summaryrefslogtreecommitdiffstats
path: root/sbin/fsck_ffs/suj.c
blob: 5fca0f5fbfdfdb1916fb7da8ad28414a8681051d (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
/*-
 * Copyright 2009, 2010 Jeffrey W. Roberson <jeff@FreeBSD.org>
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */

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

#include <sys/param.h>
#include <sys/disk.h>
#include <sys/disklabel.h>
#include <sys/mount.h>
#include <sys/stat.h>

#include <ufs/ufs/ufsmount.h>
#include <ufs/ufs/dinode.h>
#include <ufs/ufs/dir.h>
#include <ufs/ffs/fs.h>

#include <assert.h>
#include <err.h>
#include <setjmp.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <libufs.h>
#include <string.h>
#include <strings.h>
#include <sysexits.h>
#include <time.h>

#include "fsck.h"

#define	DOTDOT_OFFSET	DIRECTSIZ(1)
#define	SUJ_HASHSIZE	2048
#define	SUJ_HASHMASK	(SUJ_HASHSIZE - 1)
#define	SUJ_HASH(x)	((x * 2654435761) & SUJ_HASHMASK)

struct suj_seg {
	TAILQ_ENTRY(suj_seg) ss_next;
	struct jsegrec	ss_rec;
	uint8_t		*ss_blk;
};

struct suj_rec {
	TAILQ_ENTRY(suj_rec) sr_next;
	union jrec	*sr_rec;
};
TAILQ_HEAD(srechd, suj_rec);

struct suj_ino {
	LIST_ENTRY(suj_ino)	si_next;
	struct srechd		si_recs;
	struct srechd		si_newrecs;
	struct srechd		si_movs;
	struct jtrncrec		*si_trunc;
	ino_t			si_ino;
	char			si_skipparent;
	char			si_hasrecs;
	char			si_blkadj;
	char			si_linkadj;
	int			si_mode;
	nlink_t			si_nlinkadj;
	nlink_t			si_nlink;
	nlink_t			si_dotlinks;
};
LIST_HEAD(inohd, suj_ino);

struct suj_blk {
	LIST_ENTRY(suj_blk)	sb_next;
	struct srechd		sb_recs;
	ufs2_daddr_t		sb_blk;
};
LIST_HEAD(blkhd, suj_blk);

struct data_blk {
	LIST_ENTRY(data_blk)	db_next;
	uint8_t			*db_buf;
	ufs2_daddr_t		db_blk;
	int			db_size;
	int			db_dirty;
};

struct ino_blk {
	LIST_ENTRY(ino_blk)	ib_next;
	uint8_t			*ib_buf;
	int			ib_dirty;
	ufs2_daddr_t		ib_blk;
};
LIST_HEAD(iblkhd, ino_blk);

struct suj_cg {
	LIST_ENTRY(suj_cg)	sc_next;
	struct blkhd		sc_blkhash[SUJ_HASHSIZE];
	struct inohd		sc_inohash[SUJ_HASHSIZE];
	struct iblkhd		sc_iblkhash[SUJ_HASHSIZE];
	struct ino_blk		*sc_lastiblk;
	struct suj_ino		*sc_lastino;
	struct suj_blk		*sc_lastblk;
	uint8_t			*sc_cgbuf;
	struct cg		*sc_cgp;
	int			sc_dirty;
	int			sc_cgx;
};

LIST_HEAD(cghd, suj_cg) cghash[SUJ_HASHSIZE];
LIST_HEAD(dblkhd, data_blk) dbhash[SUJ_HASHSIZE];
struct suj_cg *lastcg;
struct data_blk *lastblk;

TAILQ_HEAD(seghd, suj_seg) allsegs;
uint64_t oldseq;
static struct uufsd *disk = NULL;
static struct fs *fs = NULL;
ino_t sujino;

/*
 * Summary statistics.
 */
uint64_t freefrags;
uint64_t freeblocks;
uint64_t freeinos;
uint64_t freedir;
uint64_t jbytes;
uint64_t jrecs;

static jmp_buf	jmpbuf;

typedef void (*ino_visitor)(ino_t, ufs_lbn_t, ufs2_daddr_t, int);
static void err_suj(const char *, ...) __dead2;
static void ino_trunc(ino_t, off_t);
static void ino_decr(ino_t);
static void ino_adjust(struct suj_ino *);
static void ino_build(struct suj_ino *);
static int blk_isfree(ufs2_daddr_t);

static void *
errmalloc(size_t n)
{
	void *a;

	a = Malloc(n);
	if (a == NULL)
		err(EX_OSERR, "malloc(%zu)", n);
	return (a);
}

/*
 * When hit a fatal error in journalling check, print out
 * the error and then offer to fallback to normal fsck.
 */
static void
err_suj(const char * restrict fmt, ...)
{
	va_list ap;

	if (preen)
		(void)fprintf(stdout, "%s: ", cdevname);

	va_start(ap, fmt);
	(void)vfprintf(stdout, fmt, ap);
	va_end(ap);

	longjmp(jmpbuf, -1);
}

/*
 * Open the given provider, load superblock.
 */
static void
opendisk(const char *devnam)
{
	if (disk != NULL)
		return;
	disk = Malloc(sizeof(*disk));
	if (disk == NULL)
		err(EX_OSERR, "malloc(%zu)", sizeof(*disk));
	if (ufs_disk_fillout(disk, devnam) == -1) {
		err(EX_OSERR, "ufs_disk_fillout(%s) failed: %s", devnam,
		    disk->d_error);
	}
	fs = &disk->d_fs;
	if (real_dev_bsize == 0 && ioctl(disk->d_fd, DIOCGSECTORSIZE,
	    &real_dev_bsize) == -1)
		real_dev_bsize = secsize;
	if (debug)
		printf("dev_bsize %u\n", real_dev_bsize);
}

/*
 * Mark file system as clean, write the super-block back, close the disk.
 */
static void
closedisk(const char *devnam)
{
	struct csum *cgsum;
	int i;

	/*
	 * Recompute the fs summary info from correct cs summaries.
	 */
	bzero(&fs->fs_cstotal, sizeof(struct csum_total));
	for (i = 0; i < fs->fs_ncg; i++) {
		cgsum = &fs->fs_cs(fs, i);
		fs->fs_cstotal.cs_nffree += cgsum->cs_nffree;
		fs->fs_cstotal.cs_nbfree += cgsum->cs_nbfree;
		fs->fs_cstotal.cs_nifree += cgsum->cs_nifree;
		fs->fs_cstotal.cs_ndir += cgsum->cs_ndir;
	}
	fs->fs_pendinginodes = 0;
	fs->fs_pendingblocks = 0;
	fs->fs_clean = 1;
	fs->fs_time = time(NULL);
	fs->fs_mtime = time(NULL);
	if (sbwrite(disk, 0) == -1)
		err(EX_OSERR, "sbwrite(%s)", devnam);
	if (ufs_disk_close(disk) == -1)
		err(EX_OSERR, "ufs_disk_close(%s)", devnam);
	free(disk);
	disk = NULL;
	fs = NULL;
}

/*
 * Lookup a cg by number in the hash so we can keep track of which cgs
 * need stats rebuilt.
 */
static struct suj_cg *
cg_lookup(int cgx)
{
	struct cghd *hd;
	struct suj_cg *sc;

	if (cgx < 0 || cgx >= fs->fs_ncg)
		err_suj("Bad cg number %d\n", cgx);
	if (lastcg && lastcg->sc_cgx == cgx)
		return (lastcg);
	hd = &cghash[SUJ_HASH(cgx)];
	LIST_FOREACH(sc, hd, sc_next)
		if (sc->sc_cgx == cgx) {
			lastcg = sc;
			return (sc);
		}
	sc = errmalloc(sizeof(*sc));
	bzero(sc, sizeof(*sc));
	sc->sc_cgbuf = errmalloc(fs->fs_bsize);
	sc->sc_cgp = (struct cg *)sc->sc_cgbuf;
	sc->sc_cgx = cgx;
	LIST_INSERT_HEAD(hd, sc, sc_next);
	if (bread(disk, fsbtodb(fs, cgtod(fs, sc->sc_cgx)), sc->sc_cgbuf,
	    fs->fs_bsize) == -1)
		err_suj("Unable to read cylinder group %d\n", sc->sc_cgx);

	return (sc);
}

/*
 * Lookup an inode number in the hash and allocate a suj_ino if it does
 * not exist.
 */
static struct suj_ino *
ino_lookup(ino_t ino, int creat)
{
	struct suj_ino *sino;
	struct inohd *hd;
	struct suj_cg *sc;

	sc = cg_lookup(ino_to_cg(fs, ino));
	if (sc->sc_lastino && sc->sc_lastino->si_ino == ino)
		return (sc->sc_lastino);
	hd = &sc->sc_inohash[SUJ_HASH(ino)];
	LIST_FOREACH(sino, hd, si_next)
		if (sino->si_ino == ino)
			return (sino);
	if (creat == 0)
		return (NULL);
	sino = errmalloc(sizeof(*sino));
	bzero(sino, sizeof(*sino));
	sino->si_ino = ino;
	TAILQ_INIT(&sino->si_recs);
	TAILQ_INIT(&sino->si_newrecs);
	TAILQ_INIT(&sino->si_movs);
	LIST_INSERT_HEAD(hd, sino, si_next);

	return (sino);
}

/*
 * Lookup a block number in the hash and allocate a suj_blk if it does
 * not exist.
 */
static struct suj_blk *
blk_lookup(ufs2_daddr_t blk, int creat)
{
	struct suj_blk *sblk;
	struct suj_cg *sc;
	struct blkhd *hd;

	sc = cg_lookup(dtog(fs, blk));
	if (sc->sc_lastblk && sc->sc_lastblk->sb_blk == blk)
		return (sc->sc_lastblk);
	hd = &sc->sc_blkhash[SUJ_HASH(fragstoblks(fs, blk))];
	LIST_FOREACH(sblk, hd, sb_next)
		if (sblk->sb_blk == blk)
			return (sblk);
	if (creat == 0)
		return (NULL);
	sblk = errmalloc(sizeof(*sblk));
	bzero(sblk, sizeof(*sblk));
	sblk->sb_blk = blk;
	TAILQ_INIT(&sblk->sb_recs);
	LIST_INSERT_HEAD(hd, sblk, sb_next);

	return (sblk);
}

static struct data_blk *
dblk_lookup(ufs2_daddr_t blk)
{
	struct data_blk *dblk;
	struct dblkhd *hd;

	hd = &dbhash[SUJ_HASH(fragstoblks(fs, blk))];
	if (lastblk && lastblk->db_blk == blk)
		return (lastblk);
	LIST_FOREACH(dblk, hd, db_next)
		if (dblk->db_blk == blk)
			return (dblk);
	/*
	 * The inode block wasn't located, allocate a new one.
	 */
	dblk = errmalloc(sizeof(*dblk));
	bzero(dblk, sizeof(*dblk));
	LIST_INSERT_HEAD(hd, dblk, db_next);
	dblk->db_blk = blk;
	return (dblk);
}

static uint8_t *
dblk_read(ufs2_daddr_t blk, int size)
{
	struct data_blk *dblk;

	dblk = dblk_lookup(blk);
	/*
	 * I doubt size mismatches can happen in practice but it is trivial
	 * to handle.
	 */
	if (size != dblk->db_size) {
		if (dblk->db_buf)
			free(dblk->db_buf);
		dblk->db_buf = errmalloc(size);
		dblk->db_size = size;
		if (bread(disk, fsbtodb(fs, blk), dblk->db_buf, size) == -1)
			err_suj("Failed to read data block %jd\n", blk);
	}
	return (dblk->db_buf);
}

static void
dblk_dirty(ufs2_daddr_t blk)
{
	struct data_blk *dblk;

	dblk = dblk_lookup(blk);
	dblk->db_dirty = 1;
}

static void
dblk_write(void)
{
	struct data_blk *dblk;
	int i;

	for (i = 0; i < SUJ_HASHSIZE; i++) {
		LIST_FOREACH(dblk, &dbhash[i], db_next) {
			if (dblk->db_dirty == 0 || dblk->db_size == 0)
				continue;
			if (bwrite(disk, fsbtodb(fs, dblk->db_blk),
			    dblk->db_buf, dblk->db_size) == -1)
				err_suj("Unable to write block %jd\n",
				    dblk->db_blk);
		}
	}
}

static union dinode *
ino_read(ino_t ino)
{
	struct ino_blk *iblk;
	struct iblkhd *hd;
	struct suj_cg *sc;
	ufs2_daddr_t blk;
	int off;

	blk = ino_to_fsba(fs, ino);
	sc = cg_lookup(ino_to_cg(fs, ino));
	iblk = sc->sc_lastiblk;
	if (iblk && iblk->ib_blk == blk)
		goto found;
	hd = &sc->sc_iblkhash[SUJ_HASH(fragstoblks(fs, blk))];
	LIST_FOREACH(iblk, hd, ib_next)
		if (iblk->ib_blk == blk)
			goto found;
	/*
	 * The inode block wasn't located, allocate a new one.
	 */
	iblk = errmalloc(sizeof(*iblk));
	bzero(iblk, sizeof(*iblk));
	iblk->ib_buf = errmalloc(fs->fs_bsize);
	iblk->ib_blk = blk;
	LIST_INSERT_HEAD(hd, iblk, ib_next);
	if (bread(disk, fsbtodb(fs, blk), iblk->ib_buf, fs->fs_bsize) == -1)
		err_suj("Failed to read inode block %jd\n", blk);
found:
	sc->sc_lastiblk = iblk;
	off = ino_to_fsbo(fs, ino);
	if (fs->fs_magic == FS_UFS1_MAGIC)
		return (union dinode *)&((struct ufs1_dinode *)iblk->ib_buf)[off];
	else
		return (union dinode *)&((struct ufs2_dinode *)iblk->ib_buf)[off];
}

static void
ino_dirty(ino_t ino)
{
	struct ino_blk *iblk;
	struct iblkhd *hd;
	struct suj_cg *sc;
	ufs2_daddr_t blk;

	blk = ino_to_fsba(fs, ino);
	sc = cg_lookup(ino_to_cg(fs, ino));
	iblk = sc->sc_lastiblk;
	if (iblk && iblk->ib_blk == blk) {
		iblk->ib_dirty = 1;
		return;
	}
	hd = &sc->sc_iblkhash[SUJ_HASH(fragstoblks(fs, blk))];
	LIST_FOREACH(iblk, hd, ib_next) {
		if (iblk->ib_blk == blk) {
			iblk->ib_dirty = 1;
			return;
		}
	}
	ino_read(ino);
	ino_dirty(ino);
}

static void
iblk_write(struct ino_blk *iblk)
{

	if (iblk->ib_dirty == 0)
		return;
	if (bwrite(disk, fsbtodb(fs, iblk->ib_blk), iblk->ib_buf,
	    fs->fs_bsize) == -1)
		err_suj("Failed to write inode block %jd\n", iblk->ib_blk);
}

static int
blk_overlaps(struct jblkrec *brec, ufs2_daddr_t start, int frags)
{
	ufs2_daddr_t bstart;
	ufs2_daddr_t bend;
	ufs2_daddr_t end;

	end = start + frags;
	bstart = brec->jb_blkno + brec->jb_oldfrags;
	bend = bstart + brec->jb_frags;
	if (start < bend && end > bstart)
		return (1);
	return (0);
}

static int
blk_equals(struct jblkrec *brec, ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t start,
    int frags)
{

	if (brec->jb_ino != ino || brec->jb_lbn != lbn)
		return (0);
	if (brec->jb_blkno + brec->jb_oldfrags != start)
		return (0);
	if (brec->jb_frags < frags)
		return (0);
	return (1);
}

static void
blk_setmask(struct jblkrec *brec, int *mask)
{
	int i;

	for (i = brec->jb_oldfrags; i < brec->jb_oldfrags + brec->jb_frags; i++)
		*mask |= 1 << i;
}

/*
 * Determine whether a given block has been reallocated to a new location.
 * Returns a mask of overlapping bits if any frags have been reused or
 * zero if the block has not been re-used and the contents can be trusted.
 *
 * This is used to ensure that an orphaned pointer due to truncate is safe
 * to be freed.  The mask value can be used to free partial blocks.
 */
static int
blk_freemask(ufs2_daddr_t blk, ino_t ino, ufs_lbn_t lbn, int frags)
{
	struct suj_blk *sblk;
	struct suj_rec *srec;
	struct jblkrec *brec;
	int mask;
	int off;

	/*
	 * To be certain we're not freeing a reallocated block we lookup
	 * this block in the blk hash and see if there is an allocation
	 * journal record that overlaps with any fragments in the block
	 * we're concerned with.  If any fragments have ben reallocated
	 * the block has already been freed and re-used for another purpose.
	 */
	mask = 0;
	sblk = blk_lookup(blknum(fs, blk), 0);
	if (sblk == NULL)
		return (0);
	off = blk - sblk->sb_blk;
	TAILQ_FOREACH(srec, &sblk->sb_recs, sr_next) {
		brec = (struct jblkrec *)srec->sr_rec;
		/*
		 * If the block overlaps but does not match
		 * exactly this record refers to the current
		 * location.
		 */
		if (blk_overlaps(brec, blk, frags) == 0)
			continue;
		if (blk_equals(brec, ino, lbn, blk, frags) == 1)
			mask = 0;
		else
			blk_setmask(brec, &mask);
	}
	if (debug)
		printf("blk_freemask: blk %jd sblk %jd off %d mask 0x%X\n",
		    blk, sblk->sb_blk, off, mask);
	return (mask >> off);
}

/*
 * Determine whether it is safe to follow an indirect.  It is not safe
 * if any part of the indirect has been reallocated or the last journal
 * entry was an allocation.  Just allocated indirects may not have valid
 * pointers yet and all of their children will have their own records.
 * It is also not safe to follow an indirect if the cg bitmap has been
 * cleared as a new allocation may write to the block prior to the journal
 * being written.
 *
 * Returns 1 if it's safe to follow the indirect and 0 otherwise.
 */
static int
blk_isindir(ufs2_daddr_t blk, ino_t ino, ufs_lbn_t lbn)
{
	struct suj_blk *sblk;
	struct jblkrec *brec;

	sblk = blk_lookup(blk, 0);
	if (sblk == NULL)
		return (1);
	if (TAILQ_EMPTY(&sblk->sb_recs))
		return (1);
	brec = (struct jblkrec *)TAILQ_LAST(&sblk->sb_recs, srechd)->sr_rec;
	if (blk_equals(brec, ino, lbn, blk, fs->fs_frag))
		if (brec->jb_op == JOP_FREEBLK)
			return (!blk_isfree(blk));
	return (0);
}

/*
 * Clear an inode from the cg bitmap.  If the inode was already clear return
 * 0 so the caller knows it does not have to check the inode contents.
 */
static int
ino_free(ino_t ino, int mode)
{
	struct suj_cg *sc;
	uint8_t *inosused;
	struct cg *cgp;
	int cg;

	cg = ino_to_cg(fs, ino);
	ino = ino % fs->fs_ipg;
	sc = cg_lookup(cg);
	cgp = sc->sc_cgp;
	inosused = cg_inosused(cgp);
	/*
	 * The bitmap may never have made it to the disk so we have to
	 * conditionally clear.  We can avoid writing the cg in this case.
	 */
	if (isclr(inosused, ino))
		return (0);
	freeinos++;
	clrbit(inosused, ino);
	if (ino < cgp->cg_irotor)
		cgp->cg_irotor = ino;
	cgp->cg_cs.cs_nifree++;
	if ((mode & IFMT) == IFDIR) {
		freedir++;
		cgp->cg_cs.cs_ndir--;
	}
	sc->sc_dirty = 1;

	return (1);
}

/*
 * Free 'frags' frags starting at filesystem block 'bno' skipping any frags
 * set in the mask.
 */
static void
blk_free(ufs2_daddr_t bno, int mask, int frags)
{
	ufs1_daddr_t fragno, cgbno;
	struct suj_cg *sc;
	struct cg *cgp;
	int i, cg;
	uint8_t *blksfree;

	if (debug)
		printf("Freeing %d frags at blk %jd mask 0x%x\n",
		    frags, bno, mask);
	cg = dtog(fs, bno);
	sc = cg_lookup(cg);
	cgp = sc->sc_cgp;
	cgbno = dtogd(fs, bno);
	blksfree = cg_blksfree(cgp);

	/*
	 * If it's not allocated we only wrote the journal entry
	 * and never the bitmaps.  Here we unconditionally clear and
	 * resolve the cg summary later.
	 */
	if (frags == fs->fs_frag && mask == 0) {
		fragno = fragstoblks(fs, cgbno);
		ffs_setblock(fs, blksfree, fragno);
		freeblocks++;
	} else {
		/*
		 * deallocate the fragment
		 */
		for (i = 0; i < frags; i++)
			if ((mask & (1 << i)) == 0 && isclr(blksfree, cgbno +i)) {
				freefrags++;
				setbit(blksfree, cgbno + i);
			}
	}
	sc->sc_dirty = 1;
}

/*
 * Returns 1 if the whole block starting at 'bno' is marked free and 0
 * otherwise.
 */
static int
blk_isfree(ufs2_daddr_t bno)
{
	struct suj_cg *sc;

	sc = cg_lookup(dtog(fs, bno));
	return ffs_isblock(fs, cg_blksfree(sc->sc_cgp), dtogd(fs, bno));
}

/*
 * Fetch an indirect block to find the block at a given lbn.  The lbn
 * may be negative to fetch a specific indirect block pointer or positive
 * to fetch a specific block.
 */
static ufs2_daddr_t
indir_blkatoff(ufs2_daddr_t blk, ino_t ino, ufs_lbn_t cur, ufs_lbn_t lbn)
{
	ufs2_daddr_t *bap2;
	ufs2_daddr_t *bap1;
	ufs_lbn_t lbnadd;
	ufs_lbn_t base;
	int level;
	int i;

	if (blk == 0)
		return (0);
	level = lbn_level(cur);
	if (level == -1)
		err_suj("Invalid indir lbn %jd\n", lbn);
	if (level == 0 && lbn < 0)
		err_suj("Invalid lbn %jd\n", lbn);
	bap2 = (void *)dblk_read(blk, fs->fs_bsize);
	bap1 = (void *)bap2;
	lbnadd = 1;
	base = -(cur + level);
	for (i = level; i > 0; i--)
		lbnadd *= NINDIR(fs);
	if (lbn > 0)
		i = (lbn - base) / lbnadd;
	else
		i = (-lbn - base) / lbnadd;
	if (i < 0 || i >= NINDIR(fs))
		err_suj("Invalid indirect index %d produced by lbn %jd\n",
		    i, lbn);
	if (level == 0)
		cur = base + (i * lbnadd);
	else
		cur = -(base + (i * lbnadd)) - (level - 1);
	if (fs->fs_magic == FS_UFS1_MAGIC)
		blk = bap1[i];
	else
		blk = bap2[i];
	if (cur == lbn)
		return (blk);
	if (level == 0)
		err_suj("Invalid lbn %jd at level 0\n", lbn);
	return indir_blkatoff(blk, ino, cur, lbn);
}

/*
 * Finds the disk block address at the specified lbn within the inode
 * specified by ip.  This follows the whole tree and honors di_size and
 * di_extsize so it is a true test of reachability.  The lbn may be
 * negative if an extattr or indirect block is requested.
 */
static ufs2_daddr_t
ino_blkatoff(union dinode *ip, ino_t ino, ufs_lbn_t lbn, int *frags)
{
	ufs_lbn_t tmpval;
	ufs_lbn_t cur;
	ufs_lbn_t next;
	int i;

	/*
	 * Handle extattr blocks first.
	 */
	if (lbn < 0 && lbn >= -NXADDR) {
		lbn = -1 - lbn;
		if (lbn > lblkno(fs, ip->dp2.di_extsize - 1))
			return (0);
		*frags = numfrags(fs, sblksize(fs, ip->dp2.di_extsize, lbn));
		return (ip->dp2.di_extb[lbn]);
	}
	/*
	 * Now direct and indirect.
	 */
	if (DIP(ip, di_mode) == IFLNK &&
	    DIP(ip, di_size) < fs->fs_maxsymlinklen)
		return (0);
	if (lbn >= 0 && lbn < NDADDR) {
		*frags = numfrags(fs, sblksize(fs, DIP(ip, di_size), lbn));
		return (DIP(ip, di_db[lbn]));
	}
	*frags = fs->fs_frag;

	for (i = 0, tmpval = NINDIR(fs), cur = NDADDR; i < NIADDR; i++,
	    tmpval *= NINDIR(fs), cur = next) {
		next = cur + tmpval;
		if (lbn == -cur - i)
			return (DIP(ip, di_ib[i]));
		/*
		 * Determine whether the lbn in question is within this tree.
		 */
		if (lbn < 0 && -lbn >= next)
			continue;
		if (lbn > 0 && lbn >= next)
			continue;
		return indir_blkatoff(DIP(ip, di_ib[i]), ino, -cur - i, lbn);
	}
	err_suj("lbn %jd not in ino\n", lbn);
	/* NOTREACHED */
}

/*
 * Determine whether a block exists at a particular lbn in an inode.
 * Returns 1 if found, 0 if not.  lbn may be negative for indirects
 * or ext blocks.
 */
static int
blk_isat(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, int *frags)
{
	union dinode *ip;
	ufs2_daddr_t nblk;

	ip = ino_read(ino);

	if (DIP(ip, di_nlink) == 0 || DIP(ip, di_mode) == 0)
		return (0);
	nblk = ino_blkatoff(ip, ino, lbn, frags);

	return (nblk == blk);
}

/*
 * Clear the directory entry at diroff that should point to child.  Minimal
 * checking is done and it is assumed that this path was verified with isat.
 */
static void
ino_clrat(ino_t parent, off_t diroff, ino_t child)
{
	union dinode *dip;
	struct direct *dp;
	ufs2_daddr_t blk;
	uint8_t *block;
	ufs_lbn_t lbn;
	int blksize;
	int frags;
	int doff;

	if (debug)
		printf("Clearing inode %ju from parent %ju at offset %jd\n",
		    (uintmax_t)child, (uintmax_t)parent, diroff);

	lbn = lblkno(fs, diroff);
	doff = blkoff(fs, diroff);
	dip = ino_read(parent);
	blk = ino_blkatoff(dip, parent, lbn, &frags);
	blksize = sblksize(fs, DIP(dip, di_size), lbn);
	block = dblk_read(blk, blksize);
	dp = (struct direct *)&block[doff];
	if (dp->d_ino != child)
		errx(1, "Inode %ju does not exist in %ju at %jd",
		    (uintmax_t)child, (uintmax_t)parent, diroff);
	dp->d_ino = 0;
	dblk_dirty(blk);
	/*
	 * The actual .. reference count will already have been removed
	 * from the parent by the .. remref record.
	 */
}

/*
 * Determines whether a pointer to an inode exists within a directory
 * at a specified offset.  Returns the mode of the found entry.
 */
static int
ino_isat(ino_t parent, off_t diroff, ino_t child, int *mode, int *isdot)
{
	union dinode *dip;
	struct direct *dp;
	ufs2_daddr_t blk;
	uint8_t *block;
	ufs_lbn_t lbn;
	int blksize;
	int frags;
	int dpoff;
	int doff;

	*isdot = 0;
	dip = ino_read(parent);
	*mode = DIP(dip, di_mode);
	if ((*mode & IFMT) != IFDIR) {
		if (debug) {
			/*
			 * This can happen if the parent inode
			 * was reallocated.
			 */
			if (*mode != 0)
				printf("Directory %ju has bad mode %o\n",
				    (uintmax_t)parent, *mode);
			else
				printf("Directory %ju has zero mode\n",
				    (uintmax_t)parent);
		}
		return (0);
	}
	lbn = lblkno(fs, diroff);
	doff = blkoff(fs, diroff);
	blksize = sblksize(fs, DIP(dip, di_size), lbn);
	if (diroff + DIRECTSIZ(1) > DIP(dip, di_size) || doff >= blksize) {
		if (debug)
			printf("ino %ju absent from %ju due to offset %jd"
			    " exceeding size %jd\n",
			    (uintmax_t)child, (uintmax_t)parent, diroff,
			    DIP(dip, di_size));
		return (0);
	}
	blk = ino_blkatoff(dip, parent, lbn, &frags);
	if (blk <= 0) {
		if (debug)
			printf("Sparse directory %ju", (uintmax_t)parent);
		return (0);
	}
	block = dblk_read(blk, blksize);
	/*
	 * Walk through the records from the start of the block to be
	 * certain we hit a valid record and not some junk in the middle
	 * of a file name.  Stop when we reach or pass the expected offset.
	 */
	dpoff = (doff / DIRBLKSIZ) * DIRBLKSIZ;
	do {
		dp = (struct direct *)&block[dpoff];
		if (dpoff == doff)
			break;
		if (dp->d_reclen == 0)
			break;
		dpoff += dp->d_reclen;
	} while (dpoff <= doff);
	if (dpoff > fs->fs_bsize)
		err_suj("Corrupt directory block in dir ino %ju\n",
		    (uintmax_t)parent);
	/* Not found. */
	if (dpoff != doff) {
		if (debug)
			printf("ino %ju not found in %ju, lbn %jd, dpoff %d\n",
			    (uintmax_t)child, (uintmax_t)parent, lbn, dpoff);
		return (0);
	}
	/*
	 * We found the item in question.  Record the mode and whether it's
	 * a . or .. link for the caller.
	 */
	if (dp->d_ino == child) {
		if (child == parent)
			*isdot = 1;
		else if (dp->d_namlen == 2 &&
		    dp->d_name[0] == '.' && dp->d_name[1] == '.')
			*isdot = 1;
		*mode = DTTOIF(dp->d_type);
		return (1);
	}
	if (debug)
		printf("ino %ju doesn't match dirent ino %ju in parent %ju\n",
		    (uintmax_t)child, (uintmax_t)dp->d_ino, (uintmax_t)parent);
	return (0);
}

#define	VISIT_INDIR	0x0001
#define	VISIT_EXT	0x0002
#define	VISIT_ROOT	0x0004	/* Operation came via root & valid pointers. */

/*
 * Read an indirect level which may or may not be linked into an inode.
 */
static void
indir_visit(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, uint64_t *frags,
    ino_visitor visitor, int flags)
{
	ufs2_daddr_t *bap2;
	ufs1_daddr_t *bap1;
	ufs_lbn_t lbnadd;
	ufs2_daddr_t nblk;
	ufs_lbn_t nlbn;
	int level;
	int i;

	/*
	 * Don't visit indirect blocks with contents we can't trust.  This
	 * should only happen when indir_visit() is called to complete a
	 * truncate that never finished and not when a pointer is found via
	 * an inode.
	 */
	if (blk == 0)
		return;
	level = lbn_level(lbn);
	if (level == -1)
		err_suj("Invalid level for lbn %jd\n", lbn);
	if ((flags & VISIT_ROOT) == 0 && blk_isindir(blk, ino, lbn) == 0) {
		if (debug)
			printf("blk %jd ino %ju lbn %jd(%d) is not indir.\n",
			    blk, (uintmax_t)ino, lbn, level);
		goto out;
	}
	lbnadd = 1;
	for (i = level; i > 0; i--)
		lbnadd *= NINDIR(fs);
	bap1 = (void *)dblk_read(blk, fs->fs_bsize);
	bap2 = (void *)bap1;
	for (i = 0; i < NINDIR(fs); i++) {
		if (fs->fs_magic == FS_UFS1_MAGIC)
			nblk = *bap1++;
		else
			nblk = *bap2++;
		if (nblk == 0)
			continue;
		if (level == 0) {
			nlbn = -lbn + i * lbnadd;
			(*frags) += fs->fs_frag;
			visitor(ino, nlbn, nblk, fs->fs_frag);
		} else {
			nlbn = (lbn + 1) - (i * lbnadd);
			indir_visit(ino, nlbn, nblk, frags, visitor, flags);
		}
	}
out:
	if (flags & VISIT_INDIR) {
		(*frags) += fs->fs_frag;
		visitor(ino, lbn, blk, fs->fs_frag);
	}
}

/*
 * Visit each block in an inode as specified by 'flags' and call a
 * callback function.  The callback may inspect or free blocks.  The
 * count of frags found according to the size in the file is returned.
 * This is not valid for sparse files but may be used to determine
 * the correct di_blocks for a file.
 */
static uint64_t
ino_visit(union dinode *ip, ino_t ino, ino_visitor visitor, int flags)
{
	ufs_lbn_t nextlbn;
	ufs_lbn_t tmpval;
	ufs_lbn_t lbn;
	uint64_t size;
	uint64_t fragcnt;
	int mode;
	int frags;
	int i;

	size = DIP(ip, di_size);
	mode = DIP(ip, di_mode) & IFMT;
	fragcnt = 0;
	if ((flags & VISIT_EXT) &&
	    fs->fs_magic == FS_UFS2_MAGIC && ip->dp2.di_extsize) {
		for (i = 0; i < NXADDR; i++) {
			if (ip->dp2.di_extb[i] == 0)
				continue;
			frags = sblksize(fs, ip->dp2.di_extsize, i);
			frags = numfrags(fs, frags);
			fragcnt += frags;
			visitor(ino, -1 - i, ip->dp2.di_extb[i], frags);
		}
	}
	/* Skip datablocks for short links and devices. */
	if (mode == IFBLK || mode == IFCHR ||
	    (mode == IFLNK && size < fs->fs_maxsymlinklen))
		return (fragcnt);
	for (i = 0; i < NDADDR; i++) {
		if (DIP(ip, di_db[i]) == 0)
			continue;
		frags = sblksize(fs, size, i);
		frags = numfrags(fs, frags);
		fragcnt += frags;
		visitor(ino, i, DIP(ip, di_db[i]), frags);
	}
	/*
	 * We know the following indirects are real as we're following
	 * real pointers to them.
	 */
	flags |= VISIT_ROOT;
	for (i = 0, tmpval = NINDIR(fs), lbn = NDADDR; i < NIADDR; i++,
	    lbn = nextlbn) {
		nextlbn = lbn + tmpval;
		tmpval *= NINDIR(fs);
		if (DIP(ip, di_ib[i]) == 0)
			continue;
		indir_visit(ino, -lbn - i, DIP(ip, di_ib[i]), &fragcnt, visitor,
		    flags);
	}
	return (fragcnt);
}

/*
 * Null visitor function used when we just want to count blocks and
 * record the lbn.
 */
ufs_lbn_t visitlbn;
static void
null_visit(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, int frags)
{
	if (lbn > 0)
		visitlbn = lbn;
}

/*
 * Recalculate di_blocks when we discover that a block allocation or
 * free was not successfully completed.  The kernel does not roll this back
 * because it would be too expensive to compute which indirects were
 * reachable at the time the inode was written.
 */
static void
ino_adjblks(struct suj_ino *sino)
{
	union dinode *ip;
	uint64_t blocks;
	uint64_t frags;
	off_t isize;
	off_t size;
	ino_t ino;

	ino = sino->si_ino;
	ip = ino_read(ino);
	/* No need to adjust zero'd inodes. */
	if (DIP(ip, di_mode) == 0)
		return;
	/*
	 * Visit all blocks and count them as well as recording the last
	 * valid lbn in the file.  If the file size doesn't agree with the
	 * last lbn we need to truncate to fix it.  Otherwise just adjust
	 * the blocks count.
	 */
	visitlbn = 0;
	frags = ino_visit(ip, ino, null_visit, VISIT_INDIR | VISIT_EXT);
	blocks = fsbtodb(fs, frags);
	/*
	 * We assume the size and direct block list is kept coherent by
	 * softdep.  For files that have extended into indirects we truncate
	 * to the size in the inode or the maximum size permitted by
	 * populated indirects.
	 */
	if (visitlbn >= NDADDR) {
		isize = DIP(ip, di_size);
		size = lblktosize(fs, visitlbn + 1);
		if (isize > size)
			isize = size;
		/* Always truncate to free any unpopulated indirects. */
		ino_trunc(sino->si_ino, isize);
		return;
	}
	if (blocks == DIP(ip, di_blocks))
		return;
	if (debug)
		printf("ino %ju adjusting block count from %jd to %jd\n",
		    (uintmax_t)ino, DIP(ip, di_blocks), blocks);
	DIP_SET(ip, di_blocks, blocks);
	ino_dirty(ino);
}

static void
blk_free_visit(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, int frags)
{

	blk_free(blk, blk_freemask(blk, ino, lbn, frags), frags);
}

/*
 * Free a block or tree of blocks that was previously rooted in ino at
 * the given lbn.  If the lbn is an indirect all children are freed
 * recursively.
 */
static void
blk_free_lbn(ufs2_daddr_t blk, ino_t ino, ufs_lbn_t lbn, int frags, int follow)
{
	uint64_t resid;
	int mask;

	mask = blk_freemask(blk, ino, lbn, frags);
	resid = 0;
	if (lbn <= -NDADDR && follow && mask == 0)
		indir_visit(ino, lbn, blk, &resid, blk_free_visit, VISIT_INDIR);
	else
		blk_free(blk, mask, frags);
}

static void
ino_setskip(struct suj_ino *sino, ino_t parent)
{
	int isdot;
	int mode;

	if (ino_isat(sino->si_ino, DOTDOT_OFFSET, parent, &mode, &isdot))
		sino->si_skipparent = 1;
}

static void
ino_remref(ino_t parent, ino_t child, uint64_t diroff, int isdotdot)
{
	struct suj_ino *sino;
	struct suj_rec *srec;
	struct jrefrec *rrec;

	/*
	 * Lookup this inode to see if we have a record for it.
	 */
	sino = ino_lookup(child, 0);
	/*
	 * Tell any child directories we've already removed their
	 * parent link cnt.  Don't try to adjust our link down again.
	 */
	if (sino != NULL && isdotdot == 0)
		ino_setskip(sino, parent);
	/*
	 * No valid record for this inode.  Just drop the on-disk
	 * link by one.
	 */
	if (sino == NULL || sino->si_hasrecs == 0) {
		ino_decr(child);
		return;
	}
	/*
	 * Use ino_adjust() if ino_check() has already processed this
	 * child.  If we lose the last non-dot reference to a
	 * directory it will be discarded.
	 */
	if (sino->si_linkadj) {
		sino->si_nlink--;
		if (isdotdot)
			sino->si_dotlinks--;
		ino_adjust(sino);
		return;
	}
	/*
	 * If we haven't yet processed this inode we need to make
	 * sure we will successfully discover the lost path.  If not
	 * use nlinkadj to remember.
	 */
	TAILQ_FOREACH(srec, &sino->si_recs, sr_next) {
		rrec = (struct jrefrec *)srec->sr_rec;
		if (rrec->jr_parent == parent &&
		    rrec->jr_diroff == diroff)
			return;
	}
	sino->si_nlinkadj++;
}

/*
 * Free the children of a directory when the directory is discarded.
 */
static void
ino_free_children(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, int frags)
{
	struct suj_ino *sino;
	struct direct *dp;
	off_t diroff;
	uint8_t *block;
	int skipparent;
	int isdotdot;
	int dpoff;
	int size;

	sino = ino_lookup(ino, 0);
	if (sino)
		skipparent = sino->si_skipparent;
	else
		skipparent = 0;
	size = lfragtosize(fs, frags);
	block = dblk_read(blk, size);
	dp = (struct direct *)&block[0];
	for (dpoff = 0; dpoff < size && dp->d_reclen; dpoff += dp->d_reclen) {
		dp = (struct direct *)&block[dpoff];
		if (dp->d_ino == 0 || dp->d_ino == WINO)
			continue;
		if (dp->d_namlen == 1 && dp->d_name[0] == '.')
			continue;
		isdotdot = dp->d_namlen == 2 && dp->d_name[0] == '.' &&
		    dp->d_name[1] == '.';
		if (isdotdot && skipparent == 1)
			continue;
		if (debug)
			printf("Directory %ju removing ino %ju name %s\n",
			    (uintmax_t)ino, (uintmax_t)dp->d_ino, dp->d_name);
		diroff = lblktosize(fs, lbn) + dpoff;
		ino_remref(ino, dp->d_ino, diroff, isdotdot);
	}
}

/*
 * Reclaim an inode, freeing all blocks and decrementing all children's
 * link counts.  Free the inode back to the cg.
 */
static void
ino_reclaim(union dinode *ip, ino_t ino, int mode)
{
	uint32_t gen;

	if (ino == ROOTINO)
		err_suj("Attempting to free ROOTINO\n");
	if (debug)
		printf("Truncating and freeing ino %ju, nlink %d, mode %o\n",
		    (uintmax_t)ino, DIP(ip, di_nlink), DIP(ip, di_mode));

	/* We are freeing an inode or directory. */
	if ((DIP(ip, di_mode) & IFMT) == IFDIR)
		ino_visit(ip, ino, ino_free_children, 0);
	DIP_SET(ip, di_nlink, 0);
	ino_visit(ip, ino, blk_free_visit, VISIT_EXT | VISIT_INDIR);
	/* Here we have to clear the inode and release any blocks it holds. */
	gen = DIP(ip, di_gen);
	if (fs->fs_magic == FS_UFS1_MAGIC)
		bzero(ip, sizeof(struct ufs1_dinode));
	else
		bzero(ip, sizeof(struct ufs2_dinode));
	DIP_SET(ip, di_gen, gen);
	ino_dirty(ino);
	ino_free(ino, mode);
	return;
}

/*
 * Adjust an inode's link count down by one when a directory goes away.
 */
static void
ino_decr(ino_t ino)
{
	union dinode *ip;
	int reqlink;
	int nlink;
	int mode;

	ip = ino_read(ino);
	nlink = DIP(ip, di_nlink);
	mode = DIP(ip, di_mode);
	if (nlink < 1)
		err_suj("Inode %d link count %d invalid\n", ino, nlink);
	if (mode == 0)
		err_suj("Inode %d has a link of %d with 0 mode\n", ino, nlink);
	nlink--;
	if ((mode & IFMT) == IFDIR)
		reqlink = 2;
	else
		reqlink = 1;
	if (nlink < reqlink) {
		if (debug)
			printf("ino %ju not enough links to live %d < %d\n",
			    (uintmax_t)ino, nlink, reqlink);
		ino_reclaim(ip, ino, mode);
		return;
	}
	DIP_SET(ip, di_nlink, nlink);
	ino_dirty(ino);
}

/*
 * Adjust the inode link count to 'nlink'.  If the count reaches zero
 * free it.
 */
static void
ino_adjust(struct suj_ino *sino)
{
	struct jrefrec *rrec;
	struct suj_rec *srec;
	struct suj_ino *stmp;
	union dinode *ip;
	nlink_t nlink;
	int recmode;
	int reqlink;
	int isdot;
	int mode;
	ino_t ino;

	nlink = sino->si_nlink;
	ino = sino->si_ino;
	mode = sino->si_mode & IFMT;
	/*
	 * If it's a directory with no dot links, it was truncated before
	 * the name was cleared.  We need to clear the dirent that
	 * points at it.
	 */
	if (mode == IFDIR && nlink == 1 && sino->si_dotlinks == 0) {
		sino->si_nlink = nlink = 0;
		TAILQ_FOREACH(srec, &sino->si_recs, sr_next) {
			rrec = (struct jrefrec *)srec->sr_rec;
			if (ino_isat(rrec->jr_parent, rrec->jr_diroff, ino,
			    &recmode, &isdot) == 0)
				continue;
			ino_clrat(rrec->jr_parent, rrec->jr_diroff, ino);
			break;
		}
		if (srec == NULL)
			errx(1, "Directory %ju name not found", (uintmax_t)ino);
	}
	/*
	 * If it's a directory with no real names pointing to it go ahead
	 * and truncate it.  This will free any children.
	 */
	if (mode == IFDIR && nlink - sino->si_dotlinks == 0) {
		sino->si_nlink = nlink = 0;
		/*
		 * Mark any .. links so they know not to free this inode
		 * when they are removed.
		 */
		TAILQ_FOREACH(srec, &sino->si_recs, sr_next) {
			rrec = (struct jrefrec *)srec->sr_rec;
			if (rrec->jr_diroff == DOTDOT_OFFSET) {
				stmp = ino_lookup(rrec->jr_parent, 0);
				if (stmp)
					ino_setskip(stmp, ino);
			}
		}
	}
	ip = ino_read(ino);
	mode = DIP(ip, di_mode) & IFMT;
	if (nlink > LINK_MAX)
		err_suj("ino %ju nlink manipulation error, new %d, old %d\n",
		    (uintmax_t)ino, nlink, DIP(ip, di_nlink));
	if (debug)
		printf("Adjusting ino %ju, nlink %d, old link %d lastmode %o\n",
		    (uintmax_t)ino, nlink, DIP(ip, di_nlink), sino->si_mode);
	if (mode == 0) {
		if (debug)
			printf("ino %ju, zero inode freeing bitmap\n",
			    (uintmax_t)ino);
		ino_free(ino, sino->si_mode);
		return;
	}
	/* XXX Should be an assert? */
	if (mode != sino->si_mode && debug)
		printf("ino %ju, mode %o != %o\n",
		    (uintmax_t)ino, mode, sino->si_mode);
	if ((mode & IFMT) == IFDIR)
		reqlink = 2;
	else
		reqlink = 1;
	/* If the inode doesn't have enough links to live, free it. */
	if (nlink < reqlink) {
		if (debug)
			printf("ino %ju not enough links to live %d < %d\n",
			    (uintmax_t)ino, nlink, reqlink);
		ino_reclaim(ip, ino, mode);
		return;
	}
	/* If required write the updated link count. */
	if (DIP(ip, di_nlink) == nlink) {
		if (debug)
			printf("ino %ju, link matches, skipping.\n",
			    (uintmax_t)ino);
		return;
	}
	DIP_SET(ip, di_nlink, nlink);
	ino_dirty(ino);
}

/*
 * Truncate some or all blocks in an indirect, freeing any that are required
 * and zeroing the indirect.
 */
static void
indir_trunc(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, ufs_lbn_t lastlbn)
{
	ufs2_daddr_t *bap2;
	ufs1_daddr_t *bap1;
	ufs_lbn_t lbnadd;
	ufs2_daddr_t nblk;
	ufs_lbn_t next;
	ufs_lbn_t nlbn;
	int dirty;
	int level;
	int i;

	if (blk == 0)
		return;
	dirty = 0;
	level = lbn_level(lbn);
	if (level == -1)
		err_suj("Invalid level for lbn %jd\n", lbn);
	lbnadd = 1;
	for (i = level; i > 0; i--)
		lbnadd *= NINDIR(fs);
	bap1 = (void *)dblk_read(blk, fs->fs_bsize);
	bap2 = (void *)bap1;
	for (i = 0; i < NINDIR(fs); i++) {
		if (fs->fs_magic == FS_UFS1_MAGIC)
			nblk = *bap1++;
		else
			nblk = *bap2++;
		if (nblk == 0)
			continue;
		if (level != 0) {
			nlbn = (lbn + 1) - (i * lbnadd);
			/*
			 * Calculate the lbn of the next indirect to
			 * determine if any of this indirect must be
			 * reclaimed.
			 */
			next = -(lbn + level) + ((i+1) * lbnadd);
			if (next <= lastlbn)
				continue;
			indir_trunc(ino, nlbn, nblk, lastlbn);
			/* If all of this indirect was reclaimed, free it. */
			nlbn = next - lbnadd;
			if (nlbn < lastlbn)
				continue;
		} else {
			nlbn = -lbn + i * lbnadd;
			if (nlbn < lastlbn)
				continue;
		}
		dirty = 1;
		blk_free(nblk, 0, fs->fs_frag);
		if (fs->fs_magic == FS_UFS1_MAGIC)
			*(bap1 - 1) = 0;
		else
			*(bap2 - 1) = 0;
	}
	if (dirty)
		dblk_dirty(blk);
}

/*
 * Truncate an inode to the minimum of the given size or the last populated
 * block after any over size have been discarded.  The kernel would allocate
 * the last block in the file but fsck does not and neither do we.  This
 * code never extends files, only shrinks them.
 */
static void
ino_trunc(ino_t ino, off_t size)
{
	union dinode *ip;
	ufs2_daddr_t bn;
	uint64_t totalfrags;
	ufs_lbn_t nextlbn;
	ufs_lbn_t lastlbn;
	ufs_lbn_t tmpval;
	ufs_lbn_t lbn;
	ufs_lbn_t i;
	int frags;
	off_t cursize;
	off_t off;
	int mode;

	ip = ino_read(ino);
	mode = DIP(ip, di_mode) & IFMT;
	cursize = DIP(ip, di_size);
	if (debug)
		printf("Truncating ino %ju, mode %o to size %jd from size %jd\n",
		    (uintmax_t)ino, mode, size, cursize);

	/* Skip datablocks for short links and devices. */
	if (mode == 0 || mode == IFBLK || mode == IFCHR ||
	    (mode == IFLNK && cursize < fs->fs_maxsymlinklen))
		return;
	/* Don't extend. */
	if (size > cursize)
		size = cursize;
	lastlbn = lblkno(fs, blkroundup(fs, size));
	for (i = lastlbn; i < NDADDR; i++) {
		if (DIP(ip, di_db[i]) == 0)
			continue;
		frags = sblksize(fs, cursize, i);
		frags = numfrags(fs, frags);
		blk_free(DIP(ip, di_db[i]), 0, frags);
		DIP_SET(ip, di_db[i], 0);
	}
	/*
	 * Follow indirect blocks, freeing anything required.
	 */
	for (i = 0, tmpval = NINDIR(fs), lbn = NDADDR; i < NIADDR; i++,
	    lbn = nextlbn) {
		nextlbn = lbn + tmpval;
		tmpval *= NINDIR(fs);
		/* If we're not freeing any in this indirect range skip it. */
		if (lastlbn >= nextlbn)
			continue;
		if (DIP(ip, di_ib[i]) == 0)
			continue;
		indir_trunc(ino, -lbn - i, DIP(ip, di_ib[i]), lastlbn);
		/* If we freed everything in this indirect free the indir. */
		if (lastlbn > lbn)
			continue;
		blk_free(DIP(ip, di_ib[i]), 0, frags);
		DIP_SET(ip, di_ib[i], 0);
	}
	ino_dirty(ino);
	/*
	 * Now that we've freed any whole blocks that exceed the desired
	 * truncation size, figure out how many blocks remain and what the
	 * last populated lbn is.  We will set the size to this last lbn
	 * rather than worrying about allocating the final lbn as the kernel
	 * would've done.  This is consistent with normal fsck behavior.
	 */
	visitlbn = 0;
	totalfrags = ino_visit(ip, ino, null_visit, VISIT_INDIR | VISIT_EXT);
	if (size > lblktosize(fs, visitlbn + 1))
		size = lblktosize(fs, visitlbn + 1);
	/*
	 * If we're truncating direct blocks we have to adjust frags
	 * accordingly.
	 */
	if (visitlbn < NDADDR && totalfrags) {
		long oldspace, newspace;

		bn = DIP(ip, di_db[visitlbn]);
		if (bn == 0)
			err_suj("Bad blk at ino %ju lbn %jd\n",
			    (uintmax_t)ino, visitlbn);
		oldspace = sblksize(fs, cursize, visitlbn);
		newspace = sblksize(fs, size, visitlbn);
		if (oldspace != newspace) {
			bn += numfrags(fs, newspace);
			frags = numfrags(fs, oldspace - newspace);
			blk_free(bn, 0, frags);
			totalfrags -= frags;
		}
	}
	DIP_SET(ip, di_blocks, fsbtodb(fs, totalfrags));
	DIP_SET(ip, di_size, size);
	/*
	 * If we've truncated into the middle of a block or frag we have
	 * to zero it here.  Otherwise the file could extend into
	 * uninitialized space later.
	 */
	off = blkoff(fs, size);
	if (off && DIP(ip, di_mode) != IFDIR) {
		uint8_t *buf;
		long clrsize;

		bn = ino_blkatoff(ip, ino, visitlbn, &frags);
		if (bn == 0)
			err_suj("Block missing from ino %ju at lbn %jd\n",
			    (uintmax_t)ino, visitlbn);
		clrsize = frags * fs->fs_fsize;
		buf = dblk_read(bn, clrsize);
		clrsize -= off;
		buf += off;
		bzero(buf, clrsize);
		dblk_dirty(bn);
	}
	return;
}

/*
 * Process records available for one inode and determine whether the
 * link count is correct or needs adjusting.
 */
static void
ino_check(struct suj_ino *sino)
{
	struct suj_rec *srec;
	struct jrefrec *rrec;
	nlink_t dotlinks;
	int newlinks;
	int removes;
	int nlink;
	ino_t ino;
	int isdot;
	int isat;
	int mode;

	if (sino->si_hasrecs == 0)
		return;
	ino = sino->si_ino;
	rrec = (struct jrefrec *)TAILQ_FIRST(&sino->si_recs)->sr_rec;
	nlink = rrec->jr_nlink;
	newlinks = 0;
	dotlinks = 0;
	removes = sino->si_nlinkadj;
	TAILQ_FOREACH(srec, &sino->si_recs, sr_next) {
		rrec = (struct jrefrec *)srec->sr_rec;
		isat = ino_isat(rrec->jr_parent, rrec->jr_diroff,
		    rrec->jr_ino, &mode, &isdot);
		if (isat && (mode & IFMT) != (rrec->jr_mode & IFMT))
			err_suj("Inode mode/directory type mismatch %o != %o\n",
			    mode, rrec->jr_mode);
		if (debug)
			printf("jrefrec: op %d ino %ju, nlink %d, parent %d, "
			    "diroff %jd, mode %o, isat %d, isdot %d\n",
			    rrec->jr_op, (uintmax_t)rrec->jr_ino,
			    rrec->jr_nlink, rrec->jr_parent, rrec->jr_diroff,
			    rrec->jr_mode, isat, isdot);
		mode = rrec->jr_mode & IFMT;
		if (rrec->jr_op == JOP_REMREF)
			removes++;
		newlinks += isat;
		if (isdot)
			dotlinks += isat;
	}
	/*
	 * The number of links that remain are the starting link count
	 * subtracted by the total number of removes with the total
	 * links discovered back in.  An incomplete remove thus
	 * makes no change to the link count but an add increases
	 * by one.
	 */
	if (debug)
		printf("ino %ju nlink %d newlinks %d removes %d dotlinks %d\n",
		    (uintmax_t)ino, nlink, newlinks, removes, dotlinks);
	nlink += newlinks;
	nlink -= removes;
	sino->si_linkadj = 1;
	sino->si_nlink = nlink;
	sino->si_dotlinks = dotlinks;
	sino->si_mode = mode;
	ino_adjust(sino);
}

/*
 * Process records available for one block and determine whether it is
 * still allocated and whether the owning inode needs to be updated or
 * a free completed.
 */
static void
blk_check(struct suj_blk *sblk)
{
	struct suj_rec *srec;
	struct jblkrec *brec;
	struct suj_ino *sino;
	ufs2_daddr_t blk;
	int mask;
	int frags;
	int isat;

	/*
	 * Each suj_blk actually contains records for any fragments in that
	 * block.  As a result we must evaluate each record individually.
	 */
	sino = NULL;
	TAILQ_FOREACH(srec, &sblk->sb_recs, sr_next) {
		brec = (struct jblkrec *)srec->sr_rec;
		frags = brec->jb_frags;
		blk = brec->jb_blkno + brec->jb_oldfrags;
		isat = blk_isat(brec->jb_ino, brec->jb_lbn, blk, &frags);
		if (sino == NULL || sino->si_ino != brec->jb_ino) {
			sino = ino_lookup(brec->jb_ino, 1);
			sino->si_blkadj = 1;
		}
		if (debug)
			printf("op %d blk %jd ino %ju lbn %jd frags %d isat %d (%d)\n",
			    brec->jb_op, blk, (uintmax_t)brec->jb_ino,
			    brec->jb_lbn, brec->jb_frags, isat, frags);
		/*
		 * If we found the block at this address we still have to
		 * determine if we need to free the tail end that was
		 * added by adding contiguous fragments from the same block.
		 */
		if (isat == 1) {
			if (frags == brec->jb_frags)
				continue;
			mask = blk_freemask(blk, brec->jb_ino, brec->jb_lbn,
			    brec->jb_frags);
			mask >>= frags;
			blk += frags;
			frags = brec->jb_frags - frags;
			blk_free(blk, mask, frags);
			continue;
		}
		/*
	 	 * The block wasn't found, attempt to free it.  It won't be
		 * freed if it was actually reallocated.  If this was an
		 * allocation we don't want to follow indirects as they
		 * may not be written yet.  Any children of the indirect will
		 * have their own records.  If it's a free we need to
		 * recursively free children.
		 */
		blk_free_lbn(blk, brec->jb_ino, brec->jb_lbn, brec->jb_frags,
		    brec->jb_op == JOP_FREEBLK);
	}
}

/*
 * Walk the list of inode records for this cg and resolve moved and duplicate
 * inode references now that we have a complete picture.
 */
static void
cg_build(struct suj_cg *sc)
{
	struct suj_ino *sino;
	int i;

	for (i = 0; i < SUJ_HASHSIZE; i++)
		LIST_FOREACH(sino, &sc->sc_inohash[i], si_next)
			ino_build(sino);
}

/*
 * Handle inodes requiring truncation.  This must be done prior to
 * looking up any inodes in directories.
 */
static void
cg_trunc(struct suj_cg *sc)
{
	struct suj_ino *sino;
	int i;

	for (i = 0; i < SUJ_HASHSIZE; i++) {
		LIST_FOREACH(sino, &sc->sc_inohash[i], si_next) {
			if (sino->si_trunc) {
				ino_trunc(sino->si_ino,
				    sino->si_trunc->jt_size);
				sino->si_blkadj = 0;
				sino->si_trunc = NULL;
			}
			if (sino->si_blkadj)
				ino_adjblks(sino);
		}
	}
}

static void
cg_adj_blk(struct suj_cg *sc)
{
	struct suj_ino *sino;
	int i;

	for (i = 0; i < SUJ_HASHSIZE; i++) {
		LIST_FOREACH(sino, &sc->sc_inohash[i], si_next) {
			if (sino->si_blkadj)
				ino_adjblks(sino);
		}
	}
}

/*
 * Free any partially allocated blocks and then resolve inode block
 * counts.
 */
static void
cg_check_blk(struct suj_cg *sc)
{
	struct suj_blk *sblk;
	int i;


	for (i = 0; i < SUJ_HASHSIZE; i++)
		LIST_FOREACH(sblk, &sc->sc_blkhash[i], sb_next)
			blk_check(sblk);
}

/*
 * Walk the list of inode records for this cg, recovering any
 * changes which were not complete at the time of crash.
 */
static void
cg_check_ino(struct suj_cg *sc)
{
	struct suj_ino *sino;
	int i;

	for (i = 0; i < SUJ_HASHSIZE; i++)
		LIST_FOREACH(sino, &sc->sc_inohash[i], si_next)
			ino_check(sino);
}

/*
 * Write a potentially dirty cg.  Recalculate the summary information and
 * update the superblock summary.
 */
static void
cg_write(struct suj_cg *sc)
{
	ufs1_daddr_t fragno, cgbno, maxbno;
	u_int8_t *blksfree;
	struct cg *cgp;
	int blk;
	int i;

	if (sc->sc_dirty == 0)
		return;
	/*
	 * Fix the frag and cluster summary.
	 */
	cgp = sc->sc_cgp;
	cgp->cg_cs.cs_nbfree = 0;
	cgp->cg_cs.cs_nffree = 0;
	bzero(&cgp->cg_frsum, sizeof(cgp->cg_frsum));
	maxbno = fragstoblks(fs, fs->fs_fpg);
	if (fs->fs_contigsumsize > 0) {
		for (i = 1; i <= fs->fs_contigsumsize; i++)
			cg_clustersum(cgp)[i] = 0;
		bzero(cg_clustersfree(cgp), howmany(maxbno, CHAR_BIT));
	}
	blksfree = cg_blksfree(cgp);
	for (cgbno = 0; cgbno < maxbno; cgbno++) {
		if (ffs_isfreeblock(fs, blksfree, cgbno))
			continue;
		if (ffs_isblock(fs, blksfree, cgbno)) {
			ffs_clusteracct(fs, cgp, cgbno, 1);
			cgp->cg_cs.cs_nbfree++;
			continue;
		}
		fragno = blkstofrags(fs, cgbno);
		blk = blkmap(fs, blksfree, fragno);
		ffs_fragacct(fs, blk, cgp->cg_frsum, 1);
		for (i = 0; i < fs->fs_frag; i++)
			if (isset(blksfree, fragno + i))
				cgp->cg_cs.cs_nffree++;
	}
	/*
	 * Update the superblock cg summary from our now correct values
	 * before writing the block.
	 */
	fs->fs_cs(fs, sc->sc_cgx) = cgp->cg_cs;
	if (bwrite(disk, fsbtodb(fs, cgtod(fs, sc->sc_cgx)), sc->sc_cgbuf,
	    fs->fs_bsize) == -1)
		err_suj("Unable to write cylinder group %d\n", sc->sc_cgx);
}

/*
 * Write out any modified inodes.
 */
static void
cg_write_inos(struct suj_cg *sc)
{
	struct ino_blk *iblk;
	int i;

	for (i = 0; i < SUJ_HASHSIZE; i++)
		LIST_FOREACH(iblk, &sc->sc_iblkhash[i], ib_next)
			if (iblk->ib_dirty)
				iblk_write(iblk);
}

static void
cg_apply(void (*apply)(struct suj_cg *))
{
	struct suj_cg *scg;
	int i;

	for (i = 0; i < SUJ_HASHSIZE; i++)
		LIST_FOREACH(scg, &cghash[i], sc_next)
			apply(scg);
}

/*
 * Process the unlinked but referenced file list.  Freeing all inodes.
 */
static void
ino_unlinked(void)
{
	union dinode *ip;
	uint16_t mode;
	ino_t inon;
	ino_t ino;

	ino = fs->fs_sujfree;
	fs->fs_sujfree = 0;
	while (ino != 0) {
		ip = ino_read(ino);
		mode = DIP(ip, di_mode) & IFMT;
		inon = DIP(ip, di_freelink);
		DIP_SET(ip, di_freelink, 0);
		/*
		 * XXX Should this be an errx?
		 */
		if (DIP(ip, di_nlink) == 0) {
			if (debug)
				printf("Freeing unlinked ino %ju mode %o\n",
				    (uintmax_t)ino, mode);
			ino_reclaim(ip, ino, mode);
		} else if (debug)
			printf("Skipping ino %ju mode %o with link %d\n",
			    (uintmax_t)ino, mode, DIP(ip, di_nlink));
		ino = inon;
	}
}

/*
 * Append a new record to the list of records requiring processing.
 */
static void
ino_append(union jrec *rec)
{
	struct jrefrec *refrec;
	struct jmvrec *mvrec;
	struct suj_ino *sino;
	struct suj_rec *srec;

	mvrec = &rec->rec_jmvrec;
	refrec = &rec->rec_jrefrec;
	if (debug && mvrec->jm_op == JOP_MVREF)
		printf("ino move: ino %d, parent %d, diroff %jd, oldoff %jd\n",
		    mvrec->jm_ino, mvrec->jm_parent, mvrec->jm_newoff,
		    mvrec->jm_oldoff);
	else if (debug &&
	    (refrec->jr_op == JOP_ADDREF || refrec->jr_op == JOP_REMREF))
		printf("ino ref: op %d, ino %d, nlink %d, "
		    "parent %d, diroff %jd\n",
		    refrec->jr_op, refrec->jr_ino, refrec->jr_nlink,
		    refrec->jr_parent, refrec->jr_diroff);
	sino = ino_lookup(((struct jrefrec *)rec)->jr_ino, 1);
	sino->si_hasrecs = 1;
	srec = errmalloc(sizeof(*srec));
	srec->sr_rec = rec;
	TAILQ_INSERT_TAIL(&sino->si_newrecs, srec, sr_next);
}

/*
 * Add a reference adjustment to the sino list and eliminate dups.  The
 * primary loop in ino_build_ref() checks for dups but new ones may be
 * created as a result of offset adjustments.
 */
static void
ino_add_ref(struct suj_ino *sino, struct suj_rec *srec)
{
	struct jrefrec *refrec;
	struct suj_rec *srn;
	struct jrefrec *rrn;

	refrec = (struct jrefrec *)srec->sr_rec;
	/*
	 * We walk backwards so that the oldest link count is preserved.  If
	 * an add record conflicts with a remove keep the remove.  Redundant
	 * removes are eliminated in ino_build_ref.  Otherwise we keep the
	 * oldest record at a given location.
	 */
	for (srn = TAILQ_LAST(&sino->si_recs, srechd); srn;
	    srn = TAILQ_PREV(srn, srechd, sr_next)) {
		rrn = (struct jrefrec *)srn->sr_rec;
		if (rrn->jr_parent != refrec->jr_parent ||
		    rrn->jr_diroff != refrec->jr_diroff)
			continue;
		if (rrn->jr_op == JOP_REMREF || refrec->jr_op == JOP_ADDREF) {
			rrn->jr_mode = refrec->jr_mode;
			return;
		}
		/*
		 * Adding a remove.
		 *
		 * Replace the record in place with the old nlink in case
		 * we replace the head of the list.  Abandon srec as a dup.
		 */
		refrec->jr_nlink = rrn->jr_nlink;
		srn->sr_rec = srec->sr_rec;
		return;
	}
	TAILQ_INSERT_TAIL(&sino->si_recs, srec, sr_next);
}

/*
 * Create a duplicate of a reference at a previous location.
 */
static void
ino_dup_ref(struct suj_ino *sino, struct jrefrec *refrec, off_t diroff)
{
	struct jrefrec *rrn;
	struct suj_rec *srn;

	rrn = errmalloc(sizeof(*refrec));
	*rrn = *refrec;
	rrn->jr_op = JOP_ADDREF;
	rrn->jr_diroff = diroff;
	srn = errmalloc(sizeof(*srn));
	srn->sr_rec = (union jrec *)rrn;
	ino_add_ref(sino, srn);
}

/*
 * Add a reference to the list at all known locations.  We follow the offset
 * changes for a single instance and create duplicate add refs at each so
 * that we can tolerate any version of the directory block.  Eliminate
 * removes which collide with adds that are seen in the journal.  They should
 * not adjust the link count down.
 */
static void
ino_build_ref(struct suj_ino *sino, struct suj_rec *srec)
{
	struct jrefrec *refrec;
	struct jmvrec *mvrec;
	struct suj_rec *srp;
	struct suj_rec *srn;
	struct jrefrec *rrn;
	off_t diroff;

	refrec = (struct jrefrec *)srec->sr_rec;
	/*
	 * Search for a mvrec that matches this offset.  Whether it's an add
	 * or a remove we can delete the mvref after creating a dup record in
	 * the old location.
	 */
	if (!TAILQ_EMPTY(&sino->si_movs)) {
		diroff = refrec->jr_diroff;
		for (srn = TAILQ_LAST(&sino->si_movs, srechd); srn; srn = srp) {
			srp = TAILQ_PREV(srn, srechd, sr_next);
			mvrec = (struct jmvrec *)srn->sr_rec;
			if (mvrec->jm_parent != refrec->jr_parent ||
			    mvrec->jm_newoff != diroff)
				continue;
			diroff = mvrec->jm_oldoff;
			TAILQ_REMOVE(&sino->si_movs, srn, sr_next);
			free(srn);
			ino_dup_ref(sino, refrec, diroff);
		}
	}
	/*
	 * If a remove wasn't eliminated by an earlier add just append it to
	 * the list.
	 */
	if (refrec->jr_op == JOP_REMREF) {
		ino_add_ref(sino, srec);
		return;
	}
	/*
	 * Walk the list of records waiting to be added to the list.  We
	 * must check for moves that apply to our current offset and remove
	 * them from the list.  Remove any duplicates to eliminate removes
	 * with corresponding adds.
	 */
	TAILQ_FOREACH_SAFE(srn, &sino->si_newrecs, sr_next, srp) {
		switch (srn->sr_rec->rec_jrefrec.jr_op) {
		case JOP_ADDREF:
			/*
			 * This should actually be an error we should
			 * have a remove for every add journaled.
			 */
			rrn = (struct jrefrec *)srn->sr_rec;
			if (rrn->jr_parent != refrec->jr_parent ||
			    rrn->jr_diroff != refrec->jr_diroff)
				break;
			TAILQ_REMOVE(&sino->si_newrecs, srn, sr_next);
			break;
		case JOP_REMREF:
			/*
			 * Once we remove the current iteration of the
			 * record at this address we're done.
			 */
			rrn = (struct jrefrec *)srn->sr_rec;
			if (rrn->jr_parent != refrec->jr_parent ||
			    rrn->jr_diroff != refrec->jr_diroff)
				break;
			TAILQ_REMOVE(&sino->si_newrecs, srn, sr_next);
			ino_add_ref(sino, srec);
			return;
		case JOP_MVREF:
			/*
			 * Update our diroff based on any moves that match
			 * and remove the move.
			 */
			mvrec = (struct jmvrec *)srn->sr_rec;
			if (mvrec->jm_parent != refrec->jr_parent ||
			    mvrec->jm_oldoff != refrec->jr_diroff)
				break;
			ino_dup_ref(sino, refrec, mvrec->jm_oldoff);
			refrec->jr_diroff = mvrec->jm_newoff;
			TAILQ_REMOVE(&sino->si_newrecs, srn, sr_next);
			break;
		default:
			err_suj("ino_build_ref: Unknown op %d\n",
			    srn->sr_rec->rec_jrefrec.jr_op);
		}
	}
	ino_add_ref(sino, srec);
}

/*
 * Walk the list of new records and add them in-order resolving any
 * dups and adjusted offsets.
 */
static void
ino_build(struct suj_ino *sino)
{
	struct suj_rec *srec;

	while ((srec = TAILQ_FIRST(&sino->si_newrecs)) != NULL) {
		TAILQ_REMOVE(&sino->si_newrecs, srec, sr_next);
		switch (srec->sr_rec->rec_jrefrec.jr_op) {
		case JOP_ADDREF:
		case JOP_REMREF:
			ino_build_ref(sino, srec);
			break;
		case JOP_MVREF:
			/*
			 * Add this mvrec to the queue of pending mvs.
			 */
			TAILQ_INSERT_TAIL(&sino->si_movs, srec, sr_next);
			break;
		default:
			err_suj("ino_build: Unknown op %d\n",
			    srec->sr_rec->rec_jrefrec.jr_op);
		}
	}
	if (TAILQ_EMPTY(&sino->si_recs))
		sino->si_hasrecs = 0;
}

/*
 * Modify journal records so they refer to the base block number
 * and a start and end frag range.  This is to facilitate the discovery
 * of overlapping fragment allocations.
 */
static void
blk_build(struct jblkrec *blkrec)
{
	struct suj_rec *srec;
	struct suj_blk *sblk;
	struct jblkrec *blkrn;
	ufs2_daddr_t blk;
	int frag;

	if (debug)
		printf("blk_build: op %d blkno %jd frags %d oldfrags %d "
		    "ino %d lbn %jd\n",
		    blkrec->jb_op, blkrec->jb_blkno, blkrec->jb_frags,
		    blkrec->jb_oldfrags, blkrec->jb_ino, blkrec->jb_lbn);

	blk = blknum(fs, blkrec->jb_blkno);
	frag = fragnum(fs, blkrec->jb_blkno);
	sblk = blk_lookup(blk, 1);
	/*
	 * Rewrite the record using oldfrags to indicate the offset into
	 * the block.  Leave jb_frags as the actual allocated count.
	 */
	blkrec->jb_blkno -= frag;
	blkrec->jb_oldfrags = frag;
	if (blkrec->jb_oldfrags + blkrec->jb_frags > fs->fs_frag)
		err_suj("Invalid fragment count %d oldfrags %d\n",
		    blkrec->jb_frags, frag);
	/*
	 * Detect dups.  If we detect a dup we always discard the oldest
	 * record as it is superseded by the new record.  This speeds up
	 * later stages but also eliminates free records which are used
	 * to indicate that the contents of indirects can be trusted.
	 */
	TAILQ_FOREACH(srec, &sblk->sb_recs, sr_next) {
		blkrn = (struct jblkrec *)srec->sr_rec;
		if (blkrn->jb_ino != blkrec->jb_ino ||
		    blkrn->jb_lbn != blkrec->jb_lbn ||
		    blkrn->jb_blkno != blkrec->jb_blkno ||
		    blkrn->jb_frags != blkrec->jb_frags ||
		    blkrn->jb_oldfrags != blkrec->jb_oldfrags)
			continue;
		if (debug)
			printf("Removed dup.\n");
		/* Discard the free which is a dup with an alloc. */
		if (blkrec->jb_op == JOP_FREEBLK)
			return;
		TAILQ_REMOVE(&sblk->sb_recs, srec, sr_next);
		free(srec);
		break;
	}
	srec = errmalloc(sizeof(*srec));
	srec->sr_rec = (union jrec *)blkrec;
	TAILQ_INSERT_TAIL(&sblk->sb_recs, srec, sr_next);
}

static void
ino_build_trunc(struct jtrncrec *rec)
{
	struct suj_ino *sino;

	if (debug)
		printf("ino_build_trunc: op %d ino %d, size %jd\n",
		    rec->jt_op, rec->jt_ino, rec->jt_size);
	sino = ino_lookup(rec->jt_ino, 1);
	if (rec->jt_op == JOP_SYNC) {
		sino->si_trunc = NULL;
		return;
	}
	if (sino->si_trunc == NULL || sino->si_trunc->jt_size > rec->jt_size)
		sino->si_trunc = rec;
}

/*
 * Build up tables of the operations we need to recover.
 */
static void
suj_build(void)
{
	struct suj_seg *seg;
	union jrec *rec;
	int off;
	int i;

	TAILQ_FOREACH(seg, &allsegs, ss_next) {
		if (debug)
			printf("seg %jd has %d records, oldseq %jd.\n",
			    seg->ss_rec.jsr_seq, seg->ss_rec.jsr_cnt,
			    seg->ss_rec.jsr_oldest);
		off = 0;
		rec = (union jrec *)seg->ss_blk;
		for (i = 0; i < seg->ss_rec.jsr_cnt; off += JREC_SIZE, rec++) {
			/* skip the segrec. */
			if ((off % real_dev_bsize) == 0)
				continue;
			switch (rec->rec_jrefrec.jr_op) {
			case JOP_ADDREF:
			case JOP_REMREF:
			case JOP_MVREF:
				ino_append(rec);
				break;
			case JOP_NEWBLK:
			case JOP_FREEBLK:
				blk_build((struct jblkrec *)rec);
				break;
			case JOP_TRUNC:
			case JOP_SYNC:
				ino_build_trunc((struct jtrncrec *)rec);
				break;
			default:
				err_suj("Unknown journal operation %d (%d)\n",
				    rec->rec_jrefrec.jr_op, off);
			}
			i++;
		}
	}
}

/*
 * Prune the journal segments to those we care about based on the
 * oldest sequence in the newest segment.  Order the segment list
 * based on sequence number.
 */
static void
suj_prune(void)
{
	struct suj_seg *seg;
	struct suj_seg *segn;
	uint64_t newseq;
	int discard;

	if (debug)
		printf("Pruning up to %jd\n", oldseq);
	/* First free the expired segments. */
	TAILQ_FOREACH_SAFE(seg, &allsegs, ss_next, segn) {
		if (seg->ss_rec.jsr_seq >= oldseq)
			continue;
		TAILQ_REMOVE(&allsegs, seg, ss_next);
		free(seg->ss_blk);
		free(seg);
	}
	/* Next ensure that segments are ordered properly. */
	seg = TAILQ_FIRST(&allsegs);
	if (seg == NULL) {
		if (debug)
			printf("Empty journal\n");
		return;
	}
	newseq = seg->ss_rec.jsr_seq;
	for (;;) {
		seg = TAILQ_LAST(&allsegs, seghd);
		if (seg->ss_rec.jsr_seq >= newseq)
			break;
		TAILQ_REMOVE(&allsegs, seg, ss_next);
		TAILQ_INSERT_HEAD(&allsegs, seg, ss_next);
		newseq = seg->ss_rec.jsr_seq;

	}
	if (newseq != oldseq) {
		TAILQ_FOREACH(seg, &allsegs, ss_next) {
			printf("%jd, ", seg->ss_rec.jsr_seq);
		}
		printf("\n");
		err_suj("Journal file sequence mismatch %jd != %jd\n",
		    newseq, oldseq);
	}
	/*
	 * The kernel may asynchronously write segments which can create
	 * gaps in the sequence space.  Throw away any segments after the
	 * gap as the kernel guarantees only those that are contiguously
	 * reachable are marked as completed.
	 */
	discard = 0;
	TAILQ_FOREACH_SAFE(seg, &allsegs, ss_next, segn) {
		if (!discard && newseq++ == seg->ss_rec.jsr_seq) {
			jrecs += seg->ss_rec.jsr_cnt;
			jbytes += seg->ss_rec.jsr_blocks * real_dev_bsize;
			continue;
		}
		discard = 1;
		if (debug)
			printf("Journal order mismatch %jd != %jd pruning\n",
			    newseq-1, seg->ss_rec.jsr_seq);
		TAILQ_REMOVE(&allsegs, seg, ss_next);
		free(seg->ss_blk);
		free(seg);
	}
	if (debug)
		printf("Processing journal segments from %jd to %jd\n",
		    oldseq, newseq-1);
}

/*
 * Verify the journal inode before attempting to read records.
 */
static int
suj_verifyino(union dinode *ip)
{

	if (DIP(ip, di_nlink) != 1) {
		printf("Invalid link count %d for journal inode %ju\n",
		    DIP(ip, di_nlink), (uintmax_t)sujino);
		return (-1);
	}

	if ((DIP(ip, di_flags) & (SF_IMMUTABLE | SF_NOUNLINK)) !=
	    (SF_IMMUTABLE | SF_NOUNLINK)) {
		printf("Invalid flags 0x%X for journal inode %ju\n",
		    DIP(ip, di_flags), (uintmax_t)sujino);
		return (-1);
	}

	if (DIP(ip, di_mode) != (IFREG | IREAD)) {
		printf("Invalid mode %o for journal inode %ju\n",
		    DIP(ip, di_mode), (uintmax_t)sujino);
		return (-1);
	}

	if (DIP(ip, di_size) < SUJ_MIN) {
		printf("Invalid size %jd for journal inode %ju\n",
		    DIP(ip, di_size), (uintmax_t)sujino);
		return (-1);
	}

	if (DIP(ip, di_modrev) != fs->fs_mtime) {
		printf("Journal timestamp does not match fs mount time\n");
		return (-1);
	}

	return (0);
}

struct jblocks {
	struct jextent *jb_extent;	/* Extent array. */
	int		jb_avail;	/* Available extents. */
	int		jb_used;	/* Last used extent. */
	int		jb_head;	/* Allocator head. */
	int		jb_off;		/* Allocator extent offset. */
};
struct jextent {
	ufs2_daddr_t	je_daddr;	/* Disk block address. */
	int		je_blocks;	/* Disk block count. */
};

struct jblocks *suj_jblocks;

static struct jblocks *
jblocks_create(void)
{
	struct jblocks *jblocks;
	int size;

	jblocks = errmalloc(sizeof(*jblocks));
	jblocks->jb_avail = 10;
	jblocks->jb_used = 0;
	jblocks->jb_head = 0;
	jblocks->jb_off = 0;
	size = sizeof(struct jextent) * jblocks->jb_avail;
	jblocks->jb_extent = errmalloc(size);
	bzero(jblocks->jb_extent, size);

	return (jblocks);
}

/*
 * Return the next available disk block and the amount of contiguous
 * free space it contains.
 */
static ufs2_daddr_t
jblocks_next(struct jblocks *jblocks, int bytes, int *actual)
{
	struct jextent *jext;
	ufs2_daddr_t daddr;
	int freecnt;
	int blocks;

	blocks = bytes / disk->d_bsize;
	jext = &jblocks->jb_extent[jblocks->jb_head];
	freecnt = jext->je_blocks - jblocks->jb_off;
	if (freecnt == 0) {
		jblocks->jb_off = 0;
		if (++jblocks->jb_head > jblocks->jb_used)
			return (0);
		jext = &jblocks->jb_extent[jblocks->jb_head];
		freecnt = jext->je_blocks;
	}
	if (freecnt > blocks)
		freecnt = blocks;
	*actual = freecnt * disk->d_bsize;
	daddr = jext->je_daddr + jblocks->jb_off;

	return (daddr);
}

/*
 * Advance the allocation head by a specified number of bytes, consuming
 * one journal segment.
 */
static void
jblocks_advance(struct jblocks *jblocks, int bytes)
{

	jblocks->jb_off += bytes / disk->d_bsize;
}

static void
jblocks_destroy(struct jblocks *jblocks)
{

	free(jblocks->jb_extent);
	free(jblocks);
}

static void
jblocks_add(struct jblocks *jblocks, ufs2_daddr_t daddr, int blocks)
{
	struct jextent *jext;
	int size;

	jext = &jblocks->jb_extent[jblocks->jb_used];
	/* Adding the first block. */
	if (jext->je_daddr == 0) {
		jext->je_daddr = daddr;
		jext->je_blocks = blocks;
		return;
	}
	/* Extending the last extent. */
	if (jext->je_daddr + jext->je_blocks == daddr) {
		jext->je_blocks += blocks;
		return;
	}
	/* Adding a new extent. */
	if (++jblocks->jb_used == jblocks->jb_avail) {
		jblocks->jb_avail *= 2;
		size = sizeof(struct jextent) * jblocks->jb_avail;
		jext = errmalloc(size);
		bzero(jext, size);
		bcopy(jblocks->jb_extent, jext,
		    sizeof(struct jextent) * jblocks->jb_used);
		free(jblocks->jb_extent);
		jblocks->jb_extent = jext;
	}
	jext = &jblocks->jb_extent[jblocks->jb_used];
	jext->je_daddr = daddr;
	jext->je_blocks = blocks;

	return;
}

/*
 * Add a file block from the journal to the extent map.  We can't read
 * each file block individually because the kernel treats it as a circular
 * buffer and segments may span mutliple contiguous blocks.
 */
static void
suj_add_block(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, int frags)
{

	jblocks_add(suj_jblocks, fsbtodb(fs, blk), fsbtodb(fs, frags));
}

static void
suj_read(void)
{
	uint8_t block[1 * 1024 * 1024];
	struct suj_seg *seg;
	struct jsegrec *recn;
	struct jsegrec *rec;
	ufs2_daddr_t blk;
	int readsize;
	int blocks;
	int recsize;
	int size;
	int i;

	/*
	 * Read records until we exhaust the journal space.  If we find
	 * an invalid record we start searching for a valid segment header
	 * at the next block.  This is because we don't have a head/tail
	 * pointer and must recover the information indirectly.  At the gap
	 * between the head and tail we won't necessarily have a valid
	 * segment.
	 */
restart:
	for (;;) {
		size = sizeof(block);
		blk = jblocks_next(suj_jblocks, size, &readsize);
		if (blk == 0)
			return;
		size = readsize;
		/*
		 * Read 1MB at a time and scan for records within this block.
		 */
		if (bread(disk, blk, &block, size) == -1) {
			err_suj("Error reading journal block %jd\n",
			    (intmax_t)blk);
		}
		for (rec = (void *)block; size; size -= recsize,
		    rec = (struct jsegrec *)((uintptr_t)rec + recsize)) {
			recsize = real_dev_bsize;
			if (rec->jsr_time != fs->fs_mtime) {
				if (debug)
					printf("Rec time %jd != fs mtime %jd\n",
					    rec->jsr_time, fs->fs_mtime);
				jblocks_advance(suj_jblocks, recsize);
				continue;
			}
			if (rec->jsr_cnt == 0) {
				if (debug)
					printf("Found illegal count %d\n",
					    rec->jsr_cnt);
				jblocks_advance(suj_jblocks, recsize);
				continue;
			}
			blocks = rec->jsr_blocks;
			recsize = blocks * real_dev_bsize;
			if (recsize > size) {
				/*
				 * We may just have run out of buffer, restart
				 * the loop to re-read from this spot.
				 */
				if (size < fs->fs_bsize &&
				    size != readsize &&
				    recsize <= fs->fs_bsize)
					goto restart;
				if (debug)
					printf("Found invalid segsize %d > %d\n",
					    recsize, size);
				recsize = real_dev_bsize;
				jblocks_advance(suj_jblocks, recsize);
				continue;
			}
			/*
			 * Verify that all blocks in the segment are present.
			 */
			for (i = 1; i < blocks; i++) {
				recn = (void *)((uintptr_t)rec) + i *
				    real_dev_bsize;
				if (recn->jsr_seq == rec->jsr_seq &&
				    recn->jsr_time == rec->jsr_time)
					continue;
				if (debug)
					printf("Incomplete record %jd (%d)\n",
					    rec->jsr_seq, i);
				recsize = i * real_dev_bsize;
				jblocks_advance(suj_jblocks, recsize);
				goto restart;
			}
			seg = errmalloc(sizeof(*seg));
			seg->ss_blk = errmalloc(recsize);
			seg->ss_rec = *rec;
			bcopy((void *)rec, seg->ss_blk, recsize);
			if (rec->jsr_oldest > oldseq)
				oldseq = rec->jsr_oldest;
			TAILQ_INSERT_TAIL(&allsegs, seg, ss_next);
			jblocks_advance(suj_jblocks, recsize);
		}
	}
}

/*
 * Search a directory block for the SUJ_FILE.
 */
static void
suj_find(ino_t ino, ufs_lbn_t lbn, ufs2_daddr_t blk, int frags)
{
	char block[MAXBSIZE];
	struct direct *dp;
	int bytes;
	int off;

	if (sujino)
		return;
	bytes = lfragtosize(fs, frags);
	if (bread(disk, fsbtodb(fs, blk), block, bytes) <= 0)
		err_suj("Failed to read ROOTINO directory block %jd\n", blk);
	for (off = 0; off < bytes; off += dp->d_reclen) {
		dp = (struct direct *)&block[off];
		if (dp->d_reclen == 0)
			break;
		if (dp->d_ino == 0)
			continue;
		if (dp->d_namlen != strlen(SUJ_FILE))
			continue;
		if (bcmp(dp->d_name, SUJ_FILE, dp->d_namlen) != 0)
			continue;
		sujino = dp->d_ino;
		return;
	}
}

/*
 * Orchestrate the verification of a filesystem via the softupdates journal.
 */
int
suj_check(const char *filesys)
{
	union dinode *jip;
	union dinode *ip;
	uint64_t blocks;
	int retval;
	struct suj_seg *seg;
	struct suj_seg *segn;

	opendisk(filesys);
	TAILQ_INIT(&allsegs);

	/*
	 * Set an exit point when SUJ check failed
	 */
	retval = setjmp(jmpbuf);
	if (retval != 0) {
		pwarn("UNEXPECTED SU+J INCONSISTENCY\n");
		TAILQ_FOREACH_SAFE(seg, &allsegs, ss_next, segn) {
			TAILQ_REMOVE(&allsegs, seg, ss_next);
				free(seg->ss_blk);
				free(seg);
		}
		if (reply("FALLBACK TO FULL FSCK") == 0) {
			ckfini(0);
			exit(EEXIT);
		} else
			return (-1);
	}

	/*
	 * Find the journal inode.
	 */
	ip = ino_read(ROOTINO);
	sujino = 0;
	ino_visit(ip, ROOTINO, suj_find, 0);
	if (sujino == 0) {
		printf("Journal inode removed.  Use tunefs to re-create.\n");
		sblock.fs_flags &= ~FS_SUJ;
		sblock.fs_sujfree = 0;
		return (-1);
	}
	/*
	 * Fetch the journal inode and verify it.
	 */
	jip = ino_read(sujino);
	printf("** SU+J Recovering %s\n", filesys);
	if (suj_verifyino(jip) != 0)
		return (-1);
	/*
	 * Build a list of journal blocks in jblocks before parsing the
	 * available journal blocks in with suj_read().
	 */
	printf("** Reading %jd byte journal from inode %ju.\n",
	    DIP(jip, di_size), (uintmax_t)sujino);
	suj_jblocks = jblocks_create();
	blocks = ino_visit(jip, sujino, suj_add_block, 0);
	if (blocks != numfrags(fs, DIP(jip, di_size))) {
		printf("Sparse journal inode %ju.\n", (uintmax_t)sujino);
		return (-1);
	}
	suj_read();
	jblocks_destroy(suj_jblocks);
	suj_jblocks = NULL;
	if (preen || reply("RECOVER")) {
		printf("** Building recovery table.\n");
		suj_prune();
		suj_build();
		cg_apply(cg_build);
		printf("** Resolving unreferenced inode list.\n");
		ino_unlinked();
		printf("** Processing journal entries.\n");
		cg_apply(cg_trunc);
		cg_apply(cg_check_blk);
		cg_apply(cg_adj_blk);
		cg_apply(cg_check_ino);
	}
	if (preen == 0 && (jrecs > 0 || jbytes > 0) && reply("WRITE CHANGES") == 0)
		return (0);
	/*
	 * To remain idempotent with partial truncations the free bitmaps
	 * must be written followed by indirect blocks and lastly inode
	 * blocks.  This preserves access to the modified pointers until
	 * they are freed.
	 */
	cg_apply(cg_write);
	dblk_write();
	cg_apply(cg_write_inos);
	/* Write back superblock. */
	closedisk(filesys);
	if (jrecs > 0 || jbytes > 0) {
		printf("** %jd journal records in %jd bytes for %.2f%% utilization\n",
		    jrecs, jbytes, ((float)jrecs / (float)(jbytes / JREC_SIZE)) * 100);
		printf("** Freed %jd inodes (%jd dirs) %jd blocks, and %jd frags.\n",
		    freeinos, freedir, freeblocks, freefrags);
	}

	return (0);
}
OpenPOWER on IntegriCloud