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_umount (handle, "/");
17 guestfs_sync (handle);
18 guestfs_close (handle);
20 cc prog.c -o prog -lguestfs
22 cc prog.c -o prog `pkg-config libguestfs --cflags --libs`
26 Libguestfs is a library for accessing and modifying guest disk images.
27 Amongst the things this is good for: making batch configuration
28 changes to guests, getting disk used/free statistics (see also:
29 virt-df), migrating between virtualization systems (see also:
30 virt-p2v), performing partial backups, performing partial guest
31 clones, cloning guests and changing registry/UUID/hostname info, and
34 Libguestfs uses Linux kernel and qemu code, and can access any type of
35 guest filesystem that Linux and qemu can, including but not limited
36 to: ext2/3/4, btrfs, FAT and NTFS, LVM, many different disk partition
37 schemes, qcow, qcow2, vmdk.
39 Libguestfs provides ways to enumerate guest storage (eg. partitions,
40 LVs, what filesystem is in each LV, etc.). It can also run commands
41 in the context of the guest. Also you can access filesystems over
44 Libguestfs is a library that can be linked with C and C++ management
45 programs (or management programs written in OCaml, Perl, Python, Ruby,
46 Java, Haskell or C#). You can also use it from shell scripts or the
49 You don't need to be root to use libguestfs, although obviously you do
50 need enough permissions to access the disk images.
52 Libguestfs is a large API because it can do many things. For a gentle
53 introduction, please read the L</API OVERVIEW> section next.
57 This section provides a gentler overview of the libguestfs API. We
58 also try to group API calls together, where that may not be obvious
59 from reading about the individual calls below.
63 Before you can use libguestfs calls, you have to create a handle.
64 Then you must add at least one disk image to the handle, followed by
65 launching the handle, then performing whatever operations you want,
66 and finally closing the handle. So the general structure of all
67 libguestfs-using programs looks like this:
69 guestfs_h *handle = guestfs_create ();
71 /* Call guestfs_add_drive additional times if there are
72 * multiple disk images.
74 guestfs_add_drive (handle, "guest.img");
76 /* Most manipulation calls won't work until you've launched
77 * the handle. You have to do this _after_ adding drives
78 * and _before_ other commands.
80 guestfs_launch (handle);
82 /* Now you can examine what partitions, LVs etc are available.
84 char **partitions = guestfs_list_partitions (handle);
85 char **logvols = guestfs_lvs (handle);
87 /* To access a filesystem in the image, you must mount it.
89 guestfs_mount (handle, "/dev/sda1", "/");
91 /* Now you can perform filesystem actions on the guest
94 guestfs_touch (handle, "/hello");
96 /* You only need to call guestfs_sync if you have made
97 * changes to the guest image.
99 guestfs_sync (handle);
101 /* Close the handle. */
102 guestfs_close (handle);
104 The code above doesn't include any error checking. In real code you
105 should check return values carefully for errors. In general all
106 functions that return integers return C<-1> on error, and all
107 functions that return pointers return C<NULL> on error. See section
108 L</ERROR HANDLING> below for how to handle errors, and consult the
109 documentation for each function call below to see precisely how they
110 return error indications.
114 The image filename (C<"guest.img"> in the example above) could be a
115 disk image from a virtual machine, a L<dd(1)> copy of a physical hard
116 disk, an actual block device, or simply an empty file of zeroes that
117 you have created through L<posix_fallocate(3)>. Libguestfs lets you
118 do useful things to all of these.
120 You can add a disk read-only using C<guestfs_add_drive_ro>, in which
121 case libguestfs won't modify the file.
123 Be extremely cautious if the disk image is in use, eg. if it is being
124 used by a virtual machine. Adding it read-write will almost certainly
125 cause disk corruption, but adding it read-only is safe.
127 You must add at least one disk image, and you may add multiple disk
128 images. In the API, the disk images are usually referred to as
129 C</dev/sda> (for the first one you added), C</dev/sdb> (for the second
132 Once C<guestfs_launch> has been called you cannot add any more images.
133 You can call C<guestfs_list_devices> to get a list of the device
134 names, in the order that you added them. See also L</BLOCK DEVICE
139 Before you can read or write files, create directories and so on in a
140 disk image that contains filesystems, you have to mount those
141 filesystems using C<guestfs_mount>. If you already know that a disk
142 image contains (for example) one partition with a filesystem on that
143 partition, then you can mount it directly:
145 guestfs_mount (handle, "/dev/sda1", "/");
147 where C</dev/sda1> means literally the first partition (C<1>) of the
148 first disk image that we added (C</dev/sda>). If the disk contains
149 Linux LVM2 logical volumes you could refer to those instead (eg. C</dev/VG/LV>).
151 If you are given a disk image and you don't know what it contains then
152 you have to find out. Libguestfs can do that too: use
153 C<guestfs_list_partitions> and C<guestfs_lvs> to list possible
154 partitions and LVs, and either try mounting each to see what is
155 mountable, or else examine them with C<guestfs_file>. But you might
156 find it easier to look at higher level programs built on top of
157 libguestfs, in particular L<virt-inspector(1)>.
159 To mount a disk image read-only, use C<guestfs_mount_ro>. There are
160 several other variations of the C<guestfs_mount_*> call.
162 =head2 FILESYSTEM ACCESS AND MODIFICATION
164 The majority of the libguestfs API consists of fairly low-level calls
165 for accessing and modifying the files, directories, symlinks etc on
166 mounted filesystems. There are over a hundred such calls which you
167 can find listed in detail below in this man page, and we don't even
168 pretend to cover them all in this overview.
170 Specify filenames as full paths including the mount point.
172 For example, if you mounted a filesystem at C<"/"> and you want to
173 read the file called C<"etc/passwd"> then you could do:
175 char *data = guestfs_cat (handle, "/etc/passwd");
177 This would return C<data> as a newly allocated buffer containing the
178 full content of that file (with some conditions: see also
179 L</DOWNLOADING> below), or C<NULL> if there was an error.
181 As another example, to create a top-level directory on that filesystem
182 called C<"var"> you would do:
184 guestfs_mkdir (handle, "/var");
186 To create a symlink you could do:
188 guestfs_ln_s (handle, "/etc/init.d/portmap",
189 "/etc/rc3.d/S30portmap");
191 Libguestfs will reject attempts to use relative paths. There is no
192 concept of a current working directory. Libguestfs can return errors
193 in many situations: for example if the filesystem isn't writable, or
194 if a file or directory that you requested doesn't exist. If you are
195 using the C API (documented here) you have to check for those error
196 conditions after each call. (Other language bindings turn these
197 errors into exceptions).
199 File writes are affected by the per-handle umask, set by calling
200 C<guestfs_umask> and defaulting to 022.
204 Libguestfs contains API calls to read, create and modify partition
205 tables on disk images.
207 In the common case where you want to create a single partition
208 covering the whole disk, you should use the C<guestfs_part_disk>
211 const char *parttype = "mbr";
212 if (disk_is_larger_than_2TB)
214 guestfs_part_disk (g, "/dev/sda", parttype);
216 Obviously this effectively wipes anything that was on that disk image
219 In general MBR partitions are both unnecessarily complicated and
220 depend on archaic details, namely the Cylinder-Head-Sector (CHS)
221 geometry of the disk. C<guestfs_sfdiskM> can be used to
222 create more complex arrangements where the relative sizes are
223 expressed in megabytes instead of cylinders, which is a small win.
224 C<guestfs_sfdiskM> will choose the nearest cylinder to approximate the
225 requested size. There's a lot of crazy stuff to do with IDE and
226 virtio disks having different, incompatible CHS geometries, that you
227 probably don't want to know about.
229 My advice: make a single partition to cover the whole disk, then use
234 Libguestfs provides access to a large part of the LVM2 API, such as
235 C<guestfs_lvcreate> and C<guestfs_vgremove>. It won't make much sense
236 unless you familiarize yourself with the concepts of physical volumes,
237 volume groups and logical volumes.
239 This author strongly recommends reading the LVM HOWTO, online at
240 L<http://tldp.org/HOWTO/LVM-HOWTO/>.
244 Use C<guestfs_cat> to download small, text only files. This call
245 is limited to files which are less than 2 MB and which cannot contain
246 any ASCII NUL (C<\0>) characters. However it has a very simple
249 C<guestfs_read_file> can be used to read files which contain
250 arbitrary 8 bit data, since it returns a (pointer, size) pair.
251 However it is still limited to "small" files, less than 2 MB.
253 C<guestfs_download> can be used to download any file, with no
254 limits on content or size (even files larger than 4 GB).
256 To download multiple files, see C<guestfs_tar_out> and
261 It's often the case that you want to write a file or files to the disk
264 For small, single files, use C<guestfs_write_file>. This call
265 currently contains a bug which limits the call to plain text files
266 (not containing ASCII NUL characters).
268 To upload a single file, use C<guestfs_upload>. This call has no
269 limits on file content or size (even files larger than 4 GB).
271 To upload multiple files, see C<guestfs_tar_in> and C<guestfs_tgz_in>.
273 However the fastest way to upload I<large numbers of arbitrary files>
274 is to turn them into a squashfs or CD ISO (see L<mksquashfs(8)> and
275 L<mkisofs(8)>), then attach this using C<guestfs_add_drive_ro>. If
276 you add the drive in a predictable way (eg. adding it last after all
277 other drives) then you can get the device name from
278 C<guestfs_list_devices> and mount it directly using
279 C<guestfs_mount_ro>. Note that squashfs images are sometimes
280 non-portable between kernel versions, and they don't support labels or
281 UUIDs. If you want to pre-build an image or you need to mount it
282 using a label or UUID, use an ISO image instead.
286 There are various different commands for copying between files and
287 devices and in and out of the guest filesystem. These are summarised
292 =item B<file> to B<file>
294 Use L</guestfs_cp> to copy a single file, or
295 L</guestfs_cp_a> to copy directories recursively.
297 =item B<file or device> to B<file or device>
299 Use L</guestfs_dd> which efficiently uses L<dd(1)>
300 to copy between files and devices in the guest.
302 Example: duplicate the contents of an LV:
304 guestfs_dd (g, "/dev/VG/Original", "/dev/VG/Copy");
306 The destination (C</dev/VG/Copy>) must be at least as large as the
307 source (C</dev/VG/Original>).
309 =item B<file on the host> to B<file or device>
311 Use L</guestfs_upload>. See L</UPLOADING> above.
313 =item B<file or device> to B<file on the host>
315 Use L</guestfs_download>. See L</DOWNLOADING> above.
321 C<guestfs_ll> is just designed for humans to read (mainly when using
322 the L<guestfish(1)>-equivalent command C<ll>).
324 C<guestfs_ls> is a quick way to get a list of files in a directory
325 from programs, as a flat list of strings.
327 C<guestfs_readdir> is a programmatic way to get a list of files in a
328 directory, plus additional information about each one. It is more
329 equivalent to using the L<readdir(3)> call on a local filesystem.
331 C<guestfs_find> can be used to recursively list files.
333 =head2 RUNNING COMMANDS
335 Although libguestfs is a primarily an API for manipulating files
336 inside guest images, we also provide some limited facilities for
337 running commands inside guests.
339 There are many limitations to this:
345 The kernel version that the command runs under will be different
346 from what it expects.
350 If the command needs to communicate with daemons, then most likely
351 they won't be running.
355 The command will be running in limited memory.
359 Only supports Linux guests (not Windows, BSD, etc).
363 Architecture limitations (eg. won't work for a PPC guest on
368 For SELinux guests, you may need to enable SELinux and load policy
369 first. See L</SELINUX> in this manpage.
373 The two main API calls to run commands are C<guestfs_command> and
374 C<guestfs_sh> (there are also variations).
376 The difference is that C<guestfs_sh> runs commands using the shell, so
377 any shell globs, redirections, etc will work.
379 =head2 CONFIGURATION FILES
381 To read and write configuration files in Linux guest filesystems, we
382 strongly recommend using Augeas. For example, Augeas understands how
383 to read and write, say, a Linux shadow password file or X.org
384 configuration file, and so avoids you having to write that code.
386 The main Augeas calls are bound through the C<guestfs_aug_*> APIs. We
387 don't document Augeas itself here because there is excellent
388 documentation on the L<http://augeas.net/> website.
390 If you don't want to use Augeas (you fool!) then try calling
391 C<guestfs_read_lines> to get the file as a list of lines which
392 you can iterate over.
396 We support SELinux guests. To ensure that labeling happens correctly
397 in SELinux guests, you need to enable SELinux and load the guest's
404 Before launching, do:
406 guestfs_set_selinux (g, 1);
410 After mounting the guest's filesystem(s), load the policy. This
411 is best done by running the L<load_policy(8)> command in the
414 guestfs_sh (g, "/usr/sbin/load_policy");
416 (Older versions of C<load_policy> require you to specify the
417 name of the policy file).
421 Optionally, set the security context for the API. The correct
422 security context to use can only be known by inspecting the
423 guest. As an example:
425 guestfs_setcon (g, "unconfined_u:unconfined_r:unconfined_t:s0");
429 This will work for running commands and editing existing files.
431 When new files are created, you may need to label them explicitly,
432 for example by running the external command
433 C<restorecon pathname>.
435 =head2 SPECIAL CONSIDERATIONS FOR WINDOWS GUESTS
437 Libguestfs can mount NTFS partitions. It does this using the
438 L<http://www.ntfs-3g.org/> driver.
440 DOS and Windows still use drive letters, and the filesystems are
441 always treated as case insensitive by Windows itself, and therefore
442 you might find a Windows configuration file referring to a path like
443 C<c:\windows\system32>. When the filesystem is mounted in libguestfs,
444 that directory might be referred to as C</WINDOWS/System32>.
446 Drive letter mappings are outside the scope of libguestfs. You have
447 to use libguestfs to read the appropriate Windows Registry and
448 configuration files, to determine yourself how drives are mapped (see
449 also L<virt-inspector(1)>).
451 Replacing backslash characters with forward slash characters is also
452 outside the scope of libguestfs, but something that you can easily do.
454 Where we can help is in resolving the case insensitivity of paths.
455 For this, call C<guestfs_case_sensitive_path>.
457 Libguestfs also provides some help for decoding Windows Registry
458 "hive" files, through the library C<hivex> which is part of the
459 libguestfs project. You have to locate and download the hive file(s)
460 yourself, and then pass them to C<hivex> functions. See also the
461 programs L<hivexml(1)>, L<hivexsh(1)> and L<virt-win-reg(1)> for more
464 =head2 USING LIBGUESTFS WITH OTHER PROGRAMMING LANGUAGES
466 Although we don't want to discourage you from using the C API, we will
467 mention here that the same API is also available in other languages.
469 The API is broadly identical in all supported languages. This means
470 that the C call C<guestfs_mount(handle,path)> is
471 C<$handle-E<gt>mount($path)> in Perl, C<handle.mount(path)> in Python,
472 and C<Guestfs.mount handle path> in OCaml. In other words, a
473 straightforward, predictable isomorphism between each language.
475 Error messages are automatically transformed
476 into exceptions if the language supports it.
478 We don't try to "object orientify" parts of the API in OO languages,
479 although contributors are welcome to write higher level APIs above
480 what we provide in their favourite languages if they wish.
486 You can use the I<guestfs.h> header file from C++ programs. The C++
487 API is identical to the C API. C++ classes and exceptions are
492 The C# bindings are highly experimental. Please read the warnings
493 at the top of C<csharp/Libguestfs.cs>.
497 This is the only language binding that working but incomplete. Only
498 calls which return simple integers have been bound in Haskell, and we
499 are looking for help to complete this binding.
503 Full documentation is contained in the Javadoc which is distributed
508 For documentation see the file C<guestfs.mli>.
512 For documentation see L<Sys::Guestfs(3)>.
516 For documentation do:
524 Use the Guestfs module. There is no Ruby-specific documentation, but
525 you can find examples written in Ruby in the libguestfs source.
527 =item B<shell scripts>
529 For documentation see L<guestfish(1)>.
533 =head2 LIBGUESTFS GOTCHAS
535 L<http://en.wikipedia.org/wiki/Gotcha_(programming)>: "A feature of a
536 system [...] that works in the way it is documented but is
537 counterintuitive and almost invites mistakes."
539 Since we developed libguestfs and the associated tools, there are
540 several things we would have designed differently, but are now stuck
541 with for backwards compatibility or other reasons. If there is ever a
542 libguestfs 2.0 release, you can expect these to change. Beware of
547 =item Autosync / forgetting to sync.
549 When modifying a filesystem from C or another language, you B<must>
550 unmount all filesystems and call L</guestfs_sync> explicitly before
551 you close the libguestfs handle. You can also call:
553 guestfs_set_autosync (handle, 1);
555 to have the unmount/sync done automatically for you when the handle is
556 closed. (This feature is called "autosync", L</guestfs_set_autosync>
559 If you forget to do this, then it is entirely possible that your
560 changes won't be written out, or will be partially written, or (very
561 rarely) that you'll get disk corruption.
563 Note that in L<guestfish(3)> I<autosync is the default>. So quick and
564 dirty guestfish scripts that forget to sync will work just fine, which
565 can make this extra-puzzling if you are trying to debug a problem.
567 =item Mount option C<-o sync> should not be the default.
569 If you use C<guestfs_mount>, then C<-o sync,noatime> are added
570 implicitly. However C<-o sync> does not add any reliability benefit,
571 but does have a very large performance impact.
573 The work around is to use C<guestfs_mount_options> and set the mount
574 options that you actually want to use.
576 =item Read-only should be the default.
578 In L<guestfish(3)>, I<--ro> should be the default, and you should
579 have to specify I<--rw> if you want to make changes to the image.
581 This would reduce the potential to corrupt live VM images.
583 Note that many filesystems change the disk when you just mount and
584 unmount, even if you didn't perform any writes. You need to use
585 C<guestfs_add_drive_ro> to guarantee that the disk is not changed.
587 =item guestfish command line is hard to use.
589 C<guestfish disk.img> doesn't do what people expect (open C<disk.img>
590 for examination). It tries to run a guestfish command C<disk.img>
591 which doesn't exist, so it fails, and it fails with a strange and
592 unintuitive error message. Like the Bourne shell, we should have used
593 C<guestfish -c command> to run commands.
597 =head2 PROTOCOL LIMITS
599 Internally libguestfs uses a message-based protocol to pass API calls
600 and their responses to and from a small "appliance" (see L</INTERNALS>
601 for plenty more detail about this). The maximum message size used by
602 the protocol is slightly less than 4 MB. For some API calls you may
603 need to be aware of this limit. The API calls which may be affected
604 are individually documented, with a link back to this section of the
607 A simple call such as C<guestfs_cat> returns its result (the file
608 data) in a simple string. Because this string is at some point
609 internally encoded as a message, the maximum size that it can return
610 is slightly under 4 MB. If the requested file is larger than this
611 then you will get an error.
613 In order to transfer large files into and out of the guest filesystem,
614 you need to use particular calls that support this. The sections
615 L</UPLOADING> and L</DOWNLOADING> document how to do this.
617 You might also consider mounting the disk image using our FUSE
618 filesystem support (L<guestmount(1)>).
620 =head1 CONNECTION MANAGEMENT
624 C<guestfs_h> is the opaque type representing a connection handle.
625 Create a handle by calling C<guestfs_create>. Call C<guestfs_close>
626 to free the handle and release all resources used.
628 For information on using multiple handles and threads, see the section
629 L</MULTIPLE HANDLES AND MULTIPLE THREADS> below.
631 =head2 guestfs_create
633 guestfs_h *guestfs_create (void);
635 Create a connection handle.
637 You have to call C<guestfs_add_drive> on the handle at least once.
639 This function returns a non-NULL pointer to a handle on success or
642 After configuring the handle, you have to call C<guestfs_launch>.
644 You may also want to configure error handling for the handle. See
645 L</ERROR HANDLING> section below.
649 void guestfs_close (guestfs_h *handle);
651 This closes the connection handle and frees up all resources used.
653 =head1 ERROR HANDLING
655 The convention in all functions that return C<int> is that they return
656 C<-1> to indicate an error. You can get additional information on
657 errors by calling C<guestfs_last_error> and/or by setting up an error
658 handler with C<guestfs_set_error_handler>.
660 The default error handler prints the information string to C<stderr>.
662 Out of memory errors are handled differently. The default action is
663 to call L<abort(3)>. If this is undesirable, then you can set a
664 handler using C<guestfs_set_out_of_memory_handler>.
666 =head2 guestfs_last_error
668 const char *guestfs_last_error (guestfs_h *handle);
670 This returns the last error message that happened on C<handle>. If
671 there has not been an error since the handle was created, then this
674 The lifetime of the returned string is until the next error occurs, or
675 C<guestfs_close> is called.
677 The error string is not localized (ie. is always in English), because
678 this makes searching for error messages in search engines give the
679 largest number of results.
681 =head2 guestfs_set_error_handler
683 typedef void (*guestfs_error_handler_cb) (guestfs_h *handle,
686 void guestfs_set_error_handler (guestfs_h *handle,
687 guestfs_error_handler_cb cb,
690 The callback C<cb> will be called if there is an error. The
691 parameters passed to the callback are an opaque data pointer and the
692 error message string.
694 Note that the message string C<msg> is freed as soon as the callback
695 function returns, so if you want to stash it somewhere you must make
698 The default handler prints messages on C<stderr>.
700 If you set C<cb> to C<NULL> then I<no> handler is called.
702 =head2 guestfs_get_error_handler
704 guestfs_error_handler_cb guestfs_get_error_handler (guestfs_h *handle,
707 Returns the current error handler callback.
709 =head2 guestfs_set_out_of_memory_handler
711 typedef void (*guestfs_abort_cb) (void);
712 int guestfs_set_out_of_memory_handler (guestfs_h *handle,
715 The callback C<cb> will be called if there is an out of memory
716 situation. I<Note this callback must not return>.
718 The default is to call L<abort(3)>.
720 You cannot set C<cb> to C<NULL>. You can't ignore out of memory
723 =head2 guestfs_get_out_of_memory_handler
725 guestfs_abort_fn guestfs_get_out_of_memory_handler (guestfs_h *handle);
727 This returns the current out of memory handler.
731 Libguestfs needs a kernel and initrd.img, which it finds by looking
732 along an internal path.
734 By default it looks for these in the directory C<$libdir/guestfs>
735 (eg. C</usr/local/lib/guestfs> or C</usr/lib64/guestfs>).
737 Use C<guestfs_set_path> or set the environment variable
738 C<LIBGUESTFS_PATH> to change the directories that libguestfs will
739 search in. The value is a colon-separated list of paths. The current
740 directory is I<not> searched unless the path contains an empty element
741 or C<.>. For example C<LIBGUESTFS_PATH=:/usr/lib/guestfs> would
742 search the current directory and then C</usr/lib/guestfs>.
744 =head1 HIGH-LEVEL API ACTIONS
748 We guarantee the libguestfs ABI (binary interface), for public,
749 high-level actions as outlined in this section. Although we will
750 deprecate some actions, for example if they get replaced by newer
751 calls, we will keep the old actions forever. This allows you the
752 developer to program in confidence against libguestfs.
762 =head2 GROUPS OF FUNCTIONALITY IN THE APPLIANCE
764 Using L</guestfs_available> you can test availability of
765 the following groups of functions. This test queries the
766 appliance to see if the appliance you are currently using
767 supports the functionality.
771 =head2 SINGLE CALLS AT COMPILE TIME
773 If you need to test whether a single libguestfs function is
774 available at compile time, we recommend using build tools
775 such as autoconf or cmake. For example in autotools you could
778 AC_CHECK_LIB([guestfs],[guestfs_create])
779 AC_CHECK_FUNCS([guestfs_dd])
781 which would result in C<HAVE_GUESTFS_DD> being either defined
782 or not defined in your program.
784 =head2 SINGLE CALLS AT RUN TIME
786 Testing at compile time doesn't guarantee that a function really
787 exists in the library. The reason is that you might be dynamically
788 linked against a previous I<libguestfs.so> (dynamic library)
789 which doesn't have the call. This situation unfortunately results
790 in a segmentation fault, which is a shortcoming of the C dynamic
791 linking system itself.
793 You can use L<dlopen(3)> to test if a function is available
794 at run time, as in this example program (note that you still
795 need the compile time check as well):
807 #ifdef HAVE_GUESTFS_DD
811 /* Test if the function guestfs_dd is really available. */
812 dl = dlopen (NULL, RTLD_LAZY);
814 fprintf (stderr, "dlopen: %s\n", dlerror ());
817 has_function = dlsym (dl, "guestfs_dd") != NULL;
821 printf ("this libguestfs.so does NOT have guestfs_dd function\n");
823 printf ("this libguestfs.so has guestfs_dd function\n");
824 /* Now it's safe to call
825 guestfs_dd (g, "foo", "bar");
829 printf ("guestfs_dd function was not found at compile time\n");
833 You may think the above is an awful lot of hassle, and it is.
834 There are other ways outside of the C linking system to ensure
835 that this kind of incompatibility never arises, such as using
838 Requires: libguestfs >= 1.0.80
842 <!-- old anchor for the next section -->
843 <a name="state_machine_and_low_level_event_api"/>
849 Internally, libguestfs is implemented by running an appliance (a
850 special type of small virtual machine) using L<qemu(1)>. Qemu runs as
851 a child process of the main program.
857 | | child process / appliance
858 | | __________________________
860 +-------------------+ RPC | +-----------------+ |
861 | libguestfs <--------------------> guestfsd | |
862 | | | +-----------------+ |
863 \___________________/ | | Linux kernel | |
864 | +--^--------------+ |
865 \_________|________________/
873 The library, linked to the main program, creates the child process and
874 hence the appliance in the L</guestfs_launch> function.
876 Inside the appliance is a Linux kernel and a complete stack of
877 userspace tools (such as LVM and ext2 programs) and a small
878 controlling daemon called C<guestfsd>. The library talks to
879 C<guestfsd> using remote procedure calls (RPC). There is a mostly
880 one-to-one correspondence between libguestfs API calls and RPC calls
881 to the daemon. Lastly the disk image(s) are attached to the qemu
882 process which translates device access by the appliance's Linux kernel
883 into accesses to the image.
885 A common misunderstanding is that the appliance "is" the virtual
886 machine. Although the disk image you are attached to might also be
887 used by some virtual machine, libguestfs doesn't know or care about
888 this. (But you will care if both libguestfs's qemu process and your
889 virtual machine are trying to update the disk image at the same time,
890 since these usually results in massive disk corruption).
894 libguestfs uses a state machine to model the child process:
905 / | \ \ guestfs_launch
916 \______/ <------ \________/
918 The normal transitions are (1) CONFIG (when the handle is created, but
919 there is no child process), (2) LAUNCHING (when the child process is
920 booting up), (3) alternating between READY and BUSY as commands are
921 issued to, and carried out by, the child process.
923 The guest may be killed by C<guestfs_kill_subprocess>, or may die
924 asynchronously at any time (eg. due to some internal error), and that
925 causes the state to transition back to CONFIG.
927 Configuration commands for qemu such as C<guestfs_add_drive> can only
928 be issued when in the CONFIG state.
930 The high-level API offers two calls that go from CONFIG through
931 LAUNCHING to READY. C<guestfs_launch> blocks until the child process
932 is READY to accept commands (or until some failure or timeout).
933 C<guestfs_launch> internally moves the state from CONFIG to LAUNCHING
936 High-level API actions such as C<guestfs_mount> can only be issued
937 when in the READY state. These high-level API calls block waiting for
938 the command to be carried out (ie. the state to transition to BUSY and
939 then back to READY). But using the low-level event API, you get
940 non-blocking versions. (But you can still only carry out one
941 operation per handle at a time - that is a limitation of the
942 communications protocol we use).
944 Finally, the child process sends asynchronous messages back to the
945 main program, such as kernel log messages. Mostly these are ignored
946 by the high-level API, but using the low-level event API you can
947 register to receive these messages.
949 =head2 SETTING CALLBACKS TO HANDLE EVENTS
951 The child process generates events in some situations. Current events
952 include: receiving a log message, the child process exits.
954 Use the C<guestfs_set_*_callback> functions to set a callback for
955 different types of events.
957 Only I<one callback of each type> can be registered for each handle.
958 Calling C<guestfs_set_*_callback> again overwrites the previous
959 callback of that type. Cancel all callbacks of this type by calling
960 this function with C<cb> set to C<NULL>.
962 =head2 guestfs_set_log_message_callback
964 typedef void (*guestfs_log_message_cb) (guestfs_h *g, void *opaque,
966 void guestfs_set_log_message_callback (guestfs_h *handle,
967 guestfs_log_message_cb cb,
970 The callback function C<cb> will be called whenever qemu or the guest
971 writes anything to the console.
973 Use this function to capture kernel messages and similar.
975 Normally there is no log message handler, and log messages are just
978 =head2 guestfs_set_subprocess_quit_callback
980 typedef void (*guestfs_subprocess_quit_cb) (guestfs_h *g, void *opaque);
981 void guestfs_set_subprocess_quit_callback (guestfs_h *handle,
982 guestfs_subprocess_quit_cb cb,
985 The callback function C<cb> will be called when the child process
986 quits, either asynchronously or if killed by
987 C<guestfs_kill_subprocess>. (This corresponds to a transition from
988 any state to the CONFIG state).
990 =head2 guestfs_set_launch_done_callback
992 typedef void (*guestfs_launch_done_cb) (guestfs_h *g, void *opaque);
993 void guestfs_set_launch_done_callback (guestfs_h *handle,
997 The callback function C<cb> will be called when the child process
998 becomes ready first time after it has been launched. (This
999 corresponds to a transition from LAUNCHING to the READY state).
1001 =head1 BLOCK DEVICE NAMING
1003 In the kernel there is now quite a profusion of schemata for naming
1004 block devices (in this context, by I<block device> I mean a physical
1005 or virtual hard drive). The original Linux IDE driver used names
1006 starting with C</dev/hd*>. SCSI devices have historically used a
1007 different naming scheme, C</dev/sd*>. When the Linux kernel I<libata>
1008 driver became a popular replacement for the old IDE driver
1009 (particularly for SATA devices) those devices also used the
1010 C</dev/sd*> scheme. Additionally we now have virtual machines with
1011 paravirtualized drivers. This has created several different naming
1012 systems, such as C</dev/vd*> for virtio disks and C</dev/xvd*> for Xen
1015 As discussed above, libguestfs uses a qemu appliance running an
1016 embedded Linux kernel to access block devices. We can run a variety
1017 of appliances based on a variety of Linux kernels.
1019 This causes a problem for libguestfs because many API calls use device
1020 or partition names. Working scripts and the recipe (example) scripts
1021 that we make available over the internet could fail if the naming
1024 Therefore libguestfs defines C</dev/sd*> as the I<standard naming
1025 scheme>. Internally C</dev/sd*> names are translated, if necessary,
1026 to other names as required. For example, under RHEL 5 which uses the
1027 C</dev/hd*> scheme, any device parameter C</dev/sda2> is translated to
1028 C</dev/hda2> transparently.
1030 Note that this I<only> applies to parameters. The
1031 C<guestfs_list_devices>, C<guestfs_list_partitions> and similar calls
1032 return the true names of the devices and partitions as known to the
1035 =head2 ALGORITHM FOR BLOCK DEVICE NAME TRANSLATION
1037 Usually this translation is transparent. However in some (very rare)
1038 cases you may need to know the exact algorithm. Such cases include
1039 where you use C<guestfs_config> to add a mixture of virtio and IDE
1040 devices to the qemu-based appliance, so have a mixture of C</dev/sd*>
1041 and C</dev/vd*> devices.
1043 The algorithm is applied only to I<parameters> which are known to be
1044 either device or partition names. Return values from functions such
1045 as C<guestfs_list_devices> are never changed.
1051 Is the string a parameter which is a device or partition name?
1055 Does the string begin with C</dev/sd>?
1059 Does the named device exist? If so, we use that device.
1060 However if I<not> then we continue with this algorithm.
1064 Replace initial C</dev/sd> string with C</dev/hd>.
1066 For example, change C</dev/sda2> to C</dev/hda2>.
1068 If that named device exists, use it. If not, continue.
1072 Replace initial C</dev/sd> string with C</dev/vd>.
1074 If that named device exists, use it. If not, return an error.
1078 =head2 PORTABILITY CONCERNS
1080 Although the standard naming scheme and automatic translation is
1081 useful for simple programs and guestfish scripts, for larger programs
1082 it is best not to rely on this mechanism.
1084 Where possible for maximum future portability programs using
1085 libguestfs should use these future-proof techniques:
1091 Use C<guestfs_list_devices> or C<guestfs_list_partitions> to list
1092 actual device names, and then use those names directly.
1094 Since those device names exist by definition, they will never be
1099 Use higher level ways to identify filesystems, such as LVM names,
1100 UUIDs and filesystem labels.
1106 =head2 COMMUNICATION PROTOCOL
1108 Don't rely on using this protocol directly. This section documents
1109 how it currently works, but it may change at any time.
1111 The protocol used to talk between the library and the daemon running
1112 inside the qemu virtual machine is a simple RPC mechanism built on top
1113 of XDR (RFC 1014, RFC 1832, RFC 4506).
1115 The detailed format of structures is in C<src/guestfs_protocol.x>
1116 (note: this file is automatically generated).
1118 There are two broad cases, ordinary functions that don't have any
1119 C<FileIn> and C<FileOut> parameters, which are handled with very
1120 simple request/reply messages. Then there are functions that have any
1121 C<FileIn> or C<FileOut> parameters, which use the same request and
1122 reply messages, but they may also be followed by files sent using a
1125 =head3 ORDINARY FUNCTIONS (NO FILEIN/FILEOUT PARAMS)
1127 For ordinary functions, the request message is:
1129 total length (header + arguments,
1130 but not including the length word itself)
1131 struct guestfs_message_header (encoded as XDR)
1132 struct guestfs_<foo>_args (encoded as XDR)
1134 The total length field allows the daemon to allocate a fixed size
1135 buffer into which it slurps the rest of the message. As a result, the
1136 total length is limited to C<GUESTFS_MESSAGE_MAX> bytes (currently
1137 4MB), which means the effective size of any request is limited to
1138 somewhere under this size.
1140 Note also that many functions don't take any arguments, in which case
1141 the C<guestfs_I<foo>_args> is completely omitted.
1143 The header contains the procedure number (C<guestfs_proc>) which is
1144 how the receiver knows what type of args structure to expect, or none
1147 The reply message for ordinary functions is:
1149 total length (header + ret,
1150 but not including the length word itself)
1151 struct guestfs_message_header (encoded as XDR)
1152 struct guestfs_<foo>_ret (encoded as XDR)
1154 As above the C<guestfs_I<foo>_ret> structure may be completely omitted
1155 for functions that return no formal return values.
1157 As above the total length of the reply is limited to
1158 C<GUESTFS_MESSAGE_MAX>.
1160 In the case of an error, a flag is set in the header, and the reply
1161 message is slightly changed:
1163 total length (header + error,
1164 but not including the length word itself)
1165 struct guestfs_message_header (encoded as XDR)
1166 struct guestfs_message_error (encoded as XDR)
1168 The C<guestfs_message_error> structure contains the error message as a
1171 =head3 FUNCTIONS THAT HAVE FILEIN PARAMETERS
1173 A C<FileIn> parameter indicates that we transfer a file I<into> the
1174 guest. The normal request message is sent (see above). However this
1175 is followed by a sequence of file chunks.
1177 total length (header + arguments,
1178 but not including the length word itself,
1179 and not including the chunks)
1180 struct guestfs_message_header (encoded as XDR)
1181 struct guestfs_<foo>_args (encoded as XDR)
1182 sequence of chunks for FileIn param #0
1183 sequence of chunks for FileIn param #1 etc.
1185 The "sequence of chunks" is:
1187 length of chunk (not including length word itself)
1188 struct guestfs_chunk (encoded as XDR)
1190 struct guestfs_chunk (encoded as XDR)
1193 struct guestfs_chunk (with data.data_len == 0)
1195 The final chunk has the C<data_len> field set to zero. Additionally a
1196 flag is set in the final chunk to indicate either successful
1197 completion or early cancellation.
1199 At time of writing there are no functions that have more than one
1200 FileIn parameter. However this is (theoretically) supported, by
1201 sending the sequence of chunks for each FileIn parameter one after
1202 another (from left to right).
1204 Both the library (sender) I<and> the daemon (receiver) may cancel the
1205 transfer. The library does this by sending a chunk with a special
1206 flag set to indicate cancellation. When the daemon sees this, it
1207 cancels the whole RPC, does I<not> send any reply, and goes back to
1208 reading the next request.
1210 The daemon may also cancel. It does this by writing a special word
1211 C<GUESTFS_CANCEL_FLAG> to the socket. The library listens for this
1212 during the transfer, and if it gets it, it will cancel the transfer
1213 (it sends a cancel chunk). The special word is chosen so that even if
1214 cancellation happens right at the end of the transfer (after the
1215 library has finished writing and has started listening for the reply),
1216 the "spurious" cancel flag will not be confused with the reply
1219 This protocol allows the transfer of arbitrary sized files (no 32 bit
1220 limit), and also files where the size is not known in advance
1221 (eg. from pipes or sockets). However the chunks are rather small
1222 (C<GUESTFS_MAX_CHUNK_SIZE>), so that neither the library nor the
1223 daemon need to keep much in memory.
1225 =head3 FUNCTIONS THAT HAVE FILEOUT PARAMETERS
1227 The protocol for FileOut parameters is exactly the same as for FileIn
1228 parameters, but with the roles of daemon and library reversed.
1230 total length (header + ret,
1231 but not including the length word itself,
1232 and not including the chunks)
1233 struct guestfs_message_header (encoded as XDR)
1234 struct guestfs_<foo>_ret (encoded as XDR)
1235 sequence of chunks for FileOut param #0
1236 sequence of chunks for FileOut param #1 etc.
1238 =head3 INITIAL MESSAGE
1240 Because the underlying channel (QEmu -net channel) doesn't have any
1241 sort of connection control, when the daemon launches it sends an
1242 initial word (C<GUESTFS_LAUNCH_FLAG>) which indicates that the guest
1243 and daemon is alive. This is what C<guestfs_launch> waits for.
1245 =head1 MULTIPLE HANDLES AND MULTIPLE THREADS
1247 All high-level libguestfs actions are synchronous. If you want
1248 to use libguestfs asynchronously then you must create a thread.
1250 Only use the handle from a single thread. Either use the handle
1251 exclusively from one thread, or provide your own mutex so that two
1252 threads cannot issue calls on the same handle at the same time.
1254 =head1 QEMU WRAPPERS
1256 If you want to compile your own qemu, run qemu from a non-standard
1257 location, or pass extra arguments to qemu, then you can write a
1258 shell-script wrapper around qemu.
1260 There is one important rule to remember: you I<must C<exec qemu>> as
1261 the last command in the shell script (so that qemu replaces the shell
1262 and becomes the direct child of the libguestfs-using program). If you
1263 don't do this, then the qemu process won't be cleaned up correctly.
1265 Here is an example of a wrapper, where I have built my own copy of
1269 qemudir=/home/rjones/d/qemu
1270 exec $qemudir/x86_64-softmmu/qemu-system-x86_64 -L $qemudir/pc-bios "$@"
1272 Save this script as C</tmp/qemu.wrapper> (or wherever), C<chmod +x>,
1273 and then use it by setting the LIBGUESTFS_QEMU environment variable.
1276 LIBGUESTFS_QEMU=/tmp/qemu.wrapper guestfish
1278 Note that libguestfs also calls qemu with the -help and -version
1279 options in order to determine features.
1281 =head1 ENVIRONMENT VARIABLES
1285 =item LIBGUESTFS_APPEND
1287 Pass additional options to the guest kernel.
1289 =item LIBGUESTFS_DEBUG
1291 Set C<LIBGUESTFS_DEBUG=1> to enable verbose messages. This
1292 has the same effect as calling C<guestfs_set_verbose (handle, 1)>.
1294 =item LIBGUESTFS_MEMSIZE
1296 Set the memory allocated to the qemu process, in megabytes. For
1299 LIBGUESTFS_MEMSIZE=700
1301 =item LIBGUESTFS_PATH
1303 Set the path that libguestfs uses to search for kernel and initrd.img.
1304 See the discussion of paths in section PATH above.
1306 =item LIBGUESTFS_QEMU
1308 Set the default qemu binary that libguestfs uses. If not set, then
1309 the qemu which was found at compile time by the configure script is
1312 See also L</QEMU WRAPPERS> above.
1314 =item LIBGUESTFS_TRACE
1316 Set C<LIBGUESTFS_TRACE=1> to enable command traces. This
1317 has the same effect as calling C<guestfs_set_trace (handle, 1)>.
1321 Location of temporary directory, defaults to C</tmp>.
1323 If libguestfs was compiled to use the supermin appliance then each
1324 handle will require rather a large amount of space in this directory
1325 for short periods of time (~ 80 MB). You can use C<$TMPDIR> to
1326 configure another directory to use in case C</tmp> is not large
1338 L<virt-inspector(1)>,
1339 L<virt-list-filesystems(1)>,
1340 L<virt-list-partitions(1)>,
1349 L<http://libguestfs.org/>.
1351 Tools with a similar purpose:
1360 To get a list of bugs against libguestfs use this link:
1362 L<https://bugzilla.redhat.com/buglist.cgi?component=libguestfs&product=Virtualization+Tools>
1364 To report a new bug against libguestfs use this link:
1366 L<https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virtualization+Tools>
1368 When reporting a bug, please check:
1374 That the bug hasn't been reported already.
1378 That you are testing a recent version.
1382 Describe the bug accurately, and give a way to reproduce it.
1386 Run libguestfs-test-tool and paste the B<complete, unedited>
1387 output into the bug report.
1393 Richard W.M. Jones (C<rjones at redhat dot com>)
1397 Copyright (C) 2009 Red Hat Inc.
1398 L<http://libguestfs.org/>
1400 This library is free software; you can redistribute it and/or
1401 modify it under the terms of the GNU Lesser General Public
1402 License as published by the Free Software Foundation; either
1403 version 2 of the License, or (at your option) any later version.
1405 This library is distributed in the hope that it will be useful,
1406 but WITHOUT ANY WARRANTY; without even the implied warranty of
1407 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
1408 Lesser General Public License for more details.
1410 You should have received a copy of the GNU Lesser General Public
1411 License along with this library; if not, write to the Free Software
1412 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA