5 guestfs - Library for accessing and modifying virtual machine images
11 guestfs_h *handle = guestfs_create ();
12 guestfs_add_drive (handle, "guest.img");
13 guestfs_launch (handle);
14 guestfs_mount (handle, "/dev/sda1", "/");
15 guestfs_touch (handle, "/hello");
16 guestfs_sync (handle);
17 guestfs_close (handle);
21 Libguestfs is a library for accessing and modifying guest disk images.
22 Amongst the things this is good for: making batch configuration
23 changes to guests, getting disk used/free statistics (see also:
24 virt-df), migrating between virtualization systems (see also:
25 virt-p2v), performing partial backups, performing partial guest
26 clones, cloning guests and changing registry/UUID/hostname info, and
29 Libguestfs uses Linux kernel and qemu code, and can access any type of
30 guest filesystem that Linux and qemu can, including but not limited
31 to: ext2/3/4, btrfs, FAT and NTFS, LVM, many different disk partition
32 schemes, qcow, qcow2, vmdk.
34 Libguestfs provides ways to enumerate guest storage (eg. partitions,
35 LVs, what filesystem is in each LV, etc.). It can also run commands
36 in the context of the guest. Also you can access filesystems over FTP.
38 Libguestfs is a library that can be linked with C and C++ management
39 programs (or management programs written in OCaml, Perl, Python, Ruby, Java
40 or Haskell). You can also use it from shell scripts or the command line.
42 You don't need to be root to use libguestfs, although obviously you do
43 need enough permissions to access the disk images.
45 Libguestfs is a large API because it can do many things. For a gentle
46 introduction, please read the L</API OVERVIEW> section next.
50 This section provides a gentler overview of the libguestfs API. We
51 also try to group API calls together, where that may not be obvious
52 from reading about the individual calls below.
56 Before you can use libguestfs calls, you have to create a handle.
57 Then you must add at least one disk image to the handle, followed by
58 launching the handle, then performing whatever operations you want,
59 and finally closing the handle. So the general structure of all
60 libguestfs-using programs looks like this:
62 guestfs_h *handle = guestfs_create ();
64 /* Call guestfs_add_drive additional times if there are
65 * multiple disk images.
67 guestfs_add_drive (handle, "guest.img");
69 /* Most manipulation calls won't work until you've launched
70 * the handle. You have to do this _after_ adding drives
71 * and _before_ other commands.
73 guestfs_launch (handle);
75 /* Now you can examine what partitions, LVs etc are available.
77 char **partitions = guestfs_list_partitions (handle);
78 char **logvols = guestfs_lvs (handle);
80 /* To access a filesystem in the image, you must mount it.
82 guestfs_mount (handle, "/dev/sda1", "/");
84 /* Now you can perform filesystem actions on the guest disk image. */
85 guestfs_touch (handle, "/hello");
87 /* You only need to call guestfs_sync if you have made
88 * changes to the guest image.
90 guestfs_sync (handle);
92 /* Close the handle. */
93 guestfs_close (handle);
95 The code above doesn't include any error checking. In real code you
96 should check return values carefully for errors. In general all
97 functions that return integers return C<-1> on error, and all
98 functions that return pointers return C<NULL> on error. See section
99 L</ERROR HANDLING> below for how to handle errors, and consult the
100 documentation for each function call below to see precisely how they
101 return error indications.
105 The image filename (C<"guest.img"> in the example above) could be a
106 disk image from a virtual machine, a L<dd(1)> copy of a physical block
107 device, an actual block device, or simply an empty file of zeroes that
108 you have created through L<posix_fallocate(3)>. Libguestfs lets you
109 do useful things to all of these.
111 You can add a disk read-only using C<guestfs_add_drive_ro>, in which
112 case libguestfs won't modify the file.
114 Be extremely cautious if the disk image is in use, eg. if it is being
115 used by a virtual machine. Adding it read-write will almost certainly
116 cause disk corruption, but adding it read-only is safe.
118 You must add at least one disk image, and you may add multiple disk
119 images. In the API, the disk images are usually referred to as
120 C</dev/sda> (for the first one you added), C</dev/sdb> (for the second
123 Once C<guestfs_launch> has been called you cannot add any more images.
124 You can call C<guestfs_list_devices> to get a list of the device
125 names, in the order that you added them. See also L</BLOCK DEVICE
130 Before you can read or write files, create directories and so on in a
131 disk image that contains filesystems, you have to mount those
132 filesystems using C<guestfs_mount>. If you already know that a disk
133 image contains (for example) one partition with a filesystem on that
134 partition, then you can mount it directly:
136 guestfs_mount (handle, "/dev/sda1", "/");
138 where C</dev/sda1> means literally the first partition (C<1>) of the
139 first disk image that we added (C</dev/sda>). If the disk contains
140 Linux LVM2 logical volumes you could refer to those instead (eg. C</dev/VG/LV>).
142 If you are given a disk image and you don't know what it contains then
143 you have to find out. Libguestfs can also do that: use
144 C<guestfs_list_partitions> and C<guestfs_lvs> to list possible
145 partitions and LVs, and either try mounting each to see what is
146 mountable, or else examine them with C<guestfs_file>. But you might
147 find it easier to look at higher level programs built on top of
148 libguestfs, in particular L<virt-inspector(1)>.
150 To mount a disk image read-only, use C<guestfs_mount_ro>. There are
151 several other variations of the C<guestfs_mount_*> call.
153 =head2 FILESYSTEM ACCESS AND MODIFICATION
155 The majority of the libguestfs API consists of fairly low-level calls
156 for accessing and modifying the files, directories, symlinks etc on
157 mounted filesystems. There are over a hundred such calls which you
158 can find listed in detail below in this man page, and we don't even
159 pretend to cover them all in this overview.
161 Specify filenames as full paths including the mount point.
163 For example, if you mounted a filesystem at C<"/"> and you want to
164 read the file called C<"etc/passwd"> then you could do:
166 char *data = guestfs_cat (handle, "/etc/passwd");
168 This would return C<data> as a newly allocated buffer containing the
169 full content of that file (with some conditions: see also
170 L</DOWNLOADING> below), or C<NULL> if there was an error.
172 As another example, to create a top-level directory on that filesystem
173 called C<"var"> you would do:
175 guestfs_mkdir (handle, "/var");
177 To create a symlink you could do:
179 guestfs_ln_s (handle, "/etc/init.d/portmap",
180 "/etc/rc3.d/S30portmap");
182 Libguestfs will reject attempts to use relative paths. There is no
183 concept of a current working directory. Libguestfs can return errors
184 in many situations: for example if the filesystem isn't writable, or
185 if a file or directory that you requested doesn't exist. If you are
186 using the C API (documented here) you have to check for those error
187 conditions after each call. (Other language bindings turn these
188 errors into exceptions).
190 File writes are affected by the per-handle umask, set by calling
191 C<guestfs_umask> and defaulting to 022.
195 Libguestfs contains API calls to read, create and modify partition
196 tables on disk images.
198 In the common case where you want to create a single partition
199 covering the whole disk, you should use the C<guestfs_part_disk>
202 const char *parttype = "mbr";
203 if (disk_is_larger_than_2TB)
205 guestfs_part_disk (g, "/dev/sda", parttype);
207 Obviously this effectively wipes anything that was on that disk image
210 In general MBR partitions are both unnecessarily complicated and
211 depend on archaic details, namely the Cylinder-Head-Sector (CHS)
212 geometry of the disk. C<guestfs_sfdiskM> can be used to
213 create more complex arrangements where the relative sizes are
214 expressed in megabytes instead of cylinders, which is a small win.
215 C<guestfs_sfdiskM> will choose the nearest cylinder to approximate the
216 requested size. There's a lot of crazy stuff to do with IDE and
217 virtio disks having different, incompatible CHS geometries, that you
218 probably don't want to know about.
220 My advice: make a single partition to cover the whole disk, then use
225 Libguestfs provides access to a large part of the LVM2 API, such as
226 C<guestfs_lvcreate> and C<guestfs_vgremove>. It won't make much sense
227 unless you familiarize yourself with the concepts of physical volumes,
228 volume groups and logical volumes.
230 This author strongly recommends reading the LVM HOWTO, online at
231 L<http://tldp.org/HOWTO/LVM-HOWTO/>.
235 Use C<guestfs_cat> to download small, text only files. This call
236 is limited to files which are less than 2 MB and which cannot contain
237 any ASCII NUL (C<\0>) characters. However it has a very simple
240 C<guestfs_read_file> can be used to read files which contain
241 arbitrary 8 bit data, since it returns a (pointer, size) pair.
242 However it is still limited to "small" files, less than 2 MB.
244 C<guestfs_download> can be used to download any file, with no
245 limits on content or size (even files larger than 4 GB).
247 To download multiple files, see C<guestfs_tar_out> and
252 It's often the case that you want to write a file or files to the disk
255 For small, single files, use C<guestfs_write_file>. This call
256 currently contains a bug which limited the call to plain text files
257 (not containing ASCII NUL characters).
259 To upload a single file, use C<guestfs_upload>. This call has no
260 limits on file content or size (even files larger than 4 GB).
262 To upload multiple files, see C<guestfs_tar_in> and C<guestfs_tgz_in>.
264 However the fastest way to upload I<large numbers of arbitrary files>
265 is to turn them into a squashfs or CD ISO (see L<mksquashfs(8)> and
266 L<mkisofs(8)>), then attach this using C<guestfs_add_drive_ro>. If
267 you add the drive in a predictable way (eg. adding it last after all
268 other drives) then you can get the device name from
269 C<guestfs_list_devices> and mount it directly using
270 C<guestfs_mount_ro>. Note that squashfs images are sometimes
271 non-portable between kernel versions, and they don't support labels or
272 UUIDs. If you want to pre-build an image or you need to mount it
273 using a label or UUID, use an ISO image instead.
277 C<guestfs_ll> is just designed for humans to read (mainly when using
278 the L<guestfish(1)>-equivalent command C<ll>).
280 C<guestfs_ls> is a quick way to get a list of files in a directory
281 from programs, as a flat list of strings.
283 C<guestfs_readdir> is a programmatic way to get a list of files in a
284 directory, plus additional information about each one. It is more
285 equivalent to using the L<readdir(3)> call on a local filesystem.
287 C<guestfs_find> can be used to recursively list files.
289 =head2 RUNNING COMMANDS
291 Although libguestfs is a primarily an API for manipulating files
292 inside guest images, we also provide some limited facilities for
293 running commands inside guests.
295 There are many limitations to this:
301 The kernel version that the command runs under will be different
302 from what it expects.
306 If the command needs to communicate with daemons, then most likely
307 they won't be running.
311 The command will be running in limited memory.
315 Only supports Linux guests (not Windows, BSD, etc).
319 Architecture limitations (eg. won't work for a PPC guest on
324 For SELinux guests, you may need to enable SELinux and load policy
325 first. See L</SELINUX> in this manpage.
329 The two main API calls to run commands are C<guestfs_command> and
330 C<guestfs_sh> (there are also variations).
332 The difference is that C<guestfs_sh> runs commands using the shell, so
333 any shell globs, redirections, etc will work.
335 =head2 CONFIGURATION FILES
337 To read and write configuration files in Linux guest filesystems, we
338 strongly recommend using Augeas. For example, Augeas understands how
339 to read and write, say, a Linux shadow password file or X.org
340 configuration file, and so avoids you having to write that code.
342 The main Augeas calls are bound through the C<guestfs_aug_*> APIs. We
343 don't document Augeas itself here because there is excellent
344 documentation on the L<http://augeas.net/> website.
346 If you don't want to use Augeas (you fool!) then try calling
347 C<guestfs_read_lines> to get the file as a list of lines which
348 you can iterate over.
352 We support SELinux guests. To ensure that labeling happens correctly
353 in SELinux guests, you need to enable SELinux and load the guest's
360 Before launching, do:
362 guestfs_set_selinux (g, 1);
366 After mounting the guest's filesystem(s), load the policy. This
367 is best done by running the L<load_policy(8)> command in the
370 guestfs_sh (g, "/usr/sbin/load_policy");
372 (Older versions of C<load_policy> require you to specify the
373 name of the policy file).
377 Optionally, set the security context for the API. The correct
378 security context to use can only be known by inspecting the
379 guest. As an example:
381 guestfs_setcon (g, "unconfined_u:unconfined_r:unconfined_t:s0");
385 This will work for running commands and editing existing files.
387 When new files are created, you may need to label them explicitly,
388 for example by running the external command
389 C<restorecon pathname>.
391 =head2 SPECIAL CONSIDERATIONS FOR WINDOWS GUESTS
393 Libguestfs can mount NTFS partitions. It does this using the
394 L<http://www.ntfs-3g.org/> driver.
396 DOS and Windows still use drive letters, and the filesystems are
397 always treated as case insensitive by Windows itself, and therefore
398 you might find a Windows configuration file referring to a path like
399 C<c:\windows\system32>. When the filesystem is mounted in libguestfs,
400 that directory might be referred to as C</WINDOWS/System32>.
402 Drive letter mappings are outside the scope of libguestfs. You have
403 to use libguestfs to read the appropriate Windows Registry and
404 configuration files, to determine yourself how drives are mapped (see
405 also L<virt-inspector(1)>).
407 Replacing backslash characters with forward slash characters is also
408 outside the scope of libguestfs, but something that you can easily do.
410 Where we can help is in resolving the case insensitivity of paths.
411 For this, call C<guestfs_case_sensitive_path>.
413 Libguestfs also provides some help for decoding Windows Registry
414 "hive" files, through the library C<libhivex> which is part of
415 libguestfs. You have to locate and download the hive file(s)
416 yourself, and then pass them to C<libhivex> functions. See also the
417 programs L<hivexml(1)>, L<hivexget(1)> and L<virt-win-reg(1)> for more
420 =head1 CONNECTION MANAGEMENT
424 C<guestfs_h> is the opaque type representing a connection handle.
425 Create a handle by calling C<guestfs_create>. Call C<guestfs_close>
426 to free the handle and release all resources used.
428 For information on using multiple handles and threads, see the section
429 L</MULTIPLE HANDLES AND MULTIPLE THREADS> below.
431 =head2 guestfs_create
433 guestfs_h *guestfs_create (void);
435 Create a connection handle.
437 You have to call C<guestfs_add_drive> on the handle at least once.
439 This function returns a non-NULL pointer to a handle on success or
442 After configuring the handle, you have to call C<guestfs_launch>.
444 You may also want to configure error handling for the handle. See
445 L</ERROR HANDLING> section below.
449 void guestfs_close (guestfs_h *handle);
451 This closes the connection handle and frees up all resources used.
453 =head1 ERROR HANDLING
455 The convention in all functions that return C<int> is that they return
456 C<-1> to indicate an error. You can get additional information on
457 errors by calling C<guestfs_last_error> and/or by setting up an error
458 handler with C<guestfs_set_error_handler>.
460 The default error handler prints the information string to C<stderr>.
462 Out of memory errors are handled differently. The default action is
463 to call L<abort(3)>. If this is undesirable, then you can set a
464 handler using C<guestfs_set_out_of_memory_handler>.
466 =head2 guestfs_last_error
468 const char *guestfs_last_error (guestfs_h *handle);
470 This returns the last error message that happened on C<handle>. If
471 there has not been an error since the handle was created, then this
474 The lifetime of the returned string is until the next error occurs, or
475 C<guestfs_close> is called.
477 The error string is not localized (ie. is always in English), because
478 this makes searching for error messages in search engines give the
479 largest number of results.
481 =head2 guestfs_set_error_handler
483 typedef void (*guestfs_error_handler_cb) (guestfs_h *handle,
486 void guestfs_set_error_handler (guestfs_h *handle,
487 guestfs_error_handler_cb cb,
490 The callback C<cb> will be called if there is an error. The
491 parameters passed to the callback are an opaque data pointer and the
492 error message string.
494 Note that the message string C<msg> is freed as soon as the callback
495 function returns, so if you want to stash it somewhere you must make
498 The default handler prints messages on C<stderr>.
500 If you set C<cb> to C<NULL> then I<no> handler is called.
502 =head2 guestfs_get_error_handler
504 guestfs_error_handler_cb guestfs_get_error_handler (guestfs_h *handle,
507 Returns the current error handler callback.
509 =head2 guestfs_set_out_of_memory_handler
511 typedef void (*guestfs_abort_cb) (void);
512 int guestfs_set_out_of_memory_handler (guestfs_h *handle,
515 The callback C<cb> will be called if there is an out of memory
516 situation. I<Note this callback must not return>.
518 The default is to call L<abort(3)>.
520 You cannot set C<cb> to C<NULL>. You can't ignore out of memory
523 =head2 guestfs_get_out_of_memory_handler
525 guestfs_abort_fn guestfs_get_out_of_memory_handler (guestfs_h *handle);
527 This returns the current out of memory handler.
531 Libguestfs needs a kernel and initrd.img, which it finds by looking
532 along an internal path.
534 By default it looks for these in the directory C<$libdir/guestfs>
535 (eg. C</usr/local/lib/guestfs> or C</usr/lib64/guestfs>).
537 Use C<guestfs_set_path> or set the environment variable
538 C<LIBGUESTFS_PATH> to change the directories that libguestfs will
539 search in. The value is a colon-separated list of paths. The current
540 directory is I<not> searched unless the path contains an empty element
541 or C<.>. For example C<LIBGUESTFS_PATH=:/usr/lib/guestfs> would
542 search the current directory and then C</usr/lib/guestfs>.
544 =head1 HIGH-LEVEL API ACTIONS
548 We guarantee the libguestfs ABI (binary interface), for public,
549 high-level actions as outlined in this section. Although we will
550 deprecate some actions, for example if they get replaced by newer
551 calls, we will keep the old actions forever. This allows you the
552 developer to program in confidence against libguestfs.
560 =head1 STATE MACHINE AND LOW-LEVEL EVENT API
562 Internally, libguestfs is implemented by running a virtual machine
563 using L<qemu(1)>. QEmu runs as a child process of the main program,
564 and most of this discussion won't make sense unless you understand
565 that the complexity is dealing with the (asynchronous) actions of the
569 ___________________ _________________________
571 | main program | | qemu +-----------------+|
572 | | | | Linux kernel ||
573 +-------------------+ | +-----------------+|
574 | libguestfs <-------------->| guestfsd ||
575 | | | +-----------------+|
576 \___________________/ \_________________________/
578 The diagram above shows libguestfs communicating with the guestfsd
579 daemon running inside the qemu child process. There are several
580 points of failure here: qemu can fail to start, the virtual machine
581 inside qemu can fail to boot, guestfsd can fail to start or not
582 establish communication, any component can start successfully but fail
583 asynchronously later, and so on.
587 libguestfs uses a state machine to model the child process:
598 / | \ \ guestfs_launch
609 \______/ <------ \________/
611 The normal transitions are (1) CONFIG (when the handle is created, but
612 there is no child process), (2) LAUNCHING (when the child process is
613 booting up), (3) alternating between READY and BUSY as commands are
614 issued to, and carried out by, the child process.
616 The guest may be killed by C<guestfs_kill_subprocess>, or may die
617 asynchronously at any time (eg. due to some internal error), and that
618 causes the state to transition back to CONFIG.
620 Configuration commands for qemu such as C<guestfs_add_drive> can only
621 be issued when in the CONFIG state.
623 The high-level API offers two calls that go from CONFIG through
624 LAUNCHING to READY. C<guestfs_launch> blocks until the child process
625 is READY to accept commands (or until some failure or timeout).
626 C<guestfs_launch> internally moves the state from CONFIG to LAUNCHING
629 High-level API actions such as C<guestfs_mount> can only be issued
630 when in the READY state. These high-level API calls block waiting for
631 the command to be carried out (ie. the state to transition to BUSY and
632 then back to READY). But using the low-level event API, you get
633 non-blocking versions. (But you can still only carry out one
634 operation per handle at a time - that is a limitation of the
635 communications protocol we use).
637 Finally, the child process sends asynchronous messages back to the
638 main program, such as kernel log messages. Mostly these are ignored
639 by the high-level API, but using the low-level event API you can
640 register to receive these messages.
642 =head2 SETTING CALLBACKS TO HANDLE EVENTS
644 The child process generates events in some situations. Current events
645 include: receiving a log message, the child process exits.
647 Use the C<guestfs_set_*_callback> functions to set a callback for
648 different types of events.
650 Only I<one callback of each type> can be registered for each handle.
651 Calling C<guestfs_set_*_callback> again overwrites the previous
652 callback of that type. Cancel all callbacks of this type by calling
653 this function with C<cb> set to C<NULL>.
655 =head2 guestfs_set_log_message_callback
657 typedef void (*guestfs_log_message_cb) (guestfs_h *g, void *opaque,
659 void guestfs_set_log_message_callback (guestfs_h *handle,
660 guestfs_log_message_cb cb,
663 The callback function C<cb> will be called whenever qemu or the guest
664 writes anything to the console.
666 Use this function to capture kernel messages and similar.
668 Normally there is no log message handler, and log messages are just
671 =head2 guestfs_set_subprocess_quit_callback
673 typedef void (*guestfs_subprocess_quit_cb) (guestfs_h *g, void *opaque);
674 void guestfs_set_subprocess_quit_callback (guestfs_h *handle,
675 guestfs_subprocess_quit_cb cb,
678 The callback function C<cb> will be called when the child process
679 quits, either asynchronously or if killed by
680 C<guestfs_kill_subprocess>. (This corresponds to a transition from
681 any state to the CONFIG state).
683 =head2 guestfs_set_launch_done_callback
685 typedef void (*guestfs_launch_done_cb) (guestfs_h *g, void *opaque);
686 void guestfs_set_launch_done_callback (guestfs_h *handle,
690 The callback function C<cb> will be called when the child process
691 becomes ready first time after it has been launched. (This
692 corresponds to a transition from LAUNCHING to the READY state).
694 =head1 BLOCK DEVICE NAMING
696 In the kernel there is now quite a profusion of schemata for naming
697 block devices (in this context, by I<block device> I mean a physical
698 or virtual hard drive). The original Linux IDE driver used names
699 starting with C</dev/hd*>. SCSI devices have historically used a
700 different naming scheme, C</dev/sd*>. When the Linux kernel I<libata>
701 driver became a popular replacement for the old IDE driver
702 (particularly for SATA devices) those devices also used the
703 C</dev/sd*> scheme. Additionally we now have virtual machines with
704 paravirtualized drivers. This has created several different naming
705 systems, such as C</dev/vd*> for virtio disks and C</dev/xvd*> for Xen
708 As discussed above, libguestfs uses a qemu appliance running an
709 embedded Linux kernel to access block devices. We can run a variety
710 of appliances based on a variety of Linux kernels.
712 This causes a problem for libguestfs because many API calls use device
713 or partition names. Working scripts and the recipe (example) scripts
714 that we make available over the internet could fail if the naming
717 Therefore libguestfs defines C</dev/sd*> as the I<standard naming
718 scheme>. Internally C</dev/sd*> names are translated, if necessary,
719 to other names as required. For example, under RHEL 5 which uses the
720 C</dev/hd*> scheme, any device parameter C</dev/sda2> is translated to
721 C</dev/hda2> transparently.
723 Note that this I<only> applies to parameters. The
724 C<guestfs_list_devices>, C<guestfs_list_partitions> and similar calls
725 return the true names of the devices and partitions as known to the
728 =head2 ALGORITHM FOR BLOCK DEVICE NAME TRANSLATION
730 Usually this translation is transparent. However in some (very rare)
731 cases you may need to know the exact algorithm. Such cases include
732 where you use C<guestfs_config> to add a mixture of virtio and IDE
733 devices to the qemu-based appliance, so have a mixture of C</dev/sd*>
734 and C</dev/vd*> devices.
736 The algorithm is applied only to I<parameters> which are known to be
737 either device or partition names. Return values from functions such
738 as C<guestfs_list_devices> are never changed.
744 Is the string a parameter which is a device or partition name?
748 Does the string begin with C</dev/sd>?
752 Does the named device exist? If so, we use that device.
753 However if I<not> then we continue with this algorithm.
757 Replace initial C</dev/sd> string with C</dev/hd>.
759 For example, change C</dev/sda2> to C</dev/hda2>.
761 If that named device exists, use it. If not, continue.
765 Replace initial C</dev/sd> string with C</dev/vd>.
767 If that named device exists, use it. If not, return an error.
771 =head2 PORTABILITY CONCERNS
773 Although the standard naming scheme and automatic translation is
774 useful for simple programs and guestfish scripts, for larger programs
775 it is best not to rely on this mechanism.
777 Where possible for maximum future portability programs using
778 libguestfs should use these future-proof techniques:
784 Use C<guestfs_list_devices> or C<guestfs_list_partitions> to list
785 actual device names, and then use those names directly.
787 Since those device names exist by definition, they will never be
792 Use higher level ways to identify filesystems, such as LVM names,
793 UUIDs and filesystem labels.
799 =head2 COMMUNICATION PROTOCOL
801 Don't rely on using this protocol directly. This section documents
802 how it currently works, but it may change at any time.
804 The protocol used to talk between the library and the daemon running
805 inside the qemu virtual machine is a simple RPC mechanism built on top
806 of XDR (RFC 1014, RFC 1832, RFC 4506).
808 The detailed format of structures is in C<src/guestfs_protocol.x>
809 (note: this file is automatically generated).
811 There are two broad cases, ordinary functions that don't have any
812 C<FileIn> and C<FileOut> parameters, which are handled with very
813 simple request/reply messages. Then there are functions that have any
814 C<FileIn> or C<FileOut> parameters, which use the same request and
815 reply messages, but they may also be followed by files sent using a
818 =head3 ORDINARY FUNCTIONS (NO FILEIN/FILEOUT PARAMS)
820 For ordinary functions, the request message is:
822 total length (header + arguments,
823 but not including the length word itself)
824 struct guestfs_message_header (encoded as XDR)
825 struct guestfs_<foo>_args (encoded as XDR)
827 The total length field allows the daemon to allocate a fixed size
828 buffer into which it slurps the rest of the message. As a result, the
829 total length is limited to C<GUESTFS_MESSAGE_MAX> bytes (currently
830 4MB), which means the effective size of any request is limited to
831 somewhere under this size.
833 Note also that many functions don't take any arguments, in which case
834 the C<guestfs_I<foo>_args> is completely omitted.
836 The header contains the procedure number (C<guestfs_proc>) which is
837 how the receiver knows what type of args structure to expect, or none
840 The reply message for ordinary functions is:
842 total length (header + ret,
843 but not including the length word itself)
844 struct guestfs_message_header (encoded as XDR)
845 struct guestfs_<foo>_ret (encoded as XDR)
847 As above the C<guestfs_I<foo>_ret> structure may be completely omitted
848 for functions that return no formal return values.
850 As above the total length of the reply is limited to
851 C<GUESTFS_MESSAGE_MAX>.
853 In the case of an error, a flag is set in the header, and the reply
854 message is slightly changed:
856 total length (header + error,
857 but not including the length word itself)
858 struct guestfs_message_header (encoded as XDR)
859 struct guestfs_message_error (encoded as XDR)
861 The C<guestfs_message_error> structure contains the error message as a
864 =head3 FUNCTIONS THAT HAVE FILEIN PARAMETERS
866 A C<FileIn> parameter indicates that we transfer a file I<into> the
867 guest. The normal request message is sent (see above). However this
868 is followed by a sequence of file chunks.
870 total length (header + arguments,
871 but not including the length word itself,
872 and not including the chunks)
873 struct guestfs_message_header (encoded as XDR)
874 struct guestfs_<foo>_args (encoded as XDR)
875 sequence of chunks for FileIn param #0
876 sequence of chunks for FileIn param #1 etc.
878 The "sequence of chunks" is:
880 length of chunk (not including length word itself)
881 struct guestfs_chunk (encoded as XDR)
883 struct guestfs_chunk (encoded as XDR)
886 struct guestfs_chunk (with data.data_len == 0)
888 The final chunk has the C<data_len> field set to zero. Additionally a
889 flag is set in the final chunk to indicate either successful
890 completion or early cancellation.
892 At time of writing there are no functions that have more than one
893 FileIn parameter. However this is (theoretically) supported, by
894 sending the sequence of chunks for each FileIn parameter one after
895 another (from left to right).
897 Both the library (sender) I<and> the daemon (receiver) may cancel the
898 transfer. The library does this by sending a chunk with a special
899 flag set to indicate cancellation. When the daemon sees this, it
900 cancels the whole RPC, does I<not> send any reply, and goes back to
901 reading the next request.
903 The daemon may also cancel. It does this by writing a special word
904 C<GUESTFS_CANCEL_FLAG> to the socket. The library listens for this
905 during the transfer, and if it gets it, it will cancel the transfer
906 (it sends a cancel chunk). The special word is chosen so that even if
907 cancellation happens right at the end of the transfer (after the
908 library has finished writing and has started listening for the reply),
909 the "spurious" cancel flag will not be confused with the reply
912 This protocol allows the transfer of arbitrary sized files (no 32 bit
913 limit), and also files where the size is not known in advance
914 (eg. from pipes or sockets). However the chunks are rather small
915 (C<GUESTFS_MAX_CHUNK_SIZE>), so that neither the library nor the
916 daemon need to keep much in memory.
918 =head3 FUNCTIONS THAT HAVE FILEOUT PARAMETERS
920 The protocol for FileOut parameters is exactly the same as for FileIn
921 parameters, but with the roles of daemon and library reversed.
923 total length (header + ret,
924 but not including the length word itself,
925 and not including the chunks)
926 struct guestfs_message_header (encoded as XDR)
927 struct guestfs_<foo>_ret (encoded as XDR)
928 sequence of chunks for FileOut param #0
929 sequence of chunks for FileOut param #1 etc.
931 =head3 INITIAL MESSAGE
933 Because the underlying channel (QEmu -net channel) doesn't have any
934 sort of connection control, when the daemon launches it sends an
935 initial word (C<GUESTFS_LAUNCH_FLAG>) which indicates that the guest
936 and daemon is alive. This is what C<guestfs_launch> waits for.
940 If you want to compile your own qemu, run qemu from a non-standard
941 location, or pass extra arguments to qemu, then you can write a
942 shell-script wrapper around qemu.
944 There is one important rule to remember: you I<must C<exec qemu>> as
945 the last command in the shell script (so that qemu replaces the shell
946 and becomes the direct child of the libguestfs-using program). If you
947 don't do this, then the qemu process won't be cleaned up correctly.
949 Here is an example of a wrapper, where I have built my own copy of
953 qemudir=/home/rjones/d/qemu
954 exec $qemudir/x86_64-softmmu/qemu-system-x86_64 -L $qemudir/pc-bios "$@"
956 Save this script as C</tmp/qemu.wrapper> (or wherever), C<chmod +x>,
957 and then use it by setting the LIBGUESTFS_QEMU environment variable.
960 LIBGUESTFS_QEMU=/tmp/qemu.wrapper guestfish
962 Note that libguestfs also calls qemu with the -help and -version
963 options in order to determine features.
965 =head1 ENVIRONMENT VARIABLES
969 =item LIBGUESTFS_APPEND
971 Pass additional options to the guest kernel.
973 =item LIBGUESTFS_DEBUG
975 Set C<LIBGUESTFS_DEBUG=1> to enable verbose messages. This
976 has the same effect as calling C<guestfs_set_verbose (handle, 1)>.
978 =item LIBGUESTFS_MEMSIZE
980 Set the memory allocated to the qemu process, in megabytes. For
983 LIBGUESTFS_MEMSIZE=700
985 =item LIBGUESTFS_PATH
987 Set the path that libguestfs uses to search for kernel and initrd.img.
988 See the discussion of paths in section PATH above.
990 =item LIBGUESTFS_QEMU
992 Set the default qemu binary that libguestfs uses. If not set, then
993 the qemu which was found at compile time by the configure script is
996 See also L</QEMU WRAPPERS> above.
998 =item LIBGUESTFS_TRACE
1000 Set C<LIBGUESTFS_TRACE=1> to enable command traces. This
1001 has the same effect as calling C<guestfs_set_trace (handle, 1)>.
1005 Location of temporary directory, defaults to C</tmp>.
1007 If libguestfs was compiled to use the supermin appliance then each
1008 handle will require rather a large amount of space in this directory
1009 for short periods of time (~ 80 MB). You can use C<$TMPDIR> to
1010 configure another directory to use in case C</tmp> is not large
1020 L<http://libguestfs.org/>.
1022 Tools with a similar purpose:
1031 To get a list of bugs against libguestfs use this link:
1033 L<https://bugzilla.redhat.com/buglist.cgi?component=libguestfs&product=Virtualization+Tools>
1035 To report a new bug against libguestfs use this link:
1037 L<https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virtualization+Tools>
1039 When reporting a bug, please check:
1045 That the bug hasn't been reported already.
1049 That you are testing a recent version.
1053 Describe the bug accurately, and give a way to reproduce it.
1057 Run libguestfs-test-tool and paste the B<complete, unedited>
1058 output into the bug report.
1064 Richard W.M. Jones (C<rjones at redhat dot com>)
1068 Copyright (C) 2009 Red Hat Inc.
1069 L<http://libguestfs.org/>
1071 This library is free software; you can redistribute it and/or
1072 modify it under the terms of the GNU Lesser General Public
1073 License as published by the Free Software Foundation; either
1074 version 2 of the License, or (at your option) any later version.
1076 This library is distributed in the hope that it will be useful,
1077 but WITHOUT ANY WARRANTY; without even the implied warranty of
1078 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
1079 Lesser General Public License for more details.
1081 You should have received a copy of the GNU Lesser General Public
1082 License along with this library; if not, write to the Free Software
1083 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA