);
EXPORT_SYMBOL(dbuf_rele);
EXPORT_SYMBOL(dbuf_rele_and_unlock);
EXPORT_SYMBOL(dbuf_refcount);
EXPORT_SYMBOL(dbuf_sync_list);
EXPORT_SYMBOL(dmu_buf_set_user);
EXPORT_SYMBOL(dmu_buf_set_user_ie);
EXPORT_SYMBOL(dmu_buf_get_user);
EXPORT_SYMBOL(dmu_buf_freeable);
EXPORT_SYMBOL(dmu_buf_get_blkptr);
/* BEGIN CSTYLED */
module_param(dbuf_cache_max_bytes, ulong, 0644);
MODULE_PARM_DESC(dbuf_cache_max_bytes,
"Maximum size in bytes of the dbuf cache.");
module_param(dbuf_cache_hiwater_pct, uint, 0644);
MODULE_PARM_DESC(dbuf_cache_hiwater_pct,
"Percentage over dbuf_cache_max_bytes when dbufs must be evicted "
"directly.");
module_param(dbuf_cache_lowater_pct, uint, 0644);
MODULE_PARM_DESC(dbuf_cache_lowater_pct,
"Percentage below dbuf_cache_max_bytes when the evict thread stops "
"evicting dbufs.");
module_param(dbuf_cache_max_shift, int, 0644);
MODULE_PARM_DESC(dbuf_cache_max_shift,
"Cap the size of the dbuf cache to a log2 fraction of arc size.");
/* END CSTYLED */
#endif
|