Linux Certif

Toute la documentation sur la certification Linux LPI

Sys::Guestfs.3pm

NAME

Sys::Guestfs - Perl bindings for libguestfs

SYNOPSIS

  use Sys::Guestfs;
 
  my $h = Sys::Guestfs->new ();
  $h->add_drive ('guest.img');
  $h->launch ();
  $h->mount ('/dev/sda1', '/');
  $h->touch ('/hello');
  $h->sync ();
 
 

DESCRIPTION

The "Sys::Guestfs" module provides a Perl XS binding to the libguestfs API for examining and modifying virtual machine disk images.

Amongst the things this is good for: making batch configuration changes to guests, getting disk used/free statistics (see also: virt-df), migrating between virtualization systems (see also: virt-p2v), performing partial backups, performing partial guest clones, cloning guests and changing registry/UUID/hostname info, and much else besides.

Libguestfs uses Linux kernel and qemu code, and can access any type of guest filesystem that Linux and qemu can, including but not limited to: ext2/3/4, btrfs, FAT and NTFS, LVM, many different disk partition schemes, qcow, qcow2, vmdk.

Libguestfs provides ways to enumerate guest storage (eg. partitions, LVs, what filesystem is in each LV, etc.). It can also run commands in the context of the guest. Also you can access filesystems over FUSE.

See also Sys::Guestfs::Lib(3) for a set of useful library functions for using libguestfs from Perl, including integration with libvirt.

ERRORS

All errors turn into calls to "croak" (see Carp(3)).

METHODS

$h = Sys::Guestfs->new ();

Create a new guestfs handle.

$h->close ();

Explicitly close the guestfs handle.

Note: You should not usually call this function. The handle will be closed implicitly when its reference count goes to zero (eg. when it goes out of scope or the program ends). This call is only required in some exceptional cases, such as where the program may contain cached references to the handle 'somewhere' and you really have to have the close happen right away. After calling "close" the program must not call any method (including "close") on the handle (but the implicit call to "DESTROY" that happens when the final reference is cleaned up is OK).

$h->set_progress_callback (\&cb);

Set the progress notification callback for this handle to the Perl closure "cb".

"cb" will be called whenever a long-running operation generates a progress notification message. The 4 parameters to the function are: "proc_nr", "serial", "position" and "total".

You should carefully read the documentation for ``guestfs_set_progress_callback'' in guestfs(3) before using this function.

$h->clear_progress_callback ();

This removes any progress callback function associated with the handle.

$h->add_cdrom ($filename);

This function adds a virtual CD-ROM disk image to the guest.

This is equivalent to the qemu parameter "-cdrom filename".

Notes:

*

This call checks for the existence of "filename". This stops you from specifying other types of drive which are supported by qemu such as "nbd:" and "http:" URLs. To specify those, use the general "$h->config" call instead.

*

If you just want to add an ISO file (often you use this as an efficient way to transfer large files into the guest), then you should probably use "$h->add_drive_ro" instead.

$h->add_drive ($filename);

This function adds a virtual machine disk image "filename" to the guest. The first time you call this function, the disk appears as IDE disk 0 ("/dev/sda") in the guest, the second time as "/dev/sdb", and so on.

You don't necessarily need to be root when using libguestfs. However you obviously do need sufficient permissions to access the filename for whatever operations you want to perform (ie. read access if you just want to read the image or write access if you want to modify the image).

This is equivalent to the qemu parameter "-drive file=filename,cache=off,if=...".

"cache=off" is omitted in cases where it is not supported by the underlying filesystem.

"if=..." is set at compile time by the configuration option "./configure --with-drive-if=...". In the rare case where you might need to change this at run time, use "$h->add_drive_with_if" or "$h->add_drive_ro_with_if".

Note that this call checks for the existence of "filename". This stops you from specifying other types of drive which are supported by qemu such as "nbd:" and "http:" URLs. To specify those, use the general "$h->config" call instead.

$h->add_drive_ro ($filename);

This adds a drive in snapshot mode, making it effectively read-only.

Note that writes to the device are allowed, and will be seen for the duration of the guestfs handle, but they are written to a temporary file which is discarded as soon as the guestfs handle is closed. We don't currently have any method to enable changes to be committed, although qemu can support this.

This is equivalent to the qemu parameter "-drive file=filename,snapshot=on,if=...".

"if=..." is set at compile time by the configuration option "./configure --with-drive-if=...". In the rare case where you might need to change this at run time, use "$h->add_drive_with_if" or "$h->add_drive_ro_with_if".

Note that this call checks for the existence of "filename". This stops you from specifying other types of drive which are supported by qemu such as "nbd:" and "http:" URLs. To specify those, use the general "$h->config" call instead.

$h->add_drive_ro_with_if ($filename, $iface);

This is the same as "$h->add_drive_ro" but it allows you to specify the QEMU interface emulation to use at run time.

$h->add_drive_with_if ($filename, $iface);

This is the same as "$h->add_drive" but it allows you to specify the QEMU interface emulation to use at run time.

$h->aug_clear ($augpath);

Set the value associated with "path" to "NULL". This is the same as the augtool(1) "clear" command.

$h->aug_close ();

Close the current Augeas handle and free up any resources used by it. After calling this, you have to call "$h->aug_init" again before you can use any other Augeas functions.

%nrnodescreated = $h->aug_defnode ($name, $expr, $val);

Defines a variable "name" whose value is the result of evaluating "expr".

If "expr" evaluates to an empty nodeset, a node is created, equivalent to calling "$h->aug_set" "expr", "value". "name" will be the nodeset containing that single node.

On success this returns a pair containing the number of nodes in the nodeset, and a boolean flag if a node was created.

$nrnodes = $h->aug_defvar ($name, $expr);

Defines an Augeas variable "name" whose value is the result of evaluating "expr". If "expr" is NULL, then "name" is undefined.

On success this returns the number of nodes in "expr", or 0 if "expr" evaluates to something which is not a nodeset.

$val = $h->aug_get ($augpath);

Look up the value associated with "path". If "path" matches exactly one node, the "value" is returned.

$h->aug_init ($root, $flags);

Create a new Augeas handle for editing configuration files. If there was any previous Augeas handle associated with this guestfs session, then it is closed.

You must call this before using any other "$h->aug_*" commands.

"root" is the filesystem root. "root" must not be NULL, use "/" instead.

The flags are the same as the flags defined in <augeas.h>, the logical or of the following integers:

AUG_SAVE_BACKUP = 1

Keep the original file with a ".augsave" extension.

AUG_SAVE_NEWFILE = 2

Save changes into a file with extension ".augnew", and do not overwrite original. Overrides "AUG_SAVE_BACKUP".

AUG_TYPE_CHECK = 4

Typecheck lenses (can be expensive).

AUG_NO_STDINC = 8

Do not use standard load path for modules.

AUG_SAVE_NOOP = 16

Make save a no-op, just record what would have been changed.

AUG_NO_LOAD = 32

Do not load the tree in "$h->aug_init".

To close the handle, you can call "$h->aug_close".
To find out more about Augeas, see <http://augeas.net/>.

$h->aug_insert ($augpath, $label, $before);

Create a new sibling "label" for "path", inserting it into the tree before or after "path" (depending on the boolean flag "before").

"path" must match exactly one existing node in the tree, and "label" must be a label, ie. not contain "/", "*" or end with a bracketed index "[N]".

$h->aug_load ();

Load files into the tree.

See "aug_load" in the Augeas documentation for the full gory details.

@matches = $h->aug_ls ($augpath);

This is just a shortcut for listing "$h->aug_match" "path/*" and sorting the resulting nodes into alphabetical order.

@matches = $h->aug_match ($augpath);

Returns a list of paths which match the path expression "path". The returned paths are sufficiently qualified so that they match exactly one node in the current tree.

$h->aug_mv ($src, $dest);

Move the node "src" to "dest". "src" must match exactly one node. "dest" is overwritten if it exists.

$nrnodes = $h->aug_rm ($augpath);

Remove "path" and all of its children.

On success this returns the number of entries which were removed.

$h->aug_save ();

This writes all pending changes to disk.

The flags which were passed to "$h->aug_init" affect exactly how files are saved.

$h->aug_set ($augpath, $val);

Set the value associated with "path" to "val".

In the Augeas API, it is possible to clear a node by setting the value to NULL. Due to an oversight in the libguestfs API you cannot do that with this call. Instead you must use the "$h->aug_clear" call.

$h->available (\@groups);

This command is used to check the availability of some groups of functionality in the appliance, which not all builds of the libguestfs appliance will be able to provide.

The libguestfs groups, and the functions that those groups correspond to, are listed in ``AVAILABILITY'' in guestfs(3). You can also fetch this list at runtime by calling "$h->available_all_groups".

The argument "groups" is a list of group names, eg: "["inotify", "augeas"]" would check for the availability of the Linux inotify functions and Augeas (configuration file editing) functions.

The command returns no error if all requested groups are available.

It fails with an error if one or more of the requested groups is unavailable in the appliance.

If an unknown group name is included in the list of groups then an error is always returned.

Notes:

*

You must call "$h->launch" before calling this function.

The reason is because we don't know what groups are supported by the appliance/daemon until it is running and can be queried.

*

If a group of functions is available, this does not necessarily mean that they will work. You still have to check for errors when calling individual API functions even if they are available.

*

It is usually the job of distro packagers to build complete functionality into the libguestfs appliance. Upstream libguestfs, if built from source with all requirements satisfied, will support everything.

*

This call was added in version 1.0.80. In previous versions of libguestfs all you could do would be to speculatively execute a command to find out if the daemon implemented it. See also "$h->version".

@groups = $h->available_all_groups ();

This command returns a list of all optional groups that this daemon knows about. Note this returns both supported and unsupported groups. To find out which ones the daemon can actually support you have to call "$h->available" on each member of the returned list.

See also "$h->available" and ``AVAILABILITY'' in guestfs(3).

$h->base64_in ($base64file, $filename);

This command uploads base64-encoded data from "base64file" to "filename".

$h->base64_out ($filename, $base64file);

This command downloads the contents of "filename", writing it out to local file "base64file" encoded as base64.

$h->blockdev_flushbufs ($device);

This tells the kernel to flush internal buffers associated with "device".

This uses the blockdev(8) command.

$blocksize = $h->blockdev_getbsz ($device);

This returns the block size of a device.

(Note this is different from both size in blocks and filesystem block size).

This uses the blockdev(8) command.

$ro = $h->blockdev_getro ($device);

Returns a boolean indicating if the block device is read-only (true if read-only, false if not).

This uses the blockdev(8) command.

$sizeinbytes = $h->blockdev_getsize64 ($device);

This returns the size of the device in bytes.

See also "$h->blockdev_getsz".

This uses the blockdev(8) command.

$sectorsize = $h->blockdev_getss ($device);

This returns the size of sectors on a block device. Usually 512, but can be larger for modern devices.

(Note, this is not the size in sectors, use "$h->blockdev_getsz" for that).

This uses the blockdev(8) command.

$sizeinsectors = $h->blockdev_getsz ($device);

This returns the size of the device in units of 512-byte sectors (even if the sectorsize isn't 512 bytes ... weird).

See also "$h->blockdev_getss" for the real sector size of the device, and "$h->blockdev_getsize64" for the more useful size in bytes.

This uses the blockdev(8) command.

$h->blockdev_rereadpt ($device);

Reread the partition table on "device".

This uses the blockdev(8) command.

$h->blockdev_setbsz ($device, $blocksize);

This sets the block size of a device.

(Note this is different from both size in blocks and filesystem block size).

This uses the blockdev(8) command.

$h->blockdev_setro ($device);

Sets the block device named "device" to read-only.

This uses the blockdev(8) command.

$h->blockdev_setrw ($device);

Sets the block device named "device" to read-write.

This uses the blockdev(8) command.

$rpath = $h->case_sensitive_path ($path);

This can be used to resolve case insensitive paths on a filesystem which is case sensitive. The use case is to resolve paths which you have read from Windows configuration files or the Windows Registry, to the true path.

The command handles a peculiarity of the Linux ntfs-3g filesystem driver (and probably others), which is that although the underlying filesystem is case-insensitive, the driver exports the filesystem to Linux as case-sensitive.

One consequence of this is that special directories such as "c:\windows" may appear as "/WINDOWS" or "/windows" (or other things) depending on the precise details of how they were created. In Windows itself this would not be a problem.

Bug or feature? You decide: http://www.tuxera.com/community/ntfs-3g-faq/#posixfilenames1 <http://www.tuxera.com/community/ntfs-3g-faq/#posixfilenames1>

This function resolves the true case of each element in the path and returns the case-sensitive path.

Thus "$h->case_sensitive_path" (``/Windows/System32'') might return "/WINDOWS/system32" (the exact return value would depend on details of how the directories were originally created under Windows).

Note: This function does not handle drive names, backslashes etc.

See also "$h->realpath".

$content = $h->cat ($path);

Return the contents of the file named "path".

Note that this function cannot correctly handle binary files (specifically, files containing "\0" character which is treated as end of string). For those you need to use the "$h->read_file" or "$h->download" functions which have a more complex interface.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$checksum = $h->checksum ($csumtype, $path);

This call computes the MD5, SHAx or CRC checksum of the file named "path".

The type of checksum to compute is given by the "csumtype" parameter which must have one of the following values:

crc

Compute the cyclic redundancy check (CRC) specified by POSIX for the "cksum" command.

md5

Compute the MD5 hash (using the "md5sum" program).

sha1

Compute the SHA1 hash (using the "sha1sum" program).

sha224

Compute the SHA224 hash (using the "sha224sum" program).

sha256

Compute the SHA256 hash (using the "sha256sum" program).

sha384

Compute the SHA384 hash (using the "sha384sum" program).

sha512

Compute the SHA512 hash (using the "sha512sum" program).

The checksum is returned as a printable string.
To get the checksum for a device, use "$h->checksum_device".
To get the checksums for many files, use "$h->checksums_out".

$checksum = $h->checksum_device ($csumtype, $device);

This call computes the MD5, SHAx or CRC checksum of the contents of the device named "device". For the types of checksums supported see the "$h->checksum" command.

$h->checksums_out ($csumtype, $directory, $sumsfile);

This command computes the checksums of all regular files in "directory" and then emits a list of those checksums to the local output file "sumsfile".

This can be used for verifying the integrity of a virtual machine. However to be properly secure you should pay attention to the output of the checksum command (it uses the ones from GNU coreutils). In particular when the filename is not printable, coreutils uses a special backslash syntax. For more information, see the GNU coreutils info file.

$h->chmod ($mode, $path);

Change the mode (permissions) of "path" to "mode". Only numeric modes are supported.

Note: When using this command from guestfish, "mode" by default would be decimal, unless you prefix it with 0 to get octal, ie. use 0700 not 700.

The mode actually set is affected by the umask.

$h->chown ($owner, $group, $path);

Change the file owner to "owner" and group to "group".

Only numeric uid and gid are supported. If you want to use names, you will need to locate and parse the password file yourself (Augeas support makes this relatively easy).

$output = $h->command (\@arguments);

This call runs a command from the guest filesystem. The filesystem must be mounted, and must contain a compatible operating system (ie. something Linux, with the same or compatible processor architecture).

The single parameter is an argv-style list of arguments. The first element is the name of the program to run. Subsequent elements are parameters. The list must be non-empty (ie. must contain a program name). Note that the command runs directly, and is not invoked via the shell (see "$h->sh").

The return value is anything printed to stdout by the command.

If the command returns a non-zero exit status, then this function returns an error message. The error message string is the content of stderr from the command.

The $PATH environment variable will contain at least "/usr/bin" and "/bin". If you require a program from another location, you should provide the full path in the first parameter.

Shared libraries and data files required by the program must be available on filesystems which are mounted in the correct places. It is the caller's responsibility to ensure all filesystems that are needed are mounted at the right locations.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@lines = $h->command_lines (\@arguments);

This is the same as "$h->command", but splits the result into a list of lines.

See also: "$h->sh_lines"

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$h->config ($qemuparam, $qemuvalue);

This can be used to add arbitrary qemu command line parameters of the form "-param value". Actually it's not quite arbitrary - we prevent you from setting some parameters which would interfere with parameters that we use.

The first character of "param" string must be a "-" (dash).

"value" can be NULL.

$h->copy_size ($src, $dest, $size);

This command copies exactly "size" bytes from one source device or file "src" to another destination device or file "dest".

Note this will fail if the source is too short or if the destination is not large enough.

$h->cp ($src, $dest);

This copies a file from "src" to "dest" where "dest" is either a destination filename or destination directory.

$h->cp_a ($src, $dest);

This copies a file or directory from "src" to "dest" recursively using the "cp -a" command.

$h->dd ($src, $dest);

This command copies from one source device or file "src" to another destination device or file "dest". Normally you would use this to copy to or from a device or partition, for example to duplicate a filesystem.

If the destination is a device, it must be as large or larger than the source file or device, otherwise the copy will fail. This command cannot do partial copies (see "$h->copy_size").

$result = $h->debug ($subcmd, \@extraargs);

The "$h->debug" command exposes some internals of "guestfsd" (the guestfs daemon) that runs inside the qemu subprocess.

There is no comprehensive help for this command. You have to look at the file "daemon/debug.c" in the libguestfs source to find out what you can do.

$h->debug_upload ($filename, $tmpname, $mode);

The "$h->debug_upload" command uploads a file to the libguestfs appliance.

There is no comprehensive help for this command. You have to look at the file "daemon/debug.c" in the libguestfs source to find out what it is for.

$output = $h->df ();

This command runs the "df" command to report disk space used.

This command is mostly useful for interactive sessions. It is not intended that you try to parse the output string. Use "statvfs" from programs.

$output = $h->df_h ();

This command runs the "df -h" command to report disk space used in human-readable format.

This command is mostly useful for interactive sessions. It is not intended that you try to parse the output string. Use "statvfs" from programs.

$kmsgs = $h->dmesg ();

This returns the kernel messages ("dmesg" output) from the guest kernel. This is sometimes useful for extended debugging of problems.

Another way to get the same information is to enable verbose messages with "$h->set_verbose" or by setting the environment variable "LIBGUESTFS_DEBUG=1" before running the program.

$h->download ($remotefilename, $filename);

Download file "remotefilename" and save it as "filename" on the local machine.

"filename" can also be a named pipe.

See also "$h->upload", "$h->cat".

$h->download_offset ($remotefilename, $filename, $offset, $size);

Download file "remotefilename" and save it as "filename" on the local machine.

"remotefilename" is read for "size" bytes starting at "offset" (this region must be within the file or device).

Note that there is no limit on the amount of data that can be downloaded with this call, unlike with "$h->pread", and this call always reads the full amount unless an error occurs.

See also "$h->download", "$h->pread".

$h->drop_caches ($whattodrop);

This instructs the guest kernel to drop its page cache, and/or dentries and inode caches. The parameter "whattodrop" tells the kernel what precisely to drop, see http://linux-mm.org/Drop_Caches <http://linux-mm.org/Drop_Caches>

Setting "whattodrop" to 3 should drop everything.

This automatically calls sync(2) before the operation, so that the maximum guest memory is freed.

$sizekb = $h->du ($path);

This command runs the "du -s" command to estimate file space usage for "path".

"path" can be a file or a directory. If "path" is a directory then the estimate includes the contents of the directory and all subdirectories (recursively).

The result is the estimated size in kilobytes (ie. units of 1024 bytes).

$h->e2fsck_f ($device);

This runs "e2fsck -p -f device", ie. runs the ext2/ext3 filesystem checker on "device", noninteractively ("-p"), even if the filesystem appears to be clean ("-f").

This command is only needed because of "$h->resize2fs" (q.v.). Normally you should use "$h->fsck".

$output = $h->echo_daemon (\@words);

This command concatenates the list of "words" passed with single spaces between them and returns the resulting string.

You can use this command to test the connection through to the daemon.

See also "$h->ping_daemon".

@lines = $h->egrep ($regex, $path);

This calls the external "egrep" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@lines = $h->egrepi ($regex, $path);

This calls the external "egrep -i" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$equality = $h->equal ($file1, $file2);

This compares the two files "file1" and "file2" and returns true if their content is exactly equal, or false otherwise.

The external cmp(1) program is used for the comparison.

$existsflag = $h->exists ($path);

This returns "true" if and only if there is a file, directory (or anything) with the given "path" name.

See also "$h->is_file", "$h->is_dir", "$h->stat".

$h->fallocate ($path, $len);

This command preallocates a file (containing zero bytes) named "path" of size "len" bytes. If the file exists already, it is overwritten.

Do not confuse this with the guestfish-specific "alloc" command which allocates a file in the host and attaches it as a device.

This function is deprecated. In new code, use the "fallocate64" call instead.

Deprecated functions will not be removed from the API, but the fact that they are deprecated indicates that there are problems with correct use of these functions.

$h->fallocate64 ($path, $len);

This command preallocates a file (containing zero bytes) named "path" of size "len" bytes. If the file exists already, it is overwritten.

Note that this call allocates disk blocks for the file. To create a sparse file use "$h->truncate_size" instead.

The deprecated call "$h->fallocate" does the same, but owing to an oversight it only allowed 30 bit lengths to be specified, effectively limiting the maximum size of files created through that call to 1GB.

Do not confuse this with the guestfish-specific "alloc" and "sparse" commands which create a file in the host and attach it as a device.

@lines = $h->fgrep ($pattern, $path);

This calls the external "fgrep" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@lines = $h->fgrepi ($pattern, $path);

This calls the external "fgrep -i" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$description = $h->file ($path);

This call uses the standard file(1) command to determine the type or contents of the file.

This call will also transparently look inside various types of compressed file.

The exact command which runs is "file -zb path". Note in particular that the filename is not prepended to the output (the "-b" option).

This command can also be used on "/dev/" devices (and partitions, LV names). You can for example use this to determine if a device contains a filesystem, although it's usually better to use "$h->vfs_type".

If the "path" does not begin with "/dev/" then this command only works for the content of regular files. For other file types (directory, symbolic link etc) it will just return the string "directory" etc.

$arch = $h->file_architecture ($filename);

This detects the architecture of the binary "filename", and returns it if known.

Currently defined architectures are:

i386

This string is returned for all 32 bit i386, i486, i586, i686 binaries irrespective of the precise processor requirements of the binary.

x86_64

64 bit x86-64.

sparc

32 bit SPARC.

sparc64

64 bit SPARC V9 and above.

ia64

Intel Itanium.

ppc

32 bit Power PC.

ppc64

64 bit Power PC.

Libguestfs may return other architecture strings in future.
The function works on at least the following types of files:

*

many types of Un*x and Linux binary

*

many types of Un*x and Linux shared library

*

Windows Win32 and Win64 binaries

*

Windows Win32 and Win64 DLLs

Win32 binaries and DLLs return "i386".

Win64 binaries and DLLs return "x86_64".

*

Linux kernel modules

*

Linux new-style initrd images

*

some non-x86 Linux vmlinuz kernels

What it can't do currently:

*

static libraries (libfoo.a)

*

Linux old-style initrd as compressed ext2 filesystem (RHEL 3)

*

x86 Linux vmlinuz kernels

x86 vmlinuz images (bzImage format) consist of a mix of 16-, 32- and compressed code, and are horribly hard to unpack. If you want to find the architecture of a kernel, use the architecture of the associated initrd or kernel module(s) instead.

$size = $h->filesize ($file);

This command returns the size of "file" in bytes.

To get other stats about a file, use "$h->stat", "$h->lstat", "$h->is_dir", "$h->is_file" etc. To get the size of block devices, use "$h->blockdev_getsize64".

$h->fill ($c, $len, $path);

This command creates a new file called "path". The initial content of the file is "len" octets of "c", where "c" must be a number in the range "[0..255]".

To fill a file with zero bytes (sparsely), it is much more efficient to use "$h->truncate_size". To create a file with a pattern of repeating bytes use "$h->fill_pattern".

$h->fill_pattern ($pattern, $len, $path);

This function is like "$h->fill" except that it creates a new file of length "len" containing the repeating pattern of bytes in "pattern". The pattern is truncated if necessary to ensure the length of the file is exactly "len" bytes.

@names = $h->find ($directory);

This command lists out all files and directories, recursively, starting at "directory". It is essentially equivalent to running the shell command "find directory -print" but some post-processing happens on the output, described below.

This returns a list of strings without any prefix. Thus if the directory structure was:

  /tmp/a
  /tmp/b
  /tmp/c/d
 
 

then the returned list from "$h->find" "/tmp" would be 4 elements:

  a
  b
  c
  c/d
 
 

If "directory" is not a directory, then this command returns an error.

The returned list is sorted.

See also "$h->find0".

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$h->find0 ($directory, $files);

This command lists out all files and directories, recursively, starting at "directory", placing the resulting list in the external file called "files".

This command works the same way as "$h->find" with the following exceptions:

*

The resulting list is written to an external file.

*

Items (filenames) in the result are separated by "\0" characters. See find(1) option -print0.

*

This command is not limited in the number of names that it can return.

*

The result list is not sorted.

$device = $h->findfs_label ($label);

This command searches the filesystems and returns the one which has the given label. An error is returned if no such filesystem can be found.

To find the label of a filesystem, use "$h->vfs_label".

$device = $h->findfs_uuid ($uuid);

This command searches the filesystems and returns the one which has the given UUID. An error is returned if no such filesystem can be found.

To find the UUID of a filesystem, use "$h->vfs_uuid".

$status = $h->fsck ($fstype, $device);

This runs the filesystem checker (fsck) on "device" which should have filesystem type "fstype".

The returned integer is the status. See fsck(8) for the list of status codes from "fsck".

Notes:

*

Multiple status codes can be summed together.

*

A non-zero return code can mean ``success'', for example if errors have been corrected on the filesystem.

*

Checking or repairing NTFS volumes is not supported (by linux-ntfs).

This command is entirely equivalent to running "fsck -a -t fstype device".

$append = $h->get_append ();

Return the additional kernel options which are added to the guest kernel command line.

If "NULL" then no options are added.

$autosync = $h->get_autosync ();

Get the autosync flag.

$direct = $h->get_direct ();

Return the direct appliance mode flag.

$label = $h->get_e2label ($device);

This returns the ext2/3/4 filesystem label of the filesystem on "device".

This function is deprecated. In new code, use the "vfs_label" call instead.

Deprecated functions will not be removed from the API, but the fact that they are deprecated indicates that there are problems with correct use of these functions.

$uuid = $h->get_e2uuid ($device);

This returns the ext2/3/4 filesystem UUID of the filesystem on "device".

This function is deprecated. In new code, use the "vfs_uuid" call instead.

Deprecated functions will not be removed from the API, but the fact that they are deprecated indicates that there are problems with correct use of these functions.

$memsize = $h->get_memsize ();

This gets the memory size in megabytes allocated to the qemu subprocess.

If "$h->set_memsize" was not called on this handle, and if "LIBGUESTFS_MEMSIZE" was not set, then this returns the compiled-in default value for memsize.

For more information on the architecture of libguestfs, see guestfs(3).

$network = $h->get_network ();

This returns the enable network flag.

$path = $h->get_path ();

Return the current search path.

This is always non-NULL. If it wasn't set already, then this will return the default path.

$pid = $h->get_pid ();

Return the process ID of the qemu subprocess. If there is no qemu subprocess, then this will return an error.

This is an internal call used for debugging and testing.

$qemu = $h->get_qemu ();

Return the current qemu binary.

This is always non-NULL. If it wasn't set already, then this will return the default qemu binary name.

$recoveryproc = $h->get_recovery_proc ();

Return the recovery process enabled flag.

$selinux = $h->get_selinux ();

This returns the current setting of the selinux flag which is passed to the appliance at boot time. See "$h->set_selinux".

For more information on the architecture of libguestfs, see guestfs(3).

$state = $h->get_state ();

This returns the current state as an opaque integer. This is only useful for printing debug and internal error messages.

For more information on states, see guestfs(3).

$trace = $h->get_trace ();

Return the command trace flag.

$mask = $h->get_umask ();

Return the current umask. By default the umask is 022 unless it has been set by calling "$h->umask".

$verbose = $h->get_verbose ();

This returns the verbose messages flag.

$context = $h->getcon ();

This gets the SELinux security context of the daemon.

See the documentation about SELINUX in guestfs(3), and "$h->setcon"

@xattrs = $h->getxattrs ($path);

This call lists the extended attributes of the file or directory "path".

At the system call level, this is a combination of the listxattr(2) and getxattr(2) calls.

See also: "$h->lgetxattrs", attr(5).

@paths = $h->glob_expand ($pattern);

This command searches for all the pathnames matching "pattern" according to the wildcard expansion rules used by the shell.

If no paths match, then this returns an empty list (note: not an error).

It is just a wrapper around the C glob(3) function with flags "GLOB_MARK|GLOB_BRACE". See that manual page for more details.

@lines = $h->grep ($regex, $path);

This calls the external "grep" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@lines = $h->grepi ($regex, $path);

This calls the external "grep -i" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$h->grub_install ($root, $device);

This command installs GRUB (the Grand Unified Bootloader) on "device", with the root directory being "root".

Note: If grub-install reports the error ``No suitable drive was found in the generated device map.'' it may be that you need to create a "/boot/grub/device.map" file first that contains the mapping between grub device names and Linux device names. It is usually sufficient to create a file containing:

  (hd0) /dev/vda
 
 

replacing "/dev/vda" with the name of the installation device.

@lines = $h->head ($path);

This command returns up to the first 10 lines of a file as a list of strings.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@lines = $h->head_n ($nrlines, $path);

If the parameter "nrlines" is a positive number, this returns the first "nrlines" lines of the file "path".

If the parameter "nrlines" is a negative number, this returns lines from the file "path", excluding the last "nrlines" lines.

If the parameter "nrlines" is zero, this returns an empty list.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$dump = $h->hexdump ($path);

This runs "hexdump -C" on the given "path". The result is the human-readable, canonical hex dump of the file.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$content = $h->initrd_cat ($initrdpath, $filename);

This command unpacks the file "filename" from the initrd file called "initrdpath". The filename must be given without the initial "/" character.

For example, in guestfish you could use the following command to examine the boot script (usually called "/init") contained in a Linux initrd or initramfs image:

  initrd-cat /boot/initrd-<version>.img init
 
 

See also "$h->initrd_list".

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@filenames = $h->initrd_list ($path);

This command lists out files contained in an initrd.

The files are listed without any initial "/" character. The files are listed in the order they appear (not necessarily alphabetical). Directory names are listed as separate items.

Old Linux kernels (2.4 and earlier) used a compressed ext2 filesystem as initrd. We only support the newer initramfs format (compressed cpio files).

$wd = $h->inotify_add_watch ($path, $mask);

Watch "path" for the events listed in "mask".

Note that if "path" is a directory then events within that directory are watched, but this does not happen recursively (in subdirectories).

Note for non-C or non-Linux callers: the inotify events are defined by the Linux kernel ABI and are listed in "/usr/include/sys/inotify.h".

$h->inotify_close ();

This closes the inotify handle which was previously opened by inotify_init. It removes all watches, throws away any pending events, and deallocates all resources.

@paths = $h->inotify_files ();

This function is a helpful wrapper around "$h->inotify_read" which just returns a list of pathnames of objects that were touched. The returned pathnames are sorted and deduplicated.

$h->inotify_init ($maxevents);

This command creates a new inotify handle. The inotify subsystem can be used to notify events which happen to objects in the guest filesystem.

"maxevents" is the maximum number of events which will be queued up between calls to "$h->inotify_read" or "$h->inotify_files". If this is passed as 0, then the kernel (or previously set) default is used. For Linux 2.6.29 the default was 16384 events. Beyond this limit, the kernel throws away events, but records the fact that it threw them away by setting a flag "IN_Q_OVERFLOW" in the returned structure list (see "$h->inotify_read").

Before any events are generated, you have to add some watches to the internal watch list. See: "$h->inotify_add_watch", "$h->inotify_rm_watch" and "$h->inotify_watch_all".

Queued up events should be read periodically by calling "$h->inotify_read" (or "$h->inotify_files" which is just a helpful wrapper around "$h->inotify_read"). If you don't read the events out often enough then you risk the internal queue overflowing.

The handle should be closed after use by calling "$h->inotify_close". This also removes any watches automatically.

See also inotify(7) for an overview of the inotify interface as exposed by the Linux kernel, which is roughly what we expose via libguestfs. Note that there is one global inotify handle per libguestfs instance.

@events = $h->inotify_read ();

Return the complete queue of events that have happened since the previous read call.

If no events have happened, this returns an empty list.

Note: In order to make sure that all events have been read, you must call this function repeatedly until it returns an empty list. The reason is that the call will read events up to the maximum appliance-to-host message size and leave remaining events in the queue.

$h->inotify_rm_watch ($wd);

Remove a previously defined inotify watch. See "$h->inotify_add_watch".

$arch = $h->inspect_get_arch ($root);

This function should only be called with a root device string as returned by "$h->inspect_os".

This returns the architecture of the inspected operating system. The possible return values are listed under "$h->file_architecture".

If the architecture could not be determined, then the string "unknown" is returned.

Please read ``INSPECTION'' in guestfs(3) for more details.

$distro = $h->inspect_get_distro ($root);

This function should only be called with a root device string as returned by "$h->inspect_os".

This returns the distro (distribution) of the inspected operating system.

Currently defined distros are:

debian

Debian or a Debian-derived distro such as Ubuntu.

fedora

Fedora.

redhat-based

Some Red Hat-derived distro.

rhel

Red Hat Enterprise Linux and some derivatives.

windows

Windows does not have distributions. This string is returned if the OS type is Windows.

unknown

The distro could not be determined.

Future versions of libguestfs may return other strings here. The caller should be prepared to handle any string.
Please read ``INSPECTION'' in guestfs(3) for more details.

@filesystems = $h->inspect_get_filesystems ($root);

This function should only be called with a root device string as returned by "$h->inspect_os".

This returns a list of all the filesystems that we think are associated with this operating system. This includes the root filesystem, other ordinary filesystems, and non-mounted devices like swap partitions.

In the case of a multi-boot virtual machine, it is possible for a filesystem to be shared between operating systems.

Please read ``INSPECTION'' in guestfs(3) for more details. See also "$h->inspect_get_mountpoints".

$major = $h->inspect_get_major_version ($root);

This function should only be called with a root device string as returned by "$h->inspect_os".

This returns the major version number of the inspected operating system.

Windows uses a consistent versioning scheme which is not reflected in the popular public names used by the operating system. Notably the operating system known as ``Windows 7'' is really version 6.1 (ie. major = 6, minor = 1). You can find out the real versions corresponding to releases of Windows by consulting Wikipedia or MSDN.

If the version could not be determined, then 0 is returned.

Please read ``INSPECTION'' in guestfs(3) for more details.

$minor = $h->inspect_get_minor_version ($root);

This function should only be called with a root device string as returned by "$h->inspect_os".

This returns the minor version number of the inspected operating system.

If the version could not be determined, then 0 is returned.

Please read ``INSPECTION'' in guestfs(3) for more details. See also "$h->inspect_get_major_version".

%mountpoints = $h->inspect_get_mountpoints ($root);

This function should only be called with a root device string as returned by "$h->inspect_os".

This returns a hash of where we think the filesystems associated with this operating system should be mounted. Callers should note that this is at best an educated guess made by reading configuration files such as "/etc/fstab".

Each element in the returned hashtable has a key which is the path of the mountpoint (eg. "/boot") and a value which is the filesystem that would be mounted there (eg. "/dev/sda1").

Non-mounted devices such as swap devices are not returned in this list.

Please read ``INSPECTION'' in guestfs(3) for more details. See also "$h->inspect_get_filesystems".

$product = $h->inspect_get_product_name ($root);

This function should only be called with a root device string as returned by "$h->inspect_os".

This returns the product name of the inspected operating system. The product name is generally some freeform string which can be displayed to the user, but should not be parsed by programs.

If the product name could not be determined, then the string "unknown" is returned.

Please read ``INSPECTION'' in guestfs(3) for more details.

$name = $h->inspect_get_type ($root);

This function should only be called with a root device string as returned by "$h->inspect_os".

This returns the type of the inspected operating system. Currently defined types are:

linux

Any Linux-based operating system.

windows

Any Microsoft Windows operating system.

unknown

The operating system type could not be determined.

Future versions of libguestfs may return other strings here. The caller should be prepared to handle any string.
Please read ``INSPECTION'' in guestfs(3) for more details.

@roots = $h->inspect_os ();

This function uses other libguestfs functions and certain heuristics to inspect the disk(s) (usually disks belonging to a virtual machine), looking for operating systems.

The list returned is empty if no operating systems were found.

If one operating system was found, then this returns a list with a single element, which is the name of the root filesystem of this operating system. It is also possible for this function to return a list containing more than one element, indicating a dual-boot or multi-boot virtual machine, with each element being the root filesystem of one of the operating systems.

You can pass the root string(s) returned to other "$h->inspect_get_*" functions in order to query further information about each operating system, such as the name and version.

This function uses other libguestfs features such as "$h->mount_ro" and "$h->umount_all" in order to mount and unmount filesystems and look at the contents. This should be called with no disks currently mounted. The function may also use Augeas, so any existing Augeas handle will be closed.

This function cannot decrypt encrypted disks. The caller must do that first (supplying the necessary keys) if the disk is encrypted.

Please read ``INSPECTION'' in guestfs(3) for more details.

See also "$h->list_filesystems".

$flag = $h->is_blockdev ($path);

This returns "true" if and only if there is a block device with the given "path" name.

See also "$h->stat".

$busy = $h->is_busy ();

This returns true iff this handle is busy processing a command (in the "BUSY" state).

For more information on states, see guestfs(3).

$flag = $h->is_chardev ($path);

This returns "true" if and only if there is a character device with the given "path" name.

See also "$h->stat".

$config = $h->is_config ();

This returns true iff this handle is being configured (in the "CONFIG" state).

For more information on states, see guestfs(3).

$dirflag = $h->is_dir ($path);

This returns "true" if and only if there is a directory with the given "path" name. Note that it returns false for other objects like files.

See also "$h->stat".

$flag = $h->is_fifo ($path);

This returns "true" if and only if there is a FIFO (named pipe) with the given "path" name.

See also "$h->stat".

$fileflag = $h->is_file ($path);

This returns "true" if and only if there is a regular file with the given "path" name. Note that it returns false for other objects like directories.

See also "$h->stat".

$launching = $h->is_launching ();

This returns true iff this handle is launching the subprocess (in the "LAUNCHING" state).

For more information on states, see guestfs(3).

$lvflag = $h->is_lv ($device);

This command tests whether "device" is a logical volume, and returns true iff this is the case.

$ready = $h->is_ready ();

This returns true iff this handle is ready to accept commands (in the "READY" state).

For more information on states, see guestfs(3).

$flag = $h->is_socket ($path);

This returns "true" if and only if there is a Unix domain socket with the given "path" name.

See also "$h->stat".

$flag = $h->is_symlink ($path);

This returns "true" if and only if there is a symbolic link with the given "path" name.

See also "$h->stat".

$h->kill_subprocess ();

This kills the qemu subprocess. You should never need to call this.

$h->launch ();

Internally libguestfs is implemented by running a virtual machine using qemu(1).

You should call this after configuring the handle (eg. adding drives) but before performing any actions.

$h->lchown ($owner, $group, $path);

Change the file owner to "owner" and group to "group". This is like "$h->chown" but if "path" is a symlink then the link itself is changed, not the target.

Only numeric uid and gid are supported. If you want to use names, you will need to locate and parse the password file yourself (Augeas support makes this relatively easy).

@xattrs = $h->lgetxattrs ($path);

This is the same as "$h->getxattrs", but if "path" is a symbolic link, then it returns the extended attributes of the link itself.

@devices = $h->list_devices ();

List all the block devices.

The full block device names are returned, eg. "/dev/sda".

See also "$h->list_filesystems".

%fses = $h->list_filesystems ();

This inspection command looks for filesystems on partitions, block devices and logical volumes, returning a list of devices containing filesystems and their type.

The return value is a hash, where the keys are the devices containing filesystems, and the values are the filesystem types. For example:

  "/dev/sda1" => "ntfs"
  "/dev/sda2" => "ext2"
  "/dev/vg_guest/lv_root" => "ext4"
  "/dev/vg_guest/lv_swap" => "swap"
 
 

The value can have the special value ``unknown'', meaning the content of the device is undetermined or empty. ``swap'' means a Linux swap partition.

This command runs other libguestfs commands, which might include "$h->mount" and "$h->umount", and therefore you should use this soon after launch and only when nothing is mounted.

Not all of the filesystems returned will be mountable. In particular, swap partitions are returned in the list. Also this command does not check that each filesystem found is valid and mountable, and some filesystems might be mountable but require special options. Filesystems may not all belong to a single logical operating system (use "$h->inspect_os" to look for OSes).

@partitions = $h->list_partitions ();

List all the partitions detected on all block devices.

The full partition device names are returned, eg. "/dev/sda1"

This does not return logical volumes. For that you will need to call "$h->lvs".

See also "$h->list_filesystems".

$listing = $h->ll ($directory);

List the files in "directory" (relative to the root directory, there is no cwd) in the format of 'ls -la'.

This command is mostly useful for interactive sessions. It is not intended that you try to parse the output string.

$h->ln ($target, $linkname);

This command creates a hard link using the "ln" command.

$h->ln_f ($target, $linkname);

This command creates a hard link using the "ln -f" command. The "-f" option removes the link ("linkname") if it exists already.

$h->ln_s ($target, $linkname);

This command creates a symbolic link using the "ln -s" command.

$h->ln_sf ($target, $linkname);

This command creates a symbolic link using the "ln -sf" command, The "-f" option removes the link ("linkname") if it exists already.

$h->lremovexattr ($xattr, $path);

This is the same as "$h->removexattr", but if "path" is a symbolic link, then it removes an extended attribute of the link itself.

@listing = $h->ls ($directory);

List the files in "directory" (relative to the root directory, there is no cwd). The '.' and '..' entries are not returned, but hidden files are shown.

This command is mostly useful for interactive sessions. Programs should probably use "$h->readdir" instead.

$h->lsetxattr ($xattr, $val, $vallen, $path);

This is the same as "$h->setxattr", but if "path" is a symbolic link, then it sets an extended attribute of the link itself.

%statbuf = $h->lstat ($path);

Returns file information for the given "path".

This is the same as "$h->stat" except that if "path" is a symbolic link, then the link is stat-ed, not the file it refers to.

This is the same as the lstat(2) system call.

@statbufs = $h->lstatlist ($path, \@names);

This call allows you to perform the "$h->lstat" operation on multiple files, where all files are in the directory "path". "names" is the list of files from this directory.

On return you get a list of stat structs, with a one-to-one correspondence to the "names" list. If any name did not exist or could not be lstat'd, then the "ino" field of that structure is set to "-1".

This call is intended for programs that want to efficiently list a directory contents without making many round-trips. See also "$h->lxattrlist" for a similarly efficient call for getting extended attributes. Very long directory listings might cause the protocol message size to be exceeded, causing this call to fail. The caller must split up such requests into smaller groups of names.

$h->luks_add_key ($device, $key, $newkey, $keyslot);

This command adds a new key on LUKS device "device". "key" is any existing key, and is used to access the device. "newkey" is the new key to add. "keyslot" is the key slot that will be replaced.

Note that if "keyslot" already contains a key, then this command will fail. You have to use "$h->luks_kill_slot" first to remove that key.

$h->luks_close ($device);

This closes a LUKS device that was created earlier by "$h->luks_open" or "$h->luks_open_ro". The "device" parameter must be the name of the LUKS mapping device (ie. "/dev/mapper/mapname") and not the name of the underlying block device.

$h->luks_format ($device, $key, $keyslot);

This command erases existing data on "device" and formats the device as a LUKS encrypted device. "key" is the initial key, which is added to key slot "slot". (LUKS supports 8 key slots, numbered 0-7).

This command is dangerous. Without careful use you can easily destroy all your data.

$h->luks_format_cipher ($device, $key, $keyslot, $cipher);

This command is the same as "$h->luks_format" but it also allows you to set the "cipher" used.

This command is dangerous. Without careful use you can easily destroy all your data.

$h->luks_kill_slot ($device, $key, $keyslot);

This command deletes the key in key slot "keyslot" from the encrypted LUKS device "device". "key" must be one of the other keys.

$h->luks_open ($device, $key, $mapname);

This command opens a block device which has been encrypted according to the Linux Unified Key Setup (LUKS) standard.

"device" is the encrypted block device or partition.

The caller must supply one of the keys associated with the LUKS block device, in the "key" parameter.

This creates a new block device called "/dev/mapper/mapname". Reads and writes to this block device are decrypted from and encrypted to the underlying "device" respectively.

If this block device contains LVM volume groups, then calling "$h->vgscan" followed by "$h->vg_activate_all" will make them visible.

$h->luks_open_ro ($device, $key, $mapname);

This is the same as "$h->luks_open" except that a read-only mapping is created.

$h->lvcreate ($logvol, $volgroup, $mbytes);

This creates an LVM logical volume called "logvol" on the volume group "volgroup", with "size" megabytes.

$h->lvm_clear_filter ();

This undoes the effect of "$h->lvm_set_filter". LVM will be able to see every block device.

This command also clears the LVM cache and performs a volume group scan.

$h->lvm_remove_all ();

This command removes all LVM logical volumes, volume groups and physical volumes.

This command is dangerous. Without careful use you can easily destroy all your data.

$h->lvm_set_filter (\@devices);

This sets the LVM device filter so that LVM will only be able to ``see'' the block devices in the list "devices", and will ignore all other attached block devices.

Where disk image(s) contain duplicate PVs or VGs, this command is useful to get LVM to ignore the duplicates, otherwise LVM can get confused. Note also there are two types of duplication possible: either cloned PVs/VGs which have identical UUIDs; or VGs that are not cloned but just happen to have the same name. In normal operation you cannot create this situation, but you can do it outside LVM, eg. by cloning disk images or by bit twiddling inside the LVM metadata.

This command also clears the LVM cache and performs a volume group scan.

You can filter whole block devices or individual partitions.

You cannot use this if any VG is currently in use (eg. contains a mounted filesystem), even if you are not filtering out that VG.

$h->lvremove ($device);

Remove an LVM logical volume "device", where "device" is the path to the LV, such as "/dev/VG/LV".

You can also remove all LVs in a volume group by specifying the VG name, "/dev/VG".

$h->lvrename ($logvol, $newlogvol);

Rename a logical volume "logvol" with the new name "newlogvol".

$h->lvresize ($device, $mbytes);

This resizes (expands or shrinks) an existing LVM logical volume to "mbytes". When reducing, data in the reduced part is lost.

$h->lvresize_free ($lv, $percent);

This expands an existing logical volume "lv" so that it fills "pc"% of the remaining free space in the volume group. Commonly you would call this with pc = 100 which expands the logical volume as much as possible, using all remaining free space in the volume group.

@logvols = $h->lvs ();

List all the logical volumes detected. This is the equivalent of the lvs(8) command.

This returns a list of the logical volume device names (eg. "/dev/VolGroup00/LogVol00").

See also "$h->lvs_full", "$h->list_filesystems".

@logvols = $h->lvs_full ();

List all the logical volumes detected. This is the equivalent of the lvs(8) command. The ``full'' version includes all fields.

$uuid = $h->lvuuid ($device);

This command returns the UUID of the LVM LV "device".

@xattrs = $h->lxattrlist ($path, \@names);

This call allows you to get the extended attributes of multiple files, where all files are in the directory "path". "names" is the list of files from this directory.

On return you get a flat list of xattr structs which must be interpreted sequentially. The first xattr struct always has a zero-length "attrname". "attrval" in this struct is zero-length to indicate there was an error doing "lgetxattr" for this file, or is a C string which is a decimal number (the number of following attributes for this file, which could be "0"). Then after the first xattr struct are the zero or more attributes for the first named file. This repeats for the second and subsequent files.

This call is intended for programs that want to efficiently list a directory contents without making many round-trips. See also "$h->lstatlist" for a similarly efficient call for getting standard stats. Very long directory listings might cause the protocol message size to be exceeded, causing this call to fail. The caller must split up such requests into smaller groups of names.

$h->mkdir ($path);

Create a directory named "path".

$h->mkdir_mode ($path, $mode);

This command creates a directory, setting the initial permissions of the directory to "mode".

For common Linux filesystems, the actual mode which is set will be "mode & ~umask & 01777". Non-native-Linux filesystems may interpret the mode in other ways.

See also "$h->mkdir", "$h->umask"

$h->mkdir_p ($path);

Create a directory named "path", creating any parent directories as necessary. This is like the "mkdir -p" shell command.

$dir = $h->mkdtemp ($template);

This command creates a temporary directory. The "template" parameter should be a full pathname for the temporary directory name with the final six characters being ``XXXXXX''.

For example: ``/tmp/myprogXXXXXX'' or ``/Temp/myprogXXXXXX'', the second one being suitable for Windows filesystems.

The name of the temporary directory that was created is returned.

The temporary directory is created with mode 0700 and is owned by root.

The caller is responsible for deleting the temporary directory and its contents after use.

See also: mkdtemp(3)

$h->mke2fs_J ($fstype, $blocksize, $device, $journal);

This creates an ext2/3/4 filesystem on "device" with an external journal on "journal". It is equivalent to the command:

  mke2fs -t fstype -b blocksize -J device=<journal> <device>
 
 

See also "$h->mke2journal".

$h->mke2fs_JL ($fstype, $blocksize, $device, $label);

This creates an ext2/3/4 filesystem on "device" with an external journal on the journal labeled "label".

See also "$h->mke2journal_L".

$h->mke2fs_JU ($fstype, $blocksize, $device, $uuid);

This creates an ext2/3/4 filesystem on "device" with an external journal on the journal with UUID "uuid".

See also "$h->mke2journal_U".

$h->mke2journal ($blocksize, $device);

This creates an ext2 external journal on "device". It is equivalent to the command:

  mke2fs -O journal_dev -b blocksize device
 
 

$h->mke2journal_L ($blocksize, $label, $device);

This creates an ext2 external journal on "device" with label "label".

$h->mke2journal_U ($blocksize, $uuid, $device);

This creates an ext2 external journal on "device" with UUID "uuid".

$h->mkfifo ($mode, $path);

This call creates a FIFO (named pipe) called "path" with mode "mode". It is just a convenient wrapper around "$h->mknod".

The mode actually set is affected by the umask.

$h->mkfs ($fstype, $device);

This creates a filesystem on "device" (usually a partition or LVM logical volume). The filesystem type is "fstype", for example "ext3".

$h->mkfs_b ($fstype, $blocksize, $device);

This call is similar to "$h->mkfs", but it allows you to control the block size of the resulting filesystem. Supported block sizes depend on the filesystem type, but typically they are 1024, 2048 or 4096 only.

For VFAT and NTFS the "blocksize" parameter is treated as the requested cluster size.

$h->mkmountpoint ($exemptpath);

"$h->mkmountpoint" and "$h->rmmountpoint" are specialized calls that can be used to create extra mountpoints before mounting the first filesystem.

These calls are only necessary in some very limited circumstances, mainly the case where you want to mount a mix of unrelated and/or read-only filesystems together.

For example, live CDs often contain a ``Russian doll'' nest of filesystems, an ISO outer layer, with a squashfs image inside, with an ext2/3 image inside that. You can unpack this as follows in guestfish:

  add-ro Fedora-11-i686-Live.iso
  run
  mkmountpoint /cd
  mkmountpoint /squash
  mkmountpoint /ext3
  mount /dev/sda /cd
  mount-loop /cd/LiveOS/squashfs.img /squash
  mount-loop /squash/LiveOS/ext3fs.img /ext3
 
 

The inner filesystem is now unpacked under the /ext3 mountpoint.

$h->mknod ($mode, $devmajor, $devminor, $path);

This call creates block or character special devices, or named pipes (FIFOs).

The "mode" parameter should be the mode, using the standard constants. "devmajor" and "devminor" are the device major and minor numbers, only used when creating block and character special devices.

Note that, just like mknod(2), the mode must be bitwise OR'd with S_IFBLK, S_IFCHR, S_IFIFO or S_IFSOCK (otherwise this call just creates a regular file). These constants are available in the standard Linux header files, or you can use "$h->mknod_b", "$h->mknod_c" or "$h->mkfifo" which are wrappers around this command which bitwise OR in the appropriate constant for you.

The mode actually set is affected by the umask.

$h->mknod_b ($mode, $devmajor, $devminor, $path);

This call creates a block device node called "path" with mode "mode" and device major/minor "devmajor" and "devminor". It is just a convenient wrapper around "$h->mknod".

The mode actually set is affected by the umask.

$h->mknod_c ($mode, $devmajor, $devminor, $path);

This call creates a char device node called "path" with mode "mode" and device major/minor "devmajor" and "devminor". It is just a convenient wrapper around "$h->mknod".

The mode actually set is affected by the umask.

$h->mkswap ($device);

Create a swap partition on "device".

$h->mkswap_L ($label, $device);

Create a swap partition on "device" with label "label".

Note that you cannot attach a swap label to a block device (eg. "/dev/sda"), just to a partition. This appears to be a limitation of the kernel or swap tools.

$h->mkswap_U ($uuid, $device);

Create a swap partition on "device" with UUID "uuid".

$h->mkswap_file ($path);

Create a swap file.

This command just writes a swap file signature to an existing file. To create the file itself, use something like "$h->fallocate".

$h->modprobe ($modulename);

This loads a kernel module in the appliance.

The kernel module must have been whitelisted when libguestfs was built (see "appliance/kmod.whitelist.in" in the source).

$h->mount ($device, $mountpoint);

Mount a guest disk at a position in the filesystem. Block devices are named "/dev/sda", "/dev/sdb" and so on, as they were added to the guest. If those block devices contain partitions, they will have the usual names (eg. "/dev/sda1"). Also LVM "/dev/VG/LV"-style names can be used.

The rules are the same as for mount(2): A filesystem must first be mounted on "/" before others can be mounted. Other filesystems can only be mounted on directories which already exist.

The mounted filesystem is writable, if we have sufficient permissions on the underlying device.

Important note: When you use this call, the filesystem options "sync" and "noatime" are set implicitly. This was originally done because we thought it would improve reliability, but it turns out that -o sync has a very large negative performance impact and negligible effect on reliability. Therefore we recommend that you avoid using "$h->mount" in any code that needs performance, and instead use "$h->mount_options" (use an empty string for the first parameter if you don't want any options).

$h->mount_loop ($file, $mountpoint);

This command lets you mount "file" (a filesystem image in a file) on a mount point. It is entirely equivalent to the command "mount -o loop file mountpoint".

$h->mount_options ($options, $device, $mountpoint);

This is the same as the "$h->mount" command, but it allows you to set the mount options as for the mount(8) -o flag.

If the "options" parameter is an empty string, then no options are passed (all options default to whatever the filesystem uses).

$h->mount_ro ($device, $mountpoint);

This is the same as the "$h->mount" command, but it mounts the filesystem with the read-only (-o ro) flag.

$h->mount_vfs ($options, $vfstype, $device, $mountpoint);

This is the same as the "$h->mount" command, but it allows you to set both the mount options and the vfstype as for the mount(8) -o and -t flags.

%mps = $h->mountpoints ();

This call is similar to "$h->mounts". That call returns a list of devices. This one returns a hash table (map) of device name to directory where the device is mounted.

@devices = $h->mounts ();

This returns the list of currently mounted filesystems. It returns the list of devices (eg. "/dev/sda1", "/dev/VG/LV").

Some internal mounts are not shown.

See also: "$h->mountpoints"

$h->mv ($src, $dest);

This moves a file from "src" to "dest" where "dest" is either a destination filename or destination directory.

$status = $h->ntfs_3g_probe ($rw, $device);

This command runs the ntfs-3g.probe(8) command which probes an NTFS "device" for mountability. (Not all NTFS volumes can be mounted read-write, and some cannot be mounted at all).

"rw" is a boolean flag. Set it to true if you want to test if the volume can be mounted read-write. Set it to false if you want to test if the volume can be mounted read-only.

The return value is an integer which 0 if the operation would succeed, or some non-zero value documented in the ntfs-3g.probe(8) manual page.

$h->ntfsresize ($device);

This command resizes an NTFS filesystem, expanding or shrinking it to the size of the underlying device. See also ntfsresize(8).

$h->ntfsresize_size ($device, $size);

This command is the same as "$h->ntfsresize" except that it allows you to specify the new size (in bytes) explicitly.

$h->part_add ($device, $prlogex, $startsect, $endsect);

This command adds a partition to "device". If there is no partition table on the device, call "$h->part_init" first.

The "prlogex" parameter is the type of partition. Normally you should pass "p" or "primary" here, but MBR partition tables also support "l" (or "logical") and "e" (or "extended") partition types.

"startsect" and "endsect" are the start and end of the partition in sectors. "endsect" may be negative, which means it counts backwards from the end of the disk ("-1" is the last sector).

Creating a partition which covers the whole disk is not so easy. Use "$h->part_disk" to do that.

$h->part_del ($device, $partnum);

This command deletes the partition numbered "partnum" on "device".

Note that in the case of MBR partitioning, deleting an extended partition also deletes any logical partitions it contains.

$h->part_disk ($device, $parttype);

This command is simply a combination of "$h->part_init" followed by "$h->part_add" to create a single primary partition covering the whole disk.

"parttype" is the partition table type, usually "mbr" or "gpt", but other possible values are described in "$h->part_init".

This command is dangerous. Without careful use you can easily destroy all your data.

$bootable = $h->part_get_bootable ($device, $partnum);

This command returns true if the partition "partnum" on "device" has the bootable flag set.

See also "$h->part_set_bootable".

$idbyte = $h->part_get_mbr_id ($device, $partnum);

Returns the MBR type byte (also known as the ID byte) from the numbered partition "partnum".

Note that only MBR (old DOS-style) partitions have type bytes. You will get undefined results for other partition table types (see "$h->part_get_parttype").

$parttype = $h->part_get_parttype ($device);

This command examines the partition table on "device" and returns the partition table type (format) being used.

Common return values include: "msdos" (a DOS/Windows style MBR partition table), "gpt" (a GPT/EFI-style partition table). Other values are possible, although unusual. See "$h->part_init" for a full list.

$h->part_init ($device, $parttype);

This creates an empty partition table on "device" of one of the partition types listed below. Usually "parttype" should be either "msdos" or "gpt" (for large disks).

Initially there are no partitions. Following this, you should call "$h->part_add" for each partition required.

Possible values for "parttype" are:

efi | gpt

Intel EFI / GPT partition table.

This is recommended for >= 2 TB partitions that will be accessed from Linux and Intel-based Mac OS X. It also has limited backwards compatibility with the "mbr" format.

mbr | msdos

The standard PC ``Master Boot Record'' (MBR) format used by MS-DOS and Windows. This partition type will only work for device sizes up to 2 TB. For large disks we recommend using "gpt".

Other partition table types that may work but are not supported include:

aix

AIX disk labels.

amiga | rdb

Amiga ``Rigid Disk Block'' format.

bsd

BSD disk labels.

dasd

DASD, used on IBM mainframes.

dvh

MIPS/SGI volumes.

mac

Old Mac partition format. Modern Macs use "gpt".

pc98

NEC PC-98 format, common in Japan apparently.

sun

Sun disk labels.

@partitions = $h->part_list ($device);

This command parses the partition table on "device" and returns the list of partitions found.

The fields in the returned structure are:

part_num

Partition number, counting from 1.

part_start

Start of the partition in bytes. To get sectors you have to divide by the device's sector size, see "$h->blockdev_getss".

part_end

End of the partition in bytes.

part_size

Size of the partition in bytes.

$h->part_set_bootable ($device, $partnum, $bootable);

This sets the bootable flag on partition numbered "partnum" on device "device". Note that partitions are numbered from 1.

The bootable flag is used by some operating systems (notably Windows) to determine which partition to boot from. It is by no means universally recognized.

$h->part_set_mbr_id ($device, $partnum, $idbyte);

Sets the MBR type byte (also known as the ID byte) of the numbered partition "partnum" to "idbyte". Note that the type bytes quoted in most documentation are in fact hexadecimal numbers, but usually documented without any leading ``0x'' which might be confusing.

Note that only MBR (old DOS-style) partitions have type bytes. You will get undefined results for other partition table types (see "$h->part_get_parttype").

$h->part_set_name ($device, $partnum, $name);

This sets the partition name on partition numbered "partnum" on device "device". Note that partitions are numbered from 1.

The partition name can only be set on certain types of partition table. This works on "gpt" but not on "mbr" partitions.

$device = $h->part_to_dev ($partition);

This function takes a partition name (eg. ``/dev/sdb1'') and removes the partition number, returning the device name (eg. ``/dev/sdb'').

The named partition must exist, for example as a string returned from "$h->list_partitions".

$h->ping_daemon ();

This is a test probe into the guestfs daemon running inside the qemu subprocess. Calling this function checks that the daemon responds to the ping message, without affecting the daemon or attached block device(s) in any other way.

$content = $h->pread ($path, $count, $offset);

This command lets you read part of a file. It reads "count" bytes of the file, starting at "offset", from file "path".

This may read fewer bytes than requested. For further details see the pread(2) system call.

See also "$h->pwrite", "$h->pread_device".

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$content = $h->pread_device ($device, $count, $offset);

This command lets you read part of a file. It reads "count" bytes of "device", starting at "offset".

This may read fewer bytes than requested. For further details see the pread(2) system call.

See also "$h->pread".

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$h->pvcreate ($device);

This creates an LVM physical volume on the named "device", where "device" should usually be a partition name such as "/dev/sda1".

$h->pvremove ($device);

This wipes a physical volume "device" so that LVM will no longer recognise it.

The implementation uses the "pvremove" command which refuses to wipe physical volumes that contain any volume groups, so you have to remove those first.

$h->pvresize ($device);

This resizes (expands or shrinks) an existing LVM physical volume to match the new size of the underlying device.

$h->pvresize_size ($device, $size);

This command is the same as "$h->pvresize" except that it allows you to specify the new size (in bytes) explicitly.

@physvols = $h->pvs ();

List all the physical volumes detected. This is the equivalent of the pvs(8) command.

This returns a list of just the device names that contain PVs (eg. "/dev/sda2").

See also "$h->pvs_full".

@physvols = $h->pvs_full ();

List all the physical volumes detected. This is the equivalent of the pvs(8) command. The ``full'' version includes all fields.

$uuid = $h->pvuuid ($device);

This command returns the UUID of the LVM PV "device".

$nbytes = $h->pwrite ($path, $content, $offset);

This command writes to part of a file. It writes the data buffer "content" to the file "path" starting at offset "offset".

This command implements the pwrite(2) system call, and like that system call it may not write the full data requested. The return value is the number of bytes that were actually written to the file. This could even be 0, although short writes are unlikely for regular files in ordinary circumstances.

See also "$h->pread", "$h->pwrite_device".

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$nbytes = $h->pwrite_device ($device, $content, $offset);

This command writes to part of a device. It writes the data buffer "content" to "device" starting at offset "offset".

This command implements the pwrite(2) system call, and like that system call it may not write the full data requested (although short writes to disk devices and partitions are probably impossible with standard Linux kernels).

See also "$h->pwrite".

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$content = $h->read_file ($path);

This calls returns the contents of the file "path" as a buffer.

Unlike "$h->cat", this function can correctly handle files that contain embedded ASCII NUL characters. However unlike "$h->download", this function is limited in the total size of file that can be handled.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@lines = $h->read_lines ($path);

Return the contents of the file named "path".

The file contents are returned as a list of lines. Trailing "LF" and "CRLF" character sequences are not returned.

Note that this function cannot correctly handle binary files (specifically, files containing "\0" character which is treated as end of line). For those you need to use the "$h->read_file" function which has a more complex interface.

@entries = $h->readdir ($dir);

This returns the list of directory entries in directory "dir".

All entries in the directory are returned, including "." and "..". The entries are not sorted, but returned in the same order as the underlying filesystem.

Also this call returns basic file type information about each file. The "ftyp" field will contain one of the following characters:

'b'

Block special

'c'

Char special

'd'

Directory

'f'

FIFO (named pipe)

'l'

Symbolic link

'r'

Regular file

's'

Socket

'u'

Unknown file type

'?'

The readdir(3) call returned a "d_type" field with an unexpected value

This function is primarily intended for use by programs. To get a simple list of names, use "$h->ls". To get a printable directory for human consumption, use "$h->ll".

$link = $h->readlink ($path);

This command reads the target of a symbolic link.

@links = $h->readlinklist ($path, \@names);

This call allows you to do a "readlink" operation on multiple files, where all files are in the directory "path". "names" is the list of files from this directory.

On return you get a list of strings, with a one-to-one correspondence to the "names" list. Each string is the value of the symbolic link.

If the readlink(2) operation fails on any name, then the corresponding result string is the empty string "". However the whole operation is completed even if there were readlink(2) errors, and so you can call this function with names where you don't know if they are symbolic links already (albeit slightly less efficient).

This call is intended for programs that want to efficiently list a directory contents without making many round-trips. Very long directory listings might cause the protocol message size to be exceeded, causing this call to fail. The caller must split up such requests into smaller groups of names.

$rpath = $h->realpath ($path);

Return the canonicalized absolute pathname of "path". The returned path has no ".", ".." or symbolic link path elements.

$h->removexattr ($xattr, $path);

This call removes the extended attribute named "xattr" of the file "path".

See also: "$h->lremovexattr", attr(5).

$h->resize2fs ($device);

This resizes an ext2, ext3 or ext4 filesystem to match the size of the underlying device.

Note: It is sometimes required that you run "$h->e2fsck_f" on the "device" before calling this command. For unknown reasons "resize2fs" sometimes gives an error about this and sometimes not. In any case, it is always safe to call "$h->e2fsck_f" before calling this function.

$h->resize2fs_size ($device, $size);

This command is the same as "$h->resize2fs" except that it allows you to specify the new size (in bytes) explicitly.

$h->rm ($path);

Remove the single file "path".

$h->rm_rf ($path);

Remove the file or directory "path", recursively removing the contents if its a directory. This is like the "rm -rf" shell command.

$h->rmdir ($path);

Remove the single directory "path".

$h->rmmountpoint ($exemptpath);

This calls removes a mountpoint that was previously created with "$h->mkmountpoint". See "$h->mkmountpoint" for full details.

$h->scrub_device ($device);

This command writes patterns over "device" to make data retrieval more difficult.

It is an interface to the scrub(1) program. See that manual page for more details.

This command is dangerous. Without careful use you can easily destroy all your data.

$h->scrub_file ($file);

This command writes patterns over a file to make data retrieval more difficult.

The file is removed after scrubbing.

It is an interface to the scrub(1) program. See that manual page for more details.

$h->scrub_freespace ($dir);

This command creates the directory "dir" and then fills it with files until the filesystem is full, and scrubs the files as for "$h->scrub_file", and deletes them. The intention is to scrub any free space on the partition containing "dir".

It is an interface to the scrub(1) program. See that manual page for more details.

$h->set_append ($append);

This function is used to add additional options to the guest kernel command line.

The default is "NULL" unless overridden by setting "LIBGUESTFS_APPEND" environment variable.

Setting "append" to "NULL" means no additional options are passed (libguestfs always adds a few of its own).

$h->set_autosync ($autosync);

If "autosync" is true, this enables autosync. Libguestfs will make a best effort attempt to run "$h->umount_all" followed by "$h->sync" when the handle is closed (also if the program exits without closing handles).

This is disabled by default (except in guestfish where it is enabled by default).

$h->set_direct ($direct);

If the direct appliance mode flag is enabled, then stdin and stdout are passed directly through to the appliance once it is launched.

One consequence of this is that log messages aren't caught by the library and handled by "$h->set_log_message_callback", but go straight to stdout.

You probably don't want to use this unless you know what you are doing.

The default is disabled.

$h->set_e2label ($device, $label);

This sets the ext2/3/4 filesystem label of the filesystem on "device" to "label". Filesystem labels are limited to 16 characters.

You can use either "$h->tune2fs_l" or "$h->get_e2label" to return the existing label on a filesystem.

$h->set_e2uuid ($device, $uuid);

This sets the ext2/3/4 filesystem UUID of the filesystem on "device" to "uuid". The format of the UUID and alternatives such as "clear", "random" and "time" are described in the tune2fs(8) manpage.

You can use either "$h->tune2fs_l" or "$h->get_e2uuid" to return the existing UUID of a filesystem.

$h->set_memsize ($memsize);

This sets the memory size in megabytes allocated to the qemu subprocess. This only has any effect if called before "$h->launch".

You can also change this by setting the environment variable "LIBGUESTFS_MEMSIZE" before the handle is created.

For more information on the architecture of libguestfs, see guestfs(3).

$h->set_network ($network);

If "network" is true, then the network is enabled in the libguestfs appliance. The default is false.

This affects whether commands are able to access the network (see ``RUNNING COMMANDS'' in guestfs(3)).

You must call this before calling "$h->launch", otherwise it has no effect.

$h->set_path ($searchpath);

Set the path that libguestfs searches for kernel and initrd.img.

The default is "$libdir/guestfs" unless overridden by setting "LIBGUESTFS_PATH" environment variable.

Setting "path" to "NULL" restores the default path.

$h->set_qemu ($qemu);

Set the qemu binary that we will use.

The default is chosen when the library was compiled by the configure script.

You can also override this by setting the "LIBGUESTFS_QEMU" environment variable.

Setting "qemu" to "NULL" restores the default qemu binary.

Note that you should call this function as early as possible after creating the handle. This is because some pre-launch operations depend on testing qemu features (by running "qemu -help"). If the qemu binary changes, we don't retest features, and so you might see inconsistent results. Using the environment variable "LIBGUESTFS_QEMU" is safest of all since that picks the qemu binary at the same time as the handle is created.

$h->set_recovery_proc ($recoveryproc);

If this is called with the parameter "false" then "$h->launch" does not create a recovery process. The purpose of the recovery process is to stop runaway qemu processes in the case where the main program aborts abruptly.

This only has any effect if called before "$h->launch", and the default is true.

About the only time when you would want to disable this is if the main process will fork itself into the background (``daemonize'' itself). In this case the recovery process thinks that the main program has disappeared and so kills qemu, which is not very helpful.

$h->set_selinux ($selinux);

This sets the selinux flag that is passed to the appliance at boot time. The default is "selinux=0" (disabled).

Note that if SELinux is enabled, it is always in Permissive mode ("enforcing=0").

For more information on the architecture of libguestfs, see guestfs(3).

$h->set_trace ($trace);

If the command trace flag is set to 1, then commands are printed on stderr before they are executed in a format which is very similar to the one used by guestfish. In other words, you can run a program with this enabled, and you will get out a script which you can feed to guestfish to perform the same set of actions.

If you want to trace C API calls into libguestfs (and other libraries) then possibly a better way is to use the external ltrace(1) command.

Command traces are disabled unless the environment variable "LIBGUESTFS_TRACE" is defined and set to 1.

$h->set_verbose ($verbose);

If "verbose" is true, this turns on verbose messages (to "stderr").

Verbose messages are disabled unless the environment variable "LIBGUESTFS_DEBUG" is defined and set to 1.

$h->setcon ($context);

This sets the SELinux security context of the daemon to the string "context".

See the documentation about SELINUX in guestfs(3).

$h->setxattr ($xattr, $val, $vallen, $path);

This call sets the extended attribute named "xattr" of the file "path" to the value "val" (of length "vallen"). The value is arbitrary 8 bit data.

See also: "$h->lsetxattr", attr(5).

$h->sfdisk ($device, $cyls, $heads, $sectors, \@lines);

This is a direct interface to the sfdisk(8) program for creating partitions on block devices.

"device" should be a block device, for example "/dev/sda".

"cyls", "heads" and "sectors" are the number of cylinders, heads and sectors on the device, which are passed directly to sfdisk as the -C, -H and -S parameters. If you pass 0 for any of these, then the corresponding parameter is omitted. Usually for 'large' disks, you can just pass 0 for these, but for small (floppy-sized) disks, sfdisk (or rather, the kernel) cannot work out the right geometry and you will need to tell it.

"lines" is a list of lines that we feed to "sfdisk". For more information refer to the sfdisk(8) manpage.

To create a single partition occupying the whole disk, you would pass "lines" as a single element list, when the single element being the string "," (comma).

See also: "$h->sfdisk_l", "$h->sfdisk_N", "$h->part_init"

This command is dangerous. Without careful use you can easily destroy all your data.

$h->sfdiskM ($device, \@lines);

This is a simplified interface to the "$h->sfdisk" command, where partition sizes are specified in megabytes only (rounded to the nearest cylinder) and you don't need to specify the cyls, heads and sectors parameters which were rarely if ever used anyway.

See also: "$h->sfdisk", the sfdisk(8) manpage and "$h->part_disk"

This command is dangerous. Without careful use you can easily destroy all your data.

$h->sfdisk_N ($device, $partnum, $cyls, $heads, $sectors, $line);

This runs sfdisk(8) option to modify just the single partition "n" (note: "n" counts from 1).

For other parameters, see "$h->sfdisk". You should usually pass 0 for the cyls/heads/sectors parameters.

See also: "$h->part_add"

This command is dangerous. Without careful use you can easily destroy all your data.

$partitions = $h->sfdisk_disk_geometry ($device);

This displays the disk geometry of "device" read from the partition table. Especially in the case where the underlying block device has been resized, this can be different from the kernel's idea of the geometry (see "$h->sfdisk_kernel_geometry").

The result is in human-readable format, and not designed to be parsed.

$partitions = $h->sfdisk_kernel_geometry ($device);

This displays the kernel's idea of the geometry of "device".

The result is in human-readable format, and not designed to be parsed.

$partitions = $h->sfdisk_l ($device);

This displays the partition table on "device", in the human-readable output of the sfdisk(8) command. It is not intended to be parsed.

See also: "$h->part_list"

$output = $h->sh ($command);

This call runs a command from the guest filesystem via the guest's "/bin/sh".

This is like "$h->command", but passes the command to:

  /bin/sh -c "command"
 
 

Depending on the guest's shell, this usually results in wildcards being expanded, shell expressions being interpolated and so on.

All the provisos about "$h->command" apply to this call.

@lines = $h->sh_lines ($command);

This is the same as "$h->sh", but splits the result into a list of lines.

See also: "$h->command_lines"

$h->sleep ($secs);

Sleep for "secs" seconds.

%statbuf = $h->stat ($path);

Returns file information for the given "path".

This is the same as the stat(2) system call.

%statbuf = $h->statvfs ($path);

Returns file system statistics for any mounted file system. "path" should be a file or directory in the mounted file system (typically it is the mount point itself, but it doesn't need to be).

This is the same as the statvfs(2) system call.

@stringsout = $h->strings ($path);

This runs the strings(1) command on a file and returns the list of printable strings found.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@stringsout = $h->strings_e ($encoding, $path);

This is like the "$h->strings" command, but allows you to specify the encoding of strings that are looked for in the source file "path".

Allowed encodings are:

s

Single 7-bit-byte characters like ASCII and the ASCII-compatible parts of ISO-8859-X (this is what "$h->strings" uses).

S

Single 8-bit-byte characters.

b

16-bit big endian strings such as those encoded in UTF-16BE or UCS-2BE.

l (lower case letter L)

16-bit little endian such as UTF-16LE and UCS-2LE. This is useful for examining binaries in Windows guests.

B

32-bit big endian such as UCS-4BE.

L

32-bit little endian such as UCS-4LE.

The returned strings are transcoded to UTF-8.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$h->swapoff_device ($device);

This command disables the libguestfs appliance swap device or partition named "device". See "$h->swapon_device".

$h->swapoff_file ($file);

This command disables the libguestfs appliance swap on file.

$h->swapoff_label ($label);

This command disables the libguestfs appliance swap on labeled swap partition.

$h->swapoff_uuid ($uuid);

This command disables the libguestfs appliance swap partition with the given UUID.

$h->swapon_device ($device);

This command enables the libguestfs appliance to use the swap device or partition named "device". The increased memory is made available for all commands, for example those run using "$h->command" or "$h->sh".

Note that you should not swap to existing guest swap partitions unless you know what you are doing. They may contain hibernation information, or other information that the guest doesn't want you to trash. You also risk leaking information about the host to the guest this way. Instead, attach a new host device to the guest and swap on that.

$h->swapon_file ($file);

This command enables swap to a file. See "$h->swapon_device" for other notes.

$h->swapon_label ($label);

This command enables swap to a labeled swap partition. See "$h->swapon_device" for other notes.

$h->swapon_uuid ($uuid);

This command enables swap to a swap partition with the given UUID. See "$h->swapon_device" for other notes.

$h->sync ();

This syncs the disk, so that any writes are flushed through to the underlying disk image.

You should always call this if you have modified a disk image, before closing the handle.

@lines = $h->tail ($path);

This command returns up to the last 10 lines of a file as a list of strings.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@lines = $h->tail_n ($nrlines, $path);

If the parameter "nrlines" is a positive number, this returns the last "nrlines" lines of the file "path".

If the parameter "nrlines" is a negative number, this returns lines from the file "path", starting with the "-nrlines"th line.

If the parameter "nrlines" is zero, this returns an empty list.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$h->tar_in ($tarfile, $directory);

This command uploads and unpacks local file "tarfile" (an uncompressed tar file) into "directory".

To upload a compressed tarball, use "$h->tgz_in" or "$h->txz_in".

$h->tar_out ($directory, $tarfile);

This command packs the contents of "directory" and downloads it to local file "tarfile".

To download a compressed tarball, use "$h->tgz_out" or "$h->txz_out".

$h->tgz_in ($tarball, $directory);

This command uploads and unpacks local file "tarball" (a gzip compressed tar file) into "directory".

To upload an uncompressed tarball, use "$h->tar_in".

$h->tgz_out ($directory, $tarball);

This command packs the contents of "directory" and downloads it to local file "tarball".

To download an uncompressed tarball, use "$h->tar_out".

$h->touch ($path);

Touch acts like the touch(1) command. It can be used to update the timestamps on a file, or, if the file does not exist, to create a new zero-length file.

This command only works on regular files, and will fail on other file types such as directories, symbolic links, block special etc.

$h->truncate ($path);

This command truncates "path" to a zero-length file. The file must exist already.

$h->truncate_size ($path, $size);

This command truncates "path" to size "size" bytes. The file must exist already.

If the current file size is less than "size" then the file is extended to the required size with zero bytes. This creates a sparse file (ie. disk blocks are not allocated for the file until you write to it). To create a non-sparse file of zeroes, use "$h->fallocate64" instead.

%superblock = $h->tune2fs_l ($device);

This returns the contents of the ext2, ext3 or ext4 filesystem superblock on "device".

It is the same as running "tune2fs -l device". See tune2fs(8) manpage for more details. The list of fields returned isn't clearly defined, and depends on both the version of "tune2fs" that libguestfs was built against, and the filesystem itself.

$h->txz_in ($tarball, $directory);

This command uploads and unpacks local file "tarball" (an xz compressed tar file) into "directory".

$h->txz_out ($directory, $tarball);

This command packs the contents of "directory" and downloads it to local file "tarball" (as an xz compressed tar archive).

$oldmask = $h->umask ($mask);

This function sets the mask used for creating new files and device nodes to "mask & 0777".

Typical umask values would be 022 which creates new files with permissions like ``-rw-r---r--'' or ``-rwxr-xr-x'', and 002 which creates new files with permissions like ``-rw-rw-r--'' or ``-rwxrwxr-x''.

The default umask is 022. This is important because it means that directories and device nodes will be created with 0644 or 0755 mode even if you specify 0777.

See also "$h->get_umask", umask(2), "$h->mknod", "$h->mkdir".

This call returns the previous umask.

$h->umount ($pathordevice);

This unmounts the given filesystem. The filesystem may be specified either by its mountpoint (path) or the device which contains the filesystem.

$h->umount_all ();

This unmounts all mounted filesystems.

Some internal mounts are not unmounted by this call.

$h->upload ($filename, $remotefilename);

Upload local file "filename" to "remotefilename" on the filesystem.

"filename" can also be a named pipe.

See also "$h->download".

$h->upload_offset ($filename, $remotefilename, $offset);

Upload local file "filename" to "remotefilename" on the filesystem.

"remotefilename" is overwritten starting at the byte "offset" specified. The intention is to overwrite parts of existing files or devices, although if a non-existant file is specified then it is created with a ``hole'' before "offset". The size of the data written is implicit in the size of the source "filename".

Note that there is no limit on the amount of data that can be uploaded with this call, unlike with "$h->pwrite", and this call always writes the full amount unless an error occurs.

See also "$h->upload", "$h->pwrite".

$h->utimens ($path, $atsecs, $atnsecs, $mtsecs, $mtnsecs);

This command sets the timestamps of a file with nanosecond precision.

"atsecs, atnsecs" are the last access time (atime) in secs and nanoseconds from the epoch.

"mtsecs, mtnsecs" are the last modification time (mtime) in secs and nanoseconds from the epoch.

If the *nsecs field contains the special value "-1" then the corresponding timestamp is set to the current time. (The *secs field is ignored in this case).

If the *nsecs field contains the special value "-2" then the corresponding timestamp is left unchanged. (The *secs field is ignored in this case).

%version = $h->version ();

Return the libguestfs version number that the program is linked against.

Note that because of dynamic linking this is not necessarily the version of libguestfs that you compiled against. You can compile the program, and then at runtime dynamically link against a completely different "libguestfs.so" library.

This call was added in version 1.0.58. In previous versions of libguestfs there was no way to get the version number. From C code you can use dynamic linker functions to find out if this symbol exists (if it doesn't, then it's an earlier version).

The call returns a structure with four elements. The first three ("major", "minor" and "release") are numbers and correspond to the usual version triplet. The fourth element ("extra") is a string and is normally empty, but may be used for distro-specific information.

To construct the original version string: "$major.$minor.$release$extra"

See also: ``LIBGUESTFS VERSION NUMBERS'' in guestfs(3).

Note: Don't use this call to test for availability of features. In enterprise distributions we backport features from later versions into earlier versions, making this an unreliable way to test for features. Use "$h->available" instead.

$label = $h->vfs_label ($device);

This returns the filesystem label of the filesystem on "device".

If the filesystem is unlabeled, this returns the empty string.

To find a filesystem from the label, use "$h->findfs_label".

$fstype = $h->vfs_type ($device);

This command gets the filesystem type corresponding to the filesystem on "device".

For most filesystems, the result is the name of the Linux VFS module which would be used to mount this filesystem if you mounted it without specifying the filesystem type. For example a string such as "ext3" or "ntfs".

$uuid = $h->vfs_uuid ($device);

This returns the filesystem UUID of the filesystem on "device".

If the filesystem does not have a UUID, this returns the empty string.

To find a filesystem from the UUID, use "$h->findfs_uuid".

$h->vg_activate ($activate, \@volgroups);

This command activates or (if "activate" is false) deactivates all logical volumes in the listed volume groups "volgroups". If activated, then they are made known to the kernel, ie. they appear as "/dev/mapper" devices. If deactivated, then those devices disappear.

This command is the same as running "vgchange -a y|n volgroups..."

Note that if "volgroups" is an empty list then all volume groups are activated or deactivated.

$h->vg_activate_all ($activate);

This command activates or (if "activate" is false) deactivates all logical volumes in all volume groups. If activated, then they are made known to the kernel, ie. they appear as "/dev/mapper" devices. If deactivated, then those devices disappear.

This command is the same as running "vgchange -a y|n"

$h->vgcreate ($volgroup, \@physvols);

This creates an LVM volume group called "volgroup" from the non-empty list of physical volumes "physvols".

@uuids = $h->vglvuuids ($vgname);

Given a VG called "vgname", this returns the UUIDs of all the logical volumes created in this volume group.

You can use this along with "$h->lvs" and "$h->lvuuid" calls to associate logical volumes and volume groups.

See also "$h->vgpvuuids".

@uuids = $h->vgpvuuids ($vgname);

Given a VG called "vgname", this returns the UUIDs of all the physical volumes that this volume group resides on.

You can use this along with "$h->pvs" and "$h->pvuuid" calls to associate physical volumes and volume groups.

See also "$h->vglvuuids".

$h->vgremove ($vgname);

Remove an LVM volume group "vgname", (for example "VG").

This also forcibly removes all logical volumes in the volume group (if any).

$h->vgrename ($volgroup, $newvolgroup);

Rename a volume group "volgroup" with the new name "newvolgroup".

@volgroups = $h->vgs ();

List all the volumes groups detected. This is the equivalent of the vgs(8) command.

This returns a list of just the volume group names that were detected (eg. "VolGroup00").

See also "$h->vgs_full".

@volgroups = $h->vgs_full ();

List all the volumes groups detected. This is the equivalent of the vgs(8) command. The ``full'' version includes all fields.

$h->vgscan ();

This rescans all block devices and rebuilds the list of LVM physical volumes, volume groups and logical volumes.

$uuid = $h->vguuid ($vgname);

This command returns the UUID of the LVM VG named "vgname".

$h->wait_ready ();

This function is a no op.

In versions of the API < 1.0.71 you had to call this function just after calling "$h->launch" to wait for the launch to complete. However this is no longer necessary because "$h->launch" now does the waiting.

If you see any calls to this function in code then you can just remove them, unless you want to retain compatibility with older versions of the API.

$chars = $h->wc_c ($path);

This command counts the characters in a file, using the "wc -c" external command.

$lines = $h->wc_l ($path);

This command counts the lines in a file, using the "wc -l" external command.

$words = $h->wc_w ($path);

This command counts the words in a file, using the "wc -w" external command.

$h->write ($path, $content);

This call creates a file called "path". The content of the file is the string "content" (which can contain any 8 bit data).

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$h->write_file ($path, $content, $size);

This call creates a file called "path". The contents of the file is the string "content" (which can contain any 8 bit data), with length "size".

As a special case, if "size" is 0 then the length is calculated using "strlen" (so in this case the content cannot contain embedded ASCII NULs).

NB. Owing to a bug, writing content containing ASCII NUL characters does not work, even if the length is specified.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

This function is deprecated. In new code, use the "write" call instead.

Deprecated functions will not be removed from the API, but the fact that they are deprecated indicates that there are problems with correct use of these functions.

@lines = $h->zegrep ($regex, $path);

This calls the external "zegrep" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@lines = $h->zegrepi ($regex, $path);

This calls the external "zegrep -i" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$h->zero ($device);

This command writes zeroes over the first few blocks of "device".

How many blocks are zeroed isn't specified (but it's not enough to securely wipe the device). It should be sufficient to remove any partition tables, filesystem superblocks and so on.

See also: "$h->zero_device", "$h->scrub_device".

$h->zero_device ($device);

This command writes zeroes over the entire "device". Compare with "$h->zero" which just zeroes the first few blocks of a device.

This command is dangerous. Without careful use you can easily destroy all your data.

$h->zerofree ($device);

This runs the zerofree program on "device". This program claims to zero unused inodes and disk blocks on an ext2/3 filesystem, thus making it possible to compress the filesystem more effectively.

You should not run this program if the filesystem is mounted.

It is possible that using this program can damage the filesystem or data on the filesystem.

@lines = $h->zfgrep ($pattern, $path);

This calls the external "zfgrep" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@lines = $h->zfgrepi ($pattern, $path);

This calls the external "zfgrep -i" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

$description = $h->zfile ($meth, $path);

This command runs "file" after first decompressing "path" using "method".

"method" must be one of "gzip", "compress" or "bzip2".

Since 1.0.63, use "$h->file" instead which can now process compressed files.

This function is deprecated. In new code, use the "file" call instead.

Deprecated functions will not be removed from the API, but the fact that they are deprecated indicates that there are problems with correct use of these functions.

@lines = $h->zgrep ($regex, $path);

This calls the external "zgrep" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

@lines = $h->zgrepi ($regex, $path);

This calls the external "zgrep -i" program and returns the matching lines.

Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See ``PROTOCOL LIMITS'' in guestfs(3).

AVAILABILITY

From time to time we add new libguestfs APIs. Also some libguestfs APIs won't be available in all builds of libguestfs (the Fedora build is full-featured, but other builds may disable features). How do you test whether the APIs that your Perl program needs are available in the version of "Sys::Guestfs" that you are using?

To test if a particular function is available in the "Sys::Guestfs" class, use the ordinary Perl UNIVERSAL method "can(METHOD)" (see perlobj(1)). For example:

  use Sys::Guestfs;
  if (defined (Sys::Guestfs->can ("set_verbose"))) {
    print "\$h->set_verbose is available\n";
  }
 
 

To test if particular features are supported by the current build, use the ``available'' method like the example below. Note that the appliance must be launched first.

  $h->available ( ["augeas"] );
 
 

Since the ``available'' method croaks if the feature is not supported, you might also want to wrap this in an eval and return a boolean. In fact this has already been done for you: use ``feature_available'' in Sys::Guestfs::Lib(3).

For further discussion on this topic, refer to ``AVAILABILITY'' in guestfs(3).

STORING DATA IN THE HANDLE

The handle returned from ``new'' is a hash reference. The hash normally contains a single element:

  {
    _g => [private data used by libguestfs]
  }
 
 

Callers can add other elements to this hash to store data for their own purposes. The data lasts for the lifetime of the handle.

Any fields whose names begin with an underscore are reserved for private use by libguestfs. We may add more in future.

It is recommended that callers prefix the name of their field(s) with some unique string, to avoid conflicts with other users.

COPYRIGHT

Copyright (C) 2009-2010 Red Hat Inc.

LICENSE

Please see the file COPYING.LIB for the full license.

SEE ALSO

guestfs(3), guestfish(1), <http://libguestfs.org>, Sys::Guestfs::Lib(3).