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);
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, 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.
56 This section provides a gentler overview of the libguestfs API. We
57 also try to group API calls together, where that may not be obvious
58 from reading about the individual calls below.
62 Before you can use libguestfs calls, you have to create a handle.
63 Then you must add at least one disk image to the handle, followed by
64 launching the handle, then performing whatever operations you want,
65 and finally closing the handle. So the general structure of all
66 libguestfs-using programs looks like this:
68 guestfs_h *handle = guestfs_create ();
70 /* Call guestfs_add_drive additional times if there are
71 * multiple disk images.
73 guestfs_add_drive (handle, "guest.img");
75 /* Most manipulation calls won't work until you've launched
76 * the handle. You have to do this _after_ adding drives
77 * and _before_ other commands.
79 guestfs_launch (handle);
81 /* Now you can examine what partitions, LVs etc are available.
83 char **partitions = guestfs_list_partitions (handle);
84 char **logvols = guestfs_lvs (handle);
86 /* To access a filesystem in the image, you must mount it.
88 guestfs_mount (handle, "/dev/sda1", "/");
90 /* Now you can perform filesystem actions on the guest
93 guestfs_touch (handle, "/hello");
95 /* You only need to call guestfs_sync if you have made
96 * changes to the guest image.
98 guestfs_sync (handle);
100 /* Close the handle. */
101 guestfs_close (handle);
103 The code above doesn't include any error checking. In real code you
104 should check return values carefully for errors. In general all
105 functions that return integers return C<-1> on error, and all
106 functions that return pointers return C<NULL> on error. See section
107 L</ERROR HANDLING> below for how to handle errors, and consult the
108 documentation for each function call below to see precisely how they
109 return error indications.
113 The image filename (C<"guest.img"> in the example above) could be a
114 disk image from a virtual machine, a L<dd(1)> copy of a physical hard
115 disk, an actual block device, or simply an empty file of zeroes that
116 you have created through L<posix_fallocate(3)>. Libguestfs lets you
117 do useful things to all of these.
119 You can add a disk read-only using C<guestfs_add_drive_ro>, in which
120 case libguestfs won't modify the file.
122 Be extremely cautious if the disk image is in use, eg. if it is being
123 used by a virtual machine. Adding it read-write will almost certainly
124 cause disk corruption, but adding it read-only is safe.
126 You must add at least one disk image, and you may add multiple disk
127 images. In the API, the disk images are usually referred to as
128 C</dev/sda> (for the first one you added), C</dev/sdb> (for the second
131 Once C<guestfs_launch> has been called you cannot add any more images.
132 You can call C<guestfs_list_devices> to get a list of the device
133 names, in the order that you added them. See also L</BLOCK DEVICE
138 Before you can read or write files, create directories and so on in a
139 disk image that contains filesystems, you have to mount those
140 filesystems using C<guestfs_mount>. If you already know that a disk
141 image contains (for example) one partition with a filesystem on that
142 partition, then you can mount it directly:
144 guestfs_mount (handle, "/dev/sda1", "/");
146 where C</dev/sda1> means literally the first partition (C<1>) of the
147 first disk image that we added (C</dev/sda>). If the disk contains
148 Linux LVM2 logical volumes you could refer to those instead (eg. C</dev/VG/LV>).
150 If you are given a disk image and you don't know what it contains then
151 you have to find out. Libguestfs can do that too: use
152 C<guestfs_list_partitions> and C<guestfs_lvs> to list possible
153 partitions and LVs, and either try mounting each to see what is
154 mountable, or else examine them with C<guestfs_file>. But you might
155 find it easier to look at higher level programs built on top of
156 libguestfs, in particular L<virt-inspector(1)>.
158 To mount a disk image read-only, use C<guestfs_mount_ro>. There are
159 several other variations of the C<guestfs_mount_*> call.
161 =head2 FILESYSTEM ACCESS AND MODIFICATION
163 The majority of the libguestfs API consists of fairly low-level calls
164 for accessing and modifying the files, directories, symlinks etc on
165 mounted filesystems. There are over a hundred such calls which you
166 can find listed in detail below in this man page, and we don't even
167 pretend to cover them all in this overview.
169 Specify filenames as full paths including the mount point.
171 For example, if you mounted a filesystem at C<"/"> and you want to
172 read the file called C<"etc/passwd"> then you could do:
174 char *data = guestfs_cat (handle, "/etc/passwd");
176 This would return C<data> as a newly allocated buffer containing the
177 full content of that file (with some conditions: see also
178 L</DOWNLOADING> below), or C<NULL> if there was an error.
180 As another example, to create a top-level directory on that filesystem
181 called C<"var"> you would do:
183 guestfs_mkdir (handle, "/var");
185 To create a symlink you could do:
187 guestfs_ln_s (handle, "/etc/init.d/portmap",
188 "/etc/rc3.d/S30portmap");
190 Libguestfs will reject attempts to use relative paths. There is no
191 concept of a current working directory. Libguestfs can return errors
192 in many situations: for example if the filesystem isn't writable, or
193 if a file or directory that you requested doesn't exist. If you are
194 using the C API (documented here) you have to check for those error
195 conditions after each call. (Other language bindings turn these
196 errors into exceptions).
198 File writes are affected by the per-handle umask, set by calling
199 C<guestfs_umask> and defaulting to 022.
203 Libguestfs contains API calls to read, create and modify partition
204 tables on disk images.
206 In the common case where you want to create a single partition
207 covering the whole disk, you should use the C<guestfs_part_disk>
210 const char *parttype = "mbr";
211 if (disk_is_larger_than_2TB)
213 guestfs_part_disk (g, "/dev/sda", parttype);
215 Obviously this effectively wipes anything that was on that disk image
218 In general MBR partitions are both unnecessarily complicated and
219 depend on archaic details, namely the Cylinder-Head-Sector (CHS)
220 geometry of the disk. C<guestfs_sfdiskM> can be used to
221 create more complex arrangements where the relative sizes are
222 expressed in megabytes instead of cylinders, which is a small win.
223 C<guestfs_sfdiskM> will choose the nearest cylinder to approximate the
224 requested size. There's a lot of crazy stuff to do with IDE and
225 virtio disks having different, incompatible CHS geometries, that you
226 probably don't want to know about.
228 My advice: make a single partition to cover the whole disk, then use
233 Libguestfs provides access to a large part of the LVM2 API, such as
234 C<guestfs_lvcreate> and C<guestfs_vgremove>. It won't make much sense
235 unless you familiarize yourself with the concepts of physical volumes,
236 volume groups and logical volumes.
238 This author strongly recommends reading the LVM HOWTO, online at
239 L<http://tldp.org/HOWTO/LVM-HOWTO/>.
243 Use C<guestfs_cat> to download small, text only files. This call
244 is limited to files which are less than 2 MB and which cannot contain
245 any ASCII NUL (C<\0>) characters. However it has a very simple
248 C<guestfs_read_file> can be used to read files which contain
249 arbitrary 8 bit data, since it returns a (pointer, size) pair.
250 However it is still limited to "small" files, less than 2 MB.
252 C<guestfs_download> can be used to download any file, with no
253 limits on content or size (even files larger than 4 GB).
255 To download multiple files, see C<guestfs_tar_out> and
260 It's often the case that you want to write a file or files to the disk
263 For small, single files, use C<guestfs_write_file>. This call
264 currently contains a bug which limits the call to plain text files
265 (not containing ASCII NUL characters).
267 To upload a single file, use C<guestfs_upload>. This call has no
268 limits on file content or size (even files larger than 4 GB).
270 To upload multiple files, see C<guestfs_tar_in> and C<guestfs_tgz_in>.
272 However the fastest way to upload I<large numbers of arbitrary files>
273 is to turn them into a squashfs or CD ISO (see L<mksquashfs(8)> and
274 L<mkisofs(8)>), then attach this using C<guestfs_add_drive_ro>. If
275 you add the drive in a predictable way (eg. adding it last after all
276 other drives) then you can get the device name from
277 C<guestfs_list_devices> and mount it directly using
278 C<guestfs_mount_ro>. Note that squashfs images are sometimes
279 non-portable between kernel versions, and they don't support labels or
280 UUIDs. If you want to pre-build an image or you need to mount it
281 using a label or UUID, use an ISO image instead.
285 There are various different commands for copying between files and
286 devices and in and out of the guest filesystem. These are summarised
291 =item B<file> to B<file>
293 Use L</guestfs_cp> to copy a single file, or
294 L</guestfs_cp_a> to copy directories recursively.
296 =item B<file or device> to B<file or device>
298 Use L</guestfs_dd> which efficiently uses L<dd(1)>
299 to copy between files and devices in the guest.
301 Example: duplicate the contents of an LV:
303 guestfs_dd (g, "/dev/VG/Original", "/dev/VG/Copy");
305 The destination (C</dev/VG/Copy>) must be at least as large as the
306 source (C</dev/VG/Original>).
308 =item B<file on the host> to B<file or device>
310 Use L</guestfs_upload>. See L</UPLOADING> above.
312 =item B<file or device> to B<file on the host>
314 Use L</guestfs_download>. See L</DOWNLOADING> above.
320 C<guestfs_ll> is just designed for humans to read (mainly when using
321 the L<guestfish(1)>-equivalent command C<ll>).
323 C<guestfs_ls> is a quick way to get a list of files in a directory
324 from programs, as a flat list of strings.
326 C<guestfs_readdir> is a programmatic way to get a list of files in a
327 directory, plus additional information about each one. It is more
328 equivalent to using the L<readdir(3)> call on a local filesystem.
330 C<guestfs_find> can be used to recursively list files.
332 =head2 RUNNING COMMANDS
334 Although libguestfs is a primarily an API for manipulating files
335 inside guest images, we also provide some limited facilities for
336 running commands inside guests.
338 There are many limitations to this:
344 The kernel version that the command runs under will be different
345 from what it expects.
349 If the command needs to communicate with daemons, then most likely
350 they won't be running.
354 The command will be running in limited memory.
358 Only supports Linux guests (not Windows, BSD, etc).
362 Architecture limitations (eg. won't work for a PPC guest on
367 For SELinux guests, you may need to enable SELinux and load policy
368 first. See L</SELINUX> in this manpage.
372 The two main API calls to run commands are C<guestfs_command> and
373 C<guestfs_sh> (there are also variations).
375 The difference is that C<guestfs_sh> runs commands using the shell, so
376 any shell globs, redirections, etc will work.
378 =head2 CONFIGURATION FILES
380 To read and write configuration files in Linux guest filesystems, we
381 strongly recommend using Augeas. For example, Augeas understands how
382 to read and write, say, a Linux shadow password file or X.org
383 configuration file, and so avoids you having to write that code.
385 The main Augeas calls are bound through the C<guestfs_aug_*> APIs. We
386 don't document Augeas itself here because there is excellent
387 documentation on the L<http://augeas.net/> website.
389 If you don't want to use Augeas (you fool!) then try calling
390 C<guestfs_read_lines> to get the file as a list of lines which
391 you can iterate over.
395 We support SELinux guests. To ensure that labeling happens correctly
396 in SELinux guests, you need to enable SELinux and load the guest's
403 Before launching, do:
405 guestfs_set_selinux (g, 1);
409 After mounting the guest's filesystem(s), load the policy. This
410 is best done by running the L<load_policy(8)> command in the
413 guestfs_sh (g, "/usr/sbin/load_policy");
415 (Older versions of C<load_policy> require you to specify the
416 name of the policy file).
420 Optionally, set the security context for the API. The correct
421 security context to use can only be known by inspecting the
422 guest. As an example:
424 guestfs_setcon (g, "unconfined_u:unconfined_r:unconfined_t:s0");
428 This will work for running commands and editing existing files.
430 When new files are created, you may need to label them explicitly,
431 for example by running the external command
432 C<restorecon pathname>.
434 =head2 SPECIAL CONSIDERATIONS FOR WINDOWS GUESTS
436 Libguestfs can mount NTFS partitions. It does this using the
437 L<http://www.ntfs-3g.org/> driver.
439 DOS and Windows still use drive letters, and the filesystems are
440 always treated as case insensitive by Windows itself, and therefore
441 you might find a Windows configuration file referring to a path like
442 C<c:\windows\system32>. When the filesystem is mounted in libguestfs,
443 that directory might be referred to as C</WINDOWS/System32>.
445 Drive letter mappings are outside the scope of libguestfs. You have
446 to use libguestfs to read the appropriate Windows Registry and
447 configuration files, to determine yourself how drives are mapped (see
448 also L<virt-inspector(1)>).
450 Replacing backslash characters with forward slash characters is also
451 outside the scope of libguestfs, but something that you can easily do.
453 Where we can help is in resolving the case insensitivity of paths.
454 For this, call C<guestfs_case_sensitive_path>.
456 Libguestfs also provides some help for decoding Windows Registry
457 "hive" files, through the library C<hivex> which is part of the
458 libguestfs project. You have to locate and download the hive file(s)
459 yourself, and then pass them to C<hivex> functions. See also the
460 programs L<hivexml(1)>, L<hivexsh(1)> and L<virt-win-reg(1)> for more
463 =head2 USING LIBGUESTFS WITH OTHER PROGRAMMING LANGUAGES
465 Although we don't want to discourage you from using the C API, we will
466 mention here that the same API is also available in other languages.
468 The API is broadly identical in all supported languages. This means
469 that the C call C<guestfs_mount(handle,path)> is
470 C<$handle-E<gt>mount($path)> in Perl, C<handle.mount(path)> in Python,
471 and C<Guestfs.mount handle path> in OCaml. In other words, a
472 straightforward, predictable isomorphism between each language.
474 Error messages are automatically transformed
475 into exceptions if the language supports it.
477 We don't try to "object orientify" parts of the API in OO languages,
478 although contributors are welcome to write higher level APIs above
479 what we provide in their favourite languages if they wish.
485 You can use the I<guestfs.h> header file from C++ programs. The C++
486 API is identical to the C API. C++ classes and exceptions are
491 The C# bindings are highly experimental. Please read the warnings
492 at the top of C<csharp/Libguestfs.cs>.
496 This is the only language binding that working but incomplete. Only
497 calls which return simple integers have been bound in Haskell, and we
498 are looking for help to complete this binding.
502 Full documentation is contained in the Javadoc which is distributed
507 For documentation see the file C<guestfs.mli>.
511 For documentation see L<Sys::Guestfs(3)>.
515 For documentation do:
523 Use the Guestfs module. There is no Ruby-specific documentation, but
524 you can find examples written in Ruby in the libguestfs source.
526 =item B<shell scripts>
528 For documentation see L<guestfish(1)>.
532 =head2 LIBGUESTFS GOTCHAS
534 L<http://en.wikipedia.org/wiki/Gotcha_(programming)>: "A feature of a
535 system [...] that works in the way it is documented but is
536 counterintuitive and almost invites mistakes."
538 Since we developed libguestfs and the associated tools, there are
539 several things we would have designed differently, but are now stuck
540 with for backwards compatibility or other reasons. If there is ever a
541 libguestfs 2.0 release, you can expect these to change. Beware of
546 =item Autosync / forgetting to sync.
548 When modifying a filesystem from C or another language, you B<must>
549 unmount all filesystems and call L</guestfs_sync> explicitly before
550 you close the libguestfs handle. You can also call:
552 guestfs_set_autosync (handle, 1);
554 to have the unmount/sync done automatically for you when the handle is
555 closed. (This feature is called "autosync", L</guestfs_set_autosync>
558 If you forget to do this, then it is entirely possible that your
559 changes won't be written out, or will be partially written, or (very
560 rarely) that you'll get disk corruption.
562 Note that in L<guestfish(3)> I<autosync is the default>. So quick and
563 dirty guestfish scripts that forget to sync will work just fine, which
564 can make this extra-puzzling if you are trying to debug a problem.
566 =item Mount option C<-o sync> should not be the default.
568 If you use C<guestfs_mount>, then C<-o sync,noatime> are added
569 implicitly. However C<-o sync> does not add any reliability benefit,
570 but does have a very large performance impact.
572 The work around is to use C<guestfs_mount_options> and set the mount
573 options that you actually want to use.
575 =item Read-only should be the default.
577 In L<guestfish(3)>, I<--ro> should be the default, and you should
578 have to specify I<--rw> if you want to make changes to the image.
580 This would reduce the potential to corrupt live VM images.
582 Note that many filesystems change the disk when you just mount and
583 unmount, even if you didn't perform any writes. You need to use
584 C<guestfs_add_drive_ro> to guarantee that the disk is not changed.
586 =item guestfish command line is hard to use.
588 C<guestfish disk.img> doesn't do what people expect (open C<disk.img>
589 for examination). It tries to run a guestfish command C<disk.img>
590 which doesn't exist, so it fails, and it fails with a strange and
591 unintuitive error message. Like the Bourne shell, we should have used
592 C<guestfish -c command> to run commands.
596 =head2 PROTOCOL LIMITS
598 Internally libguestfs uses a message-based protocol to pass API calls
599 and their responses to and from a small "appliance" (see L</INTERNALS>
600 for plenty more detail about this). The maximum message size used by
601 the protocol is slightly less than 4 MB. For some API calls you may
602 need to be aware of this limit. The API calls which may be affected
603 are individually documented, with a link back to this section of the
606 A simple call such as C<guestfs_cat> returns its result (the file
607 data) in a simple string. Because this string is at some point
608 internally encoded as a message, the maximum size that it can return
609 is slightly under 4 MB. If the requested file is larger than this
610 then you will get an error.
612 In order to transfer large files into and out of the guest filesystem,
613 you need to use particular calls that support this. The sections
614 L</UPLOADING> and L</DOWNLOADING> document how to do this.
616 You might also consider mounting the disk image using our FUSE
617 filesystem support (L<guestmount(1)>).
619 =head1 CONNECTION MANAGEMENT
623 C<guestfs_h> is the opaque type representing a connection handle.
624 Create a handle by calling C<guestfs_create>. Call C<guestfs_close>
625 to free the handle and release all resources used.
627 For information on using multiple handles and threads, see the section
628 L</MULTIPLE HANDLES AND MULTIPLE THREADS> below.
630 =head2 guestfs_create
632 guestfs_h *guestfs_create (void);
634 Create a connection handle.
636 You have to call C<guestfs_add_drive> on the handle at least once.
638 This function returns a non-NULL pointer to a handle on success or
641 After configuring the handle, you have to call C<guestfs_launch>.
643 You may also want to configure error handling for the handle. See
644 L</ERROR HANDLING> section below.
648 void guestfs_close (guestfs_h *handle);
650 This closes the connection handle and frees up all resources used.
652 =head1 ERROR HANDLING
654 The convention in all functions that return C<int> is that they return
655 C<-1> to indicate an error. You can get additional information on
656 errors by calling C<guestfs_last_error> and/or by setting up an error
657 handler with C<guestfs_set_error_handler>.
659 The default error handler prints the information string to C<stderr>.
661 Out of memory errors are handled differently. The default action is
662 to call L<abort(3)>. If this is undesirable, then you can set a
663 handler using C<guestfs_set_out_of_memory_handler>.
665 =head2 guestfs_last_error
667 const char *guestfs_last_error (guestfs_h *handle);
669 This returns the last error message that happened on C<handle>. If
670 there has not been an error since the handle was created, then this
673 The lifetime of the returned string is until the next error occurs, or
674 C<guestfs_close> is called.
676 The error string is not localized (ie. is always in English), because
677 this makes searching for error messages in search engines give the
678 largest number of results.
680 =head2 guestfs_set_error_handler
682 typedef void (*guestfs_error_handler_cb) (guestfs_h *handle,
685 void guestfs_set_error_handler (guestfs_h *handle,
686 guestfs_error_handler_cb cb,
689 The callback C<cb> will be called if there is an error. The
690 parameters passed to the callback are an opaque data pointer and the
691 error message string.
693 Note that the message string C<msg> is freed as soon as the callback
694 function returns, so if you want to stash it somewhere you must make
697 The default handler prints messages on C<stderr>.
699 If you set C<cb> to C<NULL> then I<no> handler is called.
701 =head2 guestfs_get_error_handler
703 guestfs_error_handler_cb guestfs_get_error_handler (guestfs_h *handle,
706 Returns the current error handler callback.
708 =head2 guestfs_set_out_of_memory_handler
710 typedef void (*guestfs_abort_cb) (void);
711 int guestfs_set_out_of_memory_handler (guestfs_h *handle,
714 The callback C<cb> will be called if there is an out of memory
715 situation. I<Note this callback must not return>.
717 The default is to call L<abort(3)>.
719 You cannot set C<cb> to C<NULL>. You can't ignore out of memory
722 =head2 guestfs_get_out_of_memory_handler
724 guestfs_abort_fn guestfs_get_out_of_memory_handler (guestfs_h *handle);
726 This returns the current out of memory handler.
730 Libguestfs needs a kernel and initrd.img, which it finds by looking
731 along an internal path.
733 By default it looks for these in the directory C<$libdir/guestfs>
734 (eg. C</usr/local/lib/guestfs> or C</usr/lib64/guestfs>).
736 Use C<guestfs_set_path> or set the environment variable
737 C<LIBGUESTFS_PATH> to change the directories that libguestfs will
738 search in. The value is a colon-separated list of paths. The current
739 directory is I<not> searched unless the path contains an empty element
740 or C<.>. For example C<LIBGUESTFS_PATH=:/usr/lib/guestfs> would
741 search the current directory and then C</usr/lib/guestfs>.
743 =head1 HIGH-LEVEL API ACTIONS
747 We guarantee the libguestfs ABI (binary interface), for public,
748 high-level actions as outlined in this section. Although we will
749 deprecate some actions, for example if they get replaced by newer
750 calls, we will keep the old actions forever. This allows you the
751 developer to program in confidence against libguestfs.
761 =head2 GROUPS OF FUNCTIONALITY IN THE APPLIANCE
763 Using L</guestfs_available> you can test availability of
764 the following groups of functions. This test queries the
765 appliance to see if the appliance you are currently using
766 supports the functionality.
770 =head2 SINGLE CALLS AT COMPILE TIME
772 If you need to test whether a single libguestfs function is
773 available at compile time, we recommend using build tools
774 such as autoconf or cmake. For example in autotools you could
777 AC_CHECK_LIB([guestfs],[guestfs_create])
778 AC_CHECK_FUNCS([guestfs_dd])
780 which would result in C<HAVE_GUESTFS_DD> being either defined
781 or not defined in your program.
783 =head2 SINGLE CALLS AT RUN TIME
785 Testing at compile time doesn't guarantee that a function really
786 exists in the library. The reason is that you might be dynamically
787 linked against a previous I<libguestfs.so> (dynamic library)
788 which doesn't have the call. This situation unfortunately results
789 in a segmentation fault, which is a shortcoming of the C dynamic
790 linking system itself.
792 You can use L<dlopen(3)> to test if a function is available
793 at run time, as in this example program (note that you still
794 need the compile time check as well):
806 #ifdef HAVE_GUESTFS_DD
810 /* Test if the function guestfs_dd is really available. */
811 dl = dlopen (NULL, RTLD_LAZY);
813 fprintf (stderr, "dlopen: %s\n", dlerror ());
816 has_function = dlsym (dl, "guestfs_dd") != NULL;
820 printf ("this libguestfs.so does NOT have guestfs_dd function\n");
822 printf ("this libguestfs.so has guestfs_dd function\n");
823 /* Now it's safe to call
824 guestfs_dd (g, "foo", "bar");
828 printf ("guestfs_dd function was not found at compile time\n");
832 You may think the above is an awful lot of hassle, and it is.
833 There are other ways outside of the C linking system to ensure
834 that this kind of incompatibility never arises, such as using
837 Requires: libguestfs >= 1.0.80
841 <!-- old anchor for the next section -->
842 <a name="state_machine_and_low_level_event_api"/>
848 Internally, libguestfs is implemented by running an appliance (a
849 special type of small virtual machine) using L<qemu(1)>. Qemu runs as
850 a child process of the main program.
856 | | child process / appliance
857 | | __________________________
859 +-------------------+ RPC | +-----------------+ |
860 | libguestfs <--------------------> guestfsd | |
861 | | | +-----------------+ |
862 \___________________/ | | Linux kernel | |
863 | +--^--------------+ |
864 \_________|________________/
872 The library, linked to the main program, creates the child process and
873 hence the appliance in the L</guestfs_launch> function.
875 Inside the appliance is a Linux kernel and a complete stack of
876 userspace tools (such as LVM and ext2 programs) and a small
877 controlling daemon called C<guestfsd>. The library talks to
878 C<guestfsd> using remote procedure calls (RPC). There is a mostly
879 one-to-one correspondence between libguestfs API calls and RPC calls
880 to the daemon. Lastly the disk image(s) are attached to the qemu
881 process which translates device access by the appliance's Linux kernel
882 into accesses to the image.
884 A common misunderstanding is that the appliance "is" the virtual
885 machine. Although the disk image you are attached to might also be
886 used by some virtual machine, libguestfs doesn't know or care about
887 this. (But you will care if both libguestfs's qemu process and your
888 virtual machine are trying to update the disk image at the same time,
889 since these usually results in massive disk corruption).
893 libguestfs uses a state machine to model the child process:
904 / | \ \ guestfs_launch
915 \______/ <------ \________/
917 The normal transitions are (1) CONFIG (when the handle is created, but
918 there is no child process), (2) LAUNCHING (when the child process is
919 booting up), (3) alternating between READY and BUSY as commands are
920 issued to, and carried out by, the child process.
922 The guest may be killed by C<guestfs_kill_subprocess>, or may die
923 asynchronously at any time (eg. due to some internal error), and that
924 causes the state to transition back to CONFIG.
926 Configuration commands for qemu such as C<guestfs_add_drive> can only
927 be issued when in the CONFIG state.
929 The high-level API offers two calls that go from CONFIG through
930 LAUNCHING to READY. C<guestfs_launch> blocks until the child process
931 is READY to accept commands (or until some failure or timeout).
932 C<guestfs_launch> internally moves the state from CONFIG to LAUNCHING
935 High-level API actions such as C<guestfs_mount> can only be issued
936 when in the READY state. These high-level API calls block waiting for
937 the command to be carried out (ie. the state to transition to BUSY and
938 then back to READY). But using the low-level event API, you get
939 non-blocking versions. (But you can still only carry out one
940 operation per handle at a time - that is a limitation of the
941 communications protocol we use).
943 Finally, the child process sends asynchronous messages back to the
944 main program, such as kernel log messages. Mostly these are ignored
945 by the high-level API, but using the low-level event API you can
946 register to receive these messages.
948 =head2 SETTING CALLBACKS TO HANDLE EVENTS
950 The child process generates events in some situations. Current events
951 include: receiving a log message, the child process exits.
953 Use the C<guestfs_set_*_callback> functions to set a callback for
954 different types of events.
956 Only I<one callback of each type> can be registered for each handle.
957 Calling C<guestfs_set_*_callback> again overwrites the previous
958 callback of that type. Cancel all callbacks of this type by calling
959 this function with C<cb> set to C<NULL>.
961 =head2 guestfs_set_log_message_callback
963 typedef void (*guestfs_log_message_cb) (guestfs_h *g, void *opaque,
965 void guestfs_set_log_message_callback (guestfs_h *handle,
966 guestfs_log_message_cb cb,
969 The callback function C<cb> will be called whenever qemu or the guest
970 writes anything to the console.
972 Use this function to capture kernel messages and similar.
974 Normally there is no log message handler, and log messages are just
977 =head2 guestfs_set_subprocess_quit_callback
979 typedef void (*guestfs_subprocess_quit_cb) (guestfs_h *g, void *opaque);
980 void guestfs_set_subprocess_quit_callback (guestfs_h *handle,
981 guestfs_subprocess_quit_cb cb,
984 The callback function C<cb> will be called when the child process
985 quits, either asynchronously or if killed by
986 C<guestfs_kill_subprocess>. (This corresponds to a transition from
987 any state to the CONFIG state).
989 =head2 guestfs_set_launch_done_callback
991 typedef void (*guestfs_launch_done_cb) (guestfs_h *g, void *opaque);
992 void guestfs_set_launch_done_callback (guestfs_h *handle,
996 The callback function C<cb> will be called when the child process
997 becomes ready first time after it has been launched. (This
998 corresponds to a transition from LAUNCHING to the READY state).
1000 =head1 BLOCK DEVICE NAMING
1002 In the kernel there is now quite a profusion of schemata for naming
1003 block devices (in this context, by I<block device> I mean a physical
1004 or virtual hard drive). The original Linux IDE driver used names
1005 starting with C</dev/hd*>. SCSI devices have historically used a
1006 different naming scheme, C</dev/sd*>. When the Linux kernel I<libata>
1007 driver became a popular replacement for the old IDE driver
1008 (particularly for SATA devices) those devices also used the
1009 C</dev/sd*> scheme. Additionally we now have virtual machines with
1010 paravirtualized drivers. This has created several different naming
1011 systems, such as C</dev/vd*> for virtio disks and C</dev/xvd*> for Xen
1014 As discussed above, libguestfs uses a qemu appliance running an
1015 embedded Linux kernel to access block devices. We can run a variety
1016 of appliances based on a variety of Linux kernels.
1018 This causes a problem for libguestfs because many API calls use device
1019 or partition names. Working scripts and the recipe (example) scripts
1020 that we make available over the internet could fail if the naming
1023 Therefore libguestfs defines C</dev/sd*> as the I<standard naming
1024 scheme>. Internally C</dev/sd*> names are translated, if necessary,
1025 to other names as required. For example, under RHEL 5 which uses the
1026 C</dev/hd*> scheme, any device parameter C</dev/sda2> is translated to
1027 C</dev/hda2> transparently.
1029 Note that this I<only> applies to parameters. The
1030 C<guestfs_list_devices>, C<guestfs_list_partitions> and similar calls
1031 return the true names of the devices and partitions as known to the
1034 =head2 ALGORITHM FOR BLOCK DEVICE NAME TRANSLATION
1036 Usually this translation is transparent. However in some (very rare)
1037 cases you may need to know the exact algorithm. Such cases include
1038 where you use C<guestfs_config> to add a mixture of virtio and IDE
1039 devices to the qemu-based appliance, so have a mixture of C</dev/sd*>
1040 and C</dev/vd*> devices.
1042 The algorithm is applied only to I<parameters> which are known to be
1043 either device or partition names. Return values from functions such
1044 as C<guestfs_list_devices> are never changed.
1050 Is the string a parameter which is a device or partition name?
1054 Does the string begin with C</dev/sd>?
1058 Does the named device exist? If so, we use that device.
1059 However if I<not> then we continue with this algorithm.
1063 Replace initial C</dev/sd> string with C</dev/hd>.
1065 For example, change C</dev/sda2> to C</dev/hda2>.
1067 If that named device exists, use it. If not, continue.
1071 Replace initial C</dev/sd> string with C</dev/vd>.
1073 If that named device exists, use it. If not, return an error.
1077 =head2 PORTABILITY CONCERNS
1079 Although the standard naming scheme and automatic translation is
1080 useful for simple programs and guestfish scripts, for larger programs
1081 it is best not to rely on this mechanism.
1083 Where possible for maximum future portability programs using
1084 libguestfs should use these future-proof techniques:
1090 Use C<guestfs_list_devices> or C<guestfs_list_partitions> to list
1091 actual device names, and then use those names directly.
1093 Since those device names exist by definition, they will never be
1098 Use higher level ways to identify filesystems, such as LVM names,
1099 UUIDs and filesystem labels.
1105 =head2 COMMUNICATION PROTOCOL
1107 Don't rely on using this protocol directly. This section documents
1108 how it currently works, but it may change at any time.
1110 The protocol used to talk between the library and the daemon running
1111 inside the qemu virtual machine is a simple RPC mechanism built on top
1112 of XDR (RFC 1014, RFC 1832, RFC 4506).
1114 The detailed format of structures is in C<src/guestfs_protocol.x>
1115 (note: this file is automatically generated).
1117 There are two broad cases, ordinary functions that don't have any
1118 C<FileIn> and C<FileOut> parameters, which are handled with very
1119 simple request/reply messages. Then there are functions that have any
1120 C<FileIn> or C<FileOut> parameters, which use the same request and
1121 reply messages, but they may also be followed by files sent using a
1124 =head3 ORDINARY FUNCTIONS (NO FILEIN/FILEOUT PARAMS)
1126 For ordinary functions, the request message is:
1128 total length (header + arguments,
1129 but not including the length word itself)
1130 struct guestfs_message_header (encoded as XDR)
1131 struct guestfs_<foo>_args (encoded as XDR)
1133 The total length field allows the daemon to allocate a fixed size
1134 buffer into which it slurps the rest of the message. As a result, the
1135 total length is limited to C<GUESTFS_MESSAGE_MAX> bytes (currently
1136 4MB), which means the effective size of any request is limited to
1137 somewhere under this size.
1139 Note also that many functions don't take any arguments, in which case
1140 the C<guestfs_I<foo>_args> is completely omitted.
1142 The header contains the procedure number (C<guestfs_proc>) which is
1143 how the receiver knows what type of args structure to expect, or none
1146 The reply message for ordinary functions is:
1148 total length (header + ret,
1149 but not including the length word itself)
1150 struct guestfs_message_header (encoded as XDR)
1151 struct guestfs_<foo>_ret (encoded as XDR)
1153 As above the C<guestfs_I<foo>_ret> structure may be completely omitted
1154 for functions that return no formal return values.
1156 As above the total length of the reply is limited to
1157 C<GUESTFS_MESSAGE_MAX>.
1159 In the case of an error, a flag is set in the header, and the reply
1160 message is slightly changed:
1162 total length (header + error,
1163 but not including the length word itself)
1164 struct guestfs_message_header (encoded as XDR)
1165 struct guestfs_message_error (encoded as XDR)
1167 The C<guestfs_message_error> structure contains the error message as a
1170 =head3 FUNCTIONS THAT HAVE FILEIN PARAMETERS
1172 A C<FileIn> parameter indicates that we transfer a file I<into> the
1173 guest. The normal request message is sent (see above). However this
1174 is followed by a sequence of file chunks.
1176 total length (header + arguments,
1177 but not including the length word itself,
1178 and not including the chunks)
1179 struct guestfs_message_header (encoded as XDR)
1180 struct guestfs_<foo>_args (encoded as XDR)
1181 sequence of chunks for FileIn param #0
1182 sequence of chunks for FileIn param #1 etc.
1184 The "sequence of chunks" is:
1186 length of chunk (not including length word itself)
1187 struct guestfs_chunk (encoded as XDR)
1189 struct guestfs_chunk (encoded as XDR)
1192 struct guestfs_chunk (with data.data_len == 0)
1194 The final chunk has the C<data_len> field set to zero. Additionally a
1195 flag is set in the final chunk to indicate either successful
1196 completion or early cancellation.
1198 At time of writing there are no functions that have more than one
1199 FileIn parameter. However this is (theoretically) supported, by
1200 sending the sequence of chunks for each FileIn parameter one after
1201 another (from left to right).
1203 Both the library (sender) I<and> the daemon (receiver) may cancel the
1204 transfer. The library does this by sending a chunk with a special
1205 flag set to indicate cancellation. When the daemon sees this, it
1206 cancels the whole RPC, does I<not> send any reply, and goes back to
1207 reading the next request.
1209 The daemon may also cancel. It does this by writing a special word
1210 C<GUESTFS_CANCEL_FLAG> to the socket. The library listens for this
1211 during the transfer, and if it gets it, it will cancel the transfer
1212 (it sends a cancel chunk). The special word is chosen so that even if
1213 cancellation happens right at the end of the transfer (after the
1214 library has finished writing and has started listening for the reply),
1215 the "spurious" cancel flag will not be confused with the reply
1218 This protocol allows the transfer of arbitrary sized files (no 32 bit
1219 limit), and also files where the size is not known in advance
1220 (eg. from pipes or sockets). However the chunks are rather small
1221 (C<GUESTFS_MAX_CHUNK_SIZE>), so that neither the library nor the
1222 daemon need to keep much in memory.
1224 =head3 FUNCTIONS THAT HAVE FILEOUT PARAMETERS
1226 The protocol for FileOut parameters is exactly the same as for FileIn
1227 parameters, but with the roles of daemon and library reversed.
1229 total length (header + ret,
1230 but not including the length word itself,
1231 and not including the chunks)
1232 struct guestfs_message_header (encoded as XDR)
1233 struct guestfs_<foo>_ret (encoded as XDR)
1234 sequence of chunks for FileOut param #0
1235 sequence of chunks for FileOut param #1 etc.
1237 =head3 INITIAL MESSAGE
1239 Because the underlying channel (QEmu -net channel) doesn't have any
1240 sort of connection control, when the daemon launches it sends an
1241 initial word (C<GUESTFS_LAUNCH_FLAG>) which indicates that the guest
1242 and daemon is alive. This is what C<guestfs_launch> waits for.
1244 =head1 MULTIPLE HANDLES AND MULTIPLE THREADS
1246 All high-level libguestfs actions are synchronous. If you want
1247 to use libguestfs asynchronously then you must create a thread.
1249 Only use the handle from a single thread. Either use the handle
1250 exclusively from one thread, or provide your own mutex so that two
1251 threads cannot issue calls on the same handle at the same time.
1253 =head1 QEMU WRAPPERS
1255 If you want to compile your own qemu, run qemu from a non-standard
1256 location, or pass extra arguments to qemu, then you can write a
1257 shell-script wrapper around qemu.
1259 There is one important rule to remember: you I<must C<exec qemu>> as
1260 the last command in the shell script (so that qemu replaces the shell
1261 and becomes the direct child of the libguestfs-using program). If you
1262 don't do this, then the qemu process won't be cleaned up correctly.
1264 Here is an example of a wrapper, where I have built my own copy of
1268 qemudir=/home/rjones/d/qemu
1269 exec $qemudir/x86_64-softmmu/qemu-system-x86_64 -L $qemudir/pc-bios "$@"
1271 Save this script as C</tmp/qemu.wrapper> (or wherever), C<chmod +x>,
1272 and then use it by setting the LIBGUESTFS_QEMU environment variable.
1275 LIBGUESTFS_QEMU=/tmp/qemu.wrapper guestfish
1277 Note that libguestfs also calls qemu with the -help and -version
1278 options in order to determine features.
1280 =head1 ENVIRONMENT VARIABLES
1284 =item LIBGUESTFS_APPEND
1286 Pass additional options to the guest kernel.
1288 =item LIBGUESTFS_DEBUG
1290 Set C<LIBGUESTFS_DEBUG=1> to enable verbose messages. This
1291 has the same effect as calling C<guestfs_set_verbose (handle, 1)>.
1293 =item LIBGUESTFS_MEMSIZE
1295 Set the memory allocated to the qemu process, in megabytes. For
1298 LIBGUESTFS_MEMSIZE=700
1300 =item LIBGUESTFS_PATH
1302 Set the path that libguestfs uses to search for kernel and initrd.img.
1303 See the discussion of paths in section PATH above.
1305 =item LIBGUESTFS_QEMU
1307 Set the default qemu binary that libguestfs uses. If not set, then
1308 the qemu which was found at compile time by the configure script is
1311 See also L</QEMU WRAPPERS> above.
1313 =item LIBGUESTFS_TRACE
1315 Set C<LIBGUESTFS_TRACE=1> to enable command traces. This
1316 has the same effect as calling C<guestfs_set_trace (handle, 1)>.
1320 Location of temporary directory, defaults to C</tmp>.
1322 If libguestfs was compiled to use the supermin appliance then each
1323 handle will require rather a large amount of space in this directory
1324 for short periods of time (~ 80 MB). You can use C<$TMPDIR> to
1325 configure another directory to use in case C</tmp> is not large
1337 L<virt-inspector(1)>,
1338 L<virt-list-filesystems(1)>,
1339 L<virt-list-partitions(1)>,
1348 L<http://libguestfs.org/>.
1350 Tools with a similar purpose:
1359 To get a list of bugs against libguestfs use this link:
1361 L<https://bugzilla.redhat.com/buglist.cgi?component=libguestfs&product=Virtualization+Tools>
1363 To report a new bug against libguestfs use this link:
1365 L<https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virtualization+Tools>
1367 When reporting a bug, please check:
1373 That the bug hasn't been reported already.
1377 That you are testing a recent version.
1381 Describe the bug accurately, and give a way to reproduce it.
1385 Run libguestfs-test-tool and paste the B<complete, unedited>
1386 output into the bug report.
1392 Richard W.M. Jones (C<rjones at redhat dot com>)
1396 Copyright (C) 2009 Red Hat Inc.
1397 L<http://libguestfs.org/>
1399 This library is free software; you can redistribute it and/or
1400 modify it under the terms of the GNU Lesser General Public
1401 License as published by the Free Software Foundation; either
1402 version 2 of the License, or (at your option) any later version.
1404 This library is distributed in the hope that it will be useful,
1405 but WITHOUT ANY WARRANTY; without even the implied warranty of
1406 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
1407 Lesser General Public License for more details.
1409 You should have received a copy of the GNU Lesser General Public
1410 License along with this library; if not, write to the Free Software
1411 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA