summaryrefslogtreecommitdiff
path: root/libbcache/journal.c
blob: 3bb9e3c38763107094abd427b80c12f7ea9ddc90 (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
/*
 * bcache journalling code, for btree insertions
 *
 * Copyright 2012 Google, Inc.
 */

#include "bcache.h"
#include "alloc.h"
#include "bkey_methods.h"
#include "buckets.h"
#include "btree_gc.h"
#include "btree_update.h"
#include "btree_io.h"
#include "checksum.h"
#include "debug.h"
#include "error.h"
#include "extents.h"
#include "io.h"
#include "keylist.h"
#include "journal.h"
#include "super-io.h"
#include "vstructs.h"

#include <trace/events/bcache.h>

static void journal_write(struct closure *);
static void journal_reclaim_fast(struct journal *);
static void journal_pin_add_entry(struct journal *,
				  struct journal_entry_pin_list *,
				  struct journal_entry_pin *,
				  journal_pin_flush_fn);

static inline struct journal_buf *journal_cur_buf(struct journal *j)
{
	return j->buf + j->reservations.idx;
}

static inline struct journal_buf *journal_prev_buf(struct journal *j)
{
	return j->buf + !j->reservations.idx;
}

/* Sequence number of oldest dirty journal entry */

static inline u64 last_seq(struct journal *j)
{
	return atomic64_read(&j->seq) - fifo_used(&j->pin) + 1;
}

static inline u64 journal_pin_seq(struct journal *j,
				  struct journal_entry_pin_list *pin_list)
{
	return last_seq(j) + fifo_entry_idx(&j->pin, pin_list);
}

static inline struct jset_entry *__jset_entry_type_next(struct jset *jset,
					struct jset_entry *entry, unsigned type)
{
	while (entry < vstruct_last(jset)) {
		if (JOURNAL_ENTRY_TYPE(entry) == type)
			return entry;

		entry = vstruct_next(entry);
	}

	return NULL;
}

#define for_each_jset_entry_type(entry, jset, type)			\
	for (entry = (jset)->start;					\
	     (entry = __jset_entry_type_next(jset, entry, type));	\
	     entry = vstruct_next(entry))

#define for_each_jset_key(k, _n, entry, jset)				\
	for_each_jset_entry_type(entry, jset, JOURNAL_ENTRY_BTREE_KEYS)	\
		vstruct_for_each_safe(entry, k, _n)

static inline void bch_journal_add_entry(struct journal_buf *buf,
					 const void *data, size_t u64s,
					 unsigned type, enum btree_id id,
					 unsigned level)
{
	struct jset *jset = buf->data;

	bch_journal_add_entry_at(buf, data, u64s, type, id, level,
				 le32_to_cpu(jset->u64s));
	le32_add_cpu(&jset->u64s, jset_u64s(u64s));
}

static struct jset_entry *bch_journal_find_entry(struct jset *j, unsigned type,
						 enum btree_id id)
{
	struct jset_entry *entry;

	for_each_jset_entry_type(entry, j, type)
		if (entry->btree_id == id)
			return entry;

	return NULL;
}

struct bkey_i *bch_journal_find_btree_root(struct cache_set *c, struct jset *j,
					   enum btree_id id, unsigned *level)
{
	struct bkey_i *k;
	struct jset_entry *entry =
		bch_journal_find_entry(j, JOURNAL_ENTRY_BTREE_ROOT, id);

	if (!entry)
		return NULL;

	k = entry->start;
	*level = entry->level;
	*level = entry->level;
	return k;
}

static void bch_journal_add_btree_root(struct journal_buf *buf,
				       enum btree_id id, struct bkey_i *k,
				       unsigned level)
{
	bch_journal_add_entry(buf, k, k->k.u64s,
			      JOURNAL_ENTRY_BTREE_ROOT, id, level);
}

static inline void bch_journal_add_prios(struct journal *j,
					 struct journal_buf *buf)
{
	/*
	 * no prio bucket ptrs yet... XXX should change the allocator so this
	 * can't happen:
	 */
	if (!buf->nr_prio_buckets)
		return;

	bch_journal_add_entry(buf, j->prio_buckets, buf->nr_prio_buckets,
			      JOURNAL_ENTRY_PRIO_PTRS, 0, 0);
}

static void journal_seq_blacklist_flush(struct journal *j,
					struct journal_entry_pin *pin)
{
	struct cache_set *c =
		container_of(j, struct cache_set, journal);
	struct journal_seq_blacklist *bl =
		container_of(pin, struct journal_seq_blacklist, pin);
	struct blacklisted_node n;
	struct closure cl;
	unsigned i;
	int ret;

	closure_init_stack(&cl);

	for (i = 0;; i++) {
		struct btree_iter iter;
		struct btree *b;

		mutex_lock(&j->blacklist_lock);
		if (i >= bl->nr_entries) {
			mutex_unlock(&j->blacklist_lock);
			break;
		}
		n = bl->entries[i];
		mutex_unlock(&j->blacklist_lock);

		bch_btree_iter_init(&iter, c, n.btree_id, n.pos);
		iter.is_extents = false;
redo_peek:
		b = bch_btree_iter_peek_node(&iter);

		/* The node might have already been rewritten: */

		if (b->data->keys.seq == n.seq &&
		    !bkey_cmp(b->key.k.p, n.pos)) {
			ret = bch_btree_node_rewrite(&iter, b, &cl);
			if (ret) {
				bch_btree_iter_unlock(&iter);
				closure_sync(&cl);

				if (ret == -EAGAIN ||
				    ret == -EINTR)
					goto redo_peek;

				/* -EROFS or perhaps -ENOSPC - bail out: */
				/* XXX warn here */
				return;
			}
		}

		bch_btree_iter_unlock(&iter);
	}

	closure_sync(&cl);

	for (i = 0;; i++) {
		struct btree_interior_update *as;
		struct pending_btree_node_free *d;

		mutex_lock(&j->blacklist_lock);
		if (i >= bl->nr_entries) {
			mutex_unlock(&j->blacklist_lock);
			break;
		}
		n = bl->entries[i];
		mutex_unlock(&j->blacklist_lock);
redo_wait:
		mutex_lock(&c->btree_interior_update_lock);

		/*
		 * Is the node on the list of pending interior node updates -
		 * being freed? If so, wait for that to finish:
		 */
		for_each_pending_btree_node_free(c, as, d)
			if (n.seq	== d->seq &&
			    n.btree_id	== d->btree_id &&
			    !d->level &&
			    !bkey_cmp(n.pos, d->key.k.p)) {
				closure_wait(&as->wait, &cl);
				mutex_unlock(&c->btree_interior_update_lock);
				closure_sync(&cl);
				goto redo_wait;
			}

		mutex_unlock(&c->btree_interior_update_lock);
	}

	mutex_lock(&j->blacklist_lock);

	bch_journal_pin_drop(j, &bl->pin);
	list_del(&bl->list);
	kfree(bl->entries);
	kfree(bl);

	mutex_unlock(&j->blacklist_lock);
}

static struct journal_seq_blacklist *
journal_seq_blacklist_find(struct journal *j, u64 seq)
{
	struct journal_seq_blacklist *bl;

	lockdep_assert_held(&j->blacklist_lock);

	list_for_each_entry(bl, &j->seq_blacklist, list)
		if (seq == bl->seq)
			return bl;

	return NULL;
}

static struct journal_seq_blacklist *
bch_journal_seq_blacklisted_new(struct journal *j, u64 seq)
{
	struct journal_seq_blacklist *bl;

	lockdep_assert_held(&j->blacklist_lock);

	bl = kzalloc(sizeof(*bl), GFP_KERNEL);
	if (!bl)
		return NULL;

	bl->seq = seq;
	list_add_tail(&bl->list, &j->seq_blacklist);
	return bl;
}

/*
 * Returns true if @seq is newer than the most recent journal entry that got
 * written, and data corresponding to @seq should be ignored - also marks @seq
 * as blacklisted so that on future restarts the corresponding data will still
 * be ignored:
 */
int bch_journal_seq_should_ignore(struct cache_set *c, u64 seq, struct btree *b)
{
	struct journal *j = &c->journal;
	struct journal_seq_blacklist *bl = NULL;
	struct blacklisted_node *n;
	u64 journal_seq, i;
	int ret = 0;

	if (!seq)
		return 0;

	journal_seq = atomic64_read(&j->seq);

	/* Interier updates aren't journalled: */
	BUG_ON(b->level);
	BUG_ON(seq > journal_seq && test_bit(CACHE_SET_INITIAL_GC_DONE, &c->flags));

	if (seq <= journal_seq) {
		if (list_empty_careful(&j->seq_blacklist))
			return 0;

		mutex_lock(&j->blacklist_lock);
		ret = journal_seq_blacklist_find(j, seq) != NULL;
		mutex_unlock(&j->blacklist_lock);
		return ret;
	}

	/*
	 * Decrease this back to j->seq + 2 when we next rev the on disk format:
	 * increasing it temporarily to work around bug in old kernels
	 */
	cache_set_inconsistent_on(seq > journal_seq + 4, c,
			 "bset journal seq too far in the future: %llu > %llu",
			 seq, journal_seq);

	bch_verbose(c, "btree node %u:%llu:%llu has future journal sequence number %llu, blacklisting",
		    b->btree_id, b->key.k.p.inode, b->key.k.p.offset, seq);

	/*
	 * When we start the journal, bch_journal_start() will skip over @seq:
	 */

	mutex_lock(&j->blacklist_lock);

	for (i = journal_seq + 1; i <= seq; i++) {
		bl = journal_seq_blacklist_find(j, i) ?:
			bch_journal_seq_blacklisted_new(j, i);

		if (!bl) {
			ret = -ENOMEM;
			goto out;
		}
	}

	for (n = bl->entries; n < bl->entries + bl->nr_entries; n++)
		if (b->data->keys.seq	== n->seq &&
		    b->btree_id		== n->btree_id &&
		    !bkey_cmp(b->key.k.p, n->pos))
			goto found_entry;

	if (!bl->nr_entries ||
	    is_power_of_2(bl->nr_entries)) {
		n = krealloc(bl->entries,
			     max(bl->nr_entries * 2, 8UL) * sizeof(*n),
			     GFP_KERNEL);
		if (!n) {
			ret = -ENOMEM;
			goto out;
		}
		bl->entries = n;
	}

	bl->entries[bl->nr_entries++] = (struct blacklisted_node) {
		.seq		= b->data->keys.seq,
		.btree_id	= b->btree_id,
		.pos		= b->key.k.p,
	};
found_entry:
	ret = 1;
out:
	mutex_unlock(&j->blacklist_lock);
	return ret;
}

/*
 * Journal replay/recovery:
 *
 * This code is all driven from run_cache_set(); we first read the journal
 * entries, do some other stuff, then we mark all the keys in the journal
 * entries (same as garbage collection would), then we replay them - reinserting
 * them into the cache in precisely the same order as they appear in the
 * journal.
 *
 * We only journal keys that go in leaf nodes, which simplifies things quite a
 * bit.
 */

struct journal_list {
	struct closure		cl;
	struct mutex		lock;
	struct list_head	*head;
	int			ret;
};

#define JOURNAL_ENTRY_ADD_OK		0
#define JOURNAL_ENTRY_ADD_OUT_OF_RANGE	5

/*
 * Given a journal entry we just read, add it to the list of journal entries to
 * be replayed:
 */
static int journal_entry_add(struct cache_set *c, struct journal_list *jlist,
		    struct jset *j)
{
	struct journal_replay *i, *pos;
	struct list_head *where;
	size_t bytes = vstruct_bytes(j);
	__le64 last_seq;
	int ret;

	mutex_lock(&jlist->lock);

	last_seq = !list_empty(jlist->head)
		? list_last_entry(jlist->head, struct journal_replay,
				  list)->j.last_seq
		: 0;

	/* Is this entry older than the range we need? */
	if (le64_to_cpu(j->seq) < le64_to_cpu(last_seq)) {
		ret = JOURNAL_ENTRY_ADD_OUT_OF_RANGE;
		goto out;
	}

	/* Drop entries we don't need anymore */
	list_for_each_entry_safe(i, pos, jlist->head, list) {
		if (le64_to_cpu(i->j.seq) >= le64_to_cpu(j->last_seq))
			break;
		list_del(&i->list);
		kfree(i);
	}

	list_for_each_entry_reverse(i, jlist->head, list) {
		/* Duplicate? */
		if (le64_to_cpu(j->seq) == le64_to_cpu(i->j.seq)) {
			fsck_err_on(bytes != vstruct_bytes(&i->j) ||
				    memcmp(j, &i->j, bytes), c,
				    "found duplicate but non identical journal entries (seq %llu)",
				    le64_to_cpu(j->seq));

			ret = JOURNAL_ENTRY_ADD_OK;
			goto out;
		}

		if (le64_to_cpu(j->seq) > le64_to_cpu(i->j.seq)) {
			where = &i->list;
			goto add;
		}
	}

	where = jlist->head;
add:
	i = kvmalloc(offsetof(struct journal_replay, j) + bytes, GFP_KERNEL);
	if (!i) {
		ret = -ENOMEM;
		goto out;
	}

	memcpy(&i->j, j, bytes);
	list_add(&i->list, where);
	ret = JOURNAL_ENTRY_ADD_OK;
out:
fsck_err:
	mutex_unlock(&jlist->lock);
	return ret;
}

static struct nonce journal_nonce(const struct jset *jset)
{
	return (struct nonce) {{
		[0] = 0,
		[1] = ((__le32 *) &jset->seq)[0],
		[2] = ((__le32 *) &jset->seq)[1],
		[3] = BCH_NONCE_JOURNAL,
	}};
}

static void journal_entry_null_range(void *start, void *end)
{
	struct jset_entry *entry;

	for (entry = start; entry != end; entry = vstruct_next(entry)) {
		entry->u64s	= 0;
		entry->btree_id	= 0;
		entry->level	= 0;
		entry->flags	= 0;
		SET_JOURNAL_ENTRY_TYPE(entry, 0);
	}
}

static int journal_validate_key(struct cache_set *c, struct jset *j,
				struct jset_entry *entry,
				struct bkey_i *k, enum bkey_type key_type,
				const char *type)
{
	void *next = vstruct_next(entry);
	const char *invalid;
	char buf[160];
	int ret = 0;

	if (mustfix_fsck_err_on(!k->k.u64s, c,
			"invalid %s in journal: k->u64s 0", type)) {
		entry->u64s = cpu_to_le16((u64 *) k - entry->_data);
		journal_entry_null_range(vstruct_next(entry), next);
		return 0;
	}

	if (mustfix_fsck_err_on((void *) bkey_next(k) >
				(void *) vstruct_next(entry), c,
			"invalid %s in journal: extends past end of journal entry",
			type)) {
		entry->u64s = cpu_to_le16((u64 *) k - entry->_data);
		journal_entry_null_range(vstruct_next(entry), next);
		return 0;
	}

	if (mustfix_fsck_err_on(k->k.format != KEY_FORMAT_CURRENT, c,
			"invalid %s in journal: bad format %u",
			type, k->k.format)) {
		le16_add_cpu(&entry->u64s, -k->k.u64s);
		memmove(k, bkey_next(k), next - (void *) bkey_next(k));
		journal_entry_null_range(vstruct_next(entry), next);
		return 0;
	}

	if (JSET_BIG_ENDIAN(j) != CPU_BIG_ENDIAN)
		bch_bkey_swab(key_type, NULL, bkey_to_packed(k));

	invalid = bkey_invalid(c, key_type, bkey_i_to_s_c(k));
	if (invalid) {
		bch_bkey_val_to_text(c, key_type, buf, sizeof(buf),
				     bkey_i_to_s_c(k));
		mustfix_fsck_err(c, "invalid %s in journal: %s", type, buf);

		le16_add_cpu(&entry->u64s, -k->k.u64s);
		memmove(k, bkey_next(k), next - (void *) bkey_next(k));
		journal_entry_null_range(vstruct_next(entry), next);
		return 0;
	}
fsck_err:
	return ret;
}

#define JOURNAL_ENTRY_REREAD	5
#define JOURNAL_ENTRY_NONE	6
#define JOURNAL_ENTRY_BAD	7

static int journal_entry_validate(struct cache_set *c,
				  struct jset *j, u64 sector,
				  unsigned bucket_sectors_left,
				  unsigned sectors_read)
{
	struct jset_entry *entry;
	size_t bytes = vstruct_bytes(j);
	struct bch_csum csum;
	int ret = 0;

	if (le64_to_cpu(j->magic) != jset_magic(c))
		return JOURNAL_ENTRY_NONE;

	if (le32_to_cpu(j->version) != BCACHE_JSET_VERSION) {
		bch_err(c, "unknown journal entry version %u",
			le32_to_cpu(j->version));
		return BCH_FSCK_UNKNOWN_VERSION;
	}

	if (mustfix_fsck_err_on(bytes > bucket_sectors_left << 9 ||
			bytes > c->journal.entry_size_max, c,
			"journal entry too big (%zu bytes), sector %lluu",
			bytes, sector)) {
		/* XXX: note we might have missing journal entries */
		return JOURNAL_ENTRY_BAD;
	}

	if (bytes > sectors_read << 9)
		return JOURNAL_ENTRY_REREAD;

	if (fsck_err_on(!bch_checksum_type_valid(c, JSET_CSUM_TYPE(j)), c,
			"journal entry with unknown csum type %llu sector %lluu",
			JSET_CSUM_TYPE(j), sector))
		return JOURNAL_ENTRY_BAD;

	csum = csum_vstruct(c, JSET_CSUM_TYPE(j), journal_nonce(j), j);
	if (mustfix_fsck_err_on(bch_crc_cmp(csum, j->csum), c,
			"journal checksum bad, sector %llu", sector)) {
		/* XXX: retry IO, when we start retrying checksum errors */
		/* XXX: note we might have missing journal entries */
		return JOURNAL_ENTRY_BAD;
	}

	bch_encrypt(c, JSET_CSUM_TYPE(j), journal_nonce(j),
		    j->encrypted_start,
		    vstruct_end(j) - (void *) j->encrypted_start);

	if (mustfix_fsck_err_on(le64_to_cpu(j->last_seq) > le64_to_cpu(j->seq), c,
			"invalid journal entry: last_seq > seq"))
		j->last_seq = j->seq;

	vstruct_for_each(j, entry) {
		struct bkey_i *k;

		if (mustfix_fsck_err_on(vstruct_next(entry) >
					vstruct_last(j), c,
				"journal entry extents past end of jset")) {
			j->u64s = cpu_to_le64((u64 *) entry - j->_data);
			break;
		}

		switch (JOURNAL_ENTRY_TYPE(entry)) {
		case JOURNAL_ENTRY_BTREE_KEYS:
			vstruct_for_each(entry, k) {
				ret = journal_validate_key(c, j, entry, k,
						bkey_type(entry->level,
							  entry->btree_id),
						"key");
				if (ret)
					goto fsck_err;
			}
			break;

		case JOURNAL_ENTRY_BTREE_ROOT:
			k = entry->start;

			if (mustfix_fsck_err_on(!entry->u64s ||
					le16_to_cpu(entry->u64s) != k->k.u64s, c,
					"invalid btree root journal entry: wrong number of keys")) {
				journal_entry_null_range(entry,
						vstruct_next(entry));
				continue;
			}

			ret = journal_validate_key(c, j, entry, k,
						   BKEY_TYPE_BTREE, "btree root");
			if (ret)
				goto fsck_err;
			break;

		case JOURNAL_ENTRY_PRIO_PTRS:
			break;

		case JOURNAL_ENTRY_JOURNAL_SEQ_BLACKLISTED:
			if (mustfix_fsck_err_on(le16_to_cpu(entry->u64s) != 1, c,
				"invalid journal seq blacklist entry: bad size")) {
				journal_entry_null_range(entry,
						vstruct_next(entry));
			}

			break;
		default:
			mustfix_fsck_err(c, "invalid journal entry type %llu",
				 JOURNAL_ENTRY_TYPE(entry));
			journal_entry_null_range(entry, vstruct_next(entry));
			break;
		}
	}

fsck_err:
	return ret;
}

struct journal_read_buf {
	void		*data;
	size_t		size;
};

static int journal_read_buf_realloc(struct journal_read_buf *b,
				    size_t new_size)
{
	void *n;

	new_size = roundup_pow_of_two(new_size);
	n = (void *) __get_free_pages(GFP_KERNEL, get_order(new_size));
	if (!n)
		return -ENOMEM;

	free_pages((unsigned long) b->data, get_order(b->size));
	b->data = n;
	b->size = new_size;
	return 0;
}

static int journal_read_bucket(struct cache *ca,
			       struct journal_read_buf *buf,
			       struct journal_list *jlist,
			       unsigned bucket, u64 *seq, bool *entries_found)
{
	struct cache_set *c = ca->set;
	struct journal_device *ja = &ca->journal;
	struct bio *bio = ja->bio;
	struct jset *j = NULL;
	unsigned sectors, sectors_read = 0;
	u64 offset = bucket_to_sector(ca, ja->buckets[bucket]),
	    end = offset + ca->mi.bucket_size;
	bool saw_bad = false;
	int ret = 0;

	pr_debug("reading %u", bucket);

	while (offset < end) {
		if (!sectors_read) {
reread:			sectors_read = min_t(unsigned,
				end - offset, buf->size >> 9);

			bio_reset(bio);
			bio->bi_bdev		= ca->disk_sb.bdev;
			bio->bi_iter.bi_sector	= offset;
			bio->bi_iter.bi_size	= sectors_read << 9;
			bio_set_op_attrs(bio, REQ_OP_READ, 0);
			bch_bio_map(bio, buf->data);

			ret = submit_bio_wait(bio);

			if (cache_fatal_io_err_on(ret, ca,
						  "journal read from sector %llu",
						  offset) ||
			    bch_meta_read_fault("journal"))
				return -EIO;

			j = buf->data;
		}

		ret = journal_entry_validate(c, j, offset,
					end - offset, sectors_read);
		switch (ret) {
		case BCH_FSCK_OK:
			break;
		case JOURNAL_ENTRY_REREAD:
			if (vstruct_bytes(j) > buf->size) {
				ret = journal_read_buf_realloc(buf,
							vstruct_bytes(j));
				if (ret)
					return ret;
			}
			goto reread;
		case JOURNAL_ENTRY_NONE:
			if (!saw_bad)
				return 0;
			sectors = c->sb.block_size;
			goto next_block;
		case JOURNAL_ENTRY_BAD:
			saw_bad = true;
			sectors = c->sb.block_size;
			goto next_block;
		default:
			return ret;
		}

		/*
		 * This happens sometimes if we don't have discards on -
		 * when we've partially overwritten a bucket with new
		 * journal entries. We don't need the rest of the
		 * bucket:
		 */
		if (le64_to_cpu(j->seq) < ja->bucket_seq[bucket])
			return 0;

		ja->bucket_seq[bucket] = le64_to_cpu(j->seq);

		ret = journal_entry_add(c, jlist, j);
		switch (ret) {
		case JOURNAL_ENTRY_ADD_OK:
			*entries_found = true;
			break;
		case JOURNAL_ENTRY_ADD_OUT_OF_RANGE:
			break;
		default:
			return ret;
		}

		if (le64_to_cpu(j->seq) > *seq)
			*seq = le64_to_cpu(j->seq);

		sectors = vstruct_sectors(j, c->block_bits);
next_block:
		pr_debug("next");
		offset		+= sectors;
		sectors_read	-= sectors;
		j = ((void *) j) + (sectors << 9);
	}

	return 0;
}

static void bch_journal_read_device(struct closure *cl)
{
#define read_bucket(b)							\
	({								\
		bool entries_found = false;				\
		ret = journal_read_bucket(ca, &buf, jlist, b, &seq,	\
					  &entries_found);		\
		if (ret)						\
			goto err;					\
		__set_bit(b, bitmap);					\
		entries_found;						\
	 })

	struct journal_device *ja =
		container_of(cl, struct journal_device, read);
	struct cache *ca = container_of(ja, struct cache, journal);
	struct journal_list *jlist =
		container_of(cl->parent, struct journal_list, cl);
	struct request_queue *q = bdev_get_queue(ca->disk_sb.bdev);
	struct journal_read_buf buf = { NULL, 0 };

	DECLARE_BITMAP(bitmap, ja->nr);
	unsigned i, l, r;
	u64 seq = 0;
	int ret;

	if (!ja->nr)
		goto out;

	bitmap_zero(bitmap, ja->nr);
	ret = journal_read_buf_realloc(&buf, PAGE_SIZE);
	if (ret)
		goto err;

	pr_debug("%u journal buckets", ja->nr);

	/*
	 * If the device supports discard but not secure discard, we can't do
	 * the fancy fibonacci hash/binary search because the live journal
	 * entries might not form a contiguous range:
	 */
	for (i = 0; i < ja->nr; i++)
		read_bucket(i);
	goto search_done;

	if (!blk_queue_nonrot(q))
		goto linear_scan;

	/*
	 * Read journal buckets ordered by golden ratio hash to quickly
	 * find a sequence of buckets with valid journal entries
	 */
	for (i = 0; i < ja->nr; i++) {
		l = (i * 2654435769U) % ja->nr;

		if (test_bit(l, bitmap))
			break;

		if (read_bucket(l))
			goto bsearch;
	}

	/*
	 * If that fails, check all the buckets we haven't checked
	 * already
	 */
	pr_debug("falling back to linear search");
linear_scan:
	for (l = find_first_zero_bit(bitmap, ja->nr);
	     l < ja->nr;
	     l = find_next_zero_bit(bitmap, ja->nr, l + 1))
		if (read_bucket(l))
			goto bsearch;

	/* no journal entries on this device? */
	if (l == ja->nr)
		goto out;
bsearch:
	/* Binary search */
	r = find_next_bit(bitmap, ja->nr, l + 1);
	pr_debug("starting binary search, l %u r %u", l, r);

	while (l + 1 < r) {
		unsigned m = (l + r) >> 1;
		u64 cur_seq = seq;

		read_bucket(m);

		if (cur_seq != seq)
			l = m;
		else
			r = m;
	}

search_done:
	/*
	 * Find the journal bucket with the highest sequence number:
	 *
	 * If there's duplicate journal entries in multiple buckets (which
	 * definitely isn't supposed to happen, but...) - make sure to start
	 * cur_idx at the last of those buckets, so we don't deadlock trying to
	 * allocate
	 */
	seq = 0;

	for (i = 0; i < ja->nr; i++)
		if (ja->bucket_seq[i] >= seq &&
		    ja->bucket_seq[i] != ja->bucket_seq[(i + 1) % ja->nr]) {
			/*
			 * When journal_next_bucket() goes to allocate for
			 * the first time, it'll use the bucket after
			 * ja->cur_idx
			 */
			ja->cur_idx = i;
			seq = ja->bucket_seq[i];
		}

	/*
	 * Set last_idx to indicate the entire journal is full and needs to be
	 * reclaimed - journal reclaim will immediately reclaim whatever isn't
	 * pinned when it first runs:
	 */
	ja->last_idx = (ja->cur_idx + 1) % ja->nr;

	/*
	 * Read buckets in reverse order until we stop finding more journal
	 * entries:
	 */
	for (i = (ja->cur_idx + ja->nr - 1) % ja->nr;
	     i != ja->cur_idx;
	     i = (i + ja->nr - 1) % ja->nr)
		if (!test_bit(i, bitmap) &&
		    !read_bucket(i))
			break;
out:
	free_pages((unsigned long) buf.data, get_order(buf.size));
	closure_return(cl);
err:
	mutex_lock(&jlist->lock);
	jlist->ret = ret;
	mutex_unlock(&jlist->lock);
	goto out;
#undef read_bucket
}

void bch_journal_entries_free(struct list_head *list)
{

	while (!list_empty(list)) {
		struct journal_replay *i =
			list_first_entry(list, struct journal_replay, list);
		list_del(&i->list);
		kvfree(i);
	}
}

static int journal_seq_blacklist_read(struct journal *j,
				      struct journal_replay *i,
				      struct journal_entry_pin_list *p)
{
	struct cache_set *c = container_of(j, struct cache_set, journal);
	struct jset_entry *entry;
	struct journal_seq_blacklist *bl;
	u64 seq;

	for_each_jset_entry_type(entry, &i->j,
			JOURNAL_ENTRY_JOURNAL_SEQ_BLACKLISTED) {
		seq = le64_to_cpu(entry->_data[0]);

		bch_verbose(c, "blacklisting existing journal seq %llu", seq);

		bl = bch_journal_seq_blacklisted_new(j, seq);
		if (!bl)
			return -ENOMEM;

		journal_pin_add_entry(j, p, &bl->pin,
				  journal_seq_blacklist_flush);
		bl->written = true;
	}

	return 0;
}

static inline bool journal_has_keys(struct list_head *list)
{
	struct journal_replay *i;
	struct jset_entry *entry;
	struct bkey_i *k, *_n;

	list_for_each_entry(i, list, list)
		for_each_jset_key(k, _n, entry, &i->j)
			return true;

	return false;
}

int bch_journal_read(struct cache_set *c, struct list_head *list)
{
	struct jset_entry *prio_ptrs;
	struct journal_list jlist;
	struct journal_replay *i;
	struct jset *j;
	struct journal_entry_pin_list *p;
	struct cache *ca;
	u64 cur_seq, end_seq;
	unsigned iter;
	int ret = 0;

	closure_init_stack(&jlist.cl);
	mutex_init(&jlist.lock);
	jlist.head = list;
	jlist.ret = 0;

	for_each_cache(ca, c, iter)
		closure_call(&ca->journal.read,
			     bch_journal_read_device,
			     system_unbound_wq,
			     &jlist.cl);

	closure_sync(&jlist.cl);

	if (jlist.ret)
		return jlist.ret;

	if (list_empty(list)){
		bch_err(c, "no journal entries found");
		return BCH_FSCK_REPAIR_IMPOSSIBLE;
	}

	fsck_err_on(c->sb.clean && journal_has_keys(list), c,
		    "filesystem marked clean but journal has keys to replay");

	j = &list_entry(list->prev, struct journal_replay, list)->j;

	unfixable_fsck_err_on(le64_to_cpu(j->seq) -
			le64_to_cpu(j->last_seq) + 1 >
			c->journal.pin.size, c,
			"too many journal entries open for refcount fifo");

	c->journal.pin.back = le64_to_cpu(j->seq) -
		le64_to_cpu(j->last_seq) + 1;

	atomic64_set(&c->journal.seq, le64_to_cpu(j->seq));
	c->journal.last_seq_ondisk = le64_to_cpu(j->last_seq);

	BUG_ON(last_seq(&c->journal) != le64_to_cpu(j->last_seq));

	i = list_first_entry(list, struct journal_replay, list);

	mutex_lock(&c->journal.blacklist_lock);

	fifo_for_each_entry_ptr(p, &c->journal.pin, iter) {
		u64 seq = journal_pin_seq(&c->journal, p);

		INIT_LIST_HEAD(&p->list);

		if (i && le64_to_cpu(i->j.seq) == seq) {
			atomic_set(&p->count, 1);

			if (journal_seq_blacklist_read(&c->journal, i, p)) {
				mutex_unlock(&c->journal.blacklist_lock);
				return -ENOMEM;
			}

			i = list_is_last(&i->list, list)
				? NULL
				: list_next_entry(i, list);
		} else {
			atomic_set(&p->count, 0);
		}
	}

	mutex_unlock(&c->journal.blacklist_lock);

	cur_seq = last_seq(&c->journal);
	end_seq = le64_to_cpu(list_last_entry(list,
				struct journal_replay, list)->j.seq);

	list_for_each_entry(i, list, list) {
		bool blacklisted;

		mutex_lock(&c->journal.blacklist_lock);
		while (cur_seq < le64_to_cpu(i->j.seq) &&
		       journal_seq_blacklist_find(&c->journal, cur_seq))
			cur_seq++;

		blacklisted = journal_seq_blacklist_find(&c->journal,
							 le64_to_cpu(i->j.seq));
		mutex_unlock(&c->journal.blacklist_lock);

		fsck_err_on(blacklisted, c,
			    "found blacklisted journal entry %llu",
			    le64_to_cpu(i->j.seq));

		fsck_err_on(le64_to_cpu(i->j.seq) != cur_seq, c,
			"journal entries %llu-%llu missing! (replaying %llu-%llu)",
			cur_seq, le64_to_cpu(i->j.seq) - 1,
			last_seq(&c->journal), end_seq);

		cur_seq = le64_to_cpu(i->j.seq) + 1;
	}

	prio_ptrs = bch_journal_find_entry(j, JOURNAL_ENTRY_PRIO_PTRS, 0);
	if (prio_ptrs) {
		memcpy_u64s(c->journal.prio_buckets,
			    prio_ptrs->_data,
			    le16_to_cpu(prio_ptrs->u64s));
		c->journal.nr_prio_buckets = le16_to_cpu(prio_ptrs->u64s);
	}
fsck_err:
	return ret;
}

void bch_journal_mark(struct cache_set *c, struct list_head *list)
{
	struct bkey_i *k, *n;
	struct jset_entry *j;
	struct journal_replay *r;

	list_for_each_entry(r, list, list)
		for_each_jset_key(k, n, j, &r->j) {
			enum bkey_type type = bkey_type(j->level, j->btree_id);
			struct bkey_s_c k_s_c = bkey_i_to_s_c(k);

			if (btree_type_has_ptrs(type))
				bch_btree_mark_key_initial(c, type, k_s_c);
		}
}

static bool journal_entry_is_open(struct journal *j)
{
	return j->reservations.cur_entry_offset < JOURNAL_ENTRY_CLOSED_VAL;
}

void bch_journal_buf_put_slowpath(struct journal *j, bool need_write_just_set)
{
	struct cache_set *c = container_of(j, struct cache_set, journal);

	if (!need_write_just_set &&
	    test_bit(JOURNAL_NEED_WRITE, &j->flags))
		__bch_time_stats_update(j->delay_time,
					j->need_write_time);
#if 0
	closure_call(&j->io, journal_write, NULL, &c->cl);
#else
	/* Shut sparse up: */
	closure_init(&j->io, &c->cl);
	set_closure_fn(&j->io, journal_write, NULL);
	journal_write(&j->io);
#endif
}

static void __bch_journal_next_entry(struct journal *j)
{
	struct journal_entry_pin_list pin_list, *p;
	struct journal_buf *buf;

	/*
	 * The fifo_push() needs to happen at the same time as j->seq is
	 * incremented for last_seq() to be calculated correctly
	 */
	atomic64_inc(&j->seq);
	BUG_ON(!fifo_push(&j->pin, pin_list));
	p = &fifo_peek_back(&j->pin);

	INIT_LIST_HEAD(&p->list);
	atomic_set(&p->count, 1);

	if (test_bit(JOURNAL_REPLAY_DONE, &j->flags)) {
		smp_wmb();
		j->cur_pin_list = p;
	}

	buf = journal_cur_buf(j);
	memset(buf->has_inode, 0, sizeof(buf->has_inode));

	memset(buf->data, 0, sizeof(*buf->data));
	buf->data->seq	= cpu_to_le64(atomic64_read(&j->seq));
	buf->data->u64s	= 0;

	BUG_ON(journal_pin_seq(j, p) != atomic64_read(&j->seq));
}

static inline size_t journal_entry_u64s_reserve(struct journal_buf *buf)
{
	unsigned ret = BTREE_ID_NR * (JSET_KEYS_U64s + BKEY_EXTENT_U64s_MAX);

	if (buf->nr_prio_buckets)
		ret += JSET_KEYS_U64s + buf->nr_prio_buckets;

	return ret;
}

static enum {
	JOURNAL_ENTRY_ERROR,
	JOURNAL_ENTRY_INUSE,
	JOURNAL_ENTRY_CLOSED,
	JOURNAL_UNLOCKED,
} journal_buf_switch(struct journal *j, bool need_write_just_set)
{
	struct cache_set *c = container_of(j, struct cache_set, journal);
	struct journal_buf *buf;
	union journal_res_state old, new;
	u64 v = atomic64_read(&j->reservations.counter);

	do {
		old.v = new.v = v;
		if (old.cur_entry_offset == JOURNAL_ENTRY_CLOSED_VAL)
			return JOURNAL_ENTRY_CLOSED;

		if (old.cur_entry_offset == JOURNAL_ENTRY_ERROR_VAL)
			return JOURNAL_ENTRY_ERROR;

		if (new.prev_buf_unwritten)
			return JOURNAL_ENTRY_INUSE;

		/*
		 * avoid race between setting buf->data->u64s and
		 * journal_res_put starting write:
		 */
		journal_state_inc(&new);

		new.cur_entry_offset = JOURNAL_ENTRY_CLOSED_VAL;
		new.idx++;
		new.prev_buf_unwritten = 1;

		BUG_ON(journal_state_count(new, new.idx));
	} while ((v = atomic64_cmpxchg(&j->reservations.counter,
				       old.v, new.v)) != old.v);

	journal_reclaim_fast(j);

	clear_bit(JOURNAL_NEED_WRITE, &j->flags);

	buf = &j->buf[old.idx];
	buf->data->u64s		= cpu_to_le32(old.cur_entry_offset);
	buf->data->last_seq	= cpu_to_le64(last_seq(j));

	j->prev_buf_sectors =
		vstruct_blocks_plus(buf->data, c->block_bits,
				    journal_entry_u64s_reserve(buf)) *
		c->sb.block_size;

	BUG_ON(j->prev_buf_sectors > j->cur_buf_sectors);

	atomic_dec_bug(&fifo_peek_back(&j->pin).count);
	__bch_journal_next_entry(j);

	cancel_delayed_work(&j->write_work);
	spin_unlock(&j->lock);

	if (c->bucket_journal_seq > 1 << 14) {
		c->bucket_journal_seq = 0;
		bch_bucket_seq_cleanup(c);
	}

	/* ugh - might be called from __journal_res_get() under wait_event() */
	__set_current_state(TASK_RUNNING);
	bch_journal_buf_put(j, old.idx, need_write_just_set);

	return JOURNAL_UNLOCKED;
}

void bch_journal_halt(struct journal *j)
{
	union journal_res_state old, new;
	u64 v = atomic64_read(&j->reservations.counter);

	do {
		old.v = new.v = v;
		if (old.cur_entry_offset == JOURNAL_ENTRY_ERROR_VAL)
			return;

		new.cur_entry_offset = JOURNAL_ENTRY_ERROR_VAL;
	} while ((v = atomic64_cmpxchg(&j->reservations.counter,
				       old.v, new.v)) != old.v);

	wake_up(&j->wait);
	closure_wake_up(&journal_cur_buf(j)->wait);
	closure_wake_up(&journal_prev_buf(j)->wait);
}

static unsigned journal_dev_buckets_available(struct journal *j,
					      struct cache *ca)
{
	struct journal_device *ja = &ca->journal;
	unsigned next = (ja->cur_idx + 1) % ja->nr;
	unsigned available = (ja->last_idx + ja->nr - next) % ja->nr;

	/*
	 * Hack to avoid a deadlock during journal replay:
	 * journal replay might require setting a new btree
	 * root, which requires writing another journal entry -
	 * thus, if the journal is full (and this happens when
	 * replaying the first journal bucket's entries) we're
	 * screwed.
	 *
	 * So don't let the journal fill up unless we're in
	 * replay:
	 */
	if (test_bit(JOURNAL_REPLAY_DONE, &j->flags))
		available = max((int) available - 2, 0);

	/*
	 * Don't use the last bucket unless writing the new last_seq
	 * will make another bucket available:
	 */
	if (ja->bucket_seq[ja->last_idx] >= last_seq(j))
		available = max((int) available - 1, 0);

	return available;
}

/* returns number of sectors available for next journal entry: */
static int journal_entry_sectors(struct journal *j)
{
	struct cache_set *c = container_of(j, struct cache_set, journal);
	struct cache *ca;
	struct bkey_s_extent e = bkey_i_to_s_extent(&j->key);
	unsigned sectors_available = j->entry_size_max >> 9;
	unsigned i, nr_online = 0, nr_devs = 0;

	lockdep_assert_held(&j->lock);

	rcu_read_lock();
	group_for_each_cache_rcu(ca, &j->devs, i) {
		unsigned buckets_required = 0;

		sectors_available = min_t(unsigned, sectors_available,
					  ca->mi.bucket_size);

		/*
		 * Note that we don't allocate the space for a journal entry
		 * until we write it out - thus, if we haven't started the write
		 * for the previous entry we have to make sure we have space for
		 * it too:
		 */
		if (bch_extent_has_device(e.c, ca->dev_idx)) {
			if (j->prev_buf_sectors > ca->journal.sectors_free)
				buckets_required++;

			if (j->prev_buf_sectors + sectors_available >
			    ca->journal.sectors_free)
				buckets_required++;
		} else {
			if (j->prev_buf_sectors + sectors_available >
			    ca->mi.bucket_size)
				buckets_required++;

			buckets_required++;
		}

		if (journal_dev_buckets_available(j, ca) >= buckets_required)
			nr_devs++;
		nr_online++;
	}
	rcu_read_unlock();

	if (nr_online < c->opts.metadata_replicas)
		return -EROFS;

	if (nr_devs < c->opts.metadata_replicas)
		return 0;

	return sectors_available;
}

/*
 * should _only_ called from journal_res_get() - when we actually want a
 * journal reservation - journal entry is open means journal is dirty:
 */
static int journal_entry_open(struct journal *j)
{
	struct journal_buf *buf = journal_cur_buf(j);
	ssize_t u64s;
	int ret = 0, sectors;

	lockdep_assert_held(&j->lock);
	BUG_ON(journal_entry_is_open(j));

	if (!fifo_free(&j->pin))
		return 0;

	sectors = journal_entry_sectors(j);
	if (sectors <= 0)
		return sectors;

	j->cur_buf_sectors	= sectors;
	buf->nr_prio_buckets	= j->nr_prio_buckets;

	u64s = (sectors << 9) / sizeof(u64);

	/* Subtract the journal header */
	u64s -= sizeof(struct jset) / sizeof(u64);
	/*
	 * Btree roots, prio pointers don't get added until right before we do
	 * the write:
	 */
	u64s -= journal_entry_u64s_reserve(buf);
	u64s  = max_t(ssize_t, 0L, u64s);

	BUG_ON(u64s >= JOURNAL_ENTRY_CLOSED_VAL);

	if (u64s > le32_to_cpu(buf->data->u64s)) {
		union journal_res_state old, new;
		u64 v = atomic64_read(&j->reservations.counter);

		/*
		 * Must be set before marking the journal entry as open:
		 */
		j->cur_entry_u64s = u64s;

		do {
			old.v = new.v = v;

			if (old.cur_entry_offset == JOURNAL_ENTRY_ERROR_VAL)
				return false;

			/* Handle any already added entries */
			new.cur_entry_offset = le32_to_cpu(buf->data->u64s);
		} while ((v = atomic64_cmpxchg(&j->reservations.counter,
					       old.v, new.v)) != old.v);
		ret = 1;

		wake_up(&j->wait);

		if (j->res_get_blocked_start) {
			__bch_time_stats_update(j->blocked_time,
						j->res_get_blocked_start);
			j->res_get_blocked_start = 0;
		}

		mod_delayed_work(system_freezable_wq,
				 &j->write_work,
				 msecs_to_jiffies(j->write_delay_ms));
	}

	return ret;
}

void bch_journal_start(struct cache_set *c)
{
	struct journal *j = &c->journal;
	struct journal_seq_blacklist *bl;
	struct cache *ca;
	u64 new_seq = 0;
	unsigned i;

	for_each_cache(ca, c, i)
		if (is_journal_device(ca))
			bch_cache_group_add_cache(&c->journal.devs, ca);

	list_for_each_entry(bl, &j->seq_blacklist, list)
		new_seq = max(new_seq, bl->seq);

	spin_lock(&j->lock);

	set_bit(JOURNAL_STARTED, &j->flags);

	while (atomic64_read(&j->seq) < new_seq) {
		struct journal_entry_pin_list pin_list, *p;

		BUG_ON(!fifo_push(&j->pin, pin_list));
		p = &fifo_peek_back(&j->pin);

		INIT_LIST_HEAD(&p->list);
		atomic_set(&p->count, 0);
		atomic64_inc(&j->seq);
	}

	/*
	 * journal_buf_switch() only inits the next journal entry when it
	 * closes an open journal entry - the very first journal entry gets
	 * initialized here:
	 */
	__bch_journal_next_entry(j);

	/*
	 * Adding entries to the next journal entry before allocating space on
	 * disk for the next journal entry - this is ok, because these entries
	 * only have to go down with the next journal entry we write:
	 */
	list_for_each_entry(bl, &j->seq_blacklist, list)
		if (!bl->written) {
			bch_journal_add_entry(journal_cur_buf(j), &bl->seq, 1,
					JOURNAL_ENTRY_JOURNAL_SEQ_BLACKLISTED,
					0, 0);

			journal_pin_add_entry(j,
					      &fifo_peek_back(&j->pin),
					      &bl->pin,
					      journal_seq_blacklist_flush);
			bl->written = true;
		}

	spin_unlock(&j->lock);

	queue_delayed_work(system_freezable_wq, &j->reclaim_work, 0);
}

int bch_journal_replay(struct cache_set *c, struct list_head *list)
{
	int ret = 0, keys = 0, entries = 0;
	struct journal *j = &c->journal;
	struct bkey_i *k, *_n;
	struct jset_entry *entry;
	struct journal_replay *i, *n;

	list_for_each_entry_safe(i, n, list, list) {
		j->cur_pin_list =
			&j->pin.data[((j->pin.back - 1 -
				       (atomic64_read(&j->seq) -
					le64_to_cpu(i->j.seq))) &
				      j->pin.mask)];

		for_each_jset_key(k, _n, entry, &i->j) {
			struct disk_reservation disk_res;

			/*
			 * We might cause compressed extents to be split, so we
			 * need to pass in a disk_reservation:
			 */
			BUG_ON(bch_disk_reservation_get(c, &disk_res, 0, 0));

			trace_bcache_journal_replay_key(&k->k);

			ret = bch_btree_insert(c, entry->btree_id, k,
					       &disk_res, NULL, NULL,
					       BTREE_INSERT_NOFAIL|
					       BTREE_INSERT_JOURNAL_REPLAY);
			bch_disk_reservation_put(c, &disk_res);

			if (ret)
				goto err;

			cond_resched();
			keys++;
		}

		if (atomic_dec_and_test(&j->cur_pin_list->count))
			wake_up(&j->wait);

		entries++;
	}

	if (keys) {
		bch_btree_flush(c);

		/*
		 * Write a new journal entry _before_ we start journalling new data -
		 * otherwise, we could end up with btree node bsets with journal seqs
		 * arbitrarily far in the future vs. the most recently written journal
		 * entry on disk, if we crash before writing the next journal entry:
		 */
		ret = bch_journal_meta(&c->journal);
		if (ret)
			goto err;
	}

	bch_info(c, "journal replay done, %i keys in %i entries, seq %llu",
		 keys, entries, (u64) atomic64_read(&j->seq));

	bch_journal_set_replay_done(&c->journal);
err:
	if (ret)
		bch_err(c, "journal replay error: %d", ret);

	bch_journal_entries_free(list);

	return ret;
}

static int bch_set_nr_journal_buckets(struct cache *ca, unsigned nr)
{
	struct journal_device *ja = &ca->journal;
	struct bch_sb_field_journal *journal_buckets =
		bch_sb_get_journal(ca->disk_sb.sb);
	struct bch_sb_field *f;
	u64 *p;

	p = krealloc(ja->bucket_seq, nr * sizeof(u64),
		     GFP_KERNEL|__GFP_ZERO);
	if (!p)
		return -ENOMEM;

	ja->bucket_seq = p;

	p = krealloc(ja->buckets, nr * sizeof(u64),
		     GFP_KERNEL|__GFP_ZERO);
	if (!p)
		return -ENOMEM;

	ja->buckets = p;

	f = bch_dev_sb_field_resize(&ca->disk_sb, &journal_buckets->field, nr +
				    sizeof(*journal_buckets) / sizeof(u64));
	if (!f)
		return -ENOMEM;
	f->type = BCH_SB_FIELD_journal;

	ja->nr = nr;
	return 0;
}

int bch_cache_journal_alloc(struct cache *ca)
{
	struct journal_device *ja = &ca->journal;
	struct bch_sb_field_journal *journal_buckets;
	int ret;
	unsigned i;

	if (ca->mi.tier != 0)
		return 0;

	if (dynamic_fault("bcache:add:journal_alloc"))
		return -ENOMEM;

	/*
	 * clamp journal size to 1024 buckets or 512MB (in sectors), whichever
	 * is smaller:
	 */
	ret = bch_set_nr_journal_buckets(ca,
			clamp_t(unsigned, ca->mi.nbuckets >> 8,
				BCH_JOURNAL_BUCKETS_MIN,
				min(1 << 10,
				    (1 << 20) / ca->mi.bucket_size)));
	if (ret)
		return ret;

	journal_buckets = bch_sb_get_journal(ca->disk_sb.sb);

	for (i = 0; i < ja->nr; i++) {
		u64 bucket = ca->mi.first_bucket + i;

		ja->buckets[i] = bucket;
		journal_buckets->buckets[i] = cpu_to_le64(bucket);

		bch_mark_metadata_bucket(ca, &ca->buckets[bucket], true);
	}

	return 0;
}

/* Journalling */

/**
 * journal_reclaim_fast - do the fast part of journal reclaim
 *
 * Called from IO submission context, does not block. Cleans up after btree
 * write completions by advancing the journal pin and each cache's last_idx,
 * kicking off discards and background reclaim as necessary.
 */
static void journal_reclaim_fast(struct journal *j)
{
	struct journal_entry_pin_list temp;
	bool popped = false;

	lockdep_assert_held(&j->lock);

	/*
	 * Unpin journal entries whose reference counts reached zero, meaning
	 * all btree nodes got written out
	 */
	while (!atomic_read(&fifo_peek_front(&j->pin).count)) {
		BUG_ON(!list_empty(&fifo_peek_front(&j->pin).list));
		BUG_ON(!fifo_pop(&j->pin, temp));
		popped = true;
	}

	if (popped)
		wake_up(&j->wait);
}

/*
 * Journal entry pinning - machinery for holding a reference on a given journal
 * entry, marking it as dirty:
 */

static inline void __journal_pin_add(struct journal *j,
				     struct journal_entry_pin_list *pin_list,
				     struct journal_entry_pin *pin,
				     journal_pin_flush_fn flush_fn)
{
	BUG_ON(journal_pin_active(pin));

	atomic_inc(&pin_list->count);
	pin->pin_list	= pin_list;
	pin->flush	= flush_fn;

	if (flush_fn)
		list_add(&pin->list, &pin_list->list);
	else
		INIT_LIST_HEAD(&pin->list);
}

static void journal_pin_add_entry(struct journal *j,
				  struct journal_entry_pin_list *pin_list,
				  struct journal_entry_pin *pin,
				  journal_pin_flush_fn flush_fn)
{
	spin_lock_irq(&j->pin_lock);
	__journal_pin_add(j, pin_list, pin, flush_fn);
	spin_unlock_irq(&j->pin_lock);
}

void bch_journal_pin_add(struct journal *j,
			 struct journal_entry_pin *pin,
			 journal_pin_flush_fn flush_fn)
{
	spin_lock_irq(&j->pin_lock);
	__journal_pin_add(j, j->cur_pin_list, pin, flush_fn);
	spin_unlock_irq(&j->pin_lock);
}

static inline bool __journal_pin_drop(struct journal *j,
				      struct journal_entry_pin *pin)
{
	struct journal_entry_pin_list *pin_list = pin->pin_list;

	pin->pin_list = NULL;

	/* journal_reclaim_work() might have already taken us off the list */
	if (!list_empty_careful(&pin->list))
		list_del_init(&pin->list);

	return atomic_dec_and_test(&pin_list->count);
}

void bch_journal_pin_drop(struct journal *j,
			  struct journal_entry_pin *pin)
{
	unsigned long flags;
	bool wakeup;

	if (!journal_pin_active(pin))
		return;

	spin_lock_irqsave(&j->pin_lock, flags);
	wakeup = __journal_pin_drop(j, pin);
	spin_unlock_irqrestore(&j->pin_lock, flags);

	/*
	 * Unpinning a journal entry make make journal_next_bucket() succeed, if
	 * writing a new last_seq will now make another bucket available:
	 *
	 * Nested irqsave is expensive, don't do the wakeup with lock held:
	 */
	if (wakeup)
		wake_up(&j->wait);
}

void bch_journal_pin_add_if_older(struct journal *j,
				  struct journal_entry_pin *src_pin,
				  struct journal_entry_pin *pin,
				  journal_pin_flush_fn flush_fn)
{
	spin_lock_irq(&j->pin_lock);

	if (journal_pin_active(src_pin) &&
	    (!journal_pin_active(pin) ||
	     fifo_entry_idx(&j->pin, src_pin->pin_list) <
	     fifo_entry_idx(&j->pin, pin->pin_list))) {
		if (journal_pin_active(pin))
			__journal_pin_drop(j, pin);
		__journal_pin_add(j, src_pin->pin_list,
				  pin, NULL);
	}

	spin_unlock_irq(&j->pin_lock);
}


static struct journal_entry_pin *
journal_get_next_pin(struct journal *j, u64 seq_to_flush)
{
	struct journal_entry_pin_list *pin_list;
	struct journal_entry_pin *ret = NULL;
	unsigned iter;

	/* so we don't iterate over empty fifo entries below: */
	if (!atomic_read(&fifo_peek_front(&j->pin).count)) {
		spin_lock(&j->lock);
		journal_reclaim_fast(j);
		spin_unlock(&j->lock);
	}

	spin_lock_irq(&j->pin_lock);
	fifo_for_each_entry_ptr(pin_list, &j->pin, iter) {
		if (journal_pin_seq(j, pin_list) > seq_to_flush)
			break;

		ret = list_first_entry_or_null(&pin_list->list,
				struct journal_entry_pin, list);
		if (ret) {
			/* must be list_del_init(), see bch_journal_pin_drop() */
			list_del_init(&ret->list);
			break;
		}
	}
	spin_unlock_irq(&j->pin_lock);

	return ret;
}

static bool should_discard_bucket(struct journal *j, struct journal_device *ja)
{
	bool ret;

	spin_lock(&j->lock);
	ret = (ja->last_idx != ja->cur_idx &&
	       ja->bucket_seq[ja->last_idx] < j->last_seq_ondisk);
	spin_unlock(&j->lock);

	return ret;
}

/**
 * journal_reclaim_work - free up journal buckets
 *
 * Background journal reclaim writes out btree nodes. It should be run
 * early enough so that we never completely run out of journal buckets.
 *
 * High watermarks for triggering background reclaim:
 * - FIFO has fewer than 512 entries left
 * - fewer than 25% journal buckets free
 *
 * Background reclaim runs until low watermarks are reached:
 * - FIFO has more than 1024 entries left
 * - more than 50% journal buckets free
 *
 * As long as a reclaim can complete in the time it takes to fill up
 * 512 journal entries or 25% of all journal buckets, then
 * journal_next_bucket() should not stall.
 */
static void journal_reclaim_work(struct work_struct *work)
{
	struct cache_set *c = container_of(to_delayed_work(work),
				struct cache_set, journal.reclaim_work);
	struct journal *j = &c->journal;
	struct cache *ca;
	struct journal_entry_pin *pin;
	u64 seq_to_flush = 0;
	unsigned iter, bucket_to_flush;
	unsigned long next_flush;
	bool reclaim_lock_held = false, need_flush;

	/*
	 * Advance last_idx to point to the oldest journal entry containing
	 * btree node updates that have not yet been written out
	 */
	group_for_each_cache(ca, &j->devs, iter) {
		struct journal_device *ja = &ca->journal;

		while (should_discard_bucket(j, ja)) {
			if (!reclaim_lock_held) {
				/*
				 * ugh:
				 * might be called from __journal_res_get()
				 * under wait_event() - have to go back to
				 * TASK_RUNNING before doing something that
				 * would block, but only if we're doing work:
				 */
				__set_current_state(TASK_RUNNING);

				mutex_lock(&j->reclaim_lock);
				reclaim_lock_held = true;
				/* recheck under reclaim_lock: */
				continue;
			}

			if (ca->mi.discard &&
			    blk_queue_discard(bdev_get_queue(ca->disk_sb.bdev)))
				blkdev_issue_discard(ca->disk_sb.bdev,
					bucket_to_sector(ca,
						ja->buckets[ja->last_idx]),
					ca->mi.bucket_size, GFP_NOIO, 0);

			spin_lock(&j->lock);
			ja->last_idx = (ja->last_idx + 1) % ja->nr;
			spin_unlock(&j->lock);

			wake_up(&j->wait);
		}

		/*
		 * Write out enough btree nodes to free up 50% journal
		 * buckets
		 */
		spin_lock(&j->lock);
		bucket_to_flush = (ja->cur_idx + (ja->nr >> 1)) % ja->nr;
		seq_to_flush = max_t(u64, seq_to_flush,
				     ja->bucket_seq[bucket_to_flush]);
		spin_unlock(&j->lock);
	}

	if (reclaim_lock_held)
		mutex_unlock(&j->reclaim_lock);

	/* Also flush if the pin fifo is more than half full */
	seq_to_flush = max_t(s64, seq_to_flush,
			     (s64) atomic64_read(&j->seq) -
			     (j->pin.size >> 1));

	/*
	 * If it's been longer than j->reclaim_delay_ms since we last flushed,
	 * make sure to flush at least one journal pin:
	 */
	next_flush = j->last_flushed + msecs_to_jiffies(j->reclaim_delay_ms);
	need_flush = time_after(jiffies, next_flush);

	while ((pin = journal_get_next_pin(j, need_flush
					   ? U64_MAX
					   : seq_to_flush))) {
		__set_current_state(TASK_RUNNING);
		pin->flush(j, pin);
		need_flush = false;

		j->last_flushed = jiffies;
	}

	if (!test_bit(CACHE_SET_RO, &c->flags))
		queue_delayed_work(system_freezable_wq, &j->reclaim_work,
				   msecs_to_jiffies(j->reclaim_delay_ms));
}

/**
 * journal_next_bucket - move on to the next journal bucket if possible
 */
static int journal_write_alloc(struct journal *j, unsigned sectors)
{
	struct cache_set *c = container_of(j, struct cache_set, journal);
	struct bkey_s_extent e = bkey_i_to_s_extent(&j->key);
	struct bch_extent_ptr *ptr;
	struct cache *ca;
	unsigned iter, replicas, replicas_want =
		READ_ONCE(c->opts.metadata_replicas);

	spin_lock(&j->lock);
	rcu_read_lock();

	/*
	 * Drop any pointers to devices that have been removed, are no longer
	 * empty, or filled up their current journal bucket:
	 *
	 * Note that a device may have had a small amount of free space (perhaps
	 * one sector) that wasn't enough for the smallest possible journal
	 * entry - that's why we drop pointers to devices <= current free space,
	 * i.e. whichever device was limiting the current journal entry size.
	 */
	extent_for_each_ptr_backwards(e, ptr)
		if (!(ca = PTR_CACHE(c, ptr)) ||
		    ca->mi.state != BCH_MEMBER_STATE_ACTIVE ||
		    ca->journal.sectors_free <= sectors)
			__bch_extent_drop_ptr(e, ptr);
		else
			ca->journal.sectors_free -= sectors;

	replicas = bch_extent_nr_ptrs(e.c);

	/*
	 * Determine location of the next journal write:
	 * XXX: sort caches by free journal space
	 */
	group_for_each_cache_rcu(ca, &j->devs, iter) {
		struct journal_device *ja = &ca->journal;

		if (replicas >= replicas_want)
			break;

		/*
		 * Check that we can use this device, and aren't already using
		 * it:
		 */
		if (bch_extent_has_device(e.c, ca->dev_idx) ||
		    !journal_dev_buckets_available(j, ca) ||
		    sectors > ca->mi.bucket_size)
			continue;

		ja->sectors_free = ca->mi.bucket_size - sectors;
		ja->cur_idx = (ja->cur_idx + 1) % ja->nr;
		ja->bucket_seq[ja->cur_idx] = atomic64_read(&j->seq);

		extent_ptr_append(bkey_i_to_extent(&j->key),
			(struct bch_extent_ptr) {
				  .offset = bucket_to_sector(ca,
					ja->buckets[ja->cur_idx]),
				  .dev = ca->dev_idx,
		});
		replicas++;

		trace_bcache_journal_next_bucket(ca, ja->cur_idx, ja->last_idx);
	}

	rcu_read_unlock();

	j->prev_buf_sectors = 0;
	spin_unlock(&j->lock);

	if (replicas < replicas_want)
		return -EROFS;

	return 0;
}

static void journal_write_compact(struct jset *jset)
{
	struct jset_entry *i, *next, *prev = NULL;

	/*
	 * Simple compaction, dropping empty jset_entries (from journal
	 * reservations that weren't fully used) and merging jset_entries that
	 * can be.
	 *
	 * If we wanted to be really fancy here, we could sort all the keys in
	 * the jset and drop keys that were overwritten - probably not worth it:
	 */
	vstruct_for_each_safe(jset, i, next) {
		unsigned u64s = le16_to_cpu(i->u64s);

		/* Empty entry: */
		if (!u64s)
			continue;

		/* Can we merge with previous entry? */
		if (prev &&
		    i->btree_id == prev->btree_id &&
		    i->level	== prev->level &&
		    JOURNAL_ENTRY_TYPE(i) == JOURNAL_ENTRY_TYPE(prev) &&
		    JOURNAL_ENTRY_TYPE(i) == JOURNAL_ENTRY_BTREE_KEYS &&
		    le16_to_cpu(prev->u64s) + u64s <= U16_MAX) {
			memmove_u64s_down(vstruct_next(prev),
					  i->_data,
					  u64s);
			le16_add_cpu(&prev->u64s, u64s);
			continue;
		}

		/* Couldn't merge, move i into new position (after prev): */
		prev = prev ? vstruct_next(prev) : jset->start;
		if (i != prev)
			memmove_u64s_down(prev, i, jset_u64s(u64s));
	}

	prev = prev ? vstruct_next(prev) : jset->start;
	jset->u64s = cpu_to_le32((u64 *) prev - jset->_data);
}

static void journal_write_endio(struct bio *bio)
{
	struct cache *ca = bio->bi_private;
	struct journal *j = &ca->set->journal;

	if (cache_fatal_io_err_on(bio->bi_error, ca, "journal write") ||
	    bch_meta_write_fault("journal"))
		bch_journal_halt(j);

	closure_put(&j->io);
	percpu_ref_put(&ca->ref);
}

static void journal_write_done(struct closure *cl)
{
	struct journal *j = container_of(cl, struct journal, io);
	struct journal_buf *w = journal_prev_buf(j);

	j->last_seq_ondisk = le64_to_cpu(w->data->last_seq);

	__bch_time_stats_update(j->write_time, j->write_start_time);

	BUG_ON(!j->reservations.prev_buf_unwritten);
	atomic64_sub(((union journal_res_state) { .prev_buf_unwritten = 1 }).v,
		     &j->reservations.counter);

	/*
	 * XXX: this is racy, we could technically end up doing the wake up
	 * after the journal_buf struct has been reused for the next write
	 * (because we're clearing JOURNAL_IO_IN_FLIGHT) and wake up things that
	 * are waiting on the _next_ write, not this one.
	 *
	 * The wake up can't come before, because journal_flush_seq_async() is
	 * looking at JOURNAL_IO_IN_FLIGHT when it has to wait on a journal
	 * write that was already in flight.
	 *
	 * The right fix is to use a lock here, but using j.lock here means it
	 * has to be a spin_lock_irqsave() lock which then requires propagating
	 * the irq()ness to other locks and it's all kinds of nastiness.
	 */

	closure_wake_up(&w->wait);
	wake_up(&j->wait);

	/*
	 * Updating last_seq_ondisk may let journal_reclaim_work() discard more
	 * buckets:
	 */
	mod_delayed_work(system_freezable_wq, &j->reclaim_work, 0);
}

static void journal_write(struct closure *cl)
{
	struct journal *j = container_of(cl, struct journal, io);
	struct cache_set *c = container_of(j, struct cache_set, journal);
	struct cache *ca;
	struct journal_buf *w = journal_prev_buf(j);
	struct jset *jset = w->data;
	struct bio *bio;
	struct bch_extent_ptr *ptr;
	unsigned i, sectors, bytes;

	j->write_start_time = local_clock();

	bch_journal_add_prios(j, w);

	mutex_lock(&c->btree_root_lock);
	for (i = 0; i < BTREE_ID_NR; i++) {
		struct btree_root *r = &c->btree_roots[i];

		if (r->alive)
			bch_journal_add_btree_root(w, i, &r->key, r->level);
	}
	mutex_unlock(&c->btree_root_lock);

	journal_write_compact(jset);

	jset->read_clock	= cpu_to_le16(c->prio_clock[READ].hand);
	jset->write_clock	= cpu_to_le16(c->prio_clock[WRITE].hand);
	jset->magic		= cpu_to_le64(jset_magic(c));
	jset->version		= cpu_to_le32(BCACHE_JSET_VERSION);

	SET_JSET_BIG_ENDIAN(jset, CPU_BIG_ENDIAN);
	SET_JSET_CSUM_TYPE(jset, bch_meta_checksum_type(c));

	bch_encrypt(c, JSET_CSUM_TYPE(jset), journal_nonce(jset),
		    jset->encrypted_start,
		    vstruct_end(jset) - (void *) jset->encrypted_start);

	jset->csum = csum_vstruct(c, JSET_CSUM_TYPE(jset),
				  journal_nonce(jset), jset);

	sectors = vstruct_sectors(jset, c->block_bits);
	BUG_ON(sectors > j->prev_buf_sectors);

	bytes = vstruct_bytes(w->data);
	memset((void *) w->data + bytes, 0, (sectors << 9) - bytes);

	if (journal_write_alloc(j, sectors)) {
		bch_journal_halt(j);
		bch_err(c, "Unable to allocate journal write");
		bch_fatal_error(c);
		closure_return_with_destructor(cl, journal_write_done);
	}

	bch_check_mark_super(c, &j->key, true);

	/*
	 * XXX: we really should just disable the entire journal in nochanges
	 * mode
	 */
	if (c->opts.nochanges)
		goto no_io;

	extent_for_each_ptr(bkey_i_to_s_extent(&j->key), ptr) {
		rcu_read_lock();
		ca = PTR_CACHE(c, ptr);
		if (ca)
			percpu_ref_get(&ca->ref);
		rcu_read_unlock();

		if (!ca) {
			/* XXX: fix this */
			bch_err(c, "missing device for journal write\n");
			continue;
		}

		atomic64_add(sectors, &ca->meta_sectors_written);

		bio = ca->journal.bio;
		bio_reset(bio);
		bio->bi_iter.bi_sector	= ptr->offset;
		bio->bi_bdev		= ca->disk_sb.bdev;
		bio->bi_iter.bi_size	= sectors << 9;
		bio->bi_end_io		= journal_write_endio;
		bio->bi_private		= ca;
		bio_set_op_attrs(bio, REQ_OP_WRITE,
				 REQ_SYNC|REQ_META|REQ_PREFLUSH|REQ_FUA);
		bch_bio_map(bio, jset);

		trace_bcache_journal_write(bio);
		closure_bio_submit_punt(bio, cl, c);

		ca->journal.bucket_seq[ca->journal.cur_idx] = le64_to_cpu(w->data->seq);
	}

	for_each_cache(ca, c, i)
		if (ca->mi.state == BCH_MEMBER_STATE_ACTIVE &&
		    journal_flushes_device(ca) &&
		    !bch_extent_has_device(bkey_i_to_s_c_extent(&j->key), i)) {
			percpu_ref_get(&ca->ref);

			bio = ca->journal.bio;
			bio_reset(bio);
			bio->bi_bdev		= ca->disk_sb.bdev;
			bio->bi_end_io		= journal_write_endio;
			bio->bi_private		= ca;
			bio_set_op_attrs(bio, REQ_OP_WRITE, WRITE_FLUSH);
			closure_bio_submit_punt(bio, cl, c);
		}

no_io:
	extent_for_each_ptr(bkey_i_to_s_extent(&j->key), ptr)
		ptr->offset += sectors;

	closure_return_with_destructor(cl, journal_write_done);
}

static void journal_write_work(struct work_struct *work)
{
	struct journal *j = container_of(to_delayed_work(work),
					 struct journal, write_work);
	spin_lock(&j->lock);
	set_bit(JOURNAL_NEED_WRITE, &j->flags);

	if (journal_buf_switch(j, false) != JOURNAL_UNLOCKED)
		spin_unlock(&j->lock);
}

/*
 * Given an inode number, if that inode number has data in the journal that
 * hasn't yet been flushed, return the journal sequence number that needs to be
 * flushed:
 */
u64 bch_inode_journal_seq(struct journal *j, u64 inode)
{
	size_t h = hash_64(inode, ilog2(sizeof(j->buf[0].has_inode) * 8));
	u64 seq = 0;

	if (!test_bit(h, j->buf[0].has_inode) &&
	    !test_bit(h, j->buf[1].has_inode))
		return 0;

	spin_lock(&j->lock);
	if (test_bit(h, journal_cur_buf(j)->has_inode))
		seq = atomic64_read(&j->seq);
	else if (test_bit(h, journal_prev_buf(j)->has_inode))
		seq = atomic64_read(&j->seq) - 1;
	spin_unlock(&j->lock);

	return seq;
}

static int __journal_res_get(struct journal *j, struct journal_res *res,
			      unsigned u64s_min, unsigned u64s_max)
{
	struct cache_set *c = container_of(j, struct cache_set, journal);
	int ret;
retry:
	ret = journal_res_get_fast(j, res, u64s_min, u64s_max);
	if (ret)
		return ret;

	spin_lock(&j->lock);
	/*
	 * Recheck after taking the lock, so we don't race with another thread
	 * that just did journal_entry_open() and call journal_entry_close()
	 * unnecessarily
	 */
	ret = journal_res_get_fast(j, res, u64s_min, u64s_max);
	if (ret) {
		spin_unlock(&j->lock);
		return 1;
	}

	/*
	 * Ok, no more room in the current journal entry - try to start a new
	 * one:
	 */
	switch (journal_buf_switch(j, false)) {
	case JOURNAL_ENTRY_ERROR:
		spin_unlock(&j->lock);
		return -EIO;
	case JOURNAL_ENTRY_INUSE:
		/* haven't finished writing out the previous one: */
		spin_unlock(&j->lock);
		trace_bcache_journal_entry_full(c);
		goto blocked;
	case JOURNAL_ENTRY_CLOSED:
		break;
	case JOURNAL_UNLOCKED:
		goto retry;
	}

	/* We now have a new, closed journal buf - see if we can open it: */
	ret = journal_entry_open(j);
	spin_unlock(&j->lock);

	if (ret < 0)
		return ret;
	if (ret)
		goto retry;

	/* Journal's full, we have to wait */

	/*
	 * Direct reclaim - can't rely on reclaim from work item
	 * due to freezing..
	 */
	journal_reclaim_work(&j->reclaim_work.work);

	trace_bcache_journal_full(c);
blocked:
	if (!j->res_get_blocked_start)
		j->res_get_blocked_start = local_clock() ?: 1;
	return 0;
}

/*
 * Essentially the entry function to the journaling code. When bcache is doing
 * a btree insert, it calls this function to get the current journal write.
 * Journal write is the structure used set up journal writes. The calling
 * function will then add its keys to the structure, queuing them for the
 * next write.
 *
 * To ensure forward progress, the current task must not be holding any
 * btree node write locks.
 */
int bch_journal_res_get_slowpath(struct journal *j, struct journal_res *res,
				 unsigned u64s_min, unsigned u64s_max)
{
	int ret;

	wait_event(j->wait,
		   (ret = __journal_res_get(j, res, u64s_min,
					    u64s_max)));
	return ret < 0 ? ret : 0;
}

void bch_journal_wait_on_seq(struct journal *j, u64 seq, struct closure *parent)
{
	spin_lock(&j->lock);

	BUG_ON(seq > atomic64_read(&j->seq));

	if (bch_journal_error(j)) {
		spin_unlock(&j->lock);
		return;
	}

	if (seq == atomic64_read(&j->seq)) {
		if (!closure_wait(&journal_cur_buf(j)->wait, parent))
			BUG();
	} else if (seq + 1 == atomic64_read(&j->seq) &&
		   j->reservations.prev_buf_unwritten) {
		if (!closure_wait(&journal_prev_buf(j)->wait, parent))
			BUG();

		smp_mb();

		/* check if raced with write completion (or failure) */
		if (!j->reservations.prev_buf_unwritten ||
		    bch_journal_error(j))
			closure_wake_up(&journal_prev_buf(j)->wait);
	}

	spin_unlock(&j->lock);
}

void bch_journal_flush_seq_async(struct journal *j, u64 seq, struct closure *parent)
{
	spin_lock(&j->lock);

	BUG_ON(seq > atomic64_read(&j->seq));

	if (bch_journal_error(j)) {
		spin_unlock(&j->lock);
		return;
	}

	if (seq == atomic64_read(&j->seq)) {
		bool set_need_write = false;

		if (parent &&
		    !closure_wait(&journal_cur_buf(j)->wait, parent))
			BUG();

		if (!test_and_set_bit(JOURNAL_NEED_WRITE, &j->flags)) {
			j->need_write_time = local_clock();
			set_need_write = true;
		}

		switch (journal_buf_switch(j, set_need_write)) {
		case JOURNAL_ENTRY_ERROR:
			if (parent)
				closure_wake_up(&journal_cur_buf(j)->wait);
			break;
		case JOURNAL_ENTRY_CLOSED:
			/*
			 * Journal entry hasn't been opened yet, but caller
			 * claims it has something (seq == j->seq):
			 */
			BUG();
		case JOURNAL_ENTRY_INUSE:
			break;
		case JOURNAL_UNLOCKED:
			return;
		}
	} else if (parent &&
		   seq + 1 == atomic64_read(&j->seq) &&
		   j->reservations.prev_buf_unwritten) {
		if (!closure_wait(&journal_prev_buf(j)->wait, parent))
			BUG();

		smp_mb();

		/* check if raced with write completion (or failure) */
		if (!j->reservations.prev_buf_unwritten ||
		    bch_journal_error(j))
			closure_wake_up(&journal_prev_buf(j)->wait);
	}

	spin_unlock(&j->lock);
}

int bch_journal_flush_seq(struct journal *j, u64 seq)
{
	struct closure cl;
	u64 start_time = local_clock();

	closure_init_stack(&cl);
	bch_journal_flush_seq_async(j, seq, &cl);
	closure_sync(&cl);

	bch_time_stats_update(j->flush_seq_time, start_time);

	return bch_journal_error(j);
}

void bch_journal_meta_async(struct journal *j, struct closure *parent)
{
	struct journal_res res;
	unsigned u64s = jset_u64s(0);

	memset(&res, 0, sizeof(res));

	bch_journal_res_get(j, &res, u64s, u64s);
	bch_journal_res_put(j, &res);

	bch_journal_flush_seq_async(j, res.seq, parent);
}

int bch_journal_meta(struct journal *j)
{
	struct journal_res res;
	unsigned u64s = jset_u64s(0);
	int ret;

	memset(&res, 0, sizeof(res));

	ret = bch_journal_res_get(j, &res, u64s, u64s);
	if (ret)
		return ret;

	bch_journal_res_put(j, &res);

	return bch_journal_flush_seq(j, res.seq);
}

void bch_journal_flush_async(struct journal *j, struct closure *parent)
{
	u64 seq, journal_seq;

	spin_lock(&j->lock);
	journal_seq = atomic64_read(&j->seq);

	if (journal_entry_is_open(j)) {
		seq = journal_seq;
	} else if (journal_seq) {
		seq = journal_seq - 1;
	} else {
		spin_unlock(&j->lock);
		return;
	}
	spin_unlock(&j->lock);

	bch_journal_flush_seq_async(j, seq, parent);
}

int bch_journal_flush(struct journal *j)
{
	u64 seq, journal_seq;

	spin_lock(&j->lock);
	journal_seq = atomic64_read(&j->seq);

	if (journal_entry_is_open(j)) {
		seq = journal_seq;
	} else if (journal_seq) {
		seq = journal_seq - 1;
	} else {
		spin_unlock(&j->lock);
		return 0;
	}
	spin_unlock(&j->lock);

	return bch_journal_flush_seq(j, seq);
}

void bch_journal_free(struct journal *j)
{
	unsigned order = get_order(j->entry_size_max);

	free_pages((unsigned long) j->buf[1].data, order);
	free_pages((unsigned long) j->buf[0].data, order);
	free_fifo(&j->pin);
}

int bch_journal_alloc(struct journal *j, unsigned entry_size_max)
{
	static struct lock_class_key res_key;
	unsigned order = get_order(entry_size_max);

	spin_lock_init(&j->lock);
	spin_lock_init(&j->pin_lock);
	init_waitqueue_head(&j->wait);
	INIT_DELAYED_WORK(&j->write_work, journal_write_work);
	INIT_DELAYED_WORK(&j->reclaim_work, journal_reclaim_work);
	mutex_init(&j->blacklist_lock);
	INIT_LIST_HEAD(&j->seq_blacklist);
	spin_lock_init(&j->devs.lock);
	mutex_init(&j->reclaim_lock);

	lockdep_init_map(&j->res_map, "journal res", &res_key, 0);

	j->entry_size_max	= entry_size_max;
	j->write_delay_ms	= 100;
	j->reclaim_delay_ms	= 100;

	bkey_extent_init(&j->key);

	atomic64_set(&j->reservations.counter,
		((union journal_res_state)
		 { .cur_entry_offset = JOURNAL_ENTRY_CLOSED_VAL }).v);

	if (!(init_fifo(&j->pin, JOURNAL_PIN, GFP_KERNEL)) ||
	    !(j->buf[0].data = (void *) __get_free_pages(GFP_KERNEL, order)) ||
	    !(j->buf[1].data = (void *) __get_free_pages(GFP_KERNEL, order)))
		return -ENOMEM;

	return 0;
}

ssize_t bch_journal_print_debug(struct journal *j, char *buf)
{
	union journal_res_state *s = &j->reservations;
	struct cache *ca;
	unsigned iter;
	ssize_t ret = 0;

	rcu_read_lock();
	spin_lock(&j->lock);

	ret += scnprintf(buf + ret, PAGE_SIZE - ret,
			 "active journal entries:\t%zu\n"
			 "seq:\t\t\t%llu\n"
			 "last_seq:\t\t%llu\n"
			 "last_seq_ondisk:\t%llu\n"
			 "reservation count:\t%u\n"
			 "reservation offset:\t%u\n"
			 "current entry u64s:\t%u\n"
			 "io in flight:\t\t%i\n"
			 "need write:\t\t%i\n"
			 "dirty:\t\t\t%i\n"
			 "replay done:\t\t%i\n",
			 fifo_used(&j->pin),
			 (u64) atomic64_read(&j->seq),
			 last_seq(j),
			 j->last_seq_ondisk,
			 journal_state_count(*s, s->idx),
			 s->cur_entry_offset,
			 j->cur_entry_u64s,
			 s->prev_buf_unwritten,
			 test_bit(JOURNAL_NEED_WRITE,	&j->flags),
			 journal_entry_is_open(j),
			 test_bit(JOURNAL_REPLAY_DONE,	&j->flags));

	group_for_each_cache_rcu(ca, &j->devs, iter) {
		struct journal_device *ja = &ca->journal;

		ret += scnprintf(buf + ret, PAGE_SIZE - ret,
				 "dev %u:\n"
				 "\tnr\t\t%u\n"
				 "\tcur_idx\t\t%u (seq %llu)\n"
				 "\tlast_idx\t%u (seq %llu)\n",
				 iter, ja->nr,
				 ja->cur_idx,	ja->bucket_seq[ja->cur_idx],
				 ja->last_idx,	ja->bucket_seq[ja->last_idx]);
	}

	spin_unlock(&j->lock);
	rcu_read_unlock();

	return ret;
}

static bool bch_journal_writing_to_device(struct cache *ca)
{
	struct journal *j = &ca->set->journal;
	bool ret;

	spin_lock(&j->lock);
	ret = bch_extent_has_device(bkey_i_to_s_c_extent(&j->key),
				    ca->dev_idx);
	spin_unlock(&j->lock);

	return ret;
}

/*
 * This asumes that ca has already been marked read-only so that
 * journal_next_bucket won't pick buckets out of ca any more.
 * Hence, if the journal is not currently pointing to ca, there
 * will be no new writes to journal entries in ca after all the
 * pending ones have been flushed to disk.
 *
 * If the journal is being written to ca, write a new record, and
 * journal_next_bucket will notice that the device is no longer
 * writeable and pick a new set of devices to write to.
 */

int bch_journal_move(struct cache *ca)
{
	u64 last_flushed_seq;
	struct journal_device *ja = &ca->journal;
	struct cache_set *c = ca->set;
	struct journal *j = &c->journal;
	unsigned i;
	int ret = 0;		/* Success */

	if (bch_journal_writing_to_device(ca)) {
		/*
		 * bch_journal_meta will write a record and we'll wait
		 * for the write to complete.
		 * Actually writing the journal (journal_write_locked)
		 * will call journal_next_bucket which notices that the
		 * device is no longer writeable, and picks a new one.
		 */
		bch_journal_meta(j);
		BUG_ON(bch_journal_writing_to_device(ca));
	}

	/*
	 * Flush all btree updates to backing store so that any
	 * journal entries written to ca become stale and are no
	 * longer needed.
	 */

	/*
	 * XXX: switch to normal journal reclaim machinery
	 */
	bch_btree_flush(c);

	/*
	 * Force a meta-data journal entry to be written so that
	 * we have newer journal entries in devices other than ca,
	 * and wait for the meta data write to complete.
	 */
	bch_journal_meta(j);

	/*
	 * Verify that we no longer need any of the journal entries in
	 * the device
	 */
	spin_lock(&j->lock);
	last_flushed_seq = last_seq(j);
	spin_unlock(&j->lock);

	for (i = 0; i < ja->nr; i += 1)
		BUG_ON(ja->bucket_seq[i] > last_flushed_seq);

	return ret;
}

void bch_journal_free_cache(struct cache *ca)
{
	kfree(ca->journal.buckets);
	kfree(ca->journal.bucket_seq);
}

int bch_journal_init_cache(struct cache *ca)
{
	struct journal_device *ja = &ca->journal;
	struct bch_sb_field_journal *journal_buckets =
		bch_sb_get_journal(ca->disk_sb.sb);
	unsigned i, journal_entry_pages;

	journal_entry_pages =
		DIV_ROUND_UP(1U << BCH_SB_JOURNAL_ENTRY_SIZE(ca->disk_sb.sb),
			     PAGE_SECTORS);

	ja->nr = bch_nr_journal_buckets(journal_buckets);

	ja->bucket_seq = kcalloc(ja->nr, sizeof(u64), GFP_KERNEL);
	if (!ja->bucket_seq)
		return -ENOMEM;

	ca->journal.bio = bio_kmalloc(GFP_KERNEL, journal_entry_pages);
	if (!ca->journal.bio)
		return -ENOMEM;

	ja->buckets = kcalloc(ja->nr, sizeof(u64), GFP_KERNEL);
	if (!ja->buckets)
		return -ENOMEM;

	for (i = 0; i < ja->nr; i++)
		ja->buckets[i] = le64_to_cpu(journal_buckets->buckets[i]);

	return 0;
}