5 guestfs - Library for accessing and modifying virtual machine images
11 guestfs_h *handle = guestfs_create ();
12 guestfs_add_drive (handle, "guest.img");
13 guestfs_launch (handle);
14 guestfs_mount (handle, "/dev/sda1", "/");
15 guestfs_touch (handle, "/hello");
16 guestfs_sync (handle);
17 guestfs_close (handle);
21 Libguestfs is a library for accessing and modifying guest disk images.
22 Amongst the things this is good for: making batch configuration
23 changes to guests, getting disk used/free statistics (see also:
24 virt-df), migrating between virtualization systems (see also:
25 virt-p2v), performing partial backups, performing partial guest
26 clones, cloning guests and changing registry/UUID/hostname info, and
29 Libguestfs uses Linux kernel and qemu code, and can access any type of
30 guest filesystem that Linux and qemu can, including but not limited
31 to: ext2/3/4, btrfs, FAT and NTFS, LVM, many different disk partition
32 schemes, qcow, qcow2, vmdk.
34 Libguestfs provides ways to enumerate guest storage (eg. partitions,
35 LVs, what filesystem is in each LV, etc.). It can also run commands
36 in the context of the guest. Also you can access filesystems over FTP.
38 Libguestfs is a library that can be linked with C and C++ management
39 programs (or management programs written in OCaml, Perl, Python, Ruby,
40 Java, Haskell or C#). You can also use it from shell scripts or the
43 You don't need to be root to use libguestfs, although obviously you do
44 need enough permissions to access the disk images.
46 Libguestfs is a large API because it can do many things. For a gentle
47 introduction, please read the L</API OVERVIEW> section next.
51 This section provides a gentler overview of the libguestfs API. We
52 also try to group API calls together, where that may not be obvious
53 from reading about the individual calls below.
57 Before you can use libguestfs calls, you have to create a handle.
58 Then you must add at least one disk image to the handle, followed by
59 launching the handle, then performing whatever operations you want,
60 and finally closing the handle. So the general structure of all
61 libguestfs-using programs looks like this:
63 guestfs_h *handle = guestfs_create ();
65 /* Call guestfs_add_drive additional times if there are
66 * multiple disk images.
68 guestfs_add_drive (handle, "guest.img");
70 /* Most manipulation calls won't work until you've launched
71 * the handle. You have to do this _after_ adding drives
72 * and _before_ other commands.
74 guestfs_launch (handle);
76 /* Now you can examine what partitions, LVs etc are available.
78 char **partitions = guestfs_list_partitions (handle);
79 char **logvols = guestfs_lvs (handle);
81 /* To access a filesystem in the image, you must mount it.
83 guestfs_mount (handle, "/dev/sda1", "/");
85 /* Now you can perform filesystem actions on the guest
88 guestfs_touch (handle, "/hello");
90 /* You only need to call guestfs_sync if you have made
91 * changes to the guest image.
93 guestfs_sync (handle);
95 /* Close the handle. */
96 guestfs_close (handle);
98 The code above doesn't include any error checking. In real code you
99 should check return values carefully for errors. In general all
100 functions that return integers return C<-1> on error, and all
101 functions that return pointers return C<NULL> on error. See section
102 L</ERROR HANDLING> below for how to handle errors, and consult the
103 documentation for each function call below to see precisely how they
104 return error indications.
108 The image filename (C<"guest.img"> in the example above) could be a
109 disk image from a virtual machine, a L<dd(1)> copy of a physical hard
110 disk, an actual block device, or simply an empty file of zeroes that
111 you have created through L<posix_fallocate(3)>. Libguestfs lets you
112 do useful things to all of these.
114 You can add a disk read-only using C<guestfs_add_drive_ro>, in which
115 case libguestfs won't modify the file.
117 Be extremely cautious if the disk image is in use, eg. if it is being
118 used by a virtual machine. Adding it read-write will almost certainly
119 cause disk corruption, but adding it read-only is safe.
121 You must add at least one disk image, and you may add multiple disk
122 images. In the API, the disk images are usually referred to as
123 C</dev/sda> (for the first one you added), C</dev/sdb> (for the second
126 Once C<guestfs_launch> has been called you cannot add any more images.
127 You can call C<guestfs_list_devices> to get a list of the device
128 names, in the order that you added them. See also L</BLOCK DEVICE
133 Before you can read or write files, create directories and so on in a
134 disk image that contains filesystems, you have to mount those
135 filesystems using C<guestfs_mount>. If you already know that a disk
136 image contains (for example) one partition with a filesystem on that
137 partition, then you can mount it directly:
139 guestfs_mount (handle, "/dev/sda1", "/");
141 where C</dev/sda1> means literally the first partition (C<1>) of the
142 first disk image that we added (C</dev/sda>). If the disk contains
143 Linux LVM2 logical volumes you could refer to those instead (eg. C</dev/VG/LV>).
145 If you are given a disk image and you don't know what it contains then
146 you have to find out. Libguestfs can do that too: use
147 C<guestfs_list_partitions> and C<guestfs_lvs> to list possible
148 partitions and LVs, and either try mounting each to see what is
149 mountable, or else examine them with C<guestfs_file>. But you might
150 find it easier to look at higher level programs built on top of
151 libguestfs, in particular L<virt-inspector(1)>.
153 To mount a disk image read-only, use C<guestfs_mount_ro>. There are
154 several other variations of the C<guestfs_mount_*> call.
156 =head2 FILESYSTEM ACCESS AND MODIFICATION
158 The majority of the libguestfs API consists of fairly low-level calls
159 for accessing and modifying the files, directories, symlinks etc on
160 mounted filesystems. There are over a hundred such calls which you
161 can find listed in detail below in this man page, and we don't even
162 pretend to cover them all in this overview.
164 Specify filenames as full paths including the mount point.
166 For example, if you mounted a filesystem at C<"/"> and you want to
167 read the file called C<"etc/passwd"> then you could do:
169 char *data = guestfs_cat (handle, "/etc/passwd");
171 This would return C<data> as a newly allocated buffer containing the
172 full content of that file (with some conditions: see also
173 L</DOWNLOADING> below), or C<NULL> if there was an error.
175 As another example, to create a top-level directory on that filesystem
176 called C<"var"> you would do:
178 guestfs_mkdir (handle, "/var");
180 To create a symlink you could do:
182 guestfs_ln_s (handle, "/etc/init.d/portmap",
183 "/etc/rc3.d/S30portmap");
185 Libguestfs will reject attempts to use relative paths. There is no
186 concept of a current working directory. Libguestfs can return errors
187 in many situations: for example if the filesystem isn't writable, or
188 if a file or directory that you requested doesn't exist. If you are
189 using the C API (documented here) you have to check for those error
190 conditions after each call. (Other language bindings turn these
191 errors into exceptions).
193 File writes are affected by the per-handle umask, set by calling
194 C<guestfs_umask> and defaulting to 022.
198 Libguestfs contains API calls to read, create and modify partition
199 tables on disk images.
201 In the common case where you want to create a single partition
202 covering the whole disk, you should use the C<guestfs_part_disk>
205 const char *parttype = "mbr";
206 if (disk_is_larger_than_2TB)
208 guestfs_part_disk (g, "/dev/sda", parttype);
210 Obviously this effectively wipes anything that was on that disk image
213 In general MBR partitions are both unnecessarily complicated and
214 depend on archaic details, namely the Cylinder-Head-Sector (CHS)
215 geometry of the disk. C<guestfs_sfdiskM> can be used to
216 create more complex arrangements where the relative sizes are
217 expressed in megabytes instead of cylinders, which is a small win.
218 C<guestfs_sfdiskM> will choose the nearest cylinder to approximate the
219 requested size. There's a lot of crazy stuff to do with IDE and
220 virtio disks having different, incompatible CHS geometries, that you
221 probably don't want to know about.
223 My advice: make a single partition to cover the whole disk, then use
228 Libguestfs provides access to a large part of the LVM2 API, such as
229 C<guestfs_lvcreate> and C<guestfs_vgremove>. It won't make much sense
230 unless you familiarize yourself with the concepts of physical volumes,
231 volume groups and logical volumes.
233 This author strongly recommends reading the LVM HOWTO, online at
234 L<http://tldp.org/HOWTO/LVM-HOWTO/>.
238 Use C<guestfs_cat> to download small, text only files. This call
239 is limited to files which are less than 2 MB and which cannot contain
240 any ASCII NUL (C<\0>) characters. However it has a very simple
243 C<guestfs_read_file> can be used to read files which contain
244 arbitrary 8 bit data, since it returns a (pointer, size) pair.
245 However it is still limited to "small" files, less than 2 MB.
247 C<guestfs_download> can be used to download any file, with no
248 limits on content or size (even files larger than 4 GB).
250 To download multiple files, see C<guestfs_tar_out> and
255 It's often the case that you want to write a file or files to the disk
258 For small, single files, use C<guestfs_write_file>. This call
259 currently contains a bug which limits the call to plain text files
260 (not containing ASCII NUL characters).
262 To upload a single file, use C<guestfs_upload>. This call has no
263 limits on file content or size (even files larger than 4 GB).
265 To upload multiple files, see C<guestfs_tar_in> and C<guestfs_tgz_in>.
267 However the fastest way to upload I<large numbers of arbitrary files>
268 is to turn them into a squashfs or CD ISO (see L<mksquashfs(8)> and
269 L<mkisofs(8)>), then attach this using C<guestfs_add_drive_ro>. If
270 you add the drive in a predictable way (eg. adding it last after all
271 other drives) then you can get the device name from
272 C<guestfs_list_devices> and mount it directly using
273 C<guestfs_mount_ro>. Note that squashfs images are sometimes
274 non-portable between kernel versions, and they don't support labels or
275 UUIDs. If you want to pre-build an image or you need to mount it
276 using a label or UUID, use an ISO image instead.
280 There are various different commands for copying between files and
281 devices and in and out of the guest filesystem. These are summarised
286 =item B<file> to B<file>
288 Use L</guestfs_cp> to copy a single file, or
289 L</guestfs_cp_a> to copy directories recursively.
291 =item B<file or device> to B<file or device>
293 Use L</guestfs_dd> which efficiently uses L<dd(1)>
294 to copy between files and devices in the guest.
296 Example: duplicate the contents of an LV:
298 guestfs_dd (g, "/dev/VG/Original", "/dev/VG/Copy");
300 The destination (C</dev/VG/Copy>) must be at least as large as the
301 source (C</dev/VG/Original>).
303 =item B<file on the host> to B<file or device>
305 Use L</guestfs_upload>. See L</UPLOADING> above.
307 =item B<file or device> to B<file on the host>
309 Use L</guestfs_download>. See L</DOWNLOADING> above.
315 C<guestfs_ll> is just designed for humans to read (mainly when using
316 the L<guestfish(1)>-equivalent command C<ll>).
318 C<guestfs_ls> is a quick way to get a list of files in a directory
319 from programs, as a flat list of strings.
321 C<guestfs_readdir> is a programmatic way to get a list of files in a
322 directory, plus additional information about each one. It is more
323 equivalent to using the L<readdir(3)> call on a local filesystem.
325 C<guestfs_find> can be used to recursively list files.
327 =head2 RUNNING COMMANDS
329 Although libguestfs is a primarily an API for manipulating files
330 inside guest images, we also provide some limited facilities for
331 running commands inside guests.
333 There are many limitations to this:
339 The kernel version that the command runs under will be different
340 from what it expects.
344 If the command needs to communicate with daemons, then most likely
345 they won't be running.
349 The command will be running in limited memory.
353 Only supports Linux guests (not Windows, BSD, etc).
357 Architecture limitations (eg. won't work for a PPC guest on
362 For SELinux guests, you may need to enable SELinux and load policy
363 first. See L</SELINUX> in this manpage.
367 The two main API calls to run commands are C<guestfs_command> and
368 C<guestfs_sh> (there are also variations).
370 The difference is that C<guestfs_sh> runs commands using the shell, so
371 any shell globs, redirections, etc will work.
373 =head2 CONFIGURATION FILES
375 To read and write configuration files in Linux guest filesystems, we
376 strongly recommend using Augeas. For example, Augeas understands how
377 to read and write, say, a Linux shadow password file or X.org
378 configuration file, and so avoids you having to write that code.
380 The main Augeas calls are bound through the C<guestfs_aug_*> APIs. We
381 don't document Augeas itself here because there is excellent
382 documentation on the L<http://augeas.net/> website.
384 If you don't want to use Augeas (you fool!) then try calling
385 C<guestfs_read_lines> to get the file as a list of lines which
386 you can iterate over.
390 We support SELinux guests. To ensure that labeling happens correctly
391 in SELinux guests, you need to enable SELinux and load the guest's
398 Before launching, do:
400 guestfs_set_selinux (g, 1);
404 After mounting the guest's filesystem(s), load the policy. This
405 is best done by running the L<load_policy(8)> command in the
408 guestfs_sh (g, "/usr/sbin/load_policy");
410 (Older versions of C<load_policy> require you to specify the
411 name of the policy file).
415 Optionally, set the security context for the API. The correct
416 security context to use can only be known by inspecting the
417 guest. As an example:
419 guestfs_setcon (g, "unconfined_u:unconfined_r:unconfined_t:s0");
423 This will work for running commands and editing existing files.
425 When new files are created, you may need to label them explicitly,
426 for example by running the external command
427 C<restorecon pathname>.
429 =head2 SPECIAL CONSIDERATIONS FOR WINDOWS GUESTS
431 Libguestfs can mount NTFS partitions. It does this using the
432 L<http://www.ntfs-3g.org/> driver.
434 DOS and Windows still use drive letters, and the filesystems are
435 always treated as case insensitive by Windows itself, and therefore
436 you might find a Windows configuration file referring to a path like
437 C<c:\windows\system32>. When the filesystem is mounted in libguestfs,
438 that directory might be referred to as C</WINDOWS/System32>.
440 Drive letter mappings are outside the scope of libguestfs. You have
441 to use libguestfs to read the appropriate Windows Registry and
442 configuration files, to determine yourself how drives are mapped (see
443 also L<virt-inspector(1)>).
445 Replacing backslash characters with forward slash characters is also
446 outside the scope of libguestfs, but something that you can easily do.
448 Where we can help is in resolving the case insensitivity of paths.
449 For this, call C<guestfs_case_sensitive_path>.
451 Libguestfs also provides some help for decoding Windows Registry
452 "hive" files, through the library C<libhivex> which is part of
453 libguestfs. You have to locate and download the hive file(s)
454 yourself, and then pass them to C<libhivex> functions. See also the
455 programs L<hivexml(1)>, L<hivexget(1)> and L<virt-win-reg(1)> for more
458 =head2 USING LIBGUESTFS WITH OTHER PROGRAMMING LANGUAGES
460 Although we don't want to discourage you from using the C API, we will
461 mention here that the same API is also available in other languages.
463 The API is broadly identical in all supported languages. This means
464 that the C call C<guestfs_mount(handle,path)> is
465 C<$handle-E<gt>mount($path)> in Perl, C<handle.mount(path)> in Python,
466 and C<Guestfs.mount handle path> in OCaml. In other words, a
467 straightforward, predictable isomorphism between each language.
469 Error messages are automatically transformed
470 into exceptions if the language supports it.
472 We don't try to "object orientify" parts of the API in OO languages,
473 although contributors are welcome to write higher level APIs above
474 what we provide in their favourite languages if they wish.
480 You can use the I<guestfs.h> header file from C++ programs. The C++
481 API is identical to the C API. C++ classes and exceptions are
486 The C# bindings are highly experimental. Please read the warnings
487 at the top of C<csharp/Libguestfs.cs>.
491 This is the only language binding that working but incomplete. Only
492 calls which return simple integers have been bound in Haskell, and we
493 are looking for help to complete this binding.
497 Full documentation is contained in the Javadoc which is distributed
502 For documentation see the file C<guestfs.mli>.
506 For documentation see L<Sys::Guestfs(3)>.
510 For documentation do:
518 Use the Guestfs module. There is no Ruby-specific documentation, but
519 you can find examples written in Ruby in the libguestfs source.
521 =item B<shell scripts>
523 For documentation see L<guestfish(1)>.
527 =head2 LIBGUESTFS GOTCHAS
529 L<http://en.wikipedia.org/wiki/Gotcha_(programming)>: "A feature of a
530 system [...] that works in the way it is documented but is
531 counterintuitive and almost invites mistakes."
533 Since we developed libguestfs and the associated tools, there are
534 several things we would have designed differently, but are now stuck
535 with for backwards compatibility or other reasons. If there is ever a
536 libguestfs 2.0 release, you can expect these to change. Beware of
541 =item Autosync / forgetting to sync.
543 When modifying a filesystem from C or another language, you B<must>
544 unmount all filesystems and call L</guestfs_sync> explicitly before
545 you close the libguestfs handle. You can also call:
547 guestfs_set_autosync (handle, 1);
549 to have the unmount/sync done automatically for you when the handle is
550 closed. (This feature is called "autosync", L</guestfs_set_autosync>
553 If you forget to do this, then it is entirely possible that your
554 changes won't be written out, or will be partially written, or (very
555 rarely) that you'll get disk corruption.
557 Note that in L<guestfish(3)> I<autosync is the default>. So quick and
558 dirty guestfish scripts that forget to sync will work just fine, which
559 can make this extra-puzzling if you are trying to debug a problem.
561 =item Mount option C<-o sync> should not be the default.
563 If you use C<guestfs_mount>, then C<-o sync,noatime> are added
564 implicitly. However C<-o sync> does not add any reliability benefit,
565 but does have a very large performance impact.
567 The work around is to use C<guestfs_mount_options> and set the mount
568 options that you actually want to use.
570 =item Read-only should be the default.
572 In L<guestfish(3)>, I<--ro> should be the default, and you should
573 have to specify I<--rw> if you want to make changes to the image.
575 This would reduce the potential to corrupt live VM images.
577 Note that many filesystems change the disk when you just mount and
578 unmount, even if you didn't perform any writes. You need to use
579 C<guestfs_add_drive_ro> to guarantee that the disk is not changed.
581 =item guestfish command line is hard to use.
583 C<guestfish disk.img> doesn't do what people expect (open C<disk.img>
584 for examination). It tries to run a guestfish command C<disk.img>
585 which doesn't exist, so it fails, and it fails with a strange and
586 unintuitive error message. Like the Bourne shell, we should have used
587 C<guestfish -c command> to run commands.
591 =head1 CONNECTION MANAGEMENT
595 C<guestfs_h> is the opaque type representing a connection handle.
596 Create a handle by calling C<guestfs_create>. Call C<guestfs_close>
597 to free the handle and release all resources used.
599 For information on using multiple handles and threads, see the section
600 L</MULTIPLE HANDLES AND MULTIPLE THREADS> below.
602 =head2 guestfs_create
604 guestfs_h *guestfs_create (void);
606 Create a connection handle.
608 You have to call C<guestfs_add_drive> on the handle at least once.
610 This function returns a non-NULL pointer to a handle on success or
613 After configuring the handle, you have to call C<guestfs_launch>.
615 You may also want to configure error handling for the handle. See
616 L</ERROR HANDLING> section below.
620 void guestfs_close (guestfs_h *handle);
622 This closes the connection handle and frees up all resources used.
624 =head1 ERROR HANDLING
626 The convention in all functions that return C<int> is that they return
627 C<-1> to indicate an error. You can get additional information on
628 errors by calling C<guestfs_last_error> and/or by setting up an error
629 handler with C<guestfs_set_error_handler>.
631 The default error handler prints the information string to C<stderr>.
633 Out of memory errors are handled differently. The default action is
634 to call L<abort(3)>. If this is undesirable, then you can set a
635 handler using C<guestfs_set_out_of_memory_handler>.
637 =head2 guestfs_last_error
639 const char *guestfs_last_error (guestfs_h *handle);
641 This returns the last error message that happened on C<handle>. If
642 there has not been an error since the handle was created, then this
645 The lifetime of the returned string is until the next error occurs, or
646 C<guestfs_close> is called.
648 The error string is not localized (ie. is always in English), because
649 this makes searching for error messages in search engines give the
650 largest number of results.
652 =head2 guestfs_set_error_handler
654 typedef void (*guestfs_error_handler_cb) (guestfs_h *handle,
657 void guestfs_set_error_handler (guestfs_h *handle,
658 guestfs_error_handler_cb cb,
661 The callback C<cb> will be called if there is an error. The
662 parameters passed to the callback are an opaque data pointer and the
663 error message string.
665 Note that the message string C<msg> is freed as soon as the callback
666 function returns, so if you want to stash it somewhere you must make
669 The default handler prints messages on C<stderr>.
671 If you set C<cb> to C<NULL> then I<no> handler is called.
673 =head2 guestfs_get_error_handler
675 guestfs_error_handler_cb guestfs_get_error_handler (guestfs_h *handle,
678 Returns the current error handler callback.
680 =head2 guestfs_set_out_of_memory_handler
682 typedef void (*guestfs_abort_cb) (void);
683 int guestfs_set_out_of_memory_handler (guestfs_h *handle,
686 The callback C<cb> will be called if there is an out of memory
687 situation. I<Note this callback must not return>.
689 The default is to call L<abort(3)>.
691 You cannot set C<cb> to C<NULL>. You can't ignore out of memory
694 =head2 guestfs_get_out_of_memory_handler
696 guestfs_abort_fn guestfs_get_out_of_memory_handler (guestfs_h *handle);
698 This returns the current out of memory handler.
702 Libguestfs needs a kernel and initrd.img, which it finds by looking
703 along an internal path.
705 By default it looks for these in the directory C<$libdir/guestfs>
706 (eg. C</usr/local/lib/guestfs> or C</usr/lib64/guestfs>).
708 Use C<guestfs_set_path> or set the environment variable
709 C<LIBGUESTFS_PATH> to change the directories that libguestfs will
710 search in. The value is a colon-separated list of paths. The current
711 directory is I<not> searched unless the path contains an empty element
712 or C<.>. For example C<LIBGUESTFS_PATH=:/usr/lib/guestfs> would
713 search the current directory and then C</usr/lib/guestfs>.
715 =head1 HIGH-LEVEL API ACTIONS
719 We guarantee the libguestfs ABI (binary interface), for public,
720 high-level actions as outlined in this section. Although we will
721 deprecate some actions, for example if they get replaced by newer
722 calls, we will keep the old actions forever. This allows you the
723 developer to program in confidence against libguestfs.
733 =head2 GROUPS OF FUNCTIONALITY IN THE APPLIANCE
735 Using L</guestfs_available> you can test availability of
736 the following groups of functions. This test queries the
737 appliance to see if the appliance you are currently using
738 supports the functionality.
742 =head2 SINGLE CALLS AT COMPILE TIME
744 If you need to test whether a single libguestfs function is
745 available at compile time, we recommend using build tools
746 such as autoconf or cmake. For example in autotools you could
749 AC_CHECK_LIB([guestfs],[guestfs_create])
750 AC_CHECK_FUNCS([guestfs_dd])
752 which would result in C<HAVE_GUESTFS_DD> being either defined
753 or not defined in your program.
755 =head2 SINGLE CALLS AT RUN TIME
757 Testing at compile time doesn't guarantee that a function really
758 exists in the library. The reason is that you might be dynamically
759 linked against a previous I<libguestfs.so> (dynamic library)
760 which doesn't have the call. This situation unfortunately results
761 in a segmentation fault, which is a shortcoming of the C dynamic
762 linking system itself.
764 You can use L<dlopen(3)> to test if a function is available
765 at run time, as in this example program (note that you still
766 need the compile time check as well):
778 #ifdef HAVE_GUESTFS_DD
782 /* Test if the function guestfs_dd is really available. */
783 dl = dlopen (NULL, RTLD_LAZY);
785 fprintf (stderr, "dlopen: %s\n", dlerror ());
788 has_function = dlsym (dl, "guestfs_dd") != NULL;
792 printf ("this libguestfs.so does NOT have guestfs_dd function\n");
794 printf ("this libguestfs.so has guestfs_dd function\n");
795 /* Now it's safe to call
796 guestfs_dd (g, "foo", "bar");
800 printf ("guestfs_dd function was not found at compile time\n");
804 You may think the above is an awful lot of hassle, and it is.
805 There are other ways outside of the C linking system to ensure
806 that this kind of incompatibility never arises, such as using
809 Requires: libguestfs >= 1.0.80
813 <!-- old anchor for the next section -->
814 <a name="state_machine_and_low_level_event_api"/>
820 Internally, libguestfs is implemented by running an appliance (a
821 special type of small virtual machine) using L<qemu(1)>. Qemu runs as
822 a child process of the main program.
828 | | child process / appliance
829 | | __________________________
831 +-------------------+ RPC | +-----------------+ |
832 | libguestfs <--------------------> guestfsd | |
833 | | | +-----------------+ |
834 \___________________/ | | Linux kernel | |
835 | +--^--------------+ |
836 \_________|________________/
844 The library, linked to the main program, creates the child process and
845 hence the appliance in the L</guestfs_launch> function.
847 Inside the appliance is a Linux kernel and a complete stack of
848 userspace tools (such as LVM and ext2 programs) and a small
849 controlling daemon called C<guestfsd>. The library talks to
850 C<guestfsd> using remote procedure calls (RPC). There is a mostly
851 one-to-one correspondence between libguestfs API calls and RPC calls
852 to the daemon. Lastly the disk image(s) are attached to the qemu
853 process which translates device access by the appliance's Linux kernel
854 into accesses to the image.
856 A common misunderstanding is that the appliance "is" the virtual
857 machine. Although the disk image you are attached to might also be
858 used by some virtual machine, libguestfs doesn't know or care about
859 this. (But you will care if both libguestfs's qemu process and your
860 virtual machine are trying to update the disk image at the same time,
861 since these usually results in massive disk corruption).
865 libguestfs uses a state machine to model the child process:
876 / | \ \ guestfs_launch
887 \______/ <------ \________/
889 The normal transitions are (1) CONFIG (when the handle is created, but
890 there is no child process), (2) LAUNCHING (when the child process is
891 booting up), (3) alternating between READY and BUSY as commands are
892 issued to, and carried out by, the child process.
894 The guest may be killed by C<guestfs_kill_subprocess>, or may die
895 asynchronously at any time (eg. due to some internal error), and that
896 causes the state to transition back to CONFIG.
898 Configuration commands for qemu such as C<guestfs_add_drive> can only
899 be issued when in the CONFIG state.
901 The high-level API offers two calls that go from CONFIG through
902 LAUNCHING to READY. C<guestfs_launch> blocks until the child process
903 is READY to accept commands (or until some failure or timeout).
904 C<guestfs_launch> internally moves the state from CONFIG to LAUNCHING
907 High-level API actions such as C<guestfs_mount> can only be issued
908 when in the READY state. These high-level API calls block waiting for
909 the command to be carried out (ie. the state to transition to BUSY and
910 then back to READY). But using the low-level event API, you get
911 non-blocking versions. (But you can still only carry out one
912 operation per handle at a time - that is a limitation of the
913 communications protocol we use).
915 Finally, the child process sends asynchronous messages back to the
916 main program, such as kernel log messages. Mostly these are ignored
917 by the high-level API, but using the low-level event API you can
918 register to receive these messages.
920 =head2 SETTING CALLBACKS TO HANDLE EVENTS
922 The child process generates events in some situations. Current events
923 include: receiving a log message, the child process exits.
925 Use the C<guestfs_set_*_callback> functions to set a callback for
926 different types of events.
928 Only I<one callback of each type> can be registered for each handle.
929 Calling C<guestfs_set_*_callback> again overwrites the previous
930 callback of that type. Cancel all callbacks of this type by calling
931 this function with C<cb> set to C<NULL>.
933 =head2 guestfs_set_log_message_callback
935 typedef void (*guestfs_log_message_cb) (guestfs_h *g, void *opaque,
937 void guestfs_set_log_message_callback (guestfs_h *handle,
938 guestfs_log_message_cb cb,
941 The callback function C<cb> will be called whenever qemu or the guest
942 writes anything to the console.
944 Use this function to capture kernel messages and similar.
946 Normally there is no log message handler, and log messages are just
949 =head2 guestfs_set_subprocess_quit_callback
951 typedef void (*guestfs_subprocess_quit_cb) (guestfs_h *g, void *opaque);
952 void guestfs_set_subprocess_quit_callback (guestfs_h *handle,
953 guestfs_subprocess_quit_cb cb,
956 The callback function C<cb> will be called when the child process
957 quits, either asynchronously or if killed by
958 C<guestfs_kill_subprocess>. (This corresponds to a transition from
959 any state to the CONFIG state).
961 =head2 guestfs_set_launch_done_callback
963 typedef void (*guestfs_launch_done_cb) (guestfs_h *g, void *opaque);
964 void guestfs_set_launch_done_callback (guestfs_h *handle,
968 The callback function C<cb> will be called when the child process
969 becomes ready first time after it has been launched. (This
970 corresponds to a transition from LAUNCHING to the READY state).
972 =head1 BLOCK DEVICE NAMING
974 In the kernel there is now quite a profusion of schemata for naming
975 block devices (in this context, by I<block device> I mean a physical
976 or virtual hard drive). The original Linux IDE driver used names
977 starting with C</dev/hd*>. SCSI devices have historically used a
978 different naming scheme, C</dev/sd*>. When the Linux kernel I<libata>
979 driver became a popular replacement for the old IDE driver
980 (particularly for SATA devices) those devices also used the
981 C</dev/sd*> scheme. Additionally we now have virtual machines with
982 paravirtualized drivers. This has created several different naming
983 systems, such as C</dev/vd*> for virtio disks and C</dev/xvd*> for Xen
986 As discussed above, libguestfs uses a qemu appliance running an
987 embedded Linux kernel to access block devices. We can run a variety
988 of appliances based on a variety of Linux kernels.
990 This causes a problem for libguestfs because many API calls use device
991 or partition names. Working scripts and the recipe (example) scripts
992 that we make available over the internet could fail if the naming
995 Therefore libguestfs defines C</dev/sd*> as the I<standard naming
996 scheme>. Internally C</dev/sd*> names are translated, if necessary,
997 to other names as required. For example, under RHEL 5 which uses the
998 C</dev/hd*> scheme, any device parameter C</dev/sda2> is translated to
999 C</dev/hda2> transparently.
1001 Note that this I<only> applies to parameters. The
1002 C<guestfs_list_devices>, C<guestfs_list_partitions> and similar calls
1003 return the true names of the devices and partitions as known to the
1006 =head2 ALGORITHM FOR BLOCK DEVICE NAME TRANSLATION
1008 Usually this translation is transparent. However in some (very rare)
1009 cases you may need to know the exact algorithm. Such cases include
1010 where you use C<guestfs_config> to add a mixture of virtio and IDE
1011 devices to the qemu-based appliance, so have a mixture of C</dev/sd*>
1012 and C</dev/vd*> devices.
1014 The algorithm is applied only to I<parameters> which are known to be
1015 either device or partition names. Return values from functions such
1016 as C<guestfs_list_devices> are never changed.
1022 Is the string a parameter which is a device or partition name?
1026 Does the string begin with C</dev/sd>?
1030 Does the named device exist? If so, we use that device.
1031 However if I<not> then we continue with this algorithm.
1035 Replace initial C</dev/sd> string with C</dev/hd>.
1037 For example, change C</dev/sda2> to C</dev/hda2>.
1039 If that named device exists, use it. If not, continue.
1043 Replace initial C</dev/sd> string with C</dev/vd>.
1045 If that named device exists, use it. If not, return an error.
1049 =head2 PORTABILITY CONCERNS
1051 Although the standard naming scheme and automatic translation is
1052 useful for simple programs and guestfish scripts, for larger programs
1053 it is best not to rely on this mechanism.
1055 Where possible for maximum future portability programs using
1056 libguestfs should use these future-proof techniques:
1062 Use C<guestfs_list_devices> or C<guestfs_list_partitions> to list
1063 actual device names, and then use those names directly.
1065 Since those device names exist by definition, they will never be
1070 Use higher level ways to identify filesystems, such as LVM names,
1071 UUIDs and filesystem labels.
1077 =head2 COMMUNICATION PROTOCOL
1079 Don't rely on using this protocol directly. This section documents
1080 how it currently works, but it may change at any time.
1082 The protocol used to talk between the library and the daemon running
1083 inside the qemu virtual machine is a simple RPC mechanism built on top
1084 of XDR (RFC 1014, RFC 1832, RFC 4506).
1086 The detailed format of structures is in C<src/guestfs_protocol.x>
1087 (note: this file is automatically generated).
1089 There are two broad cases, ordinary functions that don't have any
1090 C<FileIn> and C<FileOut> parameters, which are handled with very
1091 simple request/reply messages. Then there are functions that have any
1092 C<FileIn> or C<FileOut> parameters, which use the same request and
1093 reply messages, but they may also be followed by files sent using a
1096 =head3 ORDINARY FUNCTIONS (NO FILEIN/FILEOUT PARAMS)
1098 For ordinary functions, the request message is:
1100 total length (header + arguments,
1101 but not including the length word itself)
1102 struct guestfs_message_header (encoded as XDR)
1103 struct guestfs_<foo>_args (encoded as XDR)
1105 The total length field allows the daemon to allocate a fixed size
1106 buffer into which it slurps the rest of the message. As a result, the
1107 total length is limited to C<GUESTFS_MESSAGE_MAX> bytes (currently
1108 4MB), which means the effective size of any request is limited to
1109 somewhere under this size.
1111 Note also that many functions don't take any arguments, in which case
1112 the C<guestfs_I<foo>_args> is completely omitted.
1114 The header contains the procedure number (C<guestfs_proc>) which is
1115 how the receiver knows what type of args structure to expect, or none
1118 The reply message for ordinary functions is:
1120 total length (header + ret,
1121 but not including the length word itself)
1122 struct guestfs_message_header (encoded as XDR)
1123 struct guestfs_<foo>_ret (encoded as XDR)
1125 As above the C<guestfs_I<foo>_ret> structure may be completely omitted
1126 for functions that return no formal return values.
1128 As above the total length of the reply is limited to
1129 C<GUESTFS_MESSAGE_MAX>.
1131 In the case of an error, a flag is set in the header, and the reply
1132 message is slightly changed:
1134 total length (header + error,
1135 but not including the length word itself)
1136 struct guestfs_message_header (encoded as XDR)
1137 struct guestfs_message_error (encoded as XDR)
1139 The C<guestfs_message_error> structure contains the error message as a
1142 =head3 FUNCTIONS THAT HAVE FILEIN PARAMETERS
1144 A C<FileIn> parameter indicates that we transfer a file I<into> the
1145 guest. The normal request message is sent (see above). However this
1146 is followed by a sequence of file chunks.
1148 total length (header + arguments,
1149 but not including the length word itself,
1150 and not including the chunks)
1151 struct guestfs_message_header (encoded as XDR)
1152 struct guestfs_<foo>_args (encoded as XDR)
1153 sequence of chunks for FileIn param #0
1154 sequence of chunks for FileIn param #1 etc.
1156 The "sequence of chunks" is:
1158 length of chunk (not including length word itself)
1159 struct guestfs_chunk (encoded as XDR)
1161 struct guestfs_chunk (encoded as XDR)
1164 struct guestfs_chunk (with data.data_len == 0)
1166 The final chunk has the C<data_len> field set to zero. Additionally a
1167 flag is set in the final chunk to indicate either successful
1168 completion or early cancellation.
1170 At time of writing there are no functions that have more than one
1171 FileIn parameter. However this is (theoretically) supported, by
1172 sending the sequence of chunks for each FileIn parameter one after
1173 another (from left to right).
1175 Both the library (sender) I<and> the daemon (receiver) may cancel the
1176 transfer. The library does this by sending a chunk with a special
1177 flag set to indicate cancellation. When the daemon sees this, it
1178 cancels the whole RPC, does I<not> send any reply, and goes back to
1179 reading the next request.
1181 The daemon may also cancel. It does this by writing a special word
1182 C<GUESTFS_CANCEL_FLAG> to the socket. The library listens for this
1183 during the transfer, and if it gets it, it will cancel the transfer
1184 (it sends a cancel chunk). The special word is chosen so that even if
1185 cancellation happens right at the end of the transfer (after the
1186 library has finished writing and has started listening for the reply),
1187 the "spurious" cancel flag will not be confused with the reply
1190 This protocol allows the transfer of arbitrary sized files (no 32 bit
1191 limit), and also files where the size is not known in advance
1192 (eg. from pipes or sockets). However the chunks are rather small
1193 (C<GUESTFS_MAX_CHUNK_SIZE>), so that neither the library nor the
1194 daemon need to keep much in memory.
1196 =head3 FUNCTIONS THAT HAVE FILEOUT PARAMETERS
1198 The protocol for FileOut parameters is exactly the same as for FileIn
1199 parameters, but with the roles of daemon and library reversed.
1201 total length (header + ret,
1202 but not including the length word itself,
1203 and not including the chunks)
1204 struct guestfs_message_header (encoded as XDR)
1205 struct guestfs_<foo>_ret (encoded as XDR)
1206 sequence of chunks for FileOut param #0
1207 sequence of chunks for FileOut param #1 etc.
1209 =head3 INITIAL MESSAGE
1211 Because the underlying channel (QEmu -net channel) doesn't have any
1212 sort of connection control, when the daemon launches it sends an
1213 initial word (C<GUESTFS_LAUNCH_FLAG>) which indicates that the guest
1214 and daemon is alive. This is what C<guestfs_launch> waits for.
1216 =head1 MULTIPLE HANDLES AND MULTIPLE THREADS
1218 All high-level libguestfs actions are synchronous. If you want
1219 to use libguestfs asynchronously then you must create a thread.
1221 Only use the handle from a single thread. Either use the handle
1222 exclusively from one thread, or provide your own mutex so that two
1223 threads cannot issue calls on the same handle at the same time.
1225 =head1 QEMU WRAPPERS
1227 If you want to compile your own qemu, run qemu from a non-standard
1228 location, or pass extra arguments to qemu, then you can write a
1229 shell-script wrapper around qemu.
1231 There is one important rule to remember: you I<must C<exec qemu>> as
1232 the last command in the shell script (so that qemu replaces the shell
1233 and becomes the direct child of the libguestfs-using program). If you
1234 don't do this, then the qemu process won't be cleaned up correctly.
1236 Here is an example of a wrapper, where I have built my own copy of
1240 qemudir=/home/rjones/d/qemu
1241 exec $qemudir/x86_64-softmmu/qemu-system-x86_64 -L $qemudir/pc-bios "$@"
1243 Save this script as C</tmp/qemu.wrapper> (or wherever), C<chmod +x>,
1244 and then use it by setting the LIBGUESTFS_QEMU environment variable.
1247 LIBGUESTFS_QEMU=/tmp/qemu.wrapper guestfish
1249 Note that libguestfs also calls qemu with the -help and -version
1250 options in order to determine features.
1252 =head1 ENVIRONMENT VARIABLES
1256 =item LIBGUESTFS_APPEND
1258 Pass additional options to the guest kernel.
1260 =item LIBGUESTFS_DEBUG
1262 Set C<LIBGUESTFS_DEBUG=1> to enable verbose messages. This
1263 has the same effect as calling C<guestfs_set_verbose (handle, 1)>.
1265 =item LIBGUESTFS_MEMSIZE
1267 Set the memory allocated to the qemu process, in megabytes. For
1270 LIBGUESTFS_MEMSIZE=700
1272 =item LIBGUESTFS_PATH
1274 Set the path that libguestfs uses to search for kernel and initrd.img.
1275 See the discussion of paths in section PATH above.
1277 =item LIBGUESTFS_QEMU
1279 Set the default qemu binary that libguestfs uses. If not set, then
1280 the qemu which was found at compile time by the configure script is
1283 See also L</QEMU WRAPPERS> above.
1285 =item LIBGUESTFS_TRACE
1287 Set C<LIBGUESTFS_TRACE=1> to enable command traces. This
1288 has the same effect as calling C<guestfs_set_trace (handle, 1)>.
1292 Location of temporary directory, defaults to C</tmp>.
1294 If libguestfs was compiled to use the supermin appliance then each
1295 handle will require rather a large amount of space in this directory
1296 for short periods of time (~ 80 MB). You can use C<$TMPDIR> to
1297 configure another directory to use in case C</tmp> is not large
1307 L<http://libguestfs.org/>.
1309 Tools with a similar purpose:
1318 To get a list of bugs against libguestfs use this link:
1320 L<https://bugzilla.redhat.com/buglist.cgi?component=libguestfs&product=Virtualization+Tools>
1322 To report a new bug against libguestfs use this link:
1324 L<https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virtualization+Tools>
1326 When reporting a bug, please check:
1332 That the bug hasn't been reported already.
1336 That you are testing a recent version.
1340 Describe the bug accurately, and give a way to reproduce it.
1344 Run libguestfs-test-tool and paste the B<complete, unedited>
1345 output into the bug report.
1351 Richard W.M. Jones (C<rjones at redhat dot com>)
1355 Copyright (C) 2009 Red Hat Inc.
1356 L<http://libguestfs.org/>
1358 This library is free software; you can redistribute it and/or
1359 modify it under the terms of the GNU Lesser General Public
1360 License as published by the Free Software Foundation; either
1361 version 2 of the License, or (at your option) any later version.
1363 This library is distributed in the hope that it will be useful,
1364 but WITHOUT ANY WARRANTY; without even the implied warranty of
1365 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
1366 Lesser General Public License for more details.
1368 You should have received a copy of the GNU Lesser General Public
1369 License along with this library; if not, write to the Free Software
1370 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA