5 guestfs - Library for accessing and modifying virtual machine images
11 guestfs_h *g = guestfs_create ();
12 guestfs_add_drive (g, "guest.img");
14 guestfs_mount (g, "/dev/sda1", "/");
15 guestfs_touch (g, "/hello");
16 guestfs_umount (g, "/");
20 cc prog.c -o prog -lguestfs
22 cc prog.c -o prog `pkg-config libguestfs --cflags --libs`
26 Libguestfs is a library for accessing and modifying guest disk images.
27 Amongst the things this is good for: making batch configuration
28 changes to guests, getting disk used/free statistics (see also:
29 virt-df), migrating between virtualization systems (see also:
30 virt-p2v), performing partial backups, performing partial guest
31 clones, cloning guests and changing registry/UUID/hostname info, and
34 Libguestfs uses Linux kernel and qemu code, and can access any type of
35 guest filesystem that Linux and qemu can, including but not limited
36 to: ext2/3/4, btrfs, FAT and NTFS, LVM, many different disk partition
37 schemes, qcow, qcow2, vmdk.
39 Libguestfs provides ways to enumerate guest storage (eg. partitions,
40 LVs, what filesystem is in each LV, etc.). It can also run commands
41 in the context of the guest. Also you can access filesystems over
44 Libguestfs is a library that can be linked with C and C++ management
45 programs (or management programs written in OCaml, Perl, Python, Ruby,
46 Java, 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. By convention we use the single
67 letter C<g> for the name of the handle variable, although of course
68 you can use any name you want.
70 The general structure of all libguestfs-using programs looks like
73 guestfs_h *g = guestfs_create ();
75 /* Call guestfs_add_drive additional times if there are
76 * multiple disk images.
78 guestfs_add_drive (g, "guest.img");
80 /* Most manipulation calls won't work until you've launched
81 * the handle 'g'. You have to do this _after_ adding drives
82 * and _before_ other commands.
86 /* Now you can examine what partitions, LVs etc are available.
88 char **partitions = guestfs_list_partitions (g);
89 char **logvols = guestfs_lvs (g);
91 /* To access a filesystem in the image, you must mount it.
93 guestfs_mount (g, "/dev/sda1", "/");
95 /* Now you can perform filesystem actions on the guest
98 guestfs_touch (g, "/hello");
100 /* You only need to call guestfs_sync if you have made
101 * changes to the guest image. (But if you've made changes
102 * then you *must* sync).
106 /* Close the handle 'g'. */
109 The code above doesn't include any error checking. In real code you
110 should check return values carefully for errors. In general all
111 functions that return integers return C<-1> on error, and all
112 functions that return pointers return C<NULL> on error. See section
113 L</ERROR HANDLING> below for how to handle errors, and consult the
114 documentation for each function call below to see precisely how they
115 return error indications.
119 The image filename (C<"guest.img"> in the example above) could be a
120 disk image from a virtual machine, a L<dd(1)> copy of a physical hard
121 disk, an actual block device, or simply an empty file of zeroes that
122 you have created through L<posix_fallocate(3)>. Libguestfs lets you
123 do useful things to all of these.
125 You can add a disk read-only using C<guestfs_add_drive_ro>, in which
126 case libguestfs won't modify the file.
128 Be extremely cautious if the disk image is in use, eg. if it is being
129 used by a virtual machine. Adding it read-write will almost certainly
130 cause disk corruption, but adding it read-only is safe.
132 You must add at least one disk image, and you may add multiple disk
133 images. In the API, the disk images are usually referred to as
134 C</dev/sda> (for the first one you added), C</dev/sdb> (for the second
137 Once C<guestfs_launch> has been called you cannot add any more images.
138 You can call C<guestfs_list_devices> to get a list of the device
139 names, in the order that you added them. See also L</BLOCK DEVICE
144 Before you can read or write files, create directories and so on in a
145 disk image that contains filesystems, you have to mount those
146 filesystems using C<guestfs_mount>. If you already know that a disk
147 image contains (for example) one partition with a filesystem on that
148 partition, then you can mount it directly:
150 guestfs_mount (g, "/dev/sda1", "/");
152 where C</dev/sda1> means literally the first partition (C<1>) of the
153 first disk image that we added (C</dev/sda>). If the disk contains
154 Linux LVM2 logical volumes you could refer to those instead (eg. C</dev/VG/LV>).
156 If you are given a disk image and you don't know what it contains then
157 you have to find out. Libguestfs can do that too: use
158 C<guestfs_list_partitions> and C<guestfs_lvs> to list possible
159 partitions and LVs, and either try mounting each to see what is
160 mountable, or else examine them with C<guestfs_file>. But you might
161 find it easier to look at higher level programs built on top of
162 libguestfs, in particular L<virt-inspector(1)>.
164 To mount a disk image read-only, use C<guestfs_mount_ro>. There are
165 several other variations of the C<guestfs_mount_*> call.
167 =head2 FILESYSTEM ACCESS AND MODIFICATION
169 The majority of the libguestfs API consists of fairly low-level calls
170 for accessing and modifying the files, directories, symlinks etc on
171 mounted filesystems. There are over a hundred such calls which you
172 can find listed in detail below in this man page, and we don't even
173 pretend to cover them all in this overview.
175 Specify filenames as full paths including the mount point.
177 For example, if you mounted a filesystem at C<"/"> and you want to
178 read the file called C<"etc/passwd"> then you could do:
180 char *data = guestfs_cat (g, "/etc/passwd");
182 This would return C<data> as a newly allocated buffer containing the
183 full content of that file (with some conditions: see also
184 L</DOWNLOADING> below), or C<NULL> if there was an error.
186 As another example, to create a top-level directory on that filesystem
187 called C<"var"> you would do:
189 guestfs_mkdir (g, "/var");
191 To create a symlink you could do:
193 guestfs_ln_s (g, "/etc/init.d/portmap",
194 "/etc/rc3.d/S30portmap");
196 Libguestfs will reject attempts to use relative paths. There is no
197 concept of a current working directory. Libguestfs can return errors
198 in many situations: for example if the filesystem isn't writable, or
199 if a file or directory that you requested doesn't exist. If you are
200 using the C API (documented here) you have to check for those error
201 conditions after each call. (Other language bindings turn these
202 errors into exceptions).
204 File writes are affected by the per-handle umask, set by calling
205 C<guestfs_umask> and defaulting to 022.
209 Libguestfs contains API calls to read, create and modify partition
210 tables on disk images.
212 In the common case where you want to create a single partition
213 covering the whole disk, you should use the C<guestfs_part_disk>
216 const char *parttype = "mbr";
217 if (disk_is_larger_than_2TB)
219 guestfs_part_disk (g, "/dev/sda", parttype);
221 Obviously this effectively wipes anything that was on that disk image
224 In general MBR partitions are both unnecessarily complicated and
225 depend on archaic details, namely the Cylinder-Head-Sector (CHS)
226 geometry of the disk. C<guestfs_sfdiskM> can be used to
227 create more complex arrangements where the relative sizes are
228 expressed in megabytes instead of cylinders, which is a small win.
229 C<guestfs_sfdiskM> will choose the nearest cylinder to approximate the
230 requested size. There's a lot of crazy stuff to do with IDE and
231 virtio disks having different, incompatible CHS geometries, that you
232 probably don't want to know about.
234 My advice: make a single partition to cover the whole disk, then use
239 Libguestfs provides access to a large part of the LVM2 API, such as
240 C<guestfs_lvcreate> and C<guestfs_vgremove>. It won't make much sense
241 unless you familiarize yourself with the concepts of physical volumes,
242 volume groups and logical volumes.
244 This author strongly recommends reading the LVM HOWTO, online at
245 L<http://tldp.org/HOWTO/LVM-HOWTO/>.
249 Use C<guestfs_cat> to download small, text only files. This call
250 is limited to files which are less than 2 MB and which cannot contain
251 any ASCII NUL (C<\0>) characters. However it has a very simple
254 C<guestfs_read_file> can be used to read files which contain
255 arbitrary 8 bit data, since it returns a (pointer, size) pair.
256 However it is still limited to "small" files, less than 2 MB.
258 C<guestfs_download> can be used to download any file, with no
259 limits on content or size (even files larger than 4 GB).
261 To download multiple files, see C<guestfs_tar_out> and
266 It's often the case that you want to write a file or files to the disk
269 For small, single files, use C<guestfs_write_file>. This call
270 currently contains a bug which limits the call to plain text files
271 (not containing ASCII NUL characters).
273 To upload a single file, use C<guestfs_upload>. This call has no
274 limits on file content or size (even files larger than 4 GB).
276 To upload multiple files, see C<guestfs_tar_in> and C<guestfs_tgz_in>.
278 However the fastest way to upload I<large numbers of arbitrary files>
279 is to turn them into a squashfs or CD ISO (see L<mksquashfs(8)> and
280 L<mkisofs(8)>), then attach this using C<guestfs_add_drive_ro>. If
281 you add the drive in a predictable way (eg. adding it last after all
282 other drives) then you can get the device name from
283 C<guestfs_list_devices> and mount it directly using
284 C<guestfs_mount_ro>. Note that squashfs images are sometimes
285 non-portable between kernel versions, and they don't support labels or
286 UUIDs. If you want to pre-build an image or you need to mount it
287 using a label or UUID, use an ISO image instead.
291 There are various different commands for copying between files and
292 devices and in and out of the guest filesystem. These are summarised
297 =item B<file> to B<file>
299 Use L</guestfs_cp> to copy a single file, or
300 L</guestfs_cp_a> to copy directories recursively.
302 =item B<file or device> to B<file or device>
304 Use L</guestfs_dd> which efficiently uses L<dd(1)>
305 to copy between files and devices in the guest.
307 Example: duplicate the contents of an LV:
309 guestfs_dd (g, "/dev/VG/Original", "/dev/VG/Copy");
311 The destination (C</dev/VG/Copy>) must be at least as large as the
312 source (C</dev/VG/Original>).
314 =item B<file on the host> to B<file or device>
316 Use L</guestfs_upload>. See L</UPLOADING> above.
318 =item B<file or device> to B<file on the host>
320 Use L</guestfs_download>. See L</DOWNLOADING> above.
326 C<guestfs_ll> is just designed for humans to read (mainly when using
327 the L<guestfish(1)>-equivalent command C<ll>).
329 C<guestfs_ls> is a quick way to get a list of files in a directory
330 from programs, as a flat list of strings.
332 C<guestfs_readdir> is a programmatic way to get a list of files in a
333 directory, plus additional information about each one. It is more
334 equivalent to using the L<readdir(3)> call on a local filesystem.
336 C<guestfs_find> can be used to recursively list files.
338 =head2 RUNNING COMMANDS
340 Although libguestfs is a primarily an API for manipulating files
341 inside guest images, we also provide some limited facilities for
342 running commands inside guests.
344 There are many limitations to this:
350 The kernel version that the command runs under will be different
351 from what it expects.
355 If the command needs to communicate with daemons, then most likely
356 they won't be running.
360 The command will be running in limited memory.
364 Only supports Linux guests (not Windows, BSD, etc).
368 Architecture limitations (eg. won't work for a PPC guest on
373 For SELinux guests, you may need to enable SELinux and load policy
374 first. See L</SELINUX> in this manpage.
378 The two main API calls to run commands are C<guestfs_command> and
379 C<guestfs_sh> (there are also variations).
381 The difference is that C<guestfs_sh> runs commands using the shell, so
382 any shell globs, redirections, etc will work.
384 =head2 CONFIGURATION FILES
386 To read and write configuration files in Linux guest filesystems, we
387 strongly recommend using Augeas. For example, Augeas understands how
388 to read and write, say, a Linux shadow password file or X.org
389 configuration file, and so avoids you having to write that code.
391 The main Augeas calls are bound through the C<guestfs_aug_*> APIs. We
392 don't document Augeas itself here because there is excellent
393 documentation on the L<http://augeas.net/> website.
395 If you don't want to use Augeas (you fool!) then try calling
396 C<guestfs_read_lines> to get the file as a list of lines which
397 you can iterate over.
401 We support SELinux guests. To ensure that labeling happens correctly
402 in SELinux guests, you need to enable SELinux and load the guest's
409 Before launching, do:
411 guestfs_set_selinux (g, 1);
415 After mounting the guest's filesystem(s), load the policy. This
416 is best done by running the L<load_policy(8)> command in the
419 guestfs_sh (g, "/usr/sbin/load_policy");
421 (Older versions of C<load_policy> require you to specify the
422 name of the policy file).
426 Optionally, set the security context for the API. The correct
427 security context to use can only be known by inspecting the
428 guest. As an example:
430 guestfs_setcon (g, "unconfined_u:unconfined_r:unconfined_t:s0");
434 This will work for running commands and editing existing files.
436 When new files are created, you may need to label them explicitly,
437 for example by running the external command
438 C<restorecon pathname>.
440 =head2 SPECIAL CONSIDERATIONS FOR WINDOWS GUESTS
442 Libguestfs can mount NTFS partitions. It does this using the
443 L<http://www.ntfs-3g.org/> driver.
445 DOS and Windows still use drive letters, and the filesystems are
446 always treated as case insensitive by Windows itself, and therefore
447 you might find a Windows configuration file referring to a path like
448 C<c:\windows\system32>. When the filesystem is mounted in libguestfs,
449 that directory might be referred to as C</WINDOWS/System32>.
451 Drive letter mappings are outside the scope of libguestfs. You have
452 to use libguestfs to read the appropriate Windows Registry and
453 configuration files, to determine yourself how drives are mapped (see
454 also L<virt-inspector(1)>).
456 Replacing backslash characters with forward slash characters is also
457 outside the scope of libguestfs, but something that you can easily do.
459 Where we can help is in resolving the case insensitivity of paths.
460 For this, call C<guestfs_case_sensitive_path>.
462 Libguestfs also provides some help for decoding Windows Registry
463 "hive" files, through the library C<hivex> which is part of the
464 libguestfs project. You have to locate and download the hive file(s)
465 yourself, and then pass them to C<hivex> functions. See also the
466 programs L<hivexml(1)>, L<hivexsh(1)> and L<virt-win-reg(1)> for more
469 =head2 USING LIBGUESTFS WITH OTHER PROGRAMMING LANGUAGES
471 Although we don't want to discourage you from using the C API, we will
472 mention here that the same API is also available in other languages.
474 The API is broadly identical in all supported languages. This means
475 that the C call C<guestfs_mount(g,path)> is
476 C<$g-E<gt>mount($path)> in Perl, C<g.mount(path)> in Python,
477 and C<Guestfs.mount g path> in OCaml. In other words, a
478 straightforward, predictable isomorphism between each language.
480 Error messages are automatically transformed
481 into exceptions if the language supports it.
483 We don't try to "object orientify" parts of the API in OO languages,
484 although contributors are welcome to write higher level APIs above
485 what we provide in their favourite languages if they wish.
491 You can use the I<guestfs.h> header file from C++ programs. The C++
492 API is identical to the C API. C++ classes and exceptions are
497 The C# bindings are highly experimental. Please read the warnings
498 at the top of C<csharp/Libguestfs.cs>.
502 This is the only language binding that working but incomplete. Only
503 calls which return simple integers have been bound in Haskell, and we
504 are looking for help to complete this binding.
508 Full documentation is contained in the Javadoc which is distributed
513 For documentation see the file C<guestfs.mli>.
517 For documentation see L<Sys::Guestfs(3)>.
521 For documentation do:
529 Use the Guestfs module. There is no Ruby-specific documentation, but
530 you can find examples written in Ruby in the libguestfs source.
532 =item B<shell scripts>
534 For documentation see L<guestfish(1)>.
538 =head2 LIBGUESTFS GOTCHAS
540 L<http://en.wikipedia.org/wiki/Gotcha_(programming)>: "A feature of a
541 system [...] that works in the way it is documented but is
542 counterintuitive and almost invites mistakes."
544 Since we developed libguestfs and the associated tools, there are
545 several things we would have designed differently, but are now stuck
546 with for backwards compatibility or other reasons. If there is ever a
547 libguestfs 2.0 release, you can expect these to change. Beware of
552 =item Autosync / forgetting to sync.
554 When modifying a filesystem from C or another language, you B<must>
555 unmount all filesystems and call L</guestfs_sync> explicitly before
556 you close the libguestfs handle. You can also call:
558 guestfs_set_autosync (g, 1);
560 to have the unmount/sync done automatically for you when the handle 'g'
561 is closed. (This feature is called "autosync", L</guestfs_set_autosync>
564 If you forget to do this, then it is entirely possible that your
565 changes won't be written out, or will be partially written, or (very
566 rarely) that you'll get disk corruption.
568 Note that in L<guestfish(3)> I<autosync is the default>. So quick and
569 dirty guestfish scripts that forget to sync will work just fine, which
570 can make this extra-puzzling if you are trying to debug a problem.
572 =item Mount option C<-o sync> should not be the default.
574 If you use C<guestfs_mount>, then C<-o sync,noatime> are added
575 implicitly. However C<-o sync> does not add any reliability benefit,
576 but does have a very large performance impact.
578 The work around is to use C<guestfs_mount_options> and set the mount
579 options that you actually want to use.
581 =item Read-only should be the default.
583 In L<guestfish(3)>, I<--ro> should be the default, and you should
584 have to specify I<--rw> if you want to make changes to the image.
586 This would reduce the potential to corrupt live VM images.
588 Note that many filesystems change the disk when you just mount and
589 unmount, even if you didn't perform any writes. You need to use
590 C<guestfs_add_drive_ro> to guarantee that the disk is not changed.
592 =item guestfish command line is hard to use.
594 C<guestfish disk.img> doesn't do what people expect (open C<disk.img>
595 for examination). It tries to run a guestfish command C<disk.img>
596 which doesn't exist, so it fails, and it fails with a strange and
597 unintuitive error message. Like the Bourne shell, we should have used
598 C<guestfish -c command> to run commands.
602 =head2 PROTOCOL LIMITS
604 Internally libguestfs uses a message-based protocol to pass API calls
605 and their responses to and from a small "appliance" (see L</INTERNALS>
606 for plenty more detail about this). The maximum message size used by
607 the protocol is slightly less than 4 MB. For some API calls you may
608 need to be aware of this limit. The API calls which may be affected
609 are individually documented, with a link back to this section of the
612 A simple call such as C<guestfs_cat> returns its result (the file
613 data) in a simple string. Because this string is at some point
614 internally encoded as a message, the maximum size that it can return
615 is slightly under 4 MB. If the requested file is larger than this
616 then you will get an error.
618 In order to transfer large files into and out of the guest filesystem,
619 you need to use particular calls that support this. The sections
620 L</UPLOADING> and L</DOWNLOADING> document how to do this.
622 You might also consider mounting the disk image using our FUSE
623 filesystem support (L<guestmount(1)>).
625 =head1 CONNECTION MANAGEMENT
629 C<guestfs_h> is the opaque type representing a connection handle.
630 Create a handle by calling C<guestfs_create>. Call C<guestfs_close>
631 to free the handle and release all resources used.
633 For information on using multiple handles and threads, see the section
634 L</MULTIPLE HANDLES AND MULTIPLE THREADS> below.
636 =head2 guestfs_create
638 guestfs_h *guestfs_create (void);
640 Create a connection handle.
642 You have to call C<guestfs_add_drive> on the handle at least once.
644 This function returns a non-NULL pointer to a handle on success or
647 After configuring the handle, you have to call C<guestfs_launch>.
649 You may also want to configure error handling for the handle. See
650 L</ERROR HANDLING> section below.
654 void guestfs_close (guestfs_h *g);
656 This closes the connection handle and frees up all resources used.
658 =head1 ERROR HANDLING
660 The convention in all functions that return C<int> is that they return
661 C<-1> to indicate an error. You can get additional information on
662 errors by calling C<guestfs_last_error> and/or by setting up an error
663 handler with C<guestfs_set_error_handler>.
665 The default error handler prints the information string to C<stderr>.
667 Out of memory errors are handled differently. The default action is
668 to call L<abort(3)>. If this is undesirable, then you can set a
669 handler using C<guestfs_set_out_of_memory_handler>.
671 =head2 guestfs_last_error
673 const char *guestfs_last_error (guestfs_h *g);
675 This returns the last error message that happened on C<g>. If
676 there has not been an error since the handle was created, then this
679 The lifetime of the returned string is until the next error occurs, or
680 C<guestfs_close> is called.
682 The error string is not localized (ie. is always in English), because
683 this makes searching for error messages in search engines give the
684 largest number of results.
686 =head2 guestfs_set_error_handler
688 typedef void (*guestfs_error_handler_cb) (guestfs_h *g,
691 void guestfs_set_error_handler (guestfs_h *g,
692 guestfs_error_handler_cb cb,
695 The callback C<cb> will be called if there is an error. The
696 parameters passed to the callback are an opaque data pointer and the
697 error message string.
699 Note that the message string C<msg> is freed as soon as the callback
700 function returns, so if you want to stash it somewhere you must make
703 The default handler prints messages on C<stderr>.
705 If you set C<cb> to C<NULL> then I<no> handler is called.
707 =head2 guestfs_get_error_handler
709 guestfs_error_handler_cb guestfs_get_error_handler (guestfs_h *g,
712 Returns the current error handler callback.
714 =head2 guestfs_set_out_of_memory_handler
716 typedef void (*guestfs_abort_cb) (void);
717 int guestfs_set_out_of_memory_handler (guestfs_h *g,
720 The callback C<cb> will be called if there is an out of memory
721 situation. I<Note this callback must not return>.
723 The default is to call L<abort(3)>.
725 You cannot set C<cb> to C<NULL>. You can't ignore out of memory
728 =head2 guestfs_get_out_of_memory_handler
730 guestfs_abort_fn guestfs_get_out_of_memory_handler (guestfs_h *g);
732 This returns the current out of memory handler.
736 Libguestfs needs a kernel and initrd.img, which it finds by looking
737 along an internal path.
739 By default it looks for these in the directory C<$libdir/guestfs>
740 (eg. C</usr/local/lib/guestfs> or C</usr/lib64/guestfs>).
742 Use C<guestfs_set_path> or set the environment variable
743 C<LIBGUESTFS_PATH> to change the directories that libguestfs will
744 search in. The value is a colon-separated list of paths. The current
745 directory is I<not> searched unless the path contains an empty element
746 or C<.>. For example C<LIBGUESTFS_PATH=:/usr/lib/guestfs> would
747 search the current directory and then C</usr/lib/guestfs>.
749 =head1 HIGH-LEVEL API ACTIONS
753 We guarantee the libguestfs ABI (binary interface), for public,
754 high-level actions as outlined in this section. Although we will
755 deprecate some actions, for example if they get replaced by newer
756 calls, we will keep the old actions forever. This allows you the
757 developer to program in confidence against libguestfs.
767 =head2 GROUPS OF FUNCTIONALITY IN THE APPLIANCE
769 Using L</guestfs_available> you can test availability of
770 the following groups of functions. This test queries the
771 appliance to see if the appliance you are currently using
772 supports the functionality.
776 =head2 SINGLE CALLS AT COMPILE TIME
778 If you need to test whether a single libguestfs function is
779 available at compile time, we recommend using build tools
780 such as autoconf or cmake. For example in autotools you could
783 AC_CHECK_LIB([guestfs],[guestfs_create])
784 AC_CHECK_FUNCS([guestfs_dd])
786 which would result in C<HAVE_GUESTFS_DD> being either defined
787 or not defined in your program.
789 =head2 SINGLE CALLS AT RUN TIME
791 Testing at compile time doesn't guarantee that a function really
792 exists in the library. The reason is that you might be dynamically
793 linked against a previous I<libguestfs.so> (dynamic library)
794 which doesn't have the call. This situation unfortunately results
795 in a segmentation fault, which is a shortcoming of the C dynamic
796 linking system itself.
798 You can use L<dlopen(3)> to test if a function is available
799 at run time, as in this example program (note that you still
800 need the compile time check as well):
812 #ifdef HAVE_GUESTFS_DD
816 /* Test if the function guestfs_dd is really available. */
817 dl = dlopen (NULL, RTLD_LAZY);
819 fprintf (stderr, "dlopen: %s\n", dlerror ());
822 has_function = dlsym (dl, "guestfs_dd") != NULL;
826 printf ("this libguestfs.so does NOT have guestfs_dd function\n");
828 printf ("this libguestfs.so has guestfs_dd function\n");
829 /* Now it's safe to call
830 guestfs_dd (g, "foo", "bar");
834 printf ("guestfs_dd function was not found at compile time\n");
838 You may think the above is an awful lot of hassle, and it is.
839 There are other ways outside of the C linking system to ensure
840 that this kind of incompatibility never arises, such as using
843 Requires: libguestfs >= 1.0.80
847 <!-- old anchor for the next section -->
848 <a name="state_machine_and_low_level_event_api"/>
854 Internally, libguestfs is implemented by running an appliance (a
855 special type of small virtual machine) using L<qemu(1)>. Qemu runs as
856 a child process of the main program.
862 | | child process / appliance
863 | | __________________________
865 +-------------------+ RPC | +-----------------+ |
866 | libguestfs <--------------------> guestfsd | |
867 | | | +-----------------+ |
868 \___________________/ | | Linux kernel | |
869 | +--^--------------+ |
870 \_________|________________/
878 The library, linked to the main program, creates the child process and
879 hence the appliance in the L</guestfs_launch> function.
881 Inside the appliance is a Linux kernel and a complete stack of
882 userspace tools (such as LVM and ext2 programs) and a small
883 controlling daemon called C<guestfsd>. The library talks to
884 C<guestfsd> using remote procedure calls (RPC). There is a mostly
885 one-to-one correspondence between libguestfs API calls and RPC calls
886 to the daemon. Lastly the disk image(s) are attached to the qemu
887 process which translates device access by the appliance's Linux kernel
888 into accesses to the image.
890 A common misunderstanding is that the appliance "is" the virtual
891 machine. Although the disk image you are attached to might also be
892 used by some virtual machine, libguestfs doesn't know or care about
893 this. (But you will care if both libguestfs's qemu process and your
894 virtual machine are trying to update the disk image at the same time,
895 since these usually results in massive disk corruption).
899 libguestfs uses a state machine to model the child process:
910 / | \ \ guestfs_launch
921 \______/ <------ \________/
923 The normal transitions are (1) CONFIG (when the handle is created, but
924 there is no child process), (2) LAUNCHING (when the child process is
925 booting up), (3) alternating between READY and BUSY as commands are
926 issued to, and carried out by, the child process.
928 The guest may be killed by C<guestfs_kill_subprocess>, or may die
929 asynchronously at any time (eg. due to some internal error), and that
930 causes the state to transition back to CONFIG.
932 Configuration commands for qemu such as C<guestfs_add_drive> can only
933 be issued when in the CONFIG state.
935 The high-level API offers two calls that go from CONFIG through
936 LAUNCHING to READY. C<guestfs_launch> blocks until the child process
937 is READY to accept commands (or until some failure or timeout).
938 C<guestfs_launch> internally moves the state from CONFIG to LAUNCHING
941 High-level API actions such as C<guestfs_mount> can only be issued
942 when in the READY state. These high-level API calls block waiting for
943 the command to be carried out (ie. the state to transition to BUSY and
944 then back to READY). But using the low-level event API, you get
945 non-blocking versions. (But you can still only carry out one
946 operation per handle at a time - that is a limitation of the
947 communications protocol we use).
949 Finally, the child process sends asynchronous messages back to the
950 main program, such as kernel log messages. Mostly these are ignored
951 by the high-level API, but using the low-level event API you can
952 register to receive these messages.
954 =head2 SETTING CALLBACKS TO HANDLE EVENTS
956 The child process generates events in some situations. Current events
957 include: receiving a log message, the child process exits.
959 Use the C<guestfs_set_*_callback> functions to set a callback for
960 different types of events.
962 Only I<one callback of each type> can be registered for each handle.
963 Calling C<guestfs_set_*_callback> again overwrites the previous
964 callback of that type. Cancel all callbacks of this type by calling
965 this function with C<cb> set to C<NULL>.
967 =head2 guestfs_set_log_message_callback
969 typedef void (*guestfs_log_message_cb) (guestfs_h *g, void *opaque,
971 void guestfs_set_log_message_callback (guestfs_h *g,
972 guestfs_log_message_cb cb,
975 The callback function C<cb> will be called whenever qemu or the guest
976 writes anything to the console.
978 Use this function to capture kernel messages and similar.
980 Normally there is no log message handler, and log messages are just
983 =head2 guestfs_set_subprocess_quit_callback
985 typedef void (*guestfs_subprocess_quit_cb) (guestfs_h *g, void *opaque);
986 void guestfs_set_subprocess_quit_callback (guestfs_h *g,
987 guestfs_subprocess_quit_cb cb,
990 The callback function C<cb> will be called when the child process
991 quits, either asynchronously or if killed by
992 C<guestfs_kill_subprocess>. (This corresponds to a transition from
993 any state to the CONFIG state).
995 =head2 guestfs_set_launch_done_callback
997 typedef void (*guestfs_launch_done_cb) (guestfs_h *g, void *opaque);
998 void guestfs_set_launch_done_callback (guestfs_h *g,
1002 The callback function C<cb> will be called when the child process
1003 becomes ready first time after it has been launched. (This
1004 corresponds to a transition from LAUNCHING to the READY state).
1006 =head1 BLOCK DEVICE NAMING
1008 In the kernel there is now quite a profusion of schemata for naming
1009 block devices (in this context, by I<block device> I mean a physical
1010 or virtual hard drive). The original Linux IDE driver used names
1011 starting with C</dev/hd*>. SCSI devices have historically used a
1012 different naming scheme, C</dev/sd*>. When the Linux kernel I<libata>
1013 driver became a popular replacement for the old IDE driver
1014 (particularly for SATA devices) those devices also used the
1015 C</dev/sd*> scheme. Additionally we now have virtual machines with
1016 paravirtualized drivers. This has created several different naming
1017 systems, such as C</dev/vd*> for virtio disks and C</dev/xvd*> for Xen
1020 As discussed above, libguestfs uses a qemu appliance running an
1021 embedded Linux kernel to access block devices. We can run a variety
1022 of appliances based on a variety of Linux kernels.
1024 This causes a problem for libguestfs because many API calls use device
1025 or partition names. Working scripts and the recipe (example) scripts
1026 that we make available over the internet could fail if the naming
1029 Therefore libguestfs defines C</dev/sd*> as the I<standard naming
1030 scheme>. Internally C</dev/sd*> names are translated, if necessary,
1031 to other names as required. For example, under RHEL 5 which uses the
1032 C</dev/hd*> scheme, any device parameter C</dev/sda2> is translated to
1033 C</dev/hda2> transparently.
1035 Note that this I<only> applies to parameters. The
1036 C<guestfs_list_devices>, C<guestfs_list_partitions> and similar calls
1037 return the true names of the devices and partitions as known to the
1040 =head2 ALGORITHM FOR BLOCK DEVICE NAME TRANSLATION
1042 Usually this translation is transparent. However in some (very rare)
1043 cases you may need to know the exact algorithm. Such cases include
1044 where you use C<guestfs_config> to add a mixture of virtio and IDE
1045 devices to the qemu-based appliance, so have a mixture of C</dev/sd*>
1046 and C</dev/vd*> devices.
1048 The algorithm is applied only to I<parameters> which are known to be
1049 either device or partition names. Return values from functions such
1050 as C<guestfs_list_devices> are never changed.
1056 Is the string a parameter which is a device or partition name?
1060 Does the string begin with C</dev/sd>?
1064 Does the named device exist? If so, we use that device.
1065 However if I<not> then we continue with this algorithm.
1069 Replace initial C</dev/sd> string with C</dev/hd>.
1071 For example, change C</dev/sda2> to C</dev/hda2>.
1073 If that named device exists, use it. If not, continue.
1077 Replace initial C</dev/sd> string with C</dev/vd>.
1079 If that named device exists, use it. If not, return an error.
1083 =head2 PORTABILITY CONCERNS
1085 Although the standard naming scheme and automatic translation is
1086 useful for simple programs and guestfish scripts, for larger programs
1087 it is best not to rely on this mechanism.
1089 Where possible for maximum future portability programs using
1090 libguestfs should use these future-proof techniques:
1096 Use C<guestfs_list_devices> or C<guestfs_list_partitions> to list
1097 actual device names, and then use those names directly.
1099 Since those device names exist by definition, they will never be
1104 Use higher level ways to identify filesystems, such as LVM names,
1105 UUIDs and filesystem labels.
1111 =head2 COMMUNICATION PROTOCOL
1113 Don't rely on using this protocol directly. This section documents
1114 how it currently works, but it may change at any time.
1116 The protocol used to talk between the library and the daemon running
1117 inside the qemu virtual machine is a simple RPC mechanism built on top
1118 of XDR (RFC 1014, RFC 1832, RFC 4506).
1120 The detailed format of structures is in C<src/guestfs_protocol.x>
1121 (note: this file is automatically generated).
1123 There are two broad cases, ordinary functions that don't have any
1124 C<FileIn> and C<FileOut> parameters, which are handled with very
1125 simple request/reply messages. Then there are functions that have any
1126 C<FileIn> or C<FileOut> parameters, which use the same request and
1127 reply messages, but they may also be followed by files sent using a
1130 =head3 ORDINARY FUNCTIONS (NO FILEIN/FILEOUT PARAMS)
1132 For ordinary functions, the request message is:
1134 total length (header + arguments,
1135 but not including the length word itself)
1136 struct guestfs_message_header (encoded as XDR)
1137 struct guestfs_<foo>_args (encoded as XDR)
1139 The total length field allows the daemon to allocate a fixed size
1140 buffer into which it slurps the rest of the message. As a result, the
1141 total length is limited to C<GUESTFS_MESSAGE_MAX> bytes (currently
1142 4MB), which means the effective size of any request is limited to
1143 somewhere under this size.
1145 Note also that many functions don't take any arguments, in which case
1146 the C<guestfs_I<foo>_args> is completely omitted.
1148 The header contains the procedure number (C<guestfs_proc>) which is
1149 how the receiver knows what type of args structure to expect, or none
1152 The reply message for ordinary functions is:
1154 total length (header + ret,
1155 but not including the length word itself)
1156 struct guestfs_message_header (encoded as XDR)
1157 struct guestfs_<foo>_ret (encoded as XDR)
1159 As above the C<guestfs_I<foo>_ret> structure may be completely omitted
1160 for functions that return no formal return values.
1162 As above the total length of the reply is limited to
1163 C<GUESTFS_MESSAGE_MAX>.
1165 In the case of an error, a flag is set in the header, and the reply
1166 message is slightly changed:
1168 total length (header + error,
1169 but not including the length word itself)
1170 struct guestfs_message_header (encoded as XDR)
1171 struct guestfs_message_error (encoded as XDR)
1173 The C<guestfs_message_error> structure contains the error message as a
1176 =head3 FUNCTIONS THAT HAVE FILEIN PARAMETERS
1178 A C<FileIn> parameter indicates that we transfer a file I<into> the
1179 guest. The normal request message is sent (see above). However this
1180 is followed by a sequence of file chunks.
1182 total length (header + arguments,
1183 but not including the length word itself,
1184 and not including the chunks)
1185 struct guestfs_message_header (encoded as XDR)
1186 struct guestfs_<foo>_args (encoded as XDR)
1187 sequence of chunks for FileIn param #0
1188 sequence of chunks for FileIn param #1 etc.
1190 The "sequence of chunks" is:
1192 length of chunk (not including length word itself)
1193 struct guestfs_chunk (encoded as XDR)
1195 struct guestfs_chunk (encoded as XDR)
1198 struct guestfs_chunk (with data.data_len == 0)
1200 The final chunk has the C<data_len> field set to zero. Additionally a
1201 flag is set in the final chunk to indicate either successful
1202 completion or early cancellation.
1204 At time of writing there are no functions that have more than one
1205 FileIn parameter. However this is (theoretically) supported, by
1206 sending the sequence of chunks for each FileIn parameter one after
1207 another (from left to right).
1209 Both the library (sender) I<and> the daemon (receiver) may cancel the
1210 transfer. The library does this by sending a chunk with a special
1211 flag set to indicate cancellation. When the daemon sees this, it
1212 cancels the whole RPC, does I<not> send any reply, and goes back to
1213 reading the next request.
1215 The daemon may also cancel. It does this by writing a special word
1216 C<GUESTFS_CANCEL_FLAG> to the socket. The library listens for this
1217 during the transfer, and if it gets it, it will cancel the transfer
1218 (it sends a cancel chunk). The special word is chosen so that even if
1219 cancellation happens right at the end of the transfer (after the
1220 library has finished writing and has started listening for the reply),
1221 the "spurious" cancel flag will not be confused with the reply
1224 This protocol allows the transfer of arbitrary sized files (no 32 bit
1225 limit), and also files where the size is not known in advance
1226 (eg. from pipes or sockets). However the chunks are rather small
1227 (C<GUESTFS_MAX_CHUNK_SIZE>), so that neither the library nor the
1228 daemon need to keep much in memory.
1230 =head3 FUNCTIONS THAT HAVE FILEOUT PARAMETERS
1232 The protocol for FileOut parameters is exactly the same as for FileIn
1233 parameters, but with the roles of daemon and library reversed.
1235 total length (header + ret,
1236 but not including the length word itself,
1237 and not including the chunks)
1238 struct guestfs_message_header (encoded as XDR)
1239 struct guestfs_<foo>_ret (encoded as XDR)
1240 sequence of chunks for FileOut param #0
1241 sequence of chunks for FileOut param #1 etc.
1243 =head3 INITIAL MESSAGE
1245 Because the underlying channel (QEmu -net channel) doesn't have any
1246 sort of connection control, when the daemon launches it sends an
1247 initial word (C<GUESTFS_LAUNCH_FLAG>) which indicates that the guest
1248 and daemon is alive. This is what C<guestfs_launch> waits for.
1250 =head1 MULTIPLE HANDLES AND MULTIPLE THREADS
1252 All high-level libguestfs actions are synchronous. If you want
1253 to use libguestfs asynchronously then you must create a thread.
1255 Only use the handle from a single thread. Either use the handle
1256 exclusively from one thread, or provide your own mutex so that two
1257 threads cannot issue calls on the same handle at the same time.
1259 =head1 QEMU WRAPPERS
1261 If you want to compile your own qemu, run qemu from a non-standard
1262 location, or pass extra arguments to qemu, then you can write a
1263 shell-script wrapper around qemu.
1265 There is one important rule to remember: you I<must C<exec qemu>> as
1266 the last command in the shell script (so that qemu replaces the shell
1267 and becomes the direct child of the libguestfs-using program). If you
1268 don't do this, then the qemu process won't be cleaned up correctly.
1270 Here is an example of a wrapper, where I have built my own copy of
1274 qemudir=/home/rjones/d/qemu
1275 exec $qemudir/x86_64-softmmu/qemu-system-x86_64 -L $qemudir/pc-bios "$@"
1277 Save this script as C</tmp/qemu.wrapper> (or wherever), C<chmod +x>,
1278 and then use it by setting the LIBGUESTFS_QEMU environment variable.
1281 LIBGUESTFS_QEMU=/tmp/qemu.wrapper guestfish
1283 Note that libguestfs also calls qemu with the -help and -version
1284 options in order to determine features.
1286 =head1 ENVIRONMENT VARIABLES
1290 =item LIBGUESTFS_APPEND
1292 Pass additional options to the guest kernel.
1294 =item LIBGUESTFS_DEBUG
1296 Set C<LIBGUESTFS_DEBUG=1> to enable verbose messages. This
1297 has the same effect as calling C<guestfs_set_verbose (g, 1)>.
1299 =item LIBGUESTFS_MEMSIZE
1301 Set the memory allocated to the qemu process, in megabytes. For
1304 LIBGUESTFS_MEMSIZE=700
1306 =item LIBGUESTFS_PATH
1308 Set the path that libguestfs uses to search for kernel and initrd.img.
1309 See the discussion of paths in section PATH above.
1311 =item LIBGUESTFS_QEMU
1313 Set the default qemu binary that libguestfs uses. If not set, then
1314 the qemu which was found at compile time by the configure script is
1317 See also L</QEMU WRAPPERS> above.
1319 =item LIBGUESTFS_TRACE
1321 Set C<LIBGUESTFS_TRACE=1> to enable command traces. This
1322 has the same effect as calling C<guestfs_set_trace (g, 1)>.
1326 Location of temporary directory, defaults to C</tmp>.
1328 If libguestfs was compiled to use the supermin appliance then each
1329 handle will require rather a large amount of space in this directory
1330 for short periods of time (~ 80 MB). You can use C<$TMPDIR> to
1331 configure another directory to use in case C</tmp> is not large
1343 L<virt-inspector(1)>,
1344 L<virt-list-filesystems(1)>,
1345 L<virt-list-partitions(1)>,
1354 L<http://libguestfs.org/>.
1356 Tools with a similar purpose:
1365 To get a list of bugs against libguestfs use this link:
1367 L<https://bugzilla.redhat.com/buglist.cgi?component=libguestfs&product=Virtualization+Tools>
1369 To report a new bug against libguestfs use this link:
1371 L<https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virtualization+Tools>
1373 When reporting a bug, please check:
1379 That the bug hasn't been reported already.
1383 That you are testing a recent version.
1387 Describe the bug accurately, and give a way to reproduce it.
1391 Run libguestfs-test-tool and paste the B<complete, unedited>
1392 output into the bug report.
1398 Richard W.M. Jones (C<rjones at redhat dot com>)
1402 Copyright (C) 2009 Red Hat Inc.
1403 L<http://libguestfs.org/>
1405 This library is free software; you can redistribute it and/or
1406 modify it under the terms of the GNU Lesser General Public
1407 License as published by the Free Software Foundation; either
1408 version 2 of the License, or (at your option) any later version.
1410 This library is distributed in the hope that it will be useful,
1411 but WITHOUT ANY WARRANTY; without even the implied warranty of
1412 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
1413 Lesser General Public License for more details.
1415 You should have received a copy of the GNU Lesser General Public
1416 License along with this library; if not, write to the Free Software
1417 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA