xref: /openbmc/qemu/docs/interop/qcow2.txt (revision 88cd34ee)
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
82For version 2, the header is exactly 72 bytes in length, and finishes here.
83For version 3 or higher, the header length is at least 104 bytes, including
84the next fields through header_length.
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                    Bit 2:      External data file bit.  If this bit is set, an
101                                external data file is used. Guest clusters are
102                                then stored in the external data file. For such
103                                images, clusters in the external data file are
104                                not refcounted. The offset field in the
105                                Standard Cluster Descriptor must match the
106                                guest offset and neither compressed clusters
107                                nor internal snapshots are supported.
108
109                                An External Data File Name header extension may
110                                be present if this bit is set.
111
112                    Bit 3:      Compression type bit.  If this bit is set,
113                                a non-default compression is used for compressed
114                                clusters. The compression_type field must be
115                                present and not zero.
116
117                    Bits 4-63:  Reserved (set to 0)
118
119         80 -  87:  compatible_features
120                    Bitmask of compatible features. An implementation can
121                    safely ignore any unknown bits that are set.
122
123                    Bit 0:      Lazy refcounts bit.  If this bit is set then
124                                lazy refcount updates can be used.  This means
125                                marking the image file dirty and postponing
126                                refcount metadata updates.
127
128                    Bits 1-63:  Reserved (set to 0)
129
130         88 -  95:  autoclear_features
131                    Bitmask of auto-clear features. An implementation may only
132                    write to an image with unknown auto-clear features if it
133                    clears the respective bits from this field first.
134
135                    Bit 0:      Bitmaps extension bit
136                                This bit indicates consistency for the bitmaps
137                                extension data.
138
139                                It is an error if this bit is set without the
140                                bitmaps extension present.
141
142                                If the bitmaps extension is present but this
143                                bit is unset, the bitmaps extension data must be
144                                considered inconsistent.
145
146                    Bit 1:      If this bit is set, the external data file can
147                                be read as a consistent standalone raw image
148                                without looking at the qcow2 metadata.
149
150                                Setting this bit has a performance impact for
151                                some operations on the image (e.g. writing
152                                zeros requires writing to the data file instead
153                                of only setting the zero flag in the L2 table
154                                entry) and conflicts with backing files.
155
156                                This bit may only be set if the External Data
157                                File bit (incompatible feature bit 1) is also
158                                set.
159
160                    Bits 2-63:  Reserved (set to 0)
161
162         96 -  99:  refcount_order
163                    Describes the width of a reference count block entry (width
164                    in bits: refcount_bits = 1 << refcount_order). For version 2
165                    images, the order is always assumed to be 4
166                    (i.e. refcount_bits = 16).
167                    This value may not exceed 6 (i.e. refcount_bits = 64).
168
169        100 - 103:  header_length
170                    Length of the header structure in bytes. For version 2
171                    images, the length is always assumed to be 72 bytes.
172                    For version 3 it's at least 104 bytes and must be a multiple
173                    of 8.
174
175
176=== Additional fields (version 3 and higher) ===
177
178In general, these fields are optional and may be safely ignored by the software,
179as well as filled by zeros (which is equal to field absence), if software needs
180to set field B, but does not care about field A which precedes B. More
181formally, additional fields have the following compatibility rules:
182
1831. If the value of the additional field must not be ignored for correct
184handling of the file, it will be accompanied by a corresponding incompatible
185feature bit.
186
1872. If there are no unrecognized incompatible feature bits set, an unknown
188additional field may be safely ignored other than preserving its value when
189rewriting the image header.
190
1913. An explicit value of 0 will have the same behavior as when the field is not
192present*, if not altered by a specific incompatible bit.
193
194*. A field is considered not present when header_length is less than or equal
195to the field's offset. Also, all additional fields are not present for
196version 2.
197
198              104:  compression_type
199
200                    Defines the compression method used for compressed clusters.
201                    All compressed clusters in an image use the same compression
202                    type.
203
204                    If the incompatible bit "Compression type" is set: the field
205                    must be present and non-zero (which means non-zlib
206                    compression type). Otherwise, this field must not be present
207                    or must be zero (which means zlib).
208
209                    Available compression type values:
210                        0: zlib <https://www.zlib.net/>
211
212
213=== Header padding ===
214
215@header_length must be a multiple of 8, which means that if the end of the last
216additional field is not aligned, some padding is needed. This padding must be
217zeroed, so that if some existing (or future) additional field will fall into
218the padding, it will be interpreted accordingly to point [3.] of the previous
219paragraph, i.e.  in the same manner as when this field is not present.
220
221
222=== Header extensions ===
223
224Directly after the image header, optional sections called header extensions can
225be stored. Each extension has a structure like the following:
226
227    Byte  0 -  3:   Header extension type:
228                        0x00000000 - End of the header extension area
229                        0xE2792ACA - Backing file format name string
230                        0x6803f857 - Feature name table
231                        0x23852875 - Bitmaps extension
232                        0x0537be77 - Full disk encryption header pointer
233                        0x44415441 - External data file name string
234                        other      - Unknown header extension, can be safely
235                                     ignored
236
237          4 -  7:   Length of the header extension data
238
239          8 -  n:   Header extension data
240
241          n -  m:   Padding to round up the header extension size to the next
242                    multiple of 8.
243
244Unless stated otherwise, each header extension type shall appear at most once
245in the same image.
246
247If the image has a backing file then the backing file name should be stored in
248the remaining space between the end of the header extension area and the end of
249the first cluster. It is not allowed to store other data here, so that an
250implementation can safely modify the header and add extensions without harming
251data of compatible features that it doesn't support. Compatible features that
252need space for additional data can use a header extension.
253
254
255== String header extensions ==
256
257Some header extensions (such as the backing file format name and the external
258data file name) are just a single string. In this case, the header extension
259length is the string length and the string is not '\0' terminated. (The header
260extension padding can make it look like a string is '\0' terminated, but
261neither is padding always necessary nor is there a guarantee that zero bytes
262are used for padding.)
263
264
265== Feature name table ==
266
267The feature name table is an optional header extension that contains the name
268for features used by the image. It can be used by applications that don't know
269the respective feature (e.g. because the feature was introduced only later) to
270display a useful error message.
271
272The number of entries in the feature name table is determined by the length of
273the header extension data. Each entry look like this:
274
275    Byte       0:   Type of feature (select feature bitmap)
276                        0: Incompatible feature
277                        1: Compatible feature
278                        2: Autoclear feature
279
280               1:   Bit number within the selected feature bitmap (valid
281                    values: 0-63)
282
283          2 - 47:   Feature name (padded with zeros, but not necessarily null
284                    terminated if it has full length)
285
286
287== Bitmaps extension ==
288
289The bitmaps extension is an optional header extension. It provides the ability
290to store bitmaps related to a virtual disk. For now, there is only one bitmap
291type: the dirty tracking bitmap, which tracks virtual disk changes from some
292point in time.
293
294The data of the extension should be considered consistent only if the
295corresponding auto-clear feature bit is set, see autoclear_features above.
296
297The fields of the bitmaps extension are:
298
299    Byte  0 -  3:  nb_bitmaps
300                   The number of bitmaps contained in the image. Must be
301                   greater than or equal to 1.
302
303                   Note: Qemu currently only supports up to 65535 bitmaps per
304                   image.
305
306          4 -  7:  Reserved, must be zero.
307
308          8 - 15:  bitmap_directory_size
309                   Size of the bitmap directory in bytes. It is the cumulative
310                   size of all (nb_bitmaps) bitmap directory entries.
311
312         16 - 23:  bitmap_directory_offset
313                   Offset into the image file at which the bitmap directory
314                   starts. Must be aligned to a cluster boundary.
315
316== Full disk encryption header pointer ==
317
318The full disk encryption header must be present if, and only if, the
319'crypt_method' header requires metadata. Currently this is only true
320of the 'LUKS' crypt method. The header extension must be absent for
321other methods.
322
323This header provides the offset at which the crypt method can store
324its additional data, as well as the length of such data.
325
326    Byte  0 -  7:   Offset into the image file at which the encryption
327                    header starts in bytes. Must be aligned to a cluster
328                    boundary.
329    Byte  8 - 15:   Length of the written encryption header in bytes.
330                    Note actual space allocated in the qcow2 file may
331                    be larger than this value, since it will be rounded
332                    to the nearest multiple of the cluster size. Any
333                    unused bytes in the allocated space will be initialized
334                    to 0.
335
336For the LUKS crypt method, the encryption header works as follows.
337
338The first 592 bytes of the header clusters will contain the LUKS
339partition header. This is then followed by the key material data areas.
340The size of the key material data areas is determined by the number of
341stripes in the key slot and key size. Refer to the LUKS format
342specification ('docs/on-disk-format.pdf' in the cryptsetup source
343package) for details of the LUKS partition header format.
344
345In the LUKS partition header, the "payload-offset" field will be
346calculated as normal for the LUKS spec. ie the size of the LUKS
347header, plus key material regions, plus padding, relative to the
348start of the LUKS header. This offset value is not required to be
349qcow2 cluster aligned. Its value is currently never used in the
350context of qcow2, since the qcow2 file format itself defines where
351the real payload offset is, but none the less a valid payload offset
352should always be present.
353
354In the LUKS key slots header, the "key-material-offset" is relative
355to the start of the LUKS header clusters in the qcow2 container,
356not the start of the qcow2 file.
357
358Logically the layout looks like
359
360  +-----------------------------+
361  | QCow2 header                |
362  | QCow2 header extension X    |
363  | QCow2 header extension FDE  |
364  | QCow2 header extension ...  |
365  | QCow2 header extension Z    |
366  +-----------------------------+
367  | ....other QCow2 tables....  |
368  .                             .
369  .                             .
370  +-----------------------------+
371  | +-------------------------+ |
372  | | LUKS partition header   | |
373  | +-------------------------+ |
374  | | LUKS key material 1     | |
375  | +-------------------------+ |
376  | | LUKS key material 2     | |
377  | +-------------------------+ |
378  | | LUKS key material ...   | |
379  | +-------------------------+ |
380  | | LUKS key material 8     | |
381  | +-------------------------+ |
382  +-----------------------------+
383  | QCow2 cluster payload       |
384  .                             .
385  .                             .
386  .                             .
387  |                             |
388  +-----------------------------+
389
390== Data encryption ==
391
392When an encryption method is requested in the header, the image payload
393data must be encrypted/decrypted on every write/read. The image headers
394and metadata are never encrypted.
395
396The algorithms used for encryption vary depending on the method
397
398 - AES:
399
400   The AES cipher, in CBC mode, with 256 bit keys.
401
402   Initialization vectors generated using plain64 method, with
403   the virtual disk sector as the input tweak.
404
405   This format is no longer supported in QEMU system emulators, due
406   to a number of design flaws affecting its security. It is only
407   supported in the command line tools for the sake of back compatibility
408   and data liberation.
409
410 - LUKS:
411
412   The algorithms are specified in the LUKS header.
413
414   Initialization vectors generated using the method specified
415   in the LUKS header, with the physical disk sector as the
416   input tweak.
417
418== Host cluster management ==
419
420qcow2 manages the allocation of host clusters by maintaining a reference count
421for each host cluster. A refcount of 0 means that the cluster is free, 1 means
422that it is used, and >= 2 means that it is used and any write access must
423perform a COW (copy on write) operation.
424
425The refcounts are managed in a two-level table. The first level is called
426refcount table and has a variable size (which is stored in the header). The
427refcount table can cover multiple clusters, however it needs to be contiguous
428in the image file.
429
430It contains pointers to the second level structures which are called refcount
431blocks and are exactly one cluster in size.
432
433Although a large enough refcount table can reserve clusters past 64 PB
434(56 bits) (assuming the underlying protocol can even be sized that
435large), note that some qcow2 metadata such as L1/L2 tables must point
436to clusters prior to that point.
437
438Note: qemu has an implementation limit of 8 MB as the maximum refcount
439table size.  With a 2 MB cluster size and a default refcount_order of
4404, it is unable to reference host resources beyond 2 EB (61 bits); in
441the worst case, with a 512 cluster size and refcount_order of 6, it is
442unable to access beyond 32 GB (35 bits).
443
444Given an offset into the image file, the refcount of its cluster can be
445obtained as follows:
446
447    refcount_block_entries = (cluster_size * 8 / refcount_bits)
448
449    refcount_block_index = (offset / cluster_size) % refcount_block_entries
450    refcount_table_index = (offset / cluster_size) / refcount_block_entries
451
452    refcount_block = load_cluster(refcount_table[refcount_table_index]);
453    return refcount_block[refcount_block_index];
454
455Refcount table entry:
456
457    Bit  0 -  8:    Reserved (set to 0)
458
459         9 - 63:    Bits 9-63 of the offset into the image file at which the
460                    refcount block starts. Must be aligned to a cluster
461                    boundary.
462
463                    If this is 0, the corresponding refcount block has not yet
464                    been allocated. All refcounts managed by this refcount block
465                    are 0.
466
467Refcount block entry (x = refcount_bits - 1):
468
469    Bit  0 -  x:    Reference count of the cluster. If refcount_bits implies a
470                    sub-byte width, note that bit 0 means the least significant
471                    bit in this context.
472
473
474== Cluster mapping ==
475
476Just as for refcounts, qcow2 uses a two-level structure for the mapping of
477guest clusters to host clusters. They are called L1 and L2 table.
478
479The L1 table has a variable size (stored in the header) and may use multiple
480clusters, however it must be contiguous in the image file. L2 tables are
481exactly one cluster in size.
482
483The L1 and L2 tables have implications on the maximum virtual file
484size; for a given L1 table size, a larger cluster size is required for
485the guest to have access to more space.  Furthermore, a virtual
486cluster must currently map to a host offset below 64 PB (56 bits)
487(although this limit could be relaxed by putting reserved bits into
488use).  Additionally, as cluster size increases, the maximum host
489offset for a compressed cluster is reduced (a 2M cluster size requires
490compressed clusters to reside below 512 TB (49 bits), and this limit
491cannot be relaxed without an incompatible layout change).
492
493Given an offset into the virtual disk, the offset into the image file can be
494obtained as follows:
495
496    l2_entries = (cluster_size / sizeof(uint64_t))
497
498    l2_index = (offset / cluster_size) % l2_entries
499    l1_index = (offset / cluster_size) / l2_entries
500
501    l2_table = load_cluster(l1_table[l1_index]);
502    cluster_offset = l2_table[l2_index];
503
504    return cluster_offset + (offset % cluster_size)
505
506L1 table entry:
507
508    Bit  0 -  8:    Reserved (set to 0)
509
510         9 - 55:    Bits 9-55 of the offset into the image file at which the L2
511                    table starts. Must be aligned to a cluster boundary. If the
512                    offset is 0, the L2 table and all clusters described by this
513                    L2 table are unallocated.
514
515        56 - 62:    Reserved (set to 0)
516
517             63:    0 for an L2 table that is unused or requires COW, 1 if its
518                    refcount is exactly one. This information is only accurate
519                    in the active L1 table.
520
521L2 table entry:
522
523    Bit  0 -  61:   Cluster descriptor
524
525              62:   0 for standard clusters
526                    1 for compressed clusters
527
528              63:   0 for clusters that are unused, compressed or require COW.
529                    1 for standard clusters whose refcount is exactly one.
530                    This information is only accurate in L2 tables
531                    that are reachable from the active L1 table.
532
533                    With external data files, all guest clusters have an
534                    implicit refcount of 1 (because of the fixed host = guest
535                    mapping for guest cluster offsets), so this bit should be 1
536                    for all allocated clusters.
537
538Standard Cluster Descriptor:
539
540    Bit       0:    If set to 1, the cluster reads as all zeros. The host
541                    cluster offset can be used to describe a preallocation,
542                    but it won't be used for reading data from this cluster,
543                    nor is data read from the backing file if the cluster is
544                    unallocated.
545
546                    With version 2, this is always 0.
547
548         1 -  8:    Reserved (set to 0)
549
550         9 - 55:    Bits 9-55 of host cluster offset. Must be aligned to a
551                    cluster boundary. If the offset is 0 and bit 63 is clear,
552                    the cluster is unallocated. The offset may only be 0 with
553                    bit 63 set (indicating a host cluster offset of 0) when an
554                    external data file is used.
555
556        56 - 61:    Reserved (set to 0)
557
558
559Compressed Clusters Descriptor (x = 62 - (cluster_bits - 8)):
560
561    Bit  0 - x-1:   Host cluster offset. This is usually _not_ aligned to a
562                    cluster or sector boundary!  If cluster_bits is
563                    small enough that this field includes bits beyond
564                    55, those upper bits must be set to 0.
565
566         x - 61:    Number of additional 512-byte sectors used for the
567                    compressed data, beyond the sector containing the offset
568                    in the previous field. Some of these sectors may reside
569                    in the next contiguous host cluster.
570
571                    Note that the compressed data does not necessarily occupy
572                    all of the bytes in the final sector; rather, decompression
573                    stops when it has produced a cluster of data.
574
575                    Another compressed cluster may map to the tail of the final
576                    sector used by this compressed cluster.
577
578If a cluster is unallocated, read requests shall read the data from the backing
579file (except if bit 0 in the Standard Cluster Descriptor is set). If there is
580no backing file or the backing file is smaller than the image, they shall read
581zeros for all parts that are not covered by the backing file.
582
583
584== Snapshots ==
585
586qcow2 supports internal snapshots. Their basic principle of operation is to
587switch the active L1 table, so that a different set of host clusters are
588exposed to the guest.
589
590When creating a snapshot, the L1 table should be copied and the refcount of all
591L2 tables and clusters reachable from this L1 table must be increased, so that
592a write causes a COW and isn't visible in other snapshots.
593
594When loading a snapshot, bit 63 of all entries in the new active L1 table and
595all L2 tables referenced by it must be reconstructed from the refcount table
596as it doesn't need to be accurate in inactive L1 tables.
597
598A directory of all snapshots is stored in the snapshot table, a contiguous area
599in the image file, whose starting offset and length are given by the header
600fields snapshots_offset and nb_snapshots. The entries of the snapshot table
601have variable length, depending on the length of ID, name and extra data.
602
603Snapshot table entry:
604
605    Byte 0 -  7:    Offset into the image file at which the L1 table for the
606                    snapshot starts. Must be aligned to a cluster boundary.
607
608         8 - 11:    Number of entries in the L1 table of the snapshots
609
610        12 - 13:    Length of the unique ID string describing the snapshot
611
612        14 - 15:    Length of the name of the snapshot
613
614        16 - 19:    Time at which the snapshot was taken in seconds since the
615                    Epoch
616
617        20 - 23:    Subsecond part of the time at which the snapshot was taken
618                    in nanoseconds
619
620        24 - 31:    Time that the guest was running until the snapshot was
621                    taken in nanoseconds
622
623        32 - 35:    Size of the VM state in bytes. 0 if no VM state is saved.
624                    If there is VM state, it starts at the first cluster
625                    described by first L1 table entry that doesn't describe a
626                    regular guest cluster (i.e. VM state is stored like guest
627                    disk content, except that it is stored at offsets that are
628                    larger than the virtual disk presented to the guest)
629
630        36 - 39:    Size of extra data in the table entry (used for future
631                    extensions of the format)
632
633        variable:   Extra data for future extensions. Unknown fields must be
634                    ignored. Currently defined are (offset relative to snapshot
635                    table entry):
636
637                    Byte 40 - 47:   Size of the VM state in bytes. 0 if no VM
638                                    state is saved. If this field is present,
639                                    the 32-bit value in bytes 32-35 is ignored.
640
641                    Byte 48 - 55:   Virtual disk size of the snapshot in bytes
642
643                    Version 3 images must include extra data at least up to
644                    byte 55.
645
646        variable:   Unique ID string for the snapshot (not null terminated)
647
648        variable:   Name of the snapshot (not null terminated)
649
650        variable:   Padding to round up the snapshot table entry size to the
651                    next multiple of 8.
652
653
654== Bitmaps ==
655
656As mentioned above, the bitmaps extension provides the ability to store bitmaps
657related to a virtual disk. This section describes how these bitmaps are stored.
658
659All stored bitmaps are related to the virtual disk stored in the same image, so
660each bitmap size is equal to the virtual disk size.
661
662Each bit of the bitmap is responsible for strictly defined range of the virtual
663disk. For bit number bit_nr the corresponding range (in bytes) will be:
664
665    [bit_nr * bitmap_granularity .. (bit_nr + 1) * bitmap_granularity - 1]
666
667Granularity is a property of the concrete bitmap, see below.
668
669
670=== Bitmap directory ===
671
672Each bitmap saved in the image is described in a bitmap directory entry. The
673bitmap directory is a contiguous area in the image file, whose starting offset
674and length are given by the header extension fields bitmap_directory_offset and
675bitmap_directory_size. The entries of the bitmap directory have variable
676length, depending on the lengths of the bitmap name and extra data.
677
678Structure of a bitmap directory entry:
679
680    Byte 0 -  7:    bitmap_table_offset
681                    Offset into the image file at which the bitmap table
682                    (described below) for the bitmap starts. Must be aligned to
683                    a cluster boundary.
684
685         8 - 11:    bitmap_table_size
686                    Number of entries in the bitmap table of the bitmap.
687
688        12 - 15:    flags
689                    Bit
690                      0: in_use
691                         The bitmap was not saved correctly and may be
692                         inconsistent. Although the bitmap metadata is still
693                         well-formed from a qcow2 perspective, the metadata
694                         (such as the auto flag or bitmap size) or data
695                         contents may be outdated.
696
697                      1: auto
698                         The bitmap must reflect all changes of the virtual
699                         disk by any application that would write to this qcow2
700                         file (including writes, snapshot switching, etc.). The
701                         type of this bitmap must be 'dirty tracking bitmap'.
702
703                      2: extra_data_compatible
704                         This flags is meaningful when the extra data is
705                         unknown to the software (currently any extra data is
706                         unknown to Qemu).
707                         If it is set, the bitmap may be used as expected, extra
708                         data must be left as is.
709                         If it is not set, the bitmap must not be used, but
710                         both it and its extra data be left as is.
711
712                    Bits 3 - 31 are reserved and must be 0.
713
714             16:    type
715                    This field describes the sort of the bitmap.
716                    Values:
717                      1: Dirty tracking bitmap
718
719                    Values 0, 2 - 255 are reserved.
720
721             17:    granularity_bits
722                    Granularity bits. Valid values: 0 - 63.
723
724                    Note: Qemu currently supports only values 9 - 31.
725
726                    Granularity is calculated as
727                        granularity = 1 << granularity_bits
728
729                    A bitmap's granularity is how many bytes of the image
730                    accounts for one bit of the bitmap.
731
732        18 - 19:    name_size
733                    Size of the bitmap name. Must be non-zero.
734
735                    Note: Qemu currently doesn't support values greater than
736                    1023.
737
738        20 - 23:    extra_data_size
739                    Size of type-specific extra data.
740
741                    For now, as no extra data is defined, extra_data_size is
742                    reserved and should be zero. If it is non-zero the
743                    behavior is defined by extra_data_compatible flag.
744
745        variable:   extra_data
746                    Extra data for the bitmap, occupying extra_data_size bytes.
747                    Extra data must never contain references to clusters or in
748                    some other way allocate additional clusters.
749
750        variable:   name
751                    The name of the bitmap (not null terminated), occupying
752                    name_size bytes. Must be unique among all bitmap names
753                    within the bitmaps extension.
754
755        variable:   Padding to round up the bitmap directory entry size to the
756                    next multiple of 8. All bytes of the padding must be zero.
757
758
759=== Bitmap table ===
760
761Each bitmap is stored using a one-level structure (as opposed to two-level
762structures like for refcounts and guest clusters mapping) for the mapping of
763bitmap data to host clusters. This structure is called the bitmap table.
764
765Each bitmap table has a variable size (stored in the bitmap directory entry)
766and may use multiple clusters, however, it must be contiguous in the image
767file.
768
769Structure of a bitmap table entry:
770
771    Bit       0:    Reserved and must be zero if bits 9 - 55 are non-zero.
772                    If bits 9 - 55 are zero:
773                      0: Cluster should be read as all zeros.
774                      1: Cluster should be read as all ones.
775
776         1 -  8:    Reserved and must be zero.
777
778         9 - 55:    Bits 9 - 55 of the host cluster offset. Must be aligned to
779                    a cluster boundary. If the offset is 0, the cluster is
780                    unallocated; in that case, bit 0 determines how this
781                    cluster should be treated during reads.
782
783        56 - 63:    Reserved and must be zero.
784
785
786=== Bitmap data ===
787
788As noted above, bitmap data is stored in separate clusters, described by the
789bitmap table. Given an offset (in bytes) into the bitmap data, the offset into
790the image file can be obtained as follows:
791
792    image_offset(bitmap_data_offset) =
793        bitmap_table[bitmap_data_offset / cluster_size] +
794            (bitmap_data_offset % cluster_size)
795
796This offset is not defined if bits 9 - 55 of bitmap table entry are zero (see
797above).
798
799Given an offset byte_nr into the virtual disk and the bitmap's granularity, the
800bit offset into the image file to the corresponding bit of the bitmap can be
801calculated like this:
802
803    bit_offset(byte_nr) =
804        image_offset(byte_nr / granularity / 8) * 8 +
805            (byte_nr / granularity) % 8
806
807If the size of the bitmap data is not a multiple of the cluster size then the
808last cluster of the bitmap data contains some unused tail bits. These bits must
809be zero.
810
811
812=== Dirty tracking bitmaps ===
813
814Bitmaps with 'type' field equal to one are dirty tracking bitmaps.
815
816When the virtual disk is in use dirty tracking bitmap may be 'enabled' or
817'disabled'. While the bitmap is 'enabled', all writes to the virtual disk
818should be reflected in the bitmap. A set bit in the bitmap means that the
819corresponding range of the virtual disk (see above) was written to while the
820bitmap was 'enabled'. An unset bit means that this range was not written to.
821
822The software doesn't have to sync the bitmap in the image file with its
823representation in RAM after each write or metadata change. Flag 'in_use'
824should be set while the bitmap is not synced.
825
826In the image file the 'enabled' state is reflected by the 'auto' flag. If this
827flag is set, the software must consider the bitmap as 'enabled' and start
828tracking virtual disk changes to this bitmap from the first write to the
829virtual disk. If this flag is not set then the bitmap is disabled.
830