summaryrefslogtreecommitdiff
path: root/block/qcow2-refcount.c
blob: 0632fc3bc0ec5a950e8e05cc4ca1c84e057f4a34 (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
/*
 * Block driver for the QCOW version 2 format
 *
 * Copyright (c) 2004-2006 Fabrice Bellard
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 */

#include "qemu-common.h"
#include "block/block_int.h"
#include "block/qcow2.h"
#include "qemu/range.h"

static int64_t alloc_clusters_noref(BlockDriverState *bs, uint64_t size);
static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs,
                            int64_t offset, int64_t length, uint64_t addend,
                            bool decrease, enum qcow2_discard_type type);

static uint64_t get_refcount_ro0(const void *refcount_array, uint64_t index);
static uint64_t get_refcount_ro1(const void *refcount_array, uint64_t index);
static uint64_t get_refcount_ro2(const void *refcount_array, uint64_t index);
static uint64_t get_refcount_ro3(const void *refcount_array, uint64_t index);
static uint64_t get_refcount_ro4(const void *refcount_array, uint64_t index);
static uint64_t get_refcount_ro5(const void *refcount_array, uint64_t index);
static uint64_t get_refcount_ro6(const void *refcount_array, uint64_t index);

static void set_refcount_ro0(void *refcount_array, uint64_t index,
                             uint64_t value);
static void set_refcount_ro1(void *refcount_array, uint64_t index,
                             uint64_t value);
static void set_refcount_ro2(void *refcount_array, uint64_t index,
                             uint64_t value);
static void set_refcount_ro3(void *refcount_array, uint64_t index,
                             uint64_t value);
static void set_refcount_ro4(void *refcount_array, uint64_t index,
                             uint64_t value);
static void set_refcount_ro5(void *refcount_array, uint64_t index,
                             uint64_t value);
static void set_refcount_ro6(void *refcount_array, uint64_t index,
                             uint64_t value);


static Qcow2GetRefcountFunc *const get_refcount_funcs[] = {
    &get_refcount_ro0,
    &get_refcount_ro1,
    &get_refcount_ro2,
    &get_refcount_ro3,
    &get_refcount_ro4,
    &get_refcount_ro5,
    &get_refcount_ro6
};

static Qcow2SetRefcountFunc *const set_refcount_funcs[] = {
    &set_refcount_ro0,
    &set_refcount_ro1,
    &set_refcount_ro2,
    &set_refcount_ro3,
    &set_refcount_ro4,
    &set_refcount_ro5,
    &set_refcount_ro6
};


/*********************************************************/
/* refcount handling */

int qcow2_refcount_init(BlockDriverState *bs)
{
    BDRVQcowState *s = bs->opaque;
    unsigned int refcount_table_size2, i;
    int ret;

    assert(s->refcount_order >= 0 && s->refcount_order <= 6);

    s->get_refcount = get_refcount_funcs[s->refcount_order];
    s->set_refcount = set_refcount_funcs[s->refcount_order];

    assert(s->refcount_table_size <= INT_MAX / sizeof(uint64_t));
    refcount_table_size2 = s->refcount_table_size * sizeof(uint64_t);
    s->refcount_table = g_try_malloc(refcount_table_size2);

    if (s->refcount_table_size > 0) {
        if (s->refcount_table == NULL) {
            ret = -ENOMEM;
            goto fail;
        }
        BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_LOAD);
        ret = bdrv_pread(bs->file, s->refcount_table_offset,
                         s->refcount_table, refcount_table_size2);
        if (ret < 0) {
            goto fail;
        }
        for(i = 0; i < s->refcount_table_size; i++)
            be64_to_cpus(&s->refcount_table[i]);
    }
    return 0;
 fail:
    return ret;
}

void qcow2_refcount_close(BlockDriverState *bs)
{
    BDRVQcowState *s = bs->opaque;
    g_free(s->refcount_table);
}


static uint64_t get_refcount_ro0(const void *refcount_array, uint64_t index)
{
    return (((const uint8_t *)refcount_array)[index / 8] >> (index % 8)) & 0x1;
}

static void set_refcount_ro0(void *refcount_array, uint64_t index,
                             uint64_t value)
{
    assert(!(value >> 1));
    ((uint8_t *)refcount_array)[index / 8] &= ~(0x1 << (index % 8));
    ((uint8_t *)refcount_array)[index / 8] |= value << (index % 8);
}

static uint64_t get_refcount_ro1(const void *refcount_array, uint64_t index)
{
    return (((const uint8_t *)refcount_array)[index / 4] >> (2 * (index % 4)))
           & 0x3;
}

static void set_refcount_ro1(void *refcount_array, uint64_t index,
                             uint64_t value)
{
    assert(!(value >> 2));
    ((uint8_t *)refcount_array)[index / 4] &= ~(0x3 << (2 * (index % 4)));
    ((uint8_t *)refcount_array)[index / 4] |= value << (2 * (index % 4));
}

static uint64_t get_refcount_ro2(const void *refcount_array, uint64_t index)
{
    return (((const uint8_t *)refcount_array)[index / 2] >> (4 * (index % 2)))
           & 0xf;
}

static void set_refcount_ro2(void *refcount_array, uint64_t index,
                             uint64_t value)
{
    assert(!(value >> 4));
    ((uint8_t *)refcount_array)[index / 2] &= ~(0xf << (4 * (index % 2)));
    ((uint8_t *)refcount_array)[index / 2] |= value << (4 * (index % 2));
}

static uint64_t get_refcount_ro3(const void *refcount_array, uint64_t index)
{
    return ((const uint8_t *)refcount_array)[index];
}

static void set_refcount_ro3(void *refcount_array, uint64_t index,
                             uint64_t value)
{
    assert(!(value >> 8));
    ((uint8_t *)refcount_array)[index] = value;
}

static uint64_t get_refcount_ro4(const void *refcount_array, uint64_t index)
{
    return be16_to_cpu(((const uint16_t *)refcount_array)[index]);
}

static void set_refcount_ro4(void *refcount_array, uint64_t index,
                             uint64_t value)
{
    assert(!(value >> 16));
    ((uint16_t *)refcount_array)[index] = cpu_to_be16(value);
}

static uint64_t get_refcount_ro5(const void *refcount_array, uint64_t index)
{
    return be32_to_cpu(((const uint32_t *)refcount_array)[index]);
}

static void set_refcount_ro5(void *refcount_array, uint64_t index,
                             uint64_t value)
{
    assert(!(value >> 32));
    ((uint32_t *)refcount_array)[index] = cpu_to_be32(value);
}

static uint64_t get_refcount_ro6(const void *refcount_array, uint64_t index)
{
    return be64_to_cpu(((const uint64_t *)refcount_array)[index]);
}

static void set_refcount_ro6(void *refcount_array, uint64_t index,
                             uint64_t value)
{
    ((uint64_t *)refcount_array)[index] = cpu_to_be64(value);
}


static int load_refcount_block(BlockDriverState *bs,
                               int64_t refcount_block_offset,
                               void **refcount_block)
{
    BDRVQcowState *s = bs->opaque;
    int ret;

    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_LOAD);
    ret = qcow2_cache_get(bs, s->refcount_block_cache, refcount_block_offset,
        refcount_block);

    return ret;
}

/*
 * Retrieves the refcount of the cluster given by its index and stores it in
 * *refcount. Returns 0 on success and -errno on failure.
 */
int qcow2_get_refcount(BlockDriverState *bs, int64_t cluster_index,
                       uint64_t *refcount)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t refcount_table_index, block_index;
    int64_t refcount_block_offset;
    int ret;
    void *refcount_block;

    refcount_table_index = cluster_index >> s->refcount_block_bits;
    if (refcount_table_index >= s->refcount_table_size) {
        *refcount = 0;
        return 0;
    }
    refcount_block_offset =
        s->refcount_table[refcount_table_index] & REFT_OFFSET_MASK;
    if (!refcount_block_offset) {
        *refcount = 0;
        return 0;
    }

    if (offset_into_cluster(s, refcount_block_offset)) {
        qcow2_signal_corruption(bs, true, -1, -1, "Refblock offset %#" PRIx64
                                " unaligned (reftable index: %#" PRIx64 ")",
                                refcount_block_offset, refcount_table_index);
        return -EIO;
    }

    ret = qcow2_cache_get(bs, s->refcount_block_cache, refcount_block_offset,
                          &refcount_block);
    if (ret < 0) {
        return ret;
    }

    block_index = cluster_index & (s->refcount_block_size - 1);
    *refcount = s->get_refcount(refcount_block, block_index);

    qcow2_cache_put(bs, s->refcount_block_cache, &refcount_block);

    return 0;
}

/*
 * Rounds the refcount table size up to avoid growing the table for each single
 * refcount block that is allocated.
 */
static unsigned int next_refcount_table_size(BDRVQcowState *s,
    unsigned int min_size)
{
    unsigned int min_clusters = (min_size >> (s->cluster_bits - 3)) + 1;
    unsigned int refcount_table_clusters =
        MAX(1, s->refcount_table_size >> (s->cluster_bits - 3));

    while (min_clusters > refcount_table_clusters) {
        refcount_table_clusters = (refcount_table_clusters * 3 + 1) / 2;
    }

    return refcount_table_clusters << (s->cluster_bits - 3);
}


/* Checks if two offsets are described by the same refcount block */
static int in_same_refcount_block(BDRVQcowState *s, uint64_t offset_a,
    uint64_t offset_b)
{
    uint64_t block_a = offset_a >> (s->cluster_bits + s->refcount_block_bits);
    uint64_t block_b = offset_b >> (s->cluster_bits + s->refcount_block_bits);

    return (block_a == block_b);
}

/*
 * Loads a refcount block. If it doesn't exist yet, it is allocated first
 * (including growing the refcount table if needed).
 *
 * Returns 0 on success or -errno in error case
 */
static int alloc_refcount_block(BlockDriverState *bs,
                                int64_t cluster_index, void **refcount_block)
{
    BDRVQcowState *s = bs->opaque;
    unsigned int refcount_table_index;
    int ret;

    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC);

    /* Find the refcount block for the given cluster */
    refcount_table_index = cluster_index >> s->refcount_block_bits;

    if (refcount_table_index < s->refcount_table_size) {

        uint64_t refcount_block_offset =
            s->refcount_table[refcount_table_index] & REFT_OFFSET_MASK;

        /* If it's already there, we're done */
        if (refcount_block_offset) {
            if (offset_into_cluster(s, refcount_block_offset)) {
                qcow2_signal_corruption(bs, true, -1, -1, "Refblock offset %#"
                                        PRIx64 " unaligned (reftable index: "
                                        "%#x)", refcount_block_offset,
                                        refcount_table_index);
                return -EIO;
            }

             return load_refcount_block(bs, refcount_block_offset,
                                        refcount_block);
        }
    }

    /*
     * If we came here, we need to allocate something. Something is at least
     * a cluster for the new refcount block. It may also include a new refcount
     * table if the old refcount table is too small.
     *
     * Note that allocating clusters here needs some special care:
     *
     * - We can't use the normal qcow2_alloc_clusters(), it would try to
     *   increase the refcount and very likely we would end up with an endless
     *   recursion. Instead we must place the refcount blocks in a way that
     *   they can describe them themselves.
     *
     * - We need to consider that at this point we are inside update_refcounts
     *   and potentially doing an initial refcount increase. This means that
     *   some clusters have already been allocated by the caller, but their
     *   refcount isn't accurate yet. If we allocate clusters for metadata, we
     *   need to return -EAGAIN to signal the caller that it needs to restart
     *   the search for free clusters.
     *
     * - alloc_clusters_noref and qcow2_free_clusters may load a different
     *   refcount block into the cache
     */

    *refcount_block = NULL;

    /* We write to the refcount table, so we might depend on L2 tables */
    ret = qcow2_cache_flush(bs, s->l2_table_cache);
    if (ret < 0) {
        return ret;
    }

    /* Allocate the refcount block itself and mark it as used */
    int64_t new_block = alloc_clusters_noref(bs, s->cluster_size);
    if (new_block < 0) {
        return new_block;
    }

#ifdef DEBUG_ALLOC2
    fprintf(stderr, "qcow2: Allocate refcount block %d for %" PRIx64
        " at %" PRIx64 "\n",
        refcount_table_index, cluster_index << s->cluster_bits, new_block);
#endif

    if (in_same_refcount_block(s, new_block, cluster_index << s->cluster_bits)) {
        /* Zero the new refcount block before updating it */
        ret = qcow2_cache_get_empty(bs, s->refcount_block_cache, new_block,
                                    refcount_block);
        if (ret < 0) {
            goto fail_block;
        }

        memset(*refcount_block, 0, s->cluster_size);

        /* The block describes itself, need to update the cache */
        int block_index = (new_block >> s->cluster_bits) &
            (s->refcount_block_size - 1);
        s->set_refcount(*refcount_block, block_index, 1);
    } else {
        /* Described somewhere else. This can recurse at most twice before we
         * arrive at a block that describes itself. */
        ret = update_refcount(bs, new_block, s->cluster_size, 1, false,
                              QCOW2_DISCARD_NEVER);
        if (ret < 0) {
            goto fail_block;
        }

        ret = qcow2_cache_flush(bs, s->refcount_block_cache);
        if (ret < 0) {
            goto fail_block;
        }

        /* Initialize the new refcount block only after updating its refcount,
         * update_refcount uses the refcount cache itself */
        ret = qcow2_cache_get_empty(bs, s->refcount_block_cache, new_block,
                                    refcount_block);
        if (ret < 0) {
            goto fail_block;
        }

        memset(*refcount_block, 0, s->cluster_size);
    }

    /* Now the new refcount block needs to be written to disk */
    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE);
    qcow2_cache_entry_mark_dirty(bs, s->refcount_block_cache, *refcount_block);
    ret = qcow2_cache_flush(bs, s->refcount_block_cache);
    if (ret < 0) {
        goto fail_block;
    }

    /* If the refcount table is big enough, just hook the block up there */
    if (refcount_table_index < s->refcount_table_size) {
        uint64_t data64 = cpu_to_be64(new_block);
        BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_HOOKUP);
        ret = bdrv_pwrite_sync(bs->file,
            s->refcount_table_offset + refcount_table_index * sizeof(uint64_t),
            &data64, sizeof(data64));
        if (ret < 0) {
            goto fail_block;
        }

        s->refcount_table[refcount_table_index] = new_block;

        /* The new refcount block may be where the caller intended to put its
         * data, so let it restart the search. */
        return -EAGAIN;
    }

    qcow2_cache_put(bs, s->refcount_block_cache, refcount_block);

    /*
     * If we come here, we need to grow the refcount table. Again, a new
     * refcount table needs some space and we can't simply allocate to avoid
     * endless recursion.
     *
     * Therefore let's grab new refcount blocks at the end of the image, which
     * will describe themselves and the new refcount table. This way we can
     * reference them only in the new table and do the switch to the new
     * refcount table at once without producing an inconsistent state in
     * between.
     */
    BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_GROW);

    /* Calculate the number of refcount blocks needed so far; this will be the
     * basis for calculating the index of the first cluster used for the
     * self-describing refcount structures which we are about to create.
     *
     * Because we reached this point, there cannot be any refcount entries for
     * cluster_index or higher indices yet. However, because new_block has been
     * allocated to describe that cluster (and it will assume this role later
     * on), we cannot use that index; also, new_block may actually have a higher
     * cluster index than cluster_index, so it needs to be taken into account
     * here (and 1 needs to be added to its value because that cluster is used).
     */
    uint64_t blocks_used = DIV_ROUND_UP(MAX(cluster_index + 1,
                                            (new_block >> s->cluster_bits) + 1),
                                        s->refcount_block_size);

    if (blocks_used > QCOW_MAX_REFTABLE_SIZE / sizeof(uint64_t)) {
        return -EFBIG;
    }

    /* And now we need at least one block more for the new metadata */
    uint64_t table_size = next_refcount_table_size(s, blocks_used + 1);
    uint64_t last_table_size;
    uint64_t blocks_clusters;
    do {
        uint64_t table_clusters =
            size_to_clusters(s, table_size * sizeof(uint64_t));
        blocks_clusters = 1 +
            ((table_clusters + s->refcount_block_size - 1)
            / s->refcount_block_size);
        uint64_t meta_clusters = table_clusters + blocks_clusters;

        last_table_size = table_size;
        table_size = next_refcount_table_size(s, blocks_used +
            ((meta_clusters + s->refcount_block_size - 1)
            / s->refcount_block_size));

    } while (last_table_size != table_size);

#ifdef DEBUG_ALLOC2
    fprintf(stderr, "qcow2: Grow refcount table %" PRId32 " => %" PRId64 "\n",
        s->refcount_table_size, table_size);
#endif

    /* Create the new refcount table and blocks */
    uint64_t meta_offset = (blocks_used * s->refcount_block_size) *
        s->cluster_size;
    uint64_t table_offset = meta_offset + blocks_clusters * s->cluster_size;
    uint64_t *new_table = g_try_new0(uint64_t, table_size);
    void *new_blocks = g_try_malloc0(blocks_clusters * s->cluster_size);

    assert(table_size > 0 && blocks_clusters > 0);
    if (new_table == NULL || new_blocks == NULL) {
        ret = -ENOMEM;
        goto fail_table;
    }

    /* Fill the new refcount table */
    memcpy(new_table, s->refcount_table,
        s->refcount_table_size * sizeof(uint64_t));
    new_table[refcount_table_index] = new_block;

    int i;
    for (i = 0; i < blocks_clusters; i++) {
        new_table[blocks_used + i] = meta_offset + (i * s->cluster_size);
    }

    /* Fill the refcount blocks */
    uint64_t table_clusters = size_to_clusters(s, table_size * sizeof(uint64_t));
    int block = 0;
    for (i = 0; i < table_clusters + blocks_clusters; i++) {
        s->set_refcount(new_blocks, block++, 1);
    }

    /* Write refcount blocks to disk */
    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE_BLOCKS);
    ret = bdrv_pwrite_sync(bs->file, meta_offset, new_blocks,
        blocks_clusters * s->cluster_size);
    g_free(new_blocks);
    new_blocks = NULL;
    if (ret < 0) {
        goto fail_table;
    }

    /* Write refcount table to disk */
    for(i = 0; i < table_size; i++) {
        cpu_to_be64s(&new_table[i]);
    }

    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE_TABLE);
    ret = bdrv_pwrite_sync(bs->file, table_offset, new_table,
        table_size * sizeof(uint64_t));
    if (ret < 0) {
        goto fail_table;
    }

    for(i = 0; i < table_size; i++) {
        be64_to_cpus(&new_table[i]);
    }

    /* Hook up the new refcount table in the qcow2 header */
    uint8_t data[12];
    cpu_to_be64w((uint64_t*)data, table_offset);
    cpu_to_be32w((uint32_t*)(data + 8), table_clusters);
    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_SWITCH_TABLE);
    ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, refcount_table_offset),
        data, sizeof(data));
    if (ret < 0) {
        goto fail_table;
    }

    /* And switch it in memory */
    uint64_t old_table_offset = s->refcount_table_offset;
    uint64_t old_table_size = s->refcount_table_size;

    g_free(s->refcount_table);
    s->refcount_table = new_table;
    s->refcount_table_size = table_size;
    s->refcount_table_offset = table_offset;

    /* Free old table. */
    qcow2_free_clusters(bs, old_table_offset, old_table_size * sizeof(uint64_t),
                        QCOW2_DISCARD_OTHER);

    ret = load_refcount_block(bs, new_block, refcount_block);
    if (ret < 0) {
        return ret;
    }

    /* If we were trying to do the initial refcount update for some cluster
     * allocation, we might have used the same clusters to store newly
     * allocated metadata. Make the caller search some new space. */
    return -EAGAIN;

fail_table:
    g_free(new_blocks);
    g_free(new_table);
fail_block:
    if (*refcount_block != NULL) {
        qcow2_cache_put(bs, s->refcount_block_cache, refcount_block);
    }
    return ret;
}

void qcow2_process_discards(BlockDriverState *bs, int ret)
{
    BDRVQcowState *s = bs->opaque;
    Qcow2DiscardRegion *d, *next;

    QTAILQ_FOREACH_SAFE(d, &s->discards, next, next) {
        QTAILQ_REMOVE(&s->discards, d, next);

        /* Discard is optional, ignore the return value */
        if (ret >= 0) {
            bdrv_discard(bs->file,
                         d->offset >> BDRV_SECTOR_BITS,
                         d->bytes >> BDRV_SECTOR_BITS);
        }

        g_free(d);
    }
}

static void update_refcount_discard(BlockDriverState *bs,
                                    uint64_t offset, uint64_t length)
{
    BDRVQcowState *s = bs->opaque;
    Qcow2DiscardRegion *d, *p, *next;

    QTAILQ_FOREACH(d, &s->discards, next) {
        uint64_t new_start = MIN(offset, d->offset);
        uint64_t new_end = MAX(offset + length, d->offset + d->bytes);

        if (new_end - new_start <= length + d->bytes) {
            /* There can't be any overlap, areas ending up here have no
             * references any more and therefore shouldn't get freed another
             * time. */
            assert(d->bytes + length == new_end - new_start);
            d->offset = new_start;
            d->bytes = new_end - new_start;
            goto found;
        }
    }

    d = g_malloc(sizeof(*d));
    *d = (Qcow2DiscardRegion) {
        .bs     = bs,
        .offset = offset,
        .bytes  = length,
    };
    QTAILQ_INSERT_TAIL(&s->discards, d, next);

found:
    /* Merge discard requests if they are adjacent now */
    QTAILQ_FOREACH_SAFE(p, &s->discards, next, next) {
        if (p == d
            || p->offset > d->offset + d->bytes
            || d->offset > p->offset + p->bytes)
        {
            continue;
        }

        /* Still no overlap possible */
        assert(p->offset == d->offset + d->bytes
            || d->offset == p->offset + p->bytes);

        QTAILQ_REMOVE(&s->discards, p, next);
        d->offset = MIN(d->offset, p->offset);
        d->bytes += p->bytes;
        g_free(p);
    }
}

/* XXX: cache several refcount block clusters ? */
/* @addend is the absolute value of the addend; if @decrease is set, @addend
 * will be subtracted from the current refcount, otherwise it will be added */
static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs,
                                                   int64_t offset,
                                                   int64_t length,
                                                   uint64_t addend,
                                                   bool decrease,
                                                   enum qcow2_discard_type type)
{
    BDRVQcowState *s = bs->opaque;
    int64_t start, last, cluster_offset;
    void *refcount_block = NULL;
    int64_t old_table_index = -1;
    int ret;

#ifdef DEBUG_ALLOC2
    fprintf(stderr, "update_refcount: offset=%" PRId64 " size=%" PRId64
            " addend=%s%" PRIu64 "\n", offset, length, decrease ? "-" : "",
            addend);
#endif
    if (length < 0) {
        return -EINVAL;
    } else if (length == 0) {
        return 0;
    }

    if (decrease) {
        qcow2_cache_set_dependency(bs, s->refcount_block_cache,
            s->l2_table_cache);
    }

    start = start_of_cluster(s, offset);
    last = start_of_cluster(s, offset + length - 1);
    for(cluster_offset = start; cluster_offset <= last;
        cluster_offset += s->cluster_size)
    {
        int block_index;
        uint64_t refcount;
        int64_t cluster_index = cluster_offset >> s->cluster_bits;
        int64_t table_index = cluster_index >> s->refcount_block_bits;

        /* Load the refcount block and allocate it if needed */
        if (table_index != old_table_index) {
            if (refcount_block) {
                qcow2_cache_put(bs, s->refcount_block_cache, &refcount_block);
            }
            ret = alloc_refcount_block(bs, cluster_index, &refcount_block);
            if (ret < 0) {
                goto fail;
            }
        }
        old_table_index = table_index;

        qcow2_cache_entry_mark_dirty(bs, s->refcount_block_cache,
                                     refcount_block);

        /* we can update the count and save it */
        block_index = cluster_index & (s->refcount_block_size - 1);

        refcount = s->get_refcount(refcount_block, block_index);
        if (decrease ? (refcount - addend > refcount)
                     : (refcount + addend < refcount ||
                        refcount + addend > s->refcount_max))
        {
            ret = -EINVAL;
            goto fail;
        }
        if (decrease) {
            refcount -= addend;
        } else {
            refcount += addend;
        }
        if (refcount == 0 && cluster_index < s->free_cluster_index) {
            s->free_cluster_index = cluster_index;
        }
        s->set_refcount(refcount_block, block_index, refcount);

        if (refcount == 0 && s->discard_passthrough[type]) {
            update_refcount_discard(bs, cluster_offset, s->cluster_size);
        }
    }

    ret = 0;
fail:
    if (!s->cache_discards) {
        qcow2_process_discards(bs, ret);
    }

    /* Write last changed block to disk */
    if (refcount_block) {
        qcow2_cache_put(bs, s->refcount_block_cache, &refcount_block);
    }

    /*
     * Try do undo any updates if an error is returned (This may succeed in
     * some cases like ENOSPC for allocating a new refcount block)
     */
    if (ret < 0) {
        int dummy;
        dummy = update_refcount(bs, offset, cluster_offset - offset, addend,
                                !decrease, QCOW2_DISCARD_NEVER);
        (void)dummy;
    }

    return ret;
}

/*
 * Increases or decreases the refcount of a given cluster.
 *
 * @addend is the absolute value of the addend; if @decrease is set, @addend
 * will be subtracted from the current refcount, otherwise it will be added.
 *
 * On success 0 is returned; on failure -errno is returned.
 */
int qcow2_update_cluster_refcount(BlockDriverState *bs,
                                  int64_t cluster_index,
                                  uint64_t addend, bool decrease,
                                  enum qcow2_discard_type type)
{
    BDRVQcowState *s = bs->opaque;
    int ret;

    ret = update_refcount(bs, cluster_index << s->cluster_bits, 1, addend,
                          decrease, type);
    if (ret < 0) {
        return ret;
    }

    return 0;
}



/*********************************************************/
/* cluster allocation functions */



/* return < 0 if error */
static int64_t alloc_clusters_noref(BlockDriverState *bs, uint64_t size)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t i, nb_clusters, refcount;
    int ret;

    /* We can't allocate clusters if they may still be queued for discard. */
    if (s->cache_discards) {
        qcow2_process_discards(bs, 0);
    }

    nb_clusters = size_to_clusters(s, size);
retry:
    for(i = 0; i < nb_clusters; i++) {
        uint64_t next_cluster_index = s->free_cluster_index++;
        ret = qcow2_get_refcount(bs, next_cluster_index, &refcount);

        if (ret < 0) {
            return ret;
        } else if (refcount != 0) {
            goto retry;
        }
    }

    /* Make sure that all offsets in the "allocated" range are representable
     * in an int64_t */
    if (s->free_cluster_index > 0 &&
        s->free_cluster_index - 1 > (INT64_MAX >> s->cluster_bits))
    {
        return -EFBIG;
    }

#ifdef DEBUG_ALLOC2
    fprintf(stderr, "alloc_clusters: size=%" PRId64 " -> %" PRId64 "\n",
            size,
            (s->free_cluster_index - nb_clusters) << s->cluster_bits);
#endif
    return (s->free_cluster_index - nb_clusters) << s->cluster_bits;
}

int64_t qcow2_alloc_clusters(BlockDriverState *bs, uint64_t size)
{
    int64_t offset;
    int ret;

    BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_ALLOC);
    do {
        offset = alloc_clusters_noref(bs, size);
        if (offset < 0) {
            return offset;
        }

        ret = update_refcount(bs, offset, size, 1, false, QCOW2_DISCARD_NEVER);
    } while (ret == -EAGAIN);

    if (ret < 0) {
        return ret;
    }

    return offset;
}

int qcow2_alloc_clusters_at(BlockDriverState *bs, uint64_t offset,
    int nb_clusters)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t cluster_index, refcount;
    uint64_t i;
    int ret;

    assert(nb_clusters >= 0);
    if (nb_clusters == 0) {
        return 0;
    }

    do {
        /* Check how many clusters there are free */
        cluster_index = offset >> s->cluster_bits;
        for(i = 0; i < nb_clusters; i++) {
            ret = qcow2_get_refcount(bs, cluster_index++, &refcount);
            if (ret < 0) {
                return ret;
            } else if (refcount != 0) {
                break;
            }
        }

        /* And then allocate them */
        ret = update_refcount(bs, offset, i << s->cluster_bits, 1, false,
                              QCOW2_DISCARD_NEVER);
    } while (ret == -EAGAIN);

    if (ret < 0) {
        return ret;
    }

    return i;
}

/* only used to allocate compressed sectors. We try to allocate
   contiguous sectors. size must be <= cluster_size */
int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size)
{
    BDRVQcowState *s = bs->opaque;
    int64_t offset;
    size_t free_in_cluster;
    int ret;

    BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_ALLOC_BYTES);
    assert(size > 0 && size <= s->cluster_size);
    assert(!s->free_byte_offset || offset_into_cluster(s, s->free_byte_offset));

    offset = s->free_byte_offset;

    if (offset) {
        uint64_t refcount;
        ret = qcow2_get_refcount(bs, offset >> s->cluster_bits, &refcount);
        if (ret < 0) {
            return ret;
        }

        if (refcount == s->refcount_max) {
            offset = 0;
        }
    }

    free_in_cluster = s->cluster_size - offset_into_cluster(s, offset);
    if (!offset || free_in_cluster < size) {
        int64_t new_cluster = alloc_clusters_noref(bs, s->cluster_size);
        if (new_cluster < 0) {
            return new_cluster;
        }

        if (!offset || ROUND_UP(offset, s->cluster_size) != new_cluster) {
            offset = new_cluster;
        }
    }

    assert(offset);
    ret = update_refcount(bs, offset, size, 1, false, QCOW2_DISCARD_NEVER);
    if (ret < 0) {
        return ret;
    }

    /* The cluster refcount was incremented; refcount blocks must be flushed
     * before the caller's L2 table updates. */
    qcow2_cache_set_dependency(bs, s->l2_table_cache, s->refcount_block_cache);

    s->free_byte_offset = offset + size;
    if (!offset_into_cluster(s, s->free_byte_offset)) {
        s->free_byte_offset = 0;
    }

    return offset;
}

void qcow2_free_clusters(BlockDriverState *bs,
                          int64_t offset, int64_t size,
                          enum qcow2_discard_type type)
{
    int ret;

    BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_FREE);
    ret = update_refcount(bs, offset, size, 1, true, type);
    if (ret < 0) {
        fprintf(stderr, "qcow2_free_clusters failed: %s\n", strerror(-ret));
        /* TODO Remember the clusters to free them later and avoid leaking */
    }
}

/*
 * Free a cluster using its L2 entry (handles clusters of all types, e.g.
 * normal cluster, compressed cluster, etc.)
 */
void qcow2_free_any_clusters(BlockDriverState *bs, uint64_t l2_entry,
                             int nb_clusters, enum qcow2_discard_type type)
{
    BDRVQcowState *s = bs->opaque;

    switch (qcow2_get_cluster_type(l2_entry)) {
    case QCOW2_CLUSTER_COMPRESSED:
        {
            int nb_csectors;
            nb_csectors = ((l2_entry >> s->csize_shift) &
                           s->csize_mask) + 1;
            qcow2_free_clusters(bs,
                (l2_entry & s->cluster_offset_mask) & ~511,
                nb_csectors * 512, type);
        }
        break;
    case QCOW2_CLUSTER_NORMAL:
    case QCOW2_CLUSTER_ZERO:
        if (l2_entry & L2E_OFFSET_MASK) {
            if (offset_into_cluster(s, l2_entry & L2E_OFFSET_MASK)) {
                qcow2_signal_corruption(bs, false, -1, -1,
                                        "Cannot free unaligned cluster %#llx",
                                        l2_entry & L2E_OFFSET_MASK);
            } else {
                qcow2_free_clusters(bs, l2_entry & L2E_OFFSET_MASK,
                                    nb_clusters << s->cluster_bits, type);
            }
        }
        break;
    case QCOW2_CLUSTER_UNALLOCATED:
        break;
    default:
        abort();
    }
}



/*********************************************************/
/* snapshots and image creation */



/* update the refcounts of snapshots and the copied flag */
int qcow2_update_snapshot_refcount(BlockDriverState *bs,
    int64_t l1_table_offset, int l1_size, int addend)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t *l1_table, *l2_table, l2_offset, offset, l1_size2, refcount;
    bool l1_allocated = false;
    int64_t old_offset, old_l2_offset;
    int i, j, l1_modified = 0, nb_csectors;
    int ret;

    assert(addend >= -1 && addend <= 1);

    l2_table = NULL;
    l1_table = NULL;
    l1_size2 = l1_size * sizeof(uint64_t);

    s->cache_discards = true;

    /* WARNING: qcow2_snapshot_goto relies on this function not using the
     * l1_table_offset when it is the current s->l1_table_offset! Be careful
     * when changing this! */
    if (l1_table_offset != s->l1_table_offset) {
        l1_table = g_try_malloc0(align_offset(l1_size2, 512));
        if (l1_size2 && l1_table == NULL) {
            ret = -ENOMEM;
            goto fail;
        }
        l1_allocated = true;

        ret = bdrv_pread(bs->file, l1_table_offset, l1_table, l1_size2);
        if (ret < 0) {
            goto fail;
        }

        for(i = 0;i < l1_size; i++)
            be64_to_cpus(&l1_table[i]);
    } else {
        assert(l1_size == s->l1_size);
        l1_table = s->l1_table;
        l1_allocated = false;
    }

    for(i = 0; i < l1_size; i++) {
        l2_offset = l1_table[i];
        if (l2_offset) {
            old_l2_offset = l2_offset;
            l2_offset &= L1E_OFFSET_MASK;

            if (offset_into_cluster(s, l2_offset)) {
                qcow2_signal_corruption(bs, true, -1, -1, "L2 table offset %#"
                                        PRIx64 " unaligned (L1 index: %#x)",
                                        l2_offset, i);
                ret = -EIO;
                goto fail;
            }

            ret = qcow2_cache_get(bs, s->l2_table_cache, l2_offset,
                (void**) &l2_table);
            if (ret < 0) {
                goto fail;
            }

            for(j = 0; j < s->l2_size; j++) {
                uint64_t cluster_index;

                offset = be64_to_cpu(l2_table[j]);
                old_offset = offset;
                offset &= ~QCOW_OFLAG_COPIED;

                switch (qcow2_get_cluster_type(offset)) {
                    case QCOW2_CLUSTER_COMPRESSED:
                        nb_csectors = ((offset >> s->csize_shift) &
                                       s->csize_mask) + 1;
                        if (addend != 0) {
                            ret = update_refcount(bs,
                                (offset & s->cluster_offset_mask) & ~511,
                                nb_csectors * 512, abs(addend), addend < 0,
                                QCOW2_DISCARD_SNAPSHOT);
                            if (ret < 0) {
                                goto fail;
                            }
                        }
                        /* compressed clusters are never modified */
                        refcount = 2;
                        break;

                    case QCOW2_CLUSTER_NORMAL:
                    case QCOW2_CLUSTER_ZERO:
                        if (offset_into_cluster(s, offset & L2E_OFFSET_MASK)) {
                            qcow2_signal_corruption(bs, true, -1, -1, "Data "
                                                    "cluster offset %#llx "
                                                    "unaligned (L2 offset: %#"
                                                    PRIx64 ", L2 index: %#x)",
                                                    offset & L2E_OFFSET_MASK,
                                                    l2_offset, j);
                            ret = -EIO;
                            goto fail;
                        }

                        cluster_index = (offset & L2E_OFFSET_MASK) >> s->cluster_bits;
                        if (!cluster_index) {
                            /* unallocated */
                            refcount = 0;
                            break;
                        }
                        if (addend != 0) {
                            ret = qcow2_update_cluster_refcount(bs,
                                    cluster_index, abs(addend), addend < 0,
                                    QCOW2_DISCARD_SNAPSHOT);
                            if (ret < 0) {
                                goto fail;
                            }
                        }

                        ret = qcow2_get_refcount(bs, cluster_index, &refcount);
                        if (ret < 0) {
                            goto fail;
                        }
                        break;

                    case QCOW2_CLUSTER_UNALLOCATED:
                        refcount = 0;
                        break;

                    default:
                        abort();
                }

                if (refcount == 1) {
                    offset |= QCOW_OFLAG_COPIED;
                }
                if (offset != old_offset) {
                    if (addend > 0) {
                        qcow2_cache_set_dependency(bs, s->l2_table_cache,
                            s->refcount_block_cache);
                    }
                    l2_table[j] = cpu_to_be64(offset);
                    qcow2_cache_entry_mark_dirty(bs, s->l2_table_cache,
                                                 l2_table);
                }
            }

            qcow2_cache_put(bs, s->l2_table_cache, (void **) &l2_table);

            if (addend != 0) {
                ret = qcow2_update_cluster_refcount(bs, l2_offset >>
                                                        s->cluster_bits,
                                                    abs(addend), addend < 0,
                                                    QCOW2_DISCARD_SNAPSHOT);
                if (ret < 0) {
                    goto fail;
                }
            }
            ret = qcow2_get_refcount(bs, l2_offset >> s->cluster_bits,
                                     &refcount);
            if (ret < 0) {
                goto fail;
            } else if (refcount == 1) {
                l2_offset |= QCOW_OFLAG_COPIED;
            }
            if (l2_offset != old_l2_offset) {
                l1_table[i] = l2_offset;
                l1_modified = 1;
            }
        }
    }

    ret = bdrv_flush(bs);
fail:
    if (l2_table) {
        qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
    }

    s->cache_discards = false;
    qcow2_process_discards(bs, ret);

    /* Update L1 only if it isn't deleted anyway (addend = -1) */
    if (ret == 0 && addend >= 0 && l1_modified) {
        for (i = 0; i < l1_size; i++) {
            cpu_to_be64s(&l1_table[i]);
        }

        ret = bdrv_pwrite_sync(bs->file, l1_table_offset, l1_table, l1_size2);

        for (i = 0; i < l1_size; i++) {
            be64_to_cpus(&l1_table[i]);
        }
    }
    if (l1_allocated)
        g_free(l1_table);
    return ret;
}




/*********************************************************/
/* refcount checking functions */


static size_t refcount_array_byte_size(BDRVQcowState *s, uint64_t entries)
{
    /* This assertion holds because there is no way we can address more than
     * 2^(64 - 9) clusters at once (with cluster size 512 = 2^9, and because
     * offsets have to be representable in bytes); due to every cluster
     * corresponding to one refcount entry, we are well below that limit */
    assert(entries < (UINT64_C(1) << (64 - 9)));

    /* Thanks to the assertion this will not overflow, because
     * s->refcount_order < 7.
     * (note: x << s->refcount_order == x * s->refcount_bits) */
    return DIV_ROUND_UP(entries << s->refcount_order, 8);
}

/**
 * Reallocates *array so that it can hold new_size entries. *size must contain
 * the current number of entries in *array. If the reallocation fails, *array
 * and *size will not be modified and -errno will be returned. If the
 * reallocation is successful, *array will be set to the new buffer, *size
 * will be set to new_size and 0 will be returned. The size of the reallocated
 * refcount array buffer will be aligned to a cluster boundary, and the newly
 * allocated area will be zeroed.
 */
static int realloc_refcount_array(BDRVQcowState *s, void **array,
                                  int64_t *size, int64_t new_size)
{
    size_t old_byte_size, new_byte_size;
    void *new_ptr;

    /* Round to clusters so the array can be directly written to disk */
    old_byte_size = size_to_clusters(s, refcount_array_byte_size(s, *size))
                    * s->cluster_size;
    new_byte_size = size_to_clusters(s, refcount_array_byte_size(s, new_size))
                    * s->cluster_size;

    if (new_byte_size == old_byte_size) {
        *size = new_size;
        return 0;
    }

    assert(new_byte_size > 0);

    new_ptr = g_try_realloc(*array, new_byte_size);
    if (!new_ptr) {
        return -ENOMEM;
    }

    if (new_byte_size > old_byte_size) {
        memset((void *)((uintptr_t)new_ptr + old_byte_size), 0,
               new_byte_size - old_byte_size);
    }

    *array = new_ptr;
    *size  = new_size;

    return 0;
}

/*
 * Increases the refcount for a range of clusters in a given refcount table.
 * This is used to construct a temporary refcount table out of L1 and L2 tables
 * which can be compared the the refcount table saved in the image.
 *
 * Modifies the number of errors in res.
 */
static int inc_refcounts(BlockDriverState *bs,
                         BdrvCheckResult *res,
                         void **refcount_table,
                         int64_t *refcount_table_size,
                         int64_t offset, int64_t size)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t start, last, cluster_offset, k, refcount;
    int ret;

    if (size <= 0) {
        return 0;
    }

    start = start_of_cluster(s, offset);
    last = start_of_cluster(s, offset + size - 1);
    for(cluster_offset = start; cluster_offset <= last;
        cluster_offset += s->cluster_size) {
        k = cluster_offset >> s->cluster_bits;
        if (k >= *refcount_table_size) {
            ret = realloc_refcount_array(s, refcount_table,
                                         refcount_table_size, k + 1);
            if (ret < 0) {
                res->check_errors++;
                return ret;
            }
        }

        refcount = s->get_refcount(*refcount_table, k);
        if (refcount == s->refcount_max) {
            fprintf(stderr, "ERROR: overflow cluster offset=0x%" PRIx64
                    "\n", cluster_offset);
            res->corruptions++;
            continue;
        }
        s->set_refcount(*refcount_table, k, refcount + 1);
    }

    return 0;
}

/* Flags for check_refcounts_l1() and check_refcounts_l2() */
enum {
    CHECK_FRAG_INFO = 0x2,      /* update BlockFragInfo counters */
};

/*
 * Increases the refcount in the given refcount table for the all clusters
 * referenced in the L2 table. While doing so, performs some checks on L2
 * entries.
 *
 * Returns the number of errors found by the checks or -errno if an internal
 * error occurred.
 */
static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
                              void **refcount_table,
                              int64_t *refcount_table_size, int64_t l2_offset,
                              int flags)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t *l2_table, l2_entry;
    uint64_t next_contiguous_offset = 0;
    int i, l2_size, nb_csectors, ret;

    /* Read L2 table from disk */
    l2_size = s->l2_size * sizeof(uint64_t);
    l2_table = g_malloc(l2_size);

    ret = bdrv_pread(bs->file, l2_offset, l2_table, l2_size);
    if (ret < 0) {
        fprintf(stderr, "ERROR: I/O error in check_refcounts_l2\n");
        res->check_errors++;
        goto fail;
    }

    /* Do the actual checks */
    for(i = 0; i < s->l2_size; i++) {
        l2_entry = be64_to_cpu(l2_table[i]);

        switch (qcow2_get_cluster_type(l2_entry)) {
        case QCOW2_CLUSTER_COMPRESSED:
            /* Compressed clusters don't have QCOW_OFLAG_COPIED */
            if (l2_entry & QCOW_OFLAG_COPIED) {
                fprintf(stderr, "ERROR: cluster %" PRId64 ": "
                    "copied flag must never be set for compressed "
                    "clusters\n", l2_entry >> s->cluster_bits);
                l2_entry &= ~QCOW_OFLAG_COPIED;
                res->corruptions++;
            }

            /* Mark cluster as used */
            nb_csectors = ((l2_entry >> s->csize_shift) &
                           s->csize_mask) + 1;
            l2_entry &= s->cluster_offset_mask;
            ret = inc_refcounts(bs, res, refcount_table, refcount_table_size,
                                l2_entry & ~511, nb_csectors * 512);
            if (ret < 0) {
                goto fail;
            }

            if (flags & CHECK_FRAG_INFO) {
                res->bfi.allocated_clusters++;
                res->bfi.compressed_clusters++;

                /* Compressed clusters are fragmented by nature.  Since they
                 * take up sub-sector space but we only have sector granularity
                 * I/O we need to re-read the same sectors even for adjacent
                 * compressed clusters.
                 */
                res->bfi.fragmented_clusters++;
            }
            break;

        case QCOW2_CLUSTER_ZERO:
            if ((l2_entry & L2E_OFFSET_MASK) == 0) {
                break;
            }
            /* fall through */

        case QCOW2_CLUSTER_NORMAL:
        {
            uint64_t offset = l2_entry & L2E_OFFSET_MASK;

            if (flags & CHECK_FRAG_INFO) {
                res->bfi.allocated_clusters++;
                if (next_contiguous_offset &&
                    offset != next_contiguous_offset) {
                    res->bfi.fragmented_clusters++;
                }
                next_contiguous_offset = offset + s->cluster_size;
            }

            /* Mark cluster as used */
            ret = inc_refcounts(bs, res, refcount_table, refcount_table_size,
                                offset, s->cluster_size);
            if (ret < 0) {
                goto fail;
            }

            /* Correct offsets are cluster aligned */
            if (offset_into_cluster(s, offset)) {
                fprintf(stderr, "ERROR offset=%" PRIx64 ": Cluster is not "
                    "properly aligned; L2 entry corrupted.\n", offset);
                res->corruptions++;
            }
            break;
        }

        case QCOW2_CLUSTER_UNALLOCATED:
            break;

        default:
            abort();
        }
    }

    g_free(l2_table);
    return 0;

fail:
    g_free(l2_table);
    return ret;
}

/*
 * Increases the refcount for the L1 table, its L2 tables and all referenced
 * clusters in the given refcount table. While doing so, performs some checks
 * on L1 and L2 entries.
 *
 * Returns the number of errors found by the checks or -errno if an internal
 * error occurred.
 */
static int check_refcounts_l1(BlockDriverState *bs,
                              BdrvCheckResult *res,
                              void **refcount_table,
                              int64_t *refcount_table_size,
                              int64_t l1_table_offset, int l1_size,
                              int flags)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t *l1_table = NULL, l2_offset, l1_size2;
    int i, ret;

    l1_size2 = l1_size * sizeof(uint64_t);

    /* Mark L1 table as used */
    ret = inc_refcounts(bs, res, refcount_table, refcount_table_size,
                        l1_table_offset, l1_size2);
    if (ret < 0) {
        goto fail;
    }

    /* Read L1 table entries from disk */
    if (l1_size2 > 0) {
        l1_table = g_try_malloc(l1_size2);
        if (l1_table == NULL) {
            ret = -ENOMEM;
            res->check_errors++;
            goto fail;
        }
        ret = bdrv_pread(bs->file, l1_table_offset, l1_table, l1_size2);
        if (ret < 0) {
            fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n");
            res->check_errors++;
            goto fail;
        }
        for(i = 0;i < l1_size; i++)
            be64_to_cpus(&l1_table[i]);
    }

    /* Do the actual checks */
    for(i = 0; i < l1_size; i++) {
        l2_offset = l1_table[i];
        if (l2_offset) {
            /* Mark L2 table as used */
            l2_offset &= L1E_OFFSET_MASK;
            ret = inc_refcounts(bs, res, refcount_table, refcount_table_size,
                                l2_offset, s->cluster_size);
            if (ret < 0) {
                goto fail;
            }

            /* L2 tables are cluster aligned */
            if (offset_into_cluster(s, l2_offset)) {
                fprintf(stderr, "ERROR l2_offset=%" PRIx64 ": Table is not "
                    "cluster aligned; L1 entry corrupted\n", l2_offset);
                res->corruptions++;
            }

            /* Process and check L2 entries */
            ret = check_refcounts_l2(bs, res, refcount_table,
                                     refcount_table_size, l2_offset, flags);
            if (ret < 0) {
                goto fail;
            }
        }
    }
    g_free(l1_table);
    return 0;

fail:
    g_free(l1_table);
    return ret;
}

/*
 * Checks the OFLAG_COPIED flag for all L1 and L2 entries.
 *
 * This function does not print an error message nor does it increment
 * check_errors if qcow2_get_refcount fails (this is because such an error will
 * have been already detected and sufficiently signaled by the calling function
 * (qcow2_check_refcounts) by the time this function is called).
 */
static int check_oflag_copied(BlockDriverState *bs, BdrvCheckResult *res,
                              BdrvCheckMode fix)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t *l2_table = qemu_blockalign(bs, s->cluster_size);
    int ret;
    uint64_t refcount;
    int i, j;

    for (i = 0; i < s->l1_size; i++) {
        uint64_t l1_entry = s->l1_table[i];
        uint64_t l2_offset = l1_entry & L1E_OFFSET_MASK;
        bool l2_dirty = false;

        if (!l2_offset) {
            continue;
        }

        ret = qcow2_get_refcount(bs, l2_offset >> s->cluster_bits,
                                 &refcount);
        if (ret < 0) {
            /* don't print message nor increment check_errors */
            continue;
        }
        if ((refcount == 1) != ((l1_entry & QCOW_OFLAG_COPIED) != 0)) {
            fprintf(stderr, "%s OFLAG_COPIED L2 cluster: l1_index=%d "
                    "l1_entry=%" PRIx64 " refcount=%" PRIu64 "\n",
                    fix & BDRV_FIX_ERRORS ? "Repairing" :
                                            "ERROR",
                    i, l1_entry, refcount);
            if (fix & BDRV_FIX_ERRORS) {
                s->l1_table[i] = refcount == 1
                               ? l1_entry |  QCOW_OFLAG_COPIED
                               : l1_entry & ~QCOW_OFLAG_COPIED;
                ret = qcow2_write_l1_entry(bs, i);
                if (ret < 0) {
                    res->check_errors++;
                    goto fail;
                }
                res->corruptions_fixed++;
            } else {
                res->corruptions++;
            }
        }

        ret = bdrv_pread(bs->file, l2_offset, l2_table,
                         s->l2_size * sizeof(uint64_t));
        if (ret < 0) {
            fprintf(stderr, "ERROR: Could not read L2 table: %s\n",
                    strerror(-ret));
            res->check_errors++;
            goto fail;
        }

        for (j = 0; j < s->l2_size; j++) {
            uint64_t l2_entry = be64_to_cpu(l2_table[j]);
            uint64_t data_offset = l2_entry & L2E_OFFSET_MASK;
            int cluster_type = qcow2_get_cluster_type(l2_entry);

            if ((cluster_type == QCOW2_CLUSTER_NORMAL) ||
                ((cluster_type == QCOW2_CLUSTER_ZERO) && (data_offset != 0))) {
                ret = qcow2_get_refcount(bs,
                                         data_offset >> s->cluster_bits,
                                         &refcount);
                if (ret < 0) {
                    /* don't print message nor increment check_errors */
                    continue;
                }
                if ((refcount == 1) != ((l2_entry & QCOW_OFLAG_COPIED) != 0)) {
                    fprintf(stderr, "%s OFLAG_COPIED data cluster: "
                            "l2_entry=%" PRIx64 " refcount=%" PRIu64 "\n",
                            fix & BDRV_FIX_ERRORS ? "Repairing" :
                                                    "ERROR",
                            l2_entry, refcount);
                    if (fix & BDRV_FIX_ERRORS) {
                        l2_table[j] = cpu_to_be64(refcount == 1
                                    ? l2_entry |  QCOW_OFLAG_COPIED
                                    : l2_entry & ~QCOW_OFLAG_COPIED);
                        l2_dirty = true;
                        res->corruptions_fixed++;
                    } else {
                        res->corruptions++;
                    }
                }
            }
        }

        if (l2_dirty) {
            ret = qcow2_pre_write_overlap_check(bs, QCOW2_OL_ACTIVE_L2,
                                                l2_offset, s->cluster_size);
            if (ret < 0) {
                fprintf(stderr, "ERROR: Could not write L2 table; metadata "
                        "overlap check failed: %s\n", strerror(-ret));
                res->check_errors++;
                goto fail;
            }

            ret = bdrv_pwrite(bs->file, l2_offset, l2_table, s->cluster_size);
            if (ret < 0) {
                fprintf(stderr, "ERROR: Could not write L2 table: %s\n",
                        strerror(-ret));
                res->check_errors++;
                goto fail;
            }
        }
    }

    ret = 0;

fail:
    qemu_vfree(l2_table);
    return ret;
}

/*
 * Checks consistency of refblocks and accounts for each refblock in
 * *refcount_table.
 */
static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
                           BdrvCheckMode fix, bool *rebuild,
                           void **refcount_table, int64_t *nb_clusters)
{
    BDRVQcowState *s = bs->opaque;
    int64_t i, size;
    int ret;

    for(i = 0; i < s->refcount_table_size; i++) {
        uint64_t offset, cluster;
        offset = s->refcount_table[i];
        cluster = offset >> s->cluster_bits;

        /* Refcount blocks are cluster aligned */
        if (offset_into_cluster(s, offset)) {
            fprintf(stderr, "ERROR refcount block %" PRId64 " is not "
                "cluster aligned; refcount table entry corrupted\n", i);
            res->corruptions++;
            *rebuild = true;
            continue;
        }

        if (cluster >= *nb_clusters) {
            fprintf(stderr, "%s refcount block %" PRId64 " is outside image\n",
                    fix & BDRV_FIX_ERRORS ? "Repairing" : "ERROR", i);

            if (fix & BDRV_FIX_ERRORS) {
                int64_t new_nb_clusters;

                if (offset > INT64_MAX - s->cluster_size) {
                    ret = -EINVAL;
                    goto resize_fail;
                }

                ret = bdrv_truncate(bs->file, offset + s->cluster_size);
                if (ret < 0) {
                    goto resize_fail;
                }
                size = bdrv_getlength(bs->file);
                if (size < 0) {
                    ret = size;
                    goto resize_fail;
                }

                new_nb_clusters = size_to_clusters(s, size);
                assert(new_nb_clusters >= *nb_clusters);

                ret = realloc_refcount_array(s, refcount_table,
                                             nb_clusters, new_nb_clusters);
                if (ret < 0) {
                    res->check_errors++;
                    return ret;
                }

                if (cluster >= *nb_clusters) {
                    ret = -EINVAL;
                    goto resize_fail;
                }

                res->corruptions_fixed++;
                ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
                                    offset, s->cluster_size);
                if (ret < 0) {
                    return ret;
                }
                /* No need to check whether the refcount is now greater than 1:
                 * This area was just allocated and zeroed, so it can only be
                 * exactly 1 after inc_refcounts() */
                continue;

resize_fail:
                res->corruptions++;
                *rebuild = true;
                fprintf(stderr, "ERROR could not resize image: %s\n",
                        strerror(-ret));
            } else {
                res->corruptions++;
            }
            continue;
        }

        if (offset != 0) {
            ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
                                offset, s->cluster_size);
            if (ret < 0) {
                return ret;
            }
            if (s->get_refcount(*refcount_table, cluster) != 1) {
                fprintf(stderr, "ERROR refcount block %" PRId64
                        " refcount=%" PRIu64 "\n", i,
                        s->get_refcount(*refcount_table, cluster));
                res->corruptions++;
                *rebuild = true;
            }
        }
    }

    return 0;
}

/*
 * Calculates an in-memory refcount table.
 */
static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
                               BdrvCheckMode fix, bool *rebuild,
                               void **refcount_table, int64_t *nb_clusters)
{
    BDRVQcowState *s = bs->opaque;
    int64_t i;
    QCowSnapshot *sn;
    int ret;

    if (!*refcount_table) {
        int64_t old_size = 0;
        ret = realloc_refcount_array(s, refcount_table,
                                     &old_size, *nb_clusters);
        if (ret < 0) {
            res->check_errors++;
            return ret;
        }
    }

    /* header */
    ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
                        0, s->cluster_size);
    if (ret < 0) {
        return ret;
    }

    /* current L1 table */
    ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters,
                             s->l1_table_offset, s->l1_size, CHECK_FRAG_INFO);
    if (ret < 0) {
        return ret;
    }

    /* snapshots */
    for (i = 0; i < s->nb_snapshots; i++) {
        sn = s->snapshots + i;
        ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters,
                                 sn->l1_table_offset, sn->l1_size, 0);
        if (ret < 0) {
            return ret;
        }
    }
    ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
                        s->snapshots_offset, s->snapshots_size);
    if (ret < 0) {
        return ret;
    }

    /* refcount data */
    ret = inc_refcounts(bs, res, refcount_table, nb_clusters,
                        s->refcount_table_offset,
                        s->refcount_table_size * sizeof(uint64_t));
    if (ret < 0) {
        return ret;
    }

    return check_refblocks(bs, res, fix, rebuild, refcount_table, nb_clusters);
}

/*
 * Compares the actual reference count for each cluster in the image against the
 * refcount as reported by the refcount structures on-disk.
 */
static void compare_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
                              BdrvCheckMode fix, bool *rebuild,
                              int64_t *highest_cluster,
                              void *refcount_table, int64_t nb_clusters)
{
    BDRVQcowState *s = bs->opaque;
    int64_t i;
    uint64_t refcount1, refcount2;
    int ret;

    for (i = 0, *highest_cluster = 0; i < nb_clusters; i++) {
        ret = qcow2_get_refcount(bs, i, &refcount1);
        if (ret < 0) {
            fprintf(stderr, "Can't get refcount for cluster %" PRId64 ": %s\n",
                    i, strerror(-ret));
            res->check_errors++;
            continue;
        }

        refcount2 = s->get_refcount(refcount_table, i);

        if (refcount1 > 0 || refcount2 > 0) {
            *highest_cluster = i;
        }

        if (refcount1 != refcount2) {
            /* Check if we're allowed to fix the mismatch */
            int *num_fixed = NULL;
            if (refcount1 == 0) {
                *rebuild = true;
            } else if (refcount1 > refcount2 && (fix & BDRV_FIX_LEAKS)) {
                num_fixed = &res->leaks_fixed;
            } else if (refcount1 < refcount2 && (fix & BDRV_FIX_ERRORS)) {
                num_fixed = &res->corruptions_fixed;
            }

            fprintf(stderr, "%s cluster %" PRId64 " refcount=%" PRIu64
                    " reference=%" PRIu64 "\n",
                   num_fixed != NULL     ? "Repairing" :
                   refcount1 < refcount2 ? "ERROR" :
                                           "Leaked",
                   i, refcount1, refcount2);

            if (num_fixed) {
                ret = update_refcount(bs, i << s->cluster_bits, 1,
                                      refcount_diff(refcount1, refcount2),
                                      refcount1 > refcount2,
                                      QCOW2_DISCARD_ALWAYS);
                if (ret >= 0) {
                    (*num_fixed)++;
                    continue;
                }
            }

            /* And if we couldn't, print an error */
            if (refcount1 < refcount2) {
                res->corruptions++;
            } else {
                res->leaks++;
            }
        }
    }
}

/*
 * Allocates clusters using an in-memory refcount table (IMRT) in contrast to
 * the on-disk refcount structures.
 *
 * On input, *first_free_cluster tells where to start looking, and need not
 * actually be a free cluster; the returned offset will not be before that
 * cluster.  On output, *first_free_cluster points to the first gap found, even
 * if that gap was too small to be used as the returned offset.
 *
 * Note that *first_free_cluster is a cluster index whereas the return value is
 * an offset.
 */
static int64_t alloc_clusters_imrt(BlockDriverState *bs,
                                   int cluster_count,
                                   void **refcount_table,
                                   int64_t *imrt_nb_clusters,
                                   int64_t *first_free_cluster)
{
    BDRVQcowState *s = bs->opaque;
    int64_t cluster = *first_free_cluster, i;
    bool first_gap = true;
    int contiguous_free_clusters;
    int ret;

    /* Starting at *first_free_cluster, find a range of at least cluster_count
     * continuously free clusters */
    for (contiguous_free_clusters = 0;
         cluster < *imrt_nb_clusters &&
         contiguous_free_clusters < cluster_count;
         cluster++)
    {
        if (!s->get_refcount(*refcount_table, cluster)) {
            contiguous_free_clusters++;
            if (first_gap) {
                /* If this is the first free cluster found, update
                 * *first_free_cluster accordingly */
                *first_free_cluster = cluster;
                first_gap = false;
            }
        } else if (contiguous_free_clusters) {
            contiguous_free_clusters = 0;
        }
    }

    /* If contiguous_free_clusters is greater than zero, it contains the number
     * of continuously free clusters until the current cluster; the first free
     * cluster in the current "gap" is therefore
     * cluster - contiguous_free_clusters */

    /* If no such range could be found, grow the in-memory refcount table
     * accordingly to append free clusters at the end of the image */
    if (contiguous_free_clusters < cluster_count) {
        /* contiguous_free_clusters clusters are already empty at the image end;
         * we need cluster_count clusters; therefore, we have to allocate
         * cluster_count - contiguous_free_clusters new clusters at the end of
         * the image (which is the current value of cluster; note that cluster
         * may exceed old_imrt_nb_clusters if *first_free_cluster pointed beyond
         * the image end) */
        ret = realloc_refcount_array(s, refcount_table, imrt_nb_clusters,
                                     cluster + cluster_count
                                     - contiguous_free_clusters);
        if (ret < 0) {
            return ret;
        }
    }

    /* Go back to the first free cluster */
    cluster -= contiguous_free_clusters;
    for (i = 0; i < cluster_count; i++) {
        s->set_refcount(*refcount_table, cluster + i, 1);
    }

    return cluster << s->cluster_bits;
}

/*
 * Creates a new refcount structure based solely on the in-memory information
 * given through *refcount_table. All necessary allocations will be reflected
 * in that array.
 *
 * On success, the old refcount structure is leaked (it will be covered by the
 * new refcount structure).
 */
static int rebuild_refcount_structure(BlockDriverState *bs,
                                      BdrvCheckResult *res,
                                      void **refcount_table,
                                      int64_t *nb_clusters)
{
    BDRVQcowState *s = bs->opaque;
    int64_t first_free_cluster = 0, reftable_offset = -1, cluster = 0;
    int64_t refblock_offset, refblock_start, refblock_index;
    uint32_t reftable_size = 0;
    uint64_t *on_disk_reftable = NULL;
    void *on_disk_refblock;
    int ret = 0;
    struct {
        uint64_t reftable_offset;
        uint32_t reftable_clusters;
    } QEMU_PACKED reftable_offset_and_clusters;

    qcow2_cache_empty(bs, s->refcount_block_cache);

write_refblocks:
    for (; cluster < *nb_clusters; cluster++) {
        if (!s->get_refcount(*refcount_table, cluster)) {
            continue;
        }

        refblock_index = cluster >> s->refcount_block_bits;
        refblock_start = refblock_index << s->refcount_block_bits;

        /* Don't allocate a cluster in a refblock already written to disk */
        if (first_free_cluster < refblock_start) {
            first_free_cluster = refblock_start;
        }
        refblock_offset = alloc_clusters_imrt(bs, 1, refcount_table,
                                              nb_clusters, &first_free_cluster);
        if (refblock_offset < 0) {
            fprintf(stderr, "ERROR allocating refblock: %s\n",
                    strerror(-refblock_offset));
            res->check_errors++;
            ret = refblock_offset;
            goto fail;
        }

        if (reftable_size <= refblock_index) {
            uint32_t old_reftable_size = reftable_size;
            uint64_t *new_on_disk_reftable;

            reftable_size = ROUND_UP((refblock_index + 1) * sizeof(uint64_t),
                                     s->cluster_size) / sizeof(uint64_t);
            new_on_disk_reftable = g_try_realloc(on_disk_reftable,
                                                 reftable_size *
                                                 sizeof(uint64_t));
            if (!new_on_disk_reftable) {
                res->check_errors++;
                ret = -ENOMEM;
                goto fail;
            }
            on_disk_reftable = new_on_disk_reftable;

            memset(on_disk_reftable + old_reftable_size, 0,
                   (reftable_size - old_reftable_size) * sizeof(uint64_t));

            /* The offset we have for the reftable is now no longer valid;
             * this will leak that range, but we can easily fix that by running
             * a leak-fixing check after this rebuild operation */
            reftable_offset = -1;
        }
        on_disk_reftable[refblock_index] = refblock_offset;

        /* If this is apparently the last refblock (for now), try to squeeze the
         * reftable in */
        if (refblock_index == (*nb_clusters - 1) >> s->refcount_block_bits &&
            reftable_offset < 0)
        {
            uint64_t reftable_clusters = size_to_clusters(s, reftable_size *
                                                          sizeof(uint64_t));
            reftable_offset = alloc_clusters_imrt(bs, reftable_clusters,
                                                  refcount_table, nb_clusters,
                                                  &first_free_cluster);
            if (reftable_offset < 0) {
                fprintf(stderr, "ERROR allocating reftable: %s\n",
                        strerror(-reftable_offset));
                res->check_errors++;
                ret = reftable_offset;
                goto fail;
            }
        }

        ret = qcow2_pre_write_overlap_check(bs, 0, refblock_offset,
                                            s->cluster_size);
        if (ret < 0) {
            fprintf(stderr, "ERROR writing refblock: %s\n", strerror(-ret));
            goto fail;
        }

        /* The size of *refcount_table is always cluster-aligned, therefore the
         * write operation will not overflow */
        on_disk_refblock = (void *)((char *) *refcount_table +
                                    refblock_index * s->cluster_size);

        ret = bdrv_write(bs->file, refblock_offset / BDRV_SECTOR_SIZE,
                         on_disk_refblock, s->cluster_sectors);
        if (ret < 0) {
            fprintf(stderr, "ERROR writing refblock: %s\n", strerror(-ret));
            goto fail;
        }

        /* Go to the end of this refblock */
        cluster = refblock_start + s->refcount_block_size - 1;
    }

    if (reftable_offset < 0) {
        uint64_t post_refblock_start, reftable_clusters;

        post_refblock_start = ROUND_UP(*nb_clusters, s->refcount_block_size);
        reftable_clusters = size_to_clusters(s,
                                             reftable_size * sizeof(uint64_t));
        /* Not pretty but simple */
        if (first_free_cluster < post_refblock_start) {
            first_free_cluster = post_refblock_start;
        }
        reftable_offset = alloc_clusters_imrt(bs, reftable_clusters,
                                              refcount_table, nb_clusters,
                                              &first_free_cluster);
        if (reftable_offset < 0) {
            fprintf(stderr, "ERROR allocating reftable: %s\n",
                    strerror(-reftable_offset));
            res->check_errors++;
            ret = reftable_offset;
            goto fail;
        }

        goto write_refblocks;
    }

    assert(on_disk_reftable);

    for (refblock_index = 0; refblock_index < reftable_size; refblock_index++) {
        cpu_to_be64s(&on_disk_reftable[refblock_index]);
    }

    ret = qcow2_pre_write_overlap_check(bs, 0, reftable_offset,
                                        reftable_size * sizeof(uint64_t));
    if (ret < 0) {
        fprintf(stderr, "ERROR writing reftable: %s\n", strerror(-ret));
        goto fail;
    }

    assert(reftable_size < INT_MAX / sizeof(uint64_t));
    ret = bdrv_pwrite(bs->file, reftable_offset, on_disk_reftable,
                      reftable_size * sizeof(uint64_t));
    if (ret < 0) {
        fprintf(stderr, "ERROR writing reftable: %s\n", strerror(-ret));
        goto fail;
    }

    /* Enter new reftable into the image header */
    cpu_to_be64w(&reftable_offset_and_clusters.reftable_offset,
                 reftable_offset);
    cpu_to_be32w(&reftable_offset_and_clusters.reftable_clusters,
                 size_to_clusters(s, reftable_size * sizeof(uint64_t)));
    ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader,
                                              refcount_table_offset),
                           &reftable_offset_and_clusters,
                           sizeof(reftable_offset_and_clusters));
    if (ret < 0) {
        fprintf(stderr, "ERROR setting reftable: %s\n", strerror(-ret));
        goto fail;
    }

    for (refblock_index = 0; refblock_index < reftable_size; refblock_index++) {
        be64_to_cpus(&on_disk_reftable[refblock_index]);
    }
    s->refcount_table = on_disk_reftable;
    s->refcount_table_offset = reftable_offset;
    s->refcount_table_size = reftable_size;

    return 0;

fail:
    g_free(on_disk_reftable);
    return ret;
}

/*
 * Checks an image for refcount consistency.
 *
 * Returns 0 if no errors are found, the number of errors in case the image is
 * detected as corrupted, and -errno when an internal error occurred.
 */
int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
                          BdrvCheckMode fix)
{
    BDRVQcowState *s = bs->opaque;
    BdrvCheckResult pre_compare_res;
    int64_t size, highest_cluster, nb_clusters;
    void *refcount_table = NULL;
    bool rebuild = false;
    int ret;

    size = bdrv_getlength(bs->file);
    if (size < 0) {
        res->check_errors++;
        return size;
    }

    nb_clusters = size_to_clusters(s, size);
    if (nb_clusters > INT_MAX) {
        res->check_errors++;
        return -EFBIG;
    }

    res->bfi.total_clusters =
        size_to_clusters(s, bs->total_sectors * BDRV_SECTOR_SIZE);

    ret = calculate_refcounts(bs, res, fix, &rebuild, &refcount_table,
                              &nb_clusters);
    if (ret < 0) {
        goto fail;
    }

    /* In case we don't need to rebuild the refcount structure (but want to fix
     * something), this function is immediately called again, in which case the
     * result should be ignored */
    pre_compare_res = *res;
    compare_refcounts(bs, res, 0, &rebuild, &highest_cluster, refcount_table,
                      nb_clusters);

    if (rebuild && (fix & BDRV_FIX_ERRORS)) {
        BdrvCheckResult old_res = *res;
        int fresh_leaks = 0;

        fprintf(stderr, "Rebuilding refcount structure\n");
        ret = rebuild_refcount_structure(bs, res, &refcount_table,
                                         &nb_clusters);
        if (ret < 0) {
            goto fail;
        }

        res->corruptions = 0;
        res->leaks = 0;

        /* Because the old reftable has been exchanged for a new one the
         * references have to be recalculated */
        rebuild = false;
        memset(refcount_table, 0, refcount_array_byte_size(s, nb_clusters));
        ret = calculate_refcounts(bs, res, 0, &rebuild, &refcount_table,
                                  &nb_clusters);
        if (ret < 0) {
            goto fail;
        }

        if (fix & BDRV_FIX_LEAKS) {
            /* The old refcount structures are now leaked, fix it; the result
             * can be ignored, aside from leaks which were introduced by
             * rebuild_refcount_structure() that could not be fixed */
            BdrvCheckResult saved_res = *res;
            *res = (BdrvCheckResult){ 0 };

            compare_refcounts(bs, res, BDRV_FIX_LEAKS, &rebuild,
                              &highest_cluster, refcount_table, nb_clusters);
            if (rebuild) {
                fprintf(stderr, "ERROR rebuilt refcount structure is still "
                        "broken\n");
            }

            /* Any leaks accounted for here were introduced by
             * rebuild_refcount_structure() because that function has created a
             * new refcount structure from scratch */
            fresh_leaks = res->leaks;
            *res = saved_res;
        }

        if (res->corruptions < old_res.corruptions) {
            res->corruptions_fixed += old_res.corruptions - res->corruptions;
        }
        if (res->leaks < old_res.leaks) {
            res->leaks_fixed += old_res.leaks - res->leaks;
        }
        res->leaks += fresh_leaks;
    } else if (fix) {
        if (rebuild) {
            fprintf(stderr, "ERROR need to rebuild refcount structures\n");
            res->check_errors++;
            ret = -EIO;
            goto fail;
        }

        if (res->leaks || res->corruptions) {
            *res = pre_compare_res;
            compare_refcounts(bs, res, fix, &rebuild, &highest_cluster,
                              refcount_table, nb_clusters);
        }
    }

    /* check OFLAG_COPIED */
    ret = check_oflag_copied(bs, res, fix);
    if (ret < 0) {
        goto fail;
    }

    res->image_end_offset = (highest_cluster + 1) * s->cluster_size;
    ret = 0;

fail:
    g_free(refcount_table);

    return ret;
}

#define overlaps_with(ofs, sz) \
    ranges_overlap(offset, size, ofs, sz)

/*
 * Checks if the given offset into the image file is actually free to use by
 * looking for overlaps with important metadata sections (L1/L2 tables etc.),
 * i.e. a sanity check without relying on the refcount tables.
 *
 * The ign parameter specifies what checks not to perform (being a bitmask of
 * QCow2MetadataOverlap values), i.e., what sections to ignore.
 *
 * Returns:
 * - 0 if writing to this offset will not affect the mentioned metadata
 * - a positive QCow2MetadataOverlap value indicating one overlapping section
 * - a negative value (-errno) indicating an error while performing a check,
 *   e.g. when bdrv_read failed on QCOW2_OL_INACTIVE_L2
 */
int qcow2_check_metadata_overlap(BlockDriverState *bs, int ign, int64_t offset,
                                 int64_t size)
{
    BDRVQcowState *s = bs->opaque;
    int chk = s->overlap_check & ~ign;
    int i, j;

    if (!size) {
        return 0;
    }

    if (chk & QCOW2_OL_MAIN_HEADER) {
        if (offset < s->cluster_size) {
            return QCOW2_OL_MAIN_HEADER;
        }
    }

    /* align range to test to cluster boundaries */
    size = align_offset(offset_into_cluster(s, offset) + size, s->cluster_size);
    offset = start_of_cluster(s, offset);

    if ((chk & QCOW2_OL_ACTIVE_L1) && s->l1_size) {
        if (overlaps_with(s->l1_table_offset, s->l1_size * sizeof(uint64_t))) {
            return QCOW2_OL_ACTIVE_L1;
        }
    }

    if ((chk & QCOW2_OL_REFCOUNT_TABLE) && s->refcount_table_size) {
        if (overlaps_with(s->refcount_table_offset,
            s->refcount_table_size * sizeof(uint64_t))) {
            return QCOW2_OL_REFCOUNT_TABLE;
        }
    }

    if ((chk & QCOW2_OL_SNAPSHOT_TABLE) && s->snapshots_size) {
        if (overlaps_with(s->snapshots_offset, s->snapshots_size)) {
            return QCOW2_OL_SNAPSHOT_TABLE;
        }
    }

    if ((chk & QCOW2_OL_INACTIVE_L1) && s->snapshots) {
        for (i = 0; i < s->nb_snapshots; i++) {
            if (s->snapshots[i].l1_size &&
                overlaps_with(s->snapshots[i].l1_table_offset,
                s->snapshots[i].l1_size * sizeof(uint64_t))) {
                return QCOW2_OL_INACTIVE_L1;
            }
        }
    }

    if ((chk & QCOW2_OL_ACTIVE_L2) && s->l1_table) {
        for (i = 0; i < s->l1_size; i++) {
            if ((s->l1_table[i] & L1E_OFFSET_MASK) &&
                overlaps_with(s->l1_table[i] & L1E_OFFSET_MASK,
                s->cluster_size)) {
                return QCOW2_OL_ACTIVE_L2;
            }
        }
    }

    if ((chk & QCOW2_OL_REFCOUNT_BLOCK) && s->refcount_table) {
        for (i = 0; i < s->refcount_table_size; i++) {
            if ((s->refcount_table[i] & REFT_OFFSET_MASK) &&
                overlaps_with(s->refcount_table[i] & REFT_OFFSET_MASK,
                s->cluster_size)) {
                return QCOW2_OL_REFCOUNT_BLOCK;
            }
        }
    }

    if ((chk & QCOW2_OL_INACTIVE_L2) && s->snapshots) {
        for (i = 0; i < s->nb_snapshots; i++) {
            uint64_t l1_ofs = s->snapshots[i].l1_table_offset;
            uint32_t l1_sz  = s->snapshots[i].l1_size;
            uint64_t l1_sz2 = l1_sz * sizeof(uint64_t);
            uint64_t *l1 = g_try_malloc(l1_sz2);
            int ret;

            if (l1_sz2 && l1 == NULL) {
                return -ENOMEM;
            }

            ret = bdrv_pread(bs->file, l1_ofs, l1, l1_sz2);
            if (ret < 0) {
                g_free(l1);
                return ret;
            }

            for (j = 0; j < l1_sz; j++) {
                uint64_t l2_ofs = be64_to_cpu(l1[j]) & L1E_OFFSET_MASK;
                if (l2_ofs && overlaps_with(l2_ofs, s->cluster_size)) {
                    g_free(l1);
                    return QCOW2_OL_INACTIVE_L2;
                }
            }

            g_free(l1);
        }
    }

    return 0;
}

static const char *metadata_ol_names[] = {
    [QCOW2_OL_MAIN_HEADER_BITNR]    = "qcow2_header",
    [QCOW2_OL_ACTIVE_L1_BITNR]      = "active L1 table",
    [QCOW2_OL_ACTIVE_L2_BITNR]      = "active L2 table",
    [QCOW2_OL_REFCOUNT_TABLE_BITNR] = "refcount table",
    [QCOW2_OL_REFCOUNT_BLOCK_BITNR] = "refcount block",
    [QCOW2_OL_SNAPSHOT_TABLE_BITNR] = "snapshot table",
    [QCOW2_OL_INACTIVE_L1_BITNR]    = "inactive L1 table",
    [QCOW2_OL_INACTIVE_L2_BITNR]    = "inactive L2 table",
};

/*
 * First performs a check for metadata overlaps (through
 * qcow2_check_metadata_overlap); if that fails with a negative value (error
 * while performing a check), that value is returned. If an impending overlap
 * is detected, the BDS will be made unusable, the qcow2 file marked corrupt
 * and -EIO returned.
 *
 * Returns 0 if there were neither overlaps nor errors while checking for
 * overlaps; or a negative value (-errno) on error.
 */
int qcow2_pre_write_overlap_check(BlockDriverState *bs, int ign, int64_t offset,
                                  int64_t size)
{
    int ret = qcow2_check_metadata_overlap(bs, ign, offset, size);

    if (ret < 0) {
        return ret;
    } else if (ret > 0) {
        int metadata_ol_bitnr = ctz32(ret);
        assert(metadata_ol_bitnr < QCOW2_OL_MAX_BITNR);

        qcow2_signal_corruption(bs, true, offset, size, "Preventing invalid "
                                "write on metadata (overlaps with %s)",
                                metadata_ol_names[metadata_ol_bitnr]);
        return -EIO;
    }

    return 0;
}