5 guestfs - Library for accessing and modifying virtual machine images
11 guestfs_h *g = guestfs_create ();
12 guestfs_add_drive (g, "guest.img");
14 guestfs_mount (g, "/dev/sda1", "/");
15 guestfs_touch (g, "/hello");
16 guestfs_umount (g, "/");
19 cc prog.c -o prog -lguestfs
21 cc prog.c -o prog `pkg-config libguestfs --cflags --libs`
25 Libguestfs is a library for accessing and modifying guest disk images.
26 Amongst the things this is good for: making batch configuration
27 changes to guests, getting disk used/free statistics (see also:
28 virt-df), migrating between virtualization systems (see also:
29 virt-p2v), performing partial backups, performing partial guest
30 clones, cloning guests and changing registry/UUID/hostname info, and
33 Libguestfs uses Linux kernel and qemu code, and can access any type of
34 guest filesystem that Linux and qemu can, including but not limited
35 to: ext2/3/4, btrfs, FAT and NTFS, LVM, many different disk partition
36 schemes, qcow, qcow2, vmdk.
38 Libguestfs provides ways to enumerate guest storage (eg. partitions,
39 LVs, what filesystem is in each LV, etc.). It can also run commands
40 in the context of the guest. Also you can access filesystems over
43 Libguestfs is a library that can be linked with C and C++ management
44 programs (or management programs written in OCaml, Perl, Python, Ruby,
45 Java, PHP, Haskell or C#). You can also use it from shell scripts or the
48 You don't need to be root to use libguestfs, although obviously you do
49 need enough permissions to access the disk images.
51 Libguestfs is a large API because it can do many things. For a gentle
52 introduction, please read the L</API OVERVIEW> section next.
54 There are also some example programs in the L<guestfs-examples(3)>
59 This section provides a gentler overview of the libguestfs API. We
60 also try to group API calls together, where that may not be obvious
61 from reading about the individual calls in the main section of this
66 Before you can use libguestfs calls, you have to create a handle.
67 Then you must add at least one disk image to the handle, followed by
68 launching the handle, then performing whatever operations you want,
69 and finally closing the handle. By convention we use the single
70 letter C<g> for the name of the handle variable, although of course
71 you can use any name you want.
73 The general structure of all libguestfs-using programs looks like
76 guestfs_h *g = guestfs_create ();
78 /* Call guestfs_add_drive additional times if there are
79 * multiple disk images.
81 guestfs_add_drive (g, "guest.img");
83 /* Most manipulation calls won't work until you've launched
84 * the handle 'g'. You have to do this _after_ adding drives
85 * and _before_ other commands.
89 /* Now you can examine what partitions, LVs etc are available.
91 char **partitions = guestfs_list_partitions (g);
92 char **logvols = guestfs_lvs (g);
94 /* To access a filesystem in the image, you must mount it.
96 guestfs_mount (g, "/dev/sda1", "/");
98 /* Now you can perform filesystem actions on the guest
101 guestfs_touch (g, "/hello");
103 /* This is only needed for libguestfs < 1.5.24. Since then
104 * it is done automatically when you close the handle. See
105 * discussion of autosync in this page.
109 /* Close the handle 'g'. */
112 The code above doesn't include any error checking. In real code you
113 should check return values carefully for errors. In general all
114 functions that return integers return C<-1> on error, and all
115 functions that return pointers return C<NULL> on error. See section
116 L</ERROR HANDLING> below for how to handle errors, and consult the
117 documentation for each function call below to see precisely how they
118 return error indications. See L<guestfs-examples(3)> for fully worked
123 The image filename (C<"guest.img"> in the example above) could be a
124 disk image from a virtual machine, a L<dd(1)> copy of a physical hard
125 disk, an actual block device, or simply an empty file of zeroes that
126 you have created through L<posix_fallocate(3)>. Libguestfs lets you
127 do useful things to all of these.
129 The call you should use in modern code for adding drives is
130 L</guestfs_add_drive_opts>. To add a disk image, allowing writes, and
131 specifying that the format is raw, do:
133 guestfs_add_drive_opts (g, filename,
134 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "raw",
137 You can add a disk read-only using:
139 guestfs_add_drive_opts (g, filename,
140 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "raw",
141 GUESTFS_ADD_DRIVE_OPTS_READONLY, 1,
144 or by calling the older function L</guestfs_add_drive_ro>. In either
145 case libguestfs won't modify the file.
147 Be extremely cautious if the disk image is in use, eg. if it is being
148 used by a virtual machine. Adding it read-write will almost certainly
149 cause disk corruption, but adding it read-only is safe.
151 You must add at least one disk image, and you may add multiple disk
152 images. In the API, the disk images are usually referred to as
153 C</dev/sda> (for the first one you added), C</dev/sdb> (for the second
156 Once L</guestfs_launch> has been called you cannot add any more images.
157 You can call L</guestfs_list_devices> to get a list of the device
158 names, in the order that you added them. See also L</BLOCK DEVICE
163 Before you can read or write files, create directories and so on in a
164 disk image that contains filesystems, you have to mount those
165 filesystems using L</guestfs_mount>. If you already know that a disk
166 image contains (for example) one partition with a filesystem on that
167 partition, then you can mount it directly:
169 guestfs_mount (g, "/dev/sda1", "/");
171 where C</dev/sda1> means literally the first partition (C<1>) of the
172 first disk image that we added (C</dev/sda>). If the disk contains
173 Linux LVM2 logical volumes you could refer to those instead (eg. C</dev/VG/LV>).
175 If you are given a disk image and you don't know what it contains then
176 you have to find out. Libguestfs can do that too: use
177 L</guestfs_list_partitions> and L</guestfs_lvs> to list possible
178 partitions and LVs, and either try mounting each to see what is
179 mountable, or else examine them with L</guestfs_vfs_type> or
180 L</guestfs_file>. Libguestfs also has a set of APIs for inspection of
181 disk images (see L</INSPECTION> below). But you might find it easier
182 to look at higher level programs built on top of libguestfs, in
183 particular L<virt-inspector(1)>.
185 To mount a disk image read-only, use L</guestfs_mount_ro>. There are
186 several other variations of the C<guestfs_mount_*> call.
188 =head2 FILESYSTEM ACCESS AND MODIFICATION
190 The majority of the libguestfs API consists of fairly low-level calls
191 for accessing and modifying the files, directories, symlinks etc on
192 mounted filesystems. There are over a hundred such calls which you
193 can find listed in detail below in this man page, and we don't even
194 pretend to cover them all in this overview.
196 Specify filenames as full paths, starting with C<"/"> and including
199 For example, if you mounted a filesystem at C<"/"> and you want to
200 read the file called C<"etc/passwd"> then you could do:
202 char *data = guestfs_cat (g, "/etc/passwd");
204 This would return C<data> as a newly allocated buffer containing the
205 full content of that file (with some conditions: see also
206 L</DOWNLOADING> below), or C<NULL> if there was an error.
208 As another example, to create a top-level directory on that filesystem
209 called C<"var"> you would do:
211 guestfs_mkdir (g, "/var");
213 To create a symlink you could do:
215 guestfs_ln_s (g, "/etc/init.d/portmap",
216 "/etc/rc3.d/S30portmap");
218 Libguestfs will reject attempts to use relative paths and there is no
219 concept of a current working directory.
221 Libguestfs can return errors in many situations: for example if the
222 filesystem isn't writable, or if a file or directory that you
223 requested doesn't exist. If you are using the C API (documented here)
224 you have to check for those error conditions after each call. (Other
225 language bindings turn these errors into exceptions).
227 File writes are affected by the per-handle umask, set by calling
228 L</guestfs_umask> and defaulting to 022. See L</UMASK>.
232 Libguestfs contains API calls to read, create and modify partition
233 tables on disk images.
235 In the common case where you want to create a single partition
236 covering the whole disk, you should use the L</guestfs_part_disk>
239 const char *parttype = "mbr";
240 if (disk_is_larger_than_2TB)
242 guestfs_part_disk (g, "/dev/sda", parttype);
244 Obviously this effectively wipes anything that was on that disk image
249 Libguestfs provides access to a large part of the LVM2 API, such as
250 L</guestfs_lvcreate> and L</guestfs_vgremove>. It won't make much sense
251 unless you familiarize yourself with the concepts of physical volumes,
252 volume groups and logical volumes.
254 This author strongly recommends reading the LVM HOWTO, online at
255 L<http://tldp.org/HOWTO/LVM-HOWTO/>.
259 Use L</guestfs_cat> to download small, text only files. This call
260 is limited to files which are less than 2 MB and which cannot contain
261 any ASCII NUL (C<\0>) characters. However it has a very simple
264 L</guestfs_read_file> can be used to read files which contain
265 arbitrary 8 bit data, since it returns a (pointer, size) pair.
266 However it is still limited to "small" files, less than 2 MB.
268 L</guestfs_download> can be used to download any file, with no
269 limits on content or size (even files larger than 4 GB).
271 To download multiple files, see L</guestfs_tar_out> and
276 It's often the case that you want to write a file or files to the disk
279 To write a small file with fixed content, use L</guestfs_write>. To
280 create a file of all zeroes, use L</guestfs_truncate_size> (sparse) or
281 L</guestfs_fallocate64> (with all disk blocks allocated). There are a
282 variety of other functions for creating test files, for example
283 L</guestfs_fill> and L</guestfs_fill_pattern>.
285 To upload a single file, use L</guestfs_upload>. This call has no
286 limits on file content or size (even files larger than 4 GB).
288 To upload multiple files, see L</guestfs_tar_in> and L</guestfs_tgz_in>.
290 However the fastest way to upload I<large numbers of arbitrary files>
291 is to turn them into a squashfs or CD ISO (see L<mksquashfs(8)> and
292 L<mkisofs(8)>), then attach this using L</guestfs_add_drive_ro>. If
293 you add the drive in a predictable way (eg. adding it last after all
294 other drives) then you can get the device name from
295 L</guestfs_list_devices> and mount it directly using
296 L</guestfs_mount_ro>. Note that squashfs images are sometimes
297 non-portable between kernel versions, and they don't support labels or
298 UUIDs. If you want to pre-build an image or you need to mount it
299 using a label or UUID, use an ISO image instead.
303 There are various different commands for copying between files and
304 devices and in and out of the guest filesystem. These are summarised
309 =item B<file> to B<file>
311 Use L</guestfs_cp> to copy a single file, or
312 L</guestfs_cp_a> to copy directories recursively.
314 =item B<file or device> to B<file or device>
316 Use L</guestfs_dd> which efficiently uses L<dd(1)>
317 to copy between files and devices in the guest.
319 Example: duplicate the contents of an LV:
321 guestfs_dd (g, "/dev/VG/Original", "/dev/VG/Copy");
323 The destination (C</dev/VG/Copy>) must be at least as large as the
324 source (C</dev/VG/Original>). To copy less than the whole
325 source device, use L</guestfs_copy_size>.
327 =item B<file on the host> to B<file or device>
329 Use L</guestfs_upload>. See L</UPLOADING> above.
331 =item B<file or device> to B<file on the host>
333 Use L</guestfs_download>. See L</DOWNLOADING> above.
339 L</guestfs_ll> is just designed for humans to read (mainly when using
340 the L<guestfish(1)>-equivalent command C<ll>).
342 L</guestfs_ls> is a quick way to get a list of files in a directory
343 from programs, as a flat list of strings.
345 L</guestfs_readdir> is a programmatic way to get a list of files in a
346 directory, plus additional information about each one. It is more
347 equivalent to using the L<readdir(3)> call on a local filesystem.
349 L</guestfs_find> and L</guestfs_find0> can be used to recursively list
352 =head2 RUNNING COMMANDS
354 Although libguestfs is primarily an API for manipulating files
355 inside guest images, we also provide some limited facilities for
356 running commands inside guests.
358 There are many limitations to this:
364 The kernel version that the command runs under will be different
365 from what it expects.
369 If the command needs to communicate with daemons, then most likely
370 they won't be running.
374 The command will be running in limited memory.
378 The network may not be available unless you enable it
379 (see L</guestfs_set_network>).
383 Only supports Linux guests (not Windows, BSD, etc).
387 Architecture limitations (eg. won't work for a PPC guest on
392 For SELinux guests, you may need to enable SELinux and load policy
393 first. See L</SELINUX> in this manpage.
397 I<Security:> It is not safe to run commands from untrusted, possibly
398 malicious guests. These commands may attempt to exploit your program
399 by sending unexpected output. They could also try to exploit the
400 Linux kernel or qemu provided by the libguestfs appliance. They could
401 use the network provided by the libguestfs appliance to bypass
402 ordinary network partitions and firewalls. They could use the
403 elevated privileges or different SELinux context of your program
406 A secure alternative is to use libguestfs to install a "firstboot"
407 script (a script which runs when the guest next boots normally), and
408 to have this script run the commands you want in the normal context of
409 the running guest, network security and so on. For information about
410 other security issues, see L</SECURITY>.
414 The two main API calls to run commands are L</guestfs_command> and
415 L</guestfs_sh> (there are also variations).
417 The difference is that L</guestfs_sh> runs commands using the shell, so
418 any shell globs, redirections, etc will work.
420 =head2 CONFIGURATION FILES
422 To read and write configuration files in Linux guest filesystems, we
423 strongly recommend using Augeas. For example, Augeas understands how
424 to read and write, say, a Linux shadow password file or X.org
425 configuration file, and so avoids you having to write that code.
427 The main Augeas calls are bound through the C<guestfs_aug_*> APIs. We
428 don't document Augeas itself here because there is excellent
429 documentation on the L<http://augeas.net/> website.
431 If you don't want to use Augeas (you fool!) then try calling
432 L</guestfs_read_lines> to get the file as a list of lines which
433 you can iterate over.
437 We support SELinux guests. To ensure that labeling happens correctly
438 in SELinux guests, you need to enable SELinux and load the guest's
445 Before launching, do:
447 guestfs_set_selinux (g, 1);
451 After mounting the guest's filesystem(s), load the policy. This
452 is best done by running the L<load_policy(8)> command in the
455 guestfs_sh (g, "/usr/sbin/load_policy");
457 (Older versions of C<load_policy> require you to specify the
458 name of the policy file).
462 Optionally, set the security context for the API. The correct
463 security context to use can only be known by inspecting the
464 guest. As an example:
466 guestfs_setcon (g, "unconfined_u:unconfined_r:unconfined_t:s0");
470 This will work for running commands and editing existing files.
472 When new files are created, you may need to label them explicitly,
473 for example by running the external command
474 C<restorecon pathname>.
478 Certain calls are affected by the current file mode creation mask (the
479 "umask"). In particular ones which create files or directories, such
480 as L</guestfs_touch>, L</guestfs_mknod> or L</guestfs_mkdir>. This
481 affects either the default mode that the file is created with or
482 modifies the mode that you supply.
484 The default umask is C<022>, so files are created with modes such as
485 C<0644> and directories with C<0755>.
487 There are two ways to avoid being affected by umask. Either set umask
488 to 0 (call C<guestfs_umask (g, 0)> early after launching). Or call
489 L</guestfs_chmod> after creating each file or directory.
491 For more information about umask, see L<umask(2)>.
493 =head2 ENCRYPTED DISKS
495 Libguestfs allows you to access Linux guests which have been
496 encrypted using whole disk encryption that conforms to the
497 Linux Unified Key Setup (LUKS) standard. This includes
498 nearly all whole disk encryption systems used by modern
501 Use L</guestfs_vfs_type> to identify LUKS-encrypted block
502 devices (it returns the string C<crypto_LUKS>).
504 Then open these devices by calling L</guestfs_luks_open>.
505 Obviously you will require the passphrase!
507 Opening a LUKS device creates a new device mapper device
508 called C</dev/mapper/mapname> (where C<mapname> is the
509 string you supply to L</guestfs_luks_open>).
510 Reads and writes to this mapper device are decrypted from and
511 encrypted to the underlying block device respectively.
513 LVM volume groups on the device can be made visible by calling
514 L</guestfs_vgscan> followed by L</guestfs_vg_activate_all>.
515 The logical volume(s) can now be mounted in the usual way.
517 Use the reverse process to close a LUKS device. Unmount
518 any logical volumes on it, deactivate the volume groups
519 by caling C<guestfs_vg_activate (g, 0, ["/dev/VG"])>.
520 Then close the mapper device by calling
521 L</guestfs_luks_close> on the C</dev/mapper/mapname>
522 device (I<not> the underlying encrypted block device).
526 Libguestfs has APIs for inspecting an unknown disk image to find out
527 if it contains operating systems. (These APIs used to be in a
528 separate Perl-only library called L<Sys::Guestfs::Lib(3)> but since
529 version 1.5.3 the most frequently used part of this library has been
530 rewritten in C and moved into the core code).
532 Add all disks belonging to the unknown virtual machine and call
533 L</guestfs_launch> in the usual way.
535 Then call L</guestfs_inspect_os>. This function uses other libguestfs
536 calls and certain heuristics, and returns a list of operating systems
537 that were found. An empty list means none were found. A single
538 element is the root filesystem of the operating system. For dual- or
539 multi-boot guests, multiple roots can be returned, each one
540 corresponding to a separate operating system. (Multi-boot virtual
541 machines are extremely rare in the world of virtualization, but since
542 this scenario can happen, we have built libguestfs to deal with it.)
544 For each root, you can then call various C<guestfs_inspect_get_*>
545 functions to get additional details about that operating system. For
546 example, call L</guestfs_inspect_get_type> to return the string
547 C<windows> or C<linux> for Windows and Linux-based operating systems
550 Un*x-like and Linux-based operating systems usually consist of several
551 filesystems which are mounted at boot time (for example, a separate
552 boot partition mounted on C</boot>). The inspection rules are able to
553 detect how filesystems correspond to mount points. Call
554 C<guestfs_inspect_get_mountpoints> to get this mapping. It might
555 return a hash table like this example:
558 / => /dev/vg_guest/lv_root
559 /usr => /dev/vg_guest/lv_usr
561 The caller can then make calls to L</guestfs_mount_options> to
562 mount the filesystems as suggested.
564 Be careful to mount filesystems in the right order (eg. C</> before
565 C</usr>). Sorting the keys of the hash by length, shortest first,
568 Inspection currently only works for some common operating systems.
569 Contributors are welcome to send patches for other operating systems
570 that we currently cannot detect.
572 Encrypted disks must be opened before inspection. See
573 L</ENCRYPTED DISKS> for more details. The L</guestfs_inspect_os>
574 function just ignores any encrypted devices.
576 A note on the implementation: The call L</guestfs_inspect_os> performs
577 inspection and caches the results in the guest handle. Subsequent
578 calls to C<guestfs_inspect_get_*> return this cached information, but
579 I<do not> re-read the disks. If you change the content of the guest
580 disks, you can redo inspection by calling L</guestfs_inspect_os>
581 again. (L</guestfs_inspect_list_applications> works a little
582 differently from the other calls and does read the disks. See
583 documentation for that function for details).
585 =head2 SPECIAL CONSIDERATIONS FOR WINDOWS GUESTS
587 Libguestfs can mount NTFS partitions. It does this using the
588 L<http://www.ntfs-3g.org/> driver.
590 =head3 DRIVE LETTERS AND PATHS
592 DOS and Windows still use drive letters, and the filesystems are
593 always treated as case insensitive by Windows itself, and therefore
594 you might find a Windows configuration file referring to a path like
595 C<c:\windows\system32>. When the filesystem is mounted in libguestfs,
596 that directory might be referred to as C</WINDOWS/System32>.
598 Drive letter mappings are outside the scope of libguestfs. You have
599 to use libguestfs to read the appropriate Windows Registry and
600 configuration files, to determine yourself how drives are mapped (see
601 also L<hivex(3)> and L<virt-inspector(1)>).
603 Replacing backslash characters with forward slash characters is also
604 outside the scope of libguestfs, but something that you can easily do.
606 Where we can help is in resolving the case insensitivity of paths.
607 For this, call L</guestfs_case_sensitive_path>.
609 =head3 ACCESSING THE WINDOWS REGISTRY
611 Libguestfs also provides some help for decoding Windows Registry
612 "hive" files, through the library C<hivex> which is part of the
613 libguestfs project although ships as a separate tarball. You have to
614 locate and download the hive file(s) yourself, and then pass them to
615 C<hivex> functions. See also the programs L<hivexml(1)>,
616 L<hivexsh(1)>, L<hivexregedit(1)> and L<virt-win-reg(1)> for more help
619 =head3 SYMLINKS ON NTFS-3G FILESYSTEMS
621 Ntfs-3g tries to rewrite "Junction Points" and NTFS "symbolic links"
622 to provide something which looks like a Linux symlink. The way it
623 tries to do the rewriting is described here:
625 L<http://www.tuxera.com/community/ntfs-3g-advanced/junction-points-and-symbolic-links/>
627 The essential problem is that ntfs-3g simply does not have enough
628 information to do a correct job. NTFS links can contain drive letters
629 and references to external device GUIDs that ntfs-3g has no way of
630 resolving. It is almost certainly the case that libguestfs callers
631 should ignore what ntfs-3g does (ie. don't use L</guestfs_readlink> on
634 Instead if you encounter a symbolic link on an ntfs-3g filesystem, use
635 L</guestfs_lgetxattr> to read the C<system.ntfs_reparse_data> extended
636 attribute, and read the raw reparse data from that (you can find the
637 format documented in various places around the web).
639 =head3 EXTENDED ATTRIBUTES ON NTFS-3G FILESYSTEMS
641 There are other useful extended attributes that can be read from
642 ntfs-3g filesystems (using L</guestfs_getxattr>). See:
644 L<http://www.tuxera.com/community/ntfs-3g-advanced/extended-attributes/>
646 =head2 USING LIBGUESTFS WITH OTHER PROGRAMMING LANGUAGES
648 Although we don't want to discourage you from using the C API, we will
649 mention here that the same API is also available in other languages.
651 The API is broadly identical in all supported languages. This means
652 that the C call C<guestfs_mount(g,path)> is
653 C<$g-E<gt>mount($path)> in Perl, C<g.mount(path)> in Python,
654 and C<Guestfs.mount g path> in OCaml. In other words, a
655 straightforward, predictable isomorphism between each language.
657 Error messages are automatically transformed
658 into exceptions if the language supports it.
660 We don't try to "object orientify" parts of the API in OO languages,
661 although contributors are welcome to write higher level APIs above
662 what we provide in their favourite languages if they wish.
668 You can use the I<guestfs.h> header file from C++ programs. The C++
669 API is identical to the C API. C++ classes and exceptions are not
674 The C# bindings are highly experimental. Please read the warnings
675 at the top of C<csharp/Libguestfs.cs>.
679 This is the only language binding that is working but incomplete.
680 Only calls which return simple integers have been bound in Haskell,
681 and we are looking for help to complete this binding.
685 Full documentation is contained in the Javadoc which is distributed
690 For documentation see L<guestfs-ocaml(3)>.
694 For documentation see L<Sys::Guestfs(3)>.
698 For documentation see C<README-PHP> supplied with libguestfs
699 sources or in the php-libguestfs package for your distribution.
701 The PHP binding only works correctly on 64 bit machines.
705 For documentation see L<guestfs-python(3)>.
709 For documentation see L<guestfs-ruby(3)>.
711 =item B<shell scripts>
713 For documentation see L<guestfish(1)>.
717 =head2 LIBGUESTFS GOTCHAS
719 L<http://en.wikipedia.org/wiki/Gotcha_(programming)>: "A feature of a
720 system [...] that works in the way it is documented but is
721 counterintuitive and almost invites mistakes."
723 Since we developed libguestfs and the associated tools, there are
724 several things we would have designed differently, but are now stuck
725 with for backwards compatibility or other reasons. If there is ever a
726 libguestfs 2.0 release, you can expect these to change. Beware of
731 =item Autosync / forgetting to sync.
733 When modifying a filesystem from C or another language, you B<must>
734 unmount all filesystems and call L</guestfs_sync> explicitly before
735 you close the libguestfs handle. You can also call:
737 guestfs_set_autosync (g, 1);
739 to have the unmount/sync done automatically for you when the handle 'g'
740 is closed. (This feature is called "autosync", L</guestfs_set_autosync>
743 If you forget to do this, then it is entirely possible that your
744 changes won't be written out, or will be partially written, or (very
745 rarely) that you'll get disk corruption.
747 Note that in L<guestfish(3)> autosync is the default. So quick and
748 dirty guestfish scripts that forget to sync will work just fine, which
749 can make this very puzzling if you are trying to debug a problem.
751 Update: Autosync is enabled by default for all API users starting from
754 =item Mount option C<-o sync> should not be the default.
756 If you use L</guestfs_mount>, then C<-o sync,noatime> are added
757 implicitly. However C<-o sync> does not add any reliability benefit,
758 but does have a very large performance impact.
760 The work around is to use L</guestfs_mount_options> and set the mount
761 options that you actually want to use.
763 =item Read-only should be the default.
765 In L<guestfish(3)>, I<--ro> should be the default, and you should
766 have to specify I<--rw> if you want to make changes to the image.
768 This would reduce the potential to corrupt live VM images.
770 Note that many filesystems change the disk when you just mount and
771 unmount, even if you didn't perform any writes. You need to use
772 L</guestfs_add_drive_ro> to guarantee that the disk is not changed.
774 =item guestfish command line is hard to use.
776 C<guestfish disk.img> doesn't do what people expect (open C<disk.img>
777 for examination). It tries to run a guestfish command C<disk.img>
778 which doesn't exist, so it fails. In earlier versions of guestfish
779 the error message was also unintuitive, but we have corrected this
780 since. Like the Bourne shell, we should have used C<guestfish -c
781 command> to run commands.
783 =item guestfish megabyte modifiers don't work right on all commands
785 In recent guestfish you can use C<1M> to mean 1 megabyte (and
786 similarly for other modifiers). What guestfish actually does is to
787 multiply the number part by the modifier part and pass the result to
788 the C API. However this doesn't work for a few APIs which aren't
789 expecting bytes, but are already expecting some other unit
792 The most common is L</guestfs_lvcreate>. The guestfish command:
796 does not do what you might expect. Instead because
797 L</guestfs_lvcreate> is already expecting megabytes, this tries to
798 create a 100 I<terabyte> (100 megabytes * megabytes) logical volume.
799 The error message you get from this is also a little obscure.
801 This could be fixed in the generator by specially marking parameters
802 and return values which take bytes or other units.
804 =item Ambiguity between devices and paths
806 There is a subtle ambiguity in the API between a device name
807 (eg. C</dev/sdb2>) and a similar pathname. A file might just happen
808 to be called C<sdb2> in the directory C</dev> (consider some non-Unix
811 In the current API we usually resolve this ambiguity by having two
812 separate calls, for example L</guestfs_checksum> and
813 L</guestfs_checksum_device>. Some API calls are ambiguous and
814 (incorrectly) resolve the problem by detecting if the path supplied
815 begins with C</dev/>.
817 To avoid both the ambiguity and the need to duplicate some calls, we
818 could make paths/devices into structured names. One way to do this
819 would be to use a notation like grub (C<hd(0,0)>), although nobody
820 really likes this aspect of grub. Another way would be to use a
821 structured type, equivalent to this OCaml type:
823 type path = Path of string | Device of int | Partition of int * int
825 which would allow you to pass arguments like:
828 Device 1 (* /dev/sdb, or perhaps /dev/sda *)
829 Partition (1, 2) (* /dev/sdb2 (or is it /dev/sda2 or /dev/sdb3?) *)
830 Path "/dev/sdb2" (* not a device *)
832 As you can see there are still problems to resolve even with this
833 representation. Also consider how it might work in guestfish.
837 =head2 PROTOCOL LIMITS
839 Internally libguestfs uses a message-based protocol to pass API calls
840 and their responses to and from a small "appliance" (see L</INTERNALS>
841 for plenty more detail about this). The maximum message size used by
842 the protocol is slightly less than 4 MB. For some API calls you may
843 need to be aware of this limit. The API calls which may be affected
844 are individually documented, with a link back to this section of the
847 A simple call such as L</guestfs_cat> returns its result (the file
848 data) in a simple string. Because this string is at some point
849 internally encoded as a message, the maximum size that it can return
850 is slightly under 4 MB. If the requested file is larger than this
851 then you will get an error.
853 In order to transfer large files into and out of the guest filesystem,
854 you need to use particular calls that support this. The sections
855 L</UPLOADING> and L</DOWNLOADING> document how to do this.
857 You might also consider mounting the disk image using our FUSE
858 filesystem support (L<guestmount(1)>).
860 =head2 KEYS AND PASSPHRASES
862 Certain libguestfs calls take a parameter that contains sensitive key
863 material, passed in as a C string.
865 In the future we would hope to change the libguestfs implementation so
866 that keys are L<mlock(2)>-ed into physical RAM, and thus can never end
867 up in swap. However this is I<not> done at the moment, because of the
868 complexity of such an implementation.
870 Therefore you should be aware that any key parameter you pass to
871 libguestfs might end up being written out to the swap partition. If
872 this is a concern, scrub the swap partition or don't use libguestfs on
875 =head2 MULTIPLE HANDLES AND MULTIPLE THREADS
877 All high-level libguestfs actions are synchronous. If you want
878 to use libguestfs asynchronously then you must create a thread.
880 Only use the handle from a single thread. Either use the handle
881 exclusively from one thread, or provide your own mutex so that two
882 threads cannot issue calls on the same handle at the same time.
884 See the graphical program guestfs-browser for one possible
885 architecture for multithreaded programs using libvirt and libguestfs.
889 Libguestfs needs a kernel and initrd.img, which it finds by looking
890 along an internal path.
892 By default it looks for these in the directory C<$libdir/guestfs>
893 (eg. C</usr/local/lib/guestfs> or C</usr/lib64/guestfs>).
895 Use L</guestfs_set_path> or set the environment variable
896 L</LIBGUESTFS_PATH> to change the directories that libguestfs will
897 search in. The value is a colon-separated list of paths. The current
898 directory is I<not> searched unless the path contains an empty element
899 or C<.>. For example C<LIBGUESTFS_PATH=:/usr/lib/guestfs> would
900 search the current directory and then C</usr/lib/guestfs>.
904 If you want to compile your own qemu, run qemu from a non-standard
905 location, or pass extra arguments to qemu, then you can write a
906 shell-script wrapper around qemu.
908 There is one important rule to remember: you I<must C<exec qemu>> as
909 the last command in the shell script (so that qemu replaces the shell
910 and becomes the direct child of the libguestfs-using program). If you
911 don't do this, then the qemu process won't be cleaned up correctly.
913 Here is an example of a wrapper, where I have built my own copy of
917 qemudir=/home/rjones/d/qemu
918 exec $qemudir/x86_64-softmmu/qemu-system-x86_64 -L $qemudir/pc-bios "$@"
920 Save this script as C</tmp/qemu.wrapper> (or wherever), C<chmod +x>,
921 and then use it by setting the LIBGUESTFS_QEMU environment variable.
924 LIBGUESTFS_QEMU=/tmp/qemu.wrapper guestfish
926 Note that libguestfs also calls qemu with the -help and -version
927 options in order to determine features.
931 We guarantee the libguestfs ABI (binary interface), for public,
932 high-level actions as outlined in this section. Although we will
933 deprecate some actions, for example if they get replaced by newer
934 calls, we will keep the old actions forever. This allows you the
935 developer to program in confidence against the libguestfs API.
937 =head2 BLOCK DEVICE NAMING
939 In the kernel there is now quite a profusion of schemata for naming
940 block devices (in this context, by I<block device> I mean a physical
941 or virtual hard drive). The original Linux IDE driver used names
942 starting with C</dev/hd*>. SCSI devices have historically used a
943 different naming scheme, C</dev/sd*>. When the Linux kernel I<libata>
944 driver became a popular replacement for the old IDE driver
945 (particularly for SATA devices) those devices also used the
946 C</dev/sd*> scheme. Additionally we now have virtual machines with
947 paravirtualized drivers. This has created several different naming
948 systems, such as C</dev/vd*> for virtio disks and C</dev/xvd*> for Xen
951 As discussed above, libguestfs uses a qemu appliance running an
952 embedded Linux kernel to access block devices. We can run a variety
953 of appliances based on a variety of Linux kernels.
955 This causes a problem for libguestfs because many API calls use device
956 or partition names. Working scripts and the recipe (example) scripts
957 that we make available over the internet could fail if the naming
960 Therefore libguestfs defines C</dev/sd*> as the I<standard naming
961 scheme>. Internally C</dev/sd*> names are translated, if necessary,
962 to other names as required. For example, under RHEL 5 which uses the
963 C</dev/hd*> scheme, any device parameter C</dev/sda2> is translated to
964 C</dev/hda2> transparently.
966 Note that this I<only> applies to parameters. The
967 L</guestfs_list_devices>, L</guestfs_list_partitions> and similar calls
968 return the true names of the devices and partitions as known to the
971 =head3 ALGORITHM FOR BLOCK DEVICE NAME TRANSLATION
973 Usually this translation is transparent. However in some (very rare)
974 cases you may need to know the exact algorithm. Such cases include
975 where you use L</guestfs_config> to add a mixture of virtio and IDE
976 devices to the qemu-based appliance, so have a mixture of C</dev/sd*>
977 and C</dev/vd*> devices.
979 The algorithm is applied only to I<parameters> which are known to be
980 either device or partition names. Return values from functions such
981 as L</guestfs_list_devices> are never changed.
987 Is the string a parameter which is a device or partition name?
991 Does the string begin with C</dev/sd>?
995 Does the named device exist? If so, we use that device.
996 However if I<not> then we continue with this algorithm.
1000 Replace initial C</dev/sd> string with C</dev/hd>.
1002 For example, change C</dev/sda2> to C</dev/hda2>.
1004 If that named device exists, use it. If not, continue.
1008 Replace initial C</dev/sd> string with C</dev/vd>.
1010 If that named device exists, use it. If not, return an error.
1014 =head3 PORTABILITY CONCERNS WITH BLOCK DEVICE NAMING
1016 Although the standard naming scheme and automatic translation is
1017 useful for simple programs and guestfish scripts, for larger programs
1018 it is best not to rely on this mechanism.
1020 Where possible for maximum future portability programs using
1021 libguestfs should use these future-proof techniques:
1027 Use L</guestfs_list_devices> or L</guestfs_list_partitions> to list
1028 actual device names, and then use those names directly.
1030 Since those device names exist by definition, they will never be
1035 Use higher level ways to identify filesystems, such as LVM names,
1036 UUIDs and filesystem labels.
1042 This section discusses security implications of using libguestfs,
1043 particularly with untrusted or malicious guests or disk images.
1045 =head2 GENERAL SECURITY CONSIDERATIONS
1047 Be careful with any files or data that you download from a guest (by
1048 "download" we mean not just the L</guestfs_download> command but any
1049 command that reads files, filenames, directories or anything else from
1050 a disk image). An attacker could manipulate the data to fool your
1051 program into doing the wrong thing. Consider cases such as:
1057 the data (file etc) not being present
1061 being present but empty
1065 being much larger than normal
1069 containing arbitrary 8 bit data
1073 being in an unexpected character encoding
1077 containing homoglyphs.
1081 =head2 SECURITY OF MOUNTING FILESYSTEMS
1083 When you mount a filesystem under Linux, mistakes in the kernel
1084 filesystem (VFS) module can sometimes be escalated into exploits by
1085 deliberately creating a malicious, malformed filesystem. These
1086 exploits are very severe for two reasons. Firstly there are very many
1087 filesystem drivers in the kernel, and many of them are infrequently
1088 used and not much developer attention has been paid to the code.
1089 Linux userspace helps potential crackers by detecting the filesystem
1090 type and automatically choosing the right VFS driver, even if that
1091 filesystem type is obscure or unexpected for the administrator.
1092 Secondly, a kernel-level exploit is like a local root exploit (worse
1093 in some ways), giving immediate and total access to the system right
1094 down to the hardware level.
1096 That explains why you should never mount a filesystem from an
1097 untrusted guest on your host kernel. How about libguestfs? We run a
1098 Linux kernel inside a qemu virtual machine, usually running as a
1099 non-root user. The attacker would need to write a filesystem which
1100 first exploited the kernel, and then exploited either qemu
1101 virtualization (eg. a faulty qemu driver) or the libguestfs protocol,
1102 and finally to be as serious as the host kernel exploit it would need
1103 to escalate its privileges to root. This multi-step escalation,
1104 performed by a static piece of data, is thought to be extremely hard
1105 to do, although we never say 'never' about security issues.
1107 In any case callers can reduce the attack surface by forcing the
1108 filesystem type when mounting (use L</guestfs_mount_vfs>).
1110 =head2 PROTOCOL SECURITY
1112 The protocol is designed to be secure, being based on RFC 4506 (XDR)
1113 with a defined upper message size. However a program that uses
1114 libguestfs must also take care - for example you can write a program
1115 that downloads a binary from a disk image and executes it locally, and
1116 no amount of protocol security will save you from the consequences.
1118 =head2 INSPECTION SECURITY
1120 Parts of the inspection API (see L</INSPECTION>) return untrusted
1121 strings directly from the guest, and these could contain any 8 bit
1122 data. Callers should be careful to escape these before printing them
1123 to a structured file (for example, use HTML escaping if creating a web
1126 Guest configuration may be altered in unusual ways by the
1127 administrator of the virtual machine, and may not reflect reality
1128 (particularly for untrusted or actively malicious guests). For
1129 example we parse the hostname from configuration files like
1130 C</etc/sysconfig/network> that we find in the guest, but the guest
1131 administrator can easily manipulate these files to provide the wrong
1134 The inspection API parses guest configuration using two external
1135 libraries: Augeas (Linux configuration) and hivex (Windows Registry).
1136 Both are designed to be robust in the face of malicious data, although
1137 denial of service attacks are still possible, for example with
1138 oversized configuration files.
1140 =head2 RUNNING UNTRUSTED GUEST COMMANDS
1142 Be very cautious about running commands from the guest. By running a
1143 command in the guest, you are giving CPU time to a binary that you do
1144 not control, under the same user account as the library, albeit
1145 wrapped in qemu virtualization. More information and alternatives can
1146 be found in the section L</RUNNING COMMANDS>.
1148 =head2 CVE-2010-3851
1150 https://bugzilla.redhat.com/642934
1152 This security bug concerns the automatic disk format detection that
1153 qemu does on disk images.
1155 A raw disk image is just the raw bytes, there is no header. Other
1156 disk images like qcow2 contain a special header. Qemu deals with this
1157 by looking for one of the known headers, and if none is found then
1158 assuming the disk image must be raw.
1160 This allows a guest which has been given a raw disk image to write
1161 some other header. At next boot (or when the disk image is accessed
1162 by libguestfs) qemu would do autodetection and think the disk image
1163 format was, say, qcow2 based on the header written by the guest.
1165 This in itself would not be a problem, but qcow2 offers many features,
1166 one of which is to allow a disk image to refer to another image
1167 (called the "backing disk"). It does this by placing the path to the
1168 backing disk into the qcow2 header. This path is not validated and
1169 could point to any host file (eg. "/etc/passwd"). The backing disk is
1170 then exposed through "holes" in the qcow2 disk image, which of course
1171 is completely under the control of the attacker.
1173 In libguestfs this is rather hard to exploit except under two
1180 You have enabled the network or have opened the disk in write mode.
1184 You are also running untrusted code from the guest (see
1185 L</RUNNING COMMANDS>).
1189 The way to avoid this is to specify the expected disk format when
1190 adding disks (the optional C<format> option to
1191 L</guestfs_add_drive_opts>). You should always do this if the disk is
1192 raw format, and it's a good idea for other cases too.
1194 For disks added from libvirt using calls like L</guestfs_add_domain>,
1195 the format is fetched from libvirt and passed through.
1197 For libguestfs tools, use the I<--format> command line parameter as
1200 =head1 CONNECTION MANAGEMENT
1204 C<guestfs_h> is the opaque type representing a connection handle.
1205 Create a handle by calling L</guestfs_create>. Call L</guestfs_close>
1206 to free the handle and release all resources used.
1208 For information on using multiple handles and threads, see the section
1209 L</MULTIPLE HANDLES AND MULTIPLE THREADS> below.
1211 =head2 guestfs_create
1213 guestfs_h *guestfs_create (void);
1215 Create a connection handle.
1217 You have to call L</guestfs_add_drive_opts> (or one of the equivalent
1218 calls) on the handle at least once.
1220 This function returns a non-NULL pointer to a handle on success or
1223 After configuring the handle, you have to call L</guestfs_launch>.
1225 You may also want to configure error handling for the handle. See
1226 L</ERROR HANDLING> section below.
1228 =head2 guestfs_close
1230 void guestfs_close (guestfs_h *g);
1232 This closes the connection handle and frees up all resources used.
1234 =head1 ERROR HANDLING
1236 API functions can return errors. For example, almost all functions
1237 that return C<int> will return C<-1> to indicate an error.
1239 Additional information is available for errors: an error message
1240 string and optionally an error number (errno) if the thing that failed
1243 You can get at the additional information about the last error on the
1244 handle by calling L</guestfs_last_error>, L</guestfs_last_errno>,
1245 and/or by setting up an error handler with
1246 L</guestfs_set_error_handler>.
1248 When the handle is created, a default error handler is installed which
1249 prints the error message string to C<stderr>. For small short-running
1250 command line programs it is sufficient to do:
1252 if (guestfs_launch (g) == -1)
1253 exit (EXIT_FAILURE);
1255 since the default error handler will ensure that an error message has
1256 been printed to C<stderr> before the program exits.
1258 For other programs the caller will almost certainly want to install an
1259 alternate error handler or do error handling in-line like this:
1261 g = guestfs_create ();
1263 /* This disables the default behaviour of printing errors
1265 guestfs_set_error_handler (g, NULL, NULL);
1267 if (guestfs_launch (g) == -1) {
1268 /* Examine the error message and print it etc. */
1269 char *msg = guestfs_last_error (g);
1270 int errnum = guestfs_last_errno (g);
1271 fprintf (stderr, "%s\n", msg);
1275 Out of memory errors are handled differently. The default action is
1276 to call L<abort(3)>. If this is undesirable, then you can set a
1277 handler using L</guestfs_set_out_of_memory_handler>.
1279 L</guestfs_create> returns C<NULL> if the handle cannot be created,
1280 and because there is no handle if this happens there is no way to get
1281 additional error information. However L</guestfs_create> is supposed
1282 to be a lightweight operation which can only fail because of
1283 insufficient memory (it returns NULL in this case).
1285 =head2 guestfs_last_error
1287 const char *guestfs_last_error (guestfs_h *g);
1289 This returns the last error message that happened on C<g>. If
1290 there has not been an error since the handle was created, then this
1293 The lifetime of the returned string is until the next error occurs, or
1294 L</guestfs_close> is called.
1296 =head2 guestfs_last_errno
1298 int guestfs_last_errno (guestfs_h *g);
1300 This returns the last error number (errno) that happened on C<g>.
1302 If successful, an errno integer not equal to zero is returned.
1304 If no error, this returns 0. This call can return 0 in three
1311 There has not been any error on the handle.
1315 There has been an error but the errno was meaningless. This
1316 corresponds to the case where the error did not come from a
1317 failed system call, but for some other reason.
1321 There was an error from a failed system call, but for some
1322 reason the errno was not captured and returned. This usually
1323 indicates a bug in libguestfs.
1327 Libguestfs tries to convert the errno from inside the applicance into
1328 a corresponding errno for the caller (not entirely trivial: the
1329 appliance might be running a completely different operating system
1330 from the library and error numbers are not standardized across
1331 Un*xen). If this could not be done, then the error is translated to
1332 C<EINVAL>. In practice this should only happen in very rare
1335 =head2 guestfs_set_error_handler
1337 typedef void (*guestfs_error_handler_cb) (guestfs_h *g,
1340 void guestfs_set_error_handler (guestfs_h *g,
1341 guestfs_error_handler_cb cb,
1344 The callback C<cb> will be called if there is an error. The
1345 parameters passed to the callback are an opaque data pointer and the
1346 error message string.
1348 C<errno> is not passed to the callback. To get that the callback must
1349 call L</guestfs_last_errno>.
1351 Note that the message string C<msg> is freed as soon as the callback
1352 function returns, so if you want to stash it somewhere you must make
1355 The default handler prints messages on C<stderr>.
1357 If you set C<cb> to C<NULL> then I<no> handler is called.
1359 =head2 guestfs_get_error_handler
1361 guestfs_error_handler_cb guestfs_get_error_handler (guestfs_h *g,
1364 Returns the current error handler callback.
1366 =head2 guestfs_set_out_of_memory_handler
1368 typedef void (*guestfs_abort_cb) (void);
1369 int guestfs_set_out_of_memory_handler (guestfs_h *g,
1372 The callback C<cb> will be called if there is an out of memory
1373 situation. I<Note this callback must not return>.
1375 The default is to call L<abort(3)>.
1377 You cannot set C<cb> to C<NULL>. You can't ignore out of memory
1380 =head2 guestfs_get_out_of_memory_handler
1382 guestfs_abort_fn guestfs_get_out_of_memory_handler (guestfs_h *g);
1384 This returns the current out of memory handler.
1396 =head2 GROUPS OF FUNCTIONALITY IN THE APPLIANCE
1398 Using L</guestfs_available> you can test availability of
1399 the following groups of functions. This test queries the
1400 appliance to see if the appliance you are currently using
1401 supports the functionality.
1405 =head2 GUESTFISH supported COMMAND
1407 In L<guestfish(3)> there is a handy interactive command
1408 C<supported> which prints out the available groups and
1409 whether they are supported by this build of libguestfs.
1410 Note however that you have to do C<run> first.
1412 =head2 SINGLE CALLS AT COMPILE TIME
1414 Since version 1.5.8, C<E<lt>guestfs.hE<gt>> defines symbols
1415 for each C API function, such as:
1417 #define LIBGUESTFS_HAVE_DD 1
1419 if L</guestfs_dd> is available.
1421 Before version 1.5.8, if you needed to test whether a single
1422 libguestfs function is available at compile time, we recommended using
1423 build tools such as autoconf or cmake. For example in autotools you
1426 AC_CHECK_LIB([guestfs],[guestfs_create])
1427 AC_CHECK_FUNCS([guestfs_dd])
1429 which would result in C<HAVE_GUESTFS_DD> being either defined
1430 or not defined in your program.
1432 =head2 SINGLE CALLS AT RUN TIME
1434 Testing at compile time doesn't guarantee that a function really
1435 exists in the library. The reason is that you might be dynamically
1436 linked against a previous I<libguestfs.so> (dynamic library)
1437 which doesn't have the call. This situation unfortunately results
1438 in a segmentation fault, which is a shortcoming of the C dynamic
1439 linking system itself.
1441 You can use L<dlopen(3)> to test if a function is available
1442 at run time, as in this example program (note that you still
1443 need the compile time check as well):
1449 #include <guestfs.h>
1453 #ifdef LIBGUESTFS_HAVE_DD
1457 /* Test if the function guestfs_dd is really available. */
1458 dl = dlopen (NULL, RTLD_LAZY);
1460 fprintf (stderr, "dlopen: %s\n", dlerror ());
1461 exit (EXIT_FAILURE);
1463 has_function = dlsym (dl, "guestfs_dd") != NULL;
1467 printf ("this libguestfs.so does NOT have guestfs_dd function\n");
1469 printf ("this libguestfs.so has guestfs_dd function\n");
1470 /* Now it's safe to call
1471 guestfs_dd (g, "foo", "bar");
1475 printf ("guestfs_dd function was not found at compile time\n");
1479 You may think the above is an awful lot of hassle, and it is.
1480 There are other ways outside of the C linking system to ensure
1481 that this kind of incompatibility never arises, such as using
1484 Requires: libguestfs >= 1.0.80
1486 =head1 CALLS WITH OPTIONAL ARGUMENTS
1488 A recent feature of the API is the introduction of calls which take
1489 optional arguments. In C these are declared 3 ways. The main way is
1490 as a call which takes variable arguments (ie. C<...>), as in this
1493 int guestfs_add_drive_opts (guestfs_h *g, const char *filename, ...);
1495 Call this with a list of optional arguments, terminated by C<-1>.
1496 So to call with no optional arguments specified:
1498 guestfs_add_drive_opts (g, filename, -1);
1500 With a single optional argument:
1502 guestfs_add_drive_opts (g, filename,
1503 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "qcow2",
1508 guestfs_add_drive_opts (g, filename,
1509 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "qcow2",
1510 GUESTFS_ADD_DRIVE_OPTS_READONLY, 1,
1513 and so forth. Don't forget the terminating C<-1> otherwise
1514 Bad Things will happen!
1516 =head2 USING va_list FOR OPTIONAL ARGUMENTS
1518 The second variant has the same name with the suffix C<_va>, which
1519 works the same way but takes a C<va_list>. See the C manual for
1520 details. For the example function, this is declared:
1522 int guestfs_add_drive_opts_va (guestfs_h *g, const char *filename,
1525 =head2 CONSTRUCTING OPTIONAL ARGUMENTS
1527 The third variant is useful where you need to construct these
1528 calls. You pass in a structure where you fill in the optional
1529 fields. The structure has a bitmask as the first element which
1530 you must set to indicate which fields you have filled in. For
1531 our example function the structure and call are declared:
1533 struct guestfs_add_drive_opts_argv {
1539 int guestfs_add_drive_opts_argv (guestfs_h *g, const char *filename,
1540 const struct guestfs_add_drive_opts_argv *optargs);
1542 You could call it like this:
1544 struct guestfs_add_drive_opts_argv optargs = {
1545 .bitmask = GUESTFS_ADD_DRIVE_OPTS_READONLY_BITMASK |
1546 GUESTFS_ADD_DRIVE_OPTS_FORMAT_BITMASK,
1551 guestfs_add_drive_opts_argv (g, filename, &optargs);
1559 The C<_BITMASK> suffix on each option name when specifying the
1564 You do not need to fill in all fields of the structure.
1568 There must be a one-to-one correspondence between fields of the
1569 structure that are filled in, and bits set in the bitmask.
1573 =head2 OPTIONAL ARGUMENTS IN OTHER LANGUAGES
1575 In other languages, optional arguments are expressed in the
1576 way that is natural for that language. We refer you to the
1577 language-specific documentation for more details on that.
1579 For guestfish, see L<guestfish(1)/OPTIONAL ARGUMENTS>.
1581 =head2 SETTING CALLBACKS TO HANDLE EVENTS
1583 The child process generates events in some situations. Current events
1584 include: receiving a log message, the child process exits.
1586 Use the C<guestfs_set_*_callback> functions to set a callback for
1587 different types of events.
1589 Only I<one callback of each type> can be registered for each handle.
1590 Calling C<guestfs_set_*_callback> again overwrites the previous
1591 callback of that type. Cancel all callbacks of this type by calling
1592 this function with C<cb> set to C<NULL>.
1594 =head2 guestfs_set_log_message_callback
1596 typedef void (*guestfs_log_message_cb) (guestfs_h *g, void *opaque,
1597 char *buf, int len);
1598 void guestfs_set_log_message_callback (guestfs_h *g,
1599 guestfs_log_message_cb cb,
1602 The callback function C<cb> will be called whenever qemu or the guest
1603 writes anything to the console.
1605 Use this function to capture kernel messages and similar.
1607 Normally there is no log message handler, and log messages are just
1610 =head2 guestfs_set_subprocess_quit_callback
1612 typedef void (*guestfs_subprocess_quit_cb) (guestfs_h *g, void *opaque);
1613 void guestfs_set_subprocess_quit_callback (guestfs_h *g,
1614 guestfs_subprocess_quit_cb cb,
1617 The callback function C<cb> will be called when the child process
1618 quits, either asynchronously or if killed by
1619 L</guestfs_kill_subprocess>. (This corresponds to a transition from
1620 any state to the CONFIG state).
1622 =head2 guestfs_set_launch_done_callback
1624 typedef void (*guestfs_launch_done_cb) (guestfs_h *g, void *opaque);
1625 void guestfs_set_launch_done_callback (guestfs_h *g,
1626 guestfs_launch_done_cb cb,
1629 The callback function C<cb> will be called when the child process
1630 becomes ready first time after it has been launched. (This
1631 corresponds to a transition from LAUNCHING to the READY state).
1633 =head2 guestfs_set_close_callback
1635 typedef void (*guestfs_close_cb) (guestfs_h *g, void *opaque);
1636 void guestfs_set_close_callback (guestfs_h *g,
1637 guestfs_close_cb cb,
1640 The callback function C<cb> will be called while the handle
1641 is being closed (synchronously from L</guestfs_close>).
1643 Note that libguestfs installs an L<atexit(3)> handler to try to
1644 clean up handles that are open when the program exits. This
1645 means that this callback might be called indirectly from
1646 L<exit(3)>, which can cause unexpected problems in higher-level
1647 languages (eg. if your HLL interpreter has already been cleaned
1648 up by the time this is called, and if your callback then jumps
1649 into some HLL function).
1651 =head2 guestfs_set_progress_callback
1653 typedef void (*guestfs_progress_cb) (guestfs_h *g, void *opaque,
1654 int proc_nr, int serial,
1655 uint64_t position, uint64_t total);
1656 void guestfs_set_progress_callback (guestfs_h *g,
1657 guestfs_progress_cb cb,
1660 Some long-running operations can generate progress messages. If
1661 this callback is registered, then it will be called each time a
1662 progress message is generated (usually two seconds after the
1663 operation started, and three times per second thereafter until
1664 it completes, although the frequency may change in future versions).
1666 The callback receives two numbers: C<position> and C<total>.
1667 The units of C<total> are not defined, although for some
1668 operations C<total> may relate in some way to the amount of
1669 data to be transferred (eg. in bytes or megabytes), and
1670 C<position> may be the portion which has been transferred.
1672 The only defined and stable parts of the API are:
1678 The callback can display to the user some type of progress bar or
1679 indicator which shows the ratio of C<position>:C<total>.
1683 0 E<lt>= C<position> E<lt>= C<total>
1687 If any progress notification is sent during a call, then a final
1688 progress notification is always sent when C<position> = C<total>.
1690 This is to simplify caller code, so callers can easily set the
1691 progress indicator to "100%" at the end of the operation, without
1692 requiring special code to detect this case.
1696 The callback also receives the procedure number and serial number of
1697 the call. These are only useful for debugging protocol issues, and
1698 the callback can normally ignore them. The callback may want to
1699 print these numbers in error messages or debugging messages.
1701 =head1 PRIVATE DATA AREA
1703 You can attach named pieces of private data to the libguestfs handle,
1704 and fetch them by name for the lifetime of the handle. This is called
1705 the private data area and is only available from the C API.
1707 To attach a named piece of data, use the following call:
1709 void guestfs_set_private (guestfs_h *g, const char *key, void *data);
1711 C<key> is the name to associate with this data, and C<data> is an
1712 arbitrary pointer (which can be C<NULL>). Any previous item with the
1713 same name is overwritten.
1715 You can use any C<key> you want, but names beginning with an
1716 underscore character are reserved for internal libguestfs purposes
1717 (for implementing language bindings). It is recommended to prefix the
1718 name with some unique string to avoid collisions with other users.
1720 To retrieve the pointer, use:
1722 void *guestfs_get_private (guestfs_h *g, const char *key);
1724 This function returns C<NULL> if either no data is found associated
1725 with C<key>, or if the user previously set the C<key>'s C<data>
1728 Libguestfs does not try to look at or interpret the C<data> pointer in
1729 any way. As far as libguestfs is concerned, it need not be a valid
1730 pointer at all. In particular, libguestfs does I<not> try to free the
1731 data when the handle is closed. If the data must be freed, then the
1732 caller must either free it before calling L</guestfs_close> or must
1733 set up a close callback to do it (see L</guestfs_set_close_callback>,
1734 and note that only one callback can be registered for a handle).
1736 The private data area is implemented using a hash table, and should be
1737 reasonably efficient for moderate numbers of keys.
1741 <!-- old anchor for the next section -->
1742 <a name="state_machine_and_low_level_event_api"/>
1748 Internally, libguestfs is implemented by running an appliance (a
1749 special type of small virtual machine) using L<qemu(1)>. Qemu runs as
1750 a child process of the main program.
1756 | | child process / appliance
1757 | | __________________________
1759 +-------------------+ RPC | +-----------------+ |
1760 | libguestfs <--------------------> guestfsd | |
1761 | | | +-----------------+ |
1762 \___________________/ | | Linux kernel | |
1763 | +--^--------------+ |
1764 \_________|________________/
1772 The library, linked to the main program, creates the child process and
1773 hence the appliance in the L</guestfs_launch> function.
1775 Inside the appliance is a Linux kernel and a complete stack of
1776 userspace tools (such as LVM and ext2 programs) and a small
1777 controlling daemon called L</guestfsd>. The library talks to
1778 L</guestfsd> using remote procedure calls (RPC). There is a mostly
1779 one-to-one correspondence between libguestfs API calls and RPC calls
1780 to the daemon. Lastly the disk image(s) are attached to the qemu
1781 process which translates device access by the appliance's Linux kernel
1782 into accesses to the image.
1784 A common misunderstanding is that the appliance "is" the virtual
1785 machine. Although the disk image you are attached to might also be
1786 used by some virtual machine, libguestfs doesn't know or care about
1787 this. (But you will care if both libguestfs's qemu process and your
1788 virtual machine are trying to update the disk image at the same time,
1789 since these usually results in massive disk corruption).
1791 =head1 STATE MACHINE
1793 libguestfs uses a state machine to model the child process:
1804 / | \ \ guestfs_launch
1815 \______/ <------ \________/
1817 The normal transitions are (1) CONFIG (when the handle is created, but
1818 there is no child process), (2) LAUNCHING (when the child process is
1819 booting up), (3) alternating between READY and BUSY as commands are
1820 issued to, and carried out by, the child process.
1822 The guest may be killed by L</guestfs_kill_subprocess>, or may die
1823 asynchronously at any time (eg. due to some internal error), and that
1824 causes the state to transition back to CONFIG.
1826 Configuration commands for qemu such as L</guestfs_add_drive> can only
1827 be issued when in the CONFIG state.
1829 The API offers one call that goes from CONFIG through LAUNCHING to
1830 READY. L</guestfs_launch> blocks until the child process is READY to
1831 accept commands (or until some failure or timeout).
1832 L</guestfs_launch> internally moves the state from CONFIG to LAUNCHING
1833 while it is running.
1835 API actions such as L</guestfs_mount> can only be issued when in the
1836 READY state. These API calls block waiting for the command to be
1837 carried out (ie. the state to transition to BUSY and then back to
1838 READY). There are no non-blocking versions, and no way to issue more
1839 than one command per handle at the same time.
1841 Finally, the child process sends asynchronous messages back to the
1842 main program, such as kernel log messages. You can register a
1843 callback to receive these messages.
1847 =head2 COMMUNICATION PROTOCOL
1849 Don't rely on using this protocol directly. This section documents
1850 how it currently works, but it may change at any time.
1852 The protocol used to talk between the library and the daemon running
1853 inside the qemu virtual machine is a simple RPC mechanism built on top
1854 of XDR (RFC 1014, RFC 1832, RFC 4506).
1856 The detailed format of structures is in C<src/guestfs_protocol.x>
1857 (note: this file is automatically generated).
1859 There are two broad cases, ordinary functions that don't have any
1860 C<FileIn> and C<FileOut> parameters, which are handled with very
1861 simple request/reply messages. Then there are functions that have any
1862 C<FileIn> or C<FileOut> parameters, which use the same request and
1863 reply messages, but they may also be followed by files sent using a
1866 =head3 ORDINARY FUNCTIONS (NO FILEIN/FILEOUT PARAMS)
1868 For ordinary functions, the request message is:
1870 total length (header + arguments,
1871 but not including the length word itself)
1872 struct guestfs_message_header (encoded as XDR)
1873 struct guestfs_<foo>_args (encoded as XDR)
1875 The total length field allows the daemon to allocate a fixed size
1876 buffer into which it slurps the rest of the message. As a result, the
1877 total length is limited to C<GUESTFS_MESSAGE_MAX> bytes (currently
1878 4MB), which means the effective size of any request is limited to
1879 somewhere under this size.
1881 Note also that many functions don't take any arguments, in which case
1882 the C<guestfs_I<foo>_args> is completely omitted.
1884 The header contains the procedure number (C<guestfs_proc>) which is
1885 how the receiver knows what type of args structure to expect, or none
1888 For functions that take optional arguments, the optional arguments are
1889 encoded in the C<guestfs_I<foo>_args> structure in the same way as
1890 ordinary arguments. A bitmask in the header indicates which optional
1891 arguments are meaningful. The bitmask is also checked to see if it
1892 contains bits set which the daemon does not know about (eg. if more
1893 optional arguments were added in a later version of the library), and
1894 this causes the call to be rejected.
1896 The reply message for ordinary functions is:
1898 total length (header + ret,
1899 but not including the length word itself)
1900 struct guestfs_message_header (encoded as XDR)
1901 struct guestfs_<foo>_ret (encoded as XDR)
1903 As above the C<guestfs_I<foo>_ret> structure may be completely omitted
1904 for functions that return no formal return values.
1906 As above the total length of the reply is limited to
1907 C<GUESTFS_MESSAGE_MAX>.
1909 In the case of an error, a flag is set in the header, and the reply
1910 message is slightly changed:
1912 total length (header + error,
1913 but not including the length word itself)
1914 struct guestfs_message_header (encoded as XDR)
1915 struct guestfs_message_error (encoded as XDR)
1917 The C<guestfs_message_error> structure contains the error message as a
1920 =head3 FUNCTIONS THAT HAVE FILEIN PARAMETERS
1922 A C<FileIn> parameter indicates that we transfer a file I<into> the
1923 guest. The normal request message is sent (see above). However this
1924 is followed by a sequence of file chunks.
1926 total length (header + arguments,
1927 but not including the length word itself,
1928 and not including the chunks)
1929 struct guestfs_message_header (encoded as XDR)
1930 struct guestfs_<foo>_args (encoded as XDR)
1931 sequence of chunks for FileIn param #0
1932 sequence of chunks for FileIn param #1 etc.
1934 The "sequence of chunks" is:
1936 length of chunk (not including length word itself)
1937 struct guestfs_chunk (encoded as XDR)
1939 struct guestfs_chunk (encoded as XDR)
1942 struct guestfs_chunk (with data.data_len == 0)
1944 The final chunk has the C<data_len> field set to zero. Additionally a
1945 flag is set in the final chunk to indicate either successful
1946 completion or early cancellation.
1948 At time of writing there are no functions that have more than one
1949 FileIn parameter. However this is (theoretically) supported, by
1950 sending the sequence of chunks for each FileIn parameter one after
1951 another (from left to right).
1953 Both the library (sender) I<and> the daemon (receiver) may cancel the
1954 transfer. The library does this by sending a chunk with a special
1955 flag set to indicate cancellation. When the daemon sees this, it
1956 cancels the whole RPC, does I<not> send any reply, and goes back to
1957 reading the next request.
1959 The daemon may also cancel. It does this by writing a special word
1960 C<GUESTFS_CANCEL_FLAG> to the socket. The library listens for this
1961 during the transfer, and if it gets it, it will cancel the transfer
1962 (it sends a cancel chunk). The special word is chosen so that even if
1963 cancellation happens right at the end of the transfer (after the
1964 library has finished writing and has started listening for the reply),
1965 the "spurious" cancel flag will not be confused with the reply
1968 This protocol allows the transfer of arbitrary sized files (no 32 bit
1969 limit), and also files where the size is not known in advance
1970 (eg. from pipes or sockets). However the chunks are rather small
1971 (C<GUESTFS_MAX_CHUNK_SIZE>), so that neither the library nor the
1972 daemon need to keep much in memory.
1974 =head3 FUNCTIONS THAT HAVE FILEOUT PARAMETERS
1976 The protocol for FileOut parameters is exactly the same as for FileIn
1977 parameters, but with the roles of daemon and library reversed.
1979 total length (header + ret,
1980 but not including the length word itself,
1981 and not including the chunks)
1982 struct guestfs_message_header (encoded as XDR)
1983 struct guestfs_<foo>_ret (encoded as XDR)
1984 sequence of chunks for FileOut param #0
1985 sequence of chunks for FileOut param #1 etc.
1987 =head3 INITIAL MESSAGE
1989 When the daemon launches it sends an initial word
1990 (C<GUESTFS_LAUNCH_FLAG>) which indicates that the guest and daemon is
1991 alive. This is what L</guestfs_launch> waits for.
1993 =head3 PROGRESS NOTIFICATION MESSAGES
1995 The daemon may send progress notification messages at any time. These
1996 are distinguished by the normal length word being replaced by
1997 C<GUESTFS_PROGRESS_FLAG>, followed by a fixed size progress message.
1999 The library turns them into progress callbacks (see
2000 C<guestfs_set_progress_callback>) if there is a callback registered,
2001 or discards them if not.
2003 The daemon self-limits the frequency of progress messages it sends
2004 (see C<daemon/proto.c:notify_progress>). Not all calls generate
2007 =head1 LIBGUESTFS VERSION NUMBERS
2009 Since April 2010, libguestfs has started to make separate development
2010 and stable releases, along with corresponding branches in our git
2011 repository. These separate releases can be identified by version
2014 even numbers for stable: 1.2.x, 1.4.x, ...
2015 .-------- odd numbers for development: 1.3.x, 1.5.x, ...
2021 | `-------- sub-version
2023 `------ always '1' because we don't change the ABI
2025 Thus "1.3.5" is the 5th update to the development branch "1.3".
2027 As time passes we cherry pick fixes from the development branch and
2028 backport those into the stable branch, the effect being that the
2029 stable branch should get more stable and less buggy over time. So the
2030 stable releases are ideal for people who don't need new features but
2031 would just like the software to work.
2033 Our criteria for backporting changes are:
2039 Documentation changes which don't affect any code are
2040 backported unless the documentation refers to a future feature
2041 which is not in stable.
2045 Bug fixes which are not controversial, fix obvious problems, and
2046 have been well tested are backported.
2050 Simple rearrangements of code which shouldn't affect how it works get
2051 backported. This is so that the code in the two branches doesn't get
2052 too far out of step, allowing us to backport future fixes more easily.
2056 We I<don't> backport new features, new APIs, new tools etc, except in
2057 one exceptional case: the new feature is required in order to
2058 implement an important bug fix.
2062 A new stable branch starts when we think the new features in
2063 development are substantial and compelling enough over the current
2064 stable branch to warrant it. When that happens we create new stable
2065 and development versions 1.N.0 and 1.(N+1).0 [N is even]. The new
2066 dot-oh release won't necessarily be so stable at this point, but by
2067 backporting fixes from development, that branch will stabilize over
2070 =head1 ENVIRONMENT VARIABLES
2074 =item LIBGUESTFS_APPEND
2076 Pass additional options to the guest kernel.
2078 =item LIBGUESTFS_DEBUG
2080 Set C<LIBGUESTFS_DEBUG=1> to enable verbose messages. This
2081 has the same effect as calling C<guestfs_set_verbose (g, 1)>.
2083 =item LIBGUESTFS_MEMSIZE
2085 Set the memory allocated to the qemu process, in megabytes. For
2088 LIBGUESTFS_MEMSIZE=700
2090 =item LIBGUESTFS_PATH
2092 Set the path that libguestfs uses to search for kernel and initrd.img.
2093 See the discussion of paths in section PATH above.
2095 =item LIBGUESTFS_QEMU
2097 Set the default qemu binary that libguestfs uses. If not set, then
2098 the qemu which was found at compile time by the configure script is
2101 See also L</QEMU WRAPPERS> above.
2103 =item LIBGUESTFS_TRACE
2105 Set C<LIBGUESTFS_TRACE=1> to enable command traces. This
2106 has the same effect as calling C<guestfs_set_trace (g, 1)>.
2110 Location of temporary directory, defaults to C</tmp>.
2112 If libguestfs was compiled to use the supermin appliance then the
2113 real appliance is cached in this directory, shared between all
2114 handles belonging to the same EUID. You can use C<$TMPDIR> to
2115 configure another directory to use in case C</tmp> is not large
2122 L<guestfs-examples(3)>,
2123 L<guestfs-ocaml(3)>,
2124 L<guestfs-python(3)>,
2131 L<virt-filesystems(1)>,
2132 L<virt-inspector(1)>,
2133 L<virt-list-filesystems(1)>,
2134 L<virt-list-partitions(1)>,
2143 L<http://libguestfs.org/>.
2145 Tools with a similar purpose:
2154 To get a list of bugs against libguestfs use this link:
2156 L<https://bugzilla.redhat.com/buglist.cgi?component=libguestfs&product=Virtualization+Tools>
2158 To report a new bug against libguestfs use this link:
2160 L<https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virtualization+Tools>
2162 When reporting a bug, please check:
2168 That the bug hasn't been reported already.
2172 That you are testing a recent version.
2176 Describe the bug accurately, and give a way to reproduce it.
2180 Run libguestfs-test-tool and paste the B<complete, unedited>
2181 output into the bug report.
2187 Richard W.M. Jones (C<rjones at redhat dot com>)
2191 Copyright (C) 2009-2010 Red Hat Inc.
2192 L<http://libguestfs.org/>
2194 This library is free software; you can redistribute it and/or
2195 modify it under the terms of the GNU Lesser General Public
2196 License as published by the Free Software Foundation; either
2197 version 2 of the License, or (at your option) any later version.
2199 This library is distributed in the hope that it will be useful,
2200 but WITHOUT ANY WARRANTY; without even the implied warranty of
2201 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
2202 Lesser General Public License for more details.
2204 You should have received a copy of the GNU Lesser General Public
2205 License along with this library; if not, write to the Free Software
2206 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA