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.
122 The image filename (C<"guest.img"> in the example above) could be a
123 disk image from a virtual machine, a L<dd(1)> copy of a physical hard
124 disk, an actual block device, or simply an empty file of zeroes that
125 you have created through L<posix_fallocate(3)>. Libguestfs lets you
126 do useful things to all of these.
128 The call you should use in modern code for adding drives is
129 L</guestfs_add_drive_opts>. To add a disk image, allowing writes, and
130 specifying that the format is raw, do:
132 guestfs_add_drive_opts (g, filename,
133 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "raw",
136 You can add a disk read-only using:
138 guestfs_add_drive_opts (g, filename,
139 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "raw",
140 GUESTFS_ADD_DRIVE_OPTS_READONLY, 1,
143 or by calling the older function L</guestfs_add_drive_ro>. In either
144 case libguestfs won't modify the file.
146 Be extremely cautious if the disk image is in use, eg. if it is being
147 used by a virtual machine. Adding it read-write will almost certainly
148 cause disk corruption, but adding it read-only is safe.
150 You must add at least one disk image, and you may add multiple disk
151 images. In the API, the disk images are usually referred to as
152 C</dev/sda> (for the first one you added), C</dev/sdb> (for the second
155 Once L</guestfs_launch> has been called you cannot add any more images.
156 You can call L</guestfs_list_devices> to get a list of the device
157 names, in the order that you added them. See also L</BLOCK DEVICE
162 Before you can read or write files, create directories and so on in a
163 disk image that contains filesystems, you have to mount those
164 filesystems using L</guestfs_mount>. If you already know that a disk
165 image contains (for example) one partition with a filesystem on that
166 partition, then you can mount it directly:
168 guestfs_mount (g, "/dev/sda1", "/");
170 where C</dev/sda1> means literally the first partition (C<1>) of the
171 first disk image that we added (C</dev/sda>). If the disk contains
172 Linux LVM2 logical volumes you could refer to those instead (eg. C</dev/VG/LV>).
174 If you are given a disk image and you don't know what it contains then
175 you have to find out. Libguestfs can do that too: use
176 L</guestfs_list_partitions> and L</guestfs_lvs> to list possible
177 partitions and LVs, and either try mounting each to see what is
178 mountable, or else examine them with L</guestfs_vfs_type> or
179 L</guestfs_file>. Libguestfs also has a set of APIs for inspection of
180 disk images (see L</INSPECTION> below). But you might find it easier
181 to look at higher level programs built on top of libguestfs, in
182 particular L<virt-inspector(1)>.
184 To mount a disk image read-only, use L</guestfs_mount_ro>. There are
185 several other variations of the C<guestfs_mount_*> call.
187 =head2 FILESYSTEM ACCESS AND MODIFICATION
189 The majority of the libguestfs API consists of fairly low-level calls
190 for accessing and modifying the files, directories, symlinks etc on
191 mounted filesystems. There are over a hundred such calls which you
192 can find listed in detail below in this man page, and we don't even
193 pretend to cover them all in this overview.
195 Specify filenames as full paths, starting with C<"/"> and including
198 For example, if you mounted a filesystem at C<"/"> and you want to
199 read the file called C<"etc/passwd"> then you could do:
201 char *data = guestfs_cat (g, "/etc/passwd");
203 This would return C<data> as a newly allocated buffer containing the
204 full content of that file (with some conditions: see also
205 L</DOWNLOADING> below), or C<NULL> if there was an error.
207 As another example, to create a top-level directory on that filesystem
208 called C<"var"> you would do:
210 guestfs_mkdir (g, "/var");
212 To create a symlink you could do:
214 guestfs_ln_s (g, "/etc/init.d/portmap",
215 "/etc/rc3.d/S30portmap");
217 Libguestfs will reject attempts to use relative paths and there is no
218 concept of a current working directory.
220 Libguestfs can return errors in many situations: for example if the
221 filesystem isn't writable, or if a file or directory that you
222 requested doesn't exist. If you are using the C API (documented here)
223 you have to check for those error conditions after each call. (Other
224 language bindings turn these errors into exceptions).
226 File writes are affected by the per-handle umask, set by calling
227 L</guestfs_umask> and defaulting to 022. See L</UMASK>.
231 Libguestfs contains API calls to read, create and modify partition
232 tables on disk images.
234 In the common case where you want to create a single partition
235 covering the whole disk, you should use the L</guestfs_part_disk>
238 const char *parttype = "mbr";
239 if (disk_is_larger_than_2TB)
241 guestfs_part_disk (g, "/dev/sda", parttype);
243 Obviously this effectively wipes anything that was on that disk image
248 Libguestfs provides access to a large part of the LVM2 API, such as
249 L</guestfs_lvcreate> and L</guestfs_vgremove>. It won't make much sense
250 unless you familiarize yourself with the concepts of physical volumes,
251 volume groups and logical volumes.
253 This author strongly recommends reading the LVM HOWTO, online at
254 L<http://tldp.org/HOWTO/LVM-HOWTO/>.
258 Use L</guestfs_cat> to download small, text only files. This call
259 is limited to files which are less than 2 MB and which cannot contain
260 any ASCII NUL (C<\0>) characters. However it has a very simple
263 L</guestfs_read_file> can be used to read files which contain
264 arbitrary 8 bit data, since it returns a (pointer, size) pair.
265 However it is still limited to "small" files, less than 2 MB.
267 L</guestfs_download> can be used to download any file, with no
268 limits on content or size (even files larger than 4 GB).
270 To download multiple files, see L</guestfs_tar_out> and
275 It's often the case that you want to write a file or files to the disk
278 To write a small file with fixed content, use L</guestfs_write>. To
279 create a file of all zeroes, use L</guestfs_truncate_size> (sparse) or
280 L</guestfs_fallocate64> (with all disk blocks allocated). There are a
281 variety of other functions for creating test files, for example
282 L</guestfs_fill> and L</guestfs_fill_pattern>.
284 To upload a single file, use L</guestfs_upload>. This call has no
285 limits on file content or size (even files larger than 4 GB).
287 To upload multiple files, see L</guestfs_tar_in> and L</guestfs_tgz_in>.
289 However the fastest way to upload I<large numbers of arbitrary files>
290 is to turn them into a squashfs or CD ISO (see L<mksquashfs(8)> and
291 L<mkisofs(8)>), then attach this using L</guestfs_add_drive_ro>. If
292 you add the drive in a predictable way (eg. adding it last after all
293 other drives) then you can get the device name from
294 L</guestfs_list_devices> and mount it directly using
295 L</guestfs_mount_ro>. Note that squashfs images are sometimes
296 non-portable between kernel versions, and they don't support labels or
297 UUIDs. If you want to pre-build an image or you need to mount it
298 using a label or UUID, use an ISO image instead.
302 There are various different commands for copying between files and
303 devices and in and out of the guest filesystem. These are summarised
308 =item B<file> to B<file>
310 Use L</guestfs_cp> to copy a single file, or
311 L</guestfs_cp_a> to copy directories recursively.
313 =item B<file or device> to B<file or device>
315 Use L</guestfs_dd> which efficiently uses L<dd(1)>
316 to copy between files and devices in the guest.
318 Example: duplicate the contents of an LV:
320 guestfs_dd (g, "/dev/VG/Original", "/dev/VG/Copy");
322 The destination (C</dev/VG/Copy>) must be at least as large as the
323 source (C</dev/VG/Original>). To copy less than the whole
324 source device, use L</guestfs_copy_size>.
326 =item B<file on the host> to B<file or device>
328 Use L</guestfs_upload>. See L</UPLOADING> above.
330 =item B<file or device> to B<file on the host>
332 Use L</guestfs_download>. See L</DOWNLOADING> above.
338 L</guestfs_ll> is just designed for humans to read (mainly when using
339 the L<guestfish(1)>-equivalent command C<ll>).
341 L</guestfs_ls> is a quick way to get a list of files in a directory
342 from programs, as a flat list of strings.
344 L</guestfs_readdir> is a programmatic way to get a list of files in a
345 directory, plus additional information about each one. It is more
346 equivalent to using the L<readdir(3)> call on a local filesystem.
348 L</guestfs_find> and L</guestfs_find0> can be used to recursively list
351 =head2 RUNNING COMMANDS
353 Although libguestfs is primarily an API for manipulating files
354 inside guest images, we also provide some limited facilities for
355 running commands inside guests.
357 There are many limitations to this:
363 The kernel version that the command runs under will be different
364 from what it expects.
368 If the command needs to communicate with daemons, then most likely
369 they won't be running.
373 The command will be running in limited memory.
377 The network may not be available unless you enable it
378 (see L</guestfs_set_network>).
382 Only supports Linux guests (not Windows, BSD, etc).
386 Architecture limitations (eg. won't work for a PPC guest on
391 For SELinux guests, you may need to enable SELinux and load policy
392 first. See L</SELINUX> in this manpage.
396 I<Security:> It is not safe to run commands from untrusted, possibly
397 malicious guests. These commands may attempt to exploit your program
398 by sending unexpected output. They could also try to exploit the
399 Linux kernel or qemu provided by the libguestfs appliance. They could
400 use the network provided by the libguestfs appliance to bypass
401 ordinary network partitions and firewalls. They could use the
402 elevated privileges or different SELinux context of your program
405 A secure alternative is to use libguestfs to install a "firstboot"
406 script (a script which runs when the guest next boots normally), and
407 to have this script run the commands you want in the normal context of
408 the running guest, network security and so on. For information about
409 other security issues, see L</SECURITY>.
413 The two main API calls to run commands are L</guestfs_command> and
414 L</guestfs_sh> (there are also variations).
416 The difference is that L</guestfs_sh> runs commands using the shell, so
417 any shell globs, redirections, etc will work.
419 =head2 CONFIGURATION FILES
421 To read and write configuration files in Linux guest filesystems, we
422 strongly recommend using Augeas. For example, Augeas understands how
423 to read and write, say, a Linux shadow password file or X.org
424 configuration file, and so avoids you having to write that code.
426 The main Augeas calls are bound through the C<guestfs_aug_*> APIs. We
427 don't document Augeas itself here because there is excellent
428 documentation on the L<http://augeas.net/> website.
430 If you don't want to use Augeas (you fool!) then try calling
431 L</guestfs_read_lines> to get the file as a list of lines which
432 you can iterate over.
436 We support SELinux guests. To ensure that labeling happens correctly
437 in SELinux guests, you need to enable SELinux and load the guest's
444 Before launching, do:
446 guestfs_set_selinux (g, 1);
450 After mounting the guest's filesystem(s), load the policy. This
451 is best done by running the L<load_policy(8)> command in the
454 guestfs_sh (g, "/usr/sbin/load_policy");
456 (Older versions of C<load_policy> require you to specify the
457 name of the policy file).
461 Optionally, set the security context for the API. The correct
462 security context to use can only be known by inspecting the
463 guest. As an example:
465 guestfs_setcon (g, "unconfined_u:unconfined_r:unconfined_t:s0");
469 This will work for running commands and editing existing files.
471 When new files are created, you may need to label them explicitly,
472 for example by running the external command
473 C<restorecon pathname>.
477 Certain calls are affected by the current file mode creation mask (the
478 "umask"). In particular ones which create files or directories, such
479 as L</guestfs_touch>, L</guestfs_mknod> or L</guestfs_mkdir>. This
480 affects either the default mode that the file is created with or
481 modifies the mode that you supply.
483 The default umask is C<022>, so files are created with modes such as
484 C<0644> and directories with C<0755>.
486 There are two ways to avoid being affected by umask. Either set umask
487 to 0 (call C<guestfs_umask (g, 0)> early after launching). Or call
488 L</guestfs_chmod> after creating each file or directory.
490 For more information about umask, see L<umask(2)>.
492 =head2 ENCRYPTED DISKS
494 Libguestfs allows you to access Linux guests which have been
495 encrypted using whole disk encryption that conforms to the
496 Linux Unified Key Setup (LUKS) standard. This includes
497 nearly all whole disk encryption systems used by modern
500 Use L</guestfs_vfs_type> to identify LUKS-encrypted block
501 devices (it returns the string C<crypto_LUKS>).
503 Then open these devices by calling L</guestfs_luks_open>.
504 Obviously you will require the passphrase!
506 Opening a LUKS device creates a new device mapper device
507 called C</dev/mapper/mapname> (where C<mapname> is the
508 string you supply to L</guestfs_luks_open>).
509 Reads and writes to this mapper device are decrypted from and
510 encrypted to the underlying block device respectively.
512 LVM volume groups on the device can be made visible by calling
513 L</guestfs_vgscan> followed by L</guestfs_vg_activate_all>.
514 The logical volume(s) can now be mounted in the usual way.
516 Use the reverse process to close a LUKS device. Unmount
517 any logical volumes on it, deactivate the volume groups
518 by caling C<guestfs_vg_activate (g, 0, ["/dev/VG"])>.
519 Then close the mapper device by calling
520 L</guestfs_luks_close> on the C</dev/mapper/mapname>
521 device (I<not> the underlying encrypted block device).
525 Libguestfs has APIs for inspecting an unknown disk image to find out
526 if it contains operating systems. (These APIs used to be in a
527 separate Perl-only library called L<Sys::Guestfs::Lib(3)> but since
528 version 1.5.3 the most frequently used part of this library has been
529 rewritten in C and moved into the core code).
531 Add all disks belonging to the unknown virtual machine and call
532 L</guestfs_launch> in the usual way.
534 Then call L</guestfs_inspect_os>. This function uses other libguestfs
535 calls and certain heuristics, and returns a list of operating systems
536 that were found. An empty list means none were found. A single
537 element is the root filesystem of the operating system. For dual- or
538 multi-boot guests, multiple roots can be returned, each one
539 corresponding to a separate operating system. (Multi-boot virtual
540 machines are extremely rare in the world of virtualization, but since
541 this scenario can happen, we have built libguestfs to deal with it.)
543 For each root, you can then call various C<guestfs_inspect_get_*>
544 functions to get additional details about that operating system. For
545 example, call L</guestfs_inspect_get_type> to return the string
546 C<windows> or C<linux> for Windows and Linux-based operating systems
549 Un*x-like and Linux-based operating systems usually consist of several
550 filesystems which are mounted at boot time (for example, a separate
551 boot partition mounted on C</boot>). The inspection rules are able to
552 detect how filesystems correspond to mount points. Call
553 C<guestfs_inspect_get_mountpoints> to get this mapping. It might
554 return a hash table like this example:
557 / => /dev/vg_guest/lv_root
558 /usr => /dev/vg_guest/lv_usr
560 The caller can then make calls to L</guestfs_mount_options> to
561 mount the filesystems as suggested.
563 Be careful to mount filesystems in the right order (eg. C</> before
564 C</usr>). Sorting the keys of the hash by length, shortest first,
567 Inspection currently only works for some common operating systems.
568 Contributors are welcome to send patches for other operating systems
569 that we currently cannot detect.
571 Encrypted disks must be opened before inspection. See
572 L</ENCRYPTED DISKS> for more details. The L</guestfs_inspect_os>
573 function just ignores any encrypted devices.
575 A note on the implementation: The call L</guestfs_inspect_os> performs
576 inspection and caches the results in the guest handle. Subsequent
577 calls to C<guestfs_inspect_get_*> return this cached information, but
578 I<do not> re-read the disks. If you change the content of the guest
579 disks, you can redo inspection by calling L</guestfs_inspect_os>
580 again. (L</guestfs_inspect_list_applications> works a little
581 differently from the other calls and does read the disks. See
582 documentation for that function for details).
584 =head2 SPECIAL CONSIDERATIONS FOR WINDOWS GUESTS
586 Libguestfs can mount NTFS partitions. It does this using the
587 L<http://www.ntfs-3g.org/> driver.
589 DOS and Windows still use drive letters, and the filesystems are
590 always treated as case insensitive by Windows itself, and therefore
591 you might find a Windows configuration file referring to a path like
592 C<c:\windows\system32>. When the filesystem is mounted in libguestfs,
593 that directory might be referred to as C</WINDOWS/System32>.
595 Drive letter mappings are outside the scope of libguestfs. You have
596 to use libguestfs to read the appropriate Windows Registry and
597 configuration files, to determine yourself how drives are mapped (see
598 also L<hivex(3)> and L<virt-inspector(1)>).
600 Replacing backslash characters with forward slash characters is also
601 outside the scope of libguestfs, but something that you can easily do.
603 Where we can help is in resolving the case insensitivity of paths.
604 For this, call L</guestfs_case_sensitive_path>.
606 Libguestfs also provides some help for decoding Windows Registry
607 "hive" files, through the library C<hivex> which is part of the
608 libguestfs project although ships as a separate tarball. You have to
609 locate and download the hive file(s) yourself, and then pass them to
610 C<hivex> functions. See also the programs L<hivexml(1)>,
611 L<hivexsh(1)>, L<hivexregedit(1)> and L<virt-win-reg(1)> for more help
614 =head2 USING LIBGUESTFS WITH OTHER PROGRAMMING LANGUAGES
616 Although we don't want to discourage you from using the C API, we will
617 mention here that the same API is also available in other languages.
619 The API is broadly identical in all supported languages. This means
620 that the C call C<guestfs_mount(g,path)> is
621 C<$g-E<gt>mount($path)> in Perl, C<g.mount(path)> in Python,
622 and C<Guestfs.mount g path> in OCaml. In other words, a
623 straightforward, predictable isomorphism between each language.
625 Error messages are automatically transformed
626 into exceptions if the language supports it.
628 We don't try to "object orientify" parts of the API in OO languages,
629 although contributors are welcome to write higher level APIs above
630 what we provide in their favourite languages if they wish.
636 You can use the I<guestfs.h> header file from C++ programs. The C++
637 API is identical to the C API. C++ classes and exceptions are not
642 The C# bindings are highly experimental. Please read the warnings
643 at the top of C<csharp/Libguestfs.cs>.
647 This is the only language binding that is working but incomplete.
648 Only calls which return simple integers have been bound in Haskell,
649 and we are looking for help to complete this binding.
653 Full documentation is contained in the Javadoc which is distributed
658 For documentation see L<guestfs-ocaml(3)>.
662 For documentation see L<Sys::Guestfs(3)>.
666 For documentation see C<README-PHP> supplied with libguestfs
667 sources or in the php-libguestfs package for your distribution.
669 The PHP binding only works correctly on 64 bit machines.
673 For documentation see L<guestfs-python(3)>.
677 For documentation see L<guestfs-ruby(3)>.
679 =item B<shell scripts>
681 For documentation see L<guestfish(1)>.
685 =head2 LIBGUESTFS GOTCHAS
687 L<http://en.wikipedia.org/wiki/Gotcha_(programming)>: "A feature of a
688 system [...] that works in the way it is documented but is
689 counterintuitive and almost invites mistakes."
691 Since we developed libguestfs and the associated tools, there are
692 several things we would have designed differently, but are now stuck
693 with for backwards compatibility or other reasons. If there is ever a
694 libguestfs 2.0 release, you can expect these to change. Beware of
699 =item Autosync / forgetting to sync.
701 When modifying a filesystem from C or another language, you B<must>
702 unmount all filesystems and call L</guestfs_sync> explicitly before
703 you close the libguestfs handle. You can also call:
705 guestfs_set_autosync (g, 1);
707 to have the unmount/sync done automatically for you when the handle 'g'
708 is closed. (This feature is called "autosync", L</guestfs_set_autosync>
711 If you forget to do this, then it is entirely possible that your
712 changes won't be written out, or will be partially written, or (very
713 rarely) that you'll get disk corruption.
715 Note that in L<guestfish(3)> autosync is the default. So quick and
716 dirty guestfish scripts that forget to sync will work just fine, which
717 can make this very puzzling if you are trying to debug a problem.
719 Update: Autosync is enabled by default for all API users starting from
722 =item Mount option C<-o sync> should not be the default.
724 If you use L</guestfs_mount>, then C<-o sync,noatime> are added
725 implicitly. However C<-o sync> does not add any reliability benefit,
726 but does have a very large performance impact.
728 The work around is to use L</guestfs_mount_options> and set the mount
729 options that you actually want to use.
731 =item Read-only should be the default.
733 In L<guestfish(3)>, I<--ro> should be the default, and you should
734 have to specify I<--rw> if you want to make changes to the image.
736 This would reduce the potential to corrupt live VM images.
738 Note that many filesystems change the disk when you just mount and
739 unmount, even if you didn't perform any writes. You need to use
740 L</guestfs_add_drive_ro> to guarantee that the disk is not changed.
742 =item guestfish command line is hard to use.
744 C<guestfish disk.img> doesn't do what people expect (open C<disk.img>
745 for examination). It tries to run a guestfish command C<disk.img>
746 which doesn't exist, so it fails. In earlier versions of guestfish
747 the error message was also unintuitive, but we have corrected this
748 since. Like the Bourne shell, we should have used C<guestfish -c
749 command> to run commands.
751 =item guestfish megabyte modifiers don't work right on all commands
753 In recent guestfish you can use C<1M> to mean 1 megabyte (and
754 similarly for other modifiers). What guestfish actually does is to
755 multiply the number part by the modifier part and pass the result to
756 the C API. However this doesn't work for a few APIs which aren't
757 expecting bytes, but are already expecting some other unit
760 The most common is L</guestfs_lvcreate>. The guestfish command:
764 does not do what you might expect. Instead because
765 L</guestfs_lvcreate> is already expecting megabytes, this tries to
766 create a 100 I<terabyte> (100 megabytes * megabytes) logical volume.
767 The error message you get from this is also a little obscure.
769 This could be fixed in the generator by specially marking parameters
770 and return values which take bytes or other units.
772 =item Ambiguity between devices and paths
774 There is a subtle ambiguity in the API between a device name
775 (eg. C</dev/sdb2>) and a similar pathname. A file might just happen
776 to be called C<sdb2> in the directory C</dev> (consider some non-Unix
779 In the current API we usually resolve this ambiguity by having two
780 separate calls, for example L</guestfs_checksum> and
781 L</guestfs_checksum_device>. Some API calls are ambiguous and
782 (incorrectly) resolve the problem by detecting if the path supplied
783 begins with C</dev/>.
785 To avoid both the ambiguity and the need to duplicate some calls, we
786 could make paths/devices into structured names. One way to do this
787 would be to use a notation like grub (C<hd(0,0)>), although nobody
788 really likes this aspect of grub. Another way would be to use a
789 structured type, equivalent to this OCaml type:
791 type path = Path of string | Device of int | Partition of int * int
793 which would allow you to pass arguments like:
796 Device 1 (* /dev/sdb, or perhaps /dev/sda *)
797 Partition (1, 2) (* /dev/sdb2 (or is it /dev/sda2 or /dev/sdb3?) *)
798 Path "/dev/sdb2" (* not a device *)
800 As you can see there are still problems to resolve even with this
801 representation. Also consider how it might work in guestfish.
805 =head2 PROTOCOL LIMITS
807 Internally libguestfs uses a message-based protocol to pass API calls
808 and their responses to and from a small "appliance" (see L</INTERNALS>
809 for plenty more detail about this). The maximum message size used by
810 the protocol is slightly less than 4 MB. For some API calls you may
811 need to be aware of this limit. The API calls which may be affected
812 are individually documented, with a link back to this section of the
815 A simple call such as L</guestfs_cat> returns its result (the file
816 data) in a simple string. Because this string is at some point
817 internally encoded as a message, the maximum size that it can return
818 is slightly under 4 MB. If the requested file is larger than this
819 then you will get an error.
821 In order to transfer large files into and out of the guest filesystem,
822 you need to use particular calls that support this. The sections
823 L</UPLOADING> and L</DOWNLOADING> document how to do this.
825 You might also consider mounting the disk image using our FUSE
826 filesystem support (L<guestmount(1)>).
828 =head2 KEYS AND PASSPHRASES
830 Certain libguestfs calls take a parameter that contains sensitive key
831 material, passed in as a C string.
833 In the future we would hope to change the libguestfs implementation so
834 that keys are L<mlock(2)>-ed into physical RAM, and thus can never end
835 up in swap. However this is I<not> done at the moment, because of the
836 complexity of such an implementation.
838 Therefore you should be aware that any key parameter you pass to
839 libguestfs might end up being written out to the swap partition. If
840 this is a concern, scrub the swap partition or don't use libguestfs on
843 =head2 MULTIPLE HANDLES AND MULTIPLE THREADS
845 All high-level libguestfs actions are synchronous. If you want
846 to use libguestfs asynchronously then you must create a thread.
848 Only use the handle from a single thread. Either use the handle
849 exclusively from one thread, or provide your own mutex so that two
850 threads cannot issue calls on the same handle at the same time.
852 See the graphical program guestfs-browser for one possible
853 architecture for multithreaded programs using libvirt and libguestfs.
857 Libguestfs needs a kernel and initrd.img, which it finds by looking
858 along an internal path.
860 By default it looks for these in the directory C<$libdir/guestfs>
861 (eg. C</usr/local/lib/guestfs> or C</usr/lib64/guestfs>).
863 Use L</guestfs_set_path> or set the environment variable
864 L</LIBGUESTFS_PATH> to change the directories that libguestfs will
865 search in. The value is a colon-separated list of paths. The current
866 directory is I<not> searched unless the path contains an empty element
867 or C<.>. For example C<LIBGUESTFS_PATH=:/usr/lib/guestfs> would
868 search the current directory and then C</usr/lib/guestfs>.
872 If you want to compile your own qemu, run qemu from a non-standard
873 location, or pass extra arguments to qemu, then you can write a
874 shell-script wrapper around qemu.
876 There is one important rule to remember: you I<must C<exec qemu>> as
877 the last command in the shell script (so that qemu replaces the shell
878 and becomes the direct child of the libguestfs-using program). If you
879 don't do this, then the qemu process won't be cleaned up correctly.
881 Here is an example of a wrapper, where I have built my own copy of
885 qemudir=/home/rjones/d/qemu
886 exec $qemudir/x86_64-softmmu/qemu-system-x86_64 -L $qemudir/pc-bios "$@"
888 Save this script as C</tmp/qemu.wrapper> (or wherever), C<chmod +x>,
889 and then use it by setting the LIBGUESTFS_QEMU environment variable.
892 LIBGUESTFS_QEMU=/tmp/qemu.wrapper guestfish
894 Note that libguestfs also calls qemu with the -help and -version
895 options in order to determine features.
899 We guarantee the libguestfs ABI (binary interface), for public,
900 high-level actions as outlined in this section. Although we will
901 deprecate some actions, for example if they get replaced by newer
902 calls, we will keep the old actions forever. This allows you the
903 developer to program in confidence against the libguestfs API.
905 =head2 BLOCK DEVICE NAMING
907 In the kernel there is now quite a profusion of schemata for naming
908 block devices (in this context, by I<block device> I mean a physical
909 or virtual hard drive). The original Linux IDE driver used names
910 starting with C</dev/hd*>. SCSI devices have historically used a
911 different naming scheme, C</dev/sd*>. When the Linux kernel I<libata>
912 driver became a popular replacement for the old IDE driver
913 (particularly for SATA devices) those devices also used the
914 C</dev/sd*> scheme. Additionally we now have virtual machines with
915 paravirtualized drivers. This has created several different naming
916 systems, such as C</dev/vd*> for virtio disks and C</dev/xvd*> for Xen
919 As discussed above, libguestfs uses a qemu appliance running an
920 embedded Linux kernel to access block devices. We can run a variety
921 of appliances based on a variety of Linux kernels.
923 This causes a problem for libguestfs because many API calls use device
924 or partition names. Working scripts and the recipe (example) scripts
925 that we make available over the internet could fail if the naming
928 Therefore libguestfs defines C</dev/sd*> as the I<standard naming
929 scheme>. Internally C</dev/sd*> names are translated, if necessary,
930 to other names as required. For example, under RHEL 5 which uses the
931 C</dev/hd*> scheme, any device parameter C</dev/sda2> is translated to
932 C</dev/hda2> transparently.
934 Note that this I<only> applies to parameters. The
935 L</guestfs_list_devices>, L</guestfs_list_partitions> and similar calls
936 return the true names of the devices and partitions as known to the
939 =head3 ALGORITHM FOR BLOCK DEVICE NAME TRANSLATION
941 Usually this translation is transparent. However in some (very rare)
942 cases you may need to know the exact algorithm. Such cases include
943 where you use L</guestfs_config> to add a mixture of virtio and IDE
944 devices to the qemu-based appliance, so have a mixture of C</dev/sd*>
945 and C</dev/vd*> devices.
947 The algorithm is applied only to I<parameters> which are known to be
948 either device or partition names. Return values from functions such
949 as L</guestfs_list_devices> are never changed.
955 Is the string a parameter which is a device or partition name?
959 Does the string begin with C</dev/sd>?
963 Does the named device exist? If so, we use that device.
964 However if I<not> then we continue with this algorithm.
968 Replace initial C</dev/sd> string with C</dev/hd>.
970 For example, change C</dev/sda2> to C</dev/hda2>.
972 If that named device exists, use it. If not, continue.
976 Replace initial C</dev/sd> string with C</dev/vd>.
978 If that named device exists, use it. If not, return an error.
982 =head3 PORTABILITY CONCERNS WITH BLOCK DEVICE NAMING
984 Although the standard naming scheme and automatic translation is
985 useful for simple programs and guestfish scripts, for larger programs
986 it is best not to rely on this mechanism.
988 Where possible for maximum future portability programs using
989 libguestfs should use these future-proof techniques:
995 Use L</guestfs_list_devices> or L</guestfs_list_partitions> to list
996 actual device names, and then use those names directly.
998 Since those device names exist by definition, they will never be
1003 Use higher level ways to identify filesystems, such as LVM names,
1004 UUIDs and filesystem labels.
1010 This section discusses security implications of using libguestfs,
1011 particularly with untrusted or malicious guests or disk images.
1013 =head2 GENERAL SECURITY CONSIDERATIONS
1015 Be careful with any files or data that you download from a guest (by
1016 "download" we mean not just the L</guestfs_download> command but any
1017 command that reads files, filenames, directories or anything else from
1018 a disk image). An attacker could manipulate the data to fool your
1019 program into doing the wrong thing. Consider cases such as:
1025 the data (file etc) not being present
1029 being present but empty
1033 being much larger than normal
1037 containing arbitrary 8 bit data
1041 being in an unexpected character encoding
1045 containing homoglyphs.
1049 =head2 SECURITY OF MOUNTING FILESYSTEMS
1051 When you mount a filesystem under Linux, mistakes in the kernel
1052 filesystem (VFS) module can sometimes be escalated into exploits by
1053 deliberately creating a malicious, malformed filesystem. These
1054 exploits are very severe for two reasons. Firstly there are very many
1055 filesystem drivers in the kernel, and many of them are infrequently
1056 used and not much developer attention has been paid to the code.
1057 Linux userspace helps potential crackers by detecting the filesystem
1058 type and automatically choosing the right VFS driver, even if that
1059 filesystem type is obscure or unexpected for the administrator.
1060 Secondly, a kernel-level exploit is like a local root exploit (worse
1061 in some ways), giving immediate and total access to the system right
1062 down to the hardware level.
1064 That explains why you should never mount a filesystem from an
1065 untrusted guest on your host kernel. How about libguestfs? We run a
1066 Linux kernel inside a qemu virtual machine, usually running as a
1067 non-root user. The attacker would need to write a filesystem which
1068 first exploited the kernel, and then exploited either qemu
1069 virtualization (eg. a faulty qemu driver) or the libguestfs protocol,
1070 and finally to be as serious as the host kernel exploit it would need
1071 to escalate its privileges to root. This multi-step escalation,
1072 performed by a static piece of data, is thought to be extremely hard
1073 to do, although we never say 'never' about security issues.
1075 In any case callers can reduce the attack surface by forcing the
1076 filesystem type when mounting (use L</guestfs_mount_vfs>).
1078 =head2 PROTOCOL SECURITY
1080 The protocol is designed to be secure, being based on RFC 4506 (XDR)
1081 with a defined upper message size. However a program that uses
1082 libguestfs must also take care - for example you can write a program
1083 that downloads a binary from a disk image and executes it locally, and
1084 no amount of protocol security will save you from the consequences.
1086 =head2 INSPECTION SECURITY
1088 Parts of the inspection API (see L</INSPECTION>) return untrusted
1089 strings directly from the guest, and these could contain any 8 bit
1090 data. Callers should be careful to escape these before printing them
1091 to a structured file (for example, use HTML escaping if creating a web
1094 Guest configuration may be altered in unusual ways by the
1095 administrator of the virtual machine, and may not reflect reality
1096 (particularly for untrusted or actively malicious guests). For
1097 example we parse the hostname from configuration files like
1098 C</etc/sysconfig/network> that we find in the guest, but the guest
1099 administrator can easily manipulate these files to provide the wrong
1102 The inspection API parses guest configuration using two external
1103 libraries: Augeas (Linux configuration) and hivex (Windows Registry).
1104 Both are designed to be robust in the face of malicious data, although
1105 denial of service attacks are still possible, for example with
1106 oversized configuration files.
1108 =head2 RUNNING UNTRUSTED GUEST COMMANDS
1110 Be very cautious about running commands from the guest. By running a
1111 command in the guest, you are giving CPU time to a binary that you do
1112 not control, under the same user account as the library, albeit
1113 wrapped in qemu virtualization. More information and alternatives can
1114 be found in the section L</RUNNING COMMANDS>.
1116 =head2 CVE-2010-3851
1118 https://bugzilla.redhat.com/642934
1120 This security bug concerns the automatic disk format detection that
1121 qemu does on disk images.
1123 A raw disk image is just the raw bytes, there is no header. Other
1124 disk images like qcow2 contain a special header. Qemu deals with this
1125 by looking for one of the known headers, and if none is found then
1126 assuming the disk image must be raw.
1128 This allows a guest which has been given a raw disk image to write
1129 some other header. At next boot (or when the disk image is accessed
1130 by libguestfs) qemu would do autodetection and think the disk image
1131 format was, say, qcow2 based on the header written by the guest.
1133 This in itself would not be a problem, but qcow2 offers many features,
1134 one of which is to allow a disk image to refer to another image
1135 (called the "backing disk"). It does this by placing the path to the
1136 backing disk into the qcow2 header. This path is not validated and
1137 could point to any host file (eg. "/etc/passwd"). The backing disk is
1138 then exposed through "holes" in the qcow2 disk image, which of course
1139 is completely under the control of the attacker.
1141 In libguestfs this is rather hard to exploit except under two
1148 You have enabled the network or have opened the disk in write mode.
1152 You are also running untrusted code from the guest (see
1153 L</RUNNING COMMANDS>).
1157 The way to avoid this is to specify the expected disk format when
1158 adding disks (the optional C<format> option to
1159 L</guestfs_add_drive_opts>). You should always do this if the disk is
1160 raw format, and it's a good idea for other cases too.
1162 For disks added from libvirt using calls like L</guestfs_add_domain>,
1163 the format is fetched from libvirt and passed through.
1165 For libguestfs tools, use the I<--format> command line parameter as
1168 =head1 CONNECTION MANAGEMENT
1172 C<guestfs_h> is the opaque type representing a connection handle.
1173 Create a handle by calling L</guestfs_create>. Call L</guestfs_close>
1174 to free the handle and release all resources used.
1176 For information on using multiple handles and threads, see the section
1177 L</MULTIPLE HANDLES AND MULTIPLE THREADS> below.
1179 =head2 guestfs_create
1181 guestfs_h *guestfs_create (void);
1183 Create a connection handle.
1185 You have to call L</guestfs_add_drive_opts> (or one of the equivalent
1186 calls) on the handle at least once.
1188 This function returns a non-NULL pointer to a handle on success or
1191 After configuring the handle, you have to call L</guestfs_launch>.
1193 You may also want to configure error handling for the handle. See
1194 L</ERROR HANDLING> section below.
1196 =head2 guestfs_close
1198 void guestfs_close (guestfs_h *g);
1200 This closes the connection handle and frees up all resources used.
1202 =head1 ERROR HANDLING
1204 API functions can return errors. For example, almost all functions
1205 that return C<int> will return C<-1> to indicate an error.
1207 Additional information is available for errors: an error message
1208 string and optionally an error number (errno) if the thing that failed
1211 You can get at the additional information about the last error on the
1212 handle by calling L</guestfs_last_error>, L</guestfs_last_errno>,
1213 and/or by setting up an error handler with
1214 L</guestfs_set_error_handler>.
1216 When the handle is created, a default error handler is installed which
1217 prints the error message string to C<stderr>. For small short-running
1218 command line programs it is sufficient to do:
1220 if (guestfs_launch (g) == -1)
1221 exit (EXIT_FAILURE);
1223 since the default error handler will ensure that an error message has
1224 been printed to C<stderr> before the program exits.
1226 For other programs the caller will almost certainly want to install an
1227 alternate error handler or do error handling in-line like this:
1229 g = guestfs_create ();
1231 /* This disables the default behaviour of printing errors
1233 guestfs_set_error_handler (g, NULL, NULL);
1235 if (guestfs_launch (g) == -1) {
1236 /* Examine the error message and print it etc. */
1237 char *msg = guestfs_last_error (g);
1238 int errnum = guestfs_last_errno (g);
1239 fprintf (stderr, "%s\n", msg);
1243 Out of memory errors are handled differently. The default action is
1244 to call L<abort(3)>. If this is undesirable, then you can set a
1245 handler using L</guestfs_set_out_of_memory_handler>.
1247 L</guestfs_create> returns C<NULL> if the handle cannot be created,
1248 and because there is no handle if this happens there is no way to get
1249 additional error information. However L</guestfs_create> is supposed
1250 to be a lightweight operation which can only fail because of
1251 insufficient memory (it returns NULL in this case).
1253 =head2 guestfs_last_error
1255 const char *guestfs_last_error (guestfs_h *g);
1257 This returns the last error message that happened on C<g>. If
1258 there has not been an error since the handle was created, then this
1261 The lifetime of the returned string is until the next error occurs, or
1262 L</guestfs_close> is called.
1264 =head2 guestfs_last_errno
1266 int guestfs_last_errno (guestfs_h *g);
1268 This returns the last error number (errno) that happened on C<g>.
1270 If successful, an errno integer not equal to zero is returned.
1272 If no error, this returns 0. This call can return 0 in three
1279 There has not been any error on the handle.
1283 There has been an error but the errno was meaningless. This
1284 corresponds to the case where the error did not come from a
1285 failed system call, but for some other reason.
1289 There was an error from a failed system call, but for some
1290 reason the errno was not captured and returned. This usually
1291 indicates a bug in libguestfs.
1295 Libguestfs tries to convert the errno from inside the applicance into
1296 a corresponding errno for the caller (not entirely trivial: the
1297 appliance might be running a completely different operating system
1298 from the library and error numbers are not standardized across
1299 Un*xen). If this could not be done, then the error is translated to
1300 C<EINVAL>. In practice this should only happen in very rare
1303 =head2 guestfs_set_error_handler
1305 typedef void (*guestfs_error_handler_cb) (guestfs_h *g,
1308 void guestfs_set_error_handler (guestfs_h *g,
1309 guestfs_error_handler_cb cb,
1312 The callback C<cb> will be called if there is an error. The
1313 parameters passed to the callback are an opaque data pointer and the
1314 error message string.
1316 C<errno> is not passed to the callback. To get that the callback must
1317 call L</guestfs_last_errno>.
1319 Note that the message string C<msg> is freed as soon as the callback
1320 function returns, so if you want to stash it somewhere you must make
1323 The default handler prints messages on C<stderr>.
1325 If you set C<cb> to C<NULL> then I<no> handler is called.
1327 =head2 guestfs_get_error_handler
1329 guestfs_error_handler_cb guestfs_get_error_handler (guestfs_h *g,
1332 Returns the current error handler callback.
1334 =head2 guestfs_set_out_of_memory_handler
1336 typedef void (*guestfs_abort_cb) (void);
1337 int guestfs_set_out_of_memory_handler (guestfs_h *g,
1340 The callback C<cb> will be called if there is an out of memory
1341 situation. I<Note this callback must not return>.
1343 The default is to call L<abort(3)>.
1345 You cannot set C<cb> to C<NULL>. You can't ignore out of memory
1348 =head2 guestfs_get_out_of_memory_handler
1350 guestfs_abort_fn guestfs_get_out_of_memory_handler (guestfs_h *g);
1352 This returns the current out of memory handler.
1364 =head2 GROUPS OF FUNCTIONALITY IN THE APPLIANCE
1366 Using L</guestfs_available> you can test availability of
1367 the following groups of functions. This test queries the
1368 appliance to see if the appliance you are currently using
1369 supports the functionality.
1373 =head2 GUESTFISH supported COMMAND
1375 In L<guestfish(3)> there is a handy interactive command
1376 C<supported> which prints out the available groups and
1377 whether they are supported by this build of libguestfs.
1378 Note however that you have to do C<run> first.
1380 =head2 SINGLE CALLS AT COMPILE TIME
1382 Since version 1.5.8, C<E<lt>guestfs.hE<gt>> defines symbols
1383 for each C API function, such as:
1385 #define LIBGUESTFS_HAVE_DD 1
1387 if L</guestfs_dd> is available.
1389 Before version 1.5.8, if you needed to test whether a single
1390 libguestfs function is available at compile time, we recommended using
1391 build tools such as autoconf or cmake. For example in autotools you
1394 AC_CHECK_LIB([guestfs],[guestfs_create])
1395 AC_CHECK_FUNCS([guestfs_dd])
1397 which would result in C<HAVE_GUESTFS_DD> being either defined
1398 or not defined in your program.
1400 =head2 SINGLE CALLS AT RUN TIME
1402 Testing at compile time doesn't guarantee that a function really
1403 exists in the library. The reason is that you might be dynamically
1404 linked against a previous I<libguestfs.so> (dynamic library)
1405 which doesn't have the call. This situation unfortunately results
1406 in a segmentation fault, which is a shortcoming of the C dynamic
1407 linking system itself.
1409 You can use L<dlopen(3)> to test if a function is available
1410 at run time, as in this example program (note that you still
1411 need the compile time check as well):
1417 #include <guestfs.h>
1421 #ifdef LIBGUESTFS_HAVE_DD
1425 /* Test if the function guestfs_dd is really available. */
1426 dl = dlopen (NULL, RTLD_LAZY);
1428 fprintf (stderr, "dlopen: %s\n", dlerror ());
1429 exit (EXIT_FAILURE);
1431 has_function = dlsym (dl, "guestfs_dd") != NULL;
1435 printf ("this libguestfs.so does NOT have guestfs_dd function\n");
1437 printf ("this libguestfs.so has guestfs_dd function\n");
1438 /* Now it's safe to call
1439 guestfs_dd (g, "foo", "bar");
1443 printf ("guestfs_dd function was not found at compile time\n");
1447 You may think the above is an awful lot of hassle, and it is.
1448 There are other ways outside of the C linking system to ensure
1449 that this kind of incompatibility never arises, such as using
1452 Requires: libguestfs >= 1.0.80
1454 =head1 CALLS WITH OPTIONAL ARGUMENTS
1456 A recent feature of the API is the introduction of calls which take
1457 optional arguments. In C these are declared 3 ways. The main way is
1458 as a call which takes variable arguments (ie. C<...>), as in this
1461 int guestfs_add_drive_opts (guestfs_h *g, const char *filename, ...);
1463 Call this with a list of optional arguments, terminated by C<-1>.
1464 So to call with no optional arguments specified:
1466 guestfs_add_drive_opts (g, filename, -1);
1468 With a single optional argument:
1470 guestfs_add_drive_opts (g, filename,
1471 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "qcow2",
1476 guestfs_add_drive_opts (g, filename,
1477 GUESTFS_ADD_DRIVE_OPTS_FORMAT, "qcow2",
1478 GUESTFS_ADD_DRIVE_OPTS_READONLY, 1,
1481 and so forth. Don't forget the terminating C<-1> otherwise
1482 Bad Things will happen!
1484 =head2 USING va_list FOR OPTIONAL ARGUMENTS
1486 The second variant has the same name with the suffix C<_va>, which
1487 works the same way but takes a C<va_list>. See the C manual for
1488 details. For the example function, this is declared:
1490 int guestfs_add_drive_opts_va (guestfs_h *g, const char *filename,
1493 =head2 CONSTRUCTING OPTIONAL ARGUMENTS
1495 The third variant is useful where you need to construct these
1496 calls. You pass in a structure where you fill in the optional
1497 fields. The structure has a bitmask as the first element which
1498 you must set to indicate which fields you have filled in. For
1499 our example function the structure and call are declared:
1501 struct guestfs_add_drive_opts_argv {
1507 int guestfs_add_drive_opts_argv (guestfs_h *g, const char *filename,
1508 const struct guestfs_add_drive_opts_argv *optargs);
1510 You could call it like this:
1512 struct guestfs_add_drive_opts_argv optargs = {
1513 .bitmask = GUESTFS_ADD_DRIVE_OPTS_READONLY_BITMASK |
1514 GUESTFS_ADD_DRIVE_OPTS_FORMAT_BITMASK,
1519 guestfs_add_drive_opts_argv (g, filename, &optargs);
1527 The C<_BITMASK> suffix on each option name when specifying the
1532 You do not need to fill in all fields of the structure.
1536 There must be a one-to-one correspondence between fields of the
1537 structure that are filled in, and bits set in the bitmask.
1541 =head2 OPTIONAL ARGUMENTS IN OTHER LANGUAGES
1543 In other languages, optional arguments are expressed in the
1544 way that is natural for that language. We refer you to the
1545 language-specific documentation for more details on that.
1547 For guestfish, see L<guestfish(1)/OPTIONAL ARGUMENTS>.
1549 =head2 SETTING CALLBACKS TO HANDLE EVENTS
1551 The child process generates events in some situations. Current events
1552 include: receiving a log message, the child process exits.
1554 Use the C<guestfs_set_*_callback> functions to set a callback for
1555 different types of events.
1557 Only I<one callback of each type> can be registered for each handle.
1558 Calling C<guestfs_set_*_callback> again overwrites the previous
1559 callback of that type. Cancel all callbacks of this type by calling
1560 this function with C<cb> set to C<NULL>.
1562 =head2 guestfs_set_log_message_callback
1564 typedef void (*guestfs_log_message_cb) (guestfs_h *g, void *opaque,
1565 char *buf, int len);
1566 void guestfs_set_log_message_callback (guestfs_h *g,
1567 guestfs_log_message_cb cb,
1570 The callback function C<cb> will be called whenever qemu or the guest
1571 writes anything to the console.
1573 Use this function to capture kernel messages and similar.
1575 Normally there is no log message handler, and log messages are just
1578 =head2 guestfs_set_subprocess_quit_callback
1580 typedef void (*guestfs_subprocess_quit_cb) (guestfs_h *g, void *opaque);
1581 void guestfs_set_subprocess_quit_callback (guestfs_h *g,
1582 guestfs_subprocess_quit_cb cb,
1585 The callback function C<cb> will be called when the child process
1586 quits, either asynchronously or if killed by
1587 L</guestfs_kill_subprocess>. (This corresponds to a transition from
1588 any state to the CONFIG state).
1590 =head2 guestfs_set_launch_done_callback
1592 typedef void (*guestfs_launch_done_cb) (guestfs_h *g, void *opaque);
1593 void guestfs_set_launch_done_callback (guestfs_h *g,
1594 guestfs_launch_done_cb cb,
1597 The callback function C<cb> will be called when the child process
1598 becomes ready first time after it has been launched. (This
1599 corresponds to a transition from LAUNCHING to the READY state).
1601 =head2 guestfs_set_close_callback
1603 typedef void (*guestfs_close_cb) (guestfs_h *g, void *opaque);
1604 void guestfs_set_close_callback (guestfs_h *g,
1605 guestfs_close_cb cb,
1608 The callback function C<cb> will be called while the handle
1609 is being closed (synchronously from L</guestfs_close>).
1611 Note that libguestfs installs an L<atexit(3)> handler to try to
1612 clean up handles that are open when the program exits. This
1613 means that this callback might be called indirectly from
1614 L<exit(3)>, which can cause unexpected problems in higher-level
1615 languages (eg. if your HLL interpreter has already been cleaned
1616 up by the time this is called, and if your callback then jumps
1617 into some HLL function).
1619 =head2 guestfs_set_progress_callback
1621 typedef void (*guestfs_progress_cb) (guestfs_h *g, void *opaque,
1622 int proc_nr, int serial,
1623 uint64_t position, uint64_t total);
1624 void guestfs_set_progress_callback (guestfs_h *g,
1625 guestfs_progress_cb cb,
1628 Some long-running operations can generate progress messages. If
1629 this callback is registered, then it will be called each time a
1630 progress message is generated (usually two seconds after the
1631 operation started, and three times per second thereafter until
1632 it completes, although the frequency may change in future versions).
1634 The callback receives two numbers: C<position> and C<total>.
1635 The units of C<total> are not defined, although for some
1636 operations C<total> may relate in some way to the amount of
1637 data to be transferred (eg. in bytes or megabytes), and
1638 C<position> may be the portion which has been transferred.
1640 The only defined and stable parts of the API are:
1646 The callback can display to the user some type of progress bar or
1647 indicator which shows the ratio of C<position>:C<total>.
1651 0 E<lt>= C<position> E<lt>= C<total>
1655 If any progress notification is sent during a call, then a final
1656 progress notification is always sent when C<position> = C<total>.
1658 This is to simplify caller code, so callers can easily set the
1659 progress indicator to "100%" at the end of the operation, without
1660 requiring special code to detect this case.
1664 The callback also receives the procedure number and serial number of
1665 the call. These are only useful for debugging protocol issues, and
1666 the callback can normally ignore them. The callback may want to
1667 print these numbers in error messages or debugging messages.
1669 =head1 PRIVATE DATA AREA
1671 You can attach named pieces of private data to the libguestfs handle,
1672 and fetch them by name for the lifetime of the handle. This is called
1673 the private data area and is only available from the C API.
1675 To attach a named piece of data, use the following call:
1677 void guestfs_set_private (guestfs_h *g, const char *key, void *data);
1679 C<key> is the name to associate with this data, and C<data> is an
1680 arbitrary pointer (which can be C<NULL>). Any previous item with the
1681 same name is overwritten.
1683 You can use any C<key> you want, but names beginning with an
1684 underscore character are reserved for internal libguestfs purposes
1685 (for implementing language bindings). It is recommended to prefix the
1686 name with some unique string to avoid collisions with other users.
1688 To retrieve the pointer, use:
1690 void *guestfs_get_private (guestfs_h *g, const char *key);
1692 This function returns C<NULL> if either no data is found associated
1693 with C<key>, or if the user previously set the C<key>'s C<data>
1696 Libguestfs does not try to look at or interpret the C<data> pointer in
1697 any way. As far as libguestfs is concerned, it need not be a valid
1698 pointer at all. In particular, libguestfs does I<not> try to free the
1699 data when the handle is closed. If the data must be freed, then the
1700 caller must either free it before calling L</guestfs_close> or must
1701 set up a close callback to do it (see L</guestfs_set_close_callback>,
1702 and note that only one callback can be registered for a handle).
1704 The private data area is implemented using a hash table, and should be
1705 reasonably efficient for moderate numbers of keys.
1709 <!-- old anchor for the next section -->
1710 <a name="state_machine_and_low_level_event_api"/>
1716 Internally, libguestfs is implemented by running an appliance (a
1717 special type of small virtual machine) using L<qemu(1)>. Qemu runs as
1718 a child process of the main program.
1724 | | child process / appliance
1725 | | __________________________
1727 +-------------------+ RPC | +-----------------+ |
1728 | libguestfs <--------------------> guestfsd | |
1729 | | | +-----------------+ |
1730 \___________________/ | | Linux kernel | |
1731 | +--^--------------+ |
1732 \_________|________________/
1740 The library, linked to the main program, creates the child process and
1741 hence the appliance in the L</guestfs_launch> function.
1743 Inside the appliance is a Linux kernel and a complete stack of
1744 userspace tools (such as LVM and ext2 programs) and a small
1745 controlling daemon called L</guestfsd>. The library talks to
1746 L</guestfsd> using remote procedure calls (RPC). There is a mostly
1747 one-to-one correspondence between libguestfs API calls and RPC calls
1748 to the daemon. Lastly the disk image(s) are attached to the qemu
1749 process which translates device access by the appliance's Linux kernel
1750 into accesses to the image.
1752 A common misunderstanding is that the appliance "is" the virtual
1753 machine. Although the disk image you are attached to might also be
1754 used by some virtual machine, libguestfs doesn't know or care about
1755 this. (But you will care if both libguestfs's qemu process and your
1756 virtual machine are trying to update the disk image at the same time,
1757 since these usually results in massive disk corruption).
1759 =head1 STATE MACHINE
1761 libguestfs uses a state machine to model the child process:
1772 / | \ \ guestfs_launch
1783 \______/ <------ \________/
1785 The normal transitions are (1) CONFIG (when the handle is created, but
1786 there is no child process), (2) LAUNCHING (when the child process is
1787 booting up), (3) alternating between READY and BUSY as commands are
1788 issued to, and carried out by, the child process.
1790 The guest may be killed by L</guestfs_kill_subprocess>, or may die
1791 asynchronously at any time (eg. due to some internal error), and that
1792 causes the state to transition back to CONFIG.
1794 Configuration commands for qemu such as L</guestfs_add_drive> can only
1795 be issued when in the CONFIG state.
1797 The API offers one call that goes from CONFIG through LAUNCHING to
1798 READY. L</guestfs_launch> blocks until the child process is READY to
1799 accept commands (or until some failure or timeout).
1800 L</guestfs_launch> internally moves the state from CONFIG to LAUNCHING
1801 while it is running.
1803 API actions such as L</guestfs_mount> can only be issued when in the
1804 READY state. These API calls block waiting for the command to be
1805 carried out (ie. the state to transition to BUSY and then back to
1806 READY). There are no non-blocking versions, and no way to issue more
1807 than one command per handle at the same time.
1809 Finally, the child process sends asynchronous messages back to the
1810 main program, such as kernel log messages. You can register a
1811 callback to receive these messages.
1815 =head2 COMMUNICATION PROTOCOL
1817 Don't rely on using this protocol directly. This section documents
1818 how it currently works, but it may change at any time.
1820 The protocol used to talk between the library and the daemon running
1821 inside the qemu virtual machine is a simple RPC mechanism built on top
1822 of XDR (RFC 1014, RFC 1832, RFC 4506).
1824 The detailed format of structures is in C<src/guestfs_protocol.x>
1825 (note: this file is automatically generated).
1827 There are two broad cases, ordinary functions that don't have any
1828 C<FileIn> and C<FileOut> parameters, which are handled with very
1829 simple request/reply messages. Then there are functions that have any
1830 C<FileIn> or C<FileOut> parameters, which use the same request and
1831 reply messages, but they may also be followed by files sent using a
1834 =head3 ORDINARY FUNCTIONS (NO FILEIN/FILEOUT PARAMS)
1836 For ordinary functions, the request message is:
1838 total length (header + arguments,
1839 but not including the length word itself)
1840 struct guestfs_message_header (encoded as XDR)
1841 struct guestfs_<foo>_args (encoded as XDR)
1843 The total length field allows the daemon to allocate a fixed size
1844 buffer into which it slurps the rest of the message. As a result, the
1845 total length is limited to C<GUESTFS_MESSAGE_MAX> bytes (currently
1846 4MB), which means the effective size of any request is limited to
1847 somewhere under this size.
1849 Note also that many functions don't take any arguments, in which case
1850 the C<guestfs_I<foo>_args> is completely omitted.
1852 The header contains the procedure number (C<guestfs_proc>) which is
1853 how the receiver knows what type of args structure to expect, or none
1856 For functions that take optional arguments, the optional arguments are
1857 encoded in the C<guestfs_I<foo>_args> structure in the same way as
1858 ordinary arguments. A bitmask in the header indicates which optional
1859 arguments are meaningful. The bitmask is also checked to see if it
1860 contains bits set which the daemon does not know about (eg. if more
1861 optional arguments were added in a later version of the library), and
1862 this causes the call to be rejected.
1864 The reply message for ordinary functions is:
1866 total length (header + ret,
1867 but not including the length word itself)
1868 struct guestfs_message_header (encoded as XDR)
1869 struct guestfs_<foo>_ret (encoded as XDR)
1871 As above the C<guestfs_I<foo>_ret> structure may be completely omitted
1872 for functions that return no formal return values.
1874 As above the total length of the reply is limited to
1875 C<GUESTFS_MESSAGE_MAX>.
1877 In the case of an error, a flag is set in the header, and the reply
1878 message is slightly changed:
1880 total length (header + error,
1881 but not including the length word itself)
1882 struct guestfs_message_header (encoded as XDR)
1883 struct guestfs_message_error (encoded as XDR)
1885 The C<guestfs_message_error> structure contains the error message as a
1888 =head3 FUNCTIONS THAT HAVE FILEIN PARAMETERS
1890 A C<FileIn> parameter indicates that we transfer a file I<into> the
1891 guest. The normal request message is sent (see above). However this
1892 is followed by a sequence of file chunks.
1894 total length (header + arguments,
1895 but not including the length word itself,
1896 and not including the chunks)
1897 struct guestfs_message_header (encoded as XDR)
1898 struct guestfs_<foo>_args (encoded as XDR)
1899 sequence of chunks for FileIn param #0
1900 sequence of chunks for FileIn param #1 etc.
1902 The "sequence of chunks" is:
1904 length of chunk (not including length word itself)
1905 struct guestfs_chunk (encoded as XDR)
1907 struct guestfs_chunk (encoded as XDR)
1910 struct guestfs_chunk (with data.data_len == 0)
1912 The final chunk has the C<data_len> field set to zero. Additionally a
1913 flag is set in the final chunk to indicate either successful
1914 completion or early cancellation.
1916 At time of writing there are no functions that have more than one
1917 FileIn parameter. However this is (theoretically) supported, by
1918 sending the sequence of chunks for each FileIn parameter one after
1919 another (from left to right).
1921 Both the library (sender) I<and> the daemon (receiver) may cancel the
1922 transfer. The library does this by sending a chunk with a special
1923 flag set to indicate cancellation. When the daemon sees this, it
1924 cancels the whole RPC, does I<not> send any reply, and goes back to
1925 reading the next request.
1927 The daemon may also cancel. It does this by writing a special word
1928 C<GUESTFS_CANCEL_FLAG> to the socket. The library listens for this
1929 during the transfer, and if it gets it, it will cancel the transfer
1930 (it sends a cancel chunk). The special word is chosen so that even if
1931 cancellation happens right at the end of the transfer (after the
1932 library has finished writing and has started listening for the reply),
1933 the "spurious" cancel flag will not be confused with the reply
1936 This protocol allows the transfer of arbitrary sized files (no 32 bit
1937 limit), and also files where the size is not known in advance
1938 (eg. from pipes or sockets). However the chunks are rather small
1939 (C<GUESTFS_MAX_CHUNK_SIZE>), so that neither the library nor the
1940 daemon need to keep much in memory.
1942 =head3 FUNCTIONS THAT HAVE FILEOUT PARAMETERS
1944 The protocol for FileOut parameters is exactly the same as for FileIn
1945 parameters, but with the roles of daemon and library reversed.
1947 total length (header + ret,
1948 but not including the length word itself,
1949 and not including the chunks)
1950 struct guestfs_message_header (encoded as XDR)
1951 struct guestfs_<foo>_ret (encoded as XDR)
1952 sequence of chunks for FileOut param #0
1953 sequence of chunks for FileOut param #1 etc.
1955 =head3 INITIAL MESSAGE
1957 When the daemon launches it sends an initial word
1958 (C<GUESTFS_LAUNCH_FLAG>) which indicates that the guest and daemon is
1959 alive. This is what L</guestfs_launch> waits for.
1961 =head3 PROGRESS NOTIFICATION MESSAGES
1963 The daemon may send progress notification messages at any time. These
1964 are distinguished by the normal length word being replaced by
1965 C<GUESTFS_PROGRESS_FLAG>, followed by a fixed size progress message.
1967 The library turns them into progress callbacks (see
1968 C<guestfs_set_progress_callback>) if there is a callback registered,
1969 or discards them if not.
1971 The daemon self-limits the frequency of progress messages it sends
1972 (see C<daemon/proto.c:notify_progress>). Not all calls generate
1975 =head1 LIBGUESTFS VERSION NUMBERS
1977 Since April 2010, libguestfs has started to make separate development
1978 and stable releases, along with corresponding branches in our git
1979 repository. These separate releases can be identified by version
1982 even numbers for stable: 1.2.x, 1.4.x, ...
1983 .-------- odd numbers for development: 1.3.x, 1.5.x, ...
1989 | `-------- sub-version
1991 `------ always '1' because we don't change the ABI
1993 Thus "1.3.5" is the 5th update to the development branch "1.3".
1995 As time passes we cherry pick fixes from the development branch and
1996 backport those into the stable branch, the effect being that the
1997 stable branch should get more stable and less buggy over time. So the
1998 stable releases are ideal for people who don't need new features but
1999 would just like the software to work.
2001 Our criteria for backporting changes are:
2007 Documentation changes which don't affect any code are
2008 backported unless the documentation refers to a future feature
2009 which is not in stable.
2013 Bug fixes which are not controversial, fix obvious problems, and
2014 have been well tested are backported.
2018 Simple rearrangements of code which shouldn't affect how it works get
2019 backported. This is so that the code in the two branches doesn't get
2020 too far out of step, allowing us to backport future fixes more easily.
2024 We I<don't> backport new features, new APIs, new tools etc, except in
2025 one exceptional case: the new feature is required in order to
2026 implement an important bug fix.
2030 A new stable branch starts when we think the new features in
2031 development are substantial and compelling enough over the current
2032 stable branch to warrant it. When that happens we create new stable
2033 and development versions 1.N.0 and 1.(N+1).0 [N is even]. The new
2034 dot-oh release won't necessarily be so stable at this point, but by
2035 backporting fixes from development, that branch will stabilize over
2038 =head1 ENVIRONMENT VARIABLES
2042 =item LIBGUESTFS_APPEND
2044 Pass additional options to the guest kernel.
2046 =item LIBGUESTFS_DEBUG
2048 Set C<LIBGUESTFS_DEBUG=1> to enable verbose messages. This
2049 has the same effect as calling C<guestfs_set_verbose (g, 1)>.
2051 =item LIBGUESTFS_MEMSIZE
2053 Set the memory allocated to the qemu process, in megabytes. For
2056 LIBGUESTFS_MEMSIZE=700
2058 =item LIBGUESTFS_PATH
2060 Set the path that libguestfs uses to search for kernel and initrd.img.
2061 See the discussion of paths in section PATH above.
2063 =item LIBGUESTFS_QEMU
2065 Set the default qemu binary that libguestfs uses. If not set, then
2066 the qemu which was found at compile time by the configure script is
2069 See also L</QEMU WRAPPERS> above.
2071 =item LIBGUESTFS_TRACE
2073 Set C<LIBGUESTFS_TRACE=1> to enable command traces. This
2074 has the same effect as calling C<guestfs_set_trace (g, 1)>.
2078 Location of temporary directory, defaults to C</tmp>.
2080 If libguestfs was compiled to use the supermin appliance then the
2081 real appliance is cached in this directory, shared between all
2082 handles belonging to the same EUID. You can use C<$TMPDIR> to
2083 configure another directory to use in case C</tmp> is not large
2090 L<guestfs-examples(3)>,
2091 L<guestfs-ocaml(3)>,
2092 L<guestfs-python(3)>,
2099 L<virt-filesystems(1)>,
2100 L<virt-inspector(1)>,
2101 L<virt-list-filesystems(1)>,
2102 L<virt-list-partitions(1)>,
2111 L<http://libguestfs.org/>.
2113 Tools with a similar purpose:
2122 To get a list of bugs against libguestfs use this link:
2124 L<https://bugzilla.redhat.com/buglist.cgi?component=libguestfs&product=Virtualization+Tools>
2126 To report a new bug against libguestfs use this link:
2128 L<https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virtualization+Tools>
2130 When reporting a bug, please check:
2136 That the bug hasn't been reported already.
2140 That you are testing a recent version.
2144 Describe the bug accurately, and give a way to reproduce it.
2148 Run libguestfs-test-tool and paste the B<complete, unedited>
2149 output into the bug report.
2155 Richard W.M. Jones (C<rjones at redhat dot com>)
2159 Copyright (C) 2009-2010 Red Hat Inc.
2160 L<http://libguestfs.org/>
2162 This library is free software; you can redistribute it and/or
2163 modify it under the terms of the GNU Lesser General Public
2164 License as published by the Free Software Foundation; either
2165 version 2 of the License, or (at your option) any later version.
2167 This library is distributed in the hope that it will be useful,
2168 but WITHOUT ANY WARRANTY; without even the implied warranty of
2169 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
2170 Lesser General Public License for more details.
2172 You should have received a copy of the GNU Lesser General Public
2173 License along with this library; if not, write to the Free Software
2174 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA