aboutsummaryrefslogtreecommitdiff
path: root/fs/ocfs2/alloc.c
blob: a9669686757612b3ee6e41c0ac9a221f4eb56af2 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
/* -*- mode: c; c-basic-offset: 8; -*-
 * vim: noexpandtab sw=8 ts=8 sts=0:
 *
 * alloc.c
 *
 * Extent allocs and frees
 *
 * Copyright (C) 2002, 2004 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

#include <linux/fs.h>
#include <linux/types.h>
#include <linux/slab.h>
#include <linux/highmem.h>

#define MLOG_MASK_PREFIX ML_DISK_ALLOC
#include <cluster/masklog.h>

#include "ocfs2.h"

#include "alloc.h"
#include "dlmglue.h"
#include "extent_map.h"
#include "inode.h"
#include "journal.h"
#include "localalloc.h"
#include "suballoc.h"
#include "sysfile.h"
#include "file.h"
#include "super.h"
#include "uptodate.h"

#include "buffer_head_io.h"

static void ocfs2_free_truncate_context(struct ocfs2_truncate_context *tc);

/*
 * Structures which describe a path through a btree, and functions to
 * manipulate them.
 *
 * The idea here is to be as generic as possible with the tree
 * manipulation code.
 */
struct ocfs2_path_item {
	struct buffer_head		*bh;
	struct ocfs2_extent_list	*el;
};

#define OCFS2_MAX_PATH_DEPTH	5

struct ocfs2_path {
	int			p_tree_depth;
	struct ocfs2_path_item	p_node[OCFS2_MAX_PATH_DEPTH];
};

#define path_root_bh(_path) ((_path)->p_node[0].bh)
#define path_root_el(_path) ((_path)->p_node[0].el)
#define path_leaf_bh(_path) ((_path)->p_node[(_path)->p_tree_depth].bh)
#define path_leaf_el(_path) ((_path)->p_node[(_path)->p_tree_depth].el)
#define path_num_items(_path) ((_path)->p_tree_depth + 1)

/*
 * Reset the actual path elements so that we can re-use the structure
 * to build another path. Generally, this involves freeing the buffer
 * heads.
 */
static void ocfs2_reinit_path(struct ocfs2_path *path, int keep_root)
{
	int i, start = 0, depth = 0;
	struct ocfs2_path_item *node;

	if (keep_root)
		start = 1;

	for(i = start; i < path_num_items(path); i++) {
		node = &path->p_node[i];

		brelse(node->bh);
		node->bh = NULL;
		node->el = NULL;
	}

	/*
	 * Tree depth may change during truncate, or insert. If we're
	 * keeping the root extent list, then make sure that our path
	 * structure reflects the proper depth.
	 */
	if (keep_root)
		depth = le16_to_cpu(path_root_el(path)->l_tree_depth);

	path->p_tree_depth = depth;
}

static void ocfs2_free_path(struct ocfs2_path *path)
{
	if (path) {
		ocfs2_reinit_path(path, 0);
		kfree(path);
	}
}

/*
 * Make the *dest path the same as src and re-initialize src path to
 * have a root only.
 */
static void ocfs2_mv_path(struct ocfs2_path *dest, struct ocfs2_path *src)
{
	int i;

	BUG_ON(path_root_bh(dest) != path_root_bh(src));

	for(i = 1; i < OCFS2_MAX_PATH_DEPTH; i++) {
		brelse(dest->p_node[i].bh);

		dest->p_node[i].bh = src->p_node[i].bh;
		dest->p_node[i].el = src->p_node[i].el;

		src->p_node[i].bh = NULL;
		src->p_node[i].el = NULL;
	}
}

/*
 * Insert an extent block at given index.
 *
 * This will not take an additional reference on eb_bh.
 */
static inline void ocfs2_path_insert_eb(struct ocfs2_path *path, int index,
					struct buffer_head *eb_bh)
{
	struct ocfs2_extent_block *eb = (struct ocfs2_extent_block *)eb_bh->b_data;

	/*
	 * Right now, no root bh is an extent block, so this helps
	 * catch code errors with dinode trees. The assertion can be
	 * safely removed if we ever need to insert extent block
	 * structures at the root.
	 */
	BUG_ON(index == 0);

	path->p_node[index].bh = eb_bh;
	path->p_node[index].el = &eb->h_list;
}

static struct ocfs2_path *ocfs2_new_path(struct buffer_head *root_bh,
					 struct ocfs2_extent_list *root_el)
{
	struct ocfs2_path *path;

	BUG_ON(le16_to_cpu(root_el->l_tree_depth) >= OCFS2_MAX_PATH_DEPTH);

	path = kzalloc(sizeof(*path), GFP_NOFS);
	if (path) {
		path->p_tree_depth = le16_to_cpu(root_el->l_tree_depth);
		get_bh(root_bh);
		path_root_bh(path) = root_bh;
		path_root_el(path) = root_el;
	}

	return path;
}

/*
 * Allocate and initialize a new path based on a disk inode tree.
 */
static struct ocfs2_path *ocfs2_new_inode_path(struct buffer_head *di_bh)
{
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	struct ocfs2_extent_list *el = &di->id2.i_list;

	return ocfs2_new_path(di_bh, el);
}

/*
 * Convenience function to journal all components in a path.
 */
static int ocfs2_journal_access_path(struct inode *inode, handle_t *handle,
				     struct ocfs2_path *path)
{
	int i, ret = 0;

	if (!path)
		goto out;

	for(i = 0; i < path_num_items(path); i++) {
		ret = ocfs2_journal_access(handle, inode, path->p_node[i].bh,
					   OCFS2_JOURNAL_ACCESS_WRITE);
		if (ret < 0) {
			mlog_errno(ret);
			goto out;
		}
	}

out:
	return ret;
}

enum ocfs2_contig_type {
	CONTIG_NONE = 0,
	CONTIG_LEFT,
	CONTIG_RIGHT
};

static int ocfs2_block_extent_contig(struct super_block *sb,
				     struct ocfs2_extent_rec *ext,
				     u64 blkno)
{
	return blkno == (le64_to_cpu(ext->e_blkno) +
			 ocfs2_clusters_to_blocks(sb,
						  le32_to_cpu(ext->e_clusters)));
}

static int ocfs2_extents_adjacent(struct ocfs2_extent_rec *left,
				  struct ocfs2_extent_rec *right)
{
	return (le32_to_cpu(left->e_cpos) + le32_to_cpu(left->e_clusters) ==
		le32_to_cpu(right->e_cpos));
}

static enum ocfs2_contig_type
	ocfs2_extent_contig(struct inode *inode,
			    struct ocfs2_extent_rec *ext,
			    struct ocfs2_extent_rec *insert_rec)
{
	u64 blkno = le64_to_cpu(insert_rec->e_blkno);

	if (ocfs2_extents_adjacent(ext, insert_rec) &&
	    ocfs2_block_extent_contig(inode->i_sb, ext, blkno))
			return CONTIG_RIGHT;

	blkno = le64_to_cpu(ext->e_blkno);
	if (ocfs2_extents_adjacent(insert_rec, ext) &&
	    ocfs2_block_extent_contig(inode->i_sb, insert_rec, blkno))
		return CONTIG_LEFT;

	return CONTIG_NONE;
}

/*
 * NOTE: We can have pretty much any combination of contiguousness and
 * appending.
 *
 * The usefulness of APPEND_TAIL is more in that it lets us know that
 * we'll have to update the path to that leaf.
 */
enum ocfs2_append_type {
	APPEND_NONE = 0,
	APPEND_TAIL,
};

struct ocfs2_insert_type {
	enum ocfs2_append_type	ins_appending;
	enum ocfs2_contig_type	ins_contig;
	int			ins_contig_index;
	int			ins_free_records;
	int			ins_tree_depth;
};

/*
 * How many free extents have we got before we need more meta data?
 */
int ocfs2_num_free_extents(struct ocfs2_super *osb,
			   struct inode *inode,
			   struct ocfs2_dinode *fe)
{
	int retval;
	struct ocfs2_extent_list *el;
	struct ocfs2_extent_block *eb;
	struct buffer_head *eb_bh = NULL;

	mlog_entry_void();

	if (!OCFS2_IS_VALID_DINODE(fe)) {
		OCFS2_RO_ON_INVALID_DINODE(inode->i_sb, fe);
		retval = -EIO;
		goto bail;
	}

	if (fe->i_last_eb_blk) {
		retval = ocfs2_read_block(osb, le64_to_cpu(fe->i_last_eb_blk),
					  &eb_bh, OCFS2_BH_CACHED, inode);
		if (retval < 0) {
			mlog_errno(retval);
			goto bail;
		}
		eb = (struct ocfs2_extent_block *) eb_bh->b_data;
		el = &eb->h_list;
	} else
		el = &fe->id2.i_list;

	BUG_ON(el->l_tree_depth != 0);

	retval = le16_to_cpu(el->l_count) - le16_to_cpu(el->l_next_free_rec);
bail:
	if (eb_bh)
		brelse(eb_bh);

	mlog_exit(retval);
	return retval;
}

/* expects array to already be allocated
 *
 * sets h_signature, h_blkno, h_suballoc_bit, h_suballoc_slot, and
 * l_count for you
 */
static int ocfs2_create_new_meta_bhs(struct ocfs2_super *osb,
				     handle_t *handle,
				     struct inode *inode,
				     int wanted,
				     struct ocfs2_alloc_context *meta_ac,
				     struct buffer_head *bhs[])
{
	int count, status, i;
	u16 suballoc_bit_start;
	u32 num_got;
	u64 first_blkno;
	struct ocfs2_extent_block *eb;

	mlog_entry_void();

	count = 0;
	while (count < wanted) {
		status = ocfs2_claim_metadata(osb,
					      handle,
					      meta_ac,
					      wanted - count,
					      &suballoc_bit_start,
					      &num_got,
					      &first_blkno);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}

		for(i = count;  i < (num_got + count); i++) {
			bhs[i] = sb_getblk(osb->sb, first_blkno);
			if (bhs[i] == NULL) {
				status = -EIO;
				mlog_errno(status);
				goto bail;
			}
			ocfs2_set_new_buffer_uptodate(inode, bhs[i]);

			status = ocfs2_journal_access(handle, inode, bhs[i],
						      OCFS2_JOURNAL_ACCESS_CREATE);
			if (status < 0) {
				mlog_errno(status);
				goto bail;
			}

			memset(bhs[i]->b_data, 0, osb->sb->s_blocksize);
			eb = (struct ocfs2_extent_block *) bhs[i]->b_data;
			/* Ok, setup the minimal stuff here. */
			strcpy(eb->h_signature, OCFS2_EXTENT_BLOCK_SIGNATURE);
			eb->h_blkno = cpu_to_le64(first_blkno);
			eb->h_fs_generation = cpu_to_le32(osb->fs_generation);

#ifndef OCFS2_USE_ALL_METADATA_SUBALLOCATORS
			/* we always use slot zero's suballocator */
			eb->h_suballoc_slot = 0;
#else
			eb->h_suballoc_slot = cpu_to_le16(osb->slot_num);
#endif
			eb->h_suballoc_bit = cpu_to_le16(suballoc_bit_start);
			eb->h_list.l_count =
				cpu_to_le16(ocfs2_extent_recs_per_eb(osb->sb));

			suballoc_bit_start++;
			first_blkno++;

			/* We'll also be dirtied by the caller, so
			 * this isn't absolutely necessary. */
			status = ocfs2_journal_dirty(handle, bhs[i]);
			if (status < 0) {
				mlog_errno(status);
				goto bail;
			}
		}

		count += num_got;
	}

	status = 0;
bail:
	if (status < 0) {
		for(i = 0; i < wanted; i++) {
			if (bhs[i])
				brelse(bhs[i]);
			bhs[i] = NULL;
		}
	}
	mlog_exit(status);
	return status;
}

/*
 * Helper function for ocfs2_add_branch() and ocfs2_shift_tree_depth().
 *
 * Returns the sum of the rightmost extent rec logical offset and
 * cluster count.
 *
 * ocfs2_add_branch() uses this to determine what logical cluster
 * value should be populated into the leftmost new branch records.
 *
 * ocfs2_shift_tree_depth() uses this to determine the # clusters
 * value for the new topmost tree record.
 */
static inline u32 ocfs2_sum_rightmost_rec(struct ocfs2_extent_list  *el)
{
	int i;

	i = le16_to_cpu(el->l_next_free_rec) - 1;

	return le32_to_cpu(el->l_recs[i].e_cpos) +
		le32_to_cpu(el->l_recs[i].e_clusters);
}

/*
 * Add an entire tree branch to our inode. eb_bh is the extent block
 * to start at, if we don't want to start the branch at the dinode
 * structure.
 *
 * last_eb_bh is required as we have to update it's next_leaf pointer
 * for the new last extent block.
 *
 * the new branch will be 'empty' in the sense that every block will
 * contain a single record with e_clusters == 0.
 */
static int ocfs2_add_branch(struct ocfs2_super *osb,
			    handle_t *handle,
			    struct inode *inode,
			    struct buffer_head *fe_bh,
			    struct buffer_head *eb_bh,
			    struct buffer_head *last_eb_bh,
			    struct ocfs2_alloc_context *meta_ac)
{
	int status, new_blocks, i;
	u64 next_blkno, new_last_eb_blk;
	struct buffer_head *bh;
	struct buffer_head **new_eb_bhs = NULL;
	struct ocfs2_dinode *fe;
	struct ocfs2_extent_block *eb;
	struct ocfs2_extent_list  *eb_el;
	struct ocfs2_extent_list  *el;
	u32 new_cpos;

	mlog_entry_void();

	BUG_ON(!last_eb_bh);

	fe = (struct ocfs2_dinode *) fe_bh->b_data;

	if (eb_bh) {
		eb = (struct ocfs2_extent_block *) eb_bh->b_data;
		el = &eb->h_list;
	} else
		el = &fe->id2.i_list;

	/* we never add a branch to a leaf. */
	BUG_ON(!el->l_tree_depth);

	new_blocks = le16_to_cpu(el->l_tree_depth);

	/* allocate the number of new eb blocks we need */
	new_eb_bhs = kcalloc(new_blocks, sizeof(struct buffer_head *),
			     GFP_KERNEL);
	if (!new_eb_bhs) {
		status = -ENOMEM;
		mlog_errno(status);
		goto bail;
	}

	status = ocfs2_create_new_meta_bhs(osb, handle, inode, new_blocks,
					   meta_ac, new_eb_bhs);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	eb = (struct ocfs2_extent_block *)last_eb_bh->b_data;
	new_cpos = ocfs2_sum_rightmost_rec(&eb->h_list);

	/* Note: new_eb_bhs[new_blocks - 1] is the guy which will be
	 * linked with the rest of the tree.
	 * conversly, new_eb_bhs[0] is the new bottommost leaf.
	 *
	 * when we leave the loop, new_last_eb_blk will point to the
	 * newest leaf, and next_blkno will point to the topmost extent
	 * block. */
	next_blkno = new_last_eb_blk = 0;
	for(i = 0; i < new_blocks; i++) {
		bh = new_eb_bhs[i];
		eb = (struct ocfs2_extent_block *) bh->b_data;
		if (!OCFS2_IS_VALID_EXTENT_BLOCK(eb)) {
			OCFS2_RO_ON_INVALID_EXTENT_BLOCK(inode->i_sb, eb);
			status = -EIO;
			goto bail;
		}
		eb_el = &eb->h_list;

		status = ocfs2_journal_access(handle, inode, bh,
					      OCFS2_JOURNAL_ACCESS_CREATE);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}

		eb->h_next_leaf_blk = 0;
		eb_el->l_tree_depth = cpu_to_le16(i);
		eb_el->l_next_free_rec = cpu_to_le16(1);
		/*
		 * This actually counts as an empty extent as
		 * c_clusters == 0
		 */
		eb_el->l_recs[0].e_cpos = cpu_to_le32(new_cpos);
		eb_el->l_recs[0].e_blkno = cpu_to_le64(next_blkno);
		eb_el->l_recs[0].e_clusters = cpu_to_le32(0);
		if (!eb_el->l_tree_depth)
			new_last_eb_blk = le64_to_cpu(eb->h_blkno);

		status = ocfs2_journal_dirty(handle, bh);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}

		next_blkno = le64_to_cpu(eb->h_blkno);
	}

	/* This is a bit hairy. We want to update up to three blocks
	 * here without leaving any of them in an inconsistent state
	 * in case of error. We don't have to worry about
	 * journal_dirty erroring as it won't unless we've aborted the
	 * handle (in which case we would never be here) so reserving
	 * the write with journal_access is all we need to do. */
	status = ocfs2_journal_access(handle, inode, last_eb_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}
	status = ocfs2_journal_access(handle, inode, fe_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}
	if (eb_bh) {
		status = ocfs2_journal_access(handle, inode, eb_bh,
					      OCFS2_JOURNAL_ACCESS_WRITE);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}
	}

	/* Link the new branch into the rest of the tree (el will
	 * either be on the fe, or the extent block passed in. */
	i = le16_to_cpu(el->l_next_free_rec);
	el->l_recs[i].e_blkno = cpu_to_le64(next_blkno);
	el->l_recs[i].e_cpos = cpu_to_le32(new_cpos);
	el->l_recs[i].e_clusters = 0;
	le16_add_cpu(&el->l_next_free_rec, 1);

	/* fe needs a new last extent block pointer, as does the
	 * next_leaf on the previously last-extent-block. */
	fe->i_last_eb_blk = cpu_to_le64(new_last_eb_blk);

	eb = (struct ocfs2_extent_block *) last_eb_bh->b_data;
	eb->h_next_leaf_blk = cpu_to_le64(new_last_eb_blk);

	status = ocfs2_journal_dirty(handle, last_eb_bh);
	if (status < 0)
		mlog_errno(status);
	status = ocfs2_journal_dirty(handle, fe_bh);
	if (status < 0)
		mlog_errno(status);
	if (eb_bh) {
		status = ocfs2_journal_dirty(handle, eb_bh);
		if (status < 0)
			mlog_errno(status);
	}

	status = 0;
bail:
	if (new_eb_bhs) {
		for (i = 0; i < new_blocks; i++)
			if (new_eb_bhs[i])
				brelse(new_eb_bhs[i]);
		kfree(new_eb_bhs);
	}

	mlog_exit(status);
	return status;
}

/*
 * adds another level to the allocation tree.
 * returns back the new extent block so you can add a branch to it
 * after this call.
 */
static int ocfs2_shift_tree_depth(struct ocfs2_super *osb,
				  handle_t *handle,
				  struct inode *inode,
				  struct buffer_head *fe_bh,
				  struct ocfs2_alloc_context *meta_ac,
				  struct buffer_head **ret_new_eb_bh)
{
	int status, i;
	u32 new_clusters;
	struct buffer_head *new_eb_bh = NULL;
	struct ocfs2_dinode *fe;
	struct ocfs2_extent_block *eb;
	struct ocfs2_extent_list  *fe_el;
	struct ocfs2_extent_list  *eb_el;

	mlog_entry_void();

	status = ocfs2_create_new_meta_bhs(osb, handle, inode, 1, meta_ac,
					   &new_eb_bh);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	eb = (struct ocfs2_extent_block *) new_eb_bh->b_data;
	if (!OCFS2_IS_VALID_EXTENT_BLOCK(eb)) {
		OCFS2_RO_ON_INVALID_EXTENT_BLOCK(inode->i_sb, eb);
		status = -EIO;
		goto bail;
	}

	eb_el = &eb->h_list;
	fe = (struct ocfs2_dinode *) fe_bh->b_data;
	fe_el = &fe->id2.i_list;

	status = ocfs2_journal_access(handle, inode, new_eb_bh,
				      OCFS2_JOURNAL_ACCESS_CREATE);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	/* copy the fe data into the new extent block */
	eb_el->l_tree_depth = fe_el->l_tree_depth;
	eb_el->l_next_free_rec = fe_el->l_next_free_rec;
	for(i = 0; i < le16_to_cpu(fe_el->l_next_free_rec); i++) {
		eb_el->l_recs[i].e_cpos = fe_el->l_recs[i].e_cpos;
		eb_el->l_recs[i].e_clusters = fe_el->l_recs[i].e_clusters;
		eb_el->l_recs[i].e_blkno = fe_el->l_recs[i].e_blkno;
	}

	status = ocfs2_journal_dirty(handle, new_eb_bh);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	status = ocfs2_journal_access(handle, inode, fe_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	new_clusters = ocfs2_sum_rightmost_rec(eb_el);

	/* update fe now */
	le16_add_cpu(&fe_el->l_tree_depth, 1);
	fe_el->l_recs[0].e_cpos = 0;
	fe_el->l_recs[0].e_blkno = eb->h_blkno;
	fe_el->l_recs[0].e_clusters = cpu_to_le32(new_clusters);
	for(i = 1; i < le16_to_cpu(fe_el->l_next_free_rec); i++) {
		fe_el->l_recs[i].e_cpos = 0;
		fe_el->l_recs[i].e_clusters = 0;
		fe_el->l_recs[i].e_blkno = 0;
	}
	fe_el->l_next_free_rec = cpu_to_le16(1);

	/* If this is our 1st tree depth shift, then last_eb_blk
	 * becomes the allocated extent block */
	if (fe_el->l_tree_depth == cpu_to_le16(1))
		fe->i_last_eb_blk = eb->h_blkno;

	status = ocfs2_journal_dirty(handle, fe_bh);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	*ret_new_eb_bh = new_eb_bh;
	new_eb_bh = NULL;
	status = 0;
bail:
	if (new_eb_bh)
		brelse(new_eb_bh);

	mlog_exit(status);
	return status;
}

/*
 * Should only be called when there is no space left in any of the
 * leaf nodes. What we want to do is find the lowest tree depth
 * non-leaf extent block with room for new records. There are three
 * valid results of this search:
 *
 * 1) a lowest extent block is found, then we pass it back in
 *    *lowest_eb_bh and return '0'
 *
 * 2) the search fails to find anything, but the dinode has room. We
 *    pass NULL back in *lowest_eb_bh, but still return '0'
 *
 * 3) the search fails to find anything AND the dinode is full, in
 *    which case we return > 0
 *
 * return status < 0 indicates an error.
 */
static int ocfs2_find_branch_target(struct ocfs2_super *osb,
				    struct inode *inode,
				    struct buffer_head *fe_bh,
				    struct buffer_head **target_bh)
{
	int status = 0, i;
	u64 blkno;
	struct ocfs2_dinode *fe;
	struct ocfs2_extent_block *eb;
	struct ocfs2_extent_list  *el;
	struct buffer_head *bh = NULL;
	struct buffer_head *lowest_bh = NULL;

	mlog_entry_void();

	*target_bh = NULL;

	fe = (struct ocfs2_dinode *) fe_bh->b_data;
	el = &fe->id2.i_list;

	while(le16_to_cpu(el->l_tree_depth) > 1) {
		if (le16_to_cpu(el->l_next_free_rec) == 0) {
			ocfs2_error(inode->i_sb, "Dinode %llu has empty "
				    "extent list (next_free_rec == 0)",
				    (unsigned long long)OCFS2_I(inode)->ip_blkno);
			status = -EIO;
			goto bail;
		}
		i = le16_to_cpu(el->l_next_free_rec) - 1;
		blkno = le64_to_cpu(el->l_recs[i].e_blkno);
		if (!blkno) {
			ocfs2_error(inode->i_sb, "Dinode %llu has extent "
				    "list where extent # %d has no physical "
				    "block start",
				    (unsigned long long)OCFS2_I(inode)->ip_blkno, i);
			status = -EIO;
			goto bail;
		}

		if (bh) {
			brelse(bh);
			bh = NULL;
		}

		status = ocfs2_read_block(osb, blkno, &bh, OCFS2_BH_CACHED,
					  inode);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}

		eb = (struct ocfs2_extent_block *) bh->b_data;
		if (!OCFS2_IS_VALID_EXTENT_BLOCK(eb)) {
			OCFS2_RO_ON_INVALID_EXTENT_BLOCK(inode->i_sb, eb);
			status = -EIO;
			goto bail;
		}
		el = &eb->h_list;

		if (le16_to_cpu(el->l_next_free_rec) <
		    le16_to_cpu(el->l_count)) {
			if (lowest_bh)
				brelse(lowest_bh);
			lowest_bh = bh;
			get_bh(lowest_bh);
		}
	}

	/* If we didn't find one and the fe doesn't have any room,
	 * then return '1' */
	if (!lowest_bh
	    && (fe->id2.i_list.l_next_free_rec == fe->id2.i_list.l_count))
		status = 1;

	*target_bh = lowest_bh;
bail:
	if (bh)
		brelse(bh);

	mlog_exit(status);
	return status;
}

static inline int ocfs2_is_empty_extent(struct ocfs2_extent_rec *rec)
{
	return !rec->e_clusters;
}

/*
 * This function will discard the rightmost extent record.
 */
static void ocfs2_shift_records_right(struct ocfs2_extent_list *el)
{
	int next_free = le16_to_cpu(el->l_next_free_rec);
	int count = le16_to_cpu(el->l_count);
	unsigned int num_bytes;

	BUG_ON(!next_free);
	/* This will cause us to go off the end of our extent list. */
	BUG_ON(next_free >= count);

	num_bytes = sizeof(struct ocfs2_extent_rec) * next_free;

	memmove(&el->l_recs[1], &el->l_recs[0], num_bytes);
}

static void ocfs2_rotate_leaf(struct ocfs2_extent_list *el,
			      struct ocfs2_extent_rec *insert_rec)
{
	int i, insert_index, next_free, has_empty, num_bytes;
	u32 insert_cpos = le32_to_cpu(insert_rec->e_cpos);
	struct ocfs2_extent_rec *rec;

	next_free = le16_to_cpu(el->l_next_free_rec);
	has_empty = ocfs2_is_empty_extent(&el->l_recs[0]);

	BUG_ON(!next_free);

	/* The tree code before us didn't allow enough room in the leaf. */
	if (el->l_next_free_rec == el->l_count && !has_empty)
		BUG();

	/*
	 * The easiest way to approach this is to just remove the
	 * empty extent and temporarily decrement next_free.
	 */
	if (has_empty) {
		/*
		 * If next_free was 1 (only an empty extent), this
		 * loop won't execute, which is fine. We still want
		 * the decrement above to happen.
		 */
		for(i = 0; i < (next_free - 1); i++)
			el->l_recs[i] = el->l_recs[i+1];

		next_free--;
	}

	/*
	 * Figure out what the new record index should be.
	 */
	for(i = 0; i < next_free; i++) {
		rec = &el->l_recs[i];

		if (insert_cpos < le32_to_cpu(rec->e_cpos))
			break;
	}
	insert_index = i;

	mlog(0, "ins %u: index %d, has_empty %d, next_free %d, count %d\n",
	     insert_cpos, insert_index, has_empty, next_free, le16_to_cpu(el->l_count));

	BUG_ON(insert_index < 0);
	BUG_ON(insert_index >= le16_to_cpu(el->l_count));
	BUG_ON(insert_index > next_free);

	/*
	 * No need to memmove if we're just adding to the tail.
	 */
	if (insert_index != next_free) {
		BUG_ON(next_free >= le16_to_cpu(el->l_count));

		num_bytes = next_free - insert_index;
		num_bytes *= sizeof(struct ocfs2_extent_rec);
		memmove(&el->l_recs[insert_index + 1],
			&el->l_recs[insert_index],
			num_bytes);
	}

	/*
	 * Either we had an empty extent, and need to re-increment or
	 * there was no empty extent on a non full rightmost leaf node,
	 * in which case we still need to increment.
	 */
	next_free++;
	el->l_next_free_rec = cpu_to_le16(next_free);
	/*
	 * Make sure none of the math above just messed up our tree.
	 */
	BUG_ON(le16_to_cpu(el->l_next_free_rec) > le16_to_cpu(el->l_count));

	el->l_recs[insert_index] = *insert_rec;

}

/*
 * Create an empty extent record .
 *
 * l_next_free_rec may be updated.
 *
 * If an empty extent already exists do nothing.
 */
static void ocfs2_create_empty_extent(struct ocfs2_extent_list *el)
{
	int next_free = le16_to_cpu(el->l_next_free_rec);

	if (next_free == 0)
		goto set_and_inc;

	if (ocfs2_is_empty_extent(&el->l_recs[0]))
		return;

	mlog_bug_on_msg(el->l_count == el->l_next_free_rec,
			"Asked to create an empty extent in a full list:\n"
			"count = %u, tree depth = %u",
			le16_to_cpu(el->l_count),
			le16_to_cpu(el->l_tree_depth));

	ocfs2_shift_records_right(el);

set_and_inc:
	le16_add_cpu(&el->l_next_free_rec, 1);
	memset(&el->l_recs[0], 0, sizeof(struct ocfs2_extent_rec));
}

/*
 * For a rotation which involves two leaf nodes, the "root node" is
 * the lowest level tree node which contains a path to both leafs. This
 * resulting set of information can be used to form a complete "subtree"
 *
 * This function is passed two full paths from the dinode down to a
 * pair of adjacent leaves. It's task is to figure out which path
 * index contains the subtree root - this can be the root index itself
 * in a worst-case rotation.
 *
 * The array index of the subtree root is passed back.
 */
static int ocfs2_find_subtree_root(struct inode *inode,
				   struct ocfs2_path *left,
				   struct ocfs2_path *right)
{
	int i = 0;

	/*
	 * Check that the caller passed in two paths from the same tree.
	 */
	BUG_ON(path_root_bh(left) != path_root_bh(right));

	do {
		i++;

		/*
		 * The caller didn't pass two adjacent paths.
		 */
		mlog_bug_on_msg(i > left->p_tree_depth,
				"Inode %lu, left depth %u, right depth %u\n"
				"left leaf blk %llu, right leaf blk %llu\n",
				inode->i_ino, left->p_tree_depth,
				right->p_tree_depth,
				(unsigned long long)path_leaf_bh(left)->b_blocknr,
				(unsigned long long)path_leaf_bh(right)->b_blocknr);
	} while (left->p_node[i].bh->b_blocknr ==
		 right->p_node[i].bh->b_blocknr);

	return i - 1;
}

typedef void (path_insert_t)(void *, struct buffer_head *);

/*
 * Traverse a btree path in search of cpos, starting at root_el.
 *
 * This code can be called with a cpos larger than the tree, in which
 * case it will return the rightmost path.
 */
static int __ocfs2_find_path(struct inode *inode,
			     struct ocfs2_extent_list *root_el, u32 cpos,
			     path_insert_t *func, void *data)
{
	int i, ret = 0;
	u32 range;
	u64 blkno;
	struct buffer_head *bh = NULL;
	struct ocfs2_extent_block *eb;
	struct ocfs2_extent_list *el;
	struct ocfs2_extent_rec *rec;
	struct ocfs2_inode_info *oi = OCFS2_I(inode);

	el = root_el;
	while (el->l_tree_depth) {
		if (le16_to_cpu(el->l_next_free_rec) == 0) {
			ocfs2_error(inode->i_sb,
				    "Inode %llu has empty extent list at "
				    "depth %u\n",
				    (unsigned long long)oi->ip_blkno,
				    le16_to_cpu(el->l_tree_depth));
			ret = -EROFS;
			goto out;

		}

		for(i = 0; i < le16_to_cpu(el->l_next_free_rec) - 1; i++) {
			rec = &el->l_recs[i];

			/*
			 * In the case that cpos is off the allocation
			 * tree, this should just wind up returning the
			 * rightmost record.
			 */
			range = le32_to_cpu(rec->e_cpos) +
				le32_to_cpu(rec->e_clusters);
			if (cpos >= le32_to_cpu(rec->e_cpos) && cpos < range)
			    break;
		}

		blkno = le64_to_cpu(el->l_recs[i].e_blkno);
		if (blkno == 0) {
			ocfs2_error(inode->i_sb,
				    "Inode %llu has bad blkno in extent list "
				    "at depth %u (index %d)\n",
				    (unsigned long long)oi->ip_blkno,
				    le16_to_cpu(el->l_tree_depth), i);
			ret = -EROFS;
			goto out;
		}

		brelse(bh);
		bh = NULL;
		ret = ocfs2_read_block(OCFS2_SB(inode->i_sb), blkno,
				       &bh, OCFS2_BH_CACHED, inode);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		eb = (struct ocfs2_extent_block *) bh->b_data;
		el = &eb->h_list;
		if (!OCFS2_IS_VALID_EXTENT_BLOCK(eb)) {
			OCFS2_RO_ON_INVALID_EXTENT_BLOCK(inode->i_sb, eb);
			ret = -EIO;
			goto out;
		}

		if (le16_to_cpu(el->l_next_free_rec) >
		    le16_to_cpu(el->l_count)) {
			ocfs2_error(inode->i_sb,
				    "Inode %llu has bad count in extent list "
				    "at block %llu (next free=%u, count=%u)\n",
				    (unsigned long long)oi->ip_blkno,
				    (unsigned long long)bh->b_blocknr,
				    le16_to_cpu(el->l_next_free_rec),
				    le16_to_cpu(el->l_count));
			ret = -EROFS;
			goto out;
		}

		if (func)
			func(data, bh);
	}

out:
	/*
	 * Catch any trailing bh that the loop didn't handle.
	 */
	brelse(bh);

	return ret;
}

/*
 * Given an initialized path (that is, it has a valid root extent
 * list), this function will traverse the btree in search of the path
 * which would contain cpos.
 *
 * The path traveled is recorded in the path structure.
 *
 * Note that this will not do any comparisons on leaf node extent
 * records, so it will work fine in the case that we just added a tree
 * branch.
 */
struct find_path_data {
	int index;
	struct ocfs2_path *path;
};
static void find_path_ins(void *data, struct buffer_head *bh)
{
	struct find_path_data *fp = data;

	get_bh(bh);
	ocfs2_path_insert_eb(fp->path, fp->index, bh);
	fp->index++;
}
static int ocfs2_find_path(struct inode *inode, struct ocfs2_path *path,
			   u32 cpos)
{
	struct find_path_data data;

	data.index = 1;
	data.path = path;
	return __ocfs2_find_path(inode, path_root_el(path), cpos,
				 find_path_ins, &data);
}

static void find_leaf_ins(void *data, struct buffer_head *bh)
{
	struct ocfs2_extent_block *eb =(struct ocfs2_extent_block *)bh->b_data;
	struct ocfs2_extent_list *el = &eb->h_list;
	struct buffer_head **ret = data;

	/* We want to retain only the leaf block. */
	if (le16_to_cpu(el->l_tree_depth) == 0) {
		get_bh(bh);
		*ret = bh;
	}
}
/*
 * Find the leaf block in the tree which would contain cpos. No
 * checking of the actual leaf is done.
 *
 * Some paths want to call this instead of allocating a path structure
 * and calling ocfs2_find_path().
 *
 * This function doesn't handle non btree extent lists.
 */
static int ocfs2_find_leaf(struct inode *inode,
			   struct ocfs2_extent_list *root_el, u32 cpos,
			   struct buffer_head **leaf_bh)
{
	int ret;
	struct buffer_head *bh = NULL;

	ret = __ocfs2_find_path(inode, root_el, cpos, find_leaf_ins, &bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	*leaf_bh = bh;
out:
	return ret;
}

/*
 * Adjust the adjacent records (left_rec, right_rec) involved in a rotation.
 *
 * Basically, we've moved stuff around at the bottom of the tree and
 * we need to fix up the extent records above the changes to reflect
 * the new changes.
 *
 * left_rec: the record on the left.
 * left_child_el: is the child list pointed to by left_rec
 * right_rec: the record to the right of left_rec
 * right_child_el: is the child list pointed to by right_rec
 *
 * By definition, this only works on interior nodes.
 */
static void ocfs2_adjust_adjacent_records(struct ocfs2_extent_rec *left_rec,
				  struct ocfs2_extent_list *left_child_el,
				  struct ocfs2_extent_rec *right_rec,
				  struct ocfs2_extent_list *right_child_el)
{
	u32 left_clusters, right_end;

	/*
	 * Interior nodes never have holes. Their cpos is the cpos of
	 * the leftmost record in their child list. Their cluster
	 * count covers the full theoretical range of their child list
	 * - the range between their cpos and the cpos of the record
	 * immediately to their right.
	 */
	left_clusters = le32_to_cpu(right_child_el->l_recs[0].e_cpos);
	left_clusters -= le32_to_cpu(left_rec->e_cpos);
	left_rec->e_clusters = cpu_to_le32(left_clusters);

	/*
	 * Calculate the rightmost cluster count boundary before
	 * moving cpos - we will need to adjust e_clusters after
	 * updating e_cpos to keep the same highest cluster count.
	 */
	right_end = le32_to_cpu(right_rec->e_cpos);
	right_end += le32_to_cpu(right_rec->e_clusters);

	right_rec->e_cpos = left_rec->e_cpos;
	le32_add_cpu(&right_rec->e_cpos, left_clusters);

	right_end -= le32_to_cpu(right_rec->e_cpos);
	right_rec->e_clusters = cpu_to_le32(right_end);
}

/*
 * Adjust the adjacent root node records involved in a
 * rotation. left_el_blkno is passed in as a key so that we can easily
 * find it's index in the root list.
 */
static void ocfs2_adjust_root_records(struct ocfs2_extent_list *root_el,
				      struct ocfs2_extent_list *left_el,
				      struct ocfs2_extent_list *right_el,
				      u64 left_el_blkno)
{
	int i;

	BUG_ON(le16_to_cpu(root_el->l_tree_depth) <=
	       le16_to_cpu(left_el->l_tree_depth));

	for(i = 0; i < le16_to_cpu(root_el->l_next_free_rec) - 1; i++) {
		if (le64_to_cpu(root_el->l_recs[i].e_blkno) == left_el_blkno)
			break;
	}

	/*
	 * The path walking code should have never returned a root and
	 * two paths which are not adjacent.
	 */
	BUG_ON(i >= (le16_to_cpu(root_el->l_next_free_rec) - 1));

	ocfs2_adjust_adjacent_records(&root_el->l_recs[i], left_el,
				      &root_el->l_recs[i + 1], right_el);
}

/*
 * We've changed a leaf block (in right_path) and need to reflect that
 * change back up the subtree.
 *
 * This happens in multiple places:
 *   - When we've moved an extent record from the left path leaf to the right
 *     path leaf to make room for an empty extent in the left path leaf.
 *   - When our insert into the right path leaf is at the leftmost edge
 *     and requires an update of the path immediately to it's left. This
 *     can occur at the end of some types of rotation and appending inserts.
 */
static void ocfs2_complete_edge_insert(struct inode *inode, handle_t *handle,
				       struct ocfs2_path *left_path,
				       struct ocfs2_path *right_path,
				       int subtree_index)
{
	int ret, i, idx;
	struct ocfs2_extent_list *el, *left_el, *right_el;
	struct ocfs2_extent_rec *left_rec, *right_rec;
	struct buffer_head *root_bh = left_path->p_node[subtree_index].bh;

	/*
	 * Update the counts and position values within all the
	 * interior nodes to reflect the leaf rotation we just did.
	 *
	 * The root node is handled below the loop.
	 *
	 * We begin the loop with right_el and left_el pointing to the
	 * leaf lists and work our way up.
	 *
	 * NOTE: within this loop, left_el and right_el always refer
	 * to the *child* lists.
	 */
	left_el = path_leaf_el(left_path);
	right_el = path_leaf_el(right_path);
	for(i = left_path->p_tree_depth - 1; i > subtree_index; i--) {
		mlog(0, "Adjust records at index %u\n", i);

		/*
		 * One nice property of knowing that all of these
		 * nodes are below the root is that we only deal with
		 * the leftmost right node record and the rightmost
		 * left node record.
		 */
		el = left_path->p_node[i].el;
		idx = le16_to_cpu(left_el->l_next_free_rec) - 1;
		left_rec = &el->l_recs[idx];

		el = right_path->p_node[i].el;
		right_rec = &el->l_recs[0];

		ocfs2_adjust_adjacent_records(left_rec, left_el, right_rec,
					      right_el);

		ret = ocfs2_journal_dirty(handle, left_path->p_node[i].bh);
		if (ret)
			mlog_errno(ret);

		ret = ocfs2_journal_dirty(handle, right_path->p_node[i].bh);
		if (ret)
			mlog_errno(ret);

		/*
		 * Setup our list pointers now so that the current
		 * parents become children in the next iteration.
		 */
		left_el = left_path->p_node[i].el;
		right_el = right_path->p_node[i].el;
	}

	/*
	 * At the root node, adjust the two adjacent records which
	 * begin our path to the leaves.
	 */

	el = left_path->p_node[subtree_index].el;
	left_el = left_path->p_node[subtree_index + 1].el;
	right_el = right_path->p_node[subtree_index + 1].el;

	ocfs2_adjust_root_records(el, left_el, right_el,
				  left_path->p_node[subtree_index + 1].bh->b_blocknr);

	root_bh = left_path->p_node[subtree_index].bh;

	ret = ocfs2_journal_dirty(handle, root_bh);
	if (ret)
		mlog_errno(ret);
}

static int ocfs2_rotate_subtree_right(struct inode *inode,
				      handle_t *handle,
				      struct ocfs2_path *left_path,
				      struct ocfs2_path *right_path,
				      int subtree_index)
{
	int ret, i;
	struct buffer_head *right_leaf_bh;
	struct buffer_head *left_leaf_bh = NULL;
	struct buffer_head *root_bh;
	struct ocfs2_extent_list *right_el, *left_el;
	struct ocfs2_extent_rec move_rec;

	left_leaf_bh = path_leaf_bh(left_path);
	left_el = path_leaf_el(left_path);

	if (left_el->l_next_free_rec != left_el->l_count) {
		ocfs2_error(inode->i_sb,
			    "Inode %llu has non-full interior leaf node %llu"
			    "(next free = %u)",
			    (unsigned long long)OCFS2_I(inode)->ip_blkno,
			    (unsigned long long)left_leaf_bh->b_blocknr,
			    le16_to_cpu(left_el->l_next_free_rec));
		return -EROFS;
	}

	/*
	 * This extent block may already have an empty record, so we
	 * return early if so.
	 */
	if (ocfs2_is_empty_extent(&left_el->l_recs[0]))
		return 0;

	root_bh = left_path->p_node[subtree_index].bh;
	BUG_ON(root_bh != right_path->p_node[subtree_index].bh);

	ret = ocfs2_journal_access(handle, inode, root_bh,
				   OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	for(i = subtree_index + 1; i < path_num_items(right_path); i++) {
		ret = ocfs2_journal_access(handle, inode,
					   right_path->p_node[i].bh,
					   OCFS2_JOURNAL_ACCESS_WRITE);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		ret = ocfs2_journal_access(handle, inode,
					   left_path->p_node[i].bh,
					   OCFS2_JOURNAL_ACCESS_WRITE);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
	}

	right_leaf_bh = path_leaf_bh(right_path);
	right_el = path_leaf_el(right_path);

	/* This is a code error, not a disk corruption. */
	mlog_bug_on_msg(!right_el->l_next_free_rec, "Inode %llu: Rotate fails "
			"because rightmost leaf block %llu is empty\n",
			(unsigned long long)OCFS2_I(inode)->ip_blkno,
			(unsigned long long)right_leaf_bh->b_blocknr);

	ocfs2_create_empty_extent(right_el);

	ret = ocfs2_journal_dirty(handle, right_leaf_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	/* Do the copy now. */
	i = le16_to_cpu(left_el->l_next_free_rec) - 1;
	move_rec = left_el->l_recs[i];
	right_el->l_recs[0] = move_rec;

	/*
	 * Clear out the record we just copied and shift everything
	 * over, leaving an empty extent in the left leaf.
	 *
	 * We temporarily subtract from next_free_rec so that the
	 * shift will lose the tail record (which is now defunct).
	 */
	le16_add_cpu(&left_el->l_next_free_rec, -1);
	ocfs2_shift_records_right(left_el);
	memset(&left_el->l_recs[0], 0, sizeof(struct ocfs2_extent_rec));
	le16_add_cpu(&left_el->l_next_free_rec, 1);

	ret = ocfs2_journal_dirty(handle, left_leaf_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ocfs2_complete_edge_insert(inode, handle, left_path, right_path,
				subtree_index);

out:
	return ret;
}

/*
 * Given a full path, determine what cpos value would return us a path
 * containing the leaf immediately to the left of the current one.
 *
 * Will return zero if the path passed in is already the leftmost path.
 */
static int ocfs2_find_cpos_for_left_leaf(struct super_block *sb,
					 struct ocfs2_path *path, u32 *cpos)
{
	int i, j, ret = 0;
	u64 blkno;
	struct ocfs2_extent_list *el;

	*cpos = 0;

	blkno = path_leaf_bh(path)->b_blocknr;

	/* Start at the tree node just above the leaf and work our way up. */
	i = path->p_tree_depth - 1;
	while (i >= 0) {
		el = path->p_node[i].el;

		/*
		 * Find the extent record just before the one in our
		 * path.
		 */
		for(j = 0; j < le16_to_cpu(el->l_next_free_rec); j++) {
			if (le64_to_cpu(el->l_recs[j].e_blkno) == blkno) {
				if (j == 0) {
					if (i == 0) {
						/*
						 * We've determined that the
						 * path specified is already
						 * the leftmost one - return a
						 * cpos of zero.
						 */
						goto out;
					}
					/*
					 * The leftmost record points to our
					 * leaf - we need to travel up the
					 * tree one level.
					 */
					goto next_node;
				}

				*cpos = le32_to_cpu(el->l_recs[j - 1].e_cpos);
				*cpos = *cpos + le32_to_cpu(el->l_recs[j - 1].e_clusters) - 1;
				goto out;
			}
		}

		/*
		 * If we got here, we never found a valid node where
		 * the tree indicated one should be.
		 */
		ocfs2_error(sb,
			    "Invalid extent tree at extent block %llu\n",
			    (unsigned long long)blkno);
		ret = -EROFS;
		goto out;

next_node:
		blkno = path->p_node[i].bh->b_blocknr;
		i--;
	}

out:
	return ret;
}

static int ocfs2_extend_rotate_transaction(handle_t *handle, int subtree_depth,
					   struct ocfs2_path *path)
{
	int credits = (path->p_tree_depth - subtree_depth) * 2 + 1;

	if (handle->h_buffer_credits < credits)
		return ocfs2_extend_trans(handle, credits);

	return 0;
}

/*
 * Trap the case where we're inserting into the theoretical range past
 * the _actual_ left leaf range. Otherwise, we'll rotate a record
 * whose cpos is less than ours into the right leaf.
 *
 * It's only necessary to look at the rightmost record of the left
 * leaf because the logic that calls us should ensure that the
 * theoretical ranges in the path components above the leaves are
 * correct.
 */
static int ocfs2_rotate_requires_path_adjustment(struct ocfs2_path *left_path,
						 u32 insert_cpos)
{
	struct ocfs2_extent_list *left_el;
	struct ocfs2_extent_rec *rec;
	int next_free;

	left_el = path_leaf_el(left_path);
	next_free = le16_to_cpu(left_el->l_next_free_rec);
	rec = &left_el->l_recs[next_free - 1];

	if (insert_cpos > le32_to_cpu(rec->e_cpos))
		return 1;
	return 0;
}

/*
 * Rotate all the records in a btree right one record, starting at insert_cpos.
 *
 * The path to the rightmost leaf should be passed in.
 *
 * The array is assumed to be large enough to hold an entire path (tree depth).
 *
 * Upon succesful return from this function:
 *
 * - The 'right_path' array will contain a path to the leaf block
 *   whose range contains e_cpos.
 * - That leaf block will have a single empty extent in list index 0.
 * - In the case that the rotation requires a post-insert update,
 *   *ret_left_path will contain a valid path which can be passed to
 *   ocfs2_insert_path().
 */
static int ocfs2_rotate_tree_right(struct inode *inode,
				   handle_t *handle,
				   u32 insert_cpos,
				   struct ocfs2_path *right_path,
				   struct ocfs2_path **ret_left_path)
{
	int ret, start;
	u32 cpos;
	struct ocfs2_path *left_path = NULL;

	*ret_left_path = NULL;

	left_path = ocfs2_new_path(path_root_bh(right_path),
				   path_root_el(right_path));
	if (!left_path) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_find_cpos_for_left_leaf(inode->i_sb, right_path, &cpos);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	mlog(0, "Insert: %u, first left path cpos: %u\n", insert_cpos, cpos);

	/*
	 * What we want to do here is:
	 *
	 * 1) Start with the rightmost path.
	 *
	 * 2) Determine a path to the leaf block directly to the left
	 *    of that leaf.
	 *
	 * 3) Determine the 'subtree root' - the lowest level tree node
	 *    which contains a path to both leaves.
	 *
	 * 4) Rotate the subtree.
	 *
	 * 5) Find the next subtree by considering the left path to be
	 *    the new right path.
	 *
	 * The check at the top of this while loop also accepts
	 * insert_cpos == cpos because cpos is only a _theoretical_
	 * value to get us the left path - insert_cpos might very well
	 * be filling that hole.
	 *
	 * Stop at a cpos of '0' because we either started at the
	 * leftmost branch (i.e., a tree with one branch and a
	 * rotation inside of it), or we've gone as far as we can in
	 * rotating subtrees.
	 */
	while (cpos && insert_cpos <= cpos) {
		mlog(0, "Rotating a tree: ins. cpos: %u, left path cpos: %u\n",
		     insert_cpos, cpos);

		ret = ocfs2_find_path(inode, left_path, cpos);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		mlog_bug_on_msg(path_leaf_bh(left_path) ==
				path_leaf_bh(right_path),
				"Inode %lu: error during insert of %u "
				"(left path cpos %u) results in two identical "
				"paths ending at %llu\n",
				inode->i_ino, insert_cpos, cpos,
				(unsigned long long)
				path_leaf_bh(left_path)->b_blocknr);

		if (ocfs2_rotate_requires_path_adjustment(left_path,
							  insert_cpos)) {
			mlog(0, "Path adjustment required\n");

			/*
			 * We've rotated the tree as much as we
			 * should. The rest is up to
			 * ocfs2_insert_path() to complete, after the
			 * record insertion. We indicate this
			 * situation by returning the left path.
			 *
			 * The reason we don't adjust the records here
			 * before the record insert is that an error
			 * later might break the rule where a parent
			 * record e_cpos will reflect the actual
			 * e_cpos of the 1st nonempty record of the
			 * child list.
			 */
			*ret_left_path = left_path;
			goto out_ret_path;
		}

		start = ocfs2_find_subtree_root(inode, left_path, right_path);

		mlog(0, "Subtree root at index %d (blk %llu, depth %d)\n",
		     start,
		     (unsigned long long) right_path->p_node[start].bh->b_blocknr,
		     right_path->p_tree_depth);

		ret = ocfs2_extend_rotate_transaction(handle, start,
						      right_path);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		ret = ocfs2_rotate_subtree_right(inode, handle, left_path,
						 right_path, start);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		/*
		 * There is no need to re-read the next right path
		 * as we know that it'll be our current left
		 * path. Optimize by copying values instead.
		 */
		ocfs2_mv_path(right_path, left_path);

		ret = ocfs2_find_cpos_for_left_leaf(inode->i_sb, right_path,
						    &cpos);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
	}

out:
	ocfs2_free_path(left_path);

out_ret_path:
	return ret;
}

/*
 * Do the final bits of extent record insertion at the target leaf
 * list. If this leaf is part of an allocation tree, it is assumed
 * that the tree above has been prepared.
 */
static void ocfs2_insert_at_leaf(struct ocfs2_extent_rec *insert_rec,
				 struct ocfs2_extent_list *el,
				 struct ocfs2_insert_type *insert,
				 struct inode *inode)
{
	int i = insert->ins_contig_index;
	unsigned int range;
	struct ocfs2_extent_rec *rec;

	BUG_ON(el->l_tree_depth);

	/*
	 * Contiguous insert - either left or right.
	 */
	if (insert->ins_contig != CONTIG_NONE) {
		rec = &el->l_recs[i];
		if (insert->ins_contig == CONTIG_LEFT) {
			rec->e_blkno = insert_rec->e_blkno;
			rec->e_cpos = insert_rec->e_cpos;
		}
		le32_add_cpu(&rec->e_clusters,
			     le32_to_cpu(insert_rec->e_clusters));
		return;
	}

	/*
	 * Handle insert into an empty leaf.
	 */
	if (le16_to_cpu(el->l_next_free_rec) == 0 ||
	    ((le16_to_cpu(el->l_next_free_rec) == 1) &&
	     ocfs2_is_empty_extent(&el->l_recs[0]))) {
		el->l_recs[0] = *insert_rec;
		el->l_next_free_rec = cpu_to_le16(1);
		return;
	}

	/*
	 * Appending insert.
	 */
	if (insert->ins_appending == APPEND_TAIL) {
		i = le16_to_cpu(el->l_next_free_rec) - 1;
		rec = &el->l_recs[i];
		range = le32_to_cpu(rec->e_cpos) + le32_to_cpu(rec->e_clusters);
		BUG_ON(le32_to_cpu(insert_rec->e_cpos) < range);

		mlog_bug_on_msg(le16_to_cpu(el->l_next_free_rec) >=
				le16_to_cpu(el->l_count),
				"inode %lu, depth %u, count %u, next free %u, "
				"rec.cpos %u, rec.clusters %u, "
				"insert.cpos %u, insert.clusters %u\n",
				inode->i_ino,
				le16_to_cpu(el->l_tree_depth),
				le16_to_cpu(el->l_count),
				le16_to_cpu(el->l_next_free_rec),
				le32_to_cpu(el->l_recs[i].e_cpos),
				le32_to_cpu(el->l_recs[i].e_clusters),
				le32_to_cpu(insert_rec->e_cpos),
				le32_to_cpu(insert_rec->e_clusters));
		i++;
		el->l_recs[i] = *insert_rec;
		le16_add_cpu(&el->l_next_free_rec, 1);
		return;
	}

	/*
	 * Ok, we have to rotate.
	 *
	 * At this point, it is safe to assume that inserting into an
	 * empty leaf and appending to a leaf have both been handled
	 * above.
	 *
	 * This leaf needs to have space, either by the empty 1st
	 * extent record, or by virtue of an l_next_rec < l_count.
	 */
	ocfs2_rotate_leaf(el, insert_rec);
}

static inline void ocfs2_update_dinode_clusters(struct inode *inode,
						struct ocfs2_dinode *di,
						u32 clusters)
{
	le32_add_cpu(&di->i_clusters, clusters);
	spin_lock(&OCFS2_I(inode)->ip_lock);
	OCFS2_I(inode)->ip_clusters = le32_to_cpu(di->i_clusters);
	spin_unlock(&OCFS2_I(inode)->ip_lock);
}

static int ocfs2_append_rec_to_path(struct inode *inode, handle_t *handle,
				    struct ocfs2_extent_rec *insert_rec,
				    struct ocfs2_path *right_path,
				    struct ocfs2_path **ret_left_path)
{
	int ret, i, next_free;
	struct buffer_head *bh;
	struct ocfs2_extent_list *el;
	struct ocfs2_path *left_path = NULL;

	*ret_left_path = NULL;

	/*
	 * If our appending insert is at the leftmost edge of a leaf,
	 * then we might need to update the rightmost records of the
	 * neighboring path.
	 */
	el = path_leaf_el(right_path);
	next_free = le16_to_cpu(el->l_next_free_rec);
	if (next_free == 0 ||
	    (next_free == 1 && ocfs2_is_empty_extent(&el->l_recs[0]))) {
		u32 left_cpos;

		ret = ocfs2_find_cpos_for_left_leaf(inode->i_sb, right_path,
						    &left_cpos);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		mlog(0, "Append may need a left path update. cpos: %u, "
		     "left_cpos: %u\n", le32_to_cpu(insert_rec->e_cpos),
		     left_cpos);

		/*
		 * No need to worry if the append is already in the
		 * leftmost leaf.
		 */
		if (left_cpos) {
			left_path = ocfs2_new_path(path_root_bh(right_path),
						   path_root_el(right_path));
			if (!left_path) {
				ret = -ENOMEM;
				mlog_errno(ret);
				goto out;
			}

			ret = ocfs2_find_path(inode, left_path, left_cpos);
			if (ret) {
				mlog_errno(ret);
				goto out;
			}

			/*
			 * ocfs2_insert_path() will pass the left_path to the
			 * journal for us.
			 */
		}
	}

	ret = ocfs2_journal_access_path(inode, handle, right_path);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	el = path_root_el(right_path);
	bh = path_root_bh(right_path);
	i = 0;
	while (1) {
		next_free = le16_to_cpu(el->l_next_free_rec);
		if (next_free == 0) {
			ocfs2_error(inode->i_sb,
				    "Dinode %llu has a bad extent list",
				    (unsigned long long)OCFS2_I(inode)->ip_blkno);
			ret = -EIO;
			goto out;
		}

		el->l_recs[next_free - 1].e_clusters = insert_rec->e_cpos;
		le32_add_cpu(&el->l_recs[next_free - 1].e_clusters,
			     le32_to_cpu(insert_rec->e_clusters));
		le32_add_cpu(&el->l_recs[next_free - 1].e_clusters,
			    -le32_to_cpu(el->l_recs[next_free - 1].e_cpos));

		ret = ocfs2_journal_dirty(handle, bh);
		if (ret)
			mlog_errno(ret);

		if (++i >= right_path->p_tree_depth)
			break;

		bh = right_path->p_node[i].bh;
		el = right_path->p_node[i].el;
	}

	*ret_left_path = left_path;
	ret = 0;
out:
	if (ret != 0)
		ocfs2_free_path(left_path);

	return ret;
}

/*
 * This function only does inserts on an allocation b-tree. For dinode
 * lists, ocfs2_insert_at_leaf() is called directly.
 *
 * right_path is the path we want to do the actual insert
 * in. left_path should only be passed in if we need to update that
 * portion of the tree after an edge insert.
 */
static int ocfs2_insert_path(struct inode *inode,
			     handle_t *handle,
			     struct ocfs2_path *left_path,
			     struct ocfs2_path *right_path,
			     struct ocfs2_extent_rec *insert_rec,
			     struct ocfs2_insert_type *insert)
{
	int ret, subtree_index;
	struct buffer_head *leaf_bh = path_leaf_bh(right_path);
	struct ocfs2_extent_list *el;

	/*
	 * Pass both paths to the journal. The majority of inserts
	 * will be touching all components anyway.
	 */
	ret = ocfs2_journal_access_path(inode, handle, right_path);
	if (ret < 0) {
		mlog_errno(ret);
		goto out;
	}

	if (left_path) {
		int credits = handle->h_buffer_credits;

		/*
		 * There's a chance that left_path got passed back to
		 * us without being accounted for in the
		 * journal. Extend our transaction here to be sure we
		 * can change those blocks.
		 */
		credits += left_path->p_tree_depth;

		ret = ocfs2_extend_trans(handle, credits);
		if (ret < 0) {
			mlog_errno(ret);
			goto out;
		}

		ret = ocfs2_journal_access_path(inode, handle, left_path);
		if (ret < 0) {
			mlog_errno(ret);
			goto out;
		}
	}

	el = path_leaf_el(right_path);

	ocfs2_insert_at_leaf(insert_rec, el, insert, inode);
	ret = ocfs2_journal_dirty(handle, leaf_bh);
	if (ret)
		mlog_errno(ret);

	if (left_path) {
		/*
		 * The rotate code has indicated that we need to fix
		 * up portions of the tree after the insert.
		 *
		 * XXX: Should we extend the transaction here?
		 */
		subtree_index = ocfs2_find_subtree_root(inode, left_path,
							right_path);
		ocfs2_complete_edge_insert(inode, handle, left_path,
					   right_path, subtree_index);
	}

	ret = 0;
out:
	return ret;
}

static int ocfs2_do_insert_extent(struct inode *inode,
				  handle_t *handle,
				  struct buffer_head *di_bh,
				  struct ocfs2_extent_rec *insert_rec,
				  struct ocfs2_insert_type *type)
{
	int ret, rotate = 0;
	u32 cpos;
	struct ocfs2_path *right_path = NULL;
	struct ocfs2_path *left_path = NULL;
	struct ocfs2_dinode *di;
	struct ocfs2_extent_list *el;

	di = (struct ocfs2_dinode *) di_bh->b_data;
	el = &di->id2.i_list;

	ret = ocfs2_journal_access(handle, inode, di_bh,
				   OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	if (le16_to_cpu(el->l_tree_depth) == 0) {
		ocfs2_insert_at_leaf(insert_rec, el, type, inode);
		goto out_update_clusters;
	}

	right_path = ocfs2_new_inode_path(di_bh);
	if (!right_path) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

	/*
	 * Determine the path to start with. Rotations need the
	 * rightmost path, everything else can go directly to the
	 * target leaf.
	 */
	cpos = le32_to_cpu(insert_rec->e_cpos);
	if (type->ins_appending == APPEND_NONE &&
	    type->ins_contig == CONTIG_NONE) {
		rotate = 1;
		cpos = UINT_MAX;
	}

	ret = ocfs2_find_path(inode, right_path, cpos);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	/*
	 * Rotations and appends need special treatment - they modify
	 * parts of the tree's above them.
	 *
	 * Both might pass back a path immediate to the left of the
	 * one being inserted to. This will be cause
	 * ocfs2_insert_path() to modify the rightmost records of
	 * left_path to account for an edge insert.
	 *
	 * XXX: When modifying this code, keep in mind that an insert
	 * can wind up skipping both of these two special cases...
	 */
	if (rotate) {
		ret = ocfs2_rotate_tree_right(inode, handle,
					      le32_to_cpu(insert_rec->e_cpos),
					      right_path, &left_path);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
	} else if (type->ins_appending == APPEND_TAIL
		   && type->ins_contig != CONTIG_LEFT) {
		ret = ocfs2_append_rec_to_path(inode, handle, insert_rec,
					       right_path, &left_path);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
	}

	ret = ocfs2_insert_path(inode, handle, left_path, right_path,
				insert_rec, type);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

out_update_clusters:
	ocfs2_update_dinode_clusters(inode, di,
				     le32_to_cpu(insert_rec->e_clusters));

	ret = ocfs2_journal_dirty(handle, di_bh);
	if (ret)
		mlog_errno(ret);

out:
	ocfs2_free_path(left_path);
	ocfs2_free_path(right_path);

	return ret;
}

static void ocfs2_figure_contig_type(struct inode *inode,
				     struct ocfs2_insert_type *insert,
				     struct ocfs2_extent_list *el,
				     struct ocfs2_extent_rec *insert_rec)
{
	int i;
	enum ocfs2_contig_type contig_type = CONTIG_NONE;

	for(i = 0; i < le16_to_cpu(el->l_next_free_rec); i++) {
		contig_type = ocfs2_extent_contig(inode, &el->l_recs[i],
						  insert_rec);
		if (contig_type != CONTIG_NONE) {
			insert->ins_contig_index = i;
			break;
		}
	}
	insert->ins_contig = contig_type;
}

/*
 * This should only be called against the righmost leaf extent list.
 *
 * ocfs2_figure_appending_type() will figure out whether we'll have to
 * insert at the tail of the rightmost leaf.
 *
 * This should also work against the dinode list for tree's with 0
 * depth. If we consider the dinode list to be the rightmost leaf node
 * then the logic here makes sense.
 */
static void ocfs2_figure_appending_type(struct ocfs2_insert_type *insert,
					struct ocfs2_extent_list *el,
					struct ocfs2_extent_rec *insert_rec)
{
	int i;
	u32 cpos = le32_to_cpu(insert_rec->e_cpos);
	struct ocfs2_extent_rec *rec;

	insert->ins_appending = APPEND_NONE;

	BUG_ON(el->l_tree_depth);

	if (!el->l_next_free_rec)
		goto set_tail_append;

	if (ocfs2_is_empty_extent(&el->l_recs[0])) {
		/* Were all records empty? */
		if (le16_to_cpu(el->l_next_free_rec) == 1)
			goto set_tail_append;
	}

	i = le16_to_cpu(el->l_next_free_rec) - 1;
	rec = &el->l_recs[i];

	if (cpos >= (le32_to_cpu(rec->e_cpos) + le32_to_cpu(rec->e_clusters)))
		goto set_tail_append;

	return;

set_tail_append:
	insert->ins_appending = APPEND_TAIL;
}

/*
 * Helper function called at the begining of an insert.
 *
 * This computes a few things that are commonly used in the process of
 * inserting into the btree:
 *   - Whether the new extent is contiguous with an existing one.
 *   - The current tree depth.
 *   - Whether the insert is an appending one.
 *   - The total # of free records in the tree.
 *
 * All of the information is stored on the ocfs2_insert_type
 * structure.
 */
static int ocfs2_figure_insert_type(struct inode *inode,
				    struct buffer_head *di_bh,
				    struct buffer_head **last_eb_bh,
				    struct ocfs2_extent_rec *insert_rec,
				    struct ocfs2_insert_type *insert)
{
	int ret;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	struct ocfs2_extent_block *eb;
	struct ocfs2_extent_list *el;
	struct ocfs2_path *path = NULL;
	struct buffer_head *bh = NULL;

	el = &di->id2.i_list;
	insert->ins_tree_depth = le16_to_cpu(el->l_tree_depth);

	if (el->l_tree_depth) {
		/*
		 * If we have tree depth, we read in the
		 * rightmost extent block ahead of time as
		 * ocfs2_figure_insert_type() and ocfs2_add_branch()
		 * may want it later.
		 */
		ret = ocfs2_read_block(OCFS2_SB(inode->i_sb),
				       le64_to_cpu(di->i_last_eb_blk), &bh,
				       OCFS2_BH_CACHED, inode);
		if (ret) {
			mlog_exit(ret);
			goto out;
		}
		eb = (struct ocfs2_extent_block *) bh->b_data;
		el = &eb->h_list;
	}

	/*
	 * Unless we have a contiguous insert, we'll need to know if
	 * there is room left in our allocation tree for another
	 * extent record.
	 *
	 * XXX: This test is simplistic, we can search for empty
	 * extent records too.
	 */
	insert->ins_free_records = le16_to_cpu(el->l_count) -
		le16_to_cpu(el->l_next_free_rec);

	if (!insert->ins_tree_depth) {
		ocfs2_figure_contig_type(inode, insert, el, insert_rec);
		ocfs2_figure_appending_type(insert, el, insert_rec);
		return 0;
	}

	path = ocfs2_new_inode_path(di_bh);
	if (!path) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

	/*
	 * In the case that we're inserting past what the tree
	 * currently accounts for, ocfs2_find_path() will return for
	 * us the rightmost tree path. This is accounted for below in
	 * the appending code.
	 */
	ret = ocfs2_find_path(inode, path, le32_to_cpu(insert_rec->e_cpos));
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	el = path_leaf_el(path);

	/*
	 * Now that we have the path, there's two things we want to determine:
	 * 1) Contiguousness (also set contig_index if this is so)
	 *
	 * 2) Are we doing an append? We can trivially break this up
         *     into two types of appends: simple record append, or a
         *     rotate inside the tail leaf.
	 */
	ocfs2_figure_contig_type(inode, insert, el, insert_rec);

	/*
	 * The insert code isn't quite ready to deal with all cases of
	 * left contiguousness. Specifically, if it's an insert into
	 * the 1st record in a leaf, it will require the adjustment of
	 * e_clusters on the last record of the path directly to it's
	 * left. For now, just catch that case and fool the layers
	 * above us. This works just fine for tree_depth == 0, which
	 * is why we allow that above.
	 */
	if (insert->ins_contig == CONTIG_LEFT &&
	    insert->ins_contig_index == 0)
		insert->ins_contig = CONTIG_NONE;

	/*
	 * Ok, so we can simply compare against last_eb to figure out
	 * whether the path doesn't exist. This will only happen in
	 * the case that we're doing a tail append, so maybe we can
	 * take advantage of that information somehow.
	 */
	if (le64_to_cpu(di->i_last_eb_blk) == path_leaf_bh(path)->b_blocknr) {
		/*
		 * Ok, ocfs2_find_path() returned us the rightmost
		 * tree path. This might be an appending insert. There are
		 * two cases:
		 *    1) We're doing a true append at the tail:
		 *	-This might even be off the end of the leaf
		 *    2) We're "appending" by rotating in the tail
		 */
		ocfs2_figure_appending_type(insert, el, insert_rec);
	}

out:
	ocfs2_free_path(path);

	if (ret == 0)
		*last_eb_bh = bh;
	else
		brelse(bh);
	return ret;
}

/*
 * Insert an extent into an inode btree.
 *
 * The caller needs to update fe->i_clusters
 */
int ocfs2_insert_extent(struct ocfs2_super *osb,
			handle_t *handle,
			struct inode *inode,
			struct buffer_head *fe_bh,
			u32 cpos,
			u64 start_blk,
			u32 new_clusters,
			struct ocfs2_alloc_context *meta_ac)
{
	int status, shift;
	struct buffer_head *last_eb_bh = NULL;
	struct buffer_head *bh = NULL;
	struct ocfs2_insert_type insert = {0, };
	struct ocfs2_extent_rec rec;

	mlog(0, "add %u clusters at position %u to inode %llu\n",
	     new_clusters, cpos, (unsigned long long)OCFS2_I(inode)->ip_blkno);

	mlog_bug_on_msg(!ocfs2_sparse_alloc(osb) &&
			(OCFS2_I(inode)->ip_clusters != cpos),
			"Device %s, asking for sparse allocation: inode %llu, "
			"cpos %u, clusters %u\n",
			osb->dev_str,
			(unsigned long long)OCFS2_I(inode)->ip_blkno, cpos,
			OCFS2_I(inode)->ip_clusters);

	rec.e_cpos = cpu_to_le32(cpos);
	rec.e_blkno = cpu_to_le64(start_blk);
	rec.e_clusters = cpu_to_le32(new_clusters);

	status = ocfs2_figure_insert_type(inode, fe_bh, &last_eb_bh, &rec,
					  &insert);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	mlog(0, "Insert.appending: %u, Insert.Contig: %u, "
	     "Insert.contig_index: %d, Insert.free_records: %d, "
	     "Insert.tree_depth: %d\n",
	     insert.ins_appending, insert.ins_contig, insert.ins_contig_index,
	     insert.ins_free_records, insert.ins_tree_depth);

	/*
	 * Avoid growing the tree unless we're out of records and the
	 * insert type requres one.
	 */
	if (insert.ins_contig != CONTIG_NONE || insert.ins_free_records)
		goto out_add;

	shift = ocfs2_find_branch_target(osb, inode, fe_bh, &bh);
	if (shift < 0) {
		status = shift;
		mlog_errno(status);
		goto bail;
	}

	/* We traveled all the way to the bottom of the allocation tree
	 * and didn't find room for any more extents - we need to add
	 * another tree level */
	if (shift) {
		BUG_ON(bh);
		mlog(0, "need to shift tree depth "
		     "(current = %d)\n", insert.ins_tree_depth);

		/* ocfs2_shift_tree_depth will return us a buffer with
		 * the new extent block (so we can pass that to
		 * ocfs2_add_branch). */
		status = ocfs2_shift_tree_depth(osb, handle, inode, fe_bh,
						meta_ac, &bh);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}
		insert.ins_tree_depth++;
		/* Special case: we have room now if we shifted from
		 * tree_depth 0 */
		if (insert.ins_tree_depth == 1)
			goto out_add;
	}

	/* call ocfs2_add_branch to add the final part of the tree with
	 * the new data. */
	mlog(0, "add branch. bh = %p\n", bh);
	status = ocfs2_add_branch(osb, handle, inode, fe_bh, bh, last_eb_bh,
				  meta_ac);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

out_add:
	/* Finally, we can add clusters. This might rotate the tree for us. */
	status = ocfs2_do_insert_extent(inode, handle, fe_bh, &rec, &insert);
	if (status < 0)
		mlog_errno(status);

bail:
	if (bh)
		brelse(bh);

	if (last_eb_bh)
		brelse(last_eb_bh);

	mlog_exit(status);
	return status;
}

static inline int ocfs2_truncate_log_needs_flush(struct ocfs2_super *osb)
{
	struct buffer_head *tl_bh = osb->osb_tl_bh;
	struct ocfs2_dinode *di;
	struct ocfs2_truncate_log *tl;

	di = (struct ocfs2_dinode *) tl_bh->b_data;
	tl = &di->id2.i_dealloc;

	mlog_bug_on_msg(le16_to_cpu(tl->tl_used) > le16_to_cpu(tl->tl_count),
			"slot %d, invalid truncate log parameters: used = "
			"%u, count = %u\n", osb->slot_num,
			le16_to_cpu(tl->tl_used), le16_to_cpu(tl->tl_count));
	return le16_to_cpu(tl->tl_used) == le16_to_cpu(tl->tl_count);
}

static int ocfs2_truncate_log_can_coalesce(struct ocfs2_truncate_log *tl,
					   unsigned int new_start)
{
	unsigned int tail_index;
	unsigned int current_tail;

	/* No records, nothing to coalesce */
	if (!le16_to_cpu(tl->tl_used))
		return 0;

	tail_index = le16_to_cpu(tl->tl_used) - 1;
	current_tail = le32_to_cpu(tl->tl_recs[tail_index].t_start);
	current_tail += le32_to_cpu(tl->tl_recs[tail_index].t_clusters);

	return current_tail == new_start;
}

static int ocfs2_truncate_log_append(struct ocfs2_super *osb,
				     handle_t *handle,
				     u64 start_blk,
				     unsigned int num_clusters)
{
	int status, index;
	unsigned int start_cluster, tl_count;
	struct inode *tl_inode = osb->osb_tl_inode;
	struct buffer_head *tl_bh = osb->osb_tl_bh;
	struct ocfs2_dinode *di;
	struct ocfs2_truncate_log *tl;

	mlog_entry("start_blk = %llu, num_clusters = %u\n",
		   (unsigned long long)start_blk, num_clusters);

	BUG_ON(mutex_trylock(&tl_inode->i_mutex));

	start_cluster = ocfs2_blocks_to_clusters(osb->sb, start_blk);

	di = (struct ocfs2_dinode *) tl_bh->b_data;
	tl = &di->id2.i_dealloc;
	if (!OCFS2_IS_VALID_DINODE(di)) {
		OCFS2_RO_ON_INVALID_DINODE(osb->sb, di);
		status = -EIO;
		goto bail;
	}

	tl_count = le16_to_cpu(tl->tl_count);
	mlog_bug_on_msg(tl_count > ocfs2_truncate_recs_per_inode(osb->sb) ||
			tl_count == 0,
			"Truncate record count on #%llu invalid "
			"wanted %u, actual %u\n",
			(unsigned long long)OCFS2_I(tl_inode)->ip_blkno,
			ocfs2_truncate_recs_per_inode(osb->sb),
			le16_to_cpu(tl->tl_count));

	/* Caller should have known to flush before calling us. */
	index = le16_to_cpu(tl->tl_used);
	if (index >= tl_count) {
		status = -ENOSPC;
		mlog_errno(status);
		goto bail;
	}

	status = ocfs2_journal_access(handle, tl_inode, tl_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	mlog(0, "Log truncate of %u clusters starting at cluster %u to "
	     "%llu (index = %d)\n", num_clusters, start_cluster,
	     (unsigned long long)OCFS2_I(tl_inode)->ip_blkno, index);

	if (ocfs2_truncate_log_can_coalesce(tl, start_cluster)) {
		/*
		 * Move index back to the record we are coalescing with.
		 * ocfs2_truncate_log_can_coalesce() guarantees nonzero
		 */
		index--;

		num_clusters += le32_to_cpu(tl->tl_recs[index].t_clusters);
		mlog(0, "Coalesce with index %u (start = %u, clusters = %u)\n",
		     index, le32_to_cpu(tl->tl_recs[index].t_start),
		     num_clusters);
	} else {
		tl->tl_recs[index].t_start = cpu_to_le32(start_cluster);
		tl->tl_used = cpu_to_le16(index + 1);
	}
	tl->tl_recs[index].t_clusters = cpu_to_le32(num_clusters);

	status = ocfs2_journal_dirty(handle, tl_bh);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

bail:
	mlog_exit(status);
	return status;
}

static int ocfs2_replay_truncate_records(struct ocfs2_super *osb,
					 handle_t *handle,
					 struct inode *data_alloc_inode,
					 struct buffer_head *data_alloc_bh)
{
	int status = 0;
	int i;
	unsigned int num_clusters;
	u64 start_blk;
	struct ocfs2_truncate_rec rec;
	struct ocfs2_dinode *di;
	struct ocfs2_truncate_log *tl;
	struct inode *tl_inode = osb->osb_tl_inode;
	struct buffer_head *tl_bh = osb->osb_tl_bh;

	mlog_entry_void();

	di = (struct ocfs2_dinode *) tl_bh->b_data;
	tl = &di->id2.i_dealloc;
	i = le16_to_cpu(tl->tl_used) - 1;
	while (i >= 0) {
		/* Caller has given us at least enough credits to
		 * update the truncate log dinode */
		status = ocfs2_journal_access(handle, tl_inode, tl_bh,
					      OCFS2_JOURNAL_ACCESS_WRITE);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}

		tl->tl_used = cpu_to_le16(i);

		status = ocfs2_journal_dirty(handle, tl_bh);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}

		/* TODO: Perhaps we can calculate the bulk of the
		 * credits up front rather than extending like
		 * this. */
		status = ocfs2_extend_trans(handle,
					    OCFS2_TRUNCATE_LOG_FLUSH_ONE_REC);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}

		rec = tl->tl_recs[i];
		start_blk = ocfs2_clusters_to_blocks(data_alloc_inode->i_sb,
						    le32_to_cpu(rec.t_start));
		num_clusters = le32_to_cpu(rec.t_clusters);

		/* if start_blk is not set, we ignore the record as
		 * invalid. */
		if (start_blk) {
			mlog(0, "free record %d, start = %u, clusters = %u\n",
			     i, le32_to_cpu(rec.t_start), num_clusters);

			status = ocfs2_free_clusters(handle, data_alloc_inode,
						     data_alloc_bh, start_blk,
						     num_clusters);
			if (status < 0) {
				mlog_errno(status);
				goto bail;
			}
		}
		i--;
	}

bail:
	mlog_exit(status);
	return status;
}

/* Expects you to already be holding tl_inode->i_mutex */
static int __ocfs2_flush_truncate_log(struct ocfs2_super *osb)
{
	int status;
	unsigned int num_to_flush;
	handle_t *handle;
	struct inode *tl_inode = osb->osb_tl_inode;
	struct inode *data_alloc_inode = NULL;
	struct buffer_head *tl_bh = osb->osb_tl_bh;
	struct buffer_head *data_alloc_bh = NULL;
	struct ocfs2_dinode *di;
	struct ocfs2_truncate_log *tl;

	mlog_entry_void();

	BUG_ON(mutex_trylock(&tl_inode->i_mutex));

	di = (struct ocfs2_dinode *) tl_bh->b_data;
	tl = &di->id2.i_dealloc;
	if (!OCFS2_IS_VALID_DINODE(di)) {
		OCFS2_RO_ON_INVALID_DINODE(osb->sb, di);
		status = -EIO;
		goto out;
	}

	num_to_flush = le16_to_cpu(tl->tl_used);
	mlog(0, "Flush %u records from truncate log #%llu\n",
	     num_to_flush, (unsigned long long)OCFS2_I(tl_inode)->ip_blkno);
	if (!num_to_flush) {
		status = 0;
		goto out;
	}

	data_alloc_inode = ocfs2_get_system_file_inode(osb,
						       GLOBAL_BITMAP_SYSTEM_INODE,
						       OCFS2_INVALID_SLOT);
	if (!data_alloc_inode) {
		status = -EINVAL;
		mlog(ML_ERROR, "Could not get bitmap inode!\n");
		goto out;
	}

	mutex_lock(&data_alloc_inode->i_mutex);

	status = ocfs2_meta_lock(data_alloc_inode, &data_alloc_bh, 1);
	if (status < 0) {
		mlog_errno(status);
		goto out_mutex;
	}

	handle = ocfs2_start_trans(osb, OCFS2_TRUNCATE_LOG_UPDATE);
	if (IS_ERR(handle)) {
		status = PTR_ERR(handle);
		mlog_errno(status);
		goto out_unlock;
	}

	status = ocfs2_replay_truncate_records(osb, handle, data_alloc_inode,
					       data_alloc_bh);
	if (status < 0)
		mlog_errno(status);

	ocfs2_commit_trans(osb, handle);

out_unlock:
	brelse(data_alloc_bh);
	ocfs2_meta_unlock(data_alloc_inode, 1);

out_mutex:
	mutex_unlock(&data_alloc_inode->i_mutex);
	iput(data_alloc_inode);

out:
	mlog_exit(status);
	return status;
}

int ocfs2_flush_truncate_log(struct ocfs2_super *osb)
{
	int status;
	struct inode *tl_inode = osb->osb_tl_inode;

	mutex_lock(&tl_inode->i_mutex);
	status = __ocfs2_flush_truncate_log(osb);
	mutex_unlock(&tl_inode->i_mutex);

	return status;
}

static void ocfs2_truncate_log_worker(struct work_struct *work)
{
	int status;
	struct ocfs2_super *osb =
		container_of(work, struct ocfs2_super,
			     osb_truncate_log_wq.work);

	mlog_entry_void();

	status = ocfs2_flush_truncate_log(osb);
	if (status < 0)
		mlog_errno(status);

	mlog_exit(status);
}

#define OCFS2_TRUNCATE_LOG_FLUSH_INTERVAL (2 * HZ)
void ocfs2_schedule_truncate_log_flush(struct ocfs2_super *osb,
				       int cancel)
{
	if (osb->osb_tl_inode) {
		/* We want to push off log flushes while truncates are
		 * still running. */
		if (cancel)
			cancel_delayed_work(&osb->osb_truncate_log_wq);

		queue_delayed_work(ocfs2_wq, &osb->osb_truncate_log_wq,
				   OCFS2_TRUNCATE_LOG_FLUSH_INTERVAL);
	}
}

static int ocfs2_get_truncate_log_info(struct ocfs2_super *osb,
				       int slot_num,
				       struct inode **tl_inode,
				       struct buffer_head **tl_bh)
{
	int status;
	struct inode *inode = NULL;
	struct buffer_head *bh = NULL;

	inode = ocfs2_get_system_file_inode(osb,
					   TRUNCATE_LOG_SYSTEM_INODE,
					   slot_num);
	if (!inode) {
		status = -EINVAL;
		mlog(ML_ERROR, "Could not get load truncate log inode!\n");
		goto bail;
	}

	status = ocfs2_read_block(osb, OCFS2_I(inode)->ip_blkno, &bh,
				  OCFS2_BH_CACHED, inode);
	if (status < 0) {
		iput(inode);
		mlog_errno(status);
		goto bail;
	}

	*tl_inode = inode;
	*tl_bh    = bh;
bail:
	mlog_exit(status);
	return status;
}

/* called during the 1st stage of node recovery. we stamp a clean
 * truncate log and pass back a copy for processing later. if the
 * truncate log does not require processing, a *tl_copy is set to
 * NULL. */
int ocfs2_begin_truncate_log_recovery(struct ocfs2_super *osb,
				      int slot_num,
				      struct ocfs2_dinode **tl_copy)
{
	int status;
	struct inode *tl_inode = NULL;
	struct buffer_head *tl_bh = NULL;
	struct ocfs2_dinode *di;
	struct ocfs2_truncate_log *tl;

	*tl_copy = NULL;

	mlog(0, "recover truncate log from slot %d\n", slot_num);

	status = ocfs2_get_truncate_log_info(osb, slot_num, &tl_inode, &tl_bh);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	di = (struct ocfs2_dinode *) tl_bh->b_data;
	tl = &di->id2.i_dealloc;
	if (!OCFS2_IS_VALID_DINODE(di)) {
		OCFS2_RO_ON_INVALID_DINODE(tl_inode->i_sb, di);
		status = -EIO;
		goto bail;
	}

	if (le16_to_cpu(tl->tl_used)) {
		mlog(0, "We'll have %u logs to recover\n",
		     le16_to_cpu(tl->tl_used));

		*tl_copy = kmalloc(tl_bh->b_size, GFP_KERNEL);
		if (!(*tl_copy)) {
			status = -ENOMEM;
			mlog_errno(status);
			goto bail;
		}

		/* Assuming the write-out below goes well, this copy
		 * will be passed back to recovery for processing. */
		memcpy(*tl_copy, tl_bh->b_data, tl_bh->b_size);

		/* All we need to do to clear the truncate log is set
		 * tl_used. */
		tl->tl_used = 0;

		status = ocfs2_write_block(osb, tl_bh, tl_inode);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}
	}

bail:
	if (tl_inode)
		iput(tl_inode);
	if (tl_bh)
		brelse(tl_bh);

	if (status < 0 && (*tl_copy)) {
		kfree(*tl_copy);
		*tl_copy = NULL;
	}

	mlog_exit(status);
	return status;
}

int ocfs2_complete_truncate_log_recovery(struct ocfs2_super *osb,
					 struct ocfs2_dinode *tl_copy)
{
	int status = 0;
	int i;
	unsigned int clusters, num_recs, start_cluster;
	u64 start_blk;
	handle_t *handle;
	struct inode *tl_inode = osb->osb_tl_inode;
	struct ocfs2_truncate_log *tl;

	mlog_entry_void();

	if (OCFS2_I(tl_inode)->ip_blkno == le64_to_cpu(tl_copy->i_blkno)) {
		mlog(ML_ERROR, "Asked to recover my own truncate log!\n");
		return -EINVAL;
	}

	tl = &tl_copy->id2.i_dealloc;
	num_recs = le16_to_cpu(tl->tl_used);
	mlog(0, "cleanup %u records from %llu\n", num_recs,
	     (unsigned long long)tl_copy->i_blkno);

	mutex_lock(&tl_inode->i_mutex);
	for(i = 0; i < num_recs; i++) {
		if (ocfs2_truncate_log_needs_flush(osb)) {
			status = __ocfs2_flush_truncate_log(osb);
			if (status < 0) {
				mlog_errno(status);
				goto bail_up;
			}
		}

		handle = ocfs2_start_trans(osb, OCFS2_TRUNCATE_LOG_UPDATE);
		if (IS_ERR(handle)) {
			status = PTR_ERR(handle);
			mlog_errno(status);
			goto bail_up;
		}

		clusters = le32_to_cpu(tl->tl_recs[i].t_clusters);
		start_cluster = le32_to_cpu(tl->tl_recs[i].t_start);
		start_blk = ocfs2_clusters_to_blocks(osb->sb, start_cluster);

		status = ocfs2_truncate_log_append(osb, handle,
						   start_blk, clusters);
		ocfs2_commit_trans(osb, handle);
		if (status < 0) {
			mlog_errno(status);
			goto bail_up;
		}
	}

bail_up:
	mutex_unlock(&tl_inode->i_mutex);

	mlog_exit(status);
	return status;
}

void ocfs2_truncate_log_shutdown(struct ocfs2_super *osb)
{
	int status;
	struct inode *tl_inode = osb->osb_tl_inode;

	mlog_entry_void();

	if (tl_inode) {
		cancel_delayed_work(&osb->osb_truncate_log_wq);
		flush_workqueue(ocfs2_wq);

		status = ocfs2_flush_truncate_log(osb);
		if (status < 0)
			mlog_errno(status);

		brelse(osb->osb_tl_bh);
		iput(osb->osb_tl_inode);
	}

	mlog_exit_void();
}

int ocfs2_truncate_log_init(struct ocfs2_super *osb)
{
	int status;
	struct inode *tl_inode = NULL;
	struct buffer_head *tl_bh = NULL;

	mlog_entry_void();

	status = ocfs2_get_truncate_log_info(osb,
					     osb->slot_num,
					     &tl_inode,
					     &tl_bh);
	if (status < 0)
		mlog_errno(status);

	/* ocfs2_truncate_log_shutdown keys on the existence of
	 * osb->osb_tl_inode so we don't set any of the osb variables
	 * until we're sure all is well. */
	INIT_DELAYED_WORK(&osb->osb_truncate_log_wq,
			  ocfs2_truncate_log_worker);
	osb->osb_tl_bh    = tl_bh;
	osb->osb_tl_inode = tl_inode;

	mlog_exit(status);
	return status;
}

/* This function will figure out whether the currently last extent
 * block will be deleted, and if it will, what the new last extent
 * block will be so we can update his h_next_leaf_blk field, as well
 * as the dinodes i_last_eb_blk */
static int ocfs2_find_new_last_ext_blk(struct inode *inode,
				       u32 new_i_clusters,
				       struct ocfs2_path *path,
				       struct buffer_head **new_last_eb)
{
	int ret = 0;
	u32 cpos;
	struct ocfs2_extent_block *eb;
	struct ocfs2_extent_list *el;
	struct buffer_head *bh = NULL;

	*new_last_eb = NULL;

	/* we have no tree, so of course, no last_eb. */
	if (!path->p_tree_depth)
		goto out;

	/* trunc to zero special case - this makes tree_depth = 0
	 * regardless of what it is.  */
	if (!new_i_clusters)
		goto out;

	el = path_leaf_el(path);
	BUG_ON(!el->l_next_free_rec);

	/* Make sure that this guy will actually be empty after we
	 * clear away the data. */
	if (ocfs2_is_empty_extent(&el->l_recs[0])) {
		if (le16_to_cpu(el->l_next_free_rec) > 1 &&
		    le32_to_cpu(el->l_recs[1].e_cpos) < new_i_clusters)
			goto out;
	} else if (le32_to_cpu(el->l_recs[0].e_cpos) < new_i_clusters)
		goto out;

	ret = ocfs2_find_cpos_for_left_leaf(inode->i_sb, path, &cpos);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_find_leaf(inode, path_root_el(path), cpos, &bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	eb = (struct ocfs2_extent_block *) bh->b_data;
	el = &eb->h_list;
	if (!OCFS2_IS_VALID_EXTENT_BLOCK(eb)) {
		OCFS2_RO_ON_INVALID_EXTENT_BLOCK(inode->i_sb, eb);
		ret = -EROFS;
		goto out;
	}

	*new_last_eb = bh;
	get_bh(*new_last_eb);
	mlog(0, "returning block %llu, (cpos: %u)\n",
	     (unsigned long long)le64_to_cpu(eb->h_blkno), cpos);
out:
	brelse(bh);

	return ret;
}

static int ocfs2_do_truncate(struct ocfs2_super *osb,
			     unsigned int clusters_to_del,
			     struct inode *inode,
			     struct buffer_head *fe_bh,
			     handle_t *handle,
			     struct ocfs2_truncate_context *tc,
			     struct ocfs2_path *path)
{
	int status, i, index;
	struct ocfs2_dinode *fe;
	struct ocfs2_extent_block *eb;
	struct ocfs2_extent_block *last_eb = NULL;
	struct ocfs2_extent_list *el;
	struct buffer_head *eb_bh = NULL;
	struct buffer_head *last_eb_bh = NULL;
	u64 delete_blk = 0;

	fe = (struct ocfs2_dinode *) fe_bh->b_data;

	status = ocfs2_find_new_last_ext_blk(inode,
					     le32_to_cpu(fe->i_clusters) -
					     clusters_to_del,
					     path, &last_eb_bh);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	/*
	 * Each component will be touched, so we might as well journal
	 * here to avoid having to handle errors later.
	 */
	for (i = 0; i < path_num_items(path); i++) {
		status = ocfs2_journal_access(handle, inode,
					      path->p_node[i].bh,
					      OCFS2_JOURNAL_ACCESS_WRITE);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}
	}

	if (last_eb_bh) {
		status = ocfs2_journal_access(handle, inode, last_eb_bh,
					      OCFS2_JOURNAL_ACCESS_WRITE);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}

		last_eb = (struct ocfs2_extent_block *) last_eb_bh->b_data;
	}

	el = &(fe->id2.i_list);

	/*
	 * Lower levels depend on this never happening, but it's best
	 * to check it up here before changing the tree.
	 */
	if (el->l_tree_depth && ocfs2_is_empty_extent(&el->l_recs[0])) {
		ocfs2_error(inode->i_sb,
			    "Inode %lu has an empty extent record, depth %u\n",
			    inode->i_ino, le16_to_cpu(el->l_tree_depth));
		goto bail;
	}

	spin_lock(&OCFS2_I(inode)->ip_lock);
	OCFS2_I(inode)->ip_clusters = le32_to_cpu(fe->i_clusters) -
				      clusters_to_del;
	spin_unlock(&OCFS2_I(inode)->ip_lock);
	le32_add_cpu(&fe->i_clusters, -clusters_to_del);

	i = le16_to_cpu(el->l_next_free_rec) - 1;

	BUG_ON(le32_to_cpu(el->l_recs[i].e_clusters) < clusters_to_del);
	le32_add_cpu(&el->l_recs[i].e_clusters, -clusters_to_del);
	/* tree depth zero, we can just delete the clusters, otherwise
	 * we need to record the offset of the next level extent block
	 * as we may overwrite it. */
	if (!el->l_tree_depth) {
		delete_blk = le64_to_cpu(el->l_recs[i].e_blkno)
			+ ocfs2_clusters_to_blocks(osb->sb,
					le32_to_cpu(el->l_recs[i].e_clusters));

		if (!el->l_recs[i].e_clusters) {
			/* if we deleted the whole extent record, then clear
			 * out the other fields and update the extent
			 * list.
			 */
			el->l_recs[i].e_cpos = 0;
			el->l_recs[i].e_blkno = 0;
			BUG_ON(!el->l_next_free_rec);
			le16_add_cpu(&el->l_next_free_rec, -1);

			/*
			 * The leftmost record might be an empty extent -
			 * delete it here too.
			 */
			if (i == 1 && ocfs2_is_empty_extent(&el->l_recs[0])) {
				el->l_recs[0].e_cpos = 0;
				el->l_recs[0].e_blkno = 0;
				el->l_next_free_rec = 0;
			}
		}
	}

	if (le32_to_cpu(fe->i_clusters) == 0) {
		/* trunc to zero is a special case. */
		el->l_tree_depth = 0;
		fe->i_last_eb_blk = 0;
	} else if (last_eb)
		fe->i_last_eb_blk = last_eb->h_blkno;

	status = ocfs2_journal_dirty(handle, fe_bh);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	if (last_eb) {
		/* If there will be a new last extent block, then by
		 * definition, there cannot be any leaves to the right of
		 * him. */
		last_eb->h_next_leaf_blk = 0;
		status = ocfs2_journal_dirty(handle, last_eb_bh);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}
	}

	index = 1;
	/* if our tree depth > 0, update all the tree blocks below us. */
	while (index <= path->p_tree_depth) {
		eb_bh = path->p_node[index].bh;
		eb = (struct ocfs2_extent_block *)eb_bh->b_data;
		el = path->p_node[index].el;

		mlog(0, "traveling tree (index = %d, extent block: %llu)\n",
		     index,  (unsigned long long)eb_bh->b_blocknr);

		BUG_ON(le16_to_cpu(el->l_next_free_rec) == 0);
		if (index !=
		    (path->p_tree_depth - le16_to_cpu(el->l_tree_depth))) {
			ocfs2_error(inode->i_sb,
				    "Inode %lu has invalid ext. block %llu\n",
				    inode->i_ino,
				    (unsigned long long)eb_bh->b_blocknr);
			goto bail;
		}

		i = le16_to_cpu(el->l_next_free_rec) - 1;

		mlog(0, "extent block %llu, before: record %d: "
		     "(%u, %u, %llu), next = %u\n",
		     (unsigned long long)le64_to_cpu(eb->h_blkno), i,
		     le32_to_cpu(el->l_recs[i].e_cpos),
		     le32_to_cpu(el->l_recs[i].e_clusters),
		     (unsigned long long)le64_to_cpu(el->l_recs[i].e_blkno),
		     le16_to_cpu(el->l_next_free_rec));

		BUG_ON(le32_to_cpu(el->l_recs[i].e_clusters) < clusters_to_del);
		le32_add_cpu(&el->l_recs[i].e_clusters, -clusters_to_del);

		/* bottom-most block requires us to delete data.*/
		if (!el->l_tree_depth)
			delete_blk = le64_to_cpu(el->l_recs[i].e_blkno)
				+ ocfs2_clusters_to_blocks(osb->sb,
					le32_to_cpu(el->l_recs[i].e_clusters));
		if (!el->l_recs[i].e_clusters) {
			el->l_recs[i].e_cpos = 0;
			el->l_recs[i].e_blkno = 0;
			BUG_ON(!el->l_next_free_rec);
			le16_add_cpu(&el->l_next_free_rec, -1);
		}
		if (i == 1 && ocfs2_is_empty_extent(&el->l_recs[0])) {
			el->l_recs[0].e_cpos = 0;
			el->l_recs[0].e_blkno = 0;
			el->l_next_free_rec = 0;
		}

		mlog(0, "extent block %llu, after: record %d: "
		     "(%u, %u, %llu), next = %u\n",
		     (unsigned long long)le64_to_cpu(eb->h_blkno), i,
		     le32_to_cpu(el->l_recs[i].e_cpos),
		     le32_to_cpu(el->l_recs[i].e_clusters),
		     (unsigned long long)le64_to_cpu(el->l_recs[i].e_blkno),
		     le16_to_cpu(el->l_next_free_rec));

		status = ocfs2_journal_dirty(handle, eb_bh);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}

		if (!el->l_next_free_rec) {
			mlog(0, "deleting this extent block.\n");

			ocfs2_remove_from_cache(inode, eb_bh);

			BUG_ON(el->l_recs[0].e_clusters);
			BUG_ON(el->l_recs[0].e_cpos);
			BUG_ON(el->l_recs[0].e_blkno);

			/*
			 * We need to remove this extent block from
			 * the list above it.
			 *
			 * Since we've passed it already in this loop,
			 * no need to worry about journaling.
			 */
			el = path->p_node[index - 1].el;
			i = le16_to_cpu(el->l_next_free_rec) - 1;
			BUG_ON(i < 0);
			el->l_recs[i].e_cpos = 0;
			el->l_recs[i].e_clusters = 0;
			el->l_recs[i].e_blkno = 0;
			le16_add_cpu(&el->l_next_free_rec, -1);

			if (eb->h_suballoc_slot == 0) {
				/*
				 * This code only understands how to
				 * lock the suballocator in slot 0,
				 * which is fine because allocation is
				 * only ever done out of that
				 * suballocator too. A future version
				 * might change that however, so avoid
				 * a free if we don't know how to
				 * handle it. This way an fs incompat
				 * bit will not be necessary.
				 */
				status = ocfs2_free_extent_block(handle,
								 tc->tc_ext_alloc_inode,
								 tc->tc_ext_alloc_bh,
								 eb);
				if (status < 0) {
					mlog_errno(status);
					goto bail;
				}
			}
		}
		index++;
	}

	BUG_ON(!delete_blk);
	status = ocfs2_truncate_log_append(osb, handle, delete_blk,
					   clusters_to_del);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}
	status = 0;
bail:

	mlog_exit(status);
	return status;
}

/*
 * It is expected, that by the time you call this function,
 * inode->i_size and fe->i_size have been adjusted.
 *
 * WARNING: This will kfree the truncate context
 */
int ocfs2_commit_truncate(struct ocfs2_super *osb,
			  struct inode *inode,
			  struct buffer_head *fe_bh,
			  struct ocfs2_truncate_context *tc)
{
	int status, i, credits, tl_sem = 0;
	u32 clusters_to_del, new_highest_cpos, range;
	struct ocfs2_extent_list *el;
	handle_t *handle = NULL;
	struct inode *tl_inode = osb->osb_tl_inode;
	struct ocfs2_path *path = NULL;

	mlog_entry_void();

	down_write(&OCFS2_I(inode)->ip_alloc_sem);

	new_highest_cpos = ocfs2_clusters_for_bytes(osb->sb,
						     i_size_read(inode));

	path = ocfs2_new_inode_path(fe_bh);
	if (!path) {
		status = -ENOMEM;
		mlog_errno(status);
		goto bail;
	}
start:
	/*
	 * Truncate always works against the rightmost tree branch.
	 */
	status = ocfs2_find_path(inode, path, UINT_MAX);
	if (status) {
		mlog_errno(status);
		goto bail;
	}

	mlog(0, "inode->ip_clusters = %u, tree_depth = %u\n",
	     OCFS2_I(inode)->ip_clusters, path->p_tree_depth);

	/*
	 * By now, el will point to the extent list on the bottom most
	 * portion of this tree. Only the tail record is considered in
	 * each pass.
	 *
	 * We handle the following cases, in order:
	 * - empty extent: delete the remaining branch
	 * - remove the entire record
	 * - remove a partial record
	 * - no record needs to be removed (truncate has completed)
	 */
	el = path_leaf_el(path);
	i = le16_to_cpu(el->l_next_free_rec) - 1;
	range = le32_to_cpu(el->l_recs[i].e_cpos) +
		le32_to_cpu(el->l_recs[i].e_clusters);
	if (i == 0 && ocfs2_is_empty_extent(&el->l_recs[i])) {
		clusters_to_del = 0;
	} else if (le32_to_cpu(el->l_recs[i].e_cpos) >= new_highest_cpos) {
		clusters_to_del = le32_to_cpu(el->l_recs[i].e_clusters);
	} else if (range > new_highest_cpos) {
		clusters_to_del = (le32_to_cpu(el->l_recs[i].e_clusters) +
				   le32_to_cpu(el->l_recs[i].e_cpos)) -
				  new_highest_cpos;
	} else {
		status = 0;
		goto bail;
	}

	mlog(0, "clusters_to_del = %u in this pass, tail blk=%llu\n",
	     clusters_to_del, (unsigned long long)path_leaf_bh(path)->b_blocknr);

	BUG_ON(clusters_to_del == 0);

	mutex_lock(&tl_inode->i_mutex);
	tl_sem = 1;
	/* ocfs2_truncate_log_needs_flush guarantees us at least one
	 * record is free for use. If there isn't any, we flush to get
	 * an empty truncate log.  */
	if (ocfs2_truncate_log_needs_flush(osb)) {
		status = __ocfs2_flush_truncate_log(osb);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}
	}

	credits = ocfs2_calc_tree_trunc_credits(osb->sb, clusters_to_del,
						(struct ocfs2_dinode *)fe_bh->b_data,
						el);
	handle = ocfs2_start_trans(osb, credits);
	if (IS_ERR(handle)) {
		status = PTR_ERR(handle);
		handle = NULL;
		mlog_errno(status);
		goto bail;
	}

	status = ocfs2_do_truncate(osb, clusters_to_del, inode, fe_bh, handle,
				   tc, path);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	mutex_unlock(&tl_inode->i_mutex);
	tl_sem = 0;

	ocfs2_commit_trans(osb, handle);
	handle = NULL;

	ocfs2_reinit_path(path, 1);

	/*
	 * Only loop if we still have allocation.
	 */
	if (OCFS2_I(inode)->ip_clusters)
		goto start;
bail:
	up_write(&OCFS2_I(inode)->ip_alloc_sem);

	ocfs2_schedule_truncate_log_flush(osb, 1);

	if (tl_sem)
		mutex_unlock(&tl_inode->i_mutex);

	if (handle)
		ocfs2_commit_trans(osb, handle);

	ocfs2_free_path(path);

	/* This will drop the ext_alloc cluster lock for us */
	ocfs2_free_truncate_context(tc);

	mlog_exit(status);
	return status;
}

/*
 * Expects the inode to already be locked. This will figure out which
 * inodes need to be locked and will put them on the returned truncate
 * context.
 */
int ocfs2_prepare_truncate(struct ocfs2_super *osb,
			   struct inode *inode,
			   struct buffer_head *fe_bh,
			   struct ocfs2_truncate_context **tc)
{
	int status, metadata_delete, i;
	unsigned int new_i_clusters;
	struct ocfs2_dinode *fe;
	struct ocfs2_extent_block *eb;
	struct ocfs2_extent_list *el;
	struct buffer_head *last_eb_bh = NULL;
	struct inode *ext_alloc_inode = NULL;
	struct buffer_head *ext_alloc_bh = NULL;

	mlog_entry_void();

	*tc = NULL;

	new_i_clusters = ocfs2_clusters_for_bytes(osb->sb,
						  i_size_read(inode));
	fe = (struct ocfs2_dinode *) fe_bh->b_data;

	mlog(0, "fe->i_clusters = %u, new_i_clusters = %u, fe->i_size ="
	     "%llu\n", fe->i_clusters, new_i_clusters,
	     (unsigned long long)fe->i_size);

	if (!ocfs2_sparse_alloc(osb) &&
	    le32_to_cpu(fe->i_clusters) <= new_i_clusters) {
		ocfs2_error(inode->i_sb, "Dinode %llu has cluster count "
			    "%u and size %llu whereas struct inode has "
			    "cluster count %u and size %llu which caused an "
			    "invalid truncate to %u clusters.",
			    (unsigned long long)le64_to_cpu(fe->i_blkno),
			    le32_to_cpu(fe->i_clusters),
			    (unsigned long long)le64_to_cpu(fe->i_size),
			    OCFS2_I(inode)->ip_clusters, i_size_read(inode),
			    new_i_clusters);
		mlog_meta_lvb(ML_ERROR, &OCFS2_I(inode)->ip_meta_lockres);
		status = -EIO;
		goto bail;
	}

	*tc = kzalloc(sizeof(struct ocfs2_truncate_context), GFP_KERNEL);
	if (!(*tc)) {
		status = -ENOMEM;
		mlog_errno(status);
		goto bail;
	}

	metadata_delete = 0;
	if (fe->id2.i_list.l_tree_depth) {
		/* If we have a tree, then the truncate may result in
		 * metadata deletes. Figure this out from the
		 * rightmost leaf block.*/
		status = ocfs2_read_block(osb, le64_to_cpu(fe->i_last_eb_blk),
					  &last_eb_bh, OCFS2_BH_CACHED, inode);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}
		eb = (struct ocfs2_extent_block *) last_eb_bh->b_data;
		if (!OCFS2_IS_VALID_EXTENT_BLOCK(eb)) {
			OCFS2_RO_ON_INVALID_EXTENT_BLOCK(inode->i_sb, eb);

			brelse(last_eb_bh);
			status = -EIO;
			goto bail;
		}
		el = &(eb->h_list);

		i = 0;
		if (ocfs2_is_empty_extent(&el->l_recs[0]))
			i = 1;
		/*
		 * XXX: Should we check that next_free_rec contains
		 * the extent?
		 */
		if (le32_to_cpu(el->l_recs[i].e_cpos) >= new_i_clusters)
			metadata_delete = 1;
	}

	(*tc)->tc_last_eb_bh = last_eb_bh;

	if (metadata_delete) {
		mlog(0, "Will have to delete metadata for this trunc. "
		     "locking allocator.\n");
		ext_alloc_inode = ocfs2_get_system_file_inode(osb, EXTENT_ALLOC_SYSTEM_INODE, 0);
		if (!ext_alloc_inode) {
			status = -ENOMEM;
			mlog_errno(status);
			goto bail;
		}

		mutex_lock(&ext_alloc_inode->i_mutex);
		(*tc)->tc_ext_alloc_inode = ext_alloc_inode;

		status = ocfs2_meta_lock(ext_alloc_inode, &ext_alloc_bh, 1);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}
		(*tc)->tc_ext_alloc_bh = ext_alloc_bh;
		(*tc)->tc_ext_alloc_locked = 1;
	}

	status = 0;
bail:
	if (status < 0) {
		if (*tc)
			ocfs2_free_truncate_context(*tc);
		*tc = NULL;
	}
	mlog_exit_void();
	return status;
}

static void ocfs2_free_truncate_context(struct ocfs2_truncate_context *tc)
{
	if (tc->tc_ext_alloc_inode) {
		if (tc->tc_ext_alloc_locked)
			ocfs2_meta_unlock(tc->tc_ext_alloc_inode, 1);

		mutex_unlock(&tc->tc_ext_alloc_inode->i_mutex);
		iput(tc->tc_ext_alloc_inode);
	}

	if (tc->tc_ext_alloc_bh)
		brelse(tc->tc_ext_alloc_bh);

	if (tc->tc_last_eb_bh)
		brelse(tc->tc_last_eb_bh);

	kfree(tc);
}