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, "/");
20 cc prog.c -o prog -lguestfs
22 cc prog.c -o prog `pkg-config libguestfs --cflags --libs`
26 Libguestfs is a library for accessing and modifying guest disk images.
27 Amongst the things this is good for: making batch configuration
28 changes to guests, getting disk used/free statistics (see also:
29 virt-df), migrating between virtualization systems (see also:
30 virt-p2v), performing partial backups, performing partial guest
31 clones, cloning guests and changing registry/UUID/hostname info, and
34 Libguestfs uses Linux kernel and qemu code, and can access any type of
35 guest filesystem that Linux and qemu can, including but not limited
36 to: ext2/3/4, btrfs, FAT and NTFS, LVM, many different disk partition
37 schemes, qcow, qcow2, vmdk.
39 Libguestfs provides ways to enumerate guest storage (eg. partitions,
40 LVs, what filesystem is in each LV, etc.). It can also run commands
41 in the context of the guest. Also you can access filesystems over
44 Libguestfs is a library that can be linked with C and C++ management
45 programs (or management programs written in OCaml, Perl, Python, Ruby,
46 Java, PHP, Haskell or C#). You can also use it from shell scripts or the
49 You don't need to be root to use libguestfs, although obviously you do
50 need enough permissions to access the disk images.
52 Libguestfs is a large API because it can do many things. For a gentle
53 introduction, please read the L</API OVERVIEW> section next.
55 There are also some example programs in the L<guestfs-examples(3)>
60 This section provides a gentler overview of the libguestfs API. We
61 also try to group API calls together, where that may not be obvious
62 from reading about the individual calls in the main section of this
67 Before you can use libguestfs calls, you have to create a handle.
68 Then you must add at least one disk image to the handle, followed by
69 launching the handle, then performing whatever operations you want,
70 and finally closing the handle. By convention we use the single
71 letter C<g> for the name of the handle variable, although of course
72 you can use any name you want.
74 The general structure of all libguestfs-using programs looks like
77 guestfs_h *g = guestfs_create ();
79 /* Call guestfs_add_drive additional times if there are
80 * multiple disk images.
82 guestfs_add_drive (g, "guest.img");
84 /* Most manipulation calls won't work until you've launched
85 * the handle 'g'. You have to do this _after_ adding drives
86 * and _before_ other commands.
90 /* Now you can examine what partitions, LVs etc are available.
92 char **partitions = guestfs_list_partitions (g);
93 char **logvols = guestfs_lvs (g);
95 /* To access a filesystem in the image, you must mount it.
97 guestfs_mount (g, "/dev/sda1", "/");
99 /* Now you can perform filesystem actions on the guest
102 guestfs_touch (g, "/hello");
104 /* You only need to call guestfs_sync if you have made
105 * changes to the guest image. (But if you've made changes
106 * then you *must* sync). See also: guestfs_umount and
107 * guestfs_umount_all calls.
111 /* Close the handle 'g'. */
114 The code above doesn't include any error checking. In real code you
115 should check return values carefully for errors. In general all
116 functions that return integers return C<-1> on error, and all
117 functions that return pointers return C<NULL> on error. See section
118 L</ERROR HANDLING> below for how to handle errors, and consult the
119 documentation for each function call below to see precisely how they
120 return error indications.
124 The image filename (C<"guest.img"> in the example above) could be a
125 disk image from a virtual machine, a L<dd(1)> copy of a physical hard
126 disk, an actual block device, or simply an empty file of zeroes that
127 you have created through L<posix_fallocate(3)>. Libguestfs lets you
128 do useful things to all of these.
130 The call you should use in modern code for adding drives is
131 L</guestfs_add_drive_opts>. To add a disk image, allowing writes, and
132 specifying that the format is raw, do:
134 guestfs_add_drive_opts (g, filename,
135 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "raw",
138 You can add a disk read-only using:
140 guestfs_add_drive_opts (g, filename,
141 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "raw",
142 GUESTFS_ADD_DRIVE_OPTS_READONLY, 1,
145 or by calling the older function L</guestfs_add_drive_ro>. In either
146 case libguestfs won't modify the file.
148 Be extremely cautious if the disk image is in use, eg. if it is being
149 used by a virtual machine. Adding it read-write will almost certainly
150 cause disk corruption, but adding it read-only is safe.
152 You must add at least one disk image, and you may add multiple disk
153 images. In the API, the disk images are usually referred to as
154 C</dev/sda> (for the first one you added), C</dev/sdb> (for the second
157 Once L</guestfs_launch> has been called you cannot add any more images.
158 You can call L</guestfs_list_devices> to get a list of the device
159 names, in the order that you added them. See also L</BLOCK DEVICE
164 Before you can read or write files, create directories and so on in a
165 disk image that contains filesystems, you have to mount those
166 filesystems using L</guestfs_mount>. If you already know that a disk
167 image contains (for example) one partition with a filesystem on that
168 partition, then you can mount it directly:
170 guestfs_mount (g, "/dev/sda1", "/");
172 where C</dev/sda1> means literally the first partition (C<1>) of the
173 first disk image that we added (C</dev/sda>). If the disk contains
174 Linux LVM2 logical volumes you could refer to those instead (eg. C</dev/VG/LV>).
176 If you are given a disk image and you don't know what it contains then
177 you have to find out. Libguestfs can do that too: use
178 L</guestfs_list_partitions> and L</guestfs_lvs> to list possible
179 partitions and LVs, and either try mounting each to see what is
180 mountable, or else examine them with L</guestfs_vfs_type> or
181 L</guestfs_file>. Libguestfs also has a set of APIs for inspection of
182 disk images (see L</INSPECTION> below). But you might find it easier
183 to look at higher level programs built on top of libguestfs, in
184 particular L<virt-inspector(1)>.
186 To mount a disk image read-only, use L</guestfs_mount_ro>. There are
187 several other variations of the C<guestfs_mount_*> call.
189 =head2 FILESYSTEM ACCESS AND MODIFICATION
191 The majority of the libguestfs API consists of fairly low-level calls
192 for accessing and modifying the files, directories, symlinks etc on
193 mounted filesystems. There are over a hundred such calls which you
194 can find listed in detail below in this man page, and we don't even
195 pretend to cover them all in this overview.
197 Specify filenames as full paths, starting with C<"/"> and including
200 For example, if you mounted a filesystem at C<"/"> and you want to
201 read the file called C<"etc/passwd"> then you could do:
203 char *data = guestfs_cat (g, "/etc/passwd");
205 This would return C<data> as a newly allocated buffer containing the
206 full content of that file (with some conditions: see also
207 L</DOWNLOADING> below), or C<NULL> if there was an error.
209 As another example, to create a top-level directory on that filesystem
210 called C<"var"> you would do:
212 guestfs_mkdir (g, "/var");
214 To create a symlink you could do:
216 guestfs_ln_s (g, "/etc/init.d/portmap",
217 "/etc/rc3.d/S30portmap");
219 Libguestfs will reject attempts to use relative paths and there is no
220 concept of a current working directory.
222 Libguestfs can return errors in many situations: for example if the
223 filesystem isn't writable, or if a file or directory that you
224 requested doesn't exist. If you are using the C API (documented here)
225 you have to check for those error conditions after each call. (Other
226 language bindings turn these errors into exceptions).
228 File writes are affected by the per-handle umask, set by calling
229 L</guestfs_umask> and defaulting to 022. See L</UMASK>.
233 Libguestfs contains API calls to read, create and modify partition
234 tables on disk images.
236 In the common case where you want to create a single partition
237 covering the whole disk, you should use the L</guestfs_part_disk>
240 const char *parttype = "mbr";
241 if (disk_is_larger_than_2TB)
243 guestfs_part_disk (g, "/dev/sda", parttype);
245 Obviously this effectively wipes anything that was on that disk image
250 Libguestfs provides access to a large part of the LVM2 API, such as
251 L</guestfs_lvcreate> and L</guestfs_vgremove>. It won't make much sense
252 unless you familiarize yourself with the concepts of physical volumes,
253 volume groups and logical volumes.
255 This author strongly recommends reading the LVM HOWTO, online at
256 L<http://tldp.org/HOWTO/LVM-HOWTO/>.
260 Use L</guestfs_cat> to download small, text only files. This call
261 is limited to files which are less than 2 MB and which cannot contain
262 any ASCII NUL (C<\0>) characters. However it has a very simple
265 L</guestfs_read_file> can be used to read files which contain
266 arbitrary 8 bit data, since it returns a (pointer, size) pair.
267 However it is still limited to "small" files, less than 2 MB.
269 L</guestfs_download> can be used to download any file, with no
270 limits on content or size (even files larger than 4 GB).
272 To download multiple files, see L</guestfs_tar_out> and
277 It's often the case that you want to write a file or files to the disk
280 To write a small file with fixed content, use L</guestfs_write>. To
281 create a file of all zeroes, use L</guestfs_truncate_size> (sparse) or
282 L</guestfs_fallocate64> (with all disk blocks allocated). There are a
283 variety of other functions for creating test files, for example
284 L</guestfs_fill> and L</guestfs_fill_pattern>.
286 To upload a single file, use L</guestfs_upload>. This call has no
287 limits on file content or size (even files larger than 4 GB).
289 To upload multiple files, see L</guestfs_tar_in> and L</guestfs_tgz_in>.
291 However the fastest way to upload I<large numbers of arbitrary files>
292 is to turn them into a squashfs or CD ISO (see L<mksquashfs(8)> and
293 L<mkisofs(8)>), then attach this using L</guestfs_add_drive_ro>. If
294 you add the drive in a predictable way (eg. adding it last after all
295 other drives) then you can get the device name from
296 L</guestfs_list_devices> and mount it directly using
297 L</guestfs_mount_ro>. Note that squashfs images are sometimes
298 non-portable between kernel versions, and they don't support labels or
299 UUIDs. If you want to pre-build an image or you need to mount it
300 using a label or UUID, use an ISO image instead.
304 There are various different commands for copying between files and
305 devices and in and out of the guest filesystem. These are summarised
310 =item B<file> to B<file>
312 Use L</guestfs_cp> to copy a single file, or
313 L</guestfs_cp_a> to copy directories recursively.
315 =item B<file or device> to B<file or device>
317 Use L</guestfs_dd> which efficiently uses L<dd(1)>
318 to copy between files and devices in the guest.
320 Example: duplicate the contents of an LV:
322 guestfs_dd (g, "/dev/VG/Original", "/dev/VG/Copy");
324 The destination (C</dev/VG/Copy>) must be at least as large as the
325 source (C</dev/VG/Original>). To copy less than the whole
326 source device, use L</guestfs_copy_size>.
328 =item B<file on the host> to B<file or device>
330 Use L</guestfs_upload>. See L</UPLOADING> above.
332 =item B<file or device> to B<file on the host>
334 Use L</guestfs_download>. See L</DOWNLOADING> above.
340 L</guestfs_ll> is just designed for humans to read (mainly when using
341 the L<guestfish(1)>-equivalent command C<ll>).
343 L</guestfs_ls> is a quick way to get a list of files in a directory
344 from programs, as a flat list of strings.
346 L</guestfs_readdir> is a programmatic way to get a list of files in a
347 directory, plus additional information about each one. It is more
348 equivalent to using the L<readdir(3)> call on a local filesystem.
350 L</guestfs_find> and L</guestfs_find0> can be used to recursively list
353 =head2 RUNNING COMMANDS
355 Although libguestfs is primarily an API for manipulating files
356 inside guest images, we also provide some limited facilities for
357 running commands inside guests.
359 There are many limitations to this:
365 The kernel version that the command runs under will be different
366 from what it expects.
370 If the command needs to communicate with daemons, then most likely
371 they won't be running.
375 The command will be running in limited memory.
379 The network may not be available unless you enable it
380 (see L</guestfs_set_network>).
384 Only supports Linux guests (not Windows, BSD, etc).
388 Architecture limitations (eg. won't work for a PPC guest on
393 For SELinux guests, you may need to enable SELinux and load policy
394 first. See L</SELINUX> in this manpage.
398 I<Security:> It is not safe to run commands from untrusted, possibly
399 malicious guests. These commands may attempt to exploit your program
400 by sending unexpected output. They could also try to exploit the
401 Linux kernel or qemu provided by the libguestfs appliance. They could
402 use the network provided by the libguestfs appliance to bypass
403 ordinary network partitions and firewalls. They could use the
404 elevated privileges or different SELinux context of your program
407 A secure alternative is to use libguestfs to install a "firstboot"
408 script (a script which runs when the guest next boots normally), and
409 to have this script run the commands you want in the normal context of
410 the running guest, network security and so on. For information about
411 other security issues, see L</SECURITY>.
415 The two main API calls to run commands are L</guestfs_command> and
416 L</guestfs_sh> (there are also variations).
418 The difference is that L</guestfs_sh> runs commands using the shell, so
419 any shell globs, redirections, etc will work.
421 =head2 CONFIGURATION FILES
423 To read and write configuration files in Linux guest filesystems, we
424 strongly recommend using Augeas. For example, Augeas understands how
425 to read and write, say, a Linux shadow password file or X.org
426 configuration file, and so avoids you having to write that code.
428 The main Augeas calls are bound through the C<guestfs_aug_*> APIs. We
429 don't document Augeas itself here because there is excellent
430 documentation on the L<http://augeas.net/> website.
432 If you don't want to use Augeas (you fool!) then try calling
433 L</guestfs_read_lines> to get the file as a list of lines which
434 you can iterate over.
438 We support SELinux guests. To ensure that labeling happens correctly
439 in SELinux guests, you need to enable SELinux and load the guest's
446 Before launching, do:
448 guestfs_set_selinux (g, 1);
452 After mounting the guest's filesystem(s), load the policy. This
453 is best done by running the L<load_policy(8)> command in the
456 guestfs_sh (g, "/usr/sbin/load_policy");
458 (Older versions of C<load_policy> require you to specify the
459 name of the policy file).
463 Optionally, set the security context for the API. The correct
464 security context to use can only be known by inspecting the
465 guest. As an example:
467 guestfs_setcon (g, "unconfined_u:unconfined_r:unconfined_t:s0");
471 This will work for running commands and editing existing files.
473 When new files are created, you may need to label them explicitly,
474 for example by running the external command
475 C<restorecon pathname>.
479 Certain calls are affected by the current file mode creation mask (the
480 "umask"). In particular ones which create files or directories, such
481 as L</guestfs_touch>, L</guestfs_mknod> or L</guestfs_mkdir>. This
482 affects either the default mode that the file is created with or
483 modifies the mode that you supply.
485 The default umask is C<022>, so files are created with modes such as
486 C<0644> and directories with C<0755>.
488 There are two ways to avoid being affected by umask. Either set umask
489 to 0 (call C<guestfs_umask (g, 0)> early after launching). Or call
490 L</guestfs_chmod> after creating each file or directory.
492 For more information about umask, see L<umask(2)>.
494 =head2 ENCRYPTED DISKS
496 Libguestfs allows you to access Linux guests which have been
497 encrypted using whole disk encryption that conforms to the
498 Linux Unified Key Setup (LUKS) standard. This includes
499 nearly all whole disk encryption systems used by modern
502 Use L</guestfs_vfs_type> to identify LUKS-encrypted block
503 devices (it returns the string C<crypto_LUKS>).
505 Then open these devices by calling L</guestfs_luks_open>.
506 Obviously you will require the passphrase!
508 Opening a LUKS device creates a new device mapper device
509 called C</dev/mapper/mapname> (where C<mapname> is the
510 string you supply to L</guestfs_luks_open>).
511 Reads and writes to this mapper device are decrypted from and
512 encrypted to the underlying block device respectively.
514 LVM volume groups on the device can be made visible by calling
515 L</guestfs_vgscan> followed by L</guestfs_vg_activate_all>.
516 The logical volume(s) can now be mounted in the usual way.
518 Use the reverse process to close a LUKS device. Unmount
519 any logical volumes on it, deactivate the volume groups
520 by caling C<guestfs_vg_activate (g, 0, ["/dev/VG"])>.
521 Then close the mapper device by calling
522 L</guestfs_luks_close> on the C</dev/mapper/mapname>
523 device (I<not> the underlying encrypted block device).
527 Libguestfs has APIs for inspecting an unknown disk image to find out
528 if it contains operating systems. (These APIs used to be in a
529 separate Perl-only library called L<Sys::Guestfs::Lib(3)> but since
530 version 1.5.3 the most frequently used part of this library has been
531 rewritten in C and moved into the core code).
533 Add all disks belonging to the unknown virtual machine and call
534 L</guestfs_launch> in the usual way.
536 Then call L</guestfs_inspect_os>. This function uses other libguestfs
537 calls and certain heuristics, and returns a list of operating systems
538 that were found. An empty list means none were found. A single
539 element is the root filesystem of the operating system. For dual- or
540 multi-boot guests, multiple roots can be returned, each one
541 corresponding to a separate operating system. (Multi-boot virtual
542 machines are extremely rare in the world of virtualization, but since
543 this scenario can happen, we have built libguestfs to deal with it.)
545 For each root, you can then call various C<guestfs_inspect_get_*>
546 functions to get additional details about that operating system. For
547 example, call L</guestfs_inspect_get_type> to return the string
548 C<windows> or C<linux> for Windows and Linux-based operating systems
551 Un*x-like and Linux-based operating systems usually consist of several
552 filesystems which are mounted at boot time (for example, a separate
553 boot partition mounted on C</boot>). The inspection rules are able to
554 detect how filesystems correspond to mount points. Call
555 C<guestfs_inspect_get_mountpoints> to get this mapping. It might
556 return a hash table like this example:
559 / => /dev/vg_guest/lv_root
560 /usr => /dev/vg_guest/lv_usr
562 The caller can then make calls to L</guestfs_mount_options> to
563 mount the filesystems as suggested.
565 Be careful to mount filesystems in the right order (eg. C</> before
566 C</usr>). Sorting the keys of the hash by length, shortest first,
569 Inspection currently only works for some common operating systems.
570 Contributors are welcome to send patches for other operating systems
571 that we currently cannot detect.
573 Encrypted disks must be opened before inspection. See
574 L</ENCRYPTED DISKS> for more details. The L</guestfs_inspect_os>
575 function just ignores any encrypted devices.
577 A note on the implementation: The call L</guestfs_inspect_os> performs
578 inspection and caches the results in the guest handle. Subsequent
579 calls to C<guestfs_inspect_get_*> return this cached information, but
580 I<do not> re-read the disks. If you change the content of the guest
581 disks, you can redo inspection by calling L</guestfs_inspect_os>
582 again. (L</guestfs_inspect_list_applications> works a little
583 differently from the other calls and does read the disks. See
584 documentation for that function for details).
586 =head2 SPECIAL CONSIDERATIONS FOR WINDOWS GUESTS
588 Libguestfs can mount NTFS partitions. It does this using the
589 L<http://www.ntfs-3g.org/> driver.
591 DOS and Windows still use drive letters, and the filesystems are
592 always treated as case insensitive by Windows itself, and therefore
593 you might find a Windows configuration file referring to a path like
594 C<c:\windows\system32>. When the filesystem is mounted in libguestfs,
595 that directory might be referred to as C</WINDOWS/System32>.
597 Drive letter mappings are outside the scope of libguestfs. You have
598 to use libguestfs to read the appropriate Windows Registry and
599 configuration files, to determine yourself how drives are mapped (see
600 also L<hivex(3)> and L<virt-inspector(1)>).
602 Replacing backslash characters with forward slash characters is also
603 outside the scope of libguestfs, but something that you can easily do.
605 Where we can help is in resolving the case insensitivity of paths.
606 For this, call L</guestfs_case_sensitive_path>.
608 Libguestfs also provides some help for decoding Windows Registry
609 "hive" files, through the library C<hivex> which is part of the
610 libguestfs project although ships as a separate tarball. You have to
611 locate and download the hive file(s) yourself, and then pass them to
612 C<hivex> functions. See also the programs L<hivexml(1)>,
613 L<hivexsh(1)>, L<hivexregedit(1)> and L<virt-win-reg(1)> for more help
616 =head2 USING LIBGUESTFS WITH OTHER PROGRAMMING LANGUAGES
618 Although we don't want to discourage you from using the C API, we will
619 mention here that the same API is also available in other languages.
621 The API is broadly identical in all supported languages. This means
622 that the C call C<guestfs_mount(g,path)> is
623 C<$g-E<gt>mount($path)> in Perl, C<g.mount(path)> in Python,
624 and C<Guestfs.mount g path> in OCaml. In other words, a
625 straightforward, predictable isomorphism between each language.
627 Error messages are automatically transformed
628 into exceptions if the language supports it.
630 We don't try to "object orientify" parts of the API in OO languages,
631 although contributors are welcome to write higher level APIs above
632 what we provide in their favourite languages if they wish.
638 You can use the I<guestfs.h> header file from C++ programs. The C++
639 API is identical to the C API. C++ classes and exceptions are not
644 The C# bindings are highly experimental. Please read the warnings
645 at the top of C<csharp/Libguestfs.cs>.
649 This is the only language binding that is working but incomplete.
650 Only calls which return simple integers have been bound in Haskell,
651 and we are looking for help to complete this binding.
655 Full documentation is contained in the Javadoc which is distributed
660 For documentation see L<guestfs-ocaml(3)>.
664 For documentation see L<Sys::Guestfs(3)>.
668 For documentation see C<README-PHP> supplied with libguestfs
669 sources or in the php-libguestfs package for your distribution.
671 The PHP binding only works correctly on 64 bit machines.
675 For documentation see L<guestfs-python(3)>.
679 Use the Guestfs module. There is no Ruby-specific documentation, but
680 you can find examples written in Ruby in the libguestfs source.
682 =item B<shell scripts>
684 For documentation see L<guestfish(1)>.
688 =head2 LIBGUESTFS GOTCHAS
690 L<http://en.wikipedia.org/wiki/Gotcha_(programming)>: "A feature of a
691 system [...] that works in the way it is documented but is
692 counterintuitive and almost invites mistakes."
694 Since we developed libguestfs and the associated tools, there are
695 several things we would have designed differently, but are now stuck
696 with for backwards compatibility or other reasons. If there is ever a
697 libguestfs 2.0 release, you can expect these to change. Beware of
702 =item Autosync / forgetting to sync.
704 When modifying a filesystem from C or another language, you B<must>
705 unmount all filesystems and call L</guestfs_sync> explicitly before
706 you close the libguestfs handle. You can also call:
708 guestfs_set_autosync (g, 1);
710 to have the unmount/sync done automatically for you when the handle 'g'
711 is closed. (This feature is called "autosync", L</guestfs_set_autosync>
714 If you forget to do this, then it is entirely possible that your
715 changes won't be written out, or will be partially written, or (very
716 rarely) that you'll get disk corruption.
718 Note that in L<guestfish(3)> autosync is the default. So quick and
719 dirty guestfish scripts that forget to sync will work just fine, which
720 can make this very puzzling if you are trying to debug a problem.
722 Update: Autosync is enabled by default for all API users starting from
725 =item Mount option C<-o sync> should not be the default.
727 If you use L</guestfs_mount>, then C<-o sync,noatime> are added
728 implicitly. However C<-o sync> does not add any reliability benefit,
729 but does have a very large performance impact.
731 The work around is to use L</guestfs_mount_options> and set the mount
732 options that you actually want to use.
734 =item Read-only should be the default.
736 In L<guestfish(3)>, I<--ro> should be the default, and you should
737 have to specify I<--rw> if you want to make changes to the image.
739 This would reduce the potential to corrupt live VM images.
741 Note that many filesystems change the disk when you just mount and
742 unmount, even if you didn't perform any writes. You need to use
743 L</guestfs_add_drive_ro> to guarantee that the disk is not changed.
745 =item guestfish command line is hard to use.
747 C<guestfish disk.img> doesn't do what people expect (open C<disk.img>
748 for examination). It tries to run a guestfish command C<disk.img>
749 which doesn't exist, so it fails. In earlier versions of guestfish
750 the error message was also unintuitive, but we have corrected this
751 since. Like the Bourne shell, we should have used C<guestfish -c
752 command> to run commands.
754 =item guestfish megabyte modifiers don't work right on all commands
756 In recent guestfish you can use C<1M> to mean 1 megabyte (and
757 similarly for other modifiers). What guestfish actually does is to
758 multiply the number part by the modifier part and pass the result to
759 the C API. However this doesn't work for a few APIs which aren't
760 expecting bytes, but are already expecting some other unit
763 The most common is L</guestfs_lvcreate>. The guestfish command:
767 does not do what you might expect. Instead because
768 L</guestfs_lvcreate> is already expecting megabytes, this tries to
769 create a 100 I<terabyte> (100 megabytes * megabytes) logical volume.
770 The error message you get from this is also a little obscure.
772 This could be fixed in the generator by specially marking parameters
773 and return values which take bytes or other units.
775 =item Ambiguity between devices and paths
777 There is a subtle ambiguity in the API between a device name
778 (eg. C</dev/sdb2>) and a similar pathname. A file might just happen
779 to be called C<sdb2> in the directory C</dev> (consider some non-Unix
782 In the current API we usually resolve this ambiguity by having two
783 separate calls, for example L</guestfs_checksum> and
784 L</guestfs_checksum_device>. Some API calls are ambiguous and
785 (incorrectly) resolve the problem by detecting if the path supplied
786 begins with C</dev/>.
788 To avoid both the ambiguity and the need to duplicate some calls, we
789 could make paths/devices into structured names. One way to do this
790 would be to use a notation like grub (C<hd(0,0)>), although nobody
791 really likes this aspect of grub. Another way would be to use a
792 structured type, equivalent to this OCaml type:
794 type path = Path of string | Device of int | Partition of int * int
796 which would allow you to pass arguments like:
799 Device 1 (* /dev/sdb, or perhaps /dev/sda *)
800 Partition (1, 2) (* /dev/sdb2 (or is it /dev/sda2 or /dev/sdb3?) *)
801 Path "/dev/sdb2" (* not a device *)
803 As you can see there are still problems to resolve even with this
804 representation. Also consider how it might work in guestfish.
808 =head2 PROTOCOL LIMITS
810 Internally libguestfs uses a message-based protocol to pass API calls
811 and their responses to and from a small "appliance" (see L</INTERNALS>
812 for plenty more detail about this). The maximum message size used by
813 the protocol is slightly less than 4 MB. For some API calls you may
814 need to be aware of this limit. The API calls which may be affected
815 are individually documented, with a link back to this section of the
818 A simple call such as L</guestfs_cat> returns its result (the file
819 data) in a simple string. Because this string is at some point
820 internally encoded as a message, the maximum size that it can return
821 is slightly under 4 MB. If the requested file is larger than this
822 then you will get an error.
824 In order to transfer large files into and out of the guest filesystem,
825 you need to use particular calls that support this. The sections
826 L</UPLOADING> and L</DOWNLOADING> document how to do this.
828 You might also consider mounting the disk image using our FUSE
829 filesystem support (L<guestmount(1)>).
831 =head2 KEYS AND PASSPHRASES
833 Certain libguestfs calls take a parameter that contains sensitive key
834 material, passed in as a C string.
836 In the future we would hope to change the libguestfs implementation so
837 that keys are L<mlock(2)>-ed into physical RAM, and thus can never end
838 up in swap. However this is I<not> done at the moment, because of the
839 complexity of such an implementation.
841 Therefore you should be aware that any key parameter you pass to
842 libguestfs might end up being written out to the swap partition. If
843 this is a concern, scrub the swap partition or don't use libguestfs on
846 =head2 MULTIPLE HANDLES AND MULTIPLE THREADS
848 All high-level libguestfs actions are synchronous. If you want
849 to use libguestfs asynchronously then you must create a thread.
851 Only use the handle from a single thread. Either use the handle
852 exclusively from one thread, or provide your own mutex so that two
853 threads cannot issue calls on the same handle at the same time.
855 See the graphical program guestfs-browser for one possible
856 architecture for multithreaded programs using libvirt and libguestfs.
860 Libguestfs needs a kernel and initrd.img, which it finds by looking
861 along an internal path.
863 By default it looks for these in the directory C<$libdir/guestfs>
864 (eg. C</usr/local/lib/guestfs> or C</usr/lib64/guestfs>).
866 Use L</guestfs_set_path> or set the environment variable
867 L</LIBGUESTFS_PATH> to change the directories that libguestfs will
868 search in. The value is a colon-separated list of paths. The current
869 directory is I<not> searched unless the path contains an empty element
870 or C<.>. For example C<LIBGUESTFS_PATH=:/usr/lib/guestfs> would
871 search the current directory and then C</usr/lib/guestfs>.
875 If you want to compile your own qemu, run qemu from a non-standard
876 location, or pass extra arguments to qemu, then you can write a
877 shell-script wrapper around qemu.
879 There is one important rule to remember: you I<must C<exec qemu>> as
880 the last command in the shell script (so that qemu replaces the shell
881 and becomes the direct child of the libguestfs-using program). If you
882 don't do this, then the qemu process won't be cleaned up correctly.
884 Here is an example of a wrapper, where I have built my own copy of
888 qemudir=/home/rjones/d/qemu
889 exec $qemudir/x86_64-softmmu/qemu-system-x86_64 -L $qemudir/pc-bios "$@"
891 Save this script as C</tmp/qemu.wrapper> (or wherever), C<chmod +x>,
892 and then use it by setting the LIBGUESTFS_QEMU environment variable.
895 LIBGUESTFS_QEMU=/tmp/qemu.wrapper guestfish
897 Note that libguestfs also calls qemu with the -help and -version
898 options in order to determine features.
902 We guarantee the libguestfs ABI (binary interface), for public,
903 high-level actions as outlined in this section. Although we will
904 deprecate some actions, for example if they get replaced by newer
905 calls, we will keep the old actions forever. This allows you the
906 developer to program in confidence against the libguestfs API.
908 =head2 BLOCK DEVICE NAMING
910 In the kernel there is now quite a profusion of schemata for naming
911 block devices (in this context, by I<block device> I mean a physical
912 or virtual hard drive). The original Linux IDE driver used names
913 starting with C</dev/hd*>. SCSI devices have historically used a
914 different naming scheme, C</dev/sd*>. When the Linux kernel I<libata>
915 driver became a popular replacement for the old IDE driver
916 (particularly for SATA devices) those devices also used the
917 C</dev/sd*> scheme. Additionally we now have virtual machines with
918 paravirtualized drivers. This has created several different naming
919 systems, such as C</dev/vd*> for virtio disks and C</dev/xvd*> for Xen
922 As discussed above, libguestfs uses a qemu appliance running an
923 embedded Linux kernel to access block devices. We can run a variety
924 of appliances based on a variety of Linux kernels.
926 This causes a problem for libguestfs because many API calls use device
927 or partition names. Working scripts and the recipe (example) scripts
928 that we make available over the internet could fail if the naming
931 Therefore libguestfs defines C</dev/sd*> as the I<standard naming
932 scheme>. Internally C</dev/sd*> names are translated, if necessary,
933 to other names as required. For example, under RHEL 5 which uses the
934 C</dev/hd*> scheme, any device parameter C</dev/sda2> is translated to
935 C</dev/hda2> transparently.
937 Note that this I<only> applies to parameters. The
938 L</guestfs_list_devices>, L</guestfs_list_partitions> and similar calls
939 return the true names of the devices and partitions as known to the
942 =head3 ALGORITHM FOR BLOCK DEVICE NAME TRANSLATION
944 Usually this translation is transparent. However in some (very rare)
945 cases you may need to know the exact algorithm. Such cases include
946 where you use L</guestfs_config> to add a mixture of virtio and IDE
947 devices to the qemu-based appliance, so have a mixture of C</dev/sd*>
948 and C</dev/vd*> devices.
950 The algorithm is applied only to I<parameters> which are known to be
951 either device or partition names. Return values from functions such
952 as L</guestfs_list_devices> are never changed.
958 Is the string a parameter which is a device or partition name?
962 Does the string begin with C</dev/sd>?
966 Does the named device exist? If so, we use that device.
967 However if I<not> then we continue with this algorithm.
971 Replace initial C</dev/sd> string with C</dev/hd>.
973 For example, change C</dev/sda2> to C</dev/hda2>.
975 If that named device exists, use it. If not, continue.
979 Replace initial C</dev/sd> string with C</dev/vd>.
981 If that named device exists, use it. If not, return an error.
985 =head3 PORTABILITY CONCERNS WITH BLOCK DEVICE NAMING
987 Although the standard naming scheme and automatic translation is
988 useful for simple programs and guestfish scripts, for larger programs
989 it is best not to rely on this mechanism.
991 Where possible for maximum future portability programs using
992 libguestfs should use these future-proof techniques:
998 Use L</guestfs_list_devices> or L</guestfs_list_partitions> to list
999 actual device names, and then use those names directly.
1001 Since those device names exist by definition, they will never be
1006 Use higher level ways to identify filesystems, such as LVM names,
1007 UUIDs and filesystem labels.
1013 This section discusses security implications of using libguestfs,
1014 particularly with untrusted or malicious guests or disk images.
1016 =head2 GENERAL SECURITY CONSIDERATIONS
1018 Be careful with any files or data that you download from a guest (by
1019 "download" we mean not just the L</guestfs_download> command but any
1020 command that reads files, filenames, directories or anything else from
1021 a disk image). An attacker could manipulate the data to fool your
1022 program into doing the wrong thing. Consider cases such as:
1028 the data (file etc) not being present
1032 being present but empty
1036 being much larger than normal
1040 containing arbitrary 8 bit data
1044 being in an unexpected character encoding
1048 containing homoglyphs.
1052 =head2 SECURITY OF MOUNTING FILESYSTEMS
1054 When you mount a filesystem under Linux, mistakes in the kernel
1055 filesystem (VFS) module can sometimes be escalated into exploits by
1056 deliberately creating a malicious, malformed filesystem. These
1057 exploits are very severe for two reasons. Firstly there are very many
1058 filesystem drivers in the kernel, and many of them are infrequently
1059 used and not much developer attention has been paid to the code.
1060 Linux userspace helps potential crackers by detecting the filesystem
1061 type and automatically choosing the right VFS driver, even if that
1062 filesystem type is obscure or unexpected for the administrator.
1063 Secondly, a kernel-level exploit is like a local root exploit (worse
1064 in some ways), giving immediate and total access to the system right
1065 down to the hardware level.
1067 That explains why you should never mount a filesystem from an
1068 untrusted guest on your host kernel. How about libguestfs? We run a
1069 Linux kernel inside a qemu virtual machine, usually running as a
1070 non-root user. The attacker would need to write a filesystem which
1071 first exploited the kernel, and then exploited either qemu
1072 virtualization (eg. a faulty qemu driver) or the libguestfs protocol,
1073 and finally to be as serious as the host kernel exploit it would need
1074 to escalate its privileges to root. This multi-step escalation,
1075 performed by a static piece of data, is thought to be extremely hard
1076 to do, although we never say 'never' about security issues.
1078 In any case callers can reduce the attack surface by forcing the
1079 filesystem type when mounting (use L</guestfs_mount_vfs>).
1081 =head2 PROTOCOL SECURITY
1083 The protocol is designed to be secure, being based on RFC 4506 (XDR)
1084 with a defined upper message size. However a program that uses
1085 libguestfs must also take care - for example you can write a program
1086 that downloads a binary from a disk image and executes it locally, and
1087 no amount of protocol security will save you from the consequences.
1089 =head2 INSPECTION SECURITY
1091 Parts of the inspection API (see L</INSPECTION>) return untrusted
1092 strings directly from the guest, and these could contain any 8 bit
1093 data. Callers should be careful to escape these before printing them
1094 to a structured file (for example, use HTML escaping if creating a web
1097 Guest configuration may be altered in unusual ways by the
1098 administrator of the virtual machine, and may not reflect reality
1099 (particularly for untrusted or actively malicious guests). For
1100 example we parse the hostname from configuration files like
1101 C</etc/sysconfig/network> that we find in the guest, but the guest
1102 administrator can easily manipulate these files to provide the wrong
1105 The inspection API parses guest configuration using two external
1106 libraries: Augeas (Linux configuration) and hivex (Windows Registry).
1107 Both are designed to be robust in the face of malicious data, although
1108 denial of service attacks are still possible, for example with
1109 oversized configuration files.
1111 =head2 RUNNING UNTRUSTED GUEST COMMANDS
1113 Be very cautious about running commands from the guest. By running a
1114 command in the guest, you are giving CPU time to a binary that you do
1115 not control, under the same user account as the library, albeit
1116 wrapped in qemu virtualization. More information and alternatives can
1117 be found in the section L</RUNNING COMMANDS>.
1119 =head2 CVE-2010-3851
1121 https://bugzilla.redhat.com/642934
1123 This security bug concerns the automatic disk format detection that
1124 qemu does on disk images.
1126 A raw disk image is just the raw bytes, there is no header. Other
1127 disk images like qcow2 contain a special header. Qemu deals with this
1128 by looking for one of the known headers, and if none is found then
1129 assuming the disk image must be raw.
1131 This allows a guest which has been given a raw disk image to write
1132 some other header. At next boot (or when the disk image is accessed
1133 by libguestfs) qemu would do autodetection and think the disk image
1134 format was, say, qcow2 based on the header written by the guest.
1136 This in itself would not be a problem, but qcow2 offers many features,
1137 one of which is to allow a disk image to refer to another image
1138 (called the "backing disk"). It does this by placing the path to the
1139 backing disk into the qcow2 header. This path is not validated and
1140 could point to any host file (eg. "/etc/passwd"). The backing disk is
1141 then exposed through "holes" in the qcow2 disk image, which of course
1142 is completely under the control of the attacker.
1144 In libguestfs this is rather hard to exploit except under two
1151 You have enabled the network or have opened the disk in write mode.
1155 You are also running untrusted code from the guest (see
1156 L</RUNNING COMMANDS>).
1160 The way to avoid this is to specify the expected disk format when
1161 adding disks (the optional C<format> option to
1162 L</guestfs_add_drive_opts>). You should always do this if the disk is
1163 raw format, and it's a good idea for other cases too.
1165 For disks added from libvirt using calls like L</guestfs_add_domain>,
1166 the format is fetched from libvirt and passed through.
1168 For libguestfs tools, use the I<--format> command line parameter as
1171 =head1 CONNECTION MANAGEMENT
1175 C<guestfs_h> is the opaque type representing a connection handle.
1176 Create a handle by calling L</guestfs_create>. Call L</guestfs_close>
1177 to free the handle and release all resources used.
1179 For information on using multiple handles and threads, see the section
1180 L</MULTIPLE HANDLES AND MULTIPLE THREADS> below.
1182 =head2 guestfs_create
1184 guestfs_h *guestfs_create (void);
1186 Create a connection handle.
1188 You have to call L</guestfs_add_drive_opts> (or one of the equivalent
1189 calls) on the handle at least once.
1191 This function returns a non-NULL pointer to a handle on success or
1194 After configuring the handle, you have to call L</guestfs_launch>.
1196 You may also want to configure error handling for the handle. See
1197 L</ERROR HANDLING> section below.
1199 =head2 guestfs_close
1201 void guestfs_close (guestfs_h *g);
1203 This closes the connection handle and frees up all resources used.
1205 =head1 ERROR HANDLING
1207 API functions can return errors. For example, almost all functions
1208 that return C<int> will return C<-1> to indicate an error.
1210 Additional information is available for errors: an error message
1211 string and optionally an error number (errno) if the thing that failed
1214 You can get at the additional information about the last error on the
1215 handle by calling L</guestfs_last_error>, L</guestfs_last_errno>,
1216 and/or by setting up an error handler with
1217 L</guestfs_set_error_handler>.
1219 When the handle is created, a default error handler is installed which
1220 prints the error message string to C<stderr>. For small short-running
1221 command line programs it is sufficient to do:
1223 if (guestfs_launch (g) == -1)
1224 exit (EXIT_FAILURE);
1226 since the default error handler will ensure that an error message has
1227 been printed to C<stderr> before the program exits.
1229 For other programs the caller will almost certainly want to install an
1230 alternate error handler or do error handling in-line like this:
1232 g = guestfs_create ();
1234 /* This disables the default behaviour of printing errors
1236 guestfs_set_error_handler (g, NULL, NULL);
1238 if (guestfs_launch (g) == -1) {
1239 /* Examine the error message and print it etc. */
1240 char *msg = guestfs_last_error (g);
1241 int errnum = guestfs_last_errno (g);
1242 fprintf (stderr, "%s\n", msg);
1246 Out of memory errors are handled differently. The default action is
1247 to call L<abort(3)>. If this is undesirable, then you can set a
1248 handler using L</guestfs_set_out_of_memory_handler>.
1250 L</guestfs_create> returns C<NULL> if the handle cannot be created,
1251 and because there is no handle if this happens there is no way to get
1252 additional error information. However L</guestfs_create> is supposed
1253 to be a lightweight operation which can only fail because of
1254 insufficient memory (it returns NULL in this case).
1256 =head2 guestfs_last_error
1258 const char *guestfs_last_error (guestfs_h *g);
1260 This returns the last error message that happened on C<g>. If
1261 there has not been an error since the handle was created, then this
1264 The lifetime of the returned string is until the next error occurs, or
1265 L</guestfs_close> is called.
1267 =head2 guestfs_last_errno
1269 int guestfs_last_errno (guestfs_h *g);
1271 This returns the last error number (errno) that happened on C<g>.
1273 If successful, an errno integer not equal to zero is returned.
1275 If no error, this returns 0. This call can return 0 in three
1282 There has not been any error on the handle.
1286 There has been an error but the errno was meaningless. This
1287 corresponds to the case where the error did not come from a
1288 failed system call, but for some other reason.
1292 There was an error from a failed system call, but for some
1293 reason the errno was not captured and returned. This usually
1294 indicates a bug in libguestfs.
1298 Libguestfs tries to convert the errno from inside the applicance into
1299 a corresponding errno for the caller (not entirely trivial: the
1300 appliance might be running a completely different operating system
1301 from the library and error numbers are not standardized across
1302 Un*xen). If this could not be done, then the error is translated to
1303 C<EINVAL>. In practice this should only happen in very rare
1306 =head2 guestfs_set_error_handler
1308 typedef void (*guestfs_error_handler_cb) (guestfs_h *g,
1311 void guestfs_set_error_handler (guestfs_h *g,
1312 guestfs_error_handler_cb cb,
1315 The callback C<cb> will be called if there is an error. The
1316 parameters passed to the callback are an opaque data pointer and the
1317 error message string.
1319 C<errno> is not passed to the callback. To get that the callback must
1320 call L</guestfs_last_errno>.
1322 Note that the message string C<msg> is freed as soon as the callback
1323 function returns, so if you want to stash it somewhere you must make
1326 The default handler prints messages on C<stderr>.
1328 If you set C<cb> to C<NULL> then I<no> handler is called.
1330 =head2 guestfs_get_error_handler
1332 guestfs_error_handler_cb guestfs_get_error_handler (guestfs_h *g,
1335 Returns the current error handler callback.
1337 =head2 guestfs_set_out_of_memory_handler
1339 typedef void (*guestfs_abort_cb) (void);
1340 int guestfs_set_out_of_memory_handler (guestfs_h *g,
1343 The callback C<cb> will be called if there is an out of memory
1344 situation. I<Note this callback must not return>.
1346 The default is to call L<abort(3)>.
1348 You cannot set C<cb> to C<NULL>. You can't ignore out of memory
1351 =head2 guestfs_get_out_of_memory_handler
1353 guestfs_abort_fn guestfs_get_out_of_memory_handler (guestfs_h *g);
1355 This returns the current out of memory handler.
1367 =head2 GROUPS OF FUNCTIONALITY IN THE APPLIANCE
1369 Using L</guestfs_available> you can test availability of
1370 the following groups of functions. This test queries the
1371 appliance to see if the appliance you are currently using
1372 supports the functionality.
1376 =head2 GUESTFISH supported COMMAND
1378 In L<guestfish(3)> there is a handy interactive command
1379 C<supported> which prints out the available groups and
1380 whether they are supported by this build of libguestfs.
1381 Note however that you have to do C<run> first.
1383 =head2 SINGLE CALLS AT COMPILE TIME
1385 Since version 1.5.8, C<E<lt>guestfs.hE<gt>> defines symbols
1386 for each C API function, such as:
1388 #define LIBGUESTFS_HAVE_DD 1
1390 if L</guestfs_dd> is available.
1392 Before version 1.5.8, if you needed to test whether a single
1393 libguestfs function is available at compile time, we recommended using
1394 build tools such as autoconf or cmake. For example in autotools you
1397 AC_CHECK_LIB([guestfs],[guestfs_create])
1398 AC_CHECK_FUNCS([guestfs_dd])
1400 which would result in C<HAVE_GUESTFS_DD> being either defined
1401 or not defined in your program.
1403 =head2 SINGLE CALLS AT RUN TIME
1405 Testing at compile time doesn't guarantee that a function really
1406 exists in the library. The reason is that you might be dynamically
1407 linked against a previous I<libguestfs.so> (dynamic library)
1408 which doesn't have the call. This situation unfortunately results
1409 in a segmentation fault, which is a shortcoming of the C dynamic
1410 linking system itself.
1412 You can use L<dlopen(3)> to test if a function is available
1413 at run time, as in this example program (note that you still
1414 need the compile time check as well):
1420 #include <guestfs.h>
1424 #ifdef LIBGUESTFS_HAVE_DD
1428 /* Test if the function guestfs_dd is really available. */
1429 dl = dlopen (NULL, RTLD_LAZY);
1431 fprintf (stderr, "dlopen: %s\n", dlerror ());
1432 exit (EXIT_FAILURE);
1434 has_function = dlsym (dl, "guestfs_dd") != NULL;
1438 printf ("this libguestfs.so does NOT have guestfs_dd function\n");
1440 printf ("this libguestfs.so has guestfs_dd function\n");
1441 /* Now it's safe to call
1442 guestfs_dd (g, "foo", "bar");
1446 printf ("guestfs_dd function was not found at compile time\n");
1450 You may think the above is an awful lot of hassle, and it is.
1451 There are other ways outside of the C linking system to ensure
1452 that this kind of incompatibility never arises, such as using
1455 Requires: libguestfs >= 1.0.80
1457 =head1 CALLS WITH OPTIONAL ARGUMENTS
1459 A recent feature of the API is the introduction of calls which take
1460 optional arguments. In C these are declared 3 ways. The main way is
1461 as a call which takes variable arguments (ie. C<...>), as in this
1464 int guestfs_add_drive_opts (guestfs_h *g, const char *filename, ...);
1466 Call this with a list of optional arguments, terminated by C<-1>.
1467 So to call with no optional arguments specified:
1469 guestfs_add_drive_opts (g, filename, -1);
1471 With a single optional argument:
1473 guestfs_add_drive_opts (g, filename,
1474 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "qcow2",
1479 guestfs_add_drive_opts (g, filename,
1480 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "qcow2",
1481 GUESTFS_ADD_DRIVE_OPTS_READONLY, 1,
1484 and so forth. Don't forget the terminating C<-1> otherwise
1485 Bad Things will happen!
1487 =head2 USING va_list FOR OPTIONAL ARGUMENTS
1489 The second variant has the same name with the suffix C<_va>, which
1490 works the same way but takes a C<va_list>. See the C manual for
1491 details. For the example function, this is declared:
1493 int guestfs_add_drive_opts_va (guestfs_h *g, const char *filename,
1496 =head2 CONSTRUCTING OPTIONAL ARGUMENTS
1498 The third variant is useful where you need to construct these
1499 calls. You pass in a structure where you fill in the optional
1500 fields. The structure has a bitmask as the first element which
1501 you must set to indicate which fields you have filled in. For
1502 our example function the structure and call are declared:
1504 struct guestfs_add_drive_opts_argv {
1510 int guestfs_add_drive_opts_argv (guestfs_h *g, const char *filename,
1511 const struct guestfs_add_drive_opts_argv *optargs);
1513 You could call it like this:
1515 struct guestfs_add_drive_opts_argv optargs = {
1516 .bitmask = GUESTFS_ADD_DRIVE_OPTS_READONLY_BITMASK |
1517 GUESTFS_ADD_DRIVE_OPTS_FORMAT_BITMASK,
1522 guestfs_add_drive_opts_argv (g, filename, &optargs);
1530 The C<_BITMASK> suffix on each option name when specifying the
1535 You do not need to fill in all fields of the structure.
1539 There must be a one-to-one correspondence between fields of the
1540 structure that are filled in, and bits set in the bitmask.
1544 =head2 OPTIONAL ARGUMENTS IN OTHER LANGUAGES
1546 In other languages, optional arguments are expressed in the
1547 way that is natural for that language. We refer you to the
1548 language-specific documentation for more details on that.
1550 For guestfish, see L<guestfish(1)/OPTIONAL ARGUMENTS>.
1552 =head2 SETTING CALLBACKS TO HANDLE EVENTS
1554 The child process generates events in some situations. Current events
1555 include: receiving a log message, the child process exits.
1557 Use the C<guestfs_set_*_callback> functions to set a callback for
1558 different types of events.
1560 Only I<one callback of each type> can be registered for each handle.
1561 Calling C<guestfs_set_*_callback> again overwrites the previous
1562 callback of that type. Cancel all callbacks of this type by calling
1563 this function with C<cb> set to C<NULL>.
1565 =head2 guestfs_set_log_message_callback
1567 typedef void (*guestfs_log_message_cb) (guestfs_h *g, void *opaque,
1568 char *buf, int len);
1569 void guestfs_set_log_message_callback (guestfs_h *g,
1570 guestfs_log_message_cb cb,
1573 The callback function C<cb> will be called whenever qemu or the guest
1574 writes anything to the console.
1576 Use this function to capture kernel messages and similar.
1578 Normally there is no log message handler, and log messages are just
1581 =head2 guestfs_set_subprocess_quit_callback
1583 typedef void (*guestfs_subprocess_quit_cb) (guestfs_h *g, void *opaque);
1584 void guestfs_set_subprocess_quit_callback (guestfs_h *g,
1585 guestfs_subprocess_quit_cb cb,
1588 The callback function C<cb> will be called when the child process
1589 quits, either asynchronously or if killed by
1590 L</guestfs_kill_subprocess>. (This corresponds to a transition from
1591 any state to the CONFIG state).
1593 =head2 guestfs_set_launch_done_callback
1595 typedef void (*guestfs_launch_done_cb) (guestfs_h *g, void *opaque);
1596 void guestfs_set_launch_done_callback (guestfs_h *g,
1597 guestfs_launch_done_cb cb,
1600 The callback function C<cb> will be called when the child process
1601 becomes ready first time after it has been launched. (This
1602 corresponds to a transition from LAUNCHING to the READY state).
1604 =head2 guestfs_set_close_callback
1606 typedef void (*guestfs_close_cb) (guestfs_h *g, void *opaque);
1607 void guestfs_set_close_callback (guestfs_h *g,
1608 guestfs_close_cb cb,
1611 The callback function C<cb> will be called while the handle
1612 is being closed (synchronously from L</guestfs_close>).
1614 Note that libguestfs installs an L<atexit(3)> handler to try to
1615 clean up handles that are open when the program exits. This
1616 means that this callback might be called indirectly from
1617 L<exit(3)>, which can cause unexpected problems in higher-level
1618 languages (eg. if your HLL interpreter has already been cleaned
1619 up by the time this is called, and if your callback then jumps
1620 into some HLL function).
1622 =head2 guestfs_set_progress_callback
1624 typedef void (*guestfs_progress_cb) (guestfs_h *g, void *opaque,
1625 int proc_nr, int serial,
1626 uint64_t position, uint64_t total);
1627 void guestfs_set_progress_callback (guestfs_h *g,
1628 guestfs_progress_cb cb,
1631 Some long-running operations can generate progress messages. If
1632 this callback is registered, then it will be called each time a
1633 progress message is generated (usually two seconds after the
1634 operation started, and three times per second thereafter until
1635 it completes, although the frequency may change in future versions).
1637 The callback receives two numbers: C<position> and C<total>.
1638 The units of C<total> are not defined, although for some
1639 operations C<total> may relate in some way to the amount of
1640 data to be transferred (eg. in bytes or megabytes), and
1641 C<position> may be the portion which has been transferred.
1643 The only defined and stable parts of the API are:
1649 The callback can display to the user some type of progress bar or
1650 indicator which shows the ratio of C<position>:C<total>.
1654 0 E<lt>= C<position> E<lt>= C<total>
1658 If any progress notification is sent during a call, then a final
1659 progress notification is always sent when C<position> = C<total>.
1661 This is to simplify caller code, so callers can easily set the
1662 progress indicator to "100%" at the end of the operation, without
1663 requiring special code to detect this case.
1667 The callback also receives the procedure number and serial number of
1668 the call. These are only useful for debugging protocol issues, and
1669 the callback can normally ignore them. The callback may want to
1670 print these numbers in error messages or debugging messages.
1672 =head1 PRIVATE DATA AREA
1674 You can attach named pieces of private data to the libguestfs handle,
1675 and fetch them by name for the lifetime of the handle. This is called
1676 the private data area and is only available from the C API.
1678 To attach a named piece of data, use the following call:
1680 void guestfs_set_private (guestfs_h *g, const char *key, void *data);
1682 C<key> is the name to associate with this data, and C<data> is an
1683 arbitrary pointer (which can be C<NULL>). Any previous item with the
1684 same name is overwritten.
1686 You can use any C<key> you want, but names beginning with an
1687 underscore character are reserved for internal libguestfs purposes
1688 (for implementing language bindings). It is recommended to prefix the
1689 name with some unique string to avoid collisions with other users.
1691 To retrieve the pointer, use:
1693 void *guestfs_get_private (guestfs_h *g, const char *key);
1695 This function returns C<NULL> if either no data is found associated
1696 with C<key>, or if the user previously set the C<key>'s C<data>
1699 Libguestfs does not try to look at or interpret the C<data> pointer in
1700 any way. As far as libguestfs is concerned, it need not be a valid
1701 pointer at all. In particular, libguestfs does I<not> try to free the
1702 data when the handle is closed. If the data must be freed, then the
1703 caller must either free it before calling L</guestfs_close> or must
1704 set up a close callback to do it (see L</guestfs_set_close_callback>,
1705 and note that only one callback can be registered for a handle).
1707 The private data area is implemented using a hash table, and should be
1708 reasonably efficient for moderate numbers of keys.
1712 <!-- old anchor for the next section -->
1713 <a name="state_machine_and_low_level_event_api"/>
1719 Internally, libguestfs is implemented by running an appliance (a
1720 special type of small virtual machine) using L<qemu(1)>. Qemu runs as
1721 a child process of the main program.
1727 | | child process / appliance
1728 | | __________________________
1730 +-------------------+ RPC | +-----------------+ |
1731 | libguestfs <--------------------> guestfsd | |
1732 | | | +-----------------+ |
1733 \___________________/ | | Linux kernel | |
1734 | +--^--------------+ |
1735 \_________|________________/
1743 The library, linked to the main program, creates the child process and
1744 hence the appliance in the L</guestfs_launch> function.
1746 Inside the appliance is a Linux kernel and a complete stack of
1747 userspace tools (such as LVM and ext2 programs) and a small
1748 controlling daemon called L</guestfsd>. The library talks to
1749 L</guestfsd> using remote procedure calls (RPC). There is a mostly
1750 one-to-one correspondence between libguestfs API calls and RPC calls
1751 to the daemon. Lastly the disk image(s) are attached to the qemu
1752 process which translates device access by the appliance's Linux kernel
1753 into accesses to the image.
1755 A common misunderstanding is that the appliance "is" the virtual
1756 machine. Although the disk image you are attached to might also be
1757 used by some virtual machine, libguestfs doesn't know or care about
1758 this. (But you will care if both libguestfs's qemu process and your
1759 virtual machine are trying to update the disk image at the same time,
1760 since these usually results in massive disk corruption).
1762 =head1 STATE MACHINE
1764 libguestfs uses a state machine to model the child process:
1775 / | \ \ guestfs_launch
1786 \______/ <------ \________/
1788 The normal transitions are (1) CONFIG (when the handle is created, but
1789 there is no child process), (2) LAUNCHING (when the child process is
1790 booting up), (3) alternating between READY and BUSY as commands are
1791 issued to, and carried out by, the child process.
1793 The guest may be killed by L</guestfs_kill_subprocess>, or may die
1794 asynchronously at any time (eg. due to some internal error), and that
1795 causes the state to transition back to CONFIG.
1797 Configuration commands for qemu such as L</guestfs_add_drive> can only
1798 be issued when in the CONFIG state.
1800 The API offers one call that goes from CONFIG through LAUNCHING to
1801 READY. L</guestfs_launch> blocks until the child process is READY to
1802 accept commands (or until some failure or timeout).
1803 L</guestfs_launch> internally moves the state from CONFIG to LAUNCHING
1804 while it is running.
1806 API actions such as L</guestfs_mount> can only be issued when in the
1807 READY state. These API calls block waiting for the command to be
1808 carried out (ie. the state to transition to BUSY and then back to
1809 READY). There are no non-blocking versions, and no way to issue more
1810 than one command per handle at the same time.
1812 Finally, the child process sends asynchronous messages back to the
1813 main program, such as kernel log messages. You can register a
1814 callback to receive these messages.
1818 =head2 COMMUNICATION PROTOCOL
1820 Don't rely on using this protocol directly. This section documents
1821 how it currently works, but it may change at any time.
1823 The protocol used to talk between the library and the daemon running
1824 inside the qemu virtual machine is a simple RPC mechanism built on top
1825 of XDR (RFC 1014, RFC 1832, RFC 4506).
1827 The detailed format of structures is in C<src/guestfs_protocol.x>
1828 (note: this file is automatically generated).
1830 There are two broad cases, ordinary functions that don't have any
1831 C<FileIn> and C<FileOut> parameters, which are handled with very
1832 simple request/reply messages. Then there are functions that have any
1833 C<FileIn> or C<FileOut> parameters, which use the same request and
1834 reply messages, but they may also be followed by files sent using a
1837 =head3 ORDINARY FUNCTIONS (NO FILEIN/FILEOUT PARAMS)
1839 For ordinary functions, the request message is:
1841 total length (header + arguments,
1842 but not including the length word itself)
1843 struct guestfs_message_header (encoded as XDR)
1844 struct guestfs_<foo>_args (encoded as XDR)
1846 The total length field allows the daemon to allocate a fixed size
1847 buffer into which it slurps the rest of the message. As a result, the
1848 total length is limited to C<GUESTFS_MESSAGE_MAX> bytes (currently
1849 4MB), which means the effective size of any request is limited to
1850 somewhere under this size.
1852 Note also that many functions don't take any arguments, in which case
1853 the C<guestfs_I<foo>_args> is completely omitted.
1855 The header contains the procedure number (C<guestfs_proc>) which is
1856 how the receiver knows what type of args structure to expect, or none
1859 The reply message for ordinary functions is:
1861 total length (header + ret,
1862 but not including the length word itself)
1863 struct guestfs_message_header (encoded as XDR)
1864 struct guestfs_<foo>_ret (encoded as XDR)
1866 As above the C<guestfs_I<foo>_ret> structure may be completely omitted
1867 for functions that return no formal return values.
1869 As above the total length of the reply is limited to
1870 C<GUESTFS_MESSAGE_MAX>.
1872 In the case of an error, a flag is set in the header, and the reply
1873 message is slightly changed:
1875 total length (header + error,
1876 but not including the length word itself)
1877 struct guestfs_message_header (encoded as XDR)
1878 struct guestfs_message_error (encoded as XDR)
1880 The C<guestfs_message_error> structure contains the error message as a
1883 =head3 FUNCTIONS THAT HAVE FILEIN PARAMETERS
1885 A C<FileIn> parameter indicates that we transfer a file I<into> the
1886 guest. The normal request message is sent (see above). However this
1887 is followed by a sequence of file chunks.
1889 total length (header + arguments,
1890 but not including the length word itself,
1891 and not including the chunks)
1892 struct guestfs_message_header (encoded as XDR)
1893 struct guestfs_<foo>_args (encoded as XDR)
1894 sequence of chunks for FileIn param #0
1895 sequence of chunks for FileIn param #1 etc.
1897 The "sequence of chunks" is:
1899 length of chunk (not including length word itself)
1900 struct guestfs_chunk (encoded as XDR)
1902 struct guestfs_chunk (encoded as XDR)
1905 struct guestfs_chunk (with data.data_len == 0)
1907 The final chunk has the C<data_len> field set to zero. Additionally a
1908 flag is set in the final chunk to indicate either successful
1909 completion or early cancellation.
1911 At time of writing there are no functions that have more than one
1912 FileIn parameter. However this is (theoretically) supported, by
1913 sending the sequence of chunks for each FileIn parameter one after
1914 another (from left to right).
1916 Both the library (sender) I<and> the daemon (receiver) may cancel the
1917 transfer. The library does this by sending a chunk with a special
1918 flag set to indicate cancellation. When the daemon sees this, it
1919 cancels the whole RPC, does I<not> send any reply, and goes back to
1920 reading the next request.
1922 The daemon may also cancel. It does this by writing a special word
1923 C<GUESTFS_CANCEL_FLAG> to the socket. The library listens for this
1924 during the transfer, and if it gets it, it will cancel the transfer
1925 (it sends a cancel chunk). The special word is chosen so that even if
1926 cancellation happens right at the end of the transfer (after the
1927 library has finished writing and has started listening for the reply),
1928 the "spurious" cancel flag will not be confused with the reply
1931 This protocol allows the transfer of arbitrary sized files (no 32 bit
1932 limit), and also files where the size is not known in advance
1933 (eg. from pipes or sockets). However the chunks are rather small
1934 (C<GUESTFS_MAX_CHUNK_SIZE>), so that neither the library nor the
1935 daemon need to keep much in memory.
1937 =head3 FUNCTIONS THAT HAVE FILEOUT PARAMETERS
1939 The protocol for FileOut parameters is exactly the same as for FileIn
1940 parameters, but with the roles of daemon and library reversed.
1942 total length (header + ret,
1943 but not including the length word itself,
1944 and not including the chunks)
1945 struct guestfs_message_header (encoded as XDR)
1946 struct guestfs_<foo>_ret (encoded as XDR)
1947 sequence of chunks for FileOut param #0
1948 sequence of chunks for FileOut param #1 etc.
1950 =head3 INITIAL MESSAGE
1952 When the daemon launches it sends an initial word
1953 (C<GUESTFS_LAUNCH_FLAG>) which indicates that the guest and daemon is
1954 alive. This is what L</guestfs_launch> waits for.
1956 =head3 PROGRESS NOTIFICATION MESSAGES
1958 The daemon may send progress notification messages at any time. These
1959 are distinguished by the normal length word being replaced by
1960 C<GUESTFS_PROGRESS_FLAG>, followed by a fixed size progress message.
1962 The library turns them into progress callbacks (see
1963 C<guestfs_set_progress_callback>) if there is a callback registered,
1964 or discards them if not.
1966 The daemon self-limits the frequency of progress messages it sends
1967 (see C<daemon/proto.c:notify_progress>). Not all calls generate
1970 =head1 LIBGUESTFS VERSION NUMBERS
1972 Since April 2010, libguestfs has started to make separate development
1973 and stable releases, along with corresponding branches in our git
1974 repository. These separate releases can be identified by version
1977 even numbers for stable: 1.2.x, 1.4.x, ...
1978 .-------- odd numbers for development: 1.3.x, 1.5.x, ...
1984 | `-------- sub-version
1986 `------ always '1' because we don't change the ABI
1988 Thus "1.3.5" is the 5th update to the development branch "1.3".
1990 As time passes we cherry pick fixes from the development branch and
1991 backport those into the stable branch, the effect being that the
1992 stable branch should get more stable and less buggy over time. So the
1993 stable releases are ideal for people who don't need new features but
1994 would just like the software to work.
1996 Our criteria for backporting changes are:
2002 Documentation changes which don't affect any code are
2003 backported unless the documentation refers to a future feature
2004 which is not in stable.
2008 Bug fixes which are not controversial, fix obvious problems, and
2009 have been well tested are backported.
2013 Simple rearrangements of code which shouldn't affect how it works get
2014 backported. This is so that the code in the two branches doesn't get
2015 too far out of step, allowing us to backport future fixes more easily.
2019 We I<don't> backport new features, new APIs, new tools etc, except in
2020 one exceptional case: the new feature is required in order to
2021 implement an important bug fix.
2025 A new stable branch starts when we think the new features in
2026 development are substantial and compelling enough over the current
2027 stable branch to warrant it. When that happens we create new stable
2028 and development versions 1.N.0 and 1.(N+1).0 [N is even]. The new
2029 dot-oh release won't necessarily be so stable at this point, but by
2030 backporting fixes from development, that branch will stabilize over
2033 =head1 ENVIRONMENT VARIABLES
2037 =item LIBGUESTFS_APPEND
2039 Pass additional options to the guest kernel.
2041 =item LIBGUESTFS_DEBUG
2043 Set C<LIBGUESTFS_DEBUG=1> to enable verbose messages. This
2044 has the same effect as calling C<guestfs_set_verbose (g, 1)>.
2046 =item LIBGUESTFS_MEMSIZE
2048 Set the memory allocated to the qemu process, in megabytes. For
2051 LIBGUESTFS_MEMSIZE=700
2053 =item LIBGUESTFS_PATH
2055 Set the path that libguestfs uses to search for kernel and initrd.img.
2056 See the discussion of paths in section PATH above.
2058 =item LIBGUESTFS_QEMU
2060 Set the default qemu binary that libguestfs uses. If not set, then
2061 the qemu which was found at compile time by the configure script is
2064 See also L</QEMU WRAPPERS> above.
2066 =item LIBGUESTFS_TRACE
2068 Set C<LIBGUESTFS_TRACE=1> to enable command traces. This
2069 has the same effect as calling C<guestfs_set_trace (g, 1)>.
2073 Location of temporary directory, defaults to C</tmp>.
2075 If libguestfs was compiled to use the supermin appliance then the
2076 real appliance is cached in this directory, shared between all
2077 handles belonging to the same EUID. You can use C<$TMPDIR> to
2078 configure another directory to use in case C</tmp> is not large
2085 L<guestfs-examples(3)>,
2086 L<guestfs-ocaml(3)>,
2087 L<guestfs-python(3)>,
2093 L<virt-filesystems(1)>,
2094 L<virt-inspector(1)>,
2095 L<virt-list-filesystems(1)>,
2096 L<virt-list-partitions(1)>,
2105 L<http://libguestfs.org/>.
2107 Tools with a similar purpose:
2116 To get a list of bugs against libguestfs use this link:
2118 L<https://bugzilla.redhat.com/buglist.cgi?component=libguestfs&product=Virtualization+Tools>
2120 To report a new bug against libguestfs use this link:
2122 L<https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virtualization+Tools>
2124 When reporting a bug, please check:
2130 That the bug hasn't been reported already.
2134 That you are testing a recent version.
2138 Describe the bug accurately, and give a way to reproduce it.
2142 Run libguestfs-test-tool and paste the B<complete, unedited>
2143 output into the bug report.
2149 Richard W.M. Jones (C<rjones at redhat dot com>)
2153 Copyright (C) 2009-2010 Red Hat Inc.
2154 L<http://libguestfs.org/>
2156 This library is free software; you can redistribute it and/or
2157 modify it under the terms of the GNU Lesser General Public
2158 License as published by the Free Software Foundation; either
2159 version 2 of the License, or (at your option) any later version.
2161 This library is distributed in the hope that it will be useful,
2162 but WITHOUT ANY WARRANTY; without even the implied warranty of
2163 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
2164 Lesser General Public License for more details.
2166 You should have received a copy of the GNU Lesser General Public
2167 License along with this library; if not, write to the Free Software
2168 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA