1QEMU-IMG.1(1) QEMU-IMG.1(1)
2
6 qemu-img - QEMU disk image utility
7
9 qemu-img [standard options] command [command options]
10
12 qemu-img allows you to create, convert and modify images offline. It
13 can handle all image formats supported by QEMU.
14 Warning: Never use qemu-img to modify images in use by a running
16 virtual machine or any other process; this may destroy the image. Also,
17 be aware that querying an image that is being modified by another
18 process may encounter inconsistent state.
19
21 Standard options:
22 -h, --help
24 Display this help and exit
25 -V, --version
27 Display version information and exit
28 -T, --trace [[enable=]pattern][,events=file][,file=file]
30 Specify tracing options.
31 [enable=]pattern
33 Immediately enable events matching pattern (either event name
34 or a globbing pattern). This option is only available if QEMU
35 has been compiled with the simple, log or ftrace tracing
36 backend. To specify multiple events or patterns, specify the
37 -trace option multiple times.
38 Use "-trace help" to print a list of names of trace points.
40 events=file
42 Immediately enable events listed in file. The file must
43 contain one event name (as listed in the trace-events-all file)
44 per line; globbing patterns are accepted too. This option is
45 only available if QEMU has been compiled with the simple, log
46 or ftrace tracing backend.
47 file=file
49 Log output traces to file. This option is only available if
50 QEMU has been compiled with the simple tracing backend.
51 The following commands are supported:
53 amend [--object objectdef] [--image-opts] [-p] [-q] [-f fmt] [-t cache]
55 -o options filename
56 bench [-c count] [-d depth] [-f fmt] [--flush-interval=flush_interval]
57 [-n] [--no-drain] [-o offset] [--pattern=pattern] [-q] [-s buffer_size]
58 [-S step_size] [-t cache] [-w] [-U] filename
59 check [--object objectdef] [--image-opts] [-q] [-f fmt] [--output=ofmt]
60 [-r [leaks | all]] [-T src_cache] [-U] filename
61 commit [--object objectdef] [--image-opts] [-q] [-f fmt] [-t cache] [-b
62 base] [-d] [-p] filename
63 compare [--object objectdef] [--image-opts] [-f fmt] [-F fmt] [-T
64 src_cache] [-p] [-q] [-s] [-U] filename1 filename2
65 convert [--object objectdef] [--image-opts] [--target-image-opts] [-U]
66 [-C] [-c] [-p] [-q] [-n] [-f fmt] [-t cache] [-T src_cache] [-O
67 output_fmt] [-B backing_file] [-o options] [-l snapshot_param] [-S
68 sparse_size] [-m num_coroutines] [-W] [--salvage] filename [filename2
69 [...]] output_filename
70 create [--object objectdef] [-q] [-f fmt] [-b backing_file] [-F
71 backing_fmt] [-u] [-o options] filename [size]
72 dd [--image-opts] [-U] [-f fmt] [-O output_fmt] [bs=block_size]
73 [count=blocks] [skip=blocks] if=input of=output
74 info [--object objectdef] [--image-opts] [-f fmt] [--output=ofmt]
75 [--backing-chain] [-U] filename
76 map [--object objectdef] [--image-opts] [-f fmt] [--output=ofmt] [-U]
77 filename
78 measure [--output=ofmt] [-O output_fmt] [-o options] [--size N |
79 [--object objectdef] [--image-opts] [-f fmt] [-l snapshot_param]
80 filename]
81 snapshot [--object objectdef] [--image-opts] [-U] [-q] [-l | -a
82 snapshot | -c snapshot | -d snapshot] filename
83 rebase [--object objectdef] [--image-opts] [-U] [-q] [-f fmt] [-t
84 cache] [-T src_cache] [-p] [-u] -b backing_file [-F backing_fmt]
85 filename
86 resize [--object objectdef] [--image-opts] [-f fmt]
87 [--preallocation=prealloc] [-q] [--shrink] filename [+ | -]size
88 Command parameters:
90 filename
92 is a disk image filename
93 fmt is the disk image format. It is guessed automatically in most
95 cases. See below for a description of the supported disk formats.
96 size
98 is the disk image size in bytes. Optional suffixes "k" or "K"
99 (kilobyte, 1024) "M" (megabyte, 1024k) and "G" (gigabyte, 1024M)
100 and T (terabyte, 1024G) are supported. "b" is ignored.
101 output_filename
103 is the destination disk image filename
104 output_fmt
106 is the destination format
107 options
109 is a comma separated list of format specific options in a
110 name=value format. Use "-o ?" for an overview of the options
111 supported by the used format or see the format descriptions below
112 for details.
113 snapshot_param
115 is param used for internal snapshot, format is
116 'snapshot.id=[ID],snapshot.name=[NAME]' or '[ID_OR_NAME]'
117 --object objectdef
119 is a QEMU user creatable object definition. See the qemu(1) manual
120 page for a description of the object properties. The most common
121 object type is a "secret", which is used to supply passwords and/or
122 encryption keys.
123 --image-opts
125 Indicates that the source filename parameter is to be interpreted
126 as a full option string, not a plain filename. This parameter is
127 mutually exclusive with the -f parameter.
128 --target-image-opts
130 Indicates that the output_filename parameter(s) are to be
131 interpreted as a full option string, not a plain filename. This
132 parameter is mutually exclusive with the -O parameters. It is
133 currently required to also use the -n parameter to skip image
134 creation. This restriction may be relaxed in a future release.
135 --force-share (-U)
137 If specified, "qemu-img" will open the image in shared mode,
138 allowing other QEMU processes to open it in write mode. For
139 example, this can be used to get the image information (with 'info'
140 subcommand) when the image is used by a running guest. Note that
141 this could produce inconsistent results because of concurrent
142 metadata changes, etc. This option is only allowed when opening
143 images in read-only mode.
144 --backing-chain
146 will enumerate information about backing files in a disk image
147 chain. Refer below for further description.
148 -c indicates that target image must be compressed (qcow format only)
150 -h with or without a command shows help and lists the supported
152 formats
153 -p display progress bar (compare, convert and rebase commands only).
155 If the -p option is not used for a command that supports it, the
156 progress is reported when the process receives a "SIGUSR1" or
157 "SIGINFO" signal.
158 -q Quiet mode - do not print any output (except errors). There's no
160 progress bar in case both -q and -p options are used.
161 -S size
163 indicates the consecutive number of bytes that must contain only
164 zeros for qemu-img to create a sparse image during conversion. This
165 value is rounded down to the nearest 512 bytes. You may use the
166 common size suffixes like "k" for kilobytes.
167 -t cache
169 specifies the cache mode that should be used with the (destination)
170 file. See the documentation of the emulator's "-drive cache=..."
171 option for allowed values.
172 -T src_cache
174 specifies the cache mode that should be used with the source
175 file(s). See the documentation of the emulator's "-drive cache=..."
176 option for allowed values.
177 Parameters to snapshot subcommand:
179 snapshot
181 is the name of the snapshot to create, apply or delete
182 -a applies a snapshot (revert disk to saved state)
184 -c creates a snapshot
186 -d deletes a snapshot
188 -l lists all snapshots in the given image
190 Parameters to compare subcommand:
192 -f First image format
194 -F Second image format
196 -s Strict mode - fail on different image size or sector allocation
198 Parameters to convert subcommand:
200 -n Skip the creation of the target volume
202 -m Number of parallel coroutines for the convert process
204 -W Allow out-of-order writes to the destination. This option improves
206 performance, but is only recommended for preallocated devices like
207 host devices or other raw block devices.
208 -C Try to use copy offloading to move data from source image to
210 target. This may improve performance if the data is remote, such as
211 with NFS or iSCSI backends, but will not automatically sparsify
212 zero sectors, and may result in a fully allocated target image
213 depending on the host support for getting allocation information.
214 --salvage
216 Try to ignore I/O errors when reading. Unless in quiet mode
217 ("-q"), errors will still be printed. Areas that cannot be read
218 from the source will be treated as containing only zeroes.
219 Parameters to dd subcommand:
221 bs=block_size
223 defines the block size
224 count=blocks
226 sets the number of input blocks to copy
227 if=input
229 sets the input file
230 of=output
232 sets the output file
233 skip=blocks
235 sets the number of input blocks to skip
236 Command description:
238 amend [--object objectdef] [--image-opts] [-p] [-q] [-f fmt] [-t cache]
240 -o options filename
241 Amends the image format specific options for the image file
242 filename. Not all file formats support this operation.
243 bench [-c count] [-d depth] [-f fmt] [--flush-interval=flush_interval]
245 [-n] [--no-drain] [-o offset] [--pattern=pattern] [-q] [-s buffer_size]
246 [-S step_size] [-t cache] [-w] [-U] filename
247 Run a simple sequential I/O benchmark on the specified image. If
248 "-w" is specified, a write test is performed, otherwise a read test
249 is performed.
250 A total number of count I/O requests is performed, each buffer_size
252 bytes in size, and with depth requests in parallel. The first
253 request starts at the position given by offset, each following
254 request increases the current position by step_size. If step_size
255 is not given, buffer_size is used for its value.
256 If flush_interval is specified for a write test, the request queue
258 is drained and a flush is issued before new writes are made
259 whenever the number of remaining requests is a multiple of
260 flush_interval. If additionally "--no-drain" is specified, a flush
261 is issued without draining the request queue first.
262 If "-n" is specified, the native AIO backend is used if possible.
264 On Linux, this option only works if "-t none" or "-t directsync" is
265 specified as well.
266 For write tests, by default a buffer filled with zeros is written.
268 This can be overridden with a pattern byte specified by pattern.
269 check [--object objectdef] [--image-opts] [-q] [-f fmt] [--output=ofmt]
271 [-r [leaks | all]] [-T src_cache] [-U] filename
272 Perform a consistency check on the disk image filename. The command
273 can output in the format ofmt which is either "human" or "json".
274 The JSON output is an object of QAPI type "ImageCheck".
275 If "-r" is specified, qemu-img tries to repair any inconsistencies
277 found during the check. "-r leaks" repairs only cluster leaks,
278 whereas "-r all" fixes all kinds of errors, with a higher risk of
279 choosing the wrong fix or hiding corruption that has already
280 occurred.
281 Only the formats "qcow2", "qed" and "vdi" support consistency
283 checks.
284 In case the image does not have any inconsistencies, check exits
286 with 0. Other exit codes indicate the kind of inconsistency found
287 or if another error occurred. The following table summarizes all
288 exit codes of the check subcommand:
289 0 Check completed, the image is (now) consistent
291 1 Check not completed because of internal errors
293 2 Check completed, image is corrupted
295 3 Check completed, image has leaked clusters, but is not
297 corrupted
298 63 Checks are not supported by the image format
300 If "-r" is specified, exit codes representing the image state refer
302 to the state after (the attempt at) repairing it. That is, a
303 successful "-r all" will yield the exit code 0, independently of
304 the image state before.
305 commit [--object objectdef] [--image-opts] [-q] [-f fmt] [-t cache] [-b
307 base] [-d] [-p] filename
308 Commit the changes recorded in filename in its base image or
309 backing file. If the backing file is smaller than the snapshot,
310 then the backing file will be resized to be the same size as the
311 snapshot. If the snapshot is smaller than the backing file, the
312 backing file will not be truncated. If you want the backing file
313 to match the size of the smaller snapshot, you can safely truncate
314 it yourself once the commit operation successfully completes.
315 The image filename is emptied after the operation has succeeded. If
317 you do not need filename afterwards and intend to drop it, you may
318 skip emptying filename by specifying the "-d" flag.
319 If the backing chain of the given image file filename has more than
321 one layer, the backing file into which the changes will be
322 committed may be specified as base (which has to be part of
323 filename's backing chain). If base is not specified, the immediate
324 backing file of the top image (which is filename) will be used.
325 Note that after a commit operation all images between base and the
326 top image will be invalid and may return garbage data when read.
327 For this reason, "-b" implies "-d" (so that the top image stays
328 valid).
329 compare [--object objectdef] [--image-opts] [-f fmt] [-F fmt] [-T
331 src_cache] [-p] [-q] [-s] [-U] filename1 filename2
332 Check if two images have the same content. You can compare images
333 with different format or settings.
334 The format is probed unless you specify it by -f (used for
336 filename1) and/or -F (used for filename2) option.
337 By default, images with different size are considered identical if
339 the larger image contains only unallocated and/or zeroed sectors in
340 the area after the end of the other image. In addition, if any
341 sector is not allocated in one image and contains only zero bytes
342 in the second one, it is evaluated as equal. You can use Strict
343 mode by specifying the -s option. When compare runs in Strict mode,
344 it fails in case image size differs or a sector is allocated in one
345 image and is not allocated in the second one.
346 By default, compare prints out a result message. This message
348 displays information that both images are same or the position of
349 the first different byte. In addition, result message can report
350 different image size in case Strict mode is used.
351 Compare exits with 0 in case the images are equal and with 1 in
353 case the images differ. Other exit codes mean an error occurred
354 during execution and standard error output should contain an error
355 message. The following table sumarizes all exit codes of the
356 compare subcommand:
357 0 Images are identical
359 1 Images differ
361 2 Error on opening an image
363 3 Error on checking a sector allocation
365 4 Error on reading data
367 convert [--object objectdef] [--image-opts] [--target-image-opts] [-U]
369 [-C] [-c] [-p] [-q] [-n] [-f fmt] [-t cache] [-T src_cache] [-O
370 output_fmt] [-B backing_file] [-o options] [-l snapshot_param] [-S
371 sparse_size] [-m num_coroutines] [-W] filename [filename2 [...]]
372 output_filename
373 Convert the disk image filename or a snapshot snapshot_param to
374 disk image output_filename using format output_fmt. It can be
375 optionally compressed ("-c" option) or use any format specific
376 options like encryption ("-o" option).
377 Only the formats "qcow" and "qcow2" support compression. The
379 compression is read-only. It means that if a compressed sector is
380 rewritten, then it is rewritten as uncompressed data.
381 Image conversion is also useful to get smaller image when using a
383 growable format such as "qcow": the empty sectors are detected and
384 suppressed from the destination image.
385 sparse_size indicates the consecutive number of bytes (defaults to
387 4k) that must contain only zeros for qemu-img to create a sparse
388 image during conversion. If sparse_size is 0, the source will not
389 be scanned for unallocated or zero sectors, and the destination
390 image will always be fully allocated.
391 You can use the backing_file option to force the output image to be
393 created as a copy on write image of the specified base image; the
394 backing_file should have the same content as the input's base
395 image, however the path, image format, etc may differ.
396 If a relative path name is given, the backing file is looked up
398 relative to the directory containing output_filename.
399 If the "-n" option is specified, the target volume creation will be
401 skipped. This is useful for formats such as "rbd" if the target
402 volume has already been created with site specific options that
403 cannot be supplied through qemu-img.
404 Out of order writes can be enabled with "-W" to improve
406 performance. This is only recommended for preallocated devices
407 like host devices or other raw block devices. Out of order write
408 does not work in combination with creating compressed images.
409 num_coroutines specifies how many coroutines work in parallel
411 during the convert process (defaults to 8).
412 create [--object objectdef] [-q] [-f fmt] [-b backing_file] [-F
414 backing_fmt] [-u] [-o options] filename [size]
415 Create the new disk image filename of size size and format fmt.
416 Depending on the file format, you can add one or more options that
417 enable additional features of this format.
418 If the option backing_file is specified, then the image will record
420 only the differences from backing_file. No size needs to be
421 specified in this case. backing_file will never be modified unless
422 you use the "commit" monitor command (or qemu-img commit).
423 If a relative path name is given, the backing file is looked up
425 relative to the directory containing filename.
426 Note that a given backing file will be opened to check that it is
428 valid. Use the "-u" option to enable unsafe backing file mode,
429 which means that the image will be created even if the associated
430 backing file cannot be opened. A matching backing file must be
431 created or additional options be used to make the backing file
432 specification valid when you want to use an image created this way.
433 The size can also be specified using the size option with "-o", it
435 doesn't need to be specified separately in this case.
436 dd [--image-opts] [-U] [-f fmt] [-O output_fmt] [bs=block_size]
438 [count=blocks] [skip=blocks] if=input of=output
439 Dd copies from input file to output file converting it from fmt
440 format to output_fmt format.
441 The data is by default read and written using blocks of 512 bytes
443 but can be modified by specifying block_size. If count=blocks is
444 specified dd will stop reading input after reading blocks input
445 blocks.
446 The size syntax is similar to dd(1)'s size syntax.
448 info [--object objectdef] [--image-opts] [-f fmt] [--output=ofmt]
450 [--backing-chain] [-U] filename
451 Give information about the disk image filename. Use it in
452 particular to know the size reserved on disk which can be different
453 from the displayed size. If VM snapshots are stored in the disk
454 image, they are displayed too.
455 If a disk image has a backing file chain, information about each
457 disk image in the chain can be recursively enumerated by using the
458 option "--backing-chain".
459 For instance, if you have an image chain like:
461 base.qcow2 <- snap1.qcow2 <- snap2.qcow2
463 To enumerate information about each disk image in the above chain,
465 starting from top to base, do:
466 qemu-img info --backing-chain snap2.qcow2
468 The command can output in the format ofmt which is either "human"
470 or "json". The JSON output is an object of QAPI type "ImageInfo";
471 with "--backing-chain", it is an array of "ImageInfo" objects.
472 "--output=human" reports the following information (for every image
474 in the chain):
475 image
477 The image file name
478 file format
480 The image format
481 virtual size
483 The size of the guest disk
484 disk size
486 How much space the image file occupies on the host file system
487 (may be shown as 0 if this information is unavailable, e.g.
488 because there is no file system)
489 cluster_size
491 Cluster size of the image format, if applicable
492 encrypted
494 Whether the image is encrypted (only present if so)
495 cleanly shut down
497 This is shown as "no" if the image is dirty and will have to be
498 auto-repaired the next time it is opened in qemu.
499 backing file
501 The backing file name, if present
502 backing file format
504 The format of the backing file, if the image enforces it
505 Snapshot list
507 A list of all internal snapshots
508 Format specific information
510 Further information whose structure depends on the image
511 format. This section is a textual representation of the
512 respective "ImageInfoSpecific*" QAPI object (e.g.
513 "ImageInfoSpecificQCow2" for qcow2 images).
514 map [--object objectdef] [--image-opts] [-f fmt] [--output=ofmt] [-U]
516 filename
517 Dump the metadata of image filename and its backing file chain. In
518 particular, this commands dumps the allocation state of every
519 sector of filename, together with the topmost file that allocates
520 it in the backing file chain.
521 Two option formats are possible. The default format ("human") only
523 dumps known-nonzero areas of the file. Known-zero parts of the
524 file are omitted altogether, and likewise for parts that are not
525 allocated throughout the chain. qemu-img output will identify a
526 file from where the data can be read, and the offset in the file.
527 Each line will include four fields, the first three of which are
528 hexadecimal numbers. For example the first line of:
529 Offset Length Mapped to File
531 0 0x20000 0x50000 /tmp/overlay.qcow2
532 0x100000 0x10000 0x95380000 /tmp/backing.qcow2
533 means that 0x20000 (131072) bytes starting at offset 0 in the image
535 are available in /tmp/overlay.qcow2 (opened in "raw" format)
536 starting at offset 0x50000 (327680). Data that is compressed,
537 encrypted, or otherwise not available in raw format will cause an
538 error if "human" format is in use. Note that file names can
539 include newlines, thus it is not safe to parse this output format
540 in scripts.
541 The alternative format "json" will return an array of dictionaries
543 in JSON format. It will include similar information in the
544 "start", "length", "offset" fields; it will also include other more
545 specific information:
546 - whether the sectors contain actual data or not (boolean field
548 "data"; if false, the sectors are either unallocated or stored
549 as optimized all-zero clusters);
550 - whether the data is known to read as zero (boolean field
552 "zero");
553 - in order to make the output shorter, the target file is
555 expressed as a "depth"; for example, a depth of 2 refers to the
556 backing file of the backing file of filename.
557 In JSON format, the "offset" field is optional; it is absent in
559 cases where "human" format would omit the entry or exit with an
560 error. If "data" is false and the "offset" field is present, the
561 corresponding sectors in the file are not yet in use, but they are
562 preallocated.
563 For more information, consult include/block/block.h in QEMU's
565 source code.
566 measure [--output=ofmt] [-O output_fmt] [-o options] [--size N |
568 [--object objectdef] [--image-opts] [-f fmt] [-l snapshot_param]
569 filename]
570 Calculate the file size required for a new image. This information
571 can be used to size logical volumes or SAN LUNs appropriately for
572 the image that will be placed in them. The values reported are
573 guaranteed to be large enough to fit the image. The command can
574 output in the format ofmt which is either "human" or "json". The
575 JSON output is an object of QAPI type "BlockMeasureInfo".
576 If the size N is given then act as if creating a new empty image
578 file using qemu-img create. If filename is given then act as if
579 converting an existing image file using qemu-img convert. The
580 format of the new file is given by output_fmt while the format of
581 an existing file is given by fmt.
582 A snapshot in an existing image can be specified using
584 snapshot_param.
585 The following fields are reported:
587 required size: 524288
589 fully allocated size: 1074069504
590 The "required size" is the file size of the new image. It may be
592 smaller than the virtual disk size if the image format supports
593 compact representation.
594 The "fully allocated size" is the file size of the new image once
596 data has been written to all sectors. This is the maximum size
597 that the image file can occupy with the exception of internal
598 snapshots, dirty bitmaps, vmstate data, and other advanced image
599 format features.
600 snapshot [--object objectdef] [--image-opts] [-U] [-q] [-l | -a
602 snapshot | -c snapshot | -d snapshot] filename
603 List, apply, create or delete snapshots in image filename.
604 rebase [--object objectdef] [--image-opts] [-U] [-q] [-f fmt] [-t
606 cache] [-T src_cache] [-p] [-u] -b backing_file [-F backing_fmt]
607 filename
608 Changes the backing file of an image. Only the formats "qcow2" and
609 "qed" support changing the backing file.
610 The backing file is changed to backing_file and (if the image
612 format of filename supports this) the backing file format is
613 changed to backing_fmt. If backing_file is specified as "" (the
614 empty string), then the image is rebased onto no backing file (i.e.
615 it will exist independently of any backing file).
616 If a relative path name is given, the backing file is looked up
618 relative to the directory containing filename.
619 cache specifies the cache mode to be used for filename, whereas
621 src_cache specifies the cache mode for reading backing files.
622 There are two different modes in which "rebase" can operate:
624 Safe mode
626 This is the default mode and performs a real rebase operation.
627 The new backing file may differ from the old one and qemu-img
628 rebase will take care of keeping the guest-visible content of
629 filename unchanged.
630 In order to achieve this, any clusters that differ between
632 backing_file and the old backing file of filename are merged
633 into filename before actually changing the backing file.
634 Note that the safe mode is an expensive operation, comparable
636 to converting an image. It only works if the old backing file
637 still exists.
638 Unsafe mode
640 qemu-img uses the unsafe mode if "-u" is specified. In this
641 mode, only the backing file name and format of filename is
642 changed without any checks on the file contents. The user must
643 take care of specifying the correct new backing file, or the
644 guest-visible content of the image will be corrupted.
645 This mode is useful for renaming or moving the backing file to
647 somewhere else. It can be used without an accessible old
648 backing file, i.e. you can use it to fix an image whose backing
649 file has already been moved/renamed.
650 You can use "rebase" to perform a "diff" operation on two disk
652 images. This can be useful when you have copied or cloned a guest,
653 and you want to get back to a thin image on top of a template or
654 base image.
655 Say that "base.img" has been cloned as "modified.img" by copying
657 it, and that the "modified.img" guest has run so there are now some
658 changes compared to "base.img". To construct a thin image called
659 "diff.qcow2" that contains just the differences, do:
660 qemu-img create -f qcow2 -b modified.img diff.qcow2
662 qemu-img rebase -b base.img diff.qcow2
663 At this point, "modified.img" can be discarded, since "base.img +
665 diff.qcow2" contains the same information.
666 resize [--object objectdef] [--image-opts] [-f fmt]
668 [--preallocation=prealloc] [-q] [--shrink] filename [+ | -]size
669 Change the disk image as if it had been created with size.
670 Before using this command to shrink a disk image, you MUST use file
672 system and partitioning tools inside the VM to reduce allocated
673 file systems and partition sizes accordingly. Failure to do so
674 will result in data loss!
675 When shrinking images, the "--shrink" option must be given. This
677 informs qemu-img that the user acknowledges all loss of data beyond
678 the truncated image's end.
679 After using this command to grow a disk image, you must use file
681 system and partitioning tools inside the VM to actually begin using
682 the new space on the device.
683 When growing an image, the "--preallocation" option may be used to
685 specify how the additional image area should be allocated on the
686 host. See the format description in the "NOTES" section which
687 values are allowed. Using this option may result in slightly more
688 data being allocated than necessary.
689
691 Supported image file formats:
692 raw Raw disk image format (default). This format has the advantage of
694 being simple and easily exportable to all other emulators. If your
695 file system supports holes (for example in ext2 or ext3 on Linux or
696 NTFS on Windows), then only the written sectors will reserve space.
697 Use "qemu-img info" to know the real size used by the image or "ls
698 -ls" on Unix/Linux.
699 Supported options:
701 "preallocation"
703 Preallocation mode (allowed values: "off", "falloc", "full").
704 "falloc" mode preallocates space for image by calling
705 posix_fallocate(). "full" mode preallocates space for image by
706 writing zeros to underlying storage.
707 qcow2
709 QEMU image format, the most versatile format. Use it to have
710 smaller images (useful if your filesystem does not supports holes,
711 for example on Windows), optional AES encryption, zlib based
712 compression and support of multiple VM snapshots.
713 Supported options:
715 "compat"
717 Determines the qcow2 version to use. "compat=0.10" uses the
718 traditional image format that can be read by any QEMU since
719 0.10. "compat=1.1" enables image format extensions that only
720 QEMU 1.1 and newer understand (this is the default). Amongst
721 others, this includes zero clusters, which allow efficient
722 copy-on-read for sparse images.
723 "backing_file"
725 File name of a base image (see create subcommand)
726 "backing_fmt"
728 Image format of the base image
729 "encryption"
731 If this option is set to "on", the image is encrypted with
732 128-bit AES-CBC.
733 The use of encryption in qcow and qcow2 images is considered to
735 be flawed by modern cryptography standards, suffering from a
736 number of design problems:
737 - The AES-CBC cipher is used with predictable initialization
739 vectors based on the sector number. This makes it
740 vulnerable to chosen plaintext attacks which can reveal the
741 existence of encrypted data.
742 - The user passphrase is directly used as the encryption key.
744 A poorly chosen or short passphrase will compromise the
745 security of the encryption.
746 - In the event of the passphrase being compromised there is
748 no way to change the passphrase to protect data in any qcow
749 images. The files must be cloned, using a different
750 encryption passphrase in the new file. The original file
751 must then be securely erased using a program like shred,
752 though even this is ineffective with many modern storage
753 technologies.
754 - Initialization vectors used to encrypt sectors are based on
756 the guest virtual sector number, instead of the host
757 physical sector. When a disk image has multiple internal
758 snapshots this means that data in multiple physical sectors
759 is encrypted with the same initialization vector. With the
760 CBC mode, this opens the possibility of watermarking
761 attacks if the attack can collect multiple sectors
762 encrypted with the same IV and some predictable data.
763 Having multiple qcow2 images with the same passphrase also
764 exposes this weakness since the passphrase is directly used
765 as the key.
766 Use of qcow / qcow2 encryption is thus strongly discouraged.
768 Users are recommended to use an alternative encryption
769 technology such as the Linux dm-crypt / LUKS system.
770 "cluster_size"
772 Changes the qcow2 cluster size (must be between 512 and 2M).
773 Smaller cluster sizes can improve the image file size whereas
774 larger cluster sizes generally provide better performance.
775 "preallocation"
777 Preallocation mode (allowed values: "off", "metadata",
778 "falloc", "full"). An image with preallocated metadata is
779 initially larger but can improve performance when the image
780 needs to grow. "falloc" and "full" preallocations are like the
781 same options of "raw" format, but sets up metadata also.
782 "lazy_refcounts"
784 If this option is set to "on", reference count updates are
785 postponed with the goal of avoiding metadata I/O and improving
786 performance. This is particularly interesting with
787 cache=writethrough which doesn't batch metadata updates. The
788 tradeoff is that after a host crash, the reference count tables
789 must be rebuilt, i.e. on the next open an (automatic) "qemu-img
790 check -r all" is required, which may take some time.
791 This option can only be enabled if "compat=1.1" is specified.
793 "nocow"
795 If this option is set to "on", it will turn off COW of the
796 file. It's only valid on btrfs, no effect on other file
797 systems.
798 Btrfs has low performance when hosting a VM image file, even
800 more when the guest on the VM also using btrfs as file system.
801 Turning off COW is a way to mitigate this bad performance.
802 Generally there are two ways to turn off COW on btrfs: a)
803 Disable it by mounting with nodatacow, then all newly created
804 files will be NOCOW. b) For an empty file, add the NOCOW file
805 attribute. That's what this option does.
806 Note: this option is only valid to new or empty files. If there
808 is an existing file which is COW and has data blocks already,
809 it couldn't be changed to NOCOW by setting "nocow=on". One can
810 issue "lsattr filename" to check if the NOCOW flag is set or
811 not (Capital 'C' is NOCOW flag).
812 Other
814 QEMU also supports various other image file formats for
815 compatibility with older QEMU versions or other hypervisors,
816 including VMDK, VDI, VHD (vpc), VHDX, qcow1 and QED. For a full
817 list of supported formats see "qemu-img --help". For a more
818 detailed description of these formats, see the QEMU Emulation User
819 Documentation.
820 The main purpose of the block drivers for these formats is image
822 conversion. For running VMs, it is recommended to convert the disk
823 images to either raw or qcow2 in order to achieve good performance.
824
826 The HTML documentation of QEMU for more precise information and Linux
827 user mode emulator invocation.
828
830 Fabrice Bellard
831 2019-11-15 QEMU-IMG.1(1)