path: root/src/guestfs.pod

=encoding utf8

=head1 NAME

guestfs - Library for accessing and modifying virtual machine images

=head1 SYNOPSIS

 #include <guestfs.h>
 
 guestfs_h *g = guestfs_create ();
 guestfs_add_drive (g, "guest.img");
 guestfs_launch (g);
 guestfs_mount (g, "/dev/sda1", "/");
 guestfs_touch (g, "/hello");
 guestfs_umount (g, "/");
 guestfs_shutdown (g);
 guestfs_close (g);

 cc prog.c -o prog -lguestfs
or:
 cc prog.c -o prog `pkg-config libguestfs --cflags --libs`

=head1 DESCRIPTION

Libguestfs is a library for accessing and modifying disk images and
virtual machines.  This manual page documents the C API.

If you are looking for an introduction to libguestfs, see the web
site: L<http://libguestfs.org/>

Each virt tool has its own man page (for a full list, go to
L</SEE ALSO> at the end of this file).

The libguestfs FAQ contains many useful answers: L<guestfs-faq(1)>.

For examples of using the API from C, see L<guestfs-examples(3)>.  For
examples in other languages, see
L</USING LIBGUESTFS WITH OTHER PROGRAMMING LANGUAGES> below.

For tips and recipes, see L<guestfs-recipes(1)>.

If you are having performance problems, read
L<guestfs-performance(1)>.  To help test libguestfs, read
L<libguestfs-test-tool(1)> and L<guestfs-testing(1)>.

=head1 API OVERVIEW

This section provides a gentler overview of the libguestfs API.  We
also try to group API calls together, where that may not be obvious
from reading about the individual calls in the main section of this
manual.

=head2 HANDLES

Before you can use libguestfs calls, you have to create a handle.
Then you must add at least one disk image to the handle, followed by
launching the handle, then performing whatever operations you want,
and finally closing the handle.  By convention we use the single
letter C<g> for the name of the handle variable, although of course
you can use any name you want.

The general structure of all libguestfs-using programs looks like
this:

 guestfs_h *g = guestfs_create ();
 
 /* Call guestfs_add_drive additional times if there are
  * multiple disk images.
  */
 guestfs_add_drive (g, "guest.img");
 
 /* Most manipulation calls won't work until you've launched
  * the handle 'g'.  You have to do this _after_ adding drives
  * and _before_ other commands.
  */
 guestfs_launch (g);
 
 /* Now you can examine what partitions, LVs etc are available.
  */
 char **partitions = guestfs_list_partitions (g);
 char **logvols = guestfs_lvs (g);
 
 /* To access a filesystem in the image, you must mount it.
  */
 guestfs_mount (g, "/dev/sda1", "/");
 
 /* Now you can perform filesystem actions on the guest
  * disk image.
  */
 guestfs_touch (g, "/hello");
 
 /* Synchronize the disk.  This is the opposite of guestfs_launch. */
 guestfs_shutdown (g);
 
 /* Close and free the handle 'g'. */
 guestfs_close (g);

The code above doesn't include any error checking.  In real code you
should check return values carefully for errors.  In general all
functions that return integers return C<-1> on error, and all
functions that return pointers return C<NULL> on error.  See section
L</ERROR HANDLING> below for how to handle errors, and consult the
documentation for each function call below to see precisely how they
return error indications.  See L<guestfs-examples(3)> for fully worked
examples.

=head2 DISK IMAGES

The image filename (C<"guest.img"> in the example above) could be a
disk image from a virtual machine, a L<dd(1)> copy of a physical hard
disk, an actual block device, or simply an empty file of zeroes that
you have created through L<posix_fallocate(3)>.  Libguestfs lets you
do useful things to all of these.

The call you should use in modern code for adding drives is
L</guestfs_add_drive_opts>.  To add a disk image, allowing writes, and
specifying that the format is raw, do:

 guestfs_add_drive_opts (g, filename,
                         GUESTFS_ADD_DRIVE_OPTS_FORMAT, "raw",
                         -1);

You can add a disk read-only using:

 guestfs_add_drive_opts (g, filename,
                         GUESTFS_ADD_DRIVE_OPTS_FORMAT, "raw",
                         GUESTFS_ADD_DRIVE_OPTS_READONLY, 1,
                         -1);

or by calling the older function L</guestfs_add_drive_ro>.  In either
case libguestfs won't modify the file.
(See also L</DISK IMAGE FORMATS> below).

Be extremely cautious if the disk image is in use, eg. if it is being
used by a virtual machine.  Adding it read-write will almost certainly
cause disk corruption, but adding it read-only is safe.

You must add at least one disk image, and you may add multiple disk
images.  In the API, the disk images are usually referred to as
C</dev/sda> (for the first one you added), C</dev/sdb> (for the second
one you added), etc.

Once L</guestfs_launch> has been called you cannot add any more images.
You can call L</guestfs_list_devices> to get a list of the device
names, in the order that you added them.
See also L</BLOCK DEVICE NAMING> below.

There are slightly different rules when hotplugging disks (in
libguestfs E<ge> 1.20).  See L</HOTPLUGGING> below.

=head2 MOUNTING

Before you can read or write files, create directories and so on in a
disk image that contains filesystems, you have to mount those
filesystems using L</guestfs_mount> or L</guestfs_mount_ro>.
If you already know that a disk image contains (for example) one
partition with a filesystem on that partition, then you can mount it
directly:

 guestfs_mount (g, "/dev/sda1", "/");

where C</dev/sda1> means literally the first partition (C<1>) of the
first disk image that we added (C</dev/sda>).  If the disk contains
Linux LVM2 logical volumes you could refer to those instead
(eg. C</dev/VG/LV>).  Note that these are libguestfs virtual devices,
and are nothing to do with host devices.

If you are given a disk image and you don't know what it contains then
you have to find out.  Libguestfs can do that too: use
L</guestfs_list_partitions> and L</guestfs_lvs> to list possible
partitions and LVs, and either try mounting each to see what is
mountable, or else examine them with L</guestfs_vfs_type> or
L</guestfs_file>.  To list just filesystems, use
L</guestfs_list_filesystems>.

Libguestfs also has a set of APIs for inspection of unknown disk
images (see L</INSPECTION> below).  But you might find it easier to
look at higher level programs built on top of libguestfs, in
particular L<virt-inspector(1)>.

To mount a filesystem read-only, use L</guestfs_mount_ro>.  There are
several other variations of the C<guestfs_mount_*> call.

=head2 FILESYSTEM ACCESS AND MODIFICATION

The majority of the libguestfs API consists of fairly low-level calls
for accessing and modifying the files, directories, symlinks etc on
mounted filesystems.  There are over a hundred such calls which you
can find listed in detail below in this man page, and we don't even
pretend to cover them all in this overview.

Specify filenames as full paths, starting with C<"/"> and including
the mount point.

For example, if you mounted a filesystem at C<"/"> and you want to
read the file called C<"etc/passwd"> then you could do:

 char *data = guestfs_cat (g, "/etc/passwd");

This would return C<data> as a newly allocated buffer containing the
full content of that file (with some conditions: see also
L</DOWNLOADING> below), or C<NULL> if there was an error.

As another example, to create a top-level directory on that filesystem
called C<"var"> you would do:

 guestfs_mkdir (g, "/var");

To create a symlink you could do:

 guestfs_ln_s (g, "/etc/init.d/portmap",
               "/etc/rc3.d/S30portmap");

Libguestfs will reject attempts to use relative paths and there is no
concept of a current working directory.

Libguestfs can return errors in many situations: for example if the
filesystem isn't writable, or if a file or directory that you
requested doesn't exist.  If you are using the C API (documented here)
you have to check for those error conditions after each call.  (Other
language bindings turn these errors into exceptions).

File writes are affected by the per-handle umask, set by calling
L</guestfs_umask> and defaulting to 022.  See L</UMASK>.

Since libguestfs 1.18, it is possible to mount the libguestfs
filesystem on a local directory, subject to some restrictions.  See
L</MOUNT LOCAL> below.

=head2 PARTITIONING

Libguestfs contains API calls to read, create and modify partition
tables on disk images.

In the common case where you want to create a single partition
covering the whole disk, you should use the L</guestfs_part_disk>
call:

 const char *parttype = "mbr";
 if (disk_is_larger_than_2TB)
   parttype = "gpt";
 guestfs_part_disk (g, "/dev/sda", parttype);

Obviously this effectively wipes anything that was on that disk image
before.

=head2 LVM2

Libguestfs provides access to a large part of the LVM2 API, such as
L</guestfs_lvcreate> and L</guestfs_vgremove>.  It won't make much sense
unless you familiarize yourself with the concepts of physical volumes,
volume groups and logical volumes.

This author strongly recommends reading the LVM HOWTO, online at
L<http://tldp.org/HOWTO/LVM-HOWTO/>.

=head2 DOWNLOADING

Use L</guestfs_cat> to download small, text only files.  This call
cannot handle files contain any ASCII NUL (C<\0>) characters.  However
the API is very simple to use.

L</guestfs_read_file> can be used to read files which contain
arbitrary 8 bit data, since it returns a (pointer, size) pair.

L</guestfs_download> can be used to download any file, with no limits
on content or size.

To download multiple files, see L</guestfs_tar_out> and
L</guestfs_tgz_out>.

=head2 UPLOADING

It's often the case that you want to write a file or files to the disk
image.

To write a small file with fixed content, use L</guestfs_write>.  To
create a file of all zeroes, use L</guestfs_truncate_size> (sparse) or
L</guestfs_fallocate64> (with all disk blocks allocated).  There are a
variety of other functions for creating test files, for example
L</guestfs_fill> and L</guestfs_fill_pattern>.

To upload a single file, use L</guestfs_upload>.  This call has no
limits on file content or size.

To upload multiple files, see L</guestfs_tar_in> and L</guestfs_tgz_in>.

However the fastest way to upload I<large numbers of arbitrary files>
is to turn them into a squashfs or CD ISO (see L<mksquashfs(8)> and
L<mkisofs(8)>), then attach this using L</guestfs_add_drive_ro>.  If
you add the drive in a predictable way (eg. adding it last after all
other drives) then you can get the device name from
L</guestfs_list_devices> and mount it directly using
L</guestfs_mount_ro>.  Note that squashfs images are sometimes
non-portable between kernel versions, and they don't support labels or
UUIDs.  If you want to pre-build an image or you need to mount it
using a label or UUID, use an ISO image instead.

=head2 COPYING

There are various different commands for copying between files and
devices and in and out of the guest filesystem.  These are summarised
in the table below.

=over 4

=item B<file> to B<file>

Use L</guestfs_cp> to copy a single file, or L</guestfs_cp_a> to copy
directories recursively.

To copy part of a file (offset and size) use
L</guestfs_copy_file_to_file>.

=item B<file> to B<device>

=item B<device> to B<file>

=item B<device> to B<device>

Use L</guestfs_copy_file_to_device>, L</guestfs_copy_device_to_file>,
or L</guestfs_copy_device_to_device>.

Example: duplicate the contents of an LV:

 guestfs_copy_device_to_device (g,
         "/dev/VG/Original", "/dev/VG/Copy",
         /* -1 marks the end of the list of optional parameters */
         -1);

The destination (C</dev/VG/Copy>) must be at least as large as the
source (C</dev/VG/Original>).  To copy less than the whole source
device, use the optional C<size> parameter:

 guestfs_copy_device_to_device (g,
         "/dev/VG/Original", "/dev/VG/Copy",
         GUESTFS_COPY_DEVICE_TO_DEVICE_SIZE, 10000,
         -1);

=item B<file on the host> to B<file or device>

Use L</guestfs_upload>.  See L</UPLOADING> above.

=item B<file or device> to B<file on the host>

Use L</guestfs_download>.  See L</DOWNLOADING> above.

=back

=head2 UPLOADING AND DOWNLOADING TO PIPES AND FILE DESCRIPTORS

Calls like L</guestfs_upload>, L</guestfs_download>,
L</guestfs_tar_in>, L</guestfs_tar_out> etc appear to only take
filenames as arguments, so it appears you can only upload and download
to files.  However many Un*x-like hosts let you use the special device
files C</dev/stdin>, C</dev/stdout>, C</dev/stderr> and C</dev/fd/N>
to read and write from stdin, stdout, stderr, and arbitrary file
descriptor N.

For example, L<virt-cat(1)> writes its output to stdout by
doing:

 guestfs_download (g, filename, "/dev/stdout");

and you can write tar output to a file descriptor C<fd> by doing:

 char devfd[64];
 snprintf (devfd, sizeof devfd, "/dev/fd/%d", fd);
 guestfs_tar_out (g, "/", devfd);

=head2 LISTING FILES

L</guestfs_ll> is just designed for humans to read (mainly when using
the L<guestfish(1)>-equivalent command C<ll>).

L</guestfs_ls> is a quick way to get a list of files in a directory
from programs, as a flat list of strings.

L</guestfs_readdir> is a programmatic way to get a list of files in a
directory, plus additional information about each one.  It is more
equivalent to using the L<readdir(3)> call on a local filesystem.

L</guestfs_find> and L</guestfs_find0> can be used to recursively list
files.

=head2 RUNNING COMMANDS

Although libguestfs is primarily an API for manipulating files
inside guest images, we also provide some limited facilities for
running commands inside guests.

There are many limitations to this:

=over 4

=item *

The kernel version that the command runs under will be different
from what it expects.

=item *

If the command needs to communicate with daemons, then most likely
they won't be running.

=item *

The command will be running in limited memory.

=item *

The network may not be available unless you enable it
(see L</guestfs_set_network>).

=item *

Only supports Linux guests (not Windows, BSD, etc).

=item *

Architecture limitations (eg. won't work for a PPC guest on
an X86 host).

=item *

For SELinux guests, you may need to enable SELinux and load policy
first.  See L</SELINUX> in this manpage.

=item *

I<Security:> It is not safe to run commands from untrusted, possibly
malicious guests.  These commands may attempt to exploit your program
by sending unexpected output.  They could also try to exploit the
Linux kernel or qemu provided by the libguestfs appliance.  They could
use the network provided by the libguestfs appliance to bypass
ordinary network partitions and firewalls.  They could use the
elevated privileges or different SELinux context of your program
to their advantage.

A secure alternative is to use libguestfs to install a "firstboot"
script (a script which runs when the guest next boots normally), and
to have this script run the commands you want in the normal context of
the running guest, network security and so on.  For information about
other security issues, see L</SECURITY>.

=back

The two main API calls to run commands are L</guestfs_command> and
L</guestfs_sh> (there are also variations).

The difference is that L</guestfs_sh> runs commands using the shell, so
any shell globs, redirections, etc will work.

=head2 CONFIGURATION FILES

To read and write configuration files in Linux guest filesystems, we
strongly recommend using Augeas.  For example, Augeas understands how
to read and write, say, a Linux shadow password file or X.org
configuration file, and so avoids you having to write that code.

The main Augeas calls are bound through the C<guestfs_aug_*> APIs.  We
don't document Augeas itself here because there is excellent
documentation on the L<http://augeas.net/> website.

If you don't want to use Augeas (you fool!) then try calling
L</guestfs_read_lines> to get the file as a list of lines which
you can iterate over.

=head2 SELINUX

We support SELinux guests.  To ensure that labeling happens correctly
in SELinux guests, you need to enable SELinux and load the guest's
policy:

=over 4

=item 1.

Before launching, do:

 guestfs_set_selinux (g, 1);

=item 2.

After mounting the guest's filesystem(s), load the policy.  This
is best done by running the L<load_policy(8)> command in the
guest itself:

 guestfs_sh (g, "/usr/sbin/load_policy");

(Older versions of C<load_policy> require you to specify the
name of the policy file).

=item 3.

Optionally, set the security context for the API.  The correct
security context to use can only be known by inspecting the
guest.  As an example:

 guestfs_setcon (g, "unconfined_u:unconfined_r:unconfined_t:s0");

=back

This will work for running commands and editing existing files.

When new files are created, you may need to label them explicitly,
for example by running the external command
C<restorecon pathname>.

=head2 UMASK

Certain calls are affected by the current file mode creation mask (the
"umask").  In particular ones which create files or directories, such
as L</guestfs_touch>, L</guestfs_mknod> or L</guestfs_mkdir>.  This
affects either the default mode that the file is created with or
modifies the mode that you supply.

The default umask is C<022>, so files are created with modes such as
C<0644> and directories with C<0755>.

There are two ways to avoid being affected by umask.  Either set umask
to 0 (call C<guestfs_umask (g, 0)> early after launching).  Or call
L</guestfs_chmod> after creating each file or directory.

For more information about umask, see L<umask(2)>.

=head2 ENCRYPTED DISKS

Libguestfs allows you to access Linux guests which have been
encrypted using whole disk encryption that conforms to the
Linux Unified Key Setup (LUKS) standard.  This includes
nearly all whole disk encryption systems used by modern
Linux guests.

Use L</guestfs_vfs_type> to identify LUKS-encrypted block
devices (it returns the string C<crypto_LUKS>).

Then open these devices by calling L</guestfs_luks_open>.
Obviously you will require the passphrase!

Opening a LUKS device creates a new device mapper device
called C</dev/mapper/mapname> (where C<mapname> is the
string you supply to L</guestfs_luks_open>).
Reads and writes to this mapper device are decrypted from and
encrypted to the underlying block device respectively.

LVM volume groups on the device can be made visible by calling
L</guestfs_vgscan> followed by L</guestfs_vg_activate_all>.
The logical volume(s) can now be mounted in the usual way.

Use the reverse process to close a LUKS device.  Unmount
any logical volumes on it, deactivate the volume groups
by caling C<guestfs_vg_activate (g, 0, ["/dev/VG"])>.
Then close the mapper device by calling
L</guestfs_luks_close> on the C</dev/mapper/mapname>
device (I<not> the underlying encrypted block device).

=head2 MOUNT LOCAL

In libguestfs E<ge> 1.18, it is possible to mount the libguestfs
filesystem on a local directory and access it using ordinary POSIX
calls and programs.

Availability of this is subject to a number of restrictions: it
requires FUSE (the Filesystem in USErspace), and libfuse must also
have been available when libguestfs was compiled.  FUSE may require
that a kernel module is loaded, and it may be necessary to add the
current user to a special C<fuse> group.  See the documentation for
your distribution and L<http://fuse.sf.net> for further information.

The call to mount the libguestfs filesystem on a local directory is
L</guestfs_mount_local> (q.v.) followed by L</guestfs_mount_local_run>.
The latter does not return until you unmount the filesystem.
The reason is that the call enters the FUSE main loop and processes
kernel requests, turning them into libguestfs calls.  An alternative
design would have been to create a background thread to do this, but
libguestfs doesn't require pthreads.  This way is also more flexible:
for example the user can create another thread for
L</guestfs_mount_local_run>.

L</guestfs_mount_local> needs a certain amount of time to set up the
mountpoint.  The mountpoint is not ready to use until the call
returns.  At this point, accesses to the filesystem will block
until the main loop is entered (ie. L</guestfs_mount_local_run>).
So if you need to start another process to access the filesystem,
put the fork between L</guestfs_mount_local> and
L</guestfs_mount_local_run>.

=head3 MOUNT LOCAL COMPATIBILITY

Since local mounting was only added in libguestfs 1.18, and may not
be available even in these builds, you should consider writing code
so that it doesn't depend on this feature, and can fall back to
using libguestfs file system calls.

If libguestfs was compiled without support for L</guestfs_mount_local>
then calling it will return an error with errno set to C<ENOTSUP> (see
L</guestfs_last_errno>).

=head3 MOUNT LOCAL PERFORMANCE

Libguestfs on top of FUSE performs quite poorly.  For best performance
do not use it.  Use ordinary libguestfs filesystem calls, upload,
download etc. instead.

=head2 HOTPLUGGING

In libguestfs E<ge> 1.20, you may add drives and remove after calling
L</guestfs_launch>.  There are some restrictions, see below.  This is
called I<hotplugging>.

Only a subset of the attach-method backends support hotplugging
(currently only the libvirt attach-method has support).  It also
requires that you use libvirt E<ge> 0.10.3 and qemu E<ge> 1.2.

To hot-add a disk, simply call L</guestfs_add_drive_opts> after
L</guestfs_launch>.  It is mandatory to specify the C<label> parameter
so that the newly added disk has a predictable name.  For example:

 if (guestfs_launch (g) == -1)
   error ("launch failed");
 
 if (guestfs_add_drive_opts (g, filename,
                             GUESTFS_ADD_DRIVE_OPTS_LABEL, "newdisk",
                             -1) == -1)
   error ("hot-add of disk failed");
 
 if (guestfs_part_disk ("/dev/disk/guestfs/newdisk", "mbr") == -1)
   error ("partitioning of hot-added disk failed");

To hot-remove a disk, call L</guestfs_remove_drive>.  You can call
this before or after L</guestfs_launch>.  You can only remove disks
that were previously added with a label.

Backends that support hotplugging do not require that you add
E<ge> 1 disk before calling launch.  When hotplugging is supported
you don't need to add any disks.

=head2 INSPECTION

Libguestfs has APIs for inspecting an unknown disk image to find out
if it contains operating systems, an install CD or a live CD.  (These
APIs used to be in a separate Perl-only library called
L<Sys::Guestfs::Lib(3)> but since version 1.5.3 the most frequently
used part of this library has been rewritten in C and moved into the
core code).

Add all disks belonging to the unknown virtual machine and call
L</guestfs_launch> in the usual way.

Then call L</guestfs_inspect_os>.  This function uses other libguestfs
calls and certain heuristics, and returns a list of operating systems
that were found.  An empty list means none were found.  A single
element is the root filesystem of the operating system.  For dual- or
multi-boot guests, multiple roots can be returned, each one
corresponding to a separate operating system.  (Multi-boot virtual
machines are extremely rare in the world of virtualization, but since
this scenario can happen, we have built libguestfs to deal with it.)

For each root, you can then call various C<guestfs_inspect_get_*>
functions to get additional details about that operating system.  For
example, call L</guestfs_inspect_get_type> to return the string
C<windows> or C<linux> for Windows and Linux-based operating systems
respectively.

Un*x-like and Linux-based operating systems usually consist of several
filesystems which are mounted at boot time (for example, a separate
boot partition mounted on C</boot>).  The inspection rules are able to
detect how filesystems correspond to mount points.  Call
C<guestfs_inspect_get_mountpoints> to get this mapping.  It might
return a hash table like this example:

 /boot => /dev/sda1
 /     => /dev/vg_guest/lv_root
 /usr  => /dev/vg_guest/lv_usr

The caller can then make calls to L</guestfs_mount> to
mount the filesystems as suggested.

Be careful to mount filesystems in the right order (eg. C</> before
C</usr>).  Sorting the keys of the hash by length, shortest first,
should work.

Inspection currently only works for some common operating systems.
Contributors are welcome to send patches for other operating systems
that we currently cannot detect.

Encrypted disks must be opened before inspection.  See
L</ENCRYPTED DISKS> for more details.  The L</guestfs_inspect_os>
function just ignores any encrypted devices.

A note on the implementation: The call L</guestfs_inspect_os> performs
inspection and caches the results in the guest handle.  Subsequent
calls to C<guestfs_inspect_get_*> return this cached information, but
I<do not> re-read the disks.  If you change the content of the guest
disks, you can redo inspection by calling L</guestfs_inspect_os>
again.  (L</guestfs_inspect_list_applications2> works a little
differently from the other calls and does read the disks.  See
documentation for that function for details).

=head3 INSPECTING INSTALL DISKS

Libguestfs (since 1.9.4) can detect some install disks, install
CDs, live CDs and more.

Call L</guestfs_inspect_get_format> to return the format of the
operating system, which currently can be C<installed> (a regular
operating system) or C<installer> (some sort of install disk).

Further information is available about the operating system that can
be installed using the regular inspection APIs like
L</guestfs_inspect_get_product_name>,
L</guestfs_inspect_get_major_version> etc.

Some additional information specific to installer disks is also
available from the L</guestfs_inspect_is_live>,
L</guestfs_inspect_is_netinst> and L</guestfs_inspect_is_multipart>
calls.

=head2 SPECIAL CONSIDERATIONS FOR WINDOWS GUESTS

Libguestfs can mount NTFS partitions.  It does this using the
L<http://www.ntfs-3g.org/> driver.

=head3 DRIVE LETTERS AND PATHS

DOS and Windows still use drive letters, and the filesystems are
always treated as case insensitive by Windows itself, and therefore
you might find a Windows configuration file referring to a path like
C<c:\windows\system32>.  When the filesystem is mounted in libguestfs,
that directory might be referred to as C</WINDOWS/System32>.

Drive letter mappings can be found using inspection
(see L</INSPECTION> and L</guestfs_inspect_get_drive_mappings>)

Dealing with separator characters (backslash vs forward slash) is
outside the scope of libguestfs, but usually a simple character
replacement will work.

To resolve the case insensitivity of paths, call
L</guestfs_case_sensitive_path>.

=head3 LONG FILENAMES ON NTFS

NTFS supports filenames up to 255 characters long.  "Character" means
a 2 byte UTF-16 codepoint which can encode the most common Unicode
codepoints.

Most Linux filesystems support filenames up to 255 I<bytes>.
This means you may get an error:

 File name too long

when you copy a file from NTFS to a Linux filesystem if the name, when
reencoded as UTF-8, would exceed 255 bytes in length.

This will most often happen when using non-ASCII names that are longer
than ~127 characters (eg. Greek, Cyrillic) or longer than ~85
characters (Asian languages).

A workaround is not to try to store such long filenames on Linux
native filesystems.  Since the L<tar(1)> format can store unlimited
length filenames, keep the files in a tarball.

=head3 ACCESSING THE WINDOWS REGISTRY

Libguestfs also provides some help for decoding Windows Registry
"hive" files, through a separate C library called L<hivex(3)>.

Before libguestfs 1.19.35 you had to download the hive file, operate
on it locally using hivex, and upload it again.  Since this version,
we have included the major hivex APIs directly in the libguestfs API
(see L</guestfs_hivex_open>).  This means that if you have opened a
Windows guest, you can read and write the registry directly.

See also L<virt-win-reg(1)>.

=head3 SYMLINKS ON NTFS-3G FILESYSTEMS

Ntfs-3g tries to rewrite "Junction Points" and NTFS "symbolic links"
to provide something which looks like a Linux symlink.  The way it
tries to do the rewriting is described here:

L<http://www.tuxera.com/community/ntfs-3g-advanced/junction-points-and-symbolic-links/>

The essential problem is that ntfs-3g simply does not have enough
information to do a correct job.  NTFS links can contain drive letters
and references to external device GUIDs that ntfs-3g has no way of
resolving.  It is almost certainly the case that libguestfs callers
should ignore what ntfs-3g does (ie. don't use L</guestfs_readlink> on
NTFS volumes).

Instead if you encounter a symbolic link on an ntfs-3g filesystem, use
L</guestfs_lgetxattr> to read the C<system.ntfs_reparse_data> extended
attribute, and read the raw reparse data from that (you can find the
format documented in various places around the web).

=head3 EXTENDED ATTRIBUTES ON NTFS-3G FILESYSTEMS

There are other useful extended attributes that can be read from
ntfs-3g filesystems (using L</guestfs_getxattr>).  See:

L<http://www.tuxera.com/community/ntfs-3g-advanced/extended-attributes/>

=head2 RESIZE2FS ERRORS

The L</guestfs_resize2fs>, L</guestfs_resize2fs_size> and
L</guestfs_resize2fs_M> calls are used to resize ext2/3/4 filesystems.

The underlying program (L<resize2fs(8)>) requires that the filesystem
is clean and recently fsck'd before you can resize it.  Also, if the
resize operation fails for some reason, then you had to call fsck the
filesystem again to fix it.

In libguestfs C<lt> 1.17.14, you usually had to call
L</guestfs_e2fsck_f> before the resize.  However, in C<ge> 1.17.14,
L<e2fsck(8)> is called automatically before the resize, so you no
longer need to do this.

The L<resize2fs(8)> program can still fail, in which case it prints an
error message similar to:

 Please run 'e2fsck -fy <device>' to fix the filesystem
 after the aborted resize operation.

You can do this by calling L</guestfs_e2fsck> with the C<forceall>
option.  However in the context of disk images, it is usually better
to avoid this situation, eg. by rolling back to an earlier snapshot,
or by copying and resizing and on failure going back to the original.

=head2 USING LIBGUESTFS WITH OTHER PROGRAMMING LANGUAGES

Although we don't want to discourage you from using the C API, we will
mention here that the same API is also available in other languages.

The API is broadly identical in all supported languages.  This means
that the C call C<guestfs_add_drive_ro(g,file)> is
C<$g-E<gt>add_drive_ro($file)> in Perl, C<g.add_drive_ro(file)> in Python,
and C<g#add_drive_ro file> in OCaml.  In other words, a
straightforward, predictable isomorphism between each language.

Error messages are automatically transformed
into exceptions if the language supports it.

We don't try to "object orientify" parts of the API in OO languages,
although contributors are welcome to write higher level APIs above
what we provide in their favourite languages if they wish.

=over 4

=item B<C++>

You can use the I<guestfs.h> header file from C++ programs.  The C++
API is identical to the C API.  C++ classes and exceptions are not
used.

=item B<C#>

The C# bindings are highly experimental.  Please read the warnings
at the top of C<csharp/Libguestfs.cs>.

=item B<Erlang>

See L<guestfs-erlang(3)>.

=item B<GObject>

Experimental GObject bindings (with GObject Introspection support) are
available.  See the C<gobject> directory in the source.

=item B<Haskell>

This is the only language binding that is working but incomplete.
Only calls which return simple integers have been bound in Haskell,
and we are looking for help to complete this binding.

=item B<Java>

Full documentation is contained in the Javadoc which is distributed
with libguestfs.  For examples, see L<guestfs-java(3)>.

=item B<OCaml>

See L<guestfs-ocaml(3)>.

=item B<Perl>

See L<guestfs-perl(3)> and L<Sys::Guestfs(3)>.

=item B<PHP>

For documentation see C<README-PHP> supplied with libguestfs
sources or in the php-libguestfs package for your distribution.

The PHP binding only works correctly on 64 bit machines.

=item B<Python>

See L<guestfs-python(3)>.

=item B<Ruby>

See L<guestfs-ruby(3)>.

For JRuby, use the Java bindings.

=item B<shell scripts>

See L<guestfish(1)>.

=back

=head2 LIBGUESTFS GOTCHAS

L<http://en.wikipedia.org/wiki/Gotcha_(programming)>: "A feature of a
system [...] that works in the way it is documented but is
counterintuitive and almost invites mistakes."

Since we developed libguestfs and the associated tools, there are
several things we would have designed differently, but are now stuck
with for backwards compatibility or other reasons.  If there is ever a
libguestfs 2.0 release, you can expect these to change.  Beware of
them.

=over 4

=item Autosync / forgetting to sync.

I<Update:> Autosync is enabled by default for all API users starting
from libguestfs 1.5.24.  This section only applies to older versions.

When modifying a filesystem from C or another language, you B<must>
unmount all filesystems and call L</guestfs_sync> explicitly before
you close the libguestfs handle.  You can also call:

 guestfs_set_autosync (g, 1);

to have the unmount/sync done automatically for you when the handle 'g'
is closed.  (This feature is called "autosync", L</guestfs_set_autosync>
q.v.)

If you forget to do this, then it is entirely possible that your
changes won't be written out, or will be partially written, or (very
rarely) that you'll get disk corruption.

Note that in L<guestfish(3)> autosync is the default.  So quick and
dirty guestfish scripts that forget to sync will work just fine, which
can make this very puzzling if you are trying to debug a problem.

=item Mount option C<-o sync> should not be the default.

I<Update:> L</guestfs_mount> no longer adds any options starting
from libguestfs 1.13.16.  This section only applies to older versions.

If you use L</guestfs_mount>, then C<-o sync,noatime> are added
implicitly.  However C<-o sync> does not add any reliability benefit,
but does have a very large performance impact.

The work around is to use L</guestfs_mount_options> and set the mount
options that you actually want to use.

=item Read-only should be the default.

In L<guestfish(3)>, I<--ro> should be the default, and you should
have to specify I<--rw> if you want to make changes to the image.

This would reduce the potential to corrupt live VM images.

Note that many filesystems change the disk when you just mount and
unmount, even if you didn't perform any writes.  You need to use
L</guestfs_add_drive_ro> to guarantee that the disk is not changed.

=item guestfish command line is hard to use.

C<guestfish disk.img> doesn't do what people expect (open C<disk.img>
for examination).  It tries to run a guestfish command C<disk.img>
which doesn't exist, so it fails.  In earlier versions of guestfish
the error message was also unintuitive, but we have corrected this
since.  Like the Bourne shell, we should have used C<guestfish -c
command> to run commands.

=item guestfish megabyte modifiers don't work right on all commands

In recent guestfish you can use C<1M> to mean 1 megabyte (and
similarly for other modifiers).  What guestfish actually does is to
multiply the number part by the modifier part and pass the result to
the C API.  However this doesn't work for a few APIs which aren't
expecting bytes, but are already expecting some other unit
(eg. megabytes).

The most common is L</guestfs_lvcreate>.  The guestfish command:

 lvcreate LV VG 100M

does not do what you might expect.  Instead because
L</guestfs_lvcreate> is already expecting megabytes, this tries to
create a 100 I<terabyte> (100 megabytes * megabytes) logical volume.
The error message you get from this is also a little obscure.

This could be fixed in the generator by specially marking parameters
and return values which take bytes or other units.

=item Ambiguity between devices and paths

There is a subtle ambiguity in the API between a device name
(eg. C</dev/sdb2>) and a similar pathname.  A file might just happen
to be called C<sdb2> in the directory C</dev> (consider some non-Unix
VM image).

In the current API we usually resolve this ambiguity by having two
separate calls, for example L</guestfs_checksum> and
L</guestfs_checksum_device>.  Some API calls are ambiguous and
(incorrectly) resolve the problem by detecting if the path supplied
begins with C</dev/>.

To avoid both the ambiguity and the need to duplicate some calls, we
could make paths/devices into structured names.  One way to do this
would be to use a notation like grub (C<hd(0,0)>), although nobody
really likes this aspect of grub.  Another way would be to use a
structured type, equivalent to this OCaml type:

 type path = Path of string | Device of int | Partition of int * int

which would allow you to pass arguments like:

 Path "/foo/bar"
 Device 1            (* /dev/sdb, or perhaps /dev/sda *)
 Partition (1, 2)    (* /dev/sdb2 (or is it /dev/sda2 or /dev/sdb3?) *)
 Path "/dev/sdb2"    (* not a device *)

As you can see there are still problems to resolve even with this
representation.  Also consider how it might work in guestfish.

=back

=head2 KEYS AND PASSPHRASES

Certain libguestfs calls take a parameter that contains sensitive key
material, passed in as a C string.

In the future we would hope to change the libguestfs implementation so
that keys are L<mlock(2)>-ed into physical RAM, and thus can never end
up in swap.  However this is I<not> done at the moment, because of the
complexity of such an implementation.

Therefore you should be aware that any key parameter you pass to
libguestfs might end up being written out to the swap partition.  If
this is a concern, scrub the swap partition or don't use libguestfs on
encrypted devices.

=head2 MULTIPLE HANDLES AND MULTIPLE THREADS

All high-level libguestfs actions are synchronous.  If you want
to use libguestfs asynchronously then you must create a thread.

Only use the handle from a single thread.  Either use the handle
exclusively from one thread, or provide your own mutex so that two
threads cannot issue calls on the same handle at the same time.

See the graphical program guestfs-browser for one possible
architecture for multithreaded programs using libvirt and libguestfs.

=head2 PATH

Libguestfs needs a supermin appliance, which it finds by looking along
an internal path.

By default it looks for these in the directory C<$libdir/guestfs>
(eg. C</usr/local/lib/guestfs> or C</usr/lib64/guestfs>).

Use L</guestfs_set_path> or set the environment variable
L</LIBGUESTFS_PATH> to change the directories that libguestfs will
search in.  The value is a colon-separated list of paths.  The current
directory is I<not> searched unless the path contains an empty element
or C<.>.  For example C<LIBGUESTFS_PATH=:/usr/lib/guestfs> would
search the current directory and then C</usr/lib/guestfs>.

=head2 QEMU WRAPPERS

If you want to compile your own qemu, run qemu from a non-standard
location, or pass extra arguments to qemu, then you can write a
shell-script wrapper around qemu.

There is one important rule to remember: you I<must C<exec qemu>> as
the last command in the shell script (so that qemu replaces the shell
and becomes the direct child of the libguestfs-using program).  If you
don't do this, then the qemu process won't be cleaned up correctly.

Here is an example of a wrapper, where I have built my own copy of
qemu from source:

 #!/bin/sh -
 qemudir=/home/rjones/d/qemu
 exec $qemudir/x86_64-softmmu/qemu-system-x86_64 -L $qemudir/pc-bios "$@"

Save this script as C</tmp/qemu.wrapper> (or wherever), C<chmod +x>,
and then use it by setting the LIBGUESTFS_QEMU environment variable.
For example:

 LIBGUESTFS_QEMU=/tmp/qemu.wrapper guestfish

Note that libguestfs also calls qemu with the -help and -version
options in order to determine features.

Wrappers can also be used to edit the options passed to qemu.  In the
following example, the C<-machine ...> option (C<-machine> and the
following argument) are removed from the command line and replaced
with C<-machine pc,accel=tcg>.  The while loop iterates over the
options until it finds the right one to remove, putting the remaining
options into the C<args> array.

 #!/bin/bash -
 
 i=0
 while [ $# -gt 0 ]; do
     case "$1" in
     -machine)
         shift 2;;
     *)
         args[i]="$1"
         (( i++ ))
         shift ;;
     esac
 done
 
 exec qemu-kvm -machine pc,accel=tcg "${args[@]}"

=head2 ATTACHING TO RUNNING DAEMONS

I<Note (1):> This is B<highly experimental> and has a tendency to eat
babies.  Use with caution.

I<Note (2):> This section explains how to attach to a running daemon
from a low level perspective.  For most users, simply using virt tools
such as L<guestfish(1)> with the I<--live> option will "just work".

=head3 Using guestfs_set_attach_method

By calling L</guestfs_set_attach_method> you can change how the
library connects to the C<guestfsd> daemon in L</guestfs_launch>
(read L</ARCHITECTURE> for some background).

The normal attach method is C<appliance>, where a small appliance is
created containing the daemon, and then the library connects to this.
C<libvirt> or C<libvirt:I<URI>> are alternatives that use libvirt to
start the appliance.

Setting attach method to C<unix:I<path>> (where I<path> is the path of
a Unix domain socket) causes L</guestfs_launch> to connect to an
existing daemon over the Unix domain socket.

The normal use for this is to connect to a running virtual machine
that contains a C<guestfsd> daemon, and send commands so you can read
and write files inside the live virtual machine.

=head3 Using guestfs_add_domain with live flag

L</guestfs_add_domain> provides some help for getting the
correct attach method.  If you pass the C<live> option to this
function, then (if the virtual machine is running) it will
examine the libvirt XML looking for a virtio-serial channel
to connect to:

 <domain>
   ...
   <devices>
     ...
     <channel type='unix'>
       <source mode='bind' path='/path/to/socket'/>
       <target type='virtio' name='org.libguestfs.channel.0'/>
     </channel>
     ...
   </devices>
 </domain>

L</guestfs_add_domain> extracts C</path/to/socket> and sets the attach
method to C<unix:/path/to/socket>.

Some of the libguestfs tools (including guestfish) support a I<--live>
option which is passed through to L</guestfs_add_domain> thus allowing
you to attach to and modify live virtual machines.

The virtual machine needs to have been set up beforehand so that it
has the virtio-serial channel and so that guestfsd is running inside
it.

=head2 ABI GUARANTEE

We guarantee the libguestfs ABI (binary interface), for public,
high-level actions as outlined in this section.  Although we will
deprecate some actions, for example if they get replaced by newer
calls, we will keep the old actions forever.  This allows you the
developer to program in confidence against the libguestfs API.

=head2 BLOCK DEVICE NAMING

In the kernel there is now quite a profusion of schemata for naming
block devices (in this context, by I<block device> I mean a physical
or virtual hard drive).  The original Linux IDE driver used names
starting with C</dev/hd*>.  SCSI devices have historically used a
different naming scheme, C</dev/sd*>.  When the Linux kernel I<libata>
driver became a popular replacement for the old IDE driver
(particularly for SATA devices) those devices also used the
C</dev/sd*> scheme.  Additionally we now have virtual machines with
paravirtualized drivers.  This has created several different naming
systems, such as C</dev/vd*> for virtio disks and C</dev/xvd*> for Xen
PV disks.

As discussed above, libguestfs uses a qemu appliance running an
embedded Linux kernel to access block devices.  We can run a variety
of appliances based on a variety of Linux kernels.

This causes a problem for libguestfs because many API calls use device
or partition names.  Working scripts and the recipe (example) scripts
that we make available over the internet could fail if the naming
scheme changes.

Therefore libguestfs defines C</dev/sd*> as the I<standard naming
scheme>.  Internally C</dev/sd*> names are translated, if necessary,
to other names as required.  For example, under RHEL 5 which uses the
C</dev/hd*> scheme, any device parameter C</dev/sda2> is translated to
C</dev/hda2> transparently.

Note that this I<only> applies to parameters.  The
L</guestfs_list_devices>, L</guestfs_list_partitions> and similar calls
return the true names of the devices and partitions as known to the
appliance, but see L</guestfs_canonical_device_name>.

=head3 DISK LABELS

In libguestfs E<ge> 1.20, you can give a label to a disk when you add
it, using the optional C<label> parameter to L</guestfs_add_drive_opts>.
(Note that disk labels are different from and not related to
filesystem labels).

Not all versions of libguestfs support setting a disk label, and when
it is supported, it is limited to 20 ASCII characters C<[a-zA-Z]>.

When you add a disk with a label, it can either be addressed
using C</dev/sd*>, or using C</dev/disk/guestfs/I<label>>.
Partitions on the disk can be addressed using
C</dev/disk/guestfs/I<label>I<partnum>>.

Listing devices (L</guestfs_list_devices>) and partitions
(L</guestfs_list_partitions>) returns the raw block device name.
However you can use L</guestfs_list_disk_labels> to map disk labels
to raw block device and partition names.

=head3 ALGORITHM FOR BLOCK DEVICE NAME TRANSLATION

Usually this translation is transparent.  However in some (very rare)
cases you may need to know the exact algorithm.  Such cases include
where you use L</guestfs_config> to add a mixture of virtio and IDE
devices to the qemu-based appliance, so have a mixture of C</dev/sd*>
and C</dev/vd*> devices.

The algorithm is applied only to I<parameters> which are known to be
either device or partition names.  Return values from functions such
as L</guestfs_list_devices> are never changed.

=over 4

=item *

Is the string a parameter which is a device or partition name?

=item *

Does the string begin with C</dev/sd>?

=item *

Does the named device exist?  If so, we use that device.
However if I<not> then we continue with this algorithm.

=item *

Replace initial C</dev/sd> string with C</dev/hd>.

For example, change C</dev/sda2> to C</dev/hda2>.

If that named device exists, use it.  If not, continue.

=item *

Replace initial C</dev/sd> string with C</dev/vd>.

If that named device exists, use it.  If not, return an error.

=back

=head3 PORTABILITY CONCERNS WITH BLOCK DEVICE NAMING

Although the standard naming scheme and automatic translation is
useful for simple programs and guestfish scripts, for larger programs
it is best not to rely on this mechanism.

Where possible for maximum future portability programs using
libguestfs should use these future-proof techniques:

=over 4

=item *

Use L</guestfs_list_devices> or L</guestfs_list_partitions> to list
actual device names, and then use those names directly.

Since those device names exist by definition, they will never be
translated.

=item *

Use higher level ways to identify filesystems, such as LVM names,
UUIDs and filesystem labels.

=back

=head2 NULL DISKS

When adding a disk using, eg., L</guestfs_add_drive>, you can
set the filename to C<"/dev/null">.  This string is treated
specially by libguestfs, causing it to add a "null disk".

A null disk has the following properties:

=over 4

=item *

A null disk will appear as a normal device, eg. in
calls to L</guestfs_list_devices>.

=item *

You may add C<"/dev/null"> multiple times.

=item *

You should not try to access a null disk in any way.  For
example, you shouldn't try to read it or mount it.

=back

Null disks are used for three main purposes:

=over 4

=item 1.

Performance testing of libguestfs (see L<guestfs-performance(1)>).

=item 2.

The internal test suite.

=item 3.

If you want to use libguestfs APIs that don't refer to disks, since
libguestfs requires that at least one disk is added, you should add a
null disk.

For example, to test if a feature is available, use code like this:

 guestfs_h *g;
 char **groups = [ "btrfs", NULL ];
 
 g = guestfs_create ();
 guestfs_add_drive (g, "/dev/null");
 guestfs_launch (g);
 if (guestfs_available (g, groups) == 0) {
   // group(s) are available
 } else {
   // group(s) are not available
 }
 guestfs_close (g);

=back

=head2 DISK IMAGE FORMATS

Virtual disks come in a variety of formats.  Some common formats
are listed below.

Note that libguestfs itself is not responsible for handling the disk
format: this is done using L<qemu(1)>.  If support for a particular
format is missing or broken, this has to be fixed in qemu.

=head3 COMMON VIRTUAL DISK IMAGE FORMATS

=over 4

=item I<raw>

Raw format is simply a dump of the sequential bytes of the virtual
hard disk.  There is no header, container, compression or processing
of any sort.

Since raw format requires no translation to read or write, it is both
fast and very well supported by qemu and all other hypervisors.  You
can consider it to be a universal format that any hypervisor can
access.

Raw format files are not compressed and so take up the full space of
the original disk image even when they are empty.  A variation (on
Linux/Unix at least) is to not store ranges of all-zero bytes by
storing the file as a sparse file.  This "variant format" is sometimes
called I<raw sparse>.  Many utilities, including L<virt-sparsify(1)>,
can make raw disk images sparse.

=item I<qcow2>

Qcow2 is the native disk image format used by qemu.  Internally it
uses a two-level directory structure so that only blocks containing
data are stored in the file.  It also has many other features such as
compression, snapshots and backing files.

There are at least two distinct variants of this format, although qemu
(and hence libguestfs) handles both transparently to the user.

=item I<vmdk>

VMDK is VMware's native disk image format.  There are many variations.
Modern qemu (hence libguestfs) supports most variations, but you
should be aware that older versions of qemu had some very bad
data-corrupting bugs in this area.

Note that VMware ESX exposes files with the name C<guest-flat.vmdk>.
These are not VMDK.  They are raw format files which happen to have a
C<.vmdk> extension.

=item I<vdi>

VDI is VirtualBox's native disk image format.  Qemu (hence libguestfs)
has generally good support for this.

=item I<vpc>

=item I<vhd>

VPC (old) and VHD (modern) are the native disk image format of
Microsoft (and previously, Connectix) Virtual PC and Hyper-V.

=item Obsolete formats

The following formats are obsolete and should not be used:
I<qcow> (aka I<qcow1>), I<cow>, I<bochs>.

=back

=head3 DETECTING THE FORMAT OF A DISK IMAGE

Firstly note there is a security issue with auto-detecting the format
of a disk image.  It may or may not apply in your use case.  Read
L</CVE-2010-3851> below.

Libguestfs offers an API to get the format of a disk image
(L</guestfs_disk_format>, and it is safest to use this.

I<Don't> be tempted to try parsing the text / human-readable output of
C<qemu-img> since it cannot be parsed reliably and securely.  Also do
not use the C<file> command since the output of that changes over
time.

=head1 SECURITY

This section discusses security implications of using libguestfs,
particularly with untrusted or malicious guests or disk images.

=head2 GENERAL SECURITY CONSIDERATIONS

Be careful with any files or data that you download from a guest (by
"download" we mean not just the L</guestfs_download> command but any
command that reads files, filenames, directories or anything else from
a disk image).  An attacker could manipulate the data to fool your
program into doing the wrong thing.  Consider cases such as:

=over 4

=item *

the data (file etc) not being present

=item *

being present but empty

=item *

being much larger than normal

=item *

containing arbitrary 8 bit data

=item *

being in an unexpected character encoding

=item *

containing homoglyphs.

=back

=head2 SECURITY OF MOUNTING FILESYSTEMS

When you mount a filesystem under Linux, mistakes in the kernel
filesystem (VFS) module can sometimes be escalated into exploits by
deliberately creating a malicious, malformed filesystem.  These
exploits are very severe for two reasons.  Firstly there are very many
filesystem drivers in the kernel, and many of them are infrequently
used and not much developer attention has been paid to the code.
Linux userspace helps potential crackers by detecting the filesystem
type and automatically choosing the right VFS driver, even if that
filesystem type is obscure or unexpected for the administrator.
Secondly, a kernel-level exploit is like a local root exploit (worse
in some ways), giving immediate and total access to the system right
down to the hardware level.

That explains why you should never mount a filesystem from an
untrusted guest on your host kernel.  How about libguestfs?  We run a
Linux kernel inside a qemu virtual machine, usually running as a
non-root user.  The attacker would need to write a filesystem which
first exploited the kernel, and then exploited either qemu
virtualization (eg. a faulty qemu driver) or the libguestfs protocol,
and finally to be as serious as the host kernel exploit it would need
to escalate its privileges to root.  This multi-step escalation,
performed by a static piece of data, is thought to be extremely hard
to do, although we never say 'never' about security issues.

In any case callers can reduce the attack surface by forcing the
filesystem type when mounting (use L</guestfs_mount_vfs>).

=head2 PROTOCOL SECURITY

The protocol is designed to be secure, being based on RFC 4506 (XDR)
with a defined upper message size.  However a program that uses
libguestfs must also take care - for example you can write a program
that downloads a binary from a disk image and executes it locally, and
no amount of protocol security will save you from the consequences.

=head2 INSPECTION SECURITY

Parts of the inspection API (see L</INSPECTION>) return untrusted
strings directly from the guest, and these could contain any 8 bit
data.  Callers should be careful to escape these before printing them
to a structured file (for example, use HTML escaping if creating a web
page).

Guest configuration may be altered in unusual ways by the
administrator of the virtual machine, and may not reflect reality
(particularly for untrusted or actively malicious guests).  For
example we parse the hostname from configuration files like
C</etc/sysconfig/network> that we find in the guest, but the guest
administrator can easily manipulate these files to provide the wrong
hostname.

The inspection API parses guest configuration using two external
libraries: Augeas (Linux configuration) and hivex (Windows Registry).
Both are designed to be robust in the face of malicious data, although
denial of service attacks are still possible, for example with
oversized configuration files.

=head2 RUNNING UNTRUSTED GUEST COMMANDS

Be very cautious about running commands from the guest.  By running a
command in the guest, you are giving CPU time to a binary that you do
not control, under the same user account as the library, albeit
wrapped in qemu virtualization.  More information and alternatives can
be found in the section L</RUNNING COMMANDS>.

=head2 CVE-2010-3851

L<https://bugzilla.redhat.com/642934>

This security bug concerns the automatic disk format detection that
qemu does on disk images.

A raw disk image is just the raw bytes, there is no header.  Other
disk images like qcow2 contain a special header.  Qemu deals with this
by looking for one of the known headers, and if none is found then
assuming the disk image must be raw.

This allows a guest which has been given a raw disk image to write
some other header.  At next boot (or when the disk image is accessed
by libguestfs) qemu would do autodetection and think the disk image
format was, say, qcow2 based on the header written by the guest.

This in itself would not be a problem, but qcow2 offers many features,
one of which is to allow a disk image to refer to another image
(called the "backing disk").  It does this by placing the path to the
backing disk into the qcow2 header.  This path is not validated and
could point to any host file (eg. "/etc/passwd").  The backing disk is
then exposed through "holes" in the qcow2 disk image, which of course
is completely under the control of the attacker.

In libguestfs this is rather hard to exploit except under two
circumstances:

=over 4

=item 1.

You have enabled the network or have opened the disk in write mode.

=item 2.

You are also running untrusted code from the guest (see
L</RUNNING COMMANDS>).

=back

The way to avoid this is to specify the expected disk format when
adding disks (the optional C<format> option to
L</guestfs_add_drive_opts>).  You should always do this if the disk is
raw format, and it's a good idea for other cases too.
(See also L</DISK IMAGE FORMATS>).

For disks added from libvirt using calls like L</guestfs_add_domain>,
the format is fetched from libvirt and passed through.

For libguestfs tools, use the I<--format> command line parameter as
appropriate.

=head1 CONNECTION MANAGEMENT

=head2 guestfs_h *

C<guestfs_h> is the opaque type representing a connection handle.
Create a handle by calling L</guestfs_create> or
L</guestfs_create_flags>.  Call L</guestfs_close> to free the handle
and release all resources used.

For information on using multiple handles and threads, see the section
L</MULTIPLE HANDLES AND MULTIPLE THREADS> above.

=head2 guestfs_create

 guestfs_h *guestfs_create (void);

Create a connection handle.

On success this returns a non-NULL pointer to a handle.  On error it
returns NULL.

You have to "configure" the handle after creating it.  This includes
calling L</guestfs_add_drive_opts> (or one of the equivalent calls) on
the handle at least once.

After configuring the handle, you have to call L</guestfs_launch>.

You may also want to configure error handling for the handle.  See the
L</ERROR HANDLING> section below.

=head2 guestfs_create_flags

 guestfs_h *guestfs_create_flags (unsigned flags [, ...]);

Create a connection handle, supplying extra flags and
extra arguments to control how the handle is created.

On success this returns a non-NULL pointer to a handle.  On error it
returns NULL.

L</guestfs_create> is equivalent to calling C<guestfs_create_flags(0)>.

The following flags may be logically ORed together.  (Currently
no extra arguments are used).

=over 4

=item C<GUESTFS_CREATE_NO_ENVIRONMENT>

Don't parse any environment variables (such as C<LIBGUESTFS_DEBUG> etc).

You can call L</guestfs_parse_environment> or
L</guestfs_parse_environment_list> afterwards to parse environment
variables.  Alternately, I<don't> call these functions if you want the
handle to be unaffected by environment variables.  See the example below.

The default (if this flag is not given) is to implicitly call
L</guestfs_parse_environment>.

=item C<GUESTFS_CREATE_NO_CLOSE_ON_EXIT>

Don't try to close the handle in an L<atexit(3)> handler if the
program exits without explicitly closing the handle.

The default (if this flag is not given) is to install such an atexit
handler.

=back

=head3 USING C<GUESTFS_CREATE_NO_ENVIRONMENT>

You might use C<GUESTFS_CREATE_NO_ENVIRONMENT> and
an explicit call to L</guestfs_parse_environment> like this:

 guestfs_h *g;
 int r;
 
 g = guestfs_create_flags (GUESTFS_CREATE_NO_ENVIRONMENT);
 if (!g) {
   perror ("guestfs_create_flags");
   exit (EXIT_FAILURE);
 }
 r = guestfs_parse_environment (g);
 if (r == -1)
   exit (EXIT_FAILURE);

Or to create a handle which is unaffected by environment variables,
omit the call to C<guestfs_parse_environment> from the above code.

The above code has another advantage which is that any errors from
parsing the environment are passed through the error handler, whereas
C<guestfs_create> prints errors on stderr and ignores them.

=head2 guestfs_close

 void guestfs_close (guestfs_h *g);

This closes the connection handle and frees up all resources used.
If a close callback was set on the handle, then it is called.

The correct way to close the handle is:

 if (guestfs_shutdown (g) == -1) {
   /* handle write errors here */
 }
 guestfs_close (g);

L</guestfs_shutdown> is only needed if B<all> of the following are true:

=over 4

=item 1

one or more disks were added in read-write mode, I<and>

=item 2

guestfs_launch was called, I<and>

=item 3

you made some changes, I<and>

=item 4

you have a way to handle write errors (eg. by exiting with an
error code or reporting something to the user).

=back

=head1 ERROR HANDLING

API functions can return errors.  For example, almost all functions
that return C<int> will return C<-1> to indicate an error.

Additional information is available for errors: an error message
string and optionally an error number (errno) if the thing that failed
was a system call.

You can get at the additional information about the last error on the
handle by calling L</guestfs_last_error>, L</guestfs_last_errno>,
and/or by setting up an error handler with
L</guestfs_set_error_handler>.

When the handle is created, a default error handler is installed which
prints the error message string to C<stderr>.  For small short-running
command line programs it is sufficient to do:

 if (guestfs_launch (g) == -1)
   exit (EXIT_FAILURE);

since the default error handler will ensure that an error message has
been printed to C<stderr> before the program exits.

For other programs the caller will almost certainly want to install an
alternate error handler or do error handling in-line as in the example
below.  The non-C language bindings all install NULL error handlers
and turn errors into exceptions using code similar to this:

 const char *msg;
 int errnum;
 
 /* This disables the default behaviour of printing errors
    on stderr. */
 guestfs_set_error_handler (g, NULL, NULL);
 
 if (guestfs_launch (g) == -1) {
   /* Examine the error message and print it, throw it,
      etc. */
   msg = guestfs_last_error (g);
   errnum = guestfs_last_errno (g);
 
   fprintf (stderr, "%s", msg);
   if (errnum != 0)
     fprintf (stderr, ": %s", strerror (errnum));
   fprintf (stderr, "\n");
 
   /* ... */
 }

L</guestfs_create> returns C<NULL> if the handle cannot be created,
and because there is no handle if this happens there is no way to get
additional error information.  Since libguestfs E<ge> 1.20, you can
use L</guestfs_create_flags> to properly deal with errors during
handle creation, although the vast majority of programs can continue
to use L</guestfs_create> and not worry about this situation.

Out of memory errors are handled differently.  The default action is
to call L<abort(3)>.  If this is undesirable, then you can set a
handler using L</guestfs_set_out_of_memory_handler>.

=head2 guestfs_last_error

 const char *guestfs_last_error (guestfs_h *g);

This returns the last error message that happened on C<g>.  If
there has not been an error since the handle was created, then this
returns C<NULL>.

Note the returned string does I<not> have a newline character at the
end.  Most error messages are single lines.  Some are split over
multiple lines and contain C<\n> characters within the string but not
at the end.

The lifetime of the returned string is until the next error occurs
on the same handle, or L</guestfs_close> is called.  If you need
to keep it longer, copy it.

=head2 guestfs_last_errno

 int guestfs_last_errno (guestfs_h *g);

This returns the last error number (errno) that happened on C<g>.

If successful, an errno integer not equal to zero is returned.

In many cases the special errno C<ENOTSUP> is returned if you tried to
call a function or use a feature which is not supported.

If no error number is available, this returns 0.  This call can return
0 in three situations:

=over 4

=item 1.

There has not been any error on the handle.

=item 2.

There has been an error but the errno was meaningless.  This
corresponds to the case where the error did not come from a
failed system call, but for some other reason.

=item 3.

There was an error from a failed system call, but for some
reason the errno was not captured and returned.  This usually
indicates a bug in libguestfs.

=back

Libguestfs tries to convert the errno from inside the applicance into
a corresponding errno for the caller (not entirely trivial: the
appliance might be running a completely different operating system
from the library and error numbers are not standardized across
Un*xen).  If this could not be done, then the error is translated to
C<EINVAL>.  In practice this should only happen in very rare
circumstances.

=head2 guestfs_set_error_handler

 typedef void (*guestfs_error_handler_cb) (guestfs_h *g,
                                           void *opaque,
                                           const char *msg);
 void guestfs_set_error_handler (guestfs_h *g,
                                 guestfs_error_handler_cb cb,
                                 void *opaque);

The callback C<cb> will be called if there is an error.  The
parameters passed to the callback are an opaque data pointer and the
error message string.

C<errno> is not passed to the callback.  To get that the callback must
call L</guestfs_last_errno>.

Note that the message string C<msg> is freed as soon as the callback
function returns, so if you want to stash it somewhere you must make
your own copy.

The default handler prints messages on C<stderr>.

If you set C<cb> to C<NULL> then I<no> handler is called.

=head2 guestfs_get_error_handler

 guestfs_error_handler_cb guestfs_get_error_handler (guestfs_h *g,
                                                     void **opaque_rtn);

Returns the current error handler callback.

=head2 guestfs_set_out_of_memory_handler

 typedef void (*guestfs_abort_cb) (void);
 void guestfs_set_out_of_memory_handler (guestfs_h *g,
                                         guestfs_abort_cb);

The callback C<cb> will be called if there is an out of memory
situation.  I<Note this callback must not return>.

The default is to call L<abort(3)>.

You cannot set C<cb> to C<NULL>.  You can't ignore out of memory
situations.

=head2 guestfs_get_out_of_memory_handler

 guestfs_abort_fn guestfs_get_out_of_memory_handler (guestfs_h *g);

This returns the current out of memory handler.

=head1 API CALLS

__ACTIONS__

=head1 STRUCTURES

__STRUCTS__

=head1 AVAILABILITY

=head2 GROUPS OF FUNCTIONALITY IN THE APPLIANCE

Using L</guestfs_available> you can test availability of
the following groups of functions.  This test queries the
appliance to see if the appliance you are currently using
supports the functionality.

__AVAILABILITY__

=head2 FILESYSTEM AVAILABLE

The L</guestfs_filesystem_available> call tests whether a
filesystem type is supported by the appliance kernel.

This is mainly useful as a negative test.  If this returns true,
it doesn't mean that a particular filesystem can be mounted,
since filesystems can fail for other reasons such as it being
a later version of the filesystem, or having incompatible features.

=head2 GUESTFISH supported COMMAND

In L<guestfish(3)> there is a handy interactive command
C<supported> which prints out the available groups and
whether they are supported by this build of libguestfs.
Note however that you have to do C<run> first.

=head2 SINGLE CALLS AT COMPILE TIME

Since version 1.5.8, C<E<lt>guestfs.hE<gt>> defines symbols
for each C API function, such as:

 #define LIBGUESTFS_HAVE_DD 1

if L</guestfs_dd> is available.

Before version 1.5.8, if you needed to test whether a single
libguestfs function is available at compile time, we recommended using
build tools such as autoconf or cmake.  For example in autotools you
could use:

 AC_CHECK_LIB([guestfs],[guestfs_create])
 AC_CHECK_FUNCS([guestfs_dd])

which would result in C<HAVE_GUESTFS_DD> being either defined
or not defined in your program.

=head2 SINGLE CALLS AT RUN TIME

Testing at compile time doesn't guarantee that a function really
exists in the library.  The reason is that you might be dynamically
linked against a previous I<libguestfs.so> (dynamic library)
which doesn't have the call.  This situation unfortunately results
in a segmentation fault, which is a shortcoming of the C dynamic
linking system itself.

You can use L<dlopen(3)> to test if a function is available
at run time, as in this example program (note that you still
need the compile time check as well):

 #include <stdio.h>
 #include <stdlib.h>
 #include <unistd.h>
 #include <dlfcn.h>
 #include <guestfs.h>
 
 main ()
 {
 #ifdef LIBGUESTFS_HAVE_DD
   void *dl;
   int has_function;
 
   /* Test if the function guestfs_dd is really available. */
   dl = dlopen (NULL, RTLD_LAZY);
   if (!dl) {
     fprintf (stderr, "dlopen: %s\n", dlerror ());
     exit (EXIT_FAILURE);
   }
   has_function = dlsym (dl, "guestfs_dd") != NULL;
   dlclose (dl);
 
   if (!has_function)
     printf ("this libguestfs.so does NOT have guestfs_dd function\n");
   else {
     printf ("this libguestfs.so has guestfs_dd function\n");
     /* Now it's safe to call
     guestfs_dd (g, "foo", "bar");
     */
   }
 #else
   printf ("guestfs_dd function was not found at compile time\n");
 #endif
  }

You may think the above is an awful lot of hassle, and it is.
There are other ways outside of the C linking system to ensure
that this kind of incompatibility never arises, such as using
package versioning:

 Requires: libguestfs >= 1.0.80

=head1 CALLS WITH OPTIONAL ARGUMENTS

A recent feature of the API is the introduction of calls which take
optional arguments.  In C these are declared 3 ways.  The main way is
as a call which takes variable arguments (ie. C<...>), as in this
example:

 int guestfs_add_drive_opts (guestfs_h *g, const char *filename, ...);

Call this with a list of optional arguments, terminated by C<-1>.
So to call with no optional arguments specified:

 guestfs_add_drive_opts (g, filename, -1);

With a single optional argument:

 guestfs_add_drive_opts (g, filename,
                         GUESTFS_ADD_DRIVE_OPTS_FORMAT, "qcow2",
                         -1);

With two:

 guestfs_add_drive_opts (g, filename,
                         GUESTFS_ADD_DRIVE_OPTS_FORMAT, "qcow2",
                         GUESTFS_ADD_DRIVE_OPTS_READONLY, 1,
                         -1);

and so forth.  Don't forget the terminating C<-1> otherwise
Bad Things will happen!

=head2 USING va_list FOR OPTIONAL ARGUMENTS

The second variant has the same name with the suffix C<_va>, which
works the same way but takes a C<va_list>.  See the C manual for
details.  For the example function, this is declared:

 int guestfs_add_drive_opts_va (guestfs_h *g, const char *filename,
                                va_list args);

=head2 CONSTRUCTING OPTIONAL ARGUMENTS

The third variant is useful where you need to construct these
calls.  You pass in a structure where you fill in the optional
fields.  The structure has a bitmask as the first element which
you must set to indicate which fields you have filled in.  For
our example function the structure and call are declared:

 struct guestfs_add_drive_opts_argv {
   uint64_t bitmask;
   int readonly;
   const char *format;
   /* ... */
 };
 int guestfs_add_drive_opts_argv (guestfs_h *g, const char *filename,
              const struct guestfs_add_drive_opts_argv *optargs);

You could call it like this:

 struct guestfs_add_drive_opts_argv optargs = {
   .bitmask = GUESTFS_ADD_DRIVE_OPTS_READONLY_BITMASK |
              GUESTFS_ADD_DRIVE_OPTS_FORMAT_BITMASK,
   .readonly = 1,
   .format = "qcow2"
 };
 
 guestfs_add_drive_opts_argv (g, filename, &optargs);

Notes:

=over 4

=item *

The C<_BITMASK> suffix on each option name when specifying the
bitmask.

=item *

You do not need to fill in all fields of the structure.

=item *

There must be a one-to-one correspondence between fields of the
structure that are filled in, and bits set in the bitmask.

=back

=head2 OPTIONAL ARGUMENTS IN OTHER LANGUAGES

In other languages, optional arguments are expressed in the
way that is natural for that language.  We refer you to the
language-specific documentation for more details on that.

For guestfish, see L<guestfish(1)/OPTIONAL ARGUMENTS>.

=head1 EVENTS

=head2 SETTING CALLBACKS TO HANDLE EVENTS

B<Note:> This section documents the generic event mechanism introduced
in libguestfs 1.10, which you should use in new code if possible.  The
old functions C<guestfs_set_log_message_callback>,
C<guestfs_set_subprocess_quit_callback>,
C<guestfs_set_launch_done_callback>, C<guestfs_set_close_callback> and
C<guestfs_set_progress_callback> are no longer documented in this
manual page.  Because of the ABI guarantee, the old functions continue
to work.

Handles generate events when certain things happen, such as log
messages being generated, progress messages during long-running
operations, or the handle being closed.  The API calls described below
let you register a callback to be called when events happen.  You can
register multiple callbacks (for the same, different or overlapping
sets of events), and individually remove callbacks.  If callbacks are
not removed, then they remain in force until the handle is closed.

In the current implementation, events are only generated
synchronously: that means that events (and hence callbacks) can only
happen while you are in the middle of making another libguestfs call.
The callback is called in the same thread.

Events may contain a payload, usually nothing (void), an array of 64
bit unsigned integers, or a message buffer.  Payloads are discussed
later on.

=head2 CLASSES OF EVENTS

=over 4

=item GUESTFS_EVENT_CLOSE
(payload type: void)

The callback function will be called while the handle is being closed
(synchronously from L</guestfs_close>).

Note that libguestfs installs an L<atexit(3)> handler to try to clean
up handles that are open when the program exits.  This means that this
callback might be called indirectly from L<exit(3)>, which can cause
unexpected problems in higher-level languages (eg. if your HLL
interpreter has already been cleaned up by the time this is called,
and if your callback then jumps into some HLL function).

If no callback is registered: the handle is closed without any
callback being invoked.

=item GUESTFS_EVENT_SUBPROCESS_QUIT
(payload type: void)

The callback function will be called when the child process quits,
either asynchronously or if killed by L</guestfs_kill_subprocess>.
(This corresponds to a transition from any state to the CONFIG state).

If no callback is registered: the event is ignored.

=item GUESTFS_EVENT_LAUNCH_DONE
(payload type: void)

The callback function will be called when the child process becomes
ready first time after it has been launched.  (This corresponds to a
transition from LAUNCHING to the READY state).

If no callback is registered: the event is ignored.

=item GUESTFS_EVENT_PROGRESS
(payload type: array of 4 x uint64_t)

Some long-running operations can generate progress messages.  If
this callback is registered, then it will be called each time a
progress message is generated (usually two seconds after the
operation started, and three times per second thereafter until
it completes, although the frequency may change in future versions).

The callback receives in the payload four unsigned 64 bit numbers
which are (in order): C<proc_nr>, C<serial>, C<position>, C<total>.

The units of C<total> are not defined, although for some
operations C<total> may relate in some way to the amount of
data to be transferred (eg. in bytes or megabytes), and
C<position> may be the portion which has been transferred.

The only defined and stable parts of the API are:

=over 4

=item *

The callback can display to the user some type of progress bar or
indicator which shows the ratio of C<position>:C<total>.

=item *

0 E<lt>= C<position> E<lt>= C<total>

=item *

If any progress notification is sent during a call, then a final
progress notification is always sent when C<position> = C<total>
(I<unless> the call fails with an error).

This is to simplify caller code, so callers can easily set the
progress indicator to "100%" at the end of the operation, without
requiring special code to detect this case.

=item *

For some calls we are unable to estimate the progress of the call, but
we can still generate progress messages to indicate activity.  This is
known as "pulse mode", and is directly supported by certain progress
bar implementations (eg. GtkProgressBar).

For these calls, zero or more progress messages are generated with
C<position = 0> and C<total = 1>, followed by a final message with
C<position = total = 1>.

As noted above, if the call fails with an error then the final message
may not be generated.

=back

The callback also receives the procedure number (C<proc_nr>) and
serial number (C<serial>) of the call.  These are only useful for
debugging protocol issues, and the callback can normally ignore them.
The callback may want to print these numbers in error messages or
debugging messages.

If no callback is registered: progress messages are discarded.

=item GUESTFS_EVENT_APPLIANCE
(payload type: message buffer)

The callback function is called whenever a log message is generated by
qemu, the appliance kernel, guestfsd (daemon), or utility programs.

If the verbose flag (L</guestfs_set_verbose>) is set before launch
(L</guestfs_launch>) then additional debug messages are generated.

If no callback is registered: the messages are discarded unless the
verbose flag is set in which case they are sent to stderr.  You can
override the printing of verbose messages to stderr by setting up a
callback.

=item GUESTFS_EVENT_LIBRARY
(payload type: message buffer)

The callback function is called whenever a log message is generated by
the library part of libguestfs.

If the verbose flag (L</guestfs_set_verbose>) is set then additional
debug messages are generated.

If no callback is registered: the messages are discarded unless the
verbose flag is set in which case they are sent to stderr.  You can
override the printing of verbose messages to stderr by setting up a
callback.

=item GUESTFS_EVENT_TRACE
(payload type: message buffer)

The callback function is called whenever a trace message is generated.
This only applies if the trace flag (L</guestfs_set_trace>) is set.

If no callback is registered: the messages are sent to stderr.  You
can override the printing of trace messages to stderr by setting up a
callback.

=item GUESTFS_EVENT_ENTER
(payload type: function name)

The callback function is called whenever a libguestfs function
is entered.

The payload is a string which contains the name of the function
that we are entering (not including C<guestfs_> prefix).

Note that libguestfs functions can call themselves, so you may
see many events from a single call.  A few libguestfs functions
do not generate this event.

If no callback is registered: the event is ignored.

=item GUESTFS_EVENT_LIBVIRT_AUTH
(payload type: libvirt URI)

For any API function that opens a libvirt connection, this
event may be generated to indicate that libvirt demands
authentication information.  See L</LIBVIRT AUTHENTICATION> below.

If no callback is registered: C<virConnectAuthPtrDefault> is
used (suitable for command-line programs only).

=back

=head2 EVENT API

=head3 guestfs_set_event_callback

 int guestfs_set_event_callback (guestfs_h *g,
                                 guestfs_event_callback cb,
                                 uint64_t event_bitmask,
                                 int flags,
                                 void *opaque);

This function registers a callback (C<cb>) for all event classes
in the C<event_bitmask>.

For example, to register for all log message events, you could call
this function with the bitmask
C<GUESTFS_EVENT_APPLIANCE|GUESTFS_EVENT_LIBRARY>.  To register a
single callback for all possible classes of events, use
C<GUESTFS_EVENT_ALL>.

C<flags> should always be passed as 0.

C<opaque> is an opaque pointer which is passed to the callback.  You
can use it for any purpose.

The return value is the event handle (an integer) which you can use to
delete the callback (see below).

If there is an error, this function returns C<-1>, and sets the error
in the handle in the usual way (see L</guestfs_last_error> etc.)

Callbacks remain in effect until they are deleted, or until the handle
is closed.

In the case where multiple callbacks are registered for a particular
event class, all of the callbacks are called.  The order in which
multiple callbacks are called is not defined.

=head3 guestfs_delete_event_callback

 void guestfs_delete_event_callback (guestfs_h *g, int event_handle);

Delete a callback that was previously registered.  C<event_handle>
should be the integer that was returned by a previous call to
C<guestfs_set_event_callback> on the same handle.

=head3 guestfs_event_callback

 typedef void (*guestfs_event_callback) (
                  guestfs_h *g,
                  void *opaque,
                  uint64_t event,
                  int event_handle,
                  int flags,
                  const char *buf, size_t buf_len,
                  const uint64_t *array, size_t array_len);

This is the type of the event callback function that you have to
provide.

The basic parameters are: the handle (C<g>), the opaque user pointer
(C<opaque>), the event class (eg. C<GUESTFS_EVENT_PROGRESS>), the
event handle, and C<flags> which in the current API you should ignore.

The remaining parameters contain the event payload (if any).  Each
event may contain a payload, which usually relates to the event class,
but for future proofing your code should be written to handle any
payload for any event class.

C<buf> and C<buf_len> contain a message buffer (if C<buf_len == 0>,
then there is no message buffer).  Note that this message buffer can
contain arbitrary 8 bit data, including NUL bytes.

C<array> and C<array_len> is an array of 64 bit unsigned integers.  At
the moment this is only used for progress messages.

=head2 EXAMPLE: CAPTURING LOG MESSAGES

One motivation for the generic event API was to allow GUI programs to
capture debug and other messages.  In libguestfs E<le> 1.8 these were
sent unconditionally to C<stderr>.

Events associated with log messages are: C<GUESTFS_EVENT_LIBRARY>,
C<GUESTFS_EVENT_APPLIANCE> and C<GUESTFS_EVENT_TRACE>.  (Note that
error messages are not events; you must capture error messages
separately).

Programs have to set up a callback to capture the classes of events of
interest:

 int eh =
   guestfs_set_event_callback
     (g, message_callback,
      GUESTFS_EVENT_LIBRARY|GUESTFS_EVENT_APPLIANCE|
      GUESTFS_EVENT_TRACE,
      0, NULL) == -1)
 if (eh == -1) {
   // handle error in the usual way
 }

The callback can then direct messages to the appropriate place.  In
this example, messages are directed to syslog:

 static void
 message_callback (
         guestfs_h *g,
         void *opaque,
         uint64_t event,
         int event_handle,
         int flags,
         const char *buf, size_t buf_len,
         const uint64_t *array, size_t array_len)
 {
   const int priority = LOG_USER|LOG_INFO;
   if (buf_len > 0)
     syslog (priority, "event 0x%lx: %s", event, buf);
 }

=head2 LIBVIRT AUTHENTICATION

Some libguestfs API calls can open libvirt connections.  Currently the
only ones are L</guestfs_add_domain>; and L</guestfs_launch> if the
libvirt attach-method has been selected.  Libvirt connections may
require authentication, for example if they need to access a remote
server or to access root services from non-root.  Libvirt
authentication happens via a callback mechanism, see
L<http://libvirt.org/guide/html/Application_Development_Guide-Connections.html>

You may provide libvirt authentication data by registering a callback
for events of type C<GUESTFS_EVENT_LIBVIRT_AUTH>.

If no such event is registered, then libguestfs uses a libvirt
function that provides command-line prompts
(C<virConnectAuthPtrDefault>).  This is only suitable for command-line
libguestfs programs.

To provide authentication, first call
L</guestfs_set_libvirt_supported_credentials> with the list of
credentials your program knows how to provide.  Second, register a
callback for the C<GUESTFS_EVENT_LIBVIRT_AUTH> event.  The event
handler will be called when libvirt is requesting authentication
information.

In the event handler, call
L</guestfs_get_libvirt_requested_credentials> to get a list of the
credentials that libvirt is asking for.  You then need to ask (eg. the
user) for each credential, and call
L</guestfs_set_libvirt_requested_credential> with the answer.  Note
that for each credential, additional information may be available
via the calls
L</guestfs_get_libvirt_requested_credential_prompt>,
L</guestfs_get_libvirt_requested_credential_challenge> or
L</guestfs_get_libvirt_requested_credential_defresult>.

The example program below should make this clearer.

There is also a more substantial working example program supplied with
the libguestfs sources, called C<libvirt_auth.c>.

 main ()
 {
   guestfs_h *g;
   char *creds[] = { "authname", "passphrase", NULL };
   int r, eh;
 
   g = guestfs_create ();
   if (!g) exit (EXIT_FAILURE);
 
   /* Tell libvirt what credentials the program supports. */
   r = guestfs_set_libvirt_supported_credentials (g, creds);
   if (r == -1)
     exit (EXIT_FAILURE);
 
   /* Set up the event handler. */
   eh = guestfs_set_event_callback (
       g, do_auth,
       GUESTFS_EVENT_LIBVIRT_AUTH, 0, NULL);
   if (eh == -1)
     exit (EXIT_FAILURE);
 
   /* An example of a call that may ask for credentials. */
   r = guestfs_add_domain (
       g, "dom",
       GUESTFS_ADD_DOMAIN_LIBVIRTURI, "qemu:///system",
       -1);
   if (r == -1)
     exit (EXIT_FAILURE);
 
   exit (EXIT_SUCCESS);
 }
 
 static void
 do_auth (guestfs_h *g,
          void *opaque,
          uint64_t event,
          int event_handle,
          int flags,
          const char *buf, size_t buf_len,
          const uint64_t *array, size_t array_len)
 {
   char **creds;
   size_t i;
   char *prompt;
   char *reply;
   size_t replylen;
   int r;
 
   // buf will be the libvirt URI.  buf_len may be ignored.
   printf ("Authentication required for libvirt conn '%s'\n",
           buf);
 
   // Ask libguestfs what credentials libvirt is demanding.
   creds = guestfs_get_libvirt_requested_credentials (g);
   if (creds == NULL)
     exit (EXIT_FAILURE);
 
   // Now ask the user for answers.
   for (i = 0; creds[i] != NULL; ++i)
   {
     if (strcmp (creds[i], "authname") == 0 ||
         strcmp (creds[i], "passphrase") == 0)
     {
       prompt =
         guestfs_get_libvirt_requested_credential_prompt (g, i);
       if (prompt && strcmp (prompt, "") != 0)
         printf ("%s: ", prompt);
       free (prompt);
 
       // Some code here to ask for the credential.
       // ...
       // Put the reply in 'reply', length 'replylen' (bytes).
 
      r = guestfs_set_libvirt_requested_credential (g, i,
          reply, replylen);
      if (r == -1)
        exit (EXIT_FAILURE);
     }
 
     free (creds[i]);
   }
 
   free (creds);
 }

=head1 CANCELLING LONG TRANSFERS

Some operations can be cancelled by the caller while they are in
progress.  Currently only operations that involve uploading or
downloading data can be cancelled (technically: operations that have
C<FileIn> or C<FileOut> parameters in the generator).

=head2 guestfs_user_cancel

 void guestfs_user_cancel (guestfs_h *g);

C<guestfs_user_cancel> cancels the current upload or download
operation.

Unlike most other libguestfs calls, this function is signal safe and
thread safe.  You can call it from a signal handler or from another
thread, without needing to do any locking.

The transfer that was in progress (if there is one) will stop shortly
afterwards, and will return an error.  The errno (see
L</guestfs_last_errno>) is set to C<EINTR>, so you can test for this
to find out if the operation was cancelled or failed because of
another error.

No cleanup is performed: for example, if a file was being uploaded
then after cancellation there may be a partially uploaded file.  It is
the caller's responsibility to clean up if necessary.

There are two common places that you might call C<guestfs_user_cancel>.

In an interactive text-based program, you might call it from a
C<SIGINT> signal handler so that pressing C<^C> cancels the current
operation.  (You also need to call L</guestfs_set_pgroup> so that
child processes don't receive the C<^C> signal).

In a graphical program, when the main thread is displaying a progress
bar with a cancel button, wire up the cancel button to call this
function.

=head1 PRIVATE DATA AREA

You can attach named pieces of private data to the libguestfs handle,
fetch them by name, and walk over them, for the lifetime of the
handle.  This is called the private data area and is only available
from the C API.

To attach a named piece of data, use the following call:

 void guestfs_set_private (guestfs_h *g, const char *key, void *data);

C<key> is the name to associate with this data, and C<data> is an
arbitrary pointer (which can be C<NULL>).  Any previous item with the
same key is overwritten.

You can use any C<key> you want, but your key should I<not> start with
an underscore character.  Keys beginning with an underscore character
are reserved for internal libguestfs purposes (eg. for implementing
language bindings).  It is recommended that you prefix the key with
some unique string to avoid collisions with other users.

To retrieve the pointer, use:

 void *guestfs_get_private (guestfs_h *g, const char *key);

This function returns C<NULL> if either no data is found associated
with C<key>, or if the user previously set the C<key>'s C<data>
pointer to C<NULL>.

Libguestfs does not try to look at or interpret the C<data> pointer in
any way.  As far as libguestfs is concerned, it need not be a valid
pointer at all.  In particular, libguestfs does I<not> try to free the
data when the handle is closed.  If the data must be freed, then the
caller must either free it before calling L</guestfs_close> or must
set up a close callback to do it (see L</GUESTFS_EVENT_CLOSE>).

To walk over all entries, use these two functions:

 void *guestfs_first_private (guestfs_h *g, const char **key_rtn);

 void *guestfs_next_private (guestfs_h *g, const char **key_rtn);

C<guestfs_first_private> returns the first key, pointer pair ("first"
does not have any particular meaning -- keys are not returned in any
defined order).  A pointer to the key is returned in C<*key_rtn> and
the corresponding data pointer is returned from the function.  C<NULL>
is returned if there are no keys stored in the handle.

C<guestfs_next_private> returns the next key, pointer pair.  The
return value of this function is also C<NULL> is there are no further
entries to return.

Notes about walking over entries:

=over 4

=item *

You must not call C<guestfs_set_private> while walking over the
entries.

=item *

The handle maintains an internal iterator which is reset when you call
C<guestfs_first_private>.  This internal iterator is invalidated when
you call C<guestfs_set_private>.

=item *

If you have set the data pointer associated with a key to C<NULL>, ie:

 guestfs_set_private (g, key, NULL);

then that C<key> is not returned when walking.

=item *

C<*key_rtn> is only valid until the next call to
C<guestfs_first_private>, C<guestfs_next_private> or
C<guestfs_set_private>.

=back

The following example code shows how to print all keys and data
pointers that are associated with the handle C<g>:

 const char *key;
 void *data = guestfs_first_private (g, &key);
 while (data != NULL)
   {
     printf ("key = %s, data = %p\n", key, data);
     data = guestfs_next_private (g, &key);
   }

More commonly you are only interested in keys that begin with an
application-specific prefix C<foo_>.  Modify the loop like so:

 const char *key;
 void *data = guestfs_first_private (g, &key);
 while (data != NULL)
   {
     if (strncmp (key, "foo_", strlen ("foo_")) == 0)
       printf ("key = %s, data = %p\n", key, data);
     data = guestfs_next_private (g, &key);
   }

If you need to modify keys while walking, then you have to jump back
to the beginning of the loop.  For example, to delete all keys
prefixed with C<foo_>:

  const char *key;
  void *data;
 again:
  data = guestfs_first_private (g, &key);
  while (data != NULL)
    {
      if (strncmp (key, "foo_", strlen ("foo_")) == 0)
        {
          guestfs_set_private (g, key, NULL);
          /* note that 'key' pointer is now invalid, and so is
             the internal iterator */
          goto again;
        }
      data = guestfs_next_private (g, &key);
    }

Note that the above loop is guaranteed to terminate because the keys
are being deleted, but other manipulations of keys within the loop
might not terminate unless you also maintain an indication of which
keys have been visited.

=head1 SYSTEMTAP

The libguestfs C library can be probed using systemtap or DTrace.
This is true of any library, not just libguestfs.  However libguestfs
also contains static markers to help in probing internal operations.

You can list all the static markers by doing:

 stap -l 'process("/usr/lib*/libguestfs.so.0")
              .provider("guestfs").mark("*")'

B<Note:> These static markers are I<not> part of the stable API and
may change in future versions.

=head2 SYSTEMTAP SCRIPT EXAMPLE

This script contains examples of displaying both the static markers
and some ordinary C entry points:

 global last;
 
 function display_time () {
       now = gettimeofday_us ();
       delta = 0;
       if (last > 0)
             delta = now - last;
       last = now;
 
       printf ("%d (+%d):", now, delta);
 }
 
 probe begin {
       last = 0;
       printf ("ready\n");
 }
 
 /* Display all calls to static markers. */
 probe process("/usr/lib*/libguestfs.so.0")
           .provider("guestfs").mark("*") ? {
       display_time();
       printf ("\t%s %s\n", $$name, $$parms);
 }
 
 /* Display all calls to guestfs_mkfs* functions. */
 probe process("/usr/lib*/libguestfs.so.0")
           .function("guestfs_mkfs*") ? {
       display_time();
       printf ("\t%s %s\n", probefunc(), $$parms);
 }

The script above can be saved to C<test.stap> and run using the
L<stap(1)> program.  Note that you either have to be root, or you have
to add yourself to several special stap groups.  Consult the systemtap
documentation for more information.

 # stap /tmp/test.stap
 ready

In another terminal, run a guestfish command such as this:

 guestfish -N fs

In the first terminal, stap trace output similar to this is shown:

 1318248056692655 (+0):	launch_start
 1318248056692850 (+195):       launch_build_appliance_start
 1318248056818285 (+125435):    launch_build_appliance_end
 1318248056838059 (+19774):     launch_run_qemu
 1318248061071167 (+4233108):   launch_end
 1318248061280324 (+209157):    guestfs_mkfs g=0x1024ab0 fstype=0x46116f device=0x1024e60

=begin html

<!-- old anchor for the next section -->
<a name="state_machine_and_low_level_event_api"/>

=end html

=head1 ARCHITECTURE

Internally, libguestfs is implemented by running an appliance (a
special type of small virtual machine) using L<qemu(1)>.  Qemu runs as
a child process of the main program.

  ___________________
 /                   \
 | main program      |
 |                   |
 |                   |           child process / appliance
 |                   |           __________________________
 |                   |          / qemu                     \
 +-------------------+   RPC    |      +-----------------+ |
 | libguestfs     <--------------------> guestfsd        | |
 |                   |          |      +-----------------+ |
 \___________________/          |      | Linux kernel    | |
                                |      +--^--------------+ |
                                \_________|________________/
                                          |
                                   _______v______
                                  /              \
                                  | Device or    |
                                  | disk image   |
                                  \______________/

The library, linked to the main program, creates the child process and
hence the appliance in the L</guestfs_launch> function.

Inside the appliance is a Linux kernel and a complete stack of
userspace tools (such as LVM and ext2 programs) and a small
controlling daemon called L</guestfsd>.  The library talks to
L</guestfsd> using remote procedure calls (RPC).  There is a mostly
one-to-one correspondence between libguestfs API calls and RPC calls
to the daemon.  Lastly the disk image(s) are attached to the qemu
process which translates device access by the appliance's Linux kernel
into accesses to the image.

A common misunderstanding is that the appliance "is" the virtual
machine.  Although the disk image you are attached to might also be
used by some virtual machine, libguestfs doesn't know or care about
this.  (But you will care if both libguestfs's qemu process and your
virtual machine are trying to update the disk image at the same time,
since these usually results in massive disk corruption).

=head1 STATE MACHINE

libguestfs uses a state machine to model the child process:

                         |
          guestfs_create / guestfs_create_flags
                         |
                         |
                     ____V_____
                    /          \
                    |  CONFIG  |
                    \__________/
                       ^   ^  \
                       |    \  \ guestfs_launch
                       |    _\__V______
                       |   /           \
                       |   | LAUNCHING |
                       |   \___________/
                       |       /
                       |  guestfs_launch
                       |     /
                     __|____V
                    /        \
                    | READY  |
                    \________/

The normal transitions are (1) CONFIG (when the handle is created, but
there is no child process), (2) LAUNCHING (when the child process is
booting up), (3) READY meaning the appliance is up, actions can be
issued to, and carried out by, the child process.

The guest may be killed by L</guestfs_kill_subprocess>, or may die
asynchronously at any time (eg. due to some internal error), and that
causes the state to transition back to CONFIG.

Configuration commands for qemu such as L</guestfs_set_path> can only
be issued when in the CONFIG state.

The API offers one call that goes from CONFIG through LAUNCHING to
READY.  L</guestfs_launch> blocks until the child process is READY to
accept commands (or until some failure or timeout).
L</guestfs_launch> internally moves the state from CONFIG to LAUNCHING
while it is running.

API actions such as L</guestfs_mount> can only be issued when in the
READY state.  These API calls block waiting for the command to be
carried out.  There are no non-blocking versions, and no way to issue
more than one command per handle at the same time.

Finally, the child process sends asynchronous messages back to the
main program, such as kernel log messages.  You can register a
callback to receive these messages.

=head1 INTERNALS

=head2 APPLIANCE BOOT PROCESS

This process has evolved and continues to evolve.  The description
here corresponds only to the current version of libguestfs and is
provided for information only.

In order to follow the stages involved below, enable libguestfs
debugging (set the environment variable C<LIBGUESTFS_DEBUG=1>).

=over 4

=item Create the appliance

C<febootstrap-supermin-helper> is invoked to create the kernel, a
small initrd and the appliance.

The appliance is cached in C</var/tmp/.guestfs-E<lt>UIDE<gt>> (or in
another directory if C<LIBGUESTFS_CACHEDIR> or C<TMPDIR> are set).

For a complete description of how the appliance is created and cached,
read the L<febootstrap(8)> and L<febootstrap-supermin-helper(8)> man
pages.

=item Start qemu and boot the kernel

qemu is invoked to boot the kernel.

=item Run the initrd

C<febootstrap-supermin-helper> builds a small initrd.  The initrd is
not the appliance.  The purpose of the initrd is to load enough kernel
modules in order that the appliance itself can be mounted and started.

The initrd is a cpio archive called
C</var/tmp/.guestfs-E<lt>UIDE<gt>/initrd>.

When the initrd has started you will see messages showing that kernel
modules are being loaded, similar to this:

 febootstrap: ext2 mini initrd starting up
 febootstrap: mounting /sys
 febootstrap: internal insmod libcrc32c.ko
 febootstrap: internal insmod crc32c-intel.ko

=item Find and mount the appliance device

The appliance is a sparse file containing an ext2 filesystem which
contains a familiar (although reduced in size) Linux operating system.
It would normally be called C</var/tmp/.guestfs-E<lt>UIDE<gt>/root>.

The regular disks being inspected by libguestfs are the first
devices exposed by qemu (eg. as C</dev/vda>).

The last disk added to qemu is the appliance itself (eg. C</dev/vdb>
if there was only one regular disk).

Thus the final job of the initrd is to locate the appliance disk,
mount it, and switch root into the appliance, and run C</init> from
the appliance.

If this works successfully you will see messages such as:

 febootstrap: picked /sys/block/vdb/dev as root device
 febootstrap: creating /dev/root as block special 252:16
 febootstrap: mounting new root on /root
 febootstrap: chroot
 Starting /init script ...

Note that C<Starting /init script ...> indicates that the appliance's
init script is now running.

=item Initialize the appliance

The appliance itself now initializes itself.  This involves starting
certain processes like C<udev>, possibly printing some debug
information, and finally running the daemon (C<guestfsd>).

=item The daemon

Finally the daemon (C<guestfsd>) runs inside the appliance.  If it
runs you should see:

 verbose daemon enabled

The daemon expects to see a named virtio-serial port exposed by qemu
and connected on the other end to the library.

The daemon connects to this port (and hence to the library) and sends
a four byte message C<GUESTFS_LAUNCH_FLAG>, which initiates the
communication protocol (see below).

=back

=head2 COMMUNICATION PROTOCOL

Don't rely on using this protocol directly.  This section documents
how it currently works, but it may change at any time.

The protocol used to talk between the library and the daemon running
inside the qemu virtual machine is a simple RPC mechanism built on top
of XDR (RFC 1014, RFC 1832, RFC 4506).

The detailed format of structures is in C<src/guestfs_protocol.x>
(note: this file is automatically generated).

There are two broad cases, ordinary functions that don't have any
C<FileIn> and C<FileOut> parameters, which are handled with very
simple request/reply messages.  Then there are functions that have any
C<FileIn> or C<FileOut> parameters, which use the same request and
reply messages, but they may also be followed by files sent using a
chunked encoding.

=head3 ORDINARY FUNCTIONS (NO FILEIN/FILEOUT PARAMS)

For ordinary functions, the request message is:

 total length (header + arguments,
      but not including the length word itself)
 struct guestfs_message_header (encoded as XDR)
 struct guestfs_<foo>_args (encoded as XDR)

The total length field allows the daemon to allocate a fixed size
buffer into which it slurps the rest of the message.  As a result, the
total length is limited to C<GUESTFS_MESSAGE_MAX> bytes (currently
4MB), which means the effective size of any request is limited to
somewhere under this size.

Note also that many functions don't take any arguments, in which case
the C<guestfs_I<foo>_args> is completely omitted.

The header contains the procedure number (C<guestfs_proc>) which is
how the receiver knows what type of args structure to expect, or none
at all.

For functions that take optional arguments, the optional arguments are
encoded in the C<guestfs_I<foo>_args> structure in the same way as
ordinary arguments.  A bitmask in the header indicates which optional
arguments are meaningful.  The bitmask is also checked to see if it
contains bits set which the daemon does not know about (eg. if more
optional arguments were added in a later version of the library), and
this causes the call to be rejected.

The reply message for ordinary functions is:

 total length (header + ret,
      but not including the length word itself)
 struct guestfs_message_header (encoded as XDR)
 struct guestfs_<foo>_ret (encoded as XDR)

As above the C<guestfs_I<foo>_ret> structure may be completely omitted
for functions that return no formal return values.

As above the total length of the reply is limited to
C<GUESTFS_MESSAGE_MAX>.

In the case of an error, a flag is set in the header, and the reply
message is slightly changed:

 total length (header + error,
      but not including the length word itself)
 struct guestfs_message_header (encoded as XDR)
 struct guestfs_message_error (encoded as XDR)

The C<guestfs_message_error> structure contains the error message as a
string.

=head3 FUNCTIONS THAT HAVE FILEIN PARAMETERS

A C<FileIn> parameter indicates that we transfer a file I<into> the
guest.  The normal request message is sent (see above).  However this
is followed by a sequence of file chunks.

 total length (header + arguments,
      but not including the length word itself,
      and not including the chunks)
 struct guestfs_message_header (encoded as XDR)
 struct guestfs_<foo>_args (encoded as XDR)
 sequence of chunks for FileIn param #0
 sequence of chunks for FileIn param #1 etc.

The "sequence of chunks" is:

 length of chunk (not including length word itself)
 struct guestfs_chunk (encoded as XDR)
 length of chunk
 struct guestfs_chunk (encoded as XDR)
   ...
 length of chunk
 struct guestfs_chunk (with data.data_len == 0)

The final chunk has the C<data_len> field set to zero.  Additionally a
flag is set in the final chunk to indicate either successful
completion or early cancellation.

At time of writing there are no functions that have more than one
FileIn parameter.  However this is (theoretically) supported, by
sending the sequence of chunks for each FileIn parameter one after
another (from left to right).

Both the library (sender) I<and> the daemon (receiver) may cancel the
transfer.  The library does this by sending a chunk with a special
flag set to indicate cancellation.  When the daemon sees this, it
cancels the whole RPC, does I<not> send any reply, and goes back to
reading the next request.

The daemon may also cancel.  It does this by writing a special word
C<GUESTFS_CANCEL_FLAG> to the socket.  The library listens for this
during the transfer, and if it gets it, it will cancel the transfer
(it sends a cancel chunk).  The special word is chosen so that even if
cancellation happens right at the end of the transfer (after the
library has finished writing and has started listening for the reply),
the "spurious" cancel flag will not be confused with the reply
message.

This protocol allows the transfer of arbitrary sized files (no 32 bit
limit), and also files where the size is not known in advance
(eg. from pipes or sockets).  However the chunks are rather small
(C<GUESTFS_MAX_CHUNK_SIZE>), so that neither the library nor the
daemon need to keep much in memory.

=head3 FUNCTIONS THAT HAVE FILEOUT PARAMETERS

The protocol for FileOut parameters is exactly the same as for FileIn
parameters, but with the roles of daemon and library reversed.

 total length (header + ret,
      but not including the length word itself,
      and not including the chunks)
 struct guestfs_message_header (encoded as XDR)
 struct guestfs_<foo>_ret (encoded as XDR)
 sequence of chunks for FileOut param #0
 sequence of chunks for FileOut param #1 etc.

=head3 INITIAL MESSAGE

When the daemon launches it sends an initial word
(C<GUESTFS_LAUNCH_FLAG>) which indicates that the guest and daemon is
alive.  This is what L</guestfs_launch> waits for.

=head3 PROGRESS NOTIFICATION MESSAGES

The daemon may send progress notification messages at any time.  These
are distinguished by the normal length word being replaced by
C<GUESTFS_PROGRESS_FLAG>, followed by a fixed size progress message.

The library turns them into progress callbacks (see
L</GUESTFS_EVENT_PROGRESS>) if there is a callback registered, or
discards them if not.

The daemon self-limits the frequency of progress messages it sends
(see C<daemon/proto.c:notify_progress>).  Not all calls generate
progress messages.

=head1 LIBGUESTFS VERSION NUMBERS

Since April 2010, libguestfs has started to make separate development
and stable releases, along with corresponding branches in our git
repository.  These separate releases can be identified by version
number:

                 even numbers for stable: 1.2.x, 1.4.x, ...
       .-------- odd numbers for development: 1.3.x, 1.5.x, ...
       |
       v
 1  .  3  .  5
 ^           ^
 |           |
 |           `-------- sub-version
 |
 `------ always '1' because we don't change the ABI

Thus "1.3.5" is the 5th update to the development branch "1.3".

As time passes we cherry pick fixes from the development branch and
backport those into the stable branch, the effect being that the
stable branch should get more stable and less buggy over time.  So the
stable releases are ideal for people who don't need new features but
would just like the software to work.

Our criteria for backporting changes are:

=over 4

=item *

Documentation changes which don't affect any code are
backported unless the documentation refers to a future feature
which is not in stable.

=item *

Bug fixes which are not controversial, fix obvious problems, and
have been well tested are backported.

=item *

Simple rearrangements of code which shouldn't affect how it works get
backported.  This is so that the code in the two branches doesn't get
too far out of step, allowing us to backport future fixes more easily.

=item *

We I<don't> backport new features, new APIs, new tools etc, except in
one exceptional case: the new feature is required in order to
implement an important bug fix.

=back

A new stable branch starts when we think the new features in
development are substantial and compelling enough over the current
stable branch to warrant it.  When that happens we create new stable
and development versions 1.N.0 and 1.(N+1).0 [N is even].  The new
dot-oh release won't necessarily be so stable at this point, but by
backporting fixes from development, that branch will stabilize over
time.

=head1 EXTENDING LIBGUESTFS

This section is for hackers who want to extend libguestfs itself.

=head2 OVERVIEW OF THE SOURCE CODE

Libguestfs source is located in the github repository
L<https://github.com/libguestfs/libguestfs>

Large amounts of boilerplate code in libguestfs (RPC, bindings,
documentation) are generated.  This means that many source files will
appear to be missing from a straightforward git checkout.  You have to
run the generator (C<./autogen.sh && make -C generator>) in order to
create those files.

Libguestfs uses an autotools-based build system, with the main files
being C<configure.ac> and C<Makefile.am>.  The C<generator>
subdirectory contains the generator, plus files describing the API.
The C<src> subdirectory contains source for the library.  The
C<appliance> and C<daemon> subdirectories contain the source for the
code that builds the appliance, and the code that runs in the
appliance respectively.  Other directories are covered in the section
L<SOURCE CODE SUBDIRECTORIES> below.

Apart from the fact that all API entry points go via some generated
code, the library is straightforward.  (In fact, even the generated
code is designed to be readable, and should be read as ordinary code).
Some actions run entirely in the library, and are written as C
functions in files under C<src>.  Others are forwarded to the daemon
where (after some generated RPC marshalling) they appear as C
functions in files under C<daemon>.

To build from source, first read the C<README> file.

=head2 C<local*> FILES

Files in the top source directory that begin with the prefix C<local*>
are ignored by git.  These files can contain local configuration or
scripts that you need to build libguestfs.

By convention, I have a file called C<localconfigure> which is a
simple wrapper around C<autogen.sh> containing local configure
customizations that I need:

 . localenv
 ./autogen.sh \
     --with-default-attach-method=libvirt \
     --enable-gcc-warnings \
     --enable-gtk-doc \
     -C \
     "$@"

So I can use this to build libguestfs:

 ./localconfigure && make

If there is a file in the top build directory called C<localenv>, then
it will be sourced by C<make>.  This file can contain any local
environment variables needed, eg. for skipping tests:

 # Use an alternate python binary.
 export PYTHON=python3
 # Skip this test, it is broken.
 export SKIP_TEST_BTRFS_FSCK=1

Note that C<localenv> is included by the top Makefile (so it's a
Makefile fragment).  But if it is also sourced by your
C<localconfigure> script then it is used as a shell script.

=head2 ADDING A NEW API ACTION

Because large amounts of boilerplate code in libguestfs are generated,
this makes it easy to extend the libguestfs API.

To add a new API action there are two changes:

=over 4

=item 1.

You need to add a description of the call (name, parameters, return
type, tests, documentation) to C<generator/actions.ml>.

There are two sorts of API action, depending on whether the call goes
through to the daemon in the appliance, or is serviced entirely by the
library (see L</ARCHITECTURE> above).  L</guestfs_sync> is an example
of the former, since the sync is done in the appliance.
L</guestfs_set_trace> is an example of the latter, since a trace flag
is maintained in the handle and all tracing is done on the library
side.

Most new actions are of the first type, and get added to the
C<daemon_functions> list.  Each function has a unique procedure number
used in the RPC protocol which is assigned to that action when we
publish libguestfs and cannot be reused.  Take the latest procedure
number and increment it.

For library-only actions of the second type, add to the
C<non_daemon_functions> list.  Since these functions are serviced by
the library and do not travel over the RPC mechanism to the daemon,
these functions do not need a procedure number, and so the procedure
number is set to C<-1>.

=item 2.

Implement the action (in C):

For daemon actions, implement the function C<do_E<lt>nameE<gt>> in the
C<daemon/> directory.

For library actions, implement the function C<guestfs__E<lt>nameE<gt>>
(note: double underscore) in the C<src/> directory.

In either case, use another function as an example of what to do.

=back

After making these changes, use C<make> to compile.

Note that you don't need to implement the RPC, language bindings,
manual pages or anything else.  It's all automatically generated from
the OCaml description.

=head2 ADDING TESTS FOR AN API ACTION

You can supply zero or as many tests as you want per API call.  The
tests can either be added as part of the API description
(C<generator/actions.ml>), or in some rarer cases you may want to drop
a script into C<tests/*/>.  Note that adding a script to C<tests/*/>
is slower, so if possible use the first method.

The following describes the test environment used when you add an API
test in C<actions.ml>.

The test environment has 4 block devices:

=over 4

=item C</dev/sda> 500MB

General block device for testing.

=item C</dev/sdb> 50MB

C</dev/sdb1> is an ext2 filesystem used for testing
filesystem write operations.

=item C</dev/sdc> 10MB

Used in a few tests where two block devices are needed.

=item C</dev/sdd>

ISO with fixed content (see C<images/test.iso>).

=back

To be able to run the tests in a reasonable amount of time, the
libguestfs appliance and block devices are reused between tests.  So
don't try testing L</guestfs_kill_subprocess> :-x

Each test starts with an initial scenario, selected using one of the
C<Init*> expressions, described in C<generator/types.ml>.  These
initialize the disks mentioned above in a particular way as documented
in C<types.ml>.  You should not assume anything about the previous
contents of other disks that are not initialized.

You can add a prerequisite clause to any individual test.  This is a
run-time check, which, if it fails, causes the test to be skipped.
Useful if testing a command which might not work on all variations of
libguestfs builds.  A test that has prerequisite of C<Always> means to
run unconditionally.

In addition, packagers can skip individual tests by setting
environment variables before running C<make check>.

 SKIP_TEST_<CMD>_<NUM>=1

eg: C<SKIP_TEST_COMMAND_3=1> skips test #3 of L</guestfs_command>.

or:

 SKIP_TEST_<CMD>=1

eg: C<SKIP_TEST_ZEROFREE=1> skips all L</guestfs_zerofree> tests.

Packagers can run only certain tests by setting for example:

 TEST_ONLY="vfs_type zerofree"

See C<tests/c-api/tests.c> for more details of how these environment
variables work.

=head2 DEBUGGING NEW API ACTIONS

Test new actions work before submitting them.

You can use guestfish to try out new commands.

Debugging the daemon is a problem because it runs inside a minimal
environment.  However you can fprintf messages in the daemon to
stderr, and they will show up if you use C<guestfish -v>.

=head2 FORMATTING CODE AND OTHER CONVENTIONS

Our C source code generally adheres to some basic code-formatting
conventions.  The existing code base is not totally consistent on this
front, but we do prefer that contributed code be formatted similarly.
In short, use spaces-not-TABs for indentation, use 2 spaces for each
indentation level, and other than that, follow the K&R style.

If you use Emacs, add the following to one of one of your start-up files
(e.g., ~/.emacs), to help ensure that you get indentation right:

 ;;; In libguestfs, indent with spaces everywhere (not TABs).
 ;;; Exceptions: Makefile and ChangeLog modes.
 (add-hook 'find-file-hook
     '(lambda () (if (and buffer-file-name
                          (string-match "/libguestfs\\>"
                              (buffer-file-name))
                          (not (string-equal mode-name "Change Log"))
                          (not (string-equal mode-name "Makefile")))
                     (setq indent-tabs-mode nil))))
 
 ;;; When editing C sources in libguestfs, use this style.
 (defun libguestfs-c-mode ()
   "C mode with adjusted defaults for use with libguestfs."
   (interactive)
   (c-set-style "K&R")
   (setq c-indent-level 2)
   (setq c-basic-offset 2))
 (add-hook 'c-mode-hook
           '(lambda () (if (string-match "/libguestfs\\>"
                               (buffer-file-name))
                           (libguestfs-c-mode))))

Enable warnings when compiling (and fix any problems this
finds):

 ./configure --enable-gcc-warnings

Useful targets are:

 make syntax-check -j1 -k # checks for various syntax and
                          # style problems
 
 make check               # runs the test suite
 
 make extra-tests         # in-depth testing using valgrind
                          # and real guests

=head2 DAEMON CUSTOM PRINTF FORMATTERS

In the daemon code we have created custom printf formatters C<%Q> and
C<%R>, which are used to do shell quoting.

=over 4

=item %Q

Simple shell quoted string.  Any spaces or other shell characters are
escaped for you.

=item %R

Same as C<%Q> except the string is treated as a path which is prefixed
by the sysroot.

=back

For example:

 asprintf (&cmd, "cat %R", path);

would produce C<cat /sysroot/some\ path\ with\ spaces>

I<Note:> Do I<not> use these when you are passing parameters to the
C<command{,r,v,rv}()> functions.  These parameters do NOT need to be
quoted because they are not passed via the shell (instead, straight to
exec).  You probably want to use the C<sysroot_path()> function
however.

=head2 SUBMITTING YOUR NEW API ACTIONS

Submit patches to the mailing list:
L<http://www.redhat.com/mailman/listinfo/libguestfs>
and CC to L<rjones@redhat.com>.

=head2 INTERNATIONALIZATION (I18N) SUPPORT

We support i18n (gettext anyhow) in the library.

However many messages come from the daemon, and we don't translate
those at the moment.  One reason is that the appliance generally has
all locale files removed from it, because they take up a lot of space.
So we'd have to readd some of those, as well as copying our PO files
into the appliance.

Debugging messages are never translated, since they are intended for
the programmers.

=head2 SOURCE CODE SUBDIRECTORIES

=over 4

=item C<align>

L<virt-alignment-scan(1)> command and documentation.

=item C<appliance>

The libguestfs appliance, build scripts and so on.

=item C<build-aux>

Various build scripts used by autotools.

=item C<cat>

The L<virt-cat(1)>, L<virt-filesystems(1)> and L<virt-ls(1)> commands
and documentation.

=item C<contrib>

Outside contributions, experimental parts.

=item C<daemon>

The daemon that runs inside the libguestfs appliance and carries out
actions.

=item C<df>

L<virt-df(1)> command and documentation.

=item C<edit>

L<virt-edit(1)> command and documentation.

=item C<examples>

C API example code.

=item C<fish>

L<guestfish(1)>, the command-line shell, and various shell scripts
built on top such as L<virt-copy-in(1)>, L<virt-copy-out(1)>,
L<virt-tar-in(1)>, L<virt-tar-out(1)>.

=item C<format>

L<virt-format(1)> command and documentation.

=item C<fuse>

L<guestmount(1)>, FUSE (userspace filesystem) built on top of libguestfs.

=item C<generator>

The crucially important generator, used to automatically generate
large amounts of boilerplate C code for things like RPC and bindings.

=item C<gnulib>

Gnulib is used as a portability library.  A copy of gnulib is included
under here.

=item C<html>

Generated HTML manual pages.

=item C<inspector>

L<virt-inspector(1)>, the virtual machine image inspector.

=item C<logo>

Logo used on the website.  The fish is called Arthur by the way.

=item C<m4>

M4 macros used by autoconf.

=item C<po>

Translations of simple gettext strings.

=item C<po-docs>

The build infrastructure and PO files for translations of manpages and
POD files.  Eventually this will be combined with the C<po> directory,
but that is rather complicated.

=item C<rescue>

L<virt-rescue(1)> command and documentation.

=item C<resize>

L<virt-resize(1)> command and documentation.

=item C<sparsify>

L<virt-sparsify(1)> command and documentation.

=item C<src>

Source code to the C library.

=item C<sysprep>

L<virt-sysprep(1)> command and documentation.

=item C<tests>

Tests.

=item C<test-tool>

Test tool for end users to test if their qemu/kernel combination
will work with libguestfs.

=item C<tmp>

Used for temporary files when running the tests (instead of C</tmp>
etc).  The reason is so that you can run multiple parallel tests of
libguestfs without having one set of tests overwriting the appliance
created by another.

=item C<tools>

Command line tools written in Perl (L<virt-win-reg(1)> and many others).

=item C<csharp>

=item C<erlang>

=item C<gobject>

=item C<haskell>

=item C<java>

=item C<ocaml>

=item C<php>

=item C<perl>

=item C<python>

=item C<ruby>

Language bindings.

=back

=head2 MAKING A STABLE RELEASE

When we make a stable release, there are several steps documented
here.  See L</LIBGUESTFS VERSION NUMBERS> for general information
about the stable branch policy.

=over 4

=item *

Check C<make && make check> works on at least Fedora, Debian and
Ubuntu.

=item *

Finalize C<guestfs-release-notes.pod>

=item *

Update ROADMAP.

=item *

Run C<src/api-support/update-from-tarballs.sh>.

=item *

Push and pull from Transifex.

Run:

 tx push -s

to push the latest POT files to Transifex.  Then run:

 ./tx-pull.sh

which is a wrapper to pull the latest translated C<*.po> files.

=item *

Create new stable and development directories under
L<http://libguestfs.org/download>.

=item *

Create the branch in git:

 git tag -a 1.XX.0 -m "Version 1.XX.0 (stable)"
 git tag -a 1.YY.0 -m "Version 1.YY.0 (development)"
 git branch stable-1.XX
 git push origin tag 1.XX.0 1.YY.0 stable-1.XX

=back

=head1 LIMITS

=head2 PROTOCOL LIMITS

Internally libguestfs uses a message-based protocol to pass API calls
and their responses to and from a small "appliance" (see L</INTERNALS>
for plenty more detail about this).  The maximum message size used by
the protocol is slightly less than 4 MB.  For some API calls you may
need to be aware of this limit.  The API calls which may be affected
are individually documented, with a link back to this section of the
documentation.

In libguestfs E<lt> 1.19.32, several calls had to encode either their
entire argument list or their entire return value (or sometimes both)
in a single protocol message, and this gave them an arbitrary
limitation on how much data they could handle.  For example,
L</guestfs_cat> could only download a file if it was less than around
4 MB in size.  In later versions of libguestfs, some of these limits
have been removed.  The APIs which were previously limited but are now
unlimited (except perhaps by available memory) are listed below.  To
find out if a specific API is subject to protocol limits, check for
the warning in the API documentation which links to this section, and
remember to check the version of the documentation that matches the
version of libguestfs you are using.

L</guestfs_cat>, L</guestfs_find>, L</guestfs_read_file>,
L</guestfs_read_lines>, L</guestfs_write>, L</guestfs_write_append>,
L</guestfs_lstatlist>, L</guestfs_lxattrlist>,
L</guestfs_readlinklist>, L</guestfs_ls>.

See also L</UPLOADING> and L</DOWNLOADING> for further information
about copying large amounts of data into or out of a filesystem.

=head2 MAXIMUM NUMBER OF DISKS

In libguestfs E<ge> 1.19.7, you can query the maximum number of disks
that may be added by calling L</guestfs_max_disks>.  In earlier
versions of libguestfs (ie. where this call is not available) you
should assume the maximum is 25.

The rest of this section covers implementation details, which could
change in future.

When using virtio-scsi disks (the default if available in qemu) the
current limit is B<255> disks.  When using virtio-blk (the old
default) the limit is around B<27> disks, but may vary according to
implementation details and whether the network is enabled.

Virtio-scsi as used by libguestfs is configured to use one target per
disk, and 256 targets are available.

Virtio-blk consumes 1 virtual PCI slot per disk, and PCI is limited
to 31 slots, but some of these are used for other purposes.

One virtual disk is used by libguestfs internally.

Before libguestfs 1.19.7, disk names had to be a single character
(eg. C</dev/sda> through C</dev/sdz>), and since one disk is reserved,
that meant the limit was 25.  This has been fixed in more recent
versions.

In libguestfs E<ge> 1.20 it is possible to hot plug disks.  See
L</HOTPLUGGING>.

=head2 MAXIMUM NUMBER OF PARTITIONS PER DISK

Virtio limits the maximum number of partitions per disk to B<15>.

This is because it reserves 4 bits for the minor device number (thus
C</dev/vda>, and C</dev/vda1> through C</dev/vda15>).

If you attach a disk with more than 15 partitions, the extra
partitions are ignored by libguestfs.

=head2 MAXIMUM SIZE OF A DISK

Probably the limit is between 2**63-1 and 2**64-1 bytes.

We have tested block devices up to 1 exabyte (2**60 or
1,152,921,504,606,846,976 bytes) using sparse files backed by an XFS
host filesystem.

Although libguestfs probably does not impose any limit, the underlying
host storage will.  If you store disk images on a host ext4
filesystem, then the maximum size will be limited by the maximum ext4
file size (currently 16 TB).  If you store disk images as host logical
volumes then you are limited by the maximum size of an LV.

For the hugest disk image files, we recommend using XFS on the host
for storage.

=head2 MAXIMUM SIZE OF A PARTITION

The MBR (ie. classic MS-DOS) partitioning scheme uses 32 bit sector
numbers.  Assuming a 512 byte sector size, this means that MBR cannot
address a partition located beyond 2 TB on the disk.

It is recommended that you use GPT partitions on disks which are
larger than this size.  GPT uses 64 bit sector numbers and so can
address partitions which are theoretically larger than the largest
disk we could support.

=head2 MAXIMUM SIZE OF A FILESYSTEM, FILES, DIRECTORIES

This depends on the filesystem type.  libguestfs itself does not
impose any known limit.  Consult Wikipedia or the filesystem
documentation to find out what these limits are.

=head2 MAXIMUM UPLOAD AND DOWNLOAD

The API functions L</guestfs_upload>, L</guestfs_download>,
L</guestfs_tar_in>, L</guestfs_tar_out> and the like allow unlimited
sized uploads and downloads.

=head2 INSPECTION LIMITS

The inspection code has several arbitrary limits on things like the
size of Windows Registry hive it will read, and the length of product
name.  These are intended to stop a malicious guest from consuming
arbitrary amounts of memory and disk space on the host, and should not
be reached in practice.  See the source code for more information.

=head1 ENVIRONMENT VARIABLES

=over 4

=item FEBOOTSTRAP_KERNEL

=item FEBOOTSTRAP_MODULES

These two environment variables allow the kernel that libguestfs uses
in the appliance to be selected.  If C<$FEBOOTSTRAP_KERNEL> is not
set, then the most recent host kernel is chosen.  For more information
about kernel selection, see L<febootstrap-supermin-helper(8)>.  This
feature is only available in febootstrap E<ge> 3.8.

=item LIBGUESTFS_APPEND

Pass additional options to the guest kernel.

=item LIBGUESTFS_ATTACH_METHOD

Choose the default way to create the appliance.  See
L</guestfs_set_attach_method>.

=item LIBGUESTFS_CACHEDIR

The location where libguestfs will cache its appliance, when
using a supermin appliance.  The appliance is cached and shared
between all handles which have the same effective user ID.

If C<LIBGUESTFS_CACHEDIR> is not set, then C<TMPDIR> is used.  If
C<TMPDIR> is not set, then C</var/tmp> is used.

See also L</LIBGUESTFS_TMPDIR>, L</guestfs_set_cachedir>.

=item LIBGUESTFS_DEBUG

Set C<LIBGUESTFS_DEBUG=1> to enable verbose messages.  This
has the same effect as calling C<guestfs_set_verbose (g, 1)>.

=item LIBGUESTFS_MEMSIZE

Set the memory allocated to the qemu process, in megabytes.  For
example:

 LIBGUESTFS_MEMSIZE=700

=item LIBGUESTFS_PATH

Set the path that libguestfs uses to search for a supermin appliance.
See the discussion of paths in section L</PATH> above.

=item LIBGUESTFS_QEMU

Set the default qemu binary that libguestfs uses.  If not set, then
the qemu which was found at compile time by the configure script is
used.

See also L</QEMU WRAPPERS> above.

=item LIBGUESTFS_TMPDIR

The location where libguestfs will store temporary files used
by each handle.

If C<LIBGUESTFS_TMPDIR> is not set, then C<TMPDIR> is used.  If
C<TMPDIR> is not set, then C</tmp> is used.

See also L</LIBGUESTFS_CACHEDIR>, L</guestfs_set_tmpdir>.

=item LIBGUESTFS_TRACE

Set C<LIBGUESTFS_TRACE=1> to enable command traces.  This
has the same effect as calling C<guestfs_set_trace (g, 1)>.

=item TMPDIR

See L</LIBGUESTFS_CACHEDIR>, L</LIBGUESTFS_TMPDIR>.

=back

=head1 SEE ALSO

L<guestfs-examples(3)>,
L<guestfs-erlang(3)>,
L<guestfs-java(3)>,
L<guestfs-ocaml(3)>,
L<guestfs-perl(3)>,
L<guestfs-python(3)>,
L<guestfs-ruby(3)>,
L<guestfish(1)>,
L<guestmount(1)>,
L<virt-alignment-scan(1)>,
L<virt-cat(1)>,
L<virt-copy-in(1)>,
L<virt-copy-out(1)>,
L<virt-df(1)>,
L<virt-edit(1)>,
L<virt-filesystems(1)>,
L<virt-format(1)>,
L<virt-inspector(1)>,
L<virt-list-filesystems(1)>,
L<virt-list-partitions(1)>,
L<virt-ls(1)>,
L<virt-make-fs(1)>,
L<virt-rescue(1)>,
L<virt-resize(1)>,
L<virt-sparsify(1)>,
L<virt-sysprep(1)>,
L<virt-tar(1)>,
L<virt-tar-in(1)>,
L<virt-tar-out(1)>,
L<virt-win-reg(1)>,
L<guestfs-faq(1)>,
L<guestfs-performance(1)>,
L<guestfs-release-notes(1)>,
L<guestfs-testing(1)>,
L<libguestfs-test-tool(1)>,
L<libguestfs-make-fixed-appliance(1)>,
L<febootstrap(8)>,
L<febootstrap-supermin-helper(8)>,
L<qemu(1)>,
L<hivex(3)>,
L<stap(1)>,
L<http://libguestfs.org/>.

Tools with a similar purpose:
L<fdisk(8)>,
L<parted(8)>,
L<kpartx(8)>,
L<lvm(8)>,
L<disktype(1)>.

=head1 AUTHORS

Richard W.M. Jones (C<rjones at redhat dot com>)

=head1 COPYRIGHT

Copyright (C) 2009-2012 Red Hat Inc.