xref: /openbmc/qemu/docs/interop/qcow2.txt (revision ac6dd31e)
1== General ==
2
3A qcow2 image file is organized in units of constant size, which are called
4(host) clusters. A cluster is the unit in which all allocations are done,
5both for actual guest data and for image metadata.
6
7Likewise, the virtual disk as seen by the guest is divided into (guest)
8clusters of the same size.
9
10All numbers in qcow2 are stored in Big Endian byte order.
11
12
13== Header ==
14
15The first cluster of a qcow2 image contains the file header:
16
17    Byte  0 -  3:   magic
18                    QCOW magic string ("QFI\xfb")
19
20          4 -  7:   version
21                    Version number (valid values are 2 and 3)
22
23          8 - 15:   backing_file_offset
24                    Offset into the image file at which the backing file name
25                    is stored (NB: The string is not null terminated). 0 if the
26                    image doesn't have a backing file.
27
28         16 - 19:   backing_file_size
29                    Length of the backing file name in bytes. Must not be
30                    longer than 1023 bytes. Undefined if the image doesn't have
31                    a backing file.
32
33         20 - 23:   cluster_bits
34                    Number of bits that are used for addressing an offset
35                    within a cluster (1 << cluster_bits is the cluster size).
36                    Must not be less than 9 (i.e. 512 byte clusters).
37
38                    Note: qemu as of today has an implementation limit of 2 MB
39                    as the maximum cluster size and won't be able to open images
40                    with larger cluster sizes.
41
42         24 - 31:   size
43                    Virtual disk size in bytes.
44
45                    Note: qemu has an implementation limit of 32 MB as
46                    the maximum L1 table size.  With a 2 MB cluster
47                    size, it is unable to populate a virtual cluster
48                    beyond 2 EB (61 bits); with a 512 byte cluster
49                    size, it is unable to populate a virtual size
50                    larger than 128 GB (37 bits).  Meanwhile, L1/L2
51                    table layouts limit an image to no more than 64 PB
52                    (56 bits) of populated clusters, and an image may
53                    hit other limits first (such as a file system's
54                    maximum size).
55
56         32 - 35:   crypt_method
57                    0 for no encryption
58                    1 for AES encryption
59                    2 for LUKS encryption
60
61         36 - 39:   l1_size
62                    Number of entries in the active L1 table
63
64         40 - 47:   l1_table_offset
65                    Offset into the image file at which the active L1 table
66                    starts. Must be aligned to a cluster boundary.
67
68         48 - 55:   refcount_table_offset
69                    Offset into the image file at which the refcount table
70                    starts. Must be aligned to a cluster boundary.
71
72         56 - 59:   refcount_table_clusters
73                    Number of clusters that the refcount table occupies
74
75         60 - 63:   nb_snapshots
76                    Number of snapshots contained in the image
77
78         64 - 71:   snapshots_offset
79                    Offset into the image file at which the snapshot table
80                    starts. Must be aligned to a cluster boundary.
81
82If the version is 3 or higher, the header has the following additional fields.
83For version 2, the values are assumed to be zero, unless specified otherwise
84in the description of a field.
85
86         72 -  79:  incompatible_features
87                    Bitmask of incompatible features. An implementation must
88                    fail to open an image if an unknown bit is set.
89
90                    Bit 0:      Dirty bit.  If this bit is set then refcounts
91                                may be inconsistent, make sure to scan L1/L2
92                                tables to repair refcounts before accessing the
93                                image.
94
95                    Bit 1:      Corrupt bit.  If this bit is set then any data
96                                structure may be corrupt and the image must not
97                                be written to (unless for regaining
98                                consistency).
99
100                    Bits 2-63:  Reserved (set to 0)
101
102         80 -  87:  compatible_features
103                    Bitmask of compatible features. An implementation can
104                    safely ignore any unknown bits that are set.
105
106                    Bit 0:      Lazy refcounts bit.  If this bit is set then
107                                lazy refcount updates can be used.  This means
108                                marking the image file dirty and postponing
109                                refcount metadata updates.
110
111                    Bits 1-63:  Reserved (set to 0)
112
113         88 -  95:  autoclear_features
114                    Bitmask of auto-clear features. An implementation may only
115                    write to an image with unknown auto-clear features if it
116                    clears the respective bits from this field first.
117
118                    Bit 0:      Bitmaps extension bit
119                                This bit indicates consistency for the bitmaps
120                                extension data.
121
122                                It is an error if this bit is set without the
123                                bitmaps extension present.
124
125                                If the bitmaps extension is present but this
126                                bit is unset, the bitmaps extension data must be
127                                considered inconsistent.
128
129                    Bits 1-63:  Reserved (set to 0)
130
131         96 -  99:  refcount_order
132                    Describes the width of a reference count block entry (width
133                    in bits: refcount_bits = 1 << refcount_order). For version 2
134                    images, the order is always assumed to be 4
135                    (i.e. refcount_bits = 16).
136                    This value may not exceed 6 (i.e. refcount_bits = 64).
137
138        100 - 103:  header_length
139                    Length of the header structure in bytes. For version 2
140                    images, the length is always assumed to be 72 bytes.
141
142Directly after the image header, optional sections called header extensions can
143be stored. Each extension has a structure like the following:
144
145    Byte  0 -  3:   Header extension type:
146                        0x00000000 - End of the header extension area
147                        0xE2792ACA - Backing file format name
148                        0x6803f857 - Feature name table
149                        0x23852875 - Bitmaps extension
150                        0x0537be77 - Full disk encryption header pointer
151                        other      - Unknown header extension, can be safely
152                                     ignored
153
154          4 -  7:   Length of the header extension data
155
156          8 -  n:   Header extension data
157
158          n -  m:   Padding to round up the header extension size to the next
159                    multiple of 8.
160
161Unless stated otherwise, each header extension type shall appear at most once
162in the same image.
163
164If the image has a backing file then the backing file name should be stored in
165the remaining space between the end of the header extension area and the end of
166the first cluster. It is not allowed to store other data here, so that an
167implementation can safely modify the header and add extensions without harming
168data of compatible features that it doesn't support. Compatible features that
169need space for additional data can use a header extension.
170
171
172== Feature name table ==
173
174The feature name table is an optional header extension that contains the name
175for features used by the image. It can be used by applications that don't know
176the respective feature (e.g. because the feature was introduced only later) to
177display a useful error message.
178
179The number of entries in the feature name table is determined by the length of
180the header extension data. Each entry look like this:
181
182    Byte       0:   Type of feature (select feature bitmap)
183                        0: Incompatible feature
184                        1: Compatible feature
185                        2: Autoclear feature
186
187               1:   Bit number within the selected feature bitmap (valid
188                    values: 0-63)
189
190          2 - 47:   Feature name (padded with zeros, but not necessarily null
191                    terminated if it has full length)
192
193
194== Bitmaps extension ==
195
196The bitmaps extension is an optional header extension. It provides the ability
197to store bitmaps related to a virtual disk. For now, there is only one bitmap
198type: the dirty tracking bitmap, which tracks virtual disk changes from some
199point in time.
200
201The data of the extension should be considered consistent only if the
202corresponding auto-clear feature bit is set, see autoclear_features above.
203
204The fields of the bitmaps extension are:
205
206    Byte  0 -  3:  nb_bitmaps
207                   The number of bitmaps contained in the image. Must be
208                   greater than or equal to 1.
209
210                   Note: Qemu currently only supports up to 65535 bitmaps per
211                   image.
212
213          4 -  7:  Reserved, must be zero.
214
215          8 - 15:  bitmap_directory_size
216                   Size of the bitmap directory in bytes. It is the cumulative
217                   size of all (nb_bitmaps) bitmap directory entries.
218
219         16 - 23:  bitmap_directory_offset
220                   Offset into the image file at which the bitmap directory
221                   starts. Must be aligned to a cluster boundary.
222
223== Full disk encryption header pointer ==
224
225The full disk encryption header must be present if, and only if, the
226'crypt_method' header requires metadata. Currently this is only true
227of the 'LUKS' crypt method. The header extension must be absent for
228other methods.
229
230This header provides the offset at which the crypt method can store
231its additional data, as well as the length of such data.
232
233    Byte  0 -  7:   Offset into the image file at which the encryption
234                    header starts in bytes. Must be aligned to a cluster
235                    boundary.
236    Byte  8 - 15:   Length of the written encryption header in bytes.
237                    Note actual space allocated in the qcow2 file may
238                    be larger than this value, since it will be rounded
239                    to the nearest multiple of the cluster size. Any
240                    unused bytes in the allocated space will be initialized
241                    to 0.
242
243For the LUKS crypt method, the encryption header works as follows.
244
245The first 592 bytes of the header clusters will contain the LUKS
246partition header. This is then followed by the key material data areas.
247The size of the key material data areas is determined by the number of
248stripes in the key slot and key size. Refer to the LUKS format
249specification ('docs/on-disk-format.pdf' in the cryptsetup source
250package) for details of the LUKS partition header format.
251
252In the LUKS partition header, the "payload-offset" field will be
253calculated as normal for the LUKS spec. ie the size of the LUKS
254header, plus key material regions, plus padding, relative to the
255start of the LUKS header. This offset value is not required to be
256qcow2 cluster aligned. Its value is currently never used in the
257context of qcow2, since the qcow2 file format itself defines where
258the real payload offset is, but none the less a valid payload offset
259should always be present.
260
261In the LUKS key slots header, the "key-material-offset" is relative
262to the start of the LUKS header clusters in the qcow2 container,
263not the start of the qcow2 file.
264
265Logically the layout looks like
266
267  +-----------------------------+
268  | QCow2 header                |
269  | QCow2 header extension X    |
270  | QCow2 header extension FDE  |
271  | QCow2 header extension ...  |
272  | QCow2 header extension Z    |
273  +-----------------------------+
274  | ....other QCow2 tables....  |
275  .                             .
276  .                             .
277  +-----------------------------+
278  | +-------------------------+ |
279  | | LUKS partition header   | |
280  | +-------------------------+ |
281  | | LUKS key material 1     | |
282  | +-------------------------+ |
283  | | LUKS key material 2     | |
284  | +-------------------------+ |
285  | | LUKS key material ...   | |
286  | +-------------------------+ |
287  | | LUKS key material 8     | |
288  | +-------------------------+ |
289  +-----------------------------+
290  | QCow2 cluster payload       |
291  .                             .
292  .                             .
293  .                             .
294  |                             |
295  +-----------------------------+
296
297== Data encryption ==
298
299When an encryption method is requested in the header, the image payload
300data must be encrypted/decrypted on every write/read. The image headers
301and metadata are never encrypted.
302
303The algorithms used for encryption vary depending on the method
304
305 - AES:
306
307   The AES cipher, in CBC mode, with 256 bit keys.
308
309   Initialization vectors generated using plain64 method, with
310   the virtual disk sector as the input tweak.
311
312   This format is no longer supported in QEMU system emulators, due
313   to a number of design flaws affecting its security. It is only
314   supported in the command line tools for the sake of back compatibility
315   and data liberation.
316
317 - LUKS:
318
319   The algorithms are specified in the LUKS header.
320
321   Initialization vectors generated using the method specified
322   in the LUKS header, with the physical disk sector as the
323   input tweak.
324
325== Host cluster management ==
326
327qcow2 manages the allocation of host clusters by maintaining a reference count
328for each host cluster. A refcount of 0 means that the cluster is free, 1 means
329that it is used, and >= 2 means that it is used and any write access must
330perform a COW (copy on write) operation.
331
332The refcounts are managed in a two-level table. The first level is called
333refcount table and has a variable size (which is stored in the header). The
334refcount table can cover multiple clusters, however it needs to be contiguous
335in the image file.
336
337It contains pointers to the second level structures which are called refcount
338blocks and are exactly one cluster in size.
339
340Although a large enough refcount table can reserve clusters past 64 PB
341(56 bits) (assuming the underlying protocol can even be sized that
342large), note that some qcow2 metadata such as L1/L2 tables must point
343to clusters prior to that point.
344
345Note: qemu has an implementation limit of 8 MB as the maximum refcount
346table size.  With a 2 MB cluster size and a default refcount_order of
3474, it is unable to reference host resources beyond 2 EB (61 bits); in
348the worst case, with a 512 cluster size and refcount_order of 6, it is
349unable to access beyond 32 GB (35 bits).
350
351Given an offset into the image file, the refcount of its cluster can be
352obtained as follows:
353
354    refcount_block_entries = (cluster_size * 8 / refcount_bits)
355
356    refcount_block_index = (offset / cluster_size) % refcount_block_entries
357    refcount_table_index = (offset / cluster_size) / refcount_block_entries
358
359    refcount_block = load_cluster(refcount_table[refcount_table_index]);
360    return refcount_block[refcount_block_index];
361
362Refcount table entry:
363
364    Bit  0 -  8:    Reserved (set to 0)
365
366         9 - 63:    Bits 9-63 of the offset into the image file at which the
367                    refcount block starts. Must be aligned to a cluster
368                    boundary.
369
370                    If this is 0, the corresponding refcount block has not yet
371                    been allocated. All refcounts managed by this refcount block
372                    are 0.
373
374Refcount block entry (x = refcount_bits - 1):
375
376    Bit  0 -  x:    Reference count of the cluster. If refcount_bits implies a
377                    sub-byte width, note that bit 0 means the least significant
378                    bit in this context.
379
380
381== Cluster mapping ==
382
383Just as for refcounts, qcow2 uses a two-level structure for the mapping of
384guest clusters to host clusters. They are called L1 and L2 table.
385
386The L1 table has a variable size (stored in the header) and may use multiple
387clusters, however it must be contiguous in the image file. L2 tables are
388exactly one cluster in size.
389
390The L1 and L2 tables have implications on the maximum virtual file
391size; for a given L1 table size, a larger cluster size is required for
392the guest to have access to more space.  Furthermore, a virtual
393cluster must currently map to a host offset below 64 PB (56 bits)
394(although this limit could be relaxed by putting reserved bits into
395use).  Additionally, as cluster size increases, the maximum host
396offset for a compressed cluster is reduced (a 2M cluster size requires
397compressed clusters to reside below 512 TB (49 bits), and this limit
398cannot be relaxed without an incompatible layout change).
399
400Given an offset into the virtual disk, the offset into the image file can be
401obtained as follows:
402
403    l2_entries = (cluster_size / sizeof(uint64_t))
404
405    l2_index = (offset / cluster_size) % l2_entries
406    l1_index = (offset / cluster_size) / l2_entries
407
408    l2_table = load_cluster(l1_table[l1_index]);
409    cluster_offset = l2_table[l2_index];
410
411    return cluster_offset + (offset % cluster_size)
412
413L1 table entry:
414
415    Bit  0 -  8:    Reserved (set to 0)
416
417         9 - 55:    Bits 9-55 of the offset into the image file at which the L2
418                    table starts. Must be aligned to a cluster boundary. If the
419                    offset is 0, the L2 table and all clusters described by this
420                    L2 table are unallocated.
421
422        56 - 62:    Reserved (set to 0)
423
424             63:    0 for an L2 table that is unused or requires COW, 1 if its
425                    refcount is exactly one. This information is only accurate
426                    in the active L1 table.
427
428L2 table entry:
429
430    Bit  0 -  61:   Cluster descriptor
431
432              62:   0 for standard clusters
433                    1 for compressed clusters
434
435              63:   0 for clusters that are unused, compressed or require COW.
436                    1 for standard clusters whose refcount is exactly one.
437                    This information is only accurate in L2 tables
438                    that are reachable from the active L1 table.
439
440Standard Cluster Descriptor:
441
442    Bit       0:    If set to 1, the cluster reads as all zeros. The host
443                    cluster offset can be used to describe a preallocation,
444                    but it won't be used for reading data from this cluster,
445                    nor is data read from the backing file if the cluster is
446                    unallocated.
447
448                    With version 2, this is always 0.
449
450         1 -  8:    Reserved (set to 0)
451
452         9 - 55:    Bits 9-55 of host cluster offset. Must be aligned to a
453                    cluster boundary. If the offset is 0, the cluster is
454                    unallocated.
455
456        56 - 61:    Reserved (set to 0)
457
458
459Compressed Clusters Descriptor (x = 62 - (cluster_bits - 8)):
460
461    Bit  0 - x-1:   Host cluster offset. This is usually _not_ aligned to a
462                    cluster or sector boundary!  If cluster_bits is
463                    small enough that this field includes bits beyond
464                    55, those upper bits must be set to 0.
465
466         x - 61:    Number of additional 512-byte sectors used for the
467                    compressed data, beyond the sector containing the offset
468                    in the previous field. Some of these sectors may reside
469                    in the next contiguous host cluster.
470
471                    Note that the compressed data does not necessarily occupy
472                    all of the bytes in the final sector; rather, decompression
473                    stops when it has produced a cluster of data.
474
475                    Another compressed cluster may map to the tail of the final
476                    sector used by this compressed cluster.
477
478If a cluster is unallocated, read requests shall read the data from the backing
479file (except if bit 0 in the Standard Cluster Descriptor is set). If there is
480no backing file or the backing file is smaller than the image, they shall read
481zeros for all parts that are not covered by the backing file.
482
483
484== Snapshots ==
485
486qcow2 supports internal snapshots. Their basic principle of operation is to
487switch the active L1 table, so that a different set of host clusters are
488exposed to the guest.
489
490When creating a snapshot, the L1 table should be copied and the refcount of all
491L2 tables and clusters reachable from this L1 table must be increased, so that
492a write causes a COW and isn't visible in other snapshots.
493
494When loading a snapshot, bit 63 of all entries in the new active L1 table and
495all L2 tables referenced by it must be reconstructed from the refcount table
496as it doesn't need to be accurate in inactive L1 tables.
497
498A directory of all snapshots is stored in the snapshot table, a contiguous area
499in the image file, whose starting offset and length are given by the header
500fields snapshots_offset and nb_snapshots. The entries of the snapshot table
501have variable length, depending on the length of ID, name and extra data.
502
503Snapshot table entry:
504
505    Byte 0 -  7:    Offset into the image file at which the L1 table for the
506                    snapshot starts. Must be aligned to a cluster boundary.
507
508         8 - 11:    Number of entries in the L1 table of the snapshots
509
510        12 - 13:    Length of the unique ID string describing the snapshot
511
512        14 - 15:    Length of the name of the snapshot
513
514        16 - 19:    Time at which the snapshot was taken in seconds since the
515                    Epoch
516
517        20 - 23:    Subsecond part of the time at which the snapshot was taken
518                    in nanoseconds
519
520        24 - 31:    Time that the guest was running until the snapshot was
521                    taken in nanoseconds
522
523        32 - 35:    Size of the VM state in bytes. 0 if no VM state is saved.
524                    If there is VM state, it starts at the first cluster
525                    described by first L1 table entry that doesn't describe a
526                    regular guest cluster (i.e. VM state is stored like guest
527                    disk content, except that it is stored at offsets that are
528                    larger than the virtual disk presented to the guest)
529
530        36 - 39:    Size of extra data in the table entry (used for future
531                    extensions of the format)
532
533        variable:   Extra data for future extensions. Unknown fields must be
534                    ignored. Currently defined are (offset relative to snapshot
535                    table entry):
536
537                    Byte 40 - 47:   Size of the VM state in bytes. 0 if no VM
538                                    state is saved. If this field is present,
539                                    the 32-bit value in bytes 32-35 is ignored.
540
541                    Byte 48 - 55:   Virtual disk size of the snapshot in bytes
542
543                    Version 3 images must include extra data at least up to
544                    byte 55.
545
546        variable:   Unique ID string for the snapshot (not null terminated)
547
548        variable:   Name of the snapshot (not null terminated)
549
550        variable:   Padding to round up the snapshot table entry size to the
551                    next multiple of 8.
552
553
554== Bitmaps ==
555
556As mentioned above, the bitmaps extension provides the ability to store bitmaps
557related to a virtual disk. This section describes how these bitmaps are stored.
558
559All stored bitmaps are related to the virtual disk stored in the same image, so
560each bitmap size is equal to the virtual disk size.
561
562Each bit of the bitmap is responsible for strictly defined range of the virtual
563disk. For bit number bit_nr the corresponding range (in bytes) will be:
564
565    [bit_nr * bitmap_granularity .. (bit_nr + 1) * bitmap_granularity - 1]
566
567Granularity is a property of the concrete bitmap, see below.
568
569
570=== Bitmap directory ===
571
572Each bitmap saved in the image is described in a bitmap directory entry. The
573bitmap directory is a contiguous area in the image file, whose starting offset
574and length are given by the header extension fields bitmap_directory_offset and
575bitmap_directory_size. The entries of the bitmap directory have variable
576length, depending on the lengths of the bitmap name and extra data.
577
578Structure of a bitmap directory entry:
579
580    Byte 0 -  7:    bitmap_table_offset
581                    Offset into the image file at which the bitmap table
582                    (described below) for the bitmap starts. Must be aligned to
583                    a cluster boundary.
584
585         8 - 11:    bitmap_table_size
586                    Number of entries in the bitmap table of the bitmap.
587
588        12 - 15:    flags
589                    Bit
590                      0: in_use
591                         The bitmap was not saved correctly and may be
592                         inconsistent.
593
594                      1: auto
595                         The bitmap must reflect all changes of the virtual
596                         disk by any application that would write to this qcow2
597                         file (including writes, snapshot switching, etc.). The
598                         type of this bitmap must be 'dirty tracking bitmap'.
599
600                      2: extra_data_compatible
601                         This flags is meaningful when the extra data is
602                         unknown to the software (currently any extra data is
603                         unknown to Qemu).
604                         If it is set, the bitmap may be used as expected, extra
605                         data must be left as is.
606                         If it is not set, the bitmap must not be used, but
607                         both it and its extra data be left as is.
608
609                    Bits 3 - 31 are reserved and must be 0.
610
611             16:    type
612                    This field describes the sort of the bitmap.
613                    Values:
614                      1: Dirty tracking bitmap
615
616                    Values 0, 2 - 255 are reserved.
617
618             17:    granularity_bits
619                    Granularity bits. Valid values: 0 - 63.
620
621                    Note: Qemu currently supports only values 9 - 31.
622
623                    Granularity is calculated as
624                        granularity = 1 << granularity_bits
625
626                    A bitmap's granularity is how many bytes of the image
627                    accounts for one bit of the bitmap.
628
629        18 - 19:    name_size
630                    Size of the bitmap name. Must be non-zero.
631
632                    Note: Qemu currently doesn't support values greater than
633                    1023.
634
635        20 - 23:    extra_data_size
636                    Size of type-specific extra data.
637
638                    For now, as no extra data is defined, extra_data_size is
639                    reserved and should be zero. If it is non-zero the
640                    behavior is defined by extra_data_compatible flag.
641
642        variable:   extra_data
643                    Extra data for the bitmap, occupying extra_data_size bytes.
644                    Extra data must never contain references to clusters or in
645                    some other way allocate additional clusters.
646
647        variable:   name
648                    The name of the bitmap (not null terminated), occupying
649                    name_size bytes. Must be unique among all bitmap names
650                    within the bitmaps extension.
651
652        variable:   Padding to round up the bitmap directory entry size to the
653                    next multiple of 8. All bytes of the padding must be zero.
654
655
656=== Bitmap table ===
657
658Each bitmap is stored using a one-level structure (as opposed to two-level
659structures like for refcounts and guest clusters mapping) for the mapping of
660bitmap data to host clusters. This structure is called the bitmap table.
661
662Each bitmap table has a variable size (stored in the bitmap directory entry)
663and may use multiple clusters, however, it must be contiguous in the image
664file.
665
666Structure of a bitmap table entry:
667
668    Bit       0:    Reserved and must be zero if bits 9 - 55 are non-zero.
669                    If bits 9 - 55 are zero:
670                      0: Cluster should be read as all zeros.
671                      1: Cluster should be read as all ones.
672
673         1 -  8:    Reserved and must be zero.
674
675         9 - 55:    Bits 9 - 55 of the host cluster offset. Must be aligned to
676                    a cluster boundary. If the offset is 0, the cluster is
677                    unallocated; in that case, bit 0 determines how this
678                    cluster should be treated during reads.
679
680        56 - 63:    Reserved and must be zero.
681
682
683=== Bitmap data ===
684
685As noted above, bitmap data is stored in separate clusters, described by the
686bitmap table. Given an offset (in bytes) into the bitmap data, the offset into
687the image file can be obtained as follows:
688
689    image_offset(bitmap_data_offset) =
690        bitmap_table[bitmap_data_offset / cluster_size] +
691            (bitmap_data_offset % cluster_size)
692
693This offset is not defined if bits 9 - 55 of bitmap table entry are zero (see
694above).
695
696Given an offset byte_nr into the virtual disk and the bitmap's granularity, the
697bit offset into the image file to the corresponding bit of the bitmap can be
698calculated like this:
699
700    bit_offset(byte_nr) =
701        image_offset(byte_nr / granularity / 8) * 8 +
702            (byte_nr / granularity) % 8
703
704If the size of the bitmap data is not a multiple of the cluster size then the
705last cluster of the bitmap data contains some unused tail bits. These bits must
706be zero.
707
708
709=== Dirty tracking bitmaps ===
710
711Bitmaps with 'type' field equal to one are dirty tracking bitmaps.
712
713When the virtual disk is in use dirty tracking bitmap may be 'enabled' or
714'disabled'. While the bitmap is 'enabled', all writes to the virtual disk
715should be reflected in the bitmap. A set bit in the bitmap means that the
716corresponding range of the virtual disk (see above) was written to while the
717bitmap was 'enabled'. An unset bit means that this range was not written to.
718
719The software doesn't have to sync the bitmap in the image file with its
720representation in RAM after each write. Flag 'in_use' should be set while the
721bitmap is not synced.
722
723In the image file the 'enabled' state is reflected by the 'auto' flag. If this
724flag is set, the software must consider the bitmap as 'enabled' and start
725tracking virtual disk changes to this bitmap from the first write to the
726virtual disk. If this flag is not set then the bitmap is disabled.
727