libvirt/docs/formatdomain.html.in
Andrea Bolognani 2f06757048 docs: Expand documentation for the tickpolicy timer attribute
The current documentation is fairly terse and not easy to decode
for someone who's not intimately familiar with the inner workings
of timer devices. Expand on it by providing a somewhat verbose
description of what behavior each policy will result in, as seen
from both the guest OS and host point of view.

This is lifted directly from QEMU commit

  commit 2a7d957596786404c4ed16b089273de95a9580ad
  Author: Andrea Bolognani <abologna@redhat.com>
  Date:   Tue Feb 11 19:37:44 2020 +0100

    qapi: Expand documentation for LostTickPolicy

  v4.2.0-1442-g2a7d957596

The original text also matched word for word the documentation
found in QEMU.

Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
2020-02-21 11:42:47 +01:00

9385 lines
394 KiB
XML

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml">
<body>
<h1>Domain XML format</h1>
<ul id="toc"></ul>
<p>
This section describes the XML format used to represent domains, there are
variations on the format based on the kind of domains run and the options
used to launch them. For hypervisor specific details consult the
<a href="drivers.html">driver docs</a>
</p>
<h2><a id="elements">Element and attribute overview</a></h2>
<p>
The root element required for all virtual machines is
named <code>domain</code>. It has two attributes, the
<a id="attributeDomainType"><code>type</code></a>
specifies the hypervisor used for running
the domain. The allowed values are driver specific, but
include "xen", "kvm", "qemu" and "lxc". The
second attribute is <code>id</code> which is a unique
integer identifier for the running guest machine. Inactive
machines have no id value.
</p>
<h3><a id="elementsMetadata">General metadata</a></h3>
<pre>
&lt;domain type='kvm' id='1'&gt;
&lt;name&gt;MyGuest&lt;/name&gt;
&lt;uuid&gt;4dea22b3-1d52-d8f3-2516-782e98ab3fa0&lt;/uuid&gt;
&lt;genid&gt;43dc0cf8-809b-4adb-9bea-a9abb5f3d90e&lt;/genid&gt;
&lt;title&gt;A short description - title - of the domain&lt;/title&gt;
&lt;description&gt;Some human readable description&lt;/description&gt;
&lt;metadata&gt;
&lt;app1:foo xmlns:app1="http://app1.org/app1/"&gt;..&lt;/app1:foo&gt;
&lt;app2:bar xmlns:app2="http://app1.org/app2/"&gt;..&lt;/app2:bar&gt;
&lt;/metadata&gt;
...</pre>
<dl>
<dt><code>name</code></dt>
<dd>The content of the <code>name</code> element provides
a short name for the virtual machine. This name should
consist only of alpha-numeric characters and is required
to be unique within the scope of a single host. It is
often used to form the filename for storing the persistent
configuration file. <span class="since">Since 0.0.1</span></dd>
<dt><code>uuid</code></dt>
<dd>The content of the <code>uuid</code> element provides
a globally unique identifier for the virtual machine.
The format must be RFC 4122 compliant,
eg <code>3e3fce45-4f53-4fa7-bb32-11f34168b82b</code>.
If omitted when defining/creating a new machine, a random
UUID is generated. It is also possible to provide the UUID
via a <a href="#elementsSysinfo"><code>sysinfo</code></a>
specification. <span class="since">Since 0.0.1, sysinfo
since 0.8.7</span></dd>
<dt><code>genid</code></dt>
<dd><span class="since">Since 4.4.0</span>, the <code>genid</code>
element can be used to add a Virtual Machine Generation ID which
exposes a 128-bit, cryptographically random, integer value identifier,
referred to as a Globally Unique Identifier (GUID) using the same
format as the <code>uuid</code>. The value is used to help notify
the guest operating system when the virtual machine is re-executing
something that has already executed before, such as:
<ul>
<li>VM starts executing a snapshot</li>
<li>VM is recovered from backup</li>
<li>VM is failover in a disaster recovery environment</li>
<li>VM is imported, copied, or cloned</li>
</ul>
The guest operating system notices the change and is then able to
react as appropriate by marking its copies of distributed databases
as dirty, re-initializing its random number generator, etc.
<p>
The libvirt XML parser will accept both a provided GUID value
or just &lt;genid/&gt; in which case a GUID will be generated
and saved in the XML. For the transitions such as above, libvirt
will change the GUID before re-executing.</p></dd>
<dt><code>title</code></dt>
<dd>The optional element <code>title</code> provides space for a
short description of the domain. The title should not contain
any newlines. <span class="since">Since 0.9.10</span>.</dd>
<dt><code>description</code></dt>
<dd>The content of the <code>description</code> element provides a
human readable description of the virtual machine. This data is not
used by libvirt in any way, it can contain any information the user
wants. <span class="since">Since 0.7.2</span></dd>
<dt><code>metadata</code></dt>
<dd>The <code>metadata</code> node can be used by applications
to store custom metadata in the form of XML
nodes/trees. Applications must use custom namespaces on their
XML nodes/trees, with only one top-level element per namespace
(if the application needs structure, they should have
sub-elements to their namespace
element). <span class="since">Since 0.9.10</span></dd>
</dl>
<h3><a id="elementsOS">Operating system booting</a></h3>
<p>
There are a number of different ways to boot virtual machines
each with their own pros and cons.
</p>
<h4><a id="elementsOSBIOS">BIOS bootloader</a></h4>
<p>
Booting via the BIOS is available for hypervisors supporting
full virtualization. In this case the BIOS has a boot order
priority (floppy, harddisk, cdrom, network) determining where
to obtain/find the boot image.
</p>
<pre>
...
&lt;os firmware='efi'&gt;
&lt;type&gt;hvm&lt;/type&gt;
&lt;loader readonly='yes' secure='no' type='rom'&gt;/usr/lib/xen/boot/hvmloader&lt;/loader&gt;
&lt;nvram template='/usr/share/OVMF/OVMF_VARS.fd'&gt;/var/lib/libvirt/nvram/guest_VARS.fd&lt;/nvram&gt;
&lt;boot dev='hd'/&gt;
&lt;boot dev='cdrom'/&gt;
&lt;bootmenu enable='yes' timeout='3000'/&gt;
&lt;smbios mode='sysinfo'/&gt;
&lt;bios useserial='yes' rebootTimeout='0'/&gt;
&lt;/os&gt;
...</pre>
<dl>
<dt><code>firmware</code></dt>
<dd>The <code>firmware</code> attribute allows management
applications to automatically fill <code>&lt;loader/&gt;</code>
and <code>&lt;nvram/&gt;</code> elements and possibly enable
some features required by selected firmware. Accepted values are
<code>bios</code> and <code>efi</code>.<br/>
The selection process scans for files describing installed
firmware images in specified location and uses the most specific
one which fulfils domain requirements. The locations in order of
preference (from generic to most specific one) are:
<ul>
<li><code>/usr/share/qemu/firmware</code></li>
<li><code>/etc/qemu/firmware</code></li>
<li><code>$XDG_CONFIG_HOME/qemu/firmware</code></li>
</ul>
For more information refer to firmware metadata specification as
described in <code>docs/interop/firmware.json</code> in QEMU
repository. Regular users do not need to bother.
<span class="since">Since 5.2.0 (QEMU and KVM only)</span><br/>
For VMware guests, this is set to <code>efi</code> when the guest
uses UEFI, and it is not set when using BIOS.
<span class="since">Since 5.3.0 (VMware ESX and Workstation/Player)</span>
</dd>
<dt><code>type</code></dt>
<dd>The content of the <code>type</code> element specifies the
type of operating system to be booted in the virtual machine.
<code>hvm</code> indicates that the OS is one designed to run
on bare metal, so requires full virtualization. <code>linux</code>
(badly named!) refers to an OS that supports the Xen 3 hypervisor
guest ABI. There are also two optional attributes, <code>arch</code>
specifying the CPU architecture to virtualization,
and <a id="attributeOSTypeMachine"><code>machine</code></a> referring
to the machine type. The <a href="formatcaps.html">Capabilities XML</a>
provides details on allowed values for
these. If <code>arch</code> is omitted then for most hypervisor
drivers, the host native arch will be chosen. For the <code>test</code>,
<code>ESX</code> and <code>VMWare</code> hypervisor drivers, however,
the <code>i686</code> arch will always be chosen even on an
<code>x86_64</code> host. <span class="since">Since 0.0.1</span></dd>
<dt><a id="elementLoader"><code>loader</code></a></dt>
<dd>The optional <code>loader</code> tag refers to a firmware blob,
which is specified by absolute path,
used to assist the domain creation process. It is used by Xen
fully virtualized domains as well as setting the QEMU BIOS file
path for QEMU/KVM domains. <span class="since">Xen since 0.1.0,
QEMU/KVM since 0.9.12</span> Then, <span class="since">since
1.2.8</span> it's possible for the element to have two
optional attributes: <code>readonly</code> (accepted values are
<code>yes</code> and <code>no</code>) to reflect the fact that the
image should be writable or read-only. The second attribute
<code>type</code> accepts values <code>rom</code> and
<code>pflash</code>. It tells the hypervisor where in the guest
memory the file should be mapped. For instance, if the loader
path points to an UEFI image, <code>type</code> should be
<code>pflash</code>. Moreover, some firmwares may
implement the Secure boot feature. Attribute
<code>secure</code> can be used then to control it.
<span class="since">Since 2.1.0</span></dd>
<dt><code>nvram</code></dt>
<dd>Some UEFI firmwares may want to use a non-volatile memory to store
some variables. In the host, this is represented as a file and the
absolute path to the file is stored in this element. Moreover, when the
domain is started up libvirt copies so called master NVRAM store file
defined in <code>qemu.conf</code>. If needed, the <code>template</code>
attribute can be used to per domain override map of master NVRAM stores
from the config file. Note, that for transient domains if the NVRAM file
has been created by libvirt it is left behind and it is management
application's responsibility to save and remove file (if needed to be
persistent). <span class="since">Since 1.2.8</span></dd>
<dt><code>boot</code></dt>
<dd>The <code>dev</code> attribute takes one of the values "fd", "hd",
"cdrom" or "network" and is used to specify the next boot device
to consider. The <code>boot</code> element can be repeated multiple
times to setup a priority list of boot devices to try in turn.
Multiple devices of the same type are sorted according to their
targets while preserving the order of buses. After defining the
domain, its XML configuration returned by libvirt (through
virDomainGetXMLDesc) lists devices in the sorted order. Once sorted,
the first device is marked as bootable. Thus, e.g., a domain
configured to boot from "hd" with vdb, hda, vda, and hdc disks
assigned to it will boot from vda (the sorted list is vda, vdb, hda,
hdc). Similar domain with hdc, vda, vdb, and hda disks will boot from
hda (sorted disks are: hda, hdc, vda, vdb). It can be tricky to
configure in the desired way, which is why per-device boot elements
(see <a href="#elementsDisks">disks</a>,
<a href="#elementsNICS">network interfaces</a>, and
<a href="#elementsHostDev">USB and PCI devices</a> sections below) were
introduced and they are the preferred way providing full control over
booting order. The <code>boot</code> element and per-device boot
elements are mutually exclusive. <span class="since">Since 0.1.3,
per-device boot since 0.8.8</span>
</dd>
<dt><code>smbios</code></dt>
<dd>How to populate SMBIOS information visible in the guest.
The <code>mode</code> attribute must be specified, and is either
"emulate" (let the hypervisor generate all values), "host" (copy
all of Block 0 and Block 1, except for the UUID, from the host's
SMBIOS values;
the <a href="html/libvirt-libvirt-host.html#virConnectGetSysinfo">
<code>virConnectGetSysinfo</code></a> call can be
used to see what values are copied), or "sysinfo" (use the values in
the <a href="#elementsSysinfo">sysinfo</a> element). If not
specified, the hypervisor default is used. <span class="since">
Since 0.8.7</span>
</dd>
</dl>
<p>Up till here the BIOS/UEFI configuration knobs are generic enough to
be implemented by majority (if not all) firmwares out there. However,
from now on not every single setting makes sense to all firmwares. For
instance, <code>rebootTimeout</code> doesn't make sense for UEFI,
<code>useserial</code> might not be usable with a BIOS firmware that
doesn't produce any output onto serial line, etc. Moreover, firmwares
don't usually export their capabilities for libvirt (or users) to check.
And the set of their capabilities can change with every new release.
Hence users are advised to try the settings they use before relying on
them in production.</p>
<dl>
<dt><code>bootmenu</code></dt>
<dd> Whether or not to enable an interactive boot menu prompt on guest
startup. The <code>enable</code> attribute can be either "yes" or "no".
If not specified, the hypervisor default is used. <span class="since">
Since 0.8.3</span>
Additional attribute <code>timeout</code> takes the number of milliseconds
the boot menu should wait until it times out. Allowed values are numbers
in range [0, 65535] inclusive and it is ignored unless <code>enable</code>
is set to "yes". <span class="since">Since 1.2.8</span>
</dd>
<dt><code>bios</code></dt>
<dd>This element has attribute <code>useserial</code> with possible
values <code>yes</code> or <code>no</code>. It enables or disables
Serial Graphics Adapter which allows users to see BIOS messages
on a serial port. Therefore, one needs to have
<a href="#elementCharSerial">serial port</a> defined.
<span class="since">Since 0.9.4</span>.
<span class="since">Since 0.10.2 (QEMU only)</span> there is
another attribute, <code>rebootTimeout</code> that controls
whether and after how long the guest should start booting
again in case the boot fails (according to BIOS). The value is
in milliseconds with maximum of <code>65535</code> and special
value <code>-1</code> disables the reboot.
</dd>
</dl>
<h4><a id="elementsOSBootloader">Host bootloader</a></h4>
<p>
Hypervisors employing paravirtualization do not usually emulate
a BIOS, and instead the host is responsible to kicking off the
operating system boot. This may use a pseudo-bootloader in the
host to provide an interface to choose a kernel for the guest.
An example is <code>pygrub</code> with Xen. The Bhyve hypervisor
also uses a host bootloader, either <code>bhyveload</code> or
<code>grub-bhyve</code>.
</p>
<pre>
...
&lt;bootloader&gt;/usr/bin/pygrub&lt;/bootloader&gt;
&lt;bootloader_args&gt;--append single&lt;/bootloader_args&gt;
...</pre>
<dl>
<dt><code>bootloader</code></dt>
<dd>The content of the <code>bootloader</code> element provides
a fully qualified path to the bootloader executable in the
host OS. This bootloader will be run to choose which kernel
to boot. The required output of the bootloader is dependent
on the hypervisor in use. <span class="since">Since 0.1.0</span></dd>
<dt><code>bootloader_args</code></dt>
<dd>The optional <code>bootloader_args</code> element allows
command line arguments to be passed to the bootloader.
<span class="since">Since 0.2.3</span>
</dd>
</dl>
<h4><a id="elementsOSKernel">Direct kernel boot</a></h4>
<p>
When installing a new guest OS it is often useful to boot directly
from a kernel and initrd stored in the host OS, allowing command
line arguments to be passed directly to the installer. This capability
is usually available for both para and full virtualized guests.
</p>
<pre>
...
&lt;os&gt;
&lt;type&gt;hvm&lt;/type&gt;
&lt;loader&gt;/usr/lib/xen/boot/hvmloader&lt;/loader&gt;
&lt;kernel&gt;/root/f8-i386-vmlinuz&lt;/kernel&gt;
&lt;initrd&gt;/root/f8-i386-initrd&lt;/initrd&gt;
&lt;cmdline&gt;console=ttyS0 ks=http://example.com/f8-i386/os/&lt;/cmdline&gt;
&lt;dtb&gt;/root/ppc.dtb&lt;/dtb&gt;
&lt;acpi&gt;
&lt;table type='slic'&gt;/path/to/slic.dat&lt;/table&gt;
&lt;/acpi&gt;
&lt;/os&gt;
...</pre>
<dl>
<dt><code>type</code></dt>
<dd>This element has the same semantics as described earlier in the
<a href="#elementsOSBIOS">BIOS boot section</a></dd>
<dt><code>loader</code></dt>
<dd>This element has the same semantics as described earlier in the
<a href="#elementsOSBIOS">BIOS boot section</a></dd>
<dt><code>kernel</code></dt>
<dd>The contents of this element specify the fully-qualified path
to the kernel image in the host OS.</dd>
<dt><code>initrd</code></dt>
<dd>The contents of this element specify the fully-qualified path
to the (optional) ramdisk image in the host OS.</dd>
<dt><code>cmdline</code></dt>
<dd>The contents of this element specify arguments to be passed to
the kernel (or installer) at boot time. This is often used to
specify an alternate primary console (eg serial port), or the
installation media source / kickstart file</dd>
<dt><code>dtb</code></dt>
<dd>The contents of this element specify the fully-qualified path
to the (optional) device tree binary (dtb) image in the host OS.
<span class="since">Since 1.0.4</span></dd>
<dt><code>acpi</code></dt>
<dd>The <code>table</code> element contains a fully-qualified path
to the ACPI table. The <code>type</code> attribute contains the
ACPI table type (currently only <code>slic</code> is supported)
<span class="since">Since 1.3.5 (QEMU)</span>
<span class="since">Since 5.9.0 (Xen)</span></dd>
</dl>
<h4><a id="elementsOSContainer">Container boot</a></h4>
<p>
When booting a domain using container based virtualization, instead
of a kernel / boot image, a path to the init binary is required, using
the <code>init</code> element. By default this will be launched with
no arguments. To specify the initial argv, use the <code>initarg</code>
element, repeated as many time as is required. The <code>cmdline</code>
element, if set will be used to provide an equivalent to <code>/proc/cmdline</code>
but will not affect init argv.
</p>
<p>
To set environment variables, use the <code>initenv</code> element, one
for each variable.
</p>
<p>
To set a custom work directory for the init, use the <code>initdir</code>
element.
</p>
<p>
To run the init command as a given user or group, use the <code>inituser</code>
or <code>initgroup</code> elements respectively. Both elements can be provided
either a user (resp. group) id or a name. Prefixing the user or group id with
a <code>+</code> will force it to be considered like a numeric value. Without
this, it will be first tried as a user or group name.
</p>
<pre>
&lt;os&gt;
&lt;type arch='x86_64'&gt;exe&lt;/type&gt;
&lt;init&gt;/bin/systemd&lt;/init&gt;
&lt;initarg&gt;--unit&lt;/initarg&gt;
&lt;initarg&gt;emergency.service&lt;/initarg&gt;
&lt;initenv name='MYENV'&gt;some value&lt;/initenv&gt;
&lt;initdir&gt;/my/custom/cwd&lt;/initdir&gt;
&lt;inituser&gt;tester&lt;/inituser&gt;
&lt;initgroup&gt;1000&lt;/initgroup&gt;
&lt;/os&gt;
</pre>
<p>
If you want to enable user namespace, set the <code>idmap</code> element.
The <code>uid</code> and <code>gid</code> elements have three attributes:
</p>
<dl>
<dt><code>start</code></dt>
<dd>First user ID in container. It must be '0'.</dd>
<dt><code>target</code></dt>
<dd>The first user ID in container will be mapped to this target user
ID in host.</dd>
<dt><code>count</code></dt>
<dd>How many users in container are allowed to map to host's user.</dd>
</dl>
<pre>
&lt;idmap&gt;
&lt;uid start='0' target='1000' count='10'/&gt;
&lt;gid start='0' target='1000' count='10'/&gt;
&lt;/idmap&gt;
</pre>
<h3><a id="elementsSysinfo">SMBIOS System Information</a></h3>
<p>
Some hypervisors allow control over what system information is
presented to the guest (for example, SMBIOS fields can be
populated by a hypervisor and inspected via
the <code>dmidecode</code> command in the guest). The
optional <code>sysinfo</code> element covers all such categories
of information. <span class="since">Since 0.8.7</span>
</p>
<pre>
...
&lt;os&gt;
&lt;smbios mode='sysinfo'/&gt;
...
&lt;/os&gt;
&lt;sysinfo type='smbios'&gt;
&lt;bios&gt;
&lt;entry name='vendor'&gt;LENOVO&lt;/entry&gt;
&lt;/bios&gt;
&lt;system&gt;
&lt;entry name='manufacturer'&gt;Fedora&lt;/entry&gt;
&lt;entry name='product'&gt;Virt-Manager&lt;/entry&gt;
&lt;entry name='version'&gt;0.9.4&lt;/entry&gt;
&lt;/system&gt;
&lt;baseBoard&gt;
&lt;entry name='manufacturer'&gt;LENOVO&lt;/entry&gt;
&lt;entry name='product'&gt;20BE0061MC&lt;/entry&gt;
&lt;entry name='version'&gt;0B98401 Pro&lt;/entry&gt;
&lt;entry name='serial'&gt;W1KS427111E&lt;/entry&gt;
&lt;/baseBoard&gt;
&lt;chassis&gt;
&lt;entry name='manufacturer'&gt;Dell Inc.&lt;/entry&gt;
&lt;entry name='version'&gt;2.12&lt;/entry&gt;
&lt;entry name='serial'&gt;65X0XF2&lt;/entry&gt;
&lt;entry name='asset'&gt;40000101&lt;/entry&gt;
&lt;entry name='sku'&gt;Type3Sku1&lt;/entry&gt;
&lt;/chassis&gt;
&lt;oemStrings&gt;
&lt;entry&gt;myappname:some arbitrary data&lt;/entry&gt;
&lt;entry&gt;otherappname:more arbitrary data&lt;/entry&gt;
&lt;/oemStrings&gt;
&lt;/sysinfo&gt;
...</pre>
<p>
The <code>sysinfo</code> element has a mandatory
attribute <code>type</code> that determine the layout of
sub-elements, with supported values of:
</p>
<dl>
<dt><code>smbios</code></dt>
<dd>Sub-elements call out specific SMBIOS values, which will
affect the guest if used in conjunction with
the <code>smbios</code> sub-element of
the <a href="#elementsOS"><code>os</code></a> element. Each
sub-element of <code>sysinfo</code> names a SMBIOS block, and
within those elements can be a list of <code>entry</code>
elements that describe a field within the block. The following
blocks and entries are recognized:
<dl>
<dt><code>bios</code></dt>
<dd>
This is block 0 of SMBIOS, with entry names drawn from:
<dl>
<dt><code>vendor</code></dt>
<dd>BIOS Vendor's Name</dd>
<dt><code>version</code></dt>
<dd>BIOS Version</dd>
<dt><code>date</code></dt>
<dd>BIOS release date. If supplied, is in either mm/dd/yy or
mm/dd/yyyy format. If the year portion of the string is
two digits, the year is assumed to be 19yy.</dd>
<dt><code>release</code></dt>
<dd>System BIOS Major and Minor release number values
concatenated together as one string separated by
a period, for example, 10.22.</dd>
</dl>
</dd>
<dt><code>system</code></dt>
<dd>
This is block 1 of SMBIOS, with entry names drawn from:
<dl>
<dt><code>manufacturer</code></dt>
<dd>Manufacturer of BIOS</dd>
<dt><code>product</code></dt>
<dd>Product Name</dd>
<dt><code>version</code></dt>
<dd>Version of the product</dd>
<dt><code>serial</code></dt>
<dd>Serial number</dd>
<dt><code>uuid</code></dt>
<dd>Universal Unique ID number. If this entry is provided
alongside a top-level
<a href="#elementsMetadata"><code>uuid</code></a> element,
then the two values must match.</dd>
<dt><code>sku</code></dt>
<dd>SKU number to identify a particular configuration.</dd>
<dt><code>family</code></dt>
<dd>Identify the family a particular computer belongs to.</dd>
</dl>
</dd>
<dt><code>baseBoard</code></dt>
<dd>
This is block 2 of SMBIOS. This element can be repeated multiple
times to describe all the base boards; however, not all
hypervisors necessarily support the repetition. The element can
have the following children:
<dl>
<dt><code>manufacturer</code></dt>
<dd>Manufacturer of BIOS</dd>
<dt><code>product</code></dt>
<dd>Product Name</dd>
<dt><code>version</code></dt>
<dd>Version of the product</dd>
<dt><code>serial</code></dt>
<dd>Serial number</dd>
<dt><code>asset</code></dt>
<dd>Asset tag</dd>
<dt><code>location</code></dt>
<dd>Location in chassis</dd>
</dl>
NB: Incorrectly supplied entries for the
<code>bios</code>, <code>system</code> or <code>baseBoard</code>
blocks will be ignored without error. Other than <code>uuid</code>
validation and <code>date</code> format checking, all values are
passed as strings to the hypervisor driver.
</dd>
<dt><code>chassis</code></dt>
<dd>
<span class="since">Since 4.1.0,</span> this is block 3 of
SMBIOS, with entry names drawn from:
<dl>
<dt><code>manufacturer</code></dt>
<dd>Manufacturer of Chassis</dd>
<dt><code>version</code></dt>
<dd>Version of the Chassis</dd>
<dt><code>serial</code></dt>
<dd>Serial number</dd>
<dt><code>asset</code></dt>
<dd>Asset tag</dd>
<dt><code>sku</code></dt>
<dd>SKU number</dd>
</dl>
</dd>
<dt><code>oemStrings</code></dt>
<dd>
This is block 11 of SMBIOS. This element should appear once and
can have multiple <code>entry</code> child elements, each providing
arbitrary string data. There are no restrictions on what data can
be provided in the entries, however, if the data is intended to be
consumed by an application in the guest, it is recommended to use
the application name as a prefix in the string. (<span class="since">Since 4.1.0</span>)
</dd>
</dl>
</dd>
</dl>
<h3><a id="elementsCPUAllocation">CPU Allocation</a></h3>
<pre>
&lt;domain&gt;
...
&lt;vcpu placement='static' cpuset="1-4,^3,6" current="1"&gt;2&lt;/vcpu&gt;
&lt;vcpus&gt;
&lt;vcpu id='0' enabled='yes' hotpluggable='no' order='1'/&gt;
&lt;vcpu id='1' enabled='no' hotpluggable='yes'/&gt;
&lt;/vcpus&gt;
...
&lt;/domain&gt;
</pre>
<dl>
<dt><code>vcpu</code></dt>
<dd>The content of this element defines the maximum number of virtual
CPUs allocated for the guest OS, which must be between 1 and
the maximum supported by the hypervisor.
<dl>
<dt><code>cpuset</code></dt>
<dd>
The optional attribute <code>cpuset</code> is a comma-separated
list of physical CPU numbers that domain process and virtual CPUs
can be pinned to by default. (NB: The pinning policy of domain
process and virtual CPUs can be specified separately by
<code>cputune</code>. If the attribute <code>emulatorpin</code>
of <code>cputune</code> is specified, the <code>cpuset</code>
specified by <code>vcpu</code> here will be ignored. Similarly,
for virtual CPUs which have the <code>vcpupin</code> specified,
the <code>cpuset</code> specified by <code>cpuset</code> here
will be ignored. For virtual CPUs which don't have
<code>vcpupin</code> specified, each will be pinned to the physical
CPUs specified by <code>cpuset</code> here).
Each element in that list is either a single CPU number,
a range of CPU numbers, or a caret followed by a CPU number to
be excluded from a previous range.
<span class="since">Since 0.4.4</span>
</dd>
<dt><code>current</code></dt>
<dd>
The optional attribute <code>current</code> can
be used to specify whether fewer than the maximum number of
virtual CPUs should be enabled.
<span class="since">Since 0.8.5</span>
</dd>
<dt><code>placement</code></dt>
<dd>
The optional attribute <code>placement</code> can be used to
indicate the CPU placement mode for domain process. The value can
be either "static" or "auto", but defaults to <code>placement</code>
of <code>numatune</code> or "static" if <code>cpuset</code> is
specified. Using "auto" indicates the domain process will be pinned
to the advisory nodeset from querying numad and the value of
attribute <code>cpuset</code> will be ignored if it's specified.
If both <code>cpuset</code> and <code>placement</code> are not
specified or if <code>placement</code> is "static", but no
<code>cpuset</code> is specified, the domain process will be
pinned to all the available physical CPUs.
<span class="since">Since 0.9.11 (QEMU and KVM only)</span>
</dd>
</dl>
</dd>
<dt><code>vcpus</code></dt>
<dd>
The vcpus element allows to control state of individual vCPUs.
The <code>id</code> attribute specifies the vCPU id as used by libvirt
in other places such as vCPU pinning, scheduler information and NUMA
assignment. Note that the vCPU ID as seen in the guest may differ from
libvirt ID in certain cases. Valid IDs are from 0 to the maximum vCPU
count as set by the <code>vcpu</code> element minus 1.
The <code>enabled</code> attribute allows to control the state of the
vCPU. Valid values are <code>yes</code> and <code>no</code>.
<code>hotpluggable</code> controls whether given vCPU can be hotplugged
and hotunplugged in cases when the CPU is enabled at boot. Note that
all disabled vCPUs must be hotpluggable. Valid values are
<code>yes</code> and <code>no</code>.
<code>order</code> allows to specify the order to add the online vCPUs.
For hypervisors/platforms that require to insert multiple vCPUs at once
the order may be duplicated across all vCPUs that need to be
enabled at once. Specifying order is not necessary, vCPUs are then
added in an arbitrary order. If order info is used, it must be used for
all online vCPUs. Hypervisors may clear or update ordering information
during certain operations to assure valid configuration.
Note that hypervisors may create hotpluggable vCPUs differently from
boot vCPUs thus special initialization may be necessary.
Hypervisors may require that vCPUs enabled on boot which are not
hotpluggable are clustered at the beginning starting with ID 0. It may
be also required that vCPU 0 is always present and non-hotpluggable.
Note that providing state for individual CPUs may be necessary to enable
support of addressable vCPU hotplug and this feature may not be
supported by all hypervisors.
For QEMU the following conditions are required. vCPU 0 needs to be
enabled and non-hotpluggable. On PPC64 along with it vCPUs that are in
the same core need to be enabled as well. All non-hotpluggable CPUs
present at boot need to be grouped after vCPU 0.
<span class="since">Since 2.2.0 (QEMU only)</span>
</dd>
</dl>
<h3><a id="elementsIOThreadsAllocation">IOThreads Allocation</a></h3>
<p>
IOThreads are dedicated event loop threads for supported disk
devices to perform block I/O requests in order to improve
scalability especially on an SMP host/guest with many LUNs.
<span class="since">Since 1.2.8 (QEMU only)</span>
</p>
<pre>
&lt;domain&gt;
...
&lt;iothreads&gt;4&lt;/iothreads&gt;
...
&lt;/domain&gt;
</pre>
<pre>
&lt;domain&gt;
...
&lt;iothreadids&gt;
&lt;iothread id="2"/&gt;
&lt;iothread id="4"/&gt;
&lt;iothread id="6"/&gt;
&lt;iothread id="8"/&gt;
&lt;/iothreadids&gt;
...
&lt;/domain&gt;
</pre>
<dl>
<dt><code>iothreads</code></dt>
<dd>
The content of this optional element defines the number
of IOThreads to be assigned to the domain for use by
supported target storage devices. There
should be only 1 or 2 IOThreads per host CPU. There may be more
than one supported device assigned to each IOThread.
<span class="since">Since 1.2.8</span>
</dd>
<dt><code>iothreadids</code></dt>
<dd>
The optional <code>iothreadids</code> element provides the capability
to specifically define the IOThread ID's for the domain. By default,
IOThread ID's are sequentially numbered starting from 1 through the
number of <code>iothreads</code> defined for the domain. The
<code>id</code> attribute is used to define the IOThread ID. The
<code>id</code> attribute must be a positive integer greater than 0.
If there are less <code>iothreadids</code> defined than
<code>iothreads</code> defined for the domain, then libvirt will
sequentially fill <code>iothreadids</code> starting at 1 avoiding
any predefined <code>id</code>. If there are more
<code>iothreadids</code> defined than <code>iothreads</code>
defined for the domain, then the <code>iothreads</code> value
will be adjusted accordingly.
<span class="since">Since 1.2.15</span>
</dd>
</dl>
<h3><a id="elementsCPUTuning">CPU Tuning</a></h3>
<pre>
&lt;domain&gt;
...
&lt;cputune&gt;
&lt;vcpupin vcpu="0" cpuset="1-4,^2"/&gt;
&lt;vcpupin vcpu="1" cpuset="0,1"/&gt;
&lt;vcpupin vcpu="2" cpuset="2,3"/&gt;
&lt;vcpupin vcpu="3" cpuset="0,4"/&gt;
&lt;emulatorpin cpuset="1-3"/&gt;
&lt;iothreadpin iothread="1" cpuset="5,6"/&gt;
&lt;iothreadpin iothread="2" cpuset="7,8"/&gt;
&lt;shares&gt;2048&lt;/shares&gt;
&lt;period&gt;1000000&lt;/period&gt;
&lt;quota&gt;-1&lt;/quota&gt;
&lt;global_period&gt;1000000&lt;/global_period&gt;
&lt;global_quota&gt;-1&lt;/global_quota&gt;
&lt;emulator_period&gt;1000000&lt;/emulator_period&gt;
&lt;emulator_quota&gt;-1&lt;/emulator_quota&gt;
&lt;iothread_period&gt;1000000&lt;/iothread_period&gt;
&lt;iothread_quota&gt;-1&lt;/iothread_quota&gt;
&lt;vcpusched vcpus='0-4,^3' scheduler='fifo' priority='1'/&gt;
&lt;iothreadsched iothreads='2' scheduler='batch'/&gt;
&lt;cachetune vcpus='0-3'&gt;
&lt;cache id='0' level='3' type='both' size='3' unit='MiB'/&gt;
&lt;cache id='1' level='3' type='both' size='3' unit='MiB'/&gt;
&lt;monitor level='3' vcpus='1'/&gt;
&lt;monitor level='3' vcpus='0-3'/&gt;
&lt;/cachetune&gt;
&lt;cachetune vcpus='4-5'&gt;
&lt;monitor level='3' vcpus='4'/&gt;
&lt;monitor level='3' vcpus='5'/&gt;
&lt;/cachetune&gt;
&lt;memorytune vcpus='0-3'&gt;
&lt;node id='0' bandwidth='60'/&gt;
&lt;/memorytune&gt;
&lt;/cputune&gt;
...
&lt;/domain&gt;
</pre>
<dl>
<dt><code>cputune</code></dt>
<dd>
The optional <code>cputune</code> element provides details
regarding the CPU tunable parameters for the domain.
<span class="since">Since 0.9.0</span>
</dd>
<dt><code>vcpupin</code></dt>
<dd>
The optional <code>vcpupin</code> element specifies which of host's
physical CPUs the domain vCPU will be pinned to. If this is omitted,
and attribute <code>cpuset</code> of element <code>vcpu</code> is
not specified, the vCPU is pinned to all the physical CPUs by default.
It contains two required attributes, the attribute <code>vcpu</code>
specifies vCPU id, and the attribute <code>cpuset</code> is same as
attribute <code>cpuset</code> of element <code>vcpu</code>.
(NB: Only qemu driver support)
<span class="since">Since 0.9.0</span>
</dd>
<dt><code>emulatorpin</code></dt>
<dd>
The optional <code>emulatorpin</code> element specifies which of host
physical CPUs the "emulator", a subset of a domain not including vCPU
or iothreads will be pinned to. If this is omitted, and attribute
<code>cpuset</code> of element <code>vcpu</code> is not specified,
"emulator" is pinned to all the physical CPUs by default. It contains
one required attribute <code>cpuset</code> specifying which physical
CPUs to pin to.
</dd>
<dt><code>iothreadpin</code></dt>
<dd>
The optional <code>iothreadpin</code> element specifies which of host
physical CPUs the IOThreads will be pinned to. If this is omitted
and attribute <code>cpuset</code> of element <code>vcpu</code> is
not specified, the IOThreads are pinned to all the physical CPUs
by default. There are two required attributes, the attribute
<code>iothread</code> specifies the IOThread ID and the attribute
<code>cpuset</code> specifying which physical CPUs to pin to. See
the <code>iothreadids</code>
<a href="#elementsIOThreadsAllocation"><code>description</code></a>
for valid <code>iothread</code> values.
<span class="since">Since 1.2.9</span>
</dd>
<dt><code>shares</code></dt>
<dd>
The optional <code>shares</code> element specifies the proportional
weighted share for the domain. If this is omitted, it defaults to
the OS provided defaults. NB, There is no unit for the value,
it's a relative measure based on the setting of other VM,
e.g. A VM configured with value
2048 will get twice as much CPU time as a VM configured with value 1024.
<span class="since">Since 0.9.0</span>
</dd>
<dt><code>period</code></dt>
<dd>
The optional <code>period</code> element specifies the enforcement
interval (unit: microseconds). Within <code>period</code>, each vCPU of
the domain will not be allowed to consume more than <code>quota</code>
worth of runtime. The value should be in range [1000, 1000000]. A period
with value 0 means no value.
<span class="since">Only QEMU driver support since 0.9.4, LXC since
0.9.10</span>
</dd>
<dt><code>quota</code></dt>
<dd>
The optional <code>quota</code> element specifies the maximum allowed
bandwidth (unit: microseconds). A domain with <code>quota</code> as any
negative value indicates that the domain has infinite bandwidth for
vCPU threads, which means that it is not bandwidth controlled. The value
should be in range [1000, 18446744073709551] or less than 0. A quota
with value 0 means no value. You can use this feature to ensure that all
vCPUs run at the same speed.
<span class="since">Only QEMU driver support since 0.9.4, LXC since
0.9.10</span>
</dd>
<dt><code>global_period</code></dt>
<dd>
The optional <code>global_period</code> element specifies the
enforcement CFS scheduler interval (unit: microseconds) for the whole
domain in contrast with <code>period</code> which enforces the interval
per vCPU. The value should be in range 1000, 1000000]. A
<code>global_period</code> with value 0 means no value.
<span class="since">Only QEMU driver support since 1.3.3</span>
</dd>
<dt><code>global_quota</code></dt>
<dd>
The optional <code>global_quota</code> element specifies the maximum
allowed bandwidth (unit: microseconds) within a period for the whole
domain. A domain with <code>global_quota</code> as any negative
value indicates that the domain has infinite bandwidth, which means that
it is not bandwidth controlled. The value should be in range
[1000, 18446744073709551] or less than 0. A <code>global_quota</code>
with value 0 means no value.
<span class="since">Only QEMU driver support since 1.3.3</span>
</dd>
<dt><code>emulator_period</code></dt>
<dd>
The optional <code>emulator_period</code> element specifies the enforcement
interval (unit: microseconds). Within <code>emulator_period</code>, emulator
threads (those excluding vCPUs) of the domain will not be allowed to consume
more than <code>emulator_quota</code> worth of runtime. The value should be
in range [1000, 1000000]. A period with value 0 means no value.
<span class="since">Only QEMU driver support since 0.10.0</span>
</dd>
<dt><code>emulator_quota</code></dt>
<dd>
The optional <code>emulator_quota</code> element specifies the maximum
allowed bandwidth (unit: microseconds) for domain's emulator threads (those
excluding vCPUs). A domain with <code>emulator_quota</code> as any negative
value indicates that the domain has infinite bandwidth for emulator threads
(those excluding vCPUs), which means that it is not bandwidth controlled.
The value should be in range [1000, 18446744073709551] or less than 0. A
quota with value 0 means no value.
<span class="since">Only QEMU driver support since 0.10.0</span>
</dd>
<dt><code>iothread_period</code></dt>
<dd>
The optional <code>iothread_period</code> element specifies the
enforcement interval (unit: microseconds) for IOThreads. Within
<code>iothread_period</code>, each IOThread of the domain will
not be allowed to consume more than <code>iothread_quota</code>
worth of runtime. The value should be in range [1000, 1000000].
An iothread_period with value 0 means no value.
<span class="since">Only QEMU driver support since 2.1.0</span>
</dd>
<dt><code>iothread_quota</code></dt>
<dd>
The optional <code>iothread_quota</code> element specifies the maximum
allowed bandwidth (unit: microseconds) for IOThreads. A domain with
<code>iothread_quota</code> as any negative value indicates that the
domain IOThreads have infinite bandwidth, which means that it is
not bandwidth controlled. The value should be in range
[1000, 18446744073709551] or less than 0. An <code>iothread_quota</code>
with value 0 means no value. You can use this feature to ensure that
all IOThreads run at the same speed.
<span class="since">Only QEMU driver support since 2.1.0</span>
</dd>
<dt><code>vcpusched</code>, <code>iothreadsched</code>
and <code>emulatorsched</code></dt>
<dd>
The optional
<code>vcpusched</code>, <code>iothreadsched</code>
and <code>emulatorsched</code> elements specify the scheduler type
(values <code>batch</code>, <code>idle</code>, <code>fifo</code>,
<code>rr</code>) for particular vCPU, IOThread and emulator threads
respecively. For <code>vcpusched</code> and <code>iothreadsched</code>
the attributes <code>vcpus</code> and <code>iothreads</code> select
which vCPUs/IOThreads this setting applies to, leaving them out sets the
default. The element <code>emulatorsched</code> does not have that
attribute. Valid <code>vcpus</code> values start at 0 through one less
than the number of vCPU's defined for the
domain. Valid <code>iothreads</code> values are described in
the <code>iothreadids</code>
<a href="#elementsIOThreadsAllocation"><code>description</code></a>.
If no <code>iothreadids</code> are defined, then libvirt numbers
IOThreads from 1 to the number of <code>iothreads</code> available
for the domain. For real-time schedulers (<code>fifo</code>,
<code>rr</code>), priority must be specified as
well (and is ignored for non-real-time ones). The value range
for the priority depends on the host kernel (usually 1-99).
<span class="since">Since 1.2.13</span>
<code>emulatorsched</code> <span class="since">since 5.3.0</span>
</dd>
<dt><code>cachetune</code><span class="since">Since 4.1.0</span></dt>
<dd>
Optional <code>cachetune</code> element can control allocations for CPU
caches using the resctrl on the host. Whether or not is this supported
can be gathered from capabilities where some limitations like minimum
size and required granularity are reported as well. The required
attribute <code>vcpus</code> specifies to which vCPUs this allocation
applies. A vCPU can only be member of one <code>cachetune</code> element
allocation. The vCPUs specified by cachetune can be identical with those
in memorytune, however they are not allowed to overlap.
Supported subelements are:
<dl>
<dt><code>cache</code></dt>
<dd>
This optional element controls the allocation of CPU cache and has
the following attributes:
<dl>
<dt><code>level</code></dt>
<dd>
Host cache level from which to allocate.
</dd>
<dt><code>id</code></dt>
<dd>
Host cache id from which to allocate.
</dd>
<dt><code>type</code></dt>
<dd>
Type of allocation. Can be <code>code</code> for code
(instructions), <code>data</code> for data or <code>both</code>
for both code and data (unified). Currently the allocation can
be done only with the same type as the host supports, meaning
you cannot request <code>both</code> for host with CDP
(code/data prioritization) enabled.
</dd>
<dt><code>size</code></dt>
<dd>
The size of the region to allocate. The value by default is in
bytes, but the <code>unit</code> attribute can be used to scale
the value.
</dd>
<dt><code>unit</code> (optional)</dt>
<dd>
If specified it is the unit such as KiB, MiB, GiB, or TiB
(described in the <code>memory</code> element
for <a href="#elementsMemoryAllocation">Memory Allocation</a>)
in which <code>size</code> is specified, defaults to bytes.
</dd>
</dl>
</dd>
<dt><code>monitor</code><span class="since">Since 4.10.0</span></dt>
<dd>
The optional element <code>monitor</code> creates the cache
monitor(s) for current cache allocation and has the following
required attributes:
<dl>
<dt><code>level</code></dt>
<dd>
Host cache level the monitor belongs to.
</dd>
<dt><code>vcpus</code></dt>
<dd>
vCPU list the monitor applies to. A monitor's vCPU list
can only be the member(s) of the vCPU list of the associated
allocation. The default monitor has the same vCPU list as the
associated allocation. For non-default monitors, overlapping
vCPUs are not permitted.
</dd>
</dl>
</dd>
</dl>
</dd>
<dt><code>memorytune</code><span class="since">Since 4.7.0</span></dt>
<dd>
Optional <code>memorytune</code> element can control allocations for
memory bandwidth using the resctrl on the host. Whether or not is this
supported can be gathered from capabilities where some limitations like
minimum bandwidth and required granularity are reported as well. The
required attribute <code>vcpus</code> specifies to which vCPUs this
allocation applies. A vCPU can only be member of one
<code>memorytune</code> element allocation. The <code>vcpus</code> specified
by <code>memorytune</code> can be identical to those specified by
<code>cachetune</code>. However they are not allowed to overlap each other.
Supported subelements are:
<dl>
<dt><code>node</code></dt>
<dd>
This element controls the allocation of CPU memory bandwidth and has the
following attributes:
<dl>
<dt><code>id</code></dt>
<dd>
Host node id from which to allocate memory bandwidth.
</dd>
<dt><code>bandwidth</code></dt>
<dd>
The memory bandwidth to allocate from this node. The value by default
is in percentage.
</dd>
</dl>
</dd>
</dl>
</dd>
</dl>
<h3><a id="elementsMemoryAllocation">Memory Allocation</a></h3>
<pre>
&lt;domain&gt;
...
&lt;maxMemory slots='16' unit='KiB'&gt;1524288&lt;/maxMemory&gt;
&lt;memory unit='KiB'&gt;524288&lt;/memory&gt;
&lt;currentMemory unit='KiB'&gt;524288&lt;/currentMemory&gt;
...
&lt;/domain&gt;
</pre>
<dl>
<dt><code>memory</code></dt>
<dd>The maximum allocation of memory for the guest at boot time. The
memory allocation includes possible additional memory devices specified
at start or hotplugged later.
The units for this value are determined by the optional
attribute <code>unit</code>, which defaults to "KiB"
(kibibytes, 2<sup>10</sup> or blocks of 1024 bytes). Valid
units are "b" or "bytes" for bytes, "KB" for kilobytes
(10<sup>3</sup> or 1,000 bytes), "k" or "KiB" for kibibytes
(1024 bytes), "MB" for megabytes (10<sup>6</sup> or 1,000,000
bytes), "M" or "MiB" for mebibytes (2<sup>20</sup> or
1,048,576 bytes), "GB" for gigabytes (10<sup>9</sup> or
1,000,000,000 bytes), "G" or "GiB" for gibibytes
(2<sup>30</sup> or 1,073,741,824 bytes), "TB" for terabytes
(10<sup>12</sup> or 1,000,000,000,000 bytes), or "T" or "TiB"
for tebibytes (2<sup>40</sup> or 1,099,511,627,776 bytes).
However, the value will be rounded up to the nearest kibibyte
by libvirt, and may be further rounded to the granularity
supported by the hypervisor. Some hypervisors also enforce a
minimum, such as 4000KiB.
In case <a href="#elementsCPU">NUMA</a> is configured for the guest the
<code>memory</code> element can be omitted.
In the case of crash, optional attribute <code>dumpCore</code>
can be used to control whether the guest memory should be
included in the generated coredump or not (values "on", "off").
<span class='since'><code>unit</code> since 0.9.11</span>,
<span class='since'><code>dumpCore</code> since 0.10.2
(QEMU only)</span></dd>
<dt><code>maxMemory</code></dt>
<dd>The run time maximum memory allocation of the guest. The initial
memory specified by either the <code>&lt;memory&gt;</code> element or
the NUMA cell size configuration can be increased by hot-plugging of
memory to the limit specified by this element.
The <code>unit</code> attribute behaves the same as for
<code>&lt;memory&gt;</code>.
The <code>slots</code> attribute specifies the number of slots
available for adding memory to the guest. The bounds are hypervisor
specific.
Note that due to alignment of the memory chunks added via memory
hotplug the full size allocation specified by this element may be
impossible to achieve.
<span class='since'>Since 1.2.14 supported by the QEMU driver.</span>
</dd>
<dt><code>currentMemory</code></dt>
<dd>The actual allocation of memory for the guest. This value can
be less than the maximum allocation, to allow for ballooning
up the guests memory on the fly. If this is omitted, it defaults
to the same value as the <code>memory</code> element.
The <code>unit</code> attribute behaves the same as
for <code>memory</code>.</dd>
</dl>
<h3><a id="elementsMemoryBacking">Memory Backing</a></h3>
<pre>
&lt;domain&gt;
...
&lt;memoryBacking&gt;
&lt;hugepages&gt;
&lt;page size="1" unit="G" nodeset="0-3,5"/&gt;
&lt;page size="2" unit="M" nodeset="4"/&gt;
&lt;/hugepages&gt;
&lt;nosharepages/&gt;
&lt;locked/&gt;
&lt;source type="file|anonymous|memfd"/&gt;
&lt;access mode="shared|private"/&gt;
&lt;allocation mode="immediate|ondemand"/&gt;
&lt;discard/&gt;
&lt;/memoryBacking&gt;
...
&lt;/domain&gt;
</pre>
<p>The optional <code>memoryBacking</code> element may contain several
elements that influence how virtual memory pages are backed by host
pages.</p>
<dl>
<dt><code>hugepages</code></dt>
<dd>This tells the hypervisor that the guest should have its memory
allocated using hugepages instead of the normal native page size.
<span class='since'>Since 1.2.5</span> it's possible to set hugepages
more specifically per numa node. The <code>page</code> element is
introduced. It has one compulsory attribute <code>size</code> which
specifies which hugepages should be used (especially useful on systems
supporting hugepages of different sizes). The default unit for the
<code>size</code> attribute is kilobytes (multiplier of 1024). If you
want to use different unit, use optional <code>unit</code> attribute.
For systems with NUMA, the optional <code>nodeset</code> attribute may
come handy as it ties given guest's NUMA nodes to certain hugepage
sizes. From the example snippet, one gigabyte hugepages are used for
every NUMA node except node number four. For the correct syntax see
<a href="#elementsNUMATuning">this</a>.</dd>
<dt><code>nosharepages</code></dt>
<dd>Instructs hypervisor to disable shared pages (memory merge, KSM) for
this domain. <span class="since">Since 1.0.6</span></dd>
<dt><code>locked</code></dt>
<dd>When set and supported by the hypervisor, memory pages belonging
to the domain will be locked in host's memory and the host will not
be allowed to swap them out, which might be required for some
workloads such as real-time. For QEMU/KVM guests, the memory used by
the QEMU process itself will be locked too: unlike guest memory, this
is an amount libvirt has no way of figuring out in advance, so it has
to remove the limit on locked memory altogether. Thus, enabling this
option opens up to a potential security risk: the host will be unable
to reclaim the locked memory back from the guest when it's running out
of memory, which means a malicious guest allocating large amounts of
locked memory could cause a denial-of-service attack on the host.
Because of this, using this option is discouraged unless your workload
demands it; even then, it's highly recommended to set a
<code>hard_limit</code> (see
<a href="#elementsMemoryTuning">memory tuning</a>) on memory allocation
suitable for the specific environment at the same time to mitigate
the risks described above. <span class="since">Since 1.0.6</span></dd>
<dt><code>source</code></dt>
<dd>Using the <code>type</code> attribute, it's possible to
provide "file" to utilize file memorybacking or keep the
default "anonymous". <span class="since">Since 4.10.0</span>,
you may choose "memfd" backing. (QEMU/KVM only)</dd>
<dt><code>access</code></dt>
<dd>Using the <code>mode</code> attribute, specify if the memory is
to be "shared" or "private". This can be overridden per numa node by
<code>memAccess</code>.</dd>
<dt><code>allocation</code></dt>
<dd>Using the <code>mode</code> attribute, specify when to allocate
the memory by supplying either "immediate" or "ondemand".</dd>
<dt><code>discard</code></dt>
<dd>When set and supported by hypervisor the memory
content is discarded just before guest shuts down (or
when DIMM module is unplugged). Please note that this is
just an optimization and is not guaranteed to work in
all cases (e.g. when hypervisor crashes).
<span class="since">Since 4.4.0</span> (QEMU/KVM only)
</dd>
</dl>
<h3><a id="elementsMemoryTuning">Memory Tuning</a></h3>
<pre>
&lt;domain&gt;
...
&lt;memtune&gt;
&lt;hard_limit unit='G'&gt;1&lt;/hard_limit&gt;
&lt;soft_limit unit='M'&gt;128&lt;/soft_limit&gt;
&lt;swap_hard_limit unit='G'&gt;2&lt;/swap_hard_limit&gt;
&lt;min_guarantee unit='bytes'&gt;67108864&lt;/min_guarantee&gt;
&lt;/memtune&gt;
...
&lt;/domain&gt;
</pre>
<dl>
<dt><code>memtune</code></dt>
<dd> The optional <code>memtune</code> element provides details
regarding the memory tunable parameters for the domain. If this is
omitted, it defaults to the OS provided defaults. For QEMU/KVM, the
parameters are applied to the QEMU process as a whole. Thus, when
counting them, one needs to add up guest RAM, guest video RAM, and
some memory overhead of QEMU itself. The last piece is hard to
determine so one needs guess and try. For each tunable, it
is possible to designate which unit the number is in on
input, using the same values as
for <code>&lt;memory&gt;</code>. For backwards
compatibility, output is always in
KiB. <span class='since'><code>unit</code>
since 0.9.11</span>
Possible values for all *_limit parameters are in range from 0 to
VIR_DOMAIN_MEMORY_PARAM_UNLIMITED.</dd>
<dt><code>hard_limit</code></dt>
<dd> The optional <code>hard_limit</code> element is the maximum memory
the guest can use. The units for this value are kibibytes (i.e. blocks
of 1024 bytes). Users of QEMU and KVM are strongly advised not to set
this limit as domain may get killed by the kernel if the guess is too
low, and determining the memory needed for a process to run is an
<a href="http://en.wikipedia.org/wiki/Undecidable_problem">
undecidable problem</a>; that said, if you already set
<code>locked</code> in
<a href="#elementsMemoryBacking">memory backing</a> because your
workload demands it, you'll have to take into account the specifics of
your deployment and figure out a value for <code>hard_limit</code> that
is large enough to support the memory requirements of your guest, but
small enough to protect your host against a malicious guest locking all
memory.</dd>
<dt><code>soft_limit</code></dt>
<dd> The optional <code>soft_limit</code> element is the memory limit to
enforce during memory contention. The units for this value are
kibibytes (i.e. blocks of 1024 bytes)</dd>
<dt><code>swap_hard_limit</code></dt>
<dd> The optional <code>swap_hard_limit</code> element is the maximum
memory plus swap the guest can use. The units for this value are
kibibytes (i.e. blocks of 1024 bytes). This has to be more than
hard_limit value provided</dd>
<dt><code>min_guarantee</code></dt>
<dd> The optional <code>min_guarantee</code> element is the guaranteed
minimum memory allocation for the guest. The units for this value are
kibibytes (i.e. blocks of 1024 bytes). This element is only supported
by VMware ESX and OpenVZ drivers.</dd>
</dl>
<h3><a id="elementsNUMATuning">NUMA Node Tuning</a></h3>
<pre>
&lt;domain&gt;
...
&lt;numatune&gt;
&lt;memory mode="strict" nodeset="1-4,^3"/&gt;
&lt;memnode cellid="0" mode="strict" nodeset="1"/&gt;
&lt;memnode cellid="2" mode="preferred" nodeset="2"/&gt;
&lt;/numatune&gt;
...
&lt;/domain&gt;
</pre>
<dl>
<dt><code>numatune</code></dt>
<dd>
The optional <code>numatune</code> element provides details of
how to tune the performance of a NUMA host via controlling NUMA policy
for domain process. NB, only supported by QEMU driver.
<span class='since'>Since 0.9.3</span>
</dd>
<dt><code>memory</code></dt>
<dd>
The optional <code>memory</code> element specifies how to allocate memory
for the domain process on a NUMA host. It contains several optional
attributes. Attribute <code>mode</code> is either 'interleave',
'strict', or 'preferred', defaults to 'strict'. Attribute
<code>nodeset</code> specifies the NUMA nodes, using the same syntax as
attribute <code>cpuset</code> of element <code>vcpu</code>. Attribute
<code>placement</code> (<span class='since'>since 0.9.12</span>) can be
used to indicate the memory placement mode for domain process, its value
can be either "static" or "auto", defaults to <code>placement</code> of
<code>vcpu</code>, or "static" if <code>nodeset</code> is specified.
"auto" indicates the domain process will only allocate memory from the
advisory nodeset returned from querying numad, and the value of attribute
<code>nodeset</code> will be ignored if it's specified.
If <code>placement</code> of <code>vcpu</code> is 'auto', and
<code>numatune</code> is not specified, a default <code>numatune</code>
with <code>placement</code> 'auto' and <code>mode</code> 'strict' will
be added implicitly.
<span class='since'>Since 0.9.3</span>
</dd>
<dt><code>memnode</code></dt>
<dd>
Optional <code>memnode</code> elements can specify memory allocation
policies per each guest NUMA node. For those nodes having no
corresponding <code>memnode</code> element, the default from
element <code>memory</code> will be used. Attribute <code>cellid</code>
addresses guest NUMA node for which the settings are applied.
Attributes <code>mode</code> and <code>nodeset</code> have the same
meaning and syntax as in <code>memory</code> element.
This setting is not compatible with automatic placement.
<span class='since'>QEMU Since 1.2.7</span>
</dd>
</dl>
<h3><a id="elementsBlockTuning">Block I/O Tuning</a></h3>
<pre>
&lt;domain&gt;
...
&lt;blkiotune&gt;
&lt;weight&gt;800&lt;/weight&gt;
&lt;device&gt;
&lt;path&gt;/dev/sda&lt;/path&gt;
&lt;weight&gt;1000&lt;/weight&gt;
&lt;/device&gt;
&lt;device&gt;
&lt;path&gt;/dev/sdb&lt;/path&gt;
&lt;weight&gt;500&lt;/weight&gt;
&lt;read_bytes_sec&gt;10000&lt;/read_bytes_sec&gt;
&lt;write_bytes_sec&gt;10000&lt;/write_bytes_sec&gt;
&lt;read_iops_sec&gt;20000&lt;/read_iops_sec&gt;
&lt;write_iops_sec&gt;20000&lt;/write_iops_sec&gt;
&lt;/device&gt;
&lt;/blkiotune&gt;
...
&lt;/domain&gt;
</pre>
<dl>
<dt><code>blkiotune</code></dt>
<dd> The optional <code>blkiotune</code> element provides the ability
to tune Blkio cgroup tunable parameters for the domain. If this is
omitted, it defaults to the OS provided
defaults. <span class="since">Since 0.8.8</span></dd>
<dt><code>weight</code></dt>
<dd> The optional <code>weight</code> element is the overall I/O
weight of the guest. The value should be in the range [100,
1000]. After kernel 2.6.39, the value could be in the
range [10, 1000].</dd>
<dt><code>device</code></dt>
<dd>The domain may have multiple <code>device</code> elements
that further tune the weights for each host block device in
use by the domain. Note that
multiple <a href="#elementsDisks">guest disks</a> can share a
single host block device, if they are backed by files within
the same host file system, which is why this tuning parameter
is at the global domain level rather than associated with each
guest disk device (contrast this to
the <a href="#elementsDisks"><code>&lt;iotune&gt;</code></a>
element which can apply to an
individual <code>&lt;disk&gt;</code>).
Each <code>device</code> element has two
mandatory sub-elements, <code>path</code> describing the
absolute path of the device, and <code>weight</code> giving
the relative weight of that device, in the range [100,
1000]. After kernel 2.6.39, the value could be in the
range [10, 1000]. <span class="since">Since 0.9.8</span><br/>
Additionally, the following optional sub-elements can be used:
<dl>
<dt><code>read_bytes_sec</code></dt>
<dd>Read throughput limit in bytes per second.
<span class="since">Since 1.2.2</span></dd>
<dt><code>write_bytes_sec</code></dt>
<dd>Write throughput limit in bytes per second.
<span class="since">Since 1.2.2</span></dd>
<dt><code>read_iops_sec</code></dt>
<dd>Read I/O operations per second limit.
<span class="since">Since 1.2.2</span></dd>
<dt><code>write_iops_sec</code></dt>
<dd>Write I/O operations per second limit.
<span class="since">Since 1.2.2</span></dd>
</dl></dd></dl>
<h3><a id="resPartition">Resource partitioning</a></h3>
<p>
Hypervisors may allow for virtual machines to be placed into
resource partitions, potentially with nesting of said partitions.
The <code>resource</code> element groups together configuration
related to resource partitioning. It currently supports a child
element <code>partition</code> whose content defines the absolute path
of the resource partition in which to place the domain. If no
partition is listed, then the domain will be placed in a default
partition. It is the responsibility of the app/admin to ensure
that the partition exists prior to starting the guest. Only the
(hypervisor specific) default partition can be assumed to exist
by default.
</p>
<pre>
...
&lt;resource&gt;
&lt;partition&gt;/virtualmachines/production&lt;/partition&gt;
&lt;/resource&gt;
...
</pre>
<p>
Resource partitions are currently supported by the QEMU and
LXC drivers, which map partition paths to cgroups directories,
in all mounted controllers. <span class="since">Since 1.0.5</span>
</p>
<h3><a id="elementsCPU">CPU model and topology</a></h3>
<p>
Requirements for CPU model, its features and topology can be specified
using the following collection of elements.
<span class="since">Since 0.7.5</span>
</p>
<pre>
...
&lt;cpu match='exact'&gt;
&lt;model fallback='allow'&gt;core2duo&lt;/model&gt;
&lt;vendor&gt;Intel&lt;/vendor&gt;
&lt;topology sockets='1' dies='1' cores='2' threads='1'/&gt;
&lt;cache level='3' mode='emulate'/&gt;
&lt;feature policy='disable' name='lahf_lm'/&gt;
&lt;/cpu&gt;
...</pre>
<pre>
&lt;cpu mode='host-model'&gt;
&lt;model fallback='forbid'/&gt;
&lt;topology sockets='1' dies='1' cores='2' threads='1'/&gt;
&lt;/cpu&gt;
...</pre>
<pre>
&lt;cpu mode='host-passthrough'&gt;
&lt;cache mode='passthrough'/&gt;
&lt;feature policy='disable' name='lahf_lm'/&gt;
...</pre>
<p>
In case no restrictions need to be put on CPU model and its features, a
simpler <code>cpu</code> element can be used.
<span class="since">Since 0.7.6</span>
</p>
<pre>
...
&lt;cpu&gt;
&lt;topology sockets='1' dies='1' cores='2' threads='1'/&gt;
&lt;/cpu&gt;
...</pre>
<dl>
<dt><code>cpu</code></dt>
<dd>The <code>cpu</code> element is the main container for describing
guest CPU requirements. Its <code>match</code> attribute specifies how
strictly the virtual CPU provided to the guest matches these
requirements. <span class="since">Since 0.7.6</span> the
<code>match</code> attribute can be omitted if <code>topology</code>
is the only element within <code>cpu</code>. Possible values for the
<code>match</code> attribute are:
<dl>
<dt><code>minimum</code></dt>
<dd>The specified CPU model and features describes the minimum
requested CPU. A better CPU will be provided to the guest if it
is possible with the requested hypervisor on the current host.
This is a constrained <code>host-model</code> mode; the domain
will not be created if the provided virtual CPU does not meet
the requirements.</dd>
<dt><code>exact</code></dt>
<dd>The virtual CPU provided to the guest should exactly match the
specification. If such CPU is not supported, libvirt will refuse
to start the domain.</dd>
<dt><code>strict</code></dt>
<dd>The domain will not be created unless the host CPU exactly
matches the specification. This is not very useful in practice
and should only be used if there is a real reason.</dd>
</dl>
<span class="since">Since 0.8.5</span> the <code>match</code>
attribute can be omitted and will default to <code>exact</code>.
Sometimes the hypervisor is not able to create a virtual CPU exactly
matching the specification passed by libvirt.
<span class="since">Since 3.2.0</span>, an optional <code>check</code>
attribute can be used to request a specific way of checking whether
the virtual CPU matches the specification. It is usually safe to omit
this attribute when starting a domain and stick with the default
value. Once the domain starts, libvirt will automatically change the
<code>check</code> attribute to the best supported value to ensure the
virtual CPU does not change when the domain is migrated to another
host. The following values can be used:
<dl>
<dt><code>none</code></dt>
<dd>Libvirt does no checking and it is up to the hypervisor to
refuse to start the domain if it cannot provide the requested CPU.
With QEMU this means no checking is done at all since the default
behavior of QEMU is to emit warnings, but start the domain anyway.
</dd>
<dt><code>partial</code></dt>
<dd>Libvirt will check the guest CPU specification before starting
a domain, but the rest is left on the hypervisor. It can still
provide a different virtual CPU.</dd>
<dt><code>full</code></dt>
<dd>The virtual CPU created by the hypervisor will be checked
against the CPU specification and the domain will not be started
unless the two CPUs match.</dd>
</dl>
<span class="since">Since 0.9.10</span>, an optional <code>mode</code>
attribute may be used to make it easier to configure a guest CPU to be
as close to host CPU as possible. Possible values for the
<code>mode</code> attribute are:
<dl>
<dt><code>custom</code></dt>
<dd>In this mode, the <code>cpu</code> element describes the CPU
that should be presented to the guest. This is the default when no
<code>mode</code> attribute is specified. This mode makes it so that
a persistent guest will see the same hardware no matter what host
the guest is booted on.</dd>
<dt><code>host-model</code></dt>
<dd>The <code>host-model</code> mode is essentially a shortcut to
copying host CPU definition from capabilities XML into domain XML.
Since the CPU definition is copied just before starting a domain,
exactly the same XML can be used on different hosts while still
providing the best guest CPU each host supports. The
<code>match</code> attribute can't be used in this mode. Specifying
CPU model is not supported either, but <code>model</code>'s
<code>fallback</code> attribute may still be used. Using the
<code>feature</code> element, specific flags may be enabled or
disabled specifically in addition to the host model. This may be
used to fine tune features that can be emulated.
<span class="since">(Since 1.1.1)</span>.
Libvirt does not model every aspect of each CPU so
the guest CPU will not match the host CPU exactly. On the other
hand, the ABI provided to the guest is reproducible. During
migration, complete CPU model definition is transferred to the
destination host so the migrated guest will see exactly the same CPU
model for the running instance of the guest, even if the destination
host contains more capable CPUs or newer kernel; but shutting down and restarting
the guest may present different hardware to the guest according to
the capabilities of the new host. Prior to libvirt 3.2.0 and QEMU
2.9.0 detection of the host CPU model via QEMU is not supported.
Thus the CPU configuration created using <code>host-model</code>
may not work as expected.
<span class="since">Since 3.2.0 and QEMU 2.9.0</span> this mode
works the way it was designed and it is indicated by the
<code>fallback</code> attribute set to <code>forbid</code> in the
host-model CPU definition advertised in
<a href="formatdomaincaps.html#elementsCPU">domain capabilities XML</a>.
When <code>fallback</code> attribute is set to <code>allow</code>
in the domain capabilities XML, it is recommended to use
<code>custom</code> mode with just the CPU model from the host
capabilities XML. <span class="since">Since 1.2.11</span> PowerISA
allows processors to run VMs in binary compatibility mode supporting
an older version of ISA. Libvirt on PowerPC architecture uses the
<code>host-model</code> to signify a guest mode CPU running in
binary compatibility mode. Example:
When a user needs a power7 VM to run in compatibility mode
on a Power8 host, this can be described in XML as follows :
<pre>
&lt;cpu mode='host-model'&gt;
&lt;model&gt;power7&lt;/model&gt;
&lt;/cpu&gt;
...</pre>
</dd>
<dt><code>host-passthrough</code></dt>
<dd>With this mode, the CPU visible to the guest should be exactly
the same as the host CPU even in the aspects that libvirt does not
understand. Though the downside of this mode is that the guest
environment cannot be reproduced on different hardware. Thus, if you
hit any bugs, you are on your own. Further details of that CPU can
be changed using <code>feature</code> elements. Migration of a guest
using host-passthrough is dangerous if the source and destination hosts
are not identical in both hardware, QEMU version, microcode version
and configuration. If such a migration is attempted then the guest may
hang or crash upon resuming execution on the destination host.</dd>
</dl>
Both <code>host-model</code> and <code>host-passthrough</code> modes
make sense when a domain can run directly on the host CPUs (for
example, domains with type <code>kvm</code>). The actual host CPU is
irrelevant for domains with emulated virtual CPUs (such as domains with
type <code>qemu</code>). However, for backward compatibility
<code>host-model</code> may be implemented even for domains running on
emulated CPUs in which case the best CPU the hypervisor is able to
emulate may be used rather then trying to mimic the host CPU model.
</dd>
<dt><code>model</code></dt>
<dd>The content of the <code>model</code> element specifies CPU model
requested by the guest. The list of available CPU models and their
definition can be found in <code>cpu_map.xml</code> file installed
in libvirt's data directory. If a hypervisor is not able to use the
exact CPU model, libvirt automatically falls back to a closest model
supported by the hypervisor while maintaining the list of CPU
features. <span class="since">Since 0.9.10</span>, an optional
<code>fallback</code> attribute can be used to forbid this behavior,
in which case an attempt to start a domain requesting an unsupported
CPU model will fail. Supported values for <code>fallback</code>
attribute are: <code>allow</code> (this is the default), and
<code>forbid</code>. The optional <code>vendor_id</code> attribute
(<span class="since">Since 0.10.0</span>) can be used to set the
vendor id seen by the guest. It must be exactly 12 characters long.
If not set the vendor id of the host is used. Typical possible
values are "AuthenticAMD" and "GenuineIntel".</dd>
<dt><code>vendor</code></dt>
<dd><span class="since">Since 0.8.3</span> the content of the
<code>vendor</code> element specifies CPU vendor requested by the
guest. If this element is missing, the guest can be run on a CPU
matching given features regardless on its vendor. The list of
supported vendors can be found in <code>cpu_map.xml</code>.</dd>
<dt><code>topology</code></dt>
<dd>The <code>topology</code> element specifies requested topology of
virtual CPU provided to the guest. Four attributes, <code>sockets</code>,
<code>dies</code>, <code>cores</code>, and <code>threads</code>,
accept non-zero positive integer values. They refer to the total number
of CPU sockets, number of dies per socket, number of cores per die, and
number of threads per core, respectively. The <code>dies</code>
attribute is optional and will default to 1 if omitted, while the other
attributes are all mandatory. Hypervisors may require that the maximum
number of vCPUs specified by the <code>cpus</code> element equals to
the number of vcpus resulting from the topology.</dd>
<dt><code>feature</code></dt>
<dd>The <code>cpu</code> element can contain zero or more
<code>elements</code> used to fine-tune features provided by the
selected CPU model. The list of known feature names can be found in
the same file as CPU models. The meaning of each <code>feature</code>
element depends on its <code>policy</code> attribute, which has to be
set to one of the following values:
<dl>
<dt><code>force</code></dt>
<dd>The virtual CPU will claim the feature is supported regardless
of it being supported by host CPU.</dd>
<dt><code>require</code></dt>
<dd>Guest creation will fail unless the feature is supported by the
host CPU or the hypervisor is able to emulate it.</dd>
<dt><code>optional</code></dt>
<dd>The feature will be supported by virtual CPU if and only if it
is supported by host CPU.</dd>
<dt><code>disable</code></dt>
<dd>The feature will not be supported by virtual CPU.</dd>
<dt><code>forbid</code></dt>
<dd>Guest creation will fail if the feature is supported by host
CPU.</dd>
</dl>
<span class="since">Since 0.8.5</span> the <code>policy</code>
attribute can be omitted and will default to <code>require</code>.
<p> Individual CPU feature names are specified as part of the
<code>name</code> attribute. For example, to explicitly specify
the 'pcid' feature with Intel IvyBridge CPU model:
</p>
<pre>
...
&lt;cpu match='exact'&gt;
&lt;model fallback='forbid'&gt;IvyBridge&lt;/model&gt;
&lt;vendor&gt;Intel&lt;/vendor&gt;
&lt;feature policy='require' name='pcid'/&gt;
&lt;/cpu&gt;
...</pre>
</dd>
<dt><code>cache</code></dt>
<dd><span class="since">Since 3.3.0</span> the <code>cache</code>
element describes the virtual CPU cache. If the element is missing,
the hypervisor will use a sensible default.
<dl>
<dt><code>level</code></dt>
<dd>This optional attribute specifies which cache level is described
by the element. Missing attribute means the element describes all
CPU cache levels at once. Mixing <code>cache</code> elements with
the <code>level</code> attribute set and those without the
attribute is forbidden.</dd>
<dt><code>mode</code></dt>
<dd>
The following values are supported:
<dl>
<dt><code>emulate</code></dt>
<dd>The hypervisor will provide a fake CPU cache data.</dd>
<dt><code>passthrough</code></dt>
<dd>The real CPU cache data reported by the host CPU will be
passed through to the virtual CPU.</dd>
<dt><code>disable</code></dt>
<dd>The virtual CPU will report no CPU cache of the specified
level (or no cache at all if the <code>level</code> attribute
is missing).</dd>
</dl>
</dd>
</dl>
</dd>
</dl>
<p>
Guest NUMA topology can be specified using the <code>numa</code> element.
<span class="since">Since 0.9.8</span>
</p>
<pre>
...
&lt;cpu&gt;
...
&lt;numa&gt;
&lt;cell id='0' cpus='0-3' memory='512000' unit='KiB' discard='yes'/&gt;
&lt;cell id='1' cpus='4-7' memory='512000' unit='KiB' memAccess='shared'/&gt;
&lt;/numa&gt;
...
&lt;/cpu&gt;
...</pre>
<p>
Each <code>cell</code> element specifies a NUMA cell or a NUMA node.
<code>cpus</code> specifies the CPU or range of CPUs that are
part of the node. <code>memory</code> specifies the node memory
in kibibytes (i.e. blocks of 1024 bytes).
<span class="since">Since 1.2.11</span> one can use an additional <a
href="#elementsMemoryAllocation"><code>unit</code></a> attribute to
define units in which <code>memory</code> is specified.
<span class="since">Since 1.2.7</span> all cells should
have <code>id</code> attribute in case referring to some cell is
necessary in the code, otherwise the cells are
assigned <code>id</code>s in the increasing order starting from
0. Mixing cells with and without the <code>id</code> attribute
is not recommended as it may result in unwanted behaviour.
<span class='since'>Since 1.2.9</span> the optional attribute
<code>memAccess</code> can control whether the memory is to be
mapped as "shared" or "private". This is valid only for
hugepages-backed memory and nvdimm modules.
Each <code>cell</code> element can have an optional
<code>discard</code> attribute which fine tunes the discard
feature for given numa node as described under
<a href="#elementsMemoryBacking">Memory Backing</a>.
Accepted values are <code>yes</code> and <code>no</code>.
<span class='since'>Since 4.4.0</span>
</p>
<p>
This guest NUMA specification is currently available only for
QEMU/KVM and Xen.
</p>
<p>
A NUMA hardware architecture supports the notion of distances
between NUMA cells. <span class="since">Since 3.10.0</span> it
is possible to define the distance between NUMA cells using the
<code>distances</code> element within a NUMA <code>cell</code>
description. The <code>sibling</code> sub-element is used to
specify the distance value between sibling NUMA cells. For more
details, see the chapter explaining the system's SLIT (System
Locality Information Table) within the ACPI (Advanced
Configuration and Power Interface) specification.
</p>
<pre>
...
&lt;cpu&gt;
...
&lt;numa&gt;
&lt;cell id='0' cpus='0,4-7' memory='512000' unit='KiB'&gt;
&lt;distances&gt;
&lt;sibling id='0' value='10'/&gt;
&lt;sibling id='1' value='21'/&gt;
&lt;sibling id='2' value='31'/&gt;
&lt;sibling id='3' value='41'/&gt;
&lt;/distances&gt;
&lt;/cell&gt;
&lt;cell id='1' cpus='1,8-10,12-15' memory='512000' unit='KiB' memAccess='shared'&gt;
&lt;distances&gt;
&lt;sibling id='0' value='21'/&gt;
&lt;sibling id='1' value='10'/&gt;
&lt;sibling id='2' value='21'/&gt;
&lt;sibling id='3' value='31'/&gt;
&lt;/distances&gt;
&lt;/cell&gt;
&lt;cell id='2' cpus='2,11' memory='512000' unit='KiB' memAccess='shared'&gt;
&lt;distances&gt;
&lt;sibling id='0' value='31'/&gt;
&lt;sibling id='1' value='21'/&gt;
&lt;sibling id='2' value='10'/&gt;
&lt;sibling id='3' value='21'/&gt;
&lt;/distances&gt;
&lt;/cell&gt;
&lt;cell id='3' cpus='3' memory='512000' unit='KiB'&gt;
&lt;distances&gt;
&lt;sibling id='0' value='41'/&gt;
&lt;sibling id='1' value='31'/&gt;
&lt;sibling id='2' value='21'/&gt;
&lt;sibling id='3' value='10'/&gt;
&lt;/distances&gt;
&lt;/cell&gt;
&lt;/numa&gt;
...
&lt;/cpu&gt;
...</pre>
<p>
Describing distances between NUMA cells is currently only supported
by Xen and QEMU. If no <code>distances</code> are given to describe
the SLIT data between different cells, it will default to a scheme
using 10 for local and 20 for remote distances.
</p>
<h3><a id="elementsEvents">Events configuration</a></h3>
<p>
It is sometimes necessary to override the default actions taken
on various events. Not all hypervisors support all events and actions.
The actions may be taken as a result of calls to libvirt APIs
<a href="html/libvirt-libvirt-domain.html#virDomainReboot">
<code>virDomainReboot</code>
</a>,
<a href="html/libvirt-libvirt-domain.html#virDomainShutdown">
<code>virDomainShutdown</code>
</a>,
or
<a href="html/libvirt-libvirt-domain.html#virDomainShutdownFlags">
<code>virDomainShutdownFlags</code>
</a>.
Using <code>virsh reboot</code> or <code>virsh shutdown</code> would
also trigger the event.
</p>
<pre>
...
&lt;on_poweroff&gt;destroy&lt;/on_poweroff&gt;
&lt;on_reboot&gt;restart&lt;/on_reboot&gt;
&lt;on_crash&gt;restart&lt;/on_crash&gt;
&lt;on_lockfailure&gt;poweroff&lt;/on_lockfailure&gt;
...</pre>
<p>
The following collections of elements allow the actions to be
specified when a guest OS triggers a lifecycle operation. A
common use case is to force a reboot to be treated as a poweroff
when doing the initial OS installation. This allows the VM to be
re-configured for the first post-install bootup.
</p>
<dl>
<dt><code>on_poweroff</code></dt>
<dd>The content of this element specifies the action to take when
the guest requests a poweroff.</dd>
<dt><code>on_reboot</code></dt>
<dd>The content of this element specifies the action to take when
the guest requests a reboot.</dd>
<dt><code>on_crash</code></dt>
<dd>The content of this element specifies the action to take when
the guest crashes.</dd>
</dl>
<p>
Each of these states allow for the same four possible actions.
</p>
<dl>
<dt><code>destroy</code></dt>
<dd>The domain will be terminated completely and all resources
released.</dd>
<dt><code>restart</code></dt>
<dd>The domain will be terminated and then restarted with
the same configuration.</dd>
<dt><code>preserve</code></dt>
<dd>The domain will be terminated and its resource preserved
to allow analysis.</dd>
<dt><code>rename-restart</code></dt>
<dd>The domain will be terminated and then restarted with
a new name.</dd>
</dl>
<p>
QEMU/KVM supports the <code>on_poweroff</code> and <code>on_reboot</code>
events handling the <code>destroy</code> and <code>restart</code> actions.
The <code>preserve</code> action for an <code>on_reboot</code> event
is treated as a <code>destroy</code> and the <code>rename-restart</code>
action for an <code>on_poweroff</code> event is treated as a
<code>restart</code> event.
</p>
<p>
The <code>on_crash</code> event supports these additional
actions <span class="since">since 0.8.4</span>.
</p>
<dl>
<dt><code>coredump-destroy</code></dt>
<dd>The crashed domain's core will be dumped, and then the
domain will be terminated completely and all resources
released</dd>
<dt><code>coredump-restart</code></dt>
<dd>The crashed domain's core will be dumped, and then the
domain will be restarted with the same configuration</dd>
</dl>
<p>
<span class="since">Since 3.9.0</span>, the lifecycle events can
be configured via the
<a href="html/libvirt-libvirt-domain.html#virDomainSetLifecycleAction">
<code>virDomainSetLifecycleAction</code></a> API.
</p>
<p>
The <code>on_lockfailure</code> element (<span class="since">since
1.0.0</span>) may be used to configure what action should be
taken when a lock manager loses resource locks. The following
actions are recognized by libvirt, although not all of them need
to be supported by individual lock managers. When no action is
specified, each lock manager will take its default action.
</p>
<dl>
<dt><code>poweroff</code></dt>
<dd>The domain will be forcefully powered off.</dd>
<dt><code>restart</code></dt>
<dd>The domain will be powered off and started up again to
reacquire its locks.</dd>
<dt><code>pause</code></dt>
<dd>The domain will be paused so that it can be manually resumed
when lock issues are solved.</dd>
<dt><code>ignore</code></dt>
<dd>Keep the domain running as if nothing happened.</dd>
</dl>
<h3><a id="elementsPowerManagement">Power Management</a></h3>
<p>
<span class="since">Since 0.10.2</span> it is possible to
forcibly enable or disable BIOS advertisements to the guest
OS. (NB: Only qemu driver support)
</p>
<pre>
...
&lt;pm&gt;
&lt;suspend-to-disk enabled='no'/&gt;
&lt;suspend-to-mem enabled='yes'/&gt;
&lt;/pm&gt;
...</pre>
<dl>
<dt><code>pm</code></dt>
<dd>These elements enable ('yes') or disable ('no') BIOS support
for S3 (suspend-to-mem) and S4 (suspend-to-disk) ACPI sleep
states. If nothing is specified, then the hypervisor will be
left with its default value.<br/>
Note: This setting cannot prevent the guest OS from performing
a suspend as the guest OS itself can choose to circumvent the
unavailability of the sleep states (e.g. S4 by turning off
completely).</dd>
</dl>
<h3><a id="elementsFeatures">Hypervisor features</a></h3>
<p>
Hypervisors may allow certain CPU / machine features to be
toggled on/off.
</p>
<pre>
...
&lt;features&gt;
&lt;pae/&gt;
&lt;acpi/&gt;
&lt;apic/&gt;
&lt;hap/&gt;
&lt;privnet/&gt;
&lt;hyperv&gt;
&lt;relaxed state='on'/&gt;
&lt;vapic state='on'/&gt;
&lt;spinlocks state='on' retries='4096'/&gt;
&lt;vpindex state='on'/&gt;
&lt;runtime state='on'/&gt;
&lt;synic state='on'/&gt;
&lt;stimer state='on'&gt;
&lt;direct state='on'/&gt;
&lt;/stimer&gt;
&lt;reset state='on'/&gt;
&lt;vendor_id state='on' value='KVM Hv'/&gt;
&lt;frequencies state='on'/&gt;
&lt;reenlightenment state='on'/&gt;
&lt;tlbflush state='on'/&gt;
&lt;ipi state='on'/&gt;
&lt;evmcs state='on'/&gt;
&lt;/hyperv&gt;
&lt;kvm&gt;
&lt;hidden state='on'/&gt;
&lt;hint-dedicated state='on'/&gt;
&lt;/kvm&gt;
&lt;pvspinlock state='on'/&gt;
&lt;gic version='2'/&gt;
&lt;ioapic driver='qemu'/&gt;
&lt;hpt resizing='required'&gt;
&lt;maxpagesize unit='MiB'&gt;16&lt;/maxpagesize&gt;
&lt;/hpt&gt;
&lt;vmcoreinfo state='on'/&gt;
&lt;smm state='on'&gt;
&lt;tseg unit='MiB'&gt;48&lt;/tseg&gt;
&lt;/smm&gt;
&lt;htm state='on'/&gt;
&lt;ccf-assist state='on'/&gt;
&lt;msrs unknown='ignore'/&gt;
&lt;/features&gt;
...</pre>
<p>
All features are listed within the <code>features</code>
element, omitting a togglable feature tag turns it off.
The available features can be found by asking
for the <a href="formatcaps.html">capabilities XML</a> and
<a href="formatdomaincaps.html">domain capabilities XML</a>,
but a common set for fully virtualized domains are:
</p>
<dl>
<dt><code>pae</code></dt>
<dd>Physical address extension mode allows 32-bit guests
to address more than 4 GB of memory.</dd>
<dt><code>acpi</code></dt>
<dd>ACPI is useful for power management, for example, with
KVM guests it is required for graceful shutdown to work.
</dd>
<dt><code>apic</code></dt>
<dd>APIC allows the use of programmable IRQ
management. <span class="since">Since 0.10.2 (QEMU only)</span> there is
an optional attribute <code>eoi</code> with values <code>on</code>
and <code>off</code> which toggles the availability of EOI (End of
Interrupt) for the guest.
</dd>
<dt><code>hap</code></dt>
<dd>Depending on the <code>state</code> attribute (values <code>on</code>,
<code>off</code>) enable or disable use of Hardware Assisted Paging.
The default is <code>on</code> if the hypervisor detects availability
of Hardware Assisted Paging.
</dd>
<dt><code>viridian</code></dt>
<dd>Enable Viridian hypervisor extensions for paravirtualizing
guest operating systems
</dd>
<dt><code>privnet</code></dt>
<dd>Always create a private network namespace. This is
automatically set if any interface devices are defined.
This feature is only relevant for container based
virtualization drivers, such as LXC.
</dd>
<dt><code>hyperv</code></dt>
<dd>Enable various features improving behavior of guests
running Microsoft Windows.
<table class="top_table">
<tr>
<th>Feature</th>
<th>Description</th>
<th>Value</th>
<th>Since</th>
</tr>
<tr>
<td>relaxed</td>
<td>Relax constraints on timers</td>
<td>on, off</td>
<td><span class="since">1.0.0 (QEMU 2.0)</span></td>
</tr>
<tr>
<td>vapic</td>
<td>Enable virtual APIC</td>
<td>on, off</td>
<td><span class="since">1.1.0 (QEMU 2.0)</span></td>
</tr>
<tr>
<td>spinlocks</td>
<td>Enable spinlock support</td>
<td>on, off; retries - at least 4095</td>
<td><span class="since">1.1.0 (QEMU 2.0)</span></td>
</tr>
<tr>
<td>vpindex</td>
<td>Virtual processor index</td>
<td>on, off</td>
<td><span class="since">1.3.3 (QEMU 2.5)</span></td>
</tr>
<tr>
<td>runtime</td>
<td>Processor time spent on running guest code and on behalf of guest code</td>
<td>on, off</td>
<td><span class="since">1.3.3 (QEMU 2.5)</span></td>
</tr>
<tr>
<td>synic</td>
<td>Enable Synthetic Interrupt Controller (SynIC)</td>
<td>on, off</td>
<td><span class="since">1.3.3 (QEMU 2.6)</span></td>
</tr>
<tr>
<td>stimer</td>
<td>Enable SynIC timers, optionally with Direct Mode support</td>
<td>on, off; direct - on,off</td>
<td><span class="since">1.3.3 (QEMU 2.6), direct mode 5.7.0 (QEMU 4.1)</span></td>
</tr>
<tr>
<td>reset</td>
<td>Enable hypervisor reset</td>
<td>on, off</td>
<td><span class="since">1.3.3 (QEMU 2.5)</span></td>
</tr>
<tr>
<td>vendor_id</td>
<td>Set hypervisor vendor id</td>
<td>on, off; value - string, up to 12 characters</td>
<td><span class="since">1.3.3 (QEMU 2.5)</span></td>
</tr>
<tr>
<td>frequencies</td>
<td>Expose frequency MSRs</td>
<td>on, off</td>
<td><span class="since">4.7.0 (QEMU 2.12)</span></td>
</tr>
<tr>
<td>reenlightenment</td>
<td>Enable re-enlightenment notification on migration</td>
<td>on, off</td>
<td><span class="since">4.7.0 (QEMU 3.0)</span></td>
</tr>
<tr>
<td>tlbflush</td>
<td>Enable PV TLB flush support</td>
<td>on, off</td>
<td><span class="since">4.7.0 (QEMU 3.0)</span></td>
</tr>
<tr>
<td>ipi</td>
<td>Enable PV IPI support</td>
<td>on, off</td>
<td><span class="since">4.10.0 (QEMU 3.1)</span></td>
</tr>
<tr>
<td>evmcs</td>
<td>Enable Enlightened VMCS</td>
<td>on, off</td>
<td><span class="since">4.10.0 (QEMU 3.1)</span></td>
</tr>
</table>
</dd>
<dt><code>pvspinlock</code></dt>
<dd>Notify the guest that the host supports paravirtual spinlocks
for example by exposing the pvticketlocks mechanism. This feature
can be explicitly disabled by using <code>state='off'</code>
attribute.
</dd>
<dt><code>kvm</code></dt>
<dd>Various features to change the behavior of the KVM hypervisor.
<table class="top_table">
<tr>
<th>Feature</th>
<th>Description</th>
<th>Value</th>
<th>Since</th>
</tr>
<tr>
<td>hidden</td>
<td>Hide the KVM hypervisor from standard MSR based discovery</td>
<td>on, off</td>
<td><span class="since">1.2.8 (QEMU 2.1.0)</span></td>
</tr>
<tr>
<td>hint-dedicated</td>
<td>Allows a guest to enable optimizations when running on dedicated vCPUs</td>
<td>on, off</td>
<td><span class="since">5.7.0 (QEMU 2.12.0)</span></td>
</tr>
</table>
</dd>
<dt><code>pmu</code></dt>
<dd>Depending on the <code>state</code> attribute (values <code>on</code>,
<code>off</code>, default <code>on</code>) enable or disable the
performance monitoring unit for the guest.
<span class="since">Since 1.2.12</span>
</dd>
<dt><code>vmport</code></dt>
<dd>Depending on the <code>state</code> attribute (values <code>on</code>,
<code>off</code>, default <code>on</code>) enable or disable
the emulation of VMware IO port, for vmmouse etc.
<span class="since">Since 1.2.16</span>
</dd>
<dt><code>gic</code></dt>
<dd>Enable for architectures using a General Interrupt
Controller instead of APIC in order to handle interrupts.
For example, the 'aarch64' architecture uses
<code>gic</code> instead of <code>apic</code>. The optional
attribute <code>version</code> specifies the GIC version;
however, it may not be supported by all hypervisors. Accepted
values are <code>2</code>, <code>3</code> and <code>host</code>.
<span class="since">Since 1.2.16</span>
</dd>
<dt><code>smm</code></dt>
<dd>
<p>
Depending on the <code>state</code> attribute (values <code>on</code>,
<code>off</code>, default <code>on</code>) enable or disable
System Management Mode.
<span class="since">Since 2.1.0</span>
</p><p> Optional sub-element <code>tseg</code> can be used to specify
the amount of memory dedicated to SMM's extended TSEG. That offers a
fourth option size apart from the existing ones (1 MiB, 2 MiB and 8
MiB) that the guest OS (or rather loader) can choose from. The size
can be specified as a value of that element, optional attribute
<code>unit</code> can be used to specify the unit of the
aforementioned value (defaults to 'MiB'). If set to 0 the extended
size is not advertised and only the default ones (see above) are
available.
</p><p>
<b>If the VM is booting you should leave this option alone, unless you
are very certain you know what you are doing.</b>
</p><p>
This value is configurable due to the fact that the calculation cannot
be done right with the guarantee that it will work correctly. In
QEMU, the user-configurable extended TSEG feature was unavailable up
to and including <code>pc-q35-2.9</code>. Starting with
<code>pc-q35-2.10</code> the feature is available, with default size
16 MiB. That should suffice for up to roughly 272 vCPUs, 5 GiB guest
RAM in total, no hotplug memory range, and 32 GiB of 64-bit PCI MMIO
aperture. Or for 48 vCPUs, with 1TB of guest RAM, no hotplug DIMM
range, and 32GB of 64-bit PCI MMIO aperture. The values may also vary
based on the loader the VM is using.
</p><p>
Additional size might be needed for significantly higher vCPU counts
or increased address space (that can be memory, maxMemory, 64-bit PCI
MMIO aperture size; roughly 8 MiB of TSEG per 1 TiB of address space)
which can also be rounded up.
</p><p>
Due to the nature of this setting being similar to "how much RAM
should the guest have" users are advised to either consult the
documentation of the guest OS or loader (if there is any), or test
this by trial-and-error changing the value until the VM boots
successfully. Yet another guiding value for users might be the fact
that 48 MiB should be enough for pretty large guests (240 vCPUs and
4TB guest RAM), but it is on purpose not set as default as 48 MiB of
unavailable RAM might be too much for small guests (e.g. with 512 MiB
of RAM).
</p><p>
See <a href="#elementsMemoryAllocation">Memory Allocation</a>
for more details about the <code>unit</code> attribute.
<span class="since">Since 4.5.0</span> (QEMU only)
</p>
</dd>
<dt><code>ioapic</code></dt>
<dd>Tune the I/O APIC. Possible values for the
<code>driver</code> attribute are:
<code>kvm</code> (default for KVM domains)
and <code>qemu</code> which puts I/O APIC in userspace
which is also known as a split I/O APIC mode.
<span class="since">Since 3.4.0</span> (QEMU/KVM only)
</dd>
<dt><code>hpt</code></dt>
<dd>Configure the HPT (Hash Page Table) of a pSeries guest. Possible
values for the <code>resizing</code> attribute are
<code>enabled</code>, which causes HPT resizing to be enabled if
both the guest and the host support it; <code>disabled</code>, which
causes HPT resizing to be disabled regardless of guest and host
support; and <code>required</code>, which prevents the guest from
starting unless both the guest and the host support HPT resizing. If
the attribute is not defined, the hypervisor default will be used.
<span class="since">Since 3.10.0</span> (QEMU/KVM only).
<p>The optional <code>maxpagesize</code> subelement can be used to
limit the usable page size for HPT guests. Common values are 64 KiB,
16 MiB and 16 GiB; when not specified, the hypervisor default will
be used. <span class="since">Since 4.5.0</span> (QEMU/KVM only).</p>
</dd>
<dt><code>vmcoreinfo</code></dt>
<dd>Enable QEMU vmcoreinfo device to let the guest kernel save debug
details. <span class="since">Since 4.4.0</span> (QEMU only)
</dd>
<dt><code>htm</code></dt>
<dd>Configure HTM (Hardware Transational Memory) availability for
pSeries guests. Possible values for the <code>state</code> attribute
are <code>on</code> and <code>off</code>. If the attribute is not
defined, the hypervisor default will be used.
<span class="since">Since 4.6.0</span> (QEMU/KVM only)
</dd>
<dt><code>nested-hv</code></dt>
<dd>Configure nested HV availability for pSeries guests. This needs to
be enabled from the host (L0) in order to be effective; having HV
support in the (L1) guest is very desiderable if it's planned to
run nested (L2) guests inside it, because it will result in those
nested guests having much better performance than they would when
using KVM PR or TCG.
Possible values for the <code>state</code> attribute are
<code>on</code> and <code>off</code>. If the attribute is not
defined, the hypervisor default will be used.
<span class="since">Since 4.10.0</span> (QEMU/KVM only)
</dd>
<dt><code>msrs</code></dt>
<dd>Some guests might require ignoring unknown
Model Specific Registers (MSRs) reads and writes. It's possible
to switch this by setting <code>unknown</code> attribute
of <code>msrs</code> to <code>ignore</code>. If the attribute is
not defined, or set to <code>fault</code>, unknown reads and writes
will not be ignored.
<span class="since">Since 5.1.0</span> (bhyve only)
</dd>
<dt><code>ccf-assist</code></dt>
<dd>Configure ccf-assist (Count Cache Flush Assist) availability for
pSeries guests.
Possible values for the <code>state</code> attribute
are <code>on</code> and <code>off</code>. If the attribute is not
defined, the hypervisor default will be used.
<span class="since">Since 5.9.0</span> (QEMU/KVM only)
</dd>
</dl>
<h3><a id="elementsTime">Time keeping</a></h3>
<p>
The guest clock is typically initialized from the host clock.
Most operating systems expect the hardware clock to be kept
in UTC, and this is the default. Windows, however, expects
it to be in so called 'localtime'.
</p>
<pre>
...
&lt;clock offset='localtime'&gt;
&lt;timer name='rtc' tickpolicy='catchup' track='guest'&gt;
&lt;catchup threshold='123' slew='120' limit='10000'/&gt;
&lt;/timer&gt;
&lt;timer name='pit' tickpolicy='delay'/&gt;
&lt;/clock&gt;
...</pre>
<dl>
<dt><code>clock</code></dt>
<dd>
<p>The <code>offset</code> attribute takes four possible
values, allowing fine grained control over how the guest
clock is synchronized to the host. NB, not all hypervisors
support all modes.</p>
<dl>
<dt><code>utc</code></dt>
<dd>
The guest clock will always be synchronized to UTC when
booted.
<span class="since">Since 0.9.11</span> 'utc' mode can be converted
to 'variable' mode, which can be controlled by using the
<code>adjustment</code> attribute. If the value is 'reset', the
conversion is never done (not all hypervisors can
synchronize to UTC on each boot; use of 'reset' will cause
an error on those hypervisors). A numeric value
forces the conversion to 'variable' mode using the value as the
initial adjustment. The default <code>adjustment</code> is
hypervisor specific.
</dd>
<dt><code>localtime</code></dt>
<dd>
The guest clock will be synchronized to the host's configured
timezone when booted, if any.
<span class="since">Since 0.9.11,</span> the <code>adjustment</code>
attribute behaves the same as in 'utc' mode.
</dd>
<dt><code>timezone</code></dt>
<dd>
The guest clock will be synchronized to the requested timezone
using the <code>timezone</code> attribute.
<span class="since">Since 0.7.7</span>
</dd>
<dt><code>variable</code></dt>
<dd>
The guest clock will have an arbitrary offset applied
relative to UTC or localtime, depending on the <code>basis</code>
attribute. The delta relative to UTC (or localtime) is specified
in seconds, using the <code>adjustment</code> attribute.
The guest is free to adjust the RTC over time and expect
that it will be honored at next reboot. This is in
contrast to 'utc' and 'localtime' mode (with the optional
attribute adjustment='reset'), where the RTC adjustments are
lost at each reboot. <span class="since">Since 0.7.7</span>
<span class="since">Since 0.9.11</span> the <code>basis</code>
attribute can be either 'utc' (default) or 'localtime'.
</dd>
</dl>
<p>
A <code>clock</code> may have zero or more
<code>timer</code> sub-elements. <span class="since">Since
0.8.0</span>
</p>
</dd>
<dt><code>timer</code></dt>
<dd>
<p>
Each timer element requires a <code>name</code> attribute,
and has other optional attributes that depend on
the <code>name</code> specified. Various hypervisors
support different combinations of attributes.
</p>
<dl>
<dt><code>name</code></dt>
<dd>
The <code>name</code> attribute selects which timer is
being modified, and can be one of
"platform" (currently unsupported),
"hpet" (libxl, xen, qemu), "kvmclock" (qemu),
"pit" (qemu), "rtc" (qemu), "tsc" (libxl, qemu -
<span class="since">since 3.2.0</span>), "hypervclock"
(qemu - <span class="since">since 1.2.2</span>) or
"armvtimer" (qemu - <span class="since">since 6.1.0</span>).
The <code>hypervclock</code> timer adds support for the
reference time counter and the reference page for iTSC
feature for guests running the Microsoft Windows
operating system.
</dd>
<dt><code>track</code></dt>
<dd>
The <code>track</code> attribute specifies what the timer
tracks, and can be "boot", "guest", or "wall".
Only valid for <code>name="rtc"</code>
or <code>name="platform"</code>.
</dd>
<dt><code>tickpolicy</code></dt>
<dd>
<p>
The <code>tickpolicy</code> attribute determines what
happens when QEMU misses a deadline for injecting a
tick to the guest. This can happen, for example, because the
guest was paused.
</p>
<dl>
<dt><code>delay</code></dt>
<dd>Continue to deliver ticks at the normal rate. The guest OS
will not notice anything is amiss, as from its point of view
time will have continued to flow normally. The time in the
guest should now be behind the time in the host by exactly
the amount of time during which ticks have been missed.</dd>
<dt><code>catchup</code></dt>
<dd>Deliver ticks at a higher rate to catch up with the missed
ticks. The guest OS will not notice anything is amiss, as
from its point of view time will have continued to flow
normally. Once the timer has managed to catch up with all
the missing ticks, the time in the guest and in the host
should match.</dd>
<dt><code>merge</code></dt>
<dd>Merge the missed tick(s) into one tick and
inject. The guest time may be delayed, depending
on how the OS reacts to the merging of ticks</dd>
<dt><code>discard</code></dt>
<dd>Throw away the missed ticks and continue with future
injection normally. The guest OS will see the timer jump
ahead by a potentially quite significant amount all at once,
as if the intervening chunk of time had simply not existed;
needless to say, such a sudden jump can easily confuse a
guest OS which is not specifically prepared to deal with it.
Assuming the guest OS can deal correctly with the time jump,
the time in the guest and in the host should now match.</dd>
</dl>
<p>If the policy is "catchup", there can be further details in
the <code>catchup</code> sub-element.</p>
<dl>
<dt><code>catchup</code></dt>
<dd>
The <code>catchup</code> element has three optional
attributes, each a positive integer. The attributes
are <code>threshold</code>, <code>slew</code>,
and <code>limit</code>.
</dd>
</dl>
<p>
Note that hypervisors are not required to support all policies across all time sources
</p>
</dd>
<dt><code>frequency</code></dt>
<dd>
The <code>frequency</code> attribute is an unsigned
integer specifying the frequency at
which <code>name="tsc"</code> runs.
</dd>
<dt><code>mode</code></dt>
<dd>
The <code>mode</code> attribute controls how
the <code>name="tsc"</code> timer is managed, and can be
"auto", "native", "emulate", "paravirt", or "smpsafe".
Other timers are always emulated.
</dd>
<dt><code>present</code></dt>
<dd>
The <code>present</code> attribute can be "yes" or "no" to
specify whether a particular timer is available to the guest.
</dd>
</dl>
</dd>
</dl>
<h3><a id="elementsPerf">Performance monitoring events</a></h3>
<p>
Some platforms allow monitoring of performance of the virtual machine and
the code executed inside. To enable the performance monitoring events
you can either specify them in the <code>perf</code> element or enable
them via <code>virDomainSetPerfEvents</code> API. The performance values
are then retrieved using the virConnectGetAllDomainStats API.
<span class="since">Since 2.0.0</span>
</p>
<pre>
...
&lt;perf&gt;
&lt;event name='cmt' enabled='yes'/&gt;
&lt;event name='mbmt' enabled='no'/&gt;
&lt;event name='mbml' enabled='yes'/&gt;
&lt;event name='cpu_cycles' enabled='no'/&gt;
&lt;event name='instructions' enabled='yes'/&gt;
&lt;event name='cache_references' enabled='no'/&gt;
&lt;event name='cache_misses' enabled='no'/&gt;
&lt;event name='branch_instructions' enabled='no'/&gt;
&lt;event name='branch_misses' enabled='no'/&gt;
&lt;event name='bus_cycles' enabled='no'/&gt;
&lt;event name='stalled_cycles_frontend' enabled='no'/&gt;
&lt;event name='stalled_cycles_backend' enabled='no'/&gt;
&lt;event name='ref_cpu_cycles' enabled='no'/&gt;
&lt;event name='cpu_clock' enabled='no'/&gt;
&lt;event name='task_clock' enabled='no'/&gt;
&lt;event name='page_faults' enabled='no'/&gt;
&lt;event name='context_switches' enabled='no'/&gt;
&lt;event name='cpu_migrations' enabled='no'/&gt;
&lt;event name='page_faults_min' enabled='no'/&gt;
&lt;event name='page_faults_maj' enabled='no'/&gt;
&lt;event name='alignment_faults' enabled='no'/&gt;
&lt;event name='emulation_faults' enabled='no'/&gt;
&lt;/perf&gt;
...
</pre>
<table class="top_table">
<tr>
<th>event name</th>
<th>Description</th>
<th>stats parameter name</th>
</tr>
<tr>
<td><code>cmt</code></td>
<td>usage of l3 cache in bytes by applications running on the platform</td>
<td><code>perf.cmt</code></td>
</tr>
<tr>
<td><code>mbmt</code></td>
<td>total system bandwidth from one level of cache</td>
<td><code>perf.mbmt</code></td>
</tr>
<tr>
<td><code>mbml</code></td>
<td>bandwidth of memory traffic for a memory controller</td>
<td><code>perf.mbml</code></td>
</tr>
<tr>
<td><code>cpu_cycles</code></td>
<td>the count of CPU cycles (total/elapsed)</td>
<td><code>perf.cpu_cycles</code></td>
</tr>
<tr>
<td><code>instructions</code></td>
<td>the count of instructions by applications running on the platform</td>
<td><code>perf.instructions</code></td>
</tr>
<tr>
<td><code>cache_references</code></td>
<td>the count of cache hits by applications running on the platform</td>
<td><code>perf.cache_references</code></td>
</tr>
<tr>
<td><code>cache_misses</code></td>
<td>the count of cache misses by applications running on the platform</td>
<td><code>perf.cache_misses</code></td>
</tr>
<tr>
<td><code>branch_instructions</code></td>
<td>the count of branch instructions by applications running on the platform</td>
<td><code>perf.branch_instructions</code></td>
</tr>
<tr>
<td><code>branch_misses</code></td>
<td>the count of branch misses by applications running on the platform</td>
<td><code>perf.branch_misses</code></td>
</tr>
<tr>
<td><code>bus_cycles</code></td>
<td>the count of bus cycles by applications running on the platform</td>
<td><code>perf.bus_cycles</code></td>
</tr>
<tr>
<td><code>stalled_cycles_frontend</code></td>
<td>the count of stalled CPU cycles in the frontend of the instruction
processor pipeline by applications running on the platform</td>
<td><code>perf.stalled_cycles_frontend</code></td>
</tr>
<tr>
<td><code>stalled_cycles_backend</code></td>
<td>the count of stalled CPU cycles in the backend of the instruction
processor pipeline by applications running on the platform</td>
<td><code>perf.stalled_cycles_backend</code></td>
</tr>
<tr>
<td><code>ref_cpu_cycles</code></td>
<td>the count of total CPU cycles not affected by CPU frequency scaling
by applications running on the platform</td>
<td><code>perf.ref_cpu_cycles</code></td>
</tr>
<tr>
<td><code>cpu_clock</code></td>
<td>the count of CPU clock time, as measured by a monotonic
high-resolution per-CPU timer, by applications running on
the platform</td>
<td><code>perf.cpu_clock</code></td>
</tr>
<tr>
<td><code>task_clock</code></td>
<td>the count of task clock time, as measured by a monotonic
high-resolution CPU timer, specific to the task that
is run by applications running on the platform</td>
<td><code>perf.task_clock</code></td>
</tr>
<tr>
<td><code>page_faults</code></td>
<td>the count of page faults by applications running on the
platform. This includes minor, major, invalid and other
types of page faults</td>
<td><code>perf.page_faults</code></td>
</tr>
<tr>
<td><code>context_switches</code></td>
<td>the count of context switches by applications running on
the platform</td>
<td><code>perf.context_switches</code></td>
</tr>
<tr>
<td><code>cpu_migrations</code></td>
<td>the count of CPU migrations, that is, where the process
moved from one logical processor to another, by
applications running on the platform</td>
<td><code>perf.cpu_migrations</code></td>
</tr>
<tr>
<td><code>page_faults_min</code></td>
<td>the count of minor page faults, that is, where the
page was present in the page cache, and therefore
the fault avoided loading it from storage, by
applications running on the platform</td>
<td><code>perf.page_faults_min</code></td>
</tr>
<tr>
<td><code>page_faults_maj</code></td>
<td>the count of major page faults, that is, where the
page was not present in the page cache, and
therefore had to be fetched from storage, by
applications running on the platform</td>
<td><code>perf.page_faults_maj</code></td>
</tr>
<tr>
<td><code>alignment_faults</code></td>
<td>the count of alignment faults, that is when
the load or store is not aligned properly, by
applications running on the platform</td>
<td><code>perf.alignment_faults</code></td>
</tr>
<tr>
<td><code>emulation_faults</code></td>
<td>the count of emulation faults, that is when
the kernel traps on unimplemented instrucions
and emulates them for user space, by
applications running on the platform</td>
<td><code>perf.emulation_faults</code></td>
</tr>
</table>
<h3><a id="elementsDevices">Devices</a></h3>
<p>
The final set of XML elements are all used to describe devices
provided to the guest domain. All devices occur as children
of the main <code>devices</code> element.
<span class="since">Since 0.1.3</span>
</p>
<pre>
...
&lt;devices&gt;
&lt;emulator&gt;/usr/lib/xen/bin/qemu-dm&lt;/emulator&gt;
&lt;/devices&gt;
...</pre>
<dl>
<dt><a id="elementEmulator"><code>emulator</code></a></dt>
<dd>
The contents of the <code>emulator</code> element specify
the fully qualified path to the device model emulator binary.
The <a href="formatcaps.html">capabilities XML</a> specifies
the recommended default emulator to use for each particular
domain type / architecture combination.
</dd>
</dl>
<p>
To help users identifying devices they care about, every
device can have direct child <code>alias</code> element
which then has <code>name</code> attribute where users can
store identifier for the device. The identifier has to have
"ua-" prefix and must be unique within the domain. Additionally, the
identifier must consist only of the following characters:
<code>[a-zA-Z0-9_-]</code>.
<span class="since">Since 3.9.0</span>
</p>
<pre>
&lt;devices&gt;
&lt;disk type='file'&gt;
&lt;alias name='ua-myDisk'/&gt;
&lt;/disk&gt;
&lt;interface type='network' trustGuestRxFilters='yes'&gt;
&lt;alias name='ua-myNIC'/&gt;
&lt;/interface&gt;
...
&lt;/devices&gt;
</pre>
<h4><a id="elementsDisks">Hard drives, floppy disks, CDROMs</a></h4>
<p>
Any device that looks like a disk, be it a floppy, harddisk,
cdrom, or paravirtualized driver is specified via the <code>disk</code>
element.
</p>
<pre>
...
&lt;devices&gt;
&lt;disk type='file' snapshot='external'&gt;
&lt;driver name="tap" type="aio" cache="default"/&gt;
&lt;source file='/var/lib/xen/images/fv0' startupPolicy='optional'&gt;
&lt;seclabel relabel='no'/&gt;
&lt;/source&gt;
&lt;target dev='hda' bus='ide'/&gt;
&lt;iotune&gt;
&lt;total_bytes_sec&gt;10000000&lt;/total_bytes_sec&gt;
&lt;read_iops_sec&gt;400000&lt;/read_iops_sec&gt;
&lt;write_iops_sec&gt;100000&lt;/write_iops_sec&gt;
&lt;/iotune&gt;
&lt;boot order='2'/&gt;
&lt;encryption type='...'&gt;
...
&lt;/encryption&gt;
&lt;shareable/&gt;
&lt;serial&gt;
...
&lt;/serial&gt;
&lt;/disk&gt;
...
&lt;disk type='network'&gt;
&lt;driver name="qemu" type="raw" io="threads" ioeventfd="on" event_idx="off"/&gt;
&lt;source protocol="sheepdog" name="image_name"&gt;
&lt;host name="hostname" port="7000"/&gt;
&lt;/source&gt;
&lt;target dev="hdb" bus="ide"/&gt;
&lt;boot order='1'/&gt;
&lt;transient/&gt;
&lt;address type='drive' controller='0' bus='1' unit='0'/&gt;
&lt;/disk&gt;
&lt;disk type='network'&gt;
&lt;driver name="qemu" type="raw"/&gt;
&lt;source protocol="rbd" name="image_name2"&gt;
&lt;host name="hostname" port="7000"/&gt;
&lt;snapshot name="snapname"/&gt;
&lt;config file="/path/to/file"/&gt;
&lt;auth username='myuser'&gt;
&lt;secret type='ceph' usage='mypassid'/&gt;
&lt;/auth&gt;
&lt;/source&gt;
&lt;target dev="hdc" bus="ide"/&gt;
&lt;/disk&gt;
&lt;disk type='block' device='cdrom'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;target dev='hdd' bus='ide' tray='open'/&gt;
&lt;readonly/&gt;
&lt;/disk&gt;
&lt;disk type='network' device='cdrom'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source protocol="http" name="url_path"&gt;
&lt;host name="hostname" port="80"/&gt;
&lt;/source&gt;
&lt;target dev='hde' bus='ide' tray='open'/&gt;
&lt;readonly/&gt;
&lt;/disk&gt;
&lt;disk type='network' device='cdrom'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source protocol="https" name="url_path"&gt;
&lt;host name="hostname" port="443"/&gt;
&lt;/source&gt;
&lt;target dev='hdf' bus='ide' tray='open'/&gt;
&lt;readonly/&gt;
&lt;/disk&gt;
&lt;disk type='network' device='cdrom'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source protocol="ftp" name="url_path"&gt;
&lt;host name="hostname" port="21"/&gt;
&lt;/source&gt;
&lt;target dev='hdg' bus='ide' tray='open'/&gt;
&lt;readonly/&gt;
&lt;/disk&gt;
&lt;disk type='network' device='cdrom'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source protocol="ftps" name="url_path"&gt;
&lt;host name="hostname" port="990"/&gt;
&lt;/source&gt;
&lt;target dev='hdh' bus='ide' tray='open'/&gt;
&lt;readonly/&gt;
&lt;/disk&gt;
&lt;disk type='network' device='cdrom'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source protocol="tftp" name="url_path"&gt;
&lt;host name="hostname" port="69"/&gt;
&lt;/source&gt;
&lt;target dev='hdi' bus='ide' tray='open'/&gt;
&lt;readonly/&gt;
&lt;/disk&gt;
&lt;disk type='block' device='lun'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source dev='/dev/sda'&gt;
&lt;slices&gt;
&lt;slice type='storage' offset='12345' size='123'/&gt;
&lt;/slices&gt;
&lt;reservations managed='no'&gt;
&lt;source type='unix' path='/path/to/qemu-pr-helper' mode='client'/&gt;
&lt;/reservations&gt;
&lt;/source&gt;
&lt;target dev='sda' bus='scsi'/&gt;
&lt;address type='drive' controller='0' bus='0' target='3' unit='0'/&gt;
&lt;/disk&gt;
&lt;disk type='block' device='disk'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source dev='/dev/sda'/&gt;
&lt;geometry cyls='16383' heads='16' secs='63' trans='lba'/&gt;
&lt;blockio logical_block_size='512' physical_block_size='4096'/&gt;
&lt;target dev='hdj' bus='ide'/&gt;
&lt;/disk&gt;
&lt;disk type='volume' device='disk'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source pool='blk-pool0' volume='blk-pool0-vol0'/&gt;
&lt;target dev='hdk' bus='ide'/&gt;
&lt;/disk&gt;
&lt;disk type='network' device='disk'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source protocol='iscsi' name='iqn.2013-07.com.example:iscsi-nopool/2'&gt;
&lt;host name='example.com' port='3260'/&gt;
&lt;auth username='myuser'&gt;
&lt;secret type='iscsi' usage='libvirtiscsi'/&gt;
&lt;/auth&gt;
&lt;/source&gt;
&lt;target dev='vda' bus='virtio'/&gt;
&lt;/disk&gt;
&lt;disk type='network' device='lun'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source protocol='iscsi' name='iqn.2013-07.com.example:iscsi-nopool/1'&gt;
&lt;host name='example.com' port='3260'/&gt;
&lt;auth username='myuser'&gt;
&lt;secret type='iscsi' usage='libvirtiscsi'/&gt;
&lt;/auth&gt;
&lt;/source&gt;
&lt;target dev='sdb' bus='scsi'/&gt;
&lt;/disk&gt;
&lt;disk type='network' device='lun'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source protocol='iscsi' name='iqn.2013-07.com.example:iscsi-nopool/0'&gt;
&lt;host name='example.com' port='3260'/&gt;
&lt;initiator&gt;
&lt;iqn name='iqn.2013-07.com.example:client'/&gt;
&lt;/initiator&gt;
&lt;/source&gt;
&lt;target dev='sdb' bus='scsi'/&gt;
&lt;/disk&gt;
&lt;disk type='volume' device='disk'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source pool='iscsi-pool' volume='unit:0:0:1' mode='host'/&gt;
&lt;target dev='vdb' bus='virtio'/&gt;
&lt;/disk&gt;
&lt;disk type='volume' device='disk'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source pool='iscsi-pool' volume='unit:0:0:2' mode='direct'/&gt;
&lt;target dev='vdc' bus='virtio'/&gt;
&lt;/disk&gt;
&lt;disk type='file' device='disk'&gt;
&lt;driver name='qemu' type='qcow2' queues='4'/&gt;
&lt;source file='/var/lib/libvirt/images/domain.qcow'/&gt;
&lt;backingStore type='file'&gt;
&lt;format type='qcow2'/&gt;
&lt;source file='/var/lib/libvirt/images/snapshot.qcow'/&gt;
&lt;backingStore type='block'&gt;
&lt;format type='raw'/&gt;
&lt;source dev='/dev/mapper/base'/&gt;
&lt;backingStore/&gt;
&lt;/backingStore&gt;
&lt;/backingStore&gt;
&lt;target dev='vdd' bus='virtio'/&gt;
&lt;/disk&gt;
&lt;disk type='nvme' device='disk'&gt;
&lt;driver name='qemu' type='raw'/&gt;
&lt;source type='pci' managed='yes' namespace='1'&gt;
&lt;address domain='0x0000' bus='0x01' slot='0x00' function='0x0'/&gt;
&lt;/source&gt;
&lt;target dev='vde' bus='virtio'/&gt;
&lt;/disk&gt;
&lt;/devices&gt;
...</pre>
<dl>
<dt><code>disk</code></dt>
<dd>The <code>disk</code> element is the main container for
describing disks and supports the following attributes:
<dl>
<dt><code>type</code></dt>
<dd>
Valid values are "file", "block",
"dir" (<span class="since">since 0.7.5</span>),
"network" (<span class="since">since 0.8.7</span>), or
"volume" (<span class="since">since 1.0.5</span>), or
"nvme" (<span class="since">since 6.0.0</span>)
and refer to the underlying source for the disk.
<span class="since">Since 0.0.3</span>
</dd>
<dt><code>device</code></dt>
<dd>
Indicates how the disk is to be exposed to the guest OS. Possible
values for this attribute are "floppy", "disk", "cdrom", and "lun",
defaulting to "disk".
<p>
Using "lun" (<span class="since">since 0.9.10</span>) is only
valid when the <code>type</code> is "block" or "network" for
<code>protocol='iscsi'</code> or when the <code>type</code>
is "volume" when using an iSCSI source <code>pool</code>
for <code>mode</code> "host" or as an
<a href="http://wiki.libvirt.org/page/NPIV_in_libvirt">NPIV</a>
virtual Host Bus Adapter (vHBA) using a Fibre Channel storage pool.
Configured in this manner, the LUN behaves identically to "disk",
except that generic SCSI commands from the guest are accepted
and passed through to the physical device. Also note that
device='lun' will only be recognized for actual raw devices,
but never for individual partitions or LVM partitions (in those
cases, the kernel will reject the generic SCSI commands, making
it identical to device='disk').
<span class="since">Since 0.1.4</span>
</p>
</dd>
<dt><code>model</code></dt>
<dd>
Indicates the emulated device model of the disk. Typically
this is indicated solely by the <code>bus</code> property but
for <code>bus</code> "virtio" the model can be specified further
with "virtio-transitional", "virtio-non-transitional", or
"virtio". See
<a href="#elementsVirtioTransitional">Virtio transitional devices</a>
for more details.
<span class="since">Since 5.2.0</span>
</dd>
<dt><code>rawio</code></dt>
<dd>
Indicates whether the disk needs rawio capability. Valid
settings are "yes" or "no" (default is "no"). If any one disk
in a domain has rawio='yes', rawio capability will be enabled
for all disks in the domain (because, in the case of QEMU, this
capability can only be set on a per-process basis). This attribute
is only valid when device is "lun". NB, <code>rawio</code> intends
to confine the capability per-device, however, current QEMU
implementation gives the domain process broader capability
than that (per-process basis, affects all the domain disks).
To confine the capability as much as possible for QEMU driver
as this stage, <code>sgio</code> is recommended, it's more
secure than <code>rawio</code>.
<span class="since">Since 0.9.10</span>
</dd>
<dt><code>sgio</code></dt>
<dd>
If supported by the hypervisor and OS, indicates whether
unprivileged SG_IO commands are filtered for the disk. Valid
settings are "filtered" or "unfiltered" where the default is
"filtered". Only available when the <code>device</code> is 'lun'.
<span class="since">Since 1.0.2</span>
</dd>
<dt><code>snapshot</code></dt>
<dd>
Indicates the default behavior of the disk during disk snapshots:
"<code>internal</code>" requires a file format such as qcow2 that
can store both the snapshot and the data changes since the snapshot;
"<code>external</code>" will separate the snapshot from the live
data; and "<code>no</code>" means the disk will not participate in
snapshots. Read-only disks default to "<code>no</code>", while the
default for other disks depends on the hypervisor's capabilities.
Some hypervisors allow a per-snapshot choice as well, during
<a href="formatsnapshot.html">domain snapshot creation</a>.
Not all snapshot modes are supported; for example, enabling
snapshots with a transient disk generally does not make sense.
<span class="since">Since 0.9.5</span>
</dd>
</dl>
</dd>
<dt><code>source</code></dt>
<dd>Representation of the disk <code>source</code> depends on the
disk <code>type</code> attribute value as follows:
<dl>
<dt><code>file</code></dt>
<dd>
The <code>file</code> attribute specifies the fully-qualified
path to the file holding the disk.
<span class="since">Since 0.0.3</span>
</dd>
<dt><code>block</code></dt>
<dd>
The <code>dev</code> attribute specifies the fully-qualified path
to the host device to serve as the disk.
<span class="since">Since 0.0.3</span>
</dd>
<dt><code>dir</code></dt>
<dd>
The <code>dir</code> attribute specifies the fully-qualified path
to the directory to use as the disk.
<span class="since">Since 0.7.5</span>
</dd>
<dt><code>network</code></dt>
<dd>
The <code>protocol</code> attribute specifies the protocol to
access to the requested image. Possible values are "nbd",
"iscsi", "rbd", "sheepdog", "gluster" or "vxhs".
<p>If the <code>protocol</code> attribute is "rbd", "sheepdog",
"gluster", or "vxhs", an additional attribute <code>name</code>
is mandatory to specify which volume/image will be used.
</p>
<p>For "nbd", the <code>name</code> attribute is optional. TLS
transport for NBD can be enabled by setting the <code>tls</code>
attribute to <code>yes</code>. For the QEMU hypervisor, usage of
a TLS environment can also be globally controlled on the host by
the <code>nbd_tls</code> and <code>nbd_tls_x509_cert_dir</code> in
/etc/libvirt/qemu.conf.
('tls' <span class="since">Since 4.5.0</span>)
</p>
<p>For "iscsi" (<span class="since">since 1.0.4</span>), the
<code>name</code> attribute may include a logical unit number,
separated from the target's name by a slash (e.g.,
<code>iqn.2013-07.com.example:iscsi-pool/1</code>). If not
specified, the default LUN is zero.
</p>
<p>For "vxhs" (<span class="since">since 3.8.0</span>), the
<code>name</code> is the UUID of the volume, assigned by the
HyperScale server. Additionally, an optional attribute
<code>tls</code> (QEMU only) can be used to control whether a
VxHS block device would utilize a hypervisor configured TLS
X.509 certificate environment in order to encrypt the data
channel. For the QEMU hypervisor, usage of a TLS environment can
also be globally controlled on the host by the
<code>vxhs_tls</code> and <code>vxhs_tls_x509_cert_dir</code> or
<code>default_tls_x509_cert_dir</code> settings in the file
/etc/libvirt/qemu.conf. If <code>vxhs_tls</code> is enabled,
then unless the domain <code>tls</code> attribute is set to "no",
libvirt will use the host configured TLS environment. If the
<code>tls</code> attribute is set to "yes", then regardless of
the qemu.conf setting, TLS authentication will be attempted.
</p>
<span class="since">Since 0.8.7</span>
</dd>
<dt><code>volume</code></dt>
<dd>
The underlying disk source is represented by attributes
<code>pool</code> and <code>volume</code>. Attribute
<code>pool</code> specifies the name of the
<a href="formatstorage.html">storage pool</a> (managed
by libvirt) where the disk source resides. Attribute
<code>volume</code> specifies the name of storage volume (managed
by libvirt) used as the disk source. The value for the
<code>volume</code> attribute will be the output from the "Name"
column of a <code>virsh vol-list [pool-name]</code> command.
<p>
Use the attribute <code>mode</code>
(<span class="since">since 1.1.1</span>) to indicate how to
represent the LUN as the disk source. Valid values are
"direct" and "host". If <code>mode</code> is not specified,
the default is to use "host".
Using "direct" as the <code>mode</code> value indicates to use
the <a href="formatstorage.html">storage pool's</a>
<code>source</code> element <code>host</code> attribute as
the disk source to generate the libiscsi URI (e.g.
'file=iscsi://example.com:3260/iqn.2013-07.com.example:iscsi-pool/1').
Using "host" as the <code>mode</code> value indicates to use the
LUN's path as it shows up on host (e.g.
'file=/dev/disk/by-path/ip-example.com:3260-iscsi-iqn.2013-07.com.example:iscsi-pool-lun-1').
Using a LUN from an iSCSI source pool provides the same
features as a <code>disk</code> configured using
<code>type</code> 'block' or 'network' and <code>device</code>
of 'lun' with respect to how the LUN is presented to and
may be used by the guest.
<span class="since">Since 1.0.5</span>
</p>
</dd>
<dt><code>nvme</code></dt>
<dd>
To specify disk source for NVMe disk the <code>source</code>
element has the following attributes:
<dl>
<dt><code>type</code></dt>
<dd>The type of address specified in <code>address</code>
sub-element. Currently, only <code>pci</code> value is
accepted.
</dd>
<dt><code>managed</code></dt>
<dd>This attribute instructs libvirt to detach NVMe
controller automatically on domain startup (<code>yes</code>)
or expect the controller to be detached by system
administrator (<code>no</code>).
</dd>
<dt><code>namespace</code></dt>
<dd>The namespace ID which should be assigned to the domain.
According to NVMe standard, namespace numbers start from 1,
including.
</dd>
</dl>
The difference between <code>&lt;disk type='nvme'&gt;</code>
and <code>&lt;hostdev/&gt;</code> is that the latter is plain
host device assignment with all its limitations (e.g. no live
migration), while the former makes hypervisor to run the NVMe
disk through hypervisor's block layer thus enabling all
features provided by the layer (e.g. snapshots, domain
migration, etc.). Moreover, since the NVMe disk is unbinded
from its PCI driver, the host kernel storage stack is not
involved (compared to passing say <code>/dev/nvme0n1</code> via
<code>&lt;disk type='block'&gt;</code> and therefore lower
latencies can be achieved.
</dd>
</dl>
With "file", "block", and "volume", one or more optional
sub-elements <code>seclabel</code>, <a href="#seclabel">described
below</a> (and <span class="since">since 0.9.9</span>), can be
used to override the domain security labeling policy for just
that source file. (NB, for "volume" type disk, <code>seclabel</code>
is only valid when the specified storage volume is of 'file' or
'block' type).
<p>
The <code>source</code> element may also have the <code>index</code>
attribute with same semantics the <a href='#elementsDiskBackingStoreIndex'>
<code>index</code></a> attribute of <code>backingStore</code>
</p>
<p>
The <code>source</code> element may contain the following sub elements:
</p>
<dl>
<dt><code>host</code></dt>
<dd>
<p>
When the disk <code>type</code> is "network", the <code>source</code>
may have zero or more <code>host</code> sub-elements used to
specify the hosts to connect.
The <code>host</code> element supports 4 attributes, viz. "name",
"port", "transport" and "socket", which specify the hostname,
the port number, transport type and path to socket, respectively.
The meaning of this element and the number of the elements depend
on the protocol attribute.
</p>
<table class="top_table">
<tr>
<th> Protocol </th>
<th> Meaning </th>
<th> Number of hosts </th>
<th> Default port </th>
</tr>
<tr>
<td> nbd </td>
<td> a server running nbd-server </td>
<td> only one </td>
<td> 10809 </td>
</tr>
<tr>
<td> iscsi </td>
<td> an iSCSI server </td>
<td> only one </td>
<td> 3260 </td>
</tr>
<tr>
<td> rbd </td>
<td> monitor servers of RBD </td>
<td> one or more </td>
<td> librados default </td>
</tr>
<tr>
<td> sheepdog </td>
<td> one of the sheepdog servers (default is localhost:7000) </td>
<td> zero or one </td>
<td> 7000 </td>
</tr>
<tr>
<td> gluster </td>
<td> a server running glusterd daemon </td>
<td> one or more (<span class="since">Since 2.1.0</span>), just one prior to that </td>
<td> 24007 </td>
</tr>
<tr>
<td> vxhs </td>
<td> a server running Veritas HyperScale daemon </td>
<td> only one </td>
<td> 9999 </td>
</tr>
</table>
<p>
gluster supports "tcp", "rdma", "unix" as valid values for the
transport attribute. nbd supports "tcp" and "unix". Others only
support "tcp". If nothing is specified, "tcp" is assumed. If the
transport is "unix", the socket attribute specifies the path to an
AF_UNIX socket.
</p>
</dd>
<dt><code>snapshot</code></dt>
<dd>
The <code>name</code> attribute of <code>snapshot</code> element can
optionally specify an internal snapshot name to be used as the
source for storage protocols.
Supported for 'rbd' <span class="since">since 1.2.11 (QEMU only).</span>
</dd>
<dt><code>config</code></dt>
<dd>
The <code>file</code> attribute for the <code>config</code> element
provides a fully qualified path to a configuration file to be
provided as a parameter to the client of a networked storage
protocol. Supported for 'rbd' <span class="since">since 1.2.11
(QEMU only).</span>
</dd>
<dt><code>auth</code></dt>
<dd><span class="since">Since libvirt 3.9.0</span>, the
<code>auth</code> element is supported for a disk
<code>type</code> "network" that is using a <code>source</code>
element with the <code>protocol</code> attributes "rbd" or "iscsi".
If present, the <code>auth</code> element provides the
authentication credentials needed to access the source. It
includes a mandatory attribute <code>username</code>, which
identifies the username to use during authentication, as well
as a sub-element <code>secret</code> with mandatory
attribute <code>type</code>, to tie back to
a <a href="formatsecret.html">libvirt secret object</a> that
holds the actual password or other credentials (the domain XML
intentionally does not expose the password, only the reference
to the object that does manage the password).
Known secret types are "ceph" for Ceph RBD network sources and
"iscsi" for CHAP authentication of iSCSI targets.
Both will require either a <code>uuid</code> attribute
with the UUID of the secret object or a <code>usage</code>
attribute matching the key that was specified in the
secret object.
</dd>
<dt><code>encryption</code></dt>
<dd><span class="since">Since libvirt 3.9.0</span>, the
<code>encryption</code> can be a sub-element of the
<code>source</code> element for encrypted storage sources.
If present, specifies how the storage source is encrypted
See the
<a href="formatstorageencryption.html">Storage Encryption</a>
page for more information.
<p/>
Note that the 'qcow' format of encryption is broken and thus is no
longer supported for use with disk images.
(<span class="since">Since libvirt 4.5.0</span>)
</dd>
<dt><code>reservations</code></dt>
<dd><span class="since">Since libvirt 4.4.0</span>, the
<code>reservations</code> can be a sub-element of the
<code>source</code> element for storage sources (QEMU driver only).
If present it enables persistent reservations for SCSI
based disks. The element has one mandatory attribute
<code>managed</code> with accepted values <code>yes</code> and
<code>no</code>. If <code>managed</code> is enabled libvirt prepares
and manages any resources needed. When the persistent reservations
are unmanaged, then the hypervisor acts as a client and the path to
the server socket must be provided in the child element
<code>source</code>, which currently accepts only the following
attributes:
<code>type</code> with one value <code>unix</code>,
<code>path</code> path to the socket, and
finally <code>mode</code> which accepts one value
<code>client</code> specifying the role of hypervisor.
It's recommended to allow libvirt manage the persistent
reservations.
</dd>
<dt><code>initiator</code></dt>
<dd><span class="since">Since libvirt 4.7.0</span>, the
<code>initiator</code> element is supported for a disk
<code>type</code> "network" that is using a <code>source</code>
element with the <code>protocol</code> attribute "iscsi".
If present, the <code>initiator</code> element provides the
initiator IQN needed to access the source via mandatory
attribute <code>name</code>.
</dd>
<dt><code>address</code></dt>
<dd>For disk of type <code>nvme</code> this element
specifies the PCI address of the host NVMe
controller.
<span class="since">Since 6.0.0</span>
</dd>
<dt><code>slices</code></dt>
<dd>The <code>slices</code> element using its <code>slice</code>
sub-elements allows configuring offset and size of either the
location of the image format (<code>slice type='storage'</code>)
inside the storage source or the guest data inside the image format
container (future expansion).
The <code>offset</code> and <code>size</code> values are in bytes.
<span class="since">Since 6.1.0</span>
</dd>
</dl>
<p>
For a "file" or "volume" disk type which represents a cdrom or floppy
(the <code>device</code> attribute), it is possible to define
policy what to do with the disk if the source file is not accessible.
(NB, <code>startupPolicy</code> is not valid for "volume" disk unless
the specified storage volume is of "file" type). This is done by the
<code>startupPolicy</code> attribute
(<span class="since">since 0.9.7</span>),
accepting these values:
</p>
<table class="top_table">
<tr>
<td> mandatory </td>
<td> fail if missing for any reason (the default) </td>
</tr>
<tr>
<td> requisite </td>
<td> fail if missing on boot up,
drop if missing on migrate/restore/revert </td>
</tr>
<tr>
<td> optional </td>
<td> drop if missing at any start attempt </td>
</tr>
</table>
<p>
<span class="since">Since 1.1.2</span> the <code>startupPolicy</code>
is extended to support hard disks besides cdrom and floppy. On guest
cold bootup, if a certain disk is not accessible or its disk chain is
broken, with startupPolicy 'optional' the guest will drop this disk.
This feature doesn't support migration currently.
</p>
</dd>
<dt><code>backingStore</code></dt>
<dd>
This element describes the backing store used by the disk
specified by sibling <code>source</code> element.
<span class="since">Since 1.2.4.</span>
If the hypervisor driver does not support the
<a href='formatdomaincaps.html#featureBackingStoreInput'>
<code>backingStoreInput</code></a>
(<span class='since'>Since 5.10.0</span>)
domain feature the <code>backingStore</code> is ignored on
input and only used for output to describe the detected
backing chains of running domains.
If <code>backingStoreInput</code> is supported
the <code>backingStore</code> is used as the backing image of
<code>source</code> or other <code>backingStore</code> overriding
any backing image information recorded in the image metadata.
An empty <code>backingStore</code> element means the sibling
source is self-contained and is not based on any backing store.
For the detected backing chain information to be accurate, the
backing format must be correctly specified in the metadata of
each file of the chain (files created by libvirt satisfy this
property, but using existing external files for snapshot or
block copy operations requires the end user to pre-create the
file correctly). The following attributes are
supported in <code>backingStore</code>:
<dl>
<dt><code>type</code></dt>
<dd>
The <code>type</code> attribute represents the type of disk used
by the backing store, see disk type attribute above for more
details and possible values.
</dd>
<dt><code><a id="elementsDiskBackingStoreIndex">index</a></code></dt>
<dd>
This attribute is only valid in output (and ignored on input) and
it can be used to refer to a specific part of the disk chain when
doing block operations (such as via the
<code>virDomainBlockRebase</code> API). For example,
<code>vda[2]</code> refers to the backing store with
<code>index='2'</code> of the disk with <code>vda</code> target.
</dd>
</dl>
Moreover, <code>backingStore</code> supports the following sub-elements:
<dl>
<dt><code>format</code></dt>
<dd>
The <code>format</code> element contains <code>type</code>
attribute which specifies the internal format of the backing
store, such as <code>raw</code> or <code>qcow2</code>.
</dd>
<dt><code>source</code></dt>
<dd>
This element has the same structure as the <code>source</code>
element in <code>disk</code>. It specifies which file, device,
or network location contains the data of the described backing
store.
</dd>
<dt><code>backingStore</code></dt>
<dd>
If the backing store is not self-contained, the next element
in the chain is described by nested <code>backingStore</code>
element.
</dd>
</dl>
</dd>
<dt><code>mirror</code></dt>
<dd>
This element is present if the hypervisor has started a
long-running block job operation, where the mirror location in
the <code>source</code> sub-element will eventually have the
same contents as the source, and with the file format in the
sub-element <code>format</code> (which might differ from the
format of the source). The details of the <code>source</code>
sub-element are determined by the <code>type</code> attribute
of the mirror, similar to what is done for the
overall <code>disk</code> device element. The <code>job</code>
attribute mentions which API started the operation ("copy" for
the <code>virDomainBlockRebase</code> API, or "active-commit"
for the <code>virDomainBlockCommit</code>
API), <span class="since">since 1.2.7</span>. The
attribute <code>ready</code>, if present, tracks progress of
the job: <code>yes</code> if the disk is known to be ready to
pivot, or, <span class="since">since
1.2.7</span>, <code>abort</code> or <code>pivot</code> if the
job is in the process of completing. If <code>ready</code> is
not present, the disk is probably still
copying. For now, this element only valid in output; it is
ignored on input. The <code>source</code> sub-element exists
for all two-phase jobs <span class="since">since 1.2.6</span>.
Older libvirt supported only block copy to a
file, <span class="since">since 0.9.12</span>; for
compatibility with older clients, such jobs include redundant
information in the attributes <code>file</code>
and <code>format</code> in the <code>mirror</code> element.
</dd>
<dt><code>target</code></dt>
<dd>The <code>target</code> element controls the bus / device
under which the disk is exposed to the guest
OS. The <code>dev</code> attribute indicates the "logical"
device name. The actual device name specified is not
guaranteed to map to the device name in the guest OS. Treat it
as a device ordering hint. The optional <code>bus</code>
attribute specifies the type of disk device to emulate;
possible values are driver specific, with typical values being
"ide", "scsi", "virtio", "xen", "usb", "sata", or
"sd" <span class="since">"sd" since 1.1.2</span>. If omitted, the bus
type is inferred from the style of the device name (e.g. a device named
'sda' will typically be exported using a SCSI bus). The optional
attribute <code>tray</code> indicates the tray status of the
removable disks (i.e. CDROM or Floppy disk), the value can be either
"open" or "closed", defaults to "closed". NB, the value of
<code>tray</code> could be updated while the domain is running.
The optional attribute <code>removable</code> sets the
removable flag for USB disks, and its value can be either "on"
or "off", defaulting to "off". <span class="since">Since
0.0.3; <code>bus</code> attribute since 0.4.3;
<code>tray</code> attribute since 0.9.11; "usb" attribute value since
after 0.4.4; "sata" attribute value since 0.9.7; "removable" attribute
value since 1.1.3</span>
</dd>
<dt><code>iotune</code></dt>
<dd>The optional <code>iotune</code> element provides the
ability to provide additional per-device I/O tuning, with
values that can vary for each device (contrast this to
the <a href="#elementsBlockTuning"><code>&lt;blkiotune&gt;</code></a>
element, which applies globally to the domain). Currently,
the only tuning available is Block I/O throttling for qemu.
This element has optional sub-elements; any sub-element not
specified or given with a value of 0 implies no
limit. <span class="since">Since 0.9.8</span>
<dl>
<dt><code>total_bytes_sec</code></dt>
<dd>The optional <code>total_bytes_sec</code> element is the
total throughput limit in bytes per second. This cannot
appear with <code>read_bytes_sec</code>
or <code>write_bytes_sec</code>.</dd>
<dt><code>read_bytes_sec</code></dt>
<dd>The optional <code>read_bytes_sec</code> element is the
read throughput limit in bytes per second.</dd>
<dt><code>write_bytes_sec</code></dt>
<dd>The optional <code>write_bytes_sec</code> element is the
write throughput limit in bytes per second.</dd>
<dt><code>total_iops_sec</code></dt>
<dd>The optional <code>total_iops_sec</code> element is the
total I/O operations per second. This cannot
appear with <code>read_iops_sec</code>
or <code>write_iops_sec</code>.</dd>
<dt><code>read_iops_sec</code></dt>
<dd>The optional <code>read_iops_sec</code> element is the
read I/O operations per second.</dd>
<dt><code>write_iops_sec</code></dt>
<dd>The optional <code>write_iops_sec</code> element is the
write I/O operations per second.</dd>
<dt><code>total_bytes_sec_max</code></dt>
<dd>The optional <code>total_bytes_sec_max</code> element is the
maximum total throughput limit in bytes per second. This cannot
appear with <code>read_bytes_sec_max</code>
or <code>write_bytes_sec_max</code>.</dd>
<dt><code>read_bytes_sec_max</code></dt>
<dd>The optional <code>read_bytes_sec_max</code> element is the
maximum read throughput limit in bytes per second.</dd>
<dt><code>write_bytes_sec_max</code></dt>
<dd>The optional <code>write_bytes_sec_max</code> element is the
maximum write throughput limit in bytes per second.</dd>
<dt><code>total_iops_sec_max</code></dt>
<dd>The optional <code>total_iops_sec_max</code> element is the
maximum total I/O operations per second. This cannot
appear with <code>read_iops_sec_max</code>
or <code>write_iops_sec_max</code>.</dd>
<dt><code>read_iops_sec_max</code></dt>
<dd>The optional <code>read_iops_sec_max</code> element is the
maximum read I/O operations per second.</dd>
<dt><code>write_iops_sec_max</code></dt>
<dd>The optional <code>write_iops_sec_max</code> element is the
maximum write I/O operations per second.</dd>
<dt><code>size_iops_sec</code></dt>
<dd>The optional <code>size_iops_sec</code> element is the
size of I/O operations per second.
<p>
<span class="since">Throughput limits since 1.2.11 and QEMU 1.7</span>
</p>
</dd>
<dt><code>group_name</code></dt>
<dd>The optional <code>group_name</code> provides the cability
to share I/O throttling quota between multiple drives. This
prevents end-users from circumventing a hosting provider's
throttling policy by splitting 1 large drive in N small drives
and getting N times the normal throttling quota. Any name may
be used.
<p>
<span class="since">group_name since 3.0.0 and QEMU 2.4</span>
</p>
</dd>
<dt><code>total_bytes_sec_max_length</code></dt>
<dd>The optional <code>total_bytes_sec_max_length</code>
element is the maximum duration in seconds for the
<code>total_bytes_sec_max</code> burst period. Only valid
when the <code>total_bytes_sec_max</code> is set.</dd>
<dt><code>read_bytes_sec_max_length</code></dt>
<dd>The optional <code>read_bytes_sec_max_length</code>
element is the maximum duration in seconds for the
<code>read_bytes_sec_max</code> burst period. Only valid
when the <code>read_bytes_sec_max</code> is set.</dd>
<dt><code>write_bytes_sec_max</code></dt>
<dd>The optional <code>write_bytes_sec_max_length</code>
element is the maximum duration in seconds for the
<code>write_bytes_sec_max</code> burst period. Only valid
when the <code>write_bytes_sec_max</code> is set.</dd>
<dt><code>total_iops_sec_max_length</code></dt>
<dd>The optional <code>total_iops_sec_max_length</code>
element is the maximum duration in seconds for the
<code>total_iops_sec_max</code> burst period. Only valid
when the <code>total_iops_sec_max</code> is set.</dd>
<dt><code>read_iops_sec_max_length</code></dt>
<dd>The optional <code>read_iops_sec_max_length</code>
element is the maximum duration in seconds for the
<code>read_iops_sec_max</code> burst period. Only valid
when the <code>read_iops_sec_max</code> is set.</dd>
<dt><code>write_iops_sec_max</code></dt>
<dd>The optional <code>write_iops_sec_max_length</code>
element is the maximum duration in seconds for the
<code>write_iops_sec_max</code> burst period. Only valid
when the <code>write_iops_sec_max</code> is set.
<p>
<span class="since">Throughput length since 2.4.0 and QEMU 2.6</span>
</p>
</dd>
</dl>
</dd>
<dt><code>driver</code></dt>
<dd>
The optional driver element allows specifying further details
related to the hypervisor driver used to provide the disk.
<span class="since">Since 0.1.8</span>
<ul>
<li>
If the hypervisor supports multiple backend drivers, then
the <code>name</code> attribute selects the primary
backend driver name, while the optional <code>type</code>
attribute provides the sub-type. For example, xen
supports a name of "tap", "tap2", "phy", or "file", with a
type of "aio", while qemu only supports a name of "qemu",
but multiple types including "raw", "bochs", "qcow2", and
"qed".
</li>
<li>
The optional <code>cache</code> attribute controls the
cache mechanism, possible values are "default", "none",
"writethrough", "writeback", "directsync" (like
"writethrough", but it bypasses the host page cache) and
"unsafe" (host may cache all disk io, and sync requests from
guest are ignored).
<span class="since">
Since 0.6.0,
"directsync" since 0.9.5,
"unsafe" since 0.9.7
</span>
</li>
<li>
The optional <code>error_policy</code> attribute controls
how the hypervisor will behave on a disk read or write
error, possible values are "stop", "report", "ignore", and
"enospace".<span class="since">Since 0.8.0, "report" since
0.9.7</span> The default is left to the discretion of the
hypervisor. There is also an
optional <code>rerror_policy</code> that controls behavior
for read errors only. <span class="since">Since
0.9.7</span>. If no rerror_policy is given, error_policy
is used for both read and write errors. If rerror_policy
is given, it overrides the <code>error_policy</code> for
read errors. Also note that "enospace" is not a valid
policy for read errors, so if <code>error_policy</code> is
set to "enospace" and no <code>rerror_policy</code> is
given, the read error policy will be left at its default.
</li>
<li>
The optional <code>io</code> attribute controls specific
policies on I/O; qemu guests support "threads" and
"native". <span class="since">Since 0.8.8</span>
</li>
<li>
The optional <code>ioeventfd</code> attribute allows users to
set <a href='https://patchwork.kernel.org/patch/43390/'>
domain I/O asynchronous handling</a> for disk device.
The default is left to the discretion of the hypervisor.
Accepted values are "on" and "off". Enabling this allows
qemu to execute VM while a separate thread handles I/O.
Typically guests experiencing high system CPU utilization
during I/O will benefit from this. On the other hand,
on overloaded host it could increase guest I/O latency.
<span class="since">Since 0.9.3 (QEMU and KVM only)</span>
<b>In general you should leave this option alone, unless you
are very certain you know what you are doing.</b>
</li>
<li>
The optional <code>event_idx</code> attribute controls
some aspects of device event processing. The value can be
either 'on' or 'off' - if it is on, it will reduce the
number of interrupts and exits for the guest. The default
is determined by QEMU; usually if the feature is
supported, default is on. In case there is a situation
where this behavior is suboptimal, this attribute provides
a way to force the feature off.
<span class="since">Since 0.9.5 (QEMU and KVM only)</span>
<b>In general you should leave this option alone, unless you
are very certain you know what you are doing.</b>
</li>
<li>
The optional <code>copy_on_read</code> attribute controls
whether to copy read backing file into the image file. The
value can be either "on" or "off".
Copy-on-read avoids accessing the same backing file sectors
repeatedly and is useful when the backing file is over a slow
network. By default copy-on-read is off.
<span class='since'>Since 0.9.10 (QEMU and KVM only)</span>
</li>
<li>
The optional <code>discard</code> attribute controls whether
discard requests (also known as "trim" or "unmap") are
ignored or passed to the filesystem. The value can be either
"unmap" (allow the discard request to be passed) or "ignore"
(ignore the discard request).
<span class='since'>Since 1.0.6 (QEMU and KVM only)</span>
</li>
<li>
The optional <code>detect_zeroes</code> attribute controls whether
to detect zero write requests. The value can be "off", "on" or
"unmap". First two values turn the detection off and on,
respectively. The third value ("unmap") turns the detection on
and additionally tries to discard such areas from the image based
on the value of <code>discard</code> above (it will act as "on"
if <code>discard</code> is set to "ignore"). NB enabling the
detection is a compute intensive operation, but can save file
space and/or time on slow media.
<span class='since'>Since 2.0.0</span>
</li>
<li>
The optional <code>iothread</code> attribute assigns the
disk to an IOThread as defined by the range for the domain
<a href="#elementsIOThreadsAllocation"><code>iothreads</code></a>
value. Multiple disks may be assigned to the same IOThread and
are numbered from 1 to the domain iothreads value. Available
for a disk device <code>target</code> configured to use "virtio"
<code>bus</code> and "pci" or "ccw" <code>address</code> types.
<span class='since'>Since 1.2.8 (QEMU 2.1)</span>
</li>
<li>
The optional <code>queues</code> attribute specifies the number of
virt queues for virtio-blk. (<span class="since">Since 3.9.0</span>)
</li>
<li>
For virtio disks,
<a href="#elementsVirtio">Virtio-specific options</a> can also be
set. (<span class="since">Since 3.5.0</span>)
</li>
</ul>
</dd>
<dt><code>backenddomain</code></dt>
<dd>The optional <code>backenddomain</code> element allows specifying a
backend domain (aka driver domain) hosting the disk. Use the
<code>name</code> attribute to specify the backend domain name.
<span class="since">Since 1.2.13 (Xen only)</span>
</dd>
<dt><code>boot</code></dt>
<dd>Specifies that the disk is bootable. The <code>order</code>
attribute determines the order in which devices will be tried during
boot sequence. On the S390 architecture only the first boot device is
used. The optional <code>loadparm</code> attribute is an 8 character
string which can be queried by guests on S390 via sclp or diag 308.
Linux guests on S390 can use <code>loadparm</code> to select a boot
entry. <span class="since">Since 3.5.0</span>
The per-device <code>boot</code> elements cannot be used together
with general boot elements in
<a href="#elementsOSBIOS">BIOS bootloader</a> section.
<span class="since">Since 0.8.8</span>
</dd>
<dt><code>encryption</code></dt>
<dd>Starting with <span class="since">libvirt 3.9.0</span> the
<code>encryption</code> element is preferred to be a sub-element
of the <code>source</code> element. If present, specifies how the
volume is encrypted using "qcow". See the
<a href="formatstorageencryption.html">Storage Encryption</a> page
for more information.
</dd>
<dt><code>readonly</code></dt>
<dd>If present, this indicates the device cannot be modified by
the guest. For now, this is the default for disks with
attribute <code>device='cdrom'</code>.
</dd>
<dt><code>shareable</code></dt>
<dd>If present, this indicates the device is expected to be shared
between domains (assuming the hypervisor and OS support this),
which means that caching should be deactivated for that device.
</dd>
<dt><code>transient</code></dt>
<dd>If present, this indicates that changes to the device
contents should be reverted automatically when the guest
exits. With some hypervisors, marking a disk transient
prevents the domain from participating in migration or
snapshots. <span class="since">Since 0.9.5</span>
</dd>
<dt><code>serial</code></dt>
<dd>If present, this specify serial number of virtual hard drive.
For example, it may look
like <code>&lt;serial&gt;WD-WMAP9A966149&lt;/serial&gt;</code>.
Not supported for scsi-block devices, that is those using
disk <code>type</code> 'block' using <code>device</code> 'lun'
on <code>bus</code> 'scsi'.
<span class="since">Since 0.7.1</span>
</dd>
<dt><code>wwn</code></dt>
<dd>If present, this element specifies the WWN (World Wide Name)
of a virtual hard disk or CD-ROM drive. It must be composed
of 16 hexadecimal digits.
<span class='since'>Since 0.10.1</span>
</dd>
<dt><code>vendor</code></dt>
<dd>If present, this element specifies the vendor of a virtual hard
disk or CD-ROM device. It must not be longer than 8 printable
characters.
<span class='since'>Since 1.0.1</span>
</dd>
<dt><code>product</code></dt>
<dd>If present, this element specifies the product of a virtual hard
disk or CD-ROM device. It must not be longer than 16 printable
characters.
<span class='since'>Since 1.0.1</span>
</dd>
<dt><code>address</code></dt>
<dd>If present, the <code>address</code> element ties the disk
to a given slot of a controller (the
actual <code>&lt;controller&gt;</code> device can often be
inferred by libvirt, although it can
be <a href="#elementsControllers">explicitly specified</a>).
The <code>type</code> attribute is mandatory, and is typically
"pci" or "drive". For a "pci" controller, additional
attributes for <code>bus</code>, <code>slot</code>,
and <code>function</code> must be present, as well as
optional <code>domain</code> and <code>multifunction</code>.
Multifunction defaults to 'off'; any other value requires
QEMU 0.1.3 and <span class="since">libvirt 0.9.7</span>. For a
"drive" controller, additional attributes
<code>controller</code>, <code>bus</code>, <code>target</code>
(<span class="since">libvirt 0.9.11</span>), and <code>unit</code>
are available, each defaulting to 0.
</dd>
<dt><code>auth</code></dt>
<dd>Starting with <span class="since">libvirt 3.9.0</span> the
<code>auth</code> element is preferred to be a sub-element of
the <code>source</code> element. The element is still read and
managed as a <code>disk</code> sub-element. It is invalid to use
<code>auth</code> as both a sub-element of <code>disk</code>
and <code>source</code>. The <code>auth</code> element was
introduced as a <code>disk</code> sub-element in
<span class="since">libvirt 0.9.7.</span>
</dd>
<dt><code>geometry</code></dt>
<dd>The optional <code>geometry</code> element provides the
ability to override geometry settings. This mostly useful for
S390 DASD-disks or older DOS-disks. <span class="since">0.10.0</span>
<dl>
<dt><code>cyls</code></dt>
<dd>The <code>cyls</code> attribute is the
number of cylinders. </dd>
<dt><code>heads</code></dt>
<dd>The <code>heads</code> attribute is the
number of heads. </dd>
<dt><code>secs</code></dt>
<dd>The <code>secs</code> attribute is the
number of sectors per track. </dd>
<dt><code>trans</code></dt>
<dd>The optional <code>trans</code> attribute is the
BIOS-Translation-Modus (none, lba or auto)</dd>
</dl>
</dd>
<dt><code>blockio</code></dt>
<dd>If present, the <code>blockio</code> element allows
to override any of the block device properties listed below.
<span class="since">Since 0.10.2 (QEMU and KVM)</span>
<dl>
<dt><code>logical_block_size</code></dt>
<dd>The logical block size the disk will report to the guest
OS. For Linux this would be the value returned by the
BLKSSZGET ioctl and describes the smallest units for disk
I/O.
</dd>
<dt><code>physical_block_size</code></dt>
<dd>The physical block size the disk will report to the guest
OS. For Linux this would be the value returned by the
BLKPBSZGET ioctl and describes the disk's hardware sector
size which can be relevant for the alignment of disk data.
</dd>
</dl>
</dd>
</dl>
<h4><a id="elementsFilesystems">Filesystems</a></h4>
<p>
A directory on the host that can be accessed directly from the guest.
<span class="since">since 0.3.3, since 0.8.5 for QEMU/KVM</span>
</p>
<pre>
...
&lt;devices&gt;
&lt;filesystem type='template'&gt;
&lt;source name='my-vm-template'/&gt;
&lt;target dir='/'/&gt;
&lt;/filesystem&gt;
&lt;filesystem type='mount' accessmode='passthrough'&gt;
&lt;driver type='path' wrpolicy='immediate'/&gt;
&lt;source dir='/export/to/guest'/&gt;
&lt;target dir='/import/from/host'/&gt;
&lt;readonly/&gt;
&lt;/filesystem&gt;
&lt;filesystem type='file' accessmode='passthrough'&gt;
&lt;driver type='loop' format='raw'/&gt;
&lt;driver type='path' wrpolicy='immediate'/&gt;
&lt;source file='/export/to/guest.img'/&gt;
&lt;target dir='/import/from/host'/&gt;
&lt;readonly/&gt;
&lt;/filesystem&gt;
...
&lt;/devices&gt;
...</pre>
<dl>
<dt><code>filesystem</code></dt>
<dd>
The filesystem attribute <code>type</code> specifies the type of the
<code>source</code>. The possible values are:
<dl>
<dt><code>mount</code></dt>
<dd>
A host directory to mount in the guest. Used by LXC,
OpenVZ <span class="since">(since 0.6.2)</span>
and QEMU/KVM <span class="since">(since 0.8.5)</span>.
This is the default <code>type</code> if one is not specified.
This mode also has an optional
sub-element <code>driver</code>, with an
attribute <code>type='path'</code>
or <code>type='handle'</code> <span class="since">(since
0.9.7)</span>. The driver block has an optional attribute
<code>wrpolicy</code> that further controls interaction with
the host page cache; omitting the attribute gives default behavior,
while the value <code>immediate</code> means that a host writeback
is immediately triggered for all pages touched during a guest file
write operation <span class="since">(since 0.9.10)</span>.
</dd>
<dt><code>template</code></dt>
<dd>
OpenVZ filesystem template. Only used by OpenVZ driver.
</dd>
<dt><code>file</code></dt>
<dd>
A host file will be treated as an image and mounted in
the guest. The filesystem format will be autodetected.
Only used by LXC driver.
</dd>
<dt><code>block</code></dt>
<dd>
A host block device to mount in the guest. The filesystem
format will be autodetected. Only used by LXC driver
<span class="since">(since 0.9.5)</span>.
</dd>
<dt><code>ram</code></dt>
<dd>
An in-memory filesystem, using memory from the host OS.
The source element has a single attribute <code>usage</code>
which gives the memory usage limit in KiB, unless units
are specified by the <code>units</code> attribute. Only used
by LXC driver.
<span class="since"> (since 0.9.13)</span></dd>
<dt><code>bind</code></dt>
<dd>
A directory inside the guest will be bound to another
directory inside the guest. Only used by LXC driver
<span class="since"> (since 0.9.13)</span></dd>
</dl>
The filesystem element has an optional attribute <code>accessmode</code>
which specifies the security mode for accessing the source
<span class="since">(since 0.8.5)</span>. Currently this only works
with <code>type='mount'</code> for the QEMU/KVM driver. The possible
values are:
<dl>
<dt><code>passthrough</code></dt>
<dd>
The <code>source</code> is accessed with the permissions of the
user inside the guest. This is the default <code>accessmode</code> if
one is not specified.
<a href="http://lists.gnu.org/archive/html/qemu-devel/2010-05/msg02673.html">More info</a>
</dd>
<dt><code>mapped</code></dt>
<dd>
The <code>source</code> is accessed with the permissions of the
hypervisor (QEMU process).
<a href="http://lists.gnu.org/archive/html/qemu-devel/2010-05/msg02673.html">More info</a>
</dd>
<dt><code>squash</code></dt>
<dd>
Similar to 'passthrough', the exception is that failure of
privileged operations like 'chown' are ignored. This makes a
passthrough-like mode usable for people who run the hypervisor
as non-root.
<a href="http://lists.gnu.org/archive/html/qemu-devel/2010-09/msg00121.html">More info</a>
</dd>
</dl>
<span class="since">Since 5.2.0</span>, the filesystem element
has an optional attribute <code>model</code> with supported values
"virtio-transitional", "virtio-non-transitional", or "virtio".
See <a href="#elementsVirtioTransitional">Virtio transitional devices</a>
for more details.
</dd>
<dt><code>driver</code></dt>
<dd>
The optional driver element allows specifying further details
related to the hypervisor driver used to provide the filesystem.
<span class="since">Since 1.0.6</span>
<ul>
<li>
If the hypervisor supports multiple backend drivers, then
the <code>type</code> attribute selects the primary
backend driver name, while the <code>format</code>
attribute provides the format type. For example, LXC
supports a type of "loop", with a format of "raw" or
"nbd" with any format. QEMU supports a type of "path"
or "handle", but no formats. Virtuozzo driver supports
a type of "ploop" with a format of "ploop".
</li>
<li>
For virtio-backed devices,
<a href="#elementsVirtio">Virtio-specific options</a> can also be
set. (<span class="since">Since 3.5.0</span>)
</li>
</ul>
</dd>
<dt><code>source</code></dt>
<dd>
The resource on the host that is being accessed in the guest. The
<code>name</code> attribute must be used with
<code>type='template'</code>, and the <code>dir</code> attribute must
be used with <code>type='mount'</code>. The <code>usage</code> attribute
is used with <code>type='ram'</code> to set the memory limit in KiB,
unless units are specified by the <code>units</code> attribute.
</dd>
<dt><code>target</code></dt>
<dd>
Where the <code>source</code> can be accessed in the guest. For
most drivers this is an automatic mount point, but for QEMU/KVM
this is merely an arbitrary string tag that is exported to the
guest as a hint for where to mount.
</dd>
<dt><code>readonly</code></dt>
<dd>
Enables exporting filesystem as a readonly mount for guest, by
default read-write access is given (currently only works for
QEMU/KVM driver).
</dd>
<dt><code>space_hard_limit</code></dt>
<dd>
Maximum space available to this guest's filesystem.
<span class="since">Since 0.9.13</span>
</dd>
<dt><code>space_soft_limit</code></dt>
<dd>
Maximum space available to this guest's filesystem. The container is
permitted to exceed its soft limits for a grace period of time. Afterwards the
hard limit is enforced.
<span class="since">Since 0.9.13</span>
</dd>
</dl>
<h4><a id="elementsAddress">Device Addresses</a></h4>
<p>
Many devices have an optional <code>&lt;address&gt;</code>
sub-element to describe where the device is placed on the
virtual bus presented to the guest. If an address (or any
optional attribute within an address) is omitted on
input, libvirt will generate an appropriate address; but an
explicit address is required if more control over layout is
required. See below for device examples including an address
element.
</p>
<p>
Every address has a mandatory attribute <code>type</code> that
describes which bus the device is on. The choice of which
address to use for a given device is constrained in part by the
device and the architecture of the guest. For example,
a <code>&lt;disk&gt;</code> device
uses <code>type='drive'</code>, while
a <code>&lt;console&gt;</code> device would
use <code>type='pci'</code> on i686 or x86_64 guests,
or <code>type='spapr-vio'</code> on PowerPC64 pseries guests.
Each address type has further optional attributes that control
where on the bus the device will be placed:
</p>
<dl>
<dt><code>pci</code></dt>
<dd>PCI addresses have the following additional
attributes: <code>domain</code> (a 2-byte hex integer, not
currently used by qemu), <code>bus</code> (a hex value between
0 and 0xff, inclusive), <code>slot</code> (a hex value between
0x0 and 0x1f, inclusive), and <code>function</code> (a value
between 0 and 7, inclusive). Also available is
the <code>multifunction</code> attribute, which controls
turning on the multifunction bit for a particular
slot/function in the PCI control register
(<span class="since">since 0.9.7, requires QEMU
0.13</span>). <code>multifunction</code> defaults to 'off',
but should be set to 'on' for function 0 of a slot that will
have multiple functions used.
(<span class="since">Since 4.10.0</span>), PCI address extensions
depending on the architecture are supported. For example, PCI
addresses for S390 guests will have a <code>zpci</code> child
element, with two attributes: <code>uid</code> (a hex value
between 0x0001 and 0xffff, inclusive), and <code>fid</code> (a
hex value between 0x00000000 and 0xffffffff, inclusive) used by
PCI devices on S390 for User-defined Identifiers and Function
Identifiers.<br/>
<span class="since">Since 1.3.5</span>, some hypervisor
drivers may accept an <code>&lt;address type='pci'/&gt;</code>
element with no other attributes as an explicit request to
assign a PCI address for the device rather than some other
type of address that may also be appropriate for that same
device (e.g. virtio-mmio).
</dd>
<dt><code>drive</code></dt>
<dd>Drive addresses have the following additional
attributes: <code>controller</code> (a 2-digit controller
number), <code>bus</code> (a 2-digit bus number),
<code>target</code> (a 2-digit target number),
and <code>unit</code> (a 2-digit unit number on the bus).
</dd>
<dt><code>virtio-serial</code></dt>
<dd>Each virtio-serial address has the following additional
attributes: <code>controller</code> (a 2-digit controller
number), <code>bus</code> (a 2-digit bus number),
and <code>slot</code> (a 2-digit slot within the bus).
</dd>
<dt><code>ccid</code></dt>
<dd>A CCID address, for smart-cards, has the following
additional attributes: <code>bus</code> (a 2-digit bus
number), and <code>slot</code> attribute (a 2-digit slot
within the bus). <span class="since">Since 0.8.8.</span>
</dd>
<dt><code>usb</code></dt>
<dd>USB addresses have the following additional
attributes: <code>bus</code> (a hex value between 0 and 0xfff,
inclusive), and <code>port</code> (a dotted notation of up to
four octets, such as 1.2 or 2.1.3.1).
</dd>
<dt><code>spapr-vio</code></dt>
<dd>On PowerPC pseries guests, devices can be assigned to the
SPAPR-VIO bus. It has a flat 32-bit address space; by
convention, devices are generally assigned at a non-zero
multiple of 0x00001000, but other addresses are valid and
permitted by libvirt. Each address has the following
additional attribute: <code>reg</code> (the hex value address
of the starting register). <span class="since">Since
0.9.9.</span>
</dd>
<dt><code>ccw</code></dt>
<dd>S390 guests with a <code>machine</code> value of
s390-ccw-virtio use the native CCW bus for I/O devices.
CCW bus addresses have the following additional attributes:
<code>cssid</code> (a hex value between 0 and 0xfe, inclusive),
<code>ssid</code> (a value between 0 and 3, inclusive) and
<code>devno</code> (a hex value between 0 and 0xffff, inclusive).
Partially specified bus addresses are not allowed.
If omitted, libvirt will assign a free bus address with
cssid=0xfe and ssid=0. Virtio-ccw devices must have their cssid
set to 0xfe.
<span class="since">Since 1.0.4</span>
</dd>
<dt><code>virtio-mmio</code></dt>
<dd>This places the device on the virtio-mmio transport, which is
currently only available for some <code>armv7l</code> and
<code>aarch64</code> virtual machines. virtio-mmio addresses
do not have any additional attributes.
<span class="since">Since 1.1.3</span><br/>
If the guest architecture is <code>aarch64</code> and the machine
type is <code>virt</code>, libvirt will automatically assign PCI
addresses to devices; however, the presence of a single device
with virtio-mmio address in the guest configuration will cause
libvirt to assign virtio-mmio addresses to all further devices.
<span class="since">Since 3.0.0</span>
</dd>
<dt><code>isa</code></dt>
<dd>ISA addresses have the following additional
attributes: <code>iobase</code> and <code>irq</code>.
<span class="since">Since 1.2.1</span>
</dd>
<dt><code>unassigned</code></dt>
<dd>For PCI hostdevs, <code>&lt;address type='unassigned'/&gt;</code>
allows the admin to include a PCI hostdev in the domain XML definition,
without making it available for the guest. This allows for configurations
in which Libvirt manages the device as a regular PCI hostdev,
regardless of whether the guest will have access to it.
<code>&lt;address type='unassigned'/&gt;</code> is an invalid address
type for all other device types.
<span class="since">Since 6.0.0</span>
</dd>
</dl>
<h4><a id="elementsVirtio">Virtio-related options</a></h4>
<p>
QEMU's virtio devices have some attributes related to the virtio transport under
the <code>driver</code> element:
The <code>iommu</code> attribute enables the use of emulated IOMMU
by the device. The attribute <code>ats</code> controls the Address
Translation Service support for PCIe devices. This is needed to make use
of IOTLB support (see <a href="#elementsIommu">IOMMU device</a>).
Possible values are <code>on</code> or <code>off</code>.
<span class="since">Since 3.5.0</span>
</p>
<h4><a id="elementsVirtioTransitional">Virtio transitional devices</a></h4>
<p>
<span class="since">Since 5.2.0</span>, some of QEMU's virtio devices,
when used with PCI/PCIe machine types, accept the following
<code>model</code> values:
</p>
<dl>
<dt><code>virtio-transitional</code></dt>
<dd>This device can work both with virtio 0.9 and virtio 1.0 guest
drivers, so it's the best choice when compatibility with older
guest operating systems is desired. libvirt will plug the device
into a conventional PCI slot.
</dd>
<dt><code>virtio-non-transitional</code></dt>
<dd>This device can only work with virtio 1.0 guest drivers, and it's
the recommended option unless compatibility with older guest
operating systems is necessary. libvirt will plug the device into
either a PCI Express slot or a conventional PCI slot based on the
machine type, resulting in a more optimized PCI topology.
</dd>
<dt><code>virtio</code></dt>
<dd>This device will work like a <code>virtio-non-transitional</code>
device when plugged into a PCI Express slot, and like a
<code>virtio-transitional</code> device otherwise; libvirt will
pick one or the other based on the machine type. This is the best
choice when compatibility with libvirt versions older than 5.2.0
is necessary, but it's otherwise not recommended to use it.
</dd>
</dl>
<p>
While the information outlined above applies to most virtio devices,
there are a few exceptions:
</p>
<ul>
<li>
for SCSI controllers, <code>virtio-scsi</code> must be used instead
of <code>virtio</code> for backwards compatibility reasons;
</li>
<li>
some devices, such as GPUs and input devices (keyboard, tablet and
mouse), are only defined in the virtio 1.0 spec and as such don't
have a transitional variant: the only accepted model is
<code>virtio</code>, which will result in a non-transitional device.
</li>
</ul>
<p>
For more details see the
<a href="https://lists.gnu.org/archive/html/qemu-devel/2018-12/msg00923.html">qemu patch posting</a> and the
<a href="http://docs.oasis-open.org/virtio/virtio/v1.0/virtio-v1.0.html">virtio-1.0 spec</a>.
</p>
<h4><a id="elementsControllers">Controllers</a></h4>
<p>
Depending on the guest architecture, some device buses can
appear more than once, with a group of virtual devices tied to a
virtual controller. Normally, libvirt can automatically infer such
controllers without requiring explicit XML markup, but sometimes
it is necessary to provide an explicit controller element, notably
when planning the <a href="pci-hotplug.html">PCI topology</a>
for guests where device hotplug is expected.
</p>
<pre>
...
&lt;devices&gt;
&lt;controller type='ide' index='0'/&gt;
&lt;controller type='virtio-serial' index='0' ports='16' vectors='4'/&gt;
&lt;controller type='virtio-serial' index='1'&gt;
&lt;address type='pci' domain='0x0000' bus='0x00' slot='0x0a' function='0x0'/&gt;
&lt;/controller&gt;
&lt;controller type='scsi' index='0' model='virtio-scsi'&gt;
&lt;driver iothread='4'/&gt;
&lt;address type='pci' domain='0x0000' bus='0x00' slot='0x0b' function='0x0'/&gt;
&lt;/controller&gt;
&lt;controller type='xenbus' maxGrantFrames='64'/&gt;
...
&lt;/devices&gt;
...</pre>
<p>
Each controller has a mandatory attribute <code>type</code>,
which must be one of 'ide', 'fdc', 'scsi', 'sata', 'usb',
'ccid', 'virtio-serial' or 'pci', and a mandatory
attribute <code>index</code> which is the decimal integer
describing in which order the bus controller is encountered (for
use in <code>controller</code> attributes of
<code>&lt;address&gt;</code> elements).
<span class="since">Since 1.3.5</span> the index is optional; if
not specified, it will be auto-assigned to be the lowest unused
index for the given controller type. Some controller types have
additional attributes that control specific features, such as:
</p>
<dl>
<dt><code>virtio-serial</code></dt>
<dd>The <code>virtio-serial</code> controller has two additional
optional attributes <code>ports</code> and <code>vectors</code>,
which control how many devices can be connected through the
controller. <span class="since">Since 5.2.0</span>, it
supports an optional attribute <code>model</code> which can
be 'virtio', 'virtio-transitional', or 'virtio-non-transitional'. See
<a href="#elementsVirtioTransitional">Virtio transitional devices</a>
for more details.
</dd>
<dt><code>scsi</code></dt>
<dd>A <code>scsi</code> controller has an optional attribute
<code>model</code>, which is one of 'auto', 'buslogic', 'ibmvscsi',
'lsilogic', 'lsisas1068', 'lsisas1078', 'virtio-scsi',
'vmpvscsi', 'virtio-transitional', 'virtio-non-transitional'. See
<a href="#elementsVirtioTransitional">Virtio transitional devices</a>
for more details.
</dd>
<dt><code>usb</code></dt>
<dd>A <code>usb</code> controller has an optional attribute
<code>model</code>, which is one of "piix3-uhci", "piix4-uhci",
"ehci", "ich9-ehci1", "ich9-uhci1", "ich9-uhci2", "ich9-uhci3",
"vt82c686b-uhci", "pci-ohci", "nec-xhci", "qusb1" (xen pvusb
with qemu backend, version 1.1), "qusb2" (xen pvusb with qemu
backend, version 2.0) or "qemu-xhci". Additionally,
<span class="since">since 0.10.0</span>, if the USB bus needs to
be explicitly disabled for the guest, <code>model='none'</code>
may be used. <span class="since">Since 1.0.5</span>, no default
USB controller will be built on s390.
<span class="since">Since 1.3.5</span>, USB controllers accept a
<code>ports</code> attribute to configure how many devices can be
connected to the controller.</dd>
<dt><code>ide</code></dt>
<dd><span class="since">Since 3.10.0</span> for the vbox driver, the
<code>ide</code> controller has an optional attribute
<code>model</code>, which is one of "piix3", "piix4" or "ich6".</dd>
<dt><code>xenbus</code></dt>
<dd><span class="since">Since 5.2.0</span>, the <code>xenbus</code>
controller has an optional attribute <code>maxGrantFrames</code>,
which specifies the maximum number of grant frames the controller
makes available for connected devices.</dd>
</dl>
<p>
Note: The PowerPC64 "spapr-vio" addresses do not have an
associated controller.
</p>
<p>
For controllers that are themselves devices on a PCI or USB bus,
an optional sub-element <code>&lt;address&gt;</code> can specify
the exact relationship of the controller to its master bus, with
semantics <a href="#elementsAddress">given above</a>.
</p>
<p>
An optional sub-element <code>driver</code> can specify the driver
specific options:
</p>
<dl>
<dt><code>queues</code></dt>
<dd>
The optional <code>queues</code> attribute specifies the number of
queues for the controller. For best performance, it's recommended to
specify a value matching the number of vCPUs.
<span class="since">Since 1.0.5 (QEMU and KVM only)</span>
</dd>
<dt><code>cmd_per_lun</code></dt>
<dd>
The optional <code>cmd_per_lun</code> attribute specifies the maximum
number of commands that can be queued on devices controlled by the
host.
<span class="since">Since 1.2.7 (QEMU and KVM only)</span>
</dd>
<dt><code>max_sectors</code></dt>
<dd>
The optional <code>max_sectors</code> attribute specifies the maximum
amount of data in bytes that will be transferred to or from the device
in a single command. The transfer length is measured in sectors, where
a sector is 512 bytes.
<span class="since">Since 1.2.7 (QEMU and KVM only)</span>
</dd>
<dt><code>ioeventfd</code></dt>
<dd>
The optional <code>ioeventfd</code> attribute specifies
whether the controller should use
<a href='https://patchwork.kernel.org/patch/43390/'>
I/O asynchronous handling</a> or not. Accepted values are
"on" and "off". <span class="since">Since 1.2.18</span>
</dd>
<dt><code>iothread</code></dt>
<dd>
Supported for controller type <code>scsi</code> using model
<code>virtio-scsi</code> for <code>address</code> types
<code>pci</code> and <code>ccw</code>
<span class="since">since 1.3.5 (QEMU 2.4)</span>.
The optional <code>iothread</code> attribute assigns the controller
to an IOThread as defined by the range for the domain
<a href="#elementsIOThreadsAllocation"><code>iothreads</code></a>
value. Each SCSI <code>disk</code> assigned to use the specified
<code>controller</code> will utilize the same IOThread. If a specific
IOThread is desired for a specific SCSI <code>disk</code>, then
multiple controllers must be defined each having a specific
<code>iothread</code> value. The <code>iothread</code> value
must be within the range 1 to the domain iothreads value.
</dd>
<dt>virtio options</dt>
<dd>
For virtio controllers,
<a href="#elementsVirtio">Virtio-specific options</a> can also be
set. (<span class="since">Since 3.5.0</span>)
</dd>
</dl>
<p>
USB companion controllers have an optional
sub-element <code>&lt;master&gt;</code> to specify the exact
relationship of the companion to its master controller.
A companion controller is on the same bus as its master, so
the companion <code>index</code> value should be equal.
Not all controller models can be used as companion controllers
and libvirt might provide some sensible defaults (settings
of <code>master startport</code> and <code>function</code> of an
address) for some particular models.
Preferred companion controllers are <code>ich-uhci[123]</code>.
</p>
<pre>
...
&lt;devices&gt;
&lt;controller type='usb' index='0' model='ich9-ehci1'&gt;
&lt;address type='pci' domain='0' bus='0' slot='4' function='7'/&gt;
&lt;/controller&gt;
&lt;controller type='usb' index='0' model='ich9-uhci1'&gt;
&lt;master startport='0'/&gt;
&lt;address type='pci' domain='0' bus='0' slot='4' function='0' multifunction='on'/&gt;
&lt;/controller&gt;
...
&lt;/devices&gt;
...</pre>
<p>
PCI controllers have an optional <code>model</code> attribute; possible
values for this attribute are
</p>
<ul>
<li>
<code>pci-root</code>, <code>pci-bridge</code>
(<span class="since">since 1.0.5</span>)
</li>
<li>
<code>pcie-root</code>, <code>dmi-to-pci-bridge</code>
(<span class="since">since 1.1.2</span>)
</li>
<li>
<code>pcie-root-port</code>, <code>pcie-switch-upstream-port</code>,
<code>pcie-switch-downstream-port</code>
(<span class="since">since 1.2.19</span>)
</li>
<li>
<code>pci-expander-bus</code>, <code>pcie-expander-bus</code>
(<span class="since">since 1.3.4</span>)
</li>
<li>
<code>pcie-to-pci-bridge</code>
(<span class="since">since 4.3.0</span>)
</li>
</ul>
<p>
The root controllers (<code>pci-root</code>
and <code>pcie-root</code>) have an
optional <code>pcihole64</code> element specifying how big (in
kilobytes, or in the unit specified by <code>pcihole64</code>'s
<code>unit</code> attribute) the 64-bit PCI hole should be. Some guests (like
Windows XP or Windows Server 2003) might crash when QEMU and Seabios
are recent enough to support 64-bit PCI holes, unless this is disabled
(set to 0). <span class="since">Since 1.1.2 (QEMU only)</span>
</p>
<p>
PCI controllers also have an optional
subelement <code>&lt;model&gt;</code> with an attribute
<code>name</code>. The name attribute holds the name of the
specific device that qemu is emulating (e.g. "i82801b11-bridge")
rather than simply the class of device ("pcie-to-pci-bridge",
"pci-bridge"), which is set in the controller element's
model <b>attribute</b>. In almost all cases, you should not
manually add a <code>&lt;model&gt;</code> subelement to a
controller, nor should you modify one that is automatically
generated by libvirt. <span class="since">Since 1.2.19 (QEMU
only).</span>
</p>
<p>
PCI controllers also have an optional
subelement <code>&lt;target&gt;</code> with the attributes and
subelements listed below. These are configurable items that 1)
are visible to the guest OS so must be preserved for guest ABI
compatibility, and 2) are usually left to default values or
derived automatically by libvirt. In almost all cases, you
should not manually add a <code>&lt;target&gt;</code> subelement
to a controller, nor should you modify the values in the those
that are automatically generated by
libvirt. <span class="since">Since 1.2.19 (QEMU only).</span>
</p>
<dl>
<dt><code>chassisNr</code></dt>
<dd>
PCI controllers that have attribute model="pci-bridge", can
also have a <code>chassisNr</code> attribute in
the <code>&lt;target&gt;</code> subelement, which is used to
control QEMU's "chassis_nr" option for the pci-bridge device
(normally libvirt automatically sets this to the same value as
the index attribute of the pci controller). If set, chassisNr
must be between 1 and 255.
</dd>
<dt><code>chassis</code></dt>
<dd>
pcie-root-port and pcie-switch-downstream-port controllers can
also have a <code>chassis</code> attribute in
the <code>&lt;target&gt;</code> subelement, which is used to
set the controller's "chassis" configuration value, which is
visible to the virtual machine. If set, chassis must be
between 0 and 255.
</dd>
<dt><code>port</code></dt>
<dd>
pcie-root-port and pcie-switch-downstream-port controllers can
also have a <code>port</code> attribute in
the <code>&lt;target&gt;</code> subelement, which
is used to set the controller's "port" configuration value,
which is visible to the virtual machine. If set, port must be
between 0 and 255.
</dd>
<dt><code>busNr</code></dt>
<dd>
pci-expander-bus and pcie-expander-bus controllers can have an
optional <code>busNr</code> attribute (1-254). This will be
the bus number of the new bus; All bus numbers between that
specified and 255 will be available only for assignment to
PCI/PCIe controllers plugged into the hierarchy starting with
this expander bus, and bus numbers less than the specified
value will be available to the next lower expander-bus (or the
root-bus if there are no lower expander buses). If you do not
specify a busNumber, libvirt will find the lowest existing
busNumber in all other expander buses (or use 256 if there are
no others) and auto-assign the busNr of that found bus - 2,
which provides one bus number for the pci-expander-bus and one
for the pci-bridge that is automatically attached to it (if
you plan on adding more pci-bridges to the hierarchy of the
bus, you should manually set busNr to a lower value).
<p>
A similar algorithm is used for automatically determining
the busNr attribute for pcie-expander-bus, but since the
pcie-expander-bus doesn't have any built-in pci-bridge, the
2nd bus-number is just being reserved for the pcie-root-port
that must necessarily be connected to the bus in order to
actually plug in an endpoint device. If you intend to plug
multiple devices into a pcie-expander-bus, you must connect
a pcie-switch-upstream-port to the pcie-root-port that is
plugged into the pcie-expander-bus, and multiple
pcie-switch-downstream-ports to the
pcie-switch-upstream-port, and of course for this to work
properly, you will need to decrease the pcie-expander-bus'
busNr accordingly so that there are enough unused bus
numbers above it to accommodate giving out one bus number for
the upstream-port and one for each downstream-port (in
addition to the pcie-root-port and the pcie-expander-bus
itself).
</p>
</dd>
<dt><code>node</code></dt>
<dd>
Some PCI controllers (<code>pci-expander-bus</code> for the pc
machine type, <code>pcie-expander-bus</code> for the q35 machine
type and, <span class="since">since 3.6.0</span>,
<code>pci-root</code> for the pseries machine type) can have an
optional <code>&lt;node&gt;</code> subelement within
the <code>&lt;target&gt;</code> subelement, which is used to
set the NUMA node reported to the guest OS for that bus - the
guest OS will then know that all devices on that bus are a
part of the specified NUMA node (it is up to the user of the
libvirt API to attach host devices to the correct
pci-expander-bus when assigning them to the domain).
</dd>
<dt><code>index</code></dt>
<dd>
pci-root controllers for pSeries guests use this attribute to
record the order they will show up in the guest.
<span class="since">Since 3.6.0</span>
</dd>
</dl>
<p>
For machine types which provide an implicit PCI bus, the pci-root
controller with index=0 is auto-added and required to use PCI devices.
pci-root has no address.
PCI bridges are auto-added if there are too many devices to fit on
the one bus provided by pci-root, or a PCI bus number greater than zero
was specified.
PCI bridges can also be specified manually, but their addresses should
only refer to PCI buses provided by already specified PCI controllers.
Leaving gaps in the PCI controller indexes might lead to an invalid
configuration.
</p>
<pre>
...
&lt;devices&gt;
&lt;controller type='pci' index='0' model='pci-root'/&gt;
&lt;controller type='pci' index='1' model='pci-bridge'&gt;
&lt;address type='pci' domain='0' bus='0' slot='5' function='0' multifunction='off'/&gt;
&lt;/controller&gt;
&lt;/devices&gt;
...</pre>
<p>
For machine types which provide an implicit PCI Express (PCIe)
bus (for example, the machine types based on the Q35 chipset),
the pcie-root controller with index=0 is auto-added to the
domain's configuration. pcie-root has also no address, provides
31 slots (numbered 1-31) that can be used to attach PCIe or PCI
devices (although libvirt will never auto-assign a PCI device to
a PCIe slot, it will allow manual specification of such an
assignment). Devices connected to pcie-root cannot be
hotplugged. If traditional PCI devices are present in the guest
configuration, a <code>pcie-to-pci-bridge</code> controller will
automatically be added: this controller, which plugs into a
<code>pcie-root-port</code>, provides 31 usable PCI slots (1-31) with
hotplug support (<span class="since">since 4.3.0</span>). If the QEMU
binary doesn't support the corresponding device, then a
<code>dmi-to-pci-bridge</code> controller will be added instead,
usually at the defacto standard location of slot=0x1e. A
dmi-to-pci-bridge controller plugs into a PCIe slot (as provided
by pcie-root), and itself provides 31 standard PCI slots (which
also do not support device hotplug). In order to have
hot-pluggable PCI slots in the guest system, a pci-bridge
controller will also be automatically created and connected to
one of the slots of the auto-created dmi-to-pci-bridge
controller; all guest PCI devices with addresses that are
auto-determined by libvirt will be placed on this pci-bridge
device. (<span class="since">since 1.1.2</span>).
</p>
<p>
Domains with an implicit pcie-root can also add controllers
with <code>model='pcie-root-port'</code>,
<code>model='pcie-switch-upstream-port'</code>,
and <code>model='pcie-switch-downstream-port'</code>. pcie-root-port
is a simple type of bridge device that can connect only to one
of the 31 slots on the pcie-root bus on its upstream side, and
makes a single (PCIe, hotpluggable) port available on the
downstream side (at slot='0'). pcie-root-port can be used to
provide a single slot to later hotplug a PCIe device (but is not
itself hotpluggable - it must be in the configuration when the
domain is started).
(<span class="since">since 1.2.19</span>)
</p>
<p>
pcie-switch-upstream-port is a more flexible (but also more
complex) device that can only plug into a pcie-root-port or
pcie-switch-downstream-port on the upstream side (and only
before the domain is started - it is not hot-pluggable), and
provides 32 ports on the downstream side (slot='0' - slot='31')
that accept only pcie-switch-downstream-port devices; each
pcie-switch-downstream-port device can only plug into a
pcie-switch-upstream-port on its upstream side (again, not
hot-pluggable), and on its downstream side provides a single
hotpluggable pcie port that can accept any standard pci or pcie
device (or another pcie-switch-upstream-port), i.e. identical in
function to a pcie-root-port. (<span class="since">since
1.2.19</span>)
</p>
<pre>
...
&lt;devices&gt;
&lt;controller type='pci' index='0' model='pcie-root'/&gt;
&lt;controller type='pci' index='1' model='pcie-root-port'&gt;
&lt;address type='pci' domain='0x0000' bus='0x00' slot='0x01' function='0x0'/&gt;
&lt;/controller&gt;
&lt;controller type='pci' index='2' model='pcie-to-pci-bridge'&gt;
&lt;address type='pci' domain='0x0000' bus='0x01' slot='0x00' function='0x0'/&gt;
&lt;/controller&gt;
&lt;/devices&gt;
...</pre>
<h4><a id="elementsLease">Device leases</a></h4>
<p>
When using a lock manager, it may be desirable to record device leases
against a VM. The lock manager will ensure the VM won't start unless
the leases can be acquired.
</p>
<pre>
...
&lt;devices&gt;
...
&lt;lease&gt;
&lt;lockspace&gt;somearea&lt;/lockspace&gt;
&lt;key&gt;somekey&lt;/key&gt;
&lt;target path='/some/lease/path' offset='1024'/&gt;
&lt;/lease&gt;
...
&lt;/devices&gt;
...</pre>
<dl>
<dt><code>lockspace</code></dt>
<dd>This is an arbitrary string, identifying the lockspace
within which the key is held. Lock managers may impose
extra restrictions on the format, or length of the lockspace
name.</dd>
<dt><code>key</code></dt>
<dd>This is an arbitrary string, uniquely identifying the
lease to be acquired. Lock managers may impose extra
restrictions on the format, or length of the key.
</dd>
<dt><code>target</code></dt>
<dd>This is the fully qualified path of the file associated
with the lockspace. The offset specifies where the lease
is stored within the file. If the lock manager does not
require an offset, just pass 0.
</dd>
</dl>
<h4><a id="elementsHostDev">Host device assignment</a></h4>
<h5><a id="elementsHostDevSubsys">USB / PCI / SCSI devices</a></h5>
<p>
USB, PCI and SCSI devices attached to the host can be passed through
to the guest using the <code>hostdev</code> element.
<span class="since">since after 0.4.4 for USB, 0.6.0 for PCI (KVM only)
and 1.0.6 for SCSI (KVM only)</span>:
</p>
<pre>
...
&lt;devices&gt;
&lt;hostdev mode='subsystem' type='usb'&gt;
&lt;source startupPolicy='optional'&gt;
&lt;vendor id='0x1234'/&gt;
&lt;product id='0xbeef'/&gt;
&lt;/source&gt;
&lt;boot order='2'/&gt;
&lt;/hostdev&gt;
&lt;/devices&gt;
...</pre>
<p>or:</p>
<pre>
...
&lt;devices&gt;
&lt;hostdev mode='subsystem' type='pci' managed='yes'&gt;
&lt;source&gt;
&lt;address domain='0x0000' bus='0x06' slot='0x02' function='0x0'/&gt;
&lt;/source&gt;
&lt;boot order='1'/&gt;
&lt;rom bar='on' file='/etc/fake/boot.bin'/&gt;
&lt;/hostdev&gt;
&lt;/devices&gt;
...</pre>
<p>or:</p>
<pre>
...
&lt;devices&gt;
&lt;hostdev mode='subsystem' type='scsi' sgio='filtered' rawio='yes'&gt;
&lt;source&gt;
&lt;adapter name='scsi_host0'/&gt;
&lt;address bus='0' target='0' unit='0'/&gt;
&lt;/source&gt;
&lt;readonly/&gt;
&lt;address type='drive' controller='0' bus='0' target='0' unit='0'/&gt;
&lt;/hostdev&gt;
&lt;/devices&gt;
...</pre>
<p>or:</p>
<pre>
...
&lt;devices&gt;
&lt;hostdev mode='subsystem' type='scsi'&gt;
&lt;source protocol='iscsi' name='iqn.2014-08.com.example:iscsi-nopool/1'&gt;
&lt;host name='example.com' port='3260'/&gt;
&lt;auth username='myuser'&gt;
&lt;secret type='iscsi' usage='libvirtiscsi'/&gt;
&lt;/auth&gt;
&lt;/source&gt;
&lt;address type='drive' controller='0' bus='0' target='0' unit='0'/&gt;
&lt;/hostdev&gt;
&lt;/devices&gt;
...</pre>
<p>or:</p>
<pre>
...
&lt;devices&gt;
&lt;hostdev mode='subsystem' type='scsi_host'&gt;
&lt;source protocol='vhost' wwpn='naa.50014057667280d8'/&gt;
&lt;/hostdev&gt;
&lt;/devices&gt;
...</pre>
<p>or:</p>
<pre>
...
&lt;devices&gt;
&lt;hostdev mode='subsystem' type='mdev' model='vfio-pci'&gt;
&lt;source&gt;
&lt;address uuid='c2177883-f1bb-47f0-914d-32a22e3a8804'/&gt;
&lt;/source&gt;
&lt;/hostdev&gt;
&lt;hostdev mode='subsystem' type='mdev' model='vfio-ccw'&gt;
&lt;source&gt;
&lt;address uuid='9063cba3-ecef-47b6-abcf-3fef4fdcad85'/&gt;
&lt;/source&gt;
&lt;address type='ccw' cssid='0xfe' ssid='0x0' devno='0x0001'/&gt;
&lt;/hostdev&gt;
&lt;/devices&gt;
...</pre>
<dl>
<dt><code>hostdev</code></dt>
<dd>The <code>hostdev</code> element is the main container for describing
host devices. For each device, the <code>mode</code> is always
"subsystem" and the <code>type</code> is one of the following values
with additional attributes noted.
<dl>
<dt><code>usb</code></dt>
<dd>USB devices are detached from the host on guest startup
and reattached after the guest exits or the device is
hot-unplugged.
</dd>
<dt><code>pci</code></dt>
<dd>For PCI devices, when <code>managed</code> is "yes" it is
detached from the host before being passed on to the guest
and reattached to the host after the guest exits. If
<code>managed</code> is omitted or "no", the user is
responsible to call <code>virNodeDeviceDetachFlags</code>
(or <code>virsh nodedev-detach</code> before starting the guest
or hot-plugging the device and <code>virNodeDeviceReAttach</code>
(or <code>virsh nodedev-reattach</code>) after hot-unplug or
stopping the guest.
</dd>
<dt><code>scsi</code></dt>
<dd>For SCSI devices, user is responsible to make sure the device
is not used by host. If supported by the hypervisor and OS, the
optional <code>sgio</code> (<span class="since">since 1.0.6</span>)
attribute indicates whether unprivileged SG_IO commands are
filtered for the disk. Valid settings are "filtered" or
"unfiltered", where the default is "filtered".
The optional <code>rawio</code>
(<span class="since">since 1.2.9</span>) attribute indicates
whether the lun needs the rawio capability. Valid settings are
"yes" or "no". See the rawio description within the
<a href="#elementsDisks">disk</a> section.
If a disk lun in the domain already has the rawio capability,
then this setting not required.
</dd>
<dt><code>scsi_host</code></dt>
<dd><span class="since">since 2.5.0</span>For SCSI devices, user
is responsible to make sure the device is not used by host. This
<code>type</code> passes all LUNs presented by a single HBA to
the guest. <span class="since">Since 5.2.0,</span> the
<code>model</code> attribute can be specified further
with "virtio-transitional", "virtio-non-transitional", or
"virtio". See
<a href="#elementsVirtioTransitional">Virtio transitional devices</a>
for more details.
</dd>
<dt><code>mdev</code></dt>
<dd>For mediated devices (<span class="since">Since 3.2.0</span>)
the <code>model</code> attribute specifies the device API which
determines how the host's vfio driver will expose the device to the
guest. Currently, <code>model='vfio-pci'</code>,
<code>model='vfio-ccw'</code> (<span class="since">Since 4.4.0</span>)
and <code>model='vfio-ap'</code> (<span class="since">Since 4.9.0</span>)
is supported. <a href="drvnodedev.html#MDEV">MDEV</a> section
provides more information about mediated devices as well as how to
create mediated devices on the host.
<span class="since">Since 4.6.0 (QEMU 2.12)</span> an optional
<code>display</code> attribute may be used to enable or disable
support for an accelerated remote desktop backed by a mediated
device (such as NVIDIA vGPU or Intel GVT-g) as an alternative to
emulated <a href="#elementsVideo">video devices</a>. This attribute
is limited to <code>model='vfio-pci'</code> only. Supported values
are either <code>on</code> or <code>off</code> (default is 'off').
It is required to use a
<a href="#elementsGraphics">graphical framebuffer</a> in order to
use this attribute, currently only supported with VNC, Spice and
egl-headless graphics devices.
<span class="since">Since version 5.10.0</span>, there is an optional
<code>ramfb</code> attribute for devices with
<code>model='vfio-pci'</code>. Supported values are either
<code>on</code> or <code>off</code> (default is 'off'). When
enabled, this attribute provides a memory framebuffer device to the
guest. This framebuffer will be used as a boot display when a vgpu
device is the primary display.
<p>
Note: There are also some implications on the usage of guest's
address type depending on the <code>model</code> attribute,
see the <code>address</code> element below.
</p>
</dd>
</dl>
<p>
Note: The <code>managed</code> attribute is only used with
<code>type='pci'</code> and is ignored by all the other device types,
thus setting <code>managed</code> explicitly with other than a PCI
device has the same effect as omitting it. Similarly,
<code>model</code> attribute is only supported by mediated devices and
ignored by all other device types.
</p>
</dd>
<dt><code>source</code></dt>
<dd>The source element describes the device as seen from the host using
the following mechanism to describe:
<dl>
<dt><code>usb</code></dt>
<dd>The USB device can either be addressed by vendor / product id
using the <code>vendor</code> and <code>product</code> elements
or by the device's address on the host using the
<code>address</code> element.
<p>
<span class="since">Since 1.0.0</span>, the <code>source</code>
element of USB devices may contain <code>startupPolicy</code>
attribute which can be used to define policy what to do if the
specified host USB device is not found. The attribute accepts
the following values:
</p>
<table class="top_table">
<tr>
<td> mandatory </td>
<td> fail if missing for any reason (the default) </td>
</tr>
<tr>
<td> requisite </td>
<td> fail if missing on boot up,
drop if missing on migrate/restore/revert </td>
</tr>
<tr>
<td> optional </td>
<td> drop if missing at any start attempt </td>
</tr>
</table>
</dd>
<dt><code>pci</code></dt>
<dd>PCI devices can only be described by their <code>address</code>.
</dd>
<dt><code>scsi</code></dt>
<dd>SCSI devices are described by both the <code>adapter</code>
and <code>address</code> elements. The <code>address</code>
element includes a <code>bus</code> attribute (a 2-digit bus
number), a <code>target</code> attribute (a 10-digit target
number), and a <code>unit</code> attribute (a 20-digit unit
number on the bus). Not all hypervisors support larger
<code>target</code> and <code>unit</code> values. It is up
to each hypervisor to determine the maximum value supported
for the adapter.
<p>
<span class="since">Since 1.2.8</span>, the <code>source</code>
element of a SCSI device may contain the <code>protocol</code>
attribute. When the attribute is set to "iscsi", the host
device XML follows the network <a href="#elementsDisks">disk</a>
device using the same <code>name</code> attribute and optionally
using the <code>auth</code> element to provide the authentication
credentials to the iSCSI server.
</p>
</dd>
<dt><code>scsi_host</code></dt>
<dd><span class="since">Since 2.5.0</span>, multiple LUNs behind a
single SCSI HBA are described by a <code>protocol</code>
attribute set to "vhost" and a <code>wwpn</code> attribute that
is the vhost_scsi wwpn (16 hexadecimal digits with a prefix of
"naa.") established in the host configfs.
</dd>
<dt><code>mdev</code></dt>
<dd>Mediated devices (<span class="since">Since 3.2.0</span>) are
described by the <code>address</code> element. The
<code>address</code> element contains a single mandatory attribute
<code>uuid</code>.
</dd>
</dl>
</dd>
<dt><code>vendor</code>, <code>product</code></dt>
<dd>The <code>vendor</code> and <code>product</code> elements each have an
<code>id</code> attribute that specifies the USB vendor and product id.
The ids can be given in decimal, hexadecimal (starting with 0x) or
octal (starting with 0) form.</dd>
<dt><code>boot</code></dt>
<dd>Specifies that the device is bootable. The <code>order</code>
attribute determines the order in which devices will be tried during
boot sequence. The per-device <code>boot</code> elements cannot be
used together with general boot elements in
<a href="#elementsOSBIOS">BIOS bootloader</a> section.
<span class="since">Since 0.8.8</span> for PCI devices,
<span class="since">Since 1.0.1</span> for USB devices.
</dd>
<dt><code>rom</code></dt>
<dd>The <code>rom</code> element is used to change how a PCI
device's ROM is presented to the guest. The optional <code>bar</code>
attribute can be set to "on" or "off", and determines whether
or not the device's ROM will be visible in the guest's memory
map. (In PCI documentation, the "rombar" setting controls the
presence of the Base Address Register for the ROM). If no rom
bar is specified, the qemu default will be used (older
versions of qemu used a default of "off", while newer qemus
have a default of "on"). <span class="since">Since
0.9.7 (QEMU and KVM only)</span>. The optional
<code>file</code> attribute contains an absolute path to a binary file
to be presented to the guest as the device's ROM BIOS. This
can be useful, for example, to provide a PXE boot ROM for a
virtual function of an sr-iov capable ethernet device (which
has no boot ROMs for the VFs).
<span class="since">Since 0.9.10 (QEMU and KVM only)</span>.
The optional <code>enabled</code> attribute can be set to
<code>no</code> to disable PCI ROM loading completely for the device;
if PCI ROM loading is disabled through this attribute, attempts to
tweak the loading process further using the <code>bar</code> or
<code>file</code> attributes will be rejected.
<span class="since">Since 4.3.0 (QEMU and KVM only)</span>.
</dd>
<dt><code>address</code></dt>
<dd>The <code>address</code> element for USB devices has a
<code>bus</code> and <code>device</code> attribute to specify the
USB bus and device number the device appears at on the host.
The values of these attributes can be given in decimal, hexadecimal
(starting with 0x) or octal (starting with 0) form.
For PCI devices the element carries 4 attributes allowing to designate
the device as can be found with the <code>lspci</code> or
with <code>virsh nodedev-list</code>. For SCSI devices a 'drive'
address type must be used. For mediated devices, which are software-only
devices defining an allocation of resources on the physical parent device,
the address type used must conform to the <code>model</code> attribute
of element <code>hostdev</code>, e.g. any address type other than PCI for
<code>vfio-pci</code> device API or any address type other than CCW for
<code>vfio-ccw</code> device API will result in an error.
<a href="#elementsAddress">See above</a> for more details on the address
element.</dd>
<dt><code>driver</code></dt>
<dd>
PCI devices can have an optional <code>driver</code>
subelement that specifies which backend driver to use for PCI
device assignment. Use the <code>name</code> attribute to
select either "vfio" (for the new VFIO device assignment
backend, which is compatible with UEFI SecureBoot) or "kvm"
(the legacy device assignment handled directly by the KVM
kernel module)<span class="since">Since 1.0.5 (QEMU and KVM
only, requires kernel 3.6 or newer)</span>. When specified,
device assignment will fail if the requested method of device
assignment isn't available on the host. When not specified,
the default is "vfio" on systems where the VFIO driver is
available and loaded, and "kvm" on older systems, or those
where the VFIO driver hasn't been
loaded <span class="since">Since 1.1.3</span> (prior to that
the default was always "kvm").
</dd>
<dt><code>readonly</code></dt>
<dd>Indicates that the device is readonly, only supported by SCSI host
device now. <span class="since">Since 1.0.6 (QEMU and KVM only)</span>
</dd>
<dt><code>shareable</code></dt>
<dd>If present, this indicates the device is expected to be shared
between domains (assuming the hypervisor and OS support this).
Only supported by SCSI host device.
<span class="since">Since 1.0.6</span>
<p>
Note: Although <code>shareable</code> was introduced
<span class="since">in 1.0.6</span>, it did not work as
as expected until <span class="since">1.2.2</span>.
</p>
</dd>
</dl>
<h5><a id="elementsHostDevCaps">Block / character devices</a></h5>
<p>
Block / character devices from the host can be passed through
to the guest using the <code>hostdev</code> element. This is
only possible with container based virtualization. Devices are specified
by a fully qualified path.
<span class="since">since after 1.0.1 for LXC</span>:
</p>
<pre>
...
&lt;hostdev mode='capabilities' type='storage'&gt;
&lt;source&gt;
&lt;block&gt;/dev/sdf1&lt;/block&gt;
&lt;/source&gt;
&lt;/hostdev&gt;
...
</pre>
<pre>
...
&lt;hostdev mode='capabilities' type='misc'&gt;
&lt;source&gt;
&lt;char&gt;/dev/input/event3&lt;/char&gt;
&lt;/source&gt;
&lt;/hostdev&gt;
...
</pre>
<pre>
...
&lt;hostdev mode='capabilities' type='net'&gt;
&lt;source&gt;
&lt;interface&gt;eth0&lt;/interface&gt;
&lt;/source&gt;
&lt;/hostdev&gt;
...
</pre>
<dl>
<dt><code>hostdev</code></dt>
<dd>The <code>hostdev</code> element is the main container for describing
host devices. For block/character device passthrough <code>mode</code> is
always "capabilities" and <code>type</code> is "storage" for a block
device, "misc" for a character device and "net" for a host network
interface.
</dd>
<dt><code>source</code></dt>
<dd>The source element describes the device as seen from the host.
For block devices, the path to the block device in the host
OS is provided in the nested "block" element, while for character
devices the "char" element is used. For network interfaces, the
name of the interface is provided in the "interface" element.
</dd>
</dl>
<h4><a id="elementsRedir">Redirected devices</a></h4>
<p>
USB device redirection through a character device is
supported <span class="since">since after 0.9.5 (KVM
only)</span>:
</p>
<pre>
...
&lt;devices&gt;
&lt;redirdev bus='usb' type='tcp'&gt;
&lt;source mode='connect' host='localhost' service='4000'/&gt;
&lt;boot order='1'/&gt;
&lt;/redirdev&gt;
&lt;redirfilter&gt;
&lt;usbdev class='0x08' vendor='0x1234' product='0xbeef' version='2.56' allow='yes'/&gt;
&lt;usbdev allow='no'/&gt;
&lt;/redirfilter&gt;
&lt;/devices&gt;
...</pre>
<dl>
<dt><code>redirdev</code></dt>
<dd>The <code>redirdev</code> element is the main container for
describing redirected devices. <code>bus</code> must be "usb"
for a USB device.
An additional attribute <code>type</code> is required,
matching one of the
supported <a href="#elementsConsole">serial device</a> types,
to describe the host side of the
tunnel; <code>type='tcp'</code>
or <code>type='spicevmc'</code> (which uses the usbredir
channel of a <a href="#elementsGraphics">SPICE graphics
device</a>) are typical. The redirdev element has an optional
sub-element <code>&lt;address&gt;</code> which can tie the
device to a particular controller. Further sub-elements,
such as <code>&lt;source&gt;</code>, may be required according
to the given type, although a <code>&lt;target&gt;</code> sub-element
is not required (since the consumer of the character device is
the hypervisor itself, rather than a device visible in the guest).
</dd>
<dt><code>boot</code></dt>
<dd>Specifies that the device is bootable.
The <code>order</code> attribute determines the order in which
devices will be tried during boot sequence. The per-device
<code>boot</code> elements cannot be used together with general
boot elements in <a href="#elementsOSBIOS">BIOS bootloader</a> section.
(<span class="since">Since 1.0.1</span>)
</dd>
<dt><code>redirfilter</code></dt>
<dd>The<code> redirfilter </code>element is used for creating the
filter rule to filter out certain devices from redirection.
It uses sub-element <code>&lt;usbdev&gt;</code> to define each filter rule.
<code>class</code> attribute is the USB Class code, for example,
0x08 represents mass storage devices. The USB device can be addressed by
vendor / product id using the <code>vendor</code> and <code>product</code> attributes.
<code>version</code> is the device revision from the bcdDevice field (not
the version of the USB protocol).
These four attributes are optional and <code>-1</code> can be used to allow
any value for them. <code>allow</code> attribute is mandatory,
'yes' means allow, 'no' for deny.
</dd>
</dl>
<h4><a id="elementsSmartcard">Smartcard devices</a></h4>
<p>
A virtual smartcard device can be supplied to the guest via the
<code>smartcard</code> element. A USB smartcard reader device on
the host cannot be used on a guest with simple device
passthrough, since it will then not be available on the host,
possibly locking the host computer when it is "removed".
Therefore, some hypervisors provide a specialized virtual device
that can present a smartcard interface to the guest, with
several modes for describing how credentials are obtained from
the host or even a from a channel created to a third-party
smartcard provider. <span class="since">Since 0.8.8</span>
</p>
<pre>
...
&lt;devices&gt;
&lt;smartcard mode='host'/&gt;
&lt;smartcard mode='host-certificates'&gt;
&lt;certificate&gt;cert1&lt;/certificate&gt;
&lt;certificate&gt;cert2&lt;/certificate&gt;
&lt;certificate&gt;cert3&lt;/certificate&gt;
&lt;database&gt;/etc/pki/nssdb/&lt;/database&gt;
&lt;/smartcard&gt;
&lt;smartcard mode='passthrough' type='tcp'&gt;
&lt;source mode='bind' host='127.0.0.1' service='2001'/&gt;
&lt;protocol type='raw'/&gt;
&lt;address type='ccid' controller='0' slot='0'/&gt;
&lt;/smartcard&gt;
&lt;smartcard mode='passthrough' type='spicevmc'/&gt;
&lt;/devices&gt;
...
</pre>
<p>
The <code>&lt;smartcard&gt;</code> element has a mandatory
attribute <code>mode</code>. The following modes are supported;
in each mode, the guest sees a device on its USB bus that
behaves like a physical USB CCID (Chip/Smart Card Interface
Device) card.
</p>
<dl>
<dt><code>host</code></dt>
<dd>The simplest operation, where the hypervisor relays all
requests from the guest into direct access to the host's
smartcard via NSS. No other attributes or sub-elements are
required. See below about the use of an
optional <code>&lt;address&gt;</code> sub-element.</dd>
<dt><code>host-certificates</code></dt>
<dd>Rather than requiring a smartcard to be plugged into the
host, it is possible to provide three NSS certificate names
residing in a database on the host. These certificates can be
generated via the command <code>certutil -d /etc/pki/nssdb -x -t
CT,CT,CT -S -s CN=cert1 -n cert1</code>, and the resulting three
certificate names must be supplied as the content of each of
three <code>&lt;certificate&gt;</code> sub-elements. An
additional sub-element <code>&lt;database&gt;</code> can specify
the absolute path to an alternate directory (matching
the <code>-d</code> option of the <code>certutil</code> command
when creating the certificates); if not present, it defaults to
/etc/pki/nssdb.</dd>
<dt><code>passthrough</code></dt>
<dd>Rather than having the hypervisor directly communicate with
the host, it is possible to tunnel all requests through a
secondary character device to a third-party provider (which may
in turn be talking to a smartcard or using three certificate
files). In this mode of operation, an additional
attribute <code>type</code> is required, matching one of the
supported <a href="#elementsConsole">serial device</a> types, to
describe the host side of the tunnel; <code>type='tcp'</code>
or <code>type='spicevmc'</code> (which uses the smartcard
channel of a <a href="#elementsGraphics">SPICE graphics
device</a>) are typical. Further sub-elements, such
as <code>&lt;source&gt;</code>, may be required according to the
given type, although a <code>&lt;target&gt;</code> sub-element
is not required (since the consumer of the character device is
the hypervisor itself, rather than a device visible in the
guest).</dd>
</dl>
<p>
Each mode supports an optional
sub-element <code>&lt;address&gt;</code>, which fine-tunes the
correlation between the smartcard and a ccid bus
controller, <a href="#elementsAddress">documented above</a>.
For now, qemu only supports at most one
smartcard, with an address of bus=0 slot=0.
</p>
<h4><a id="elementsNICS">Network interfaces</a></h4>
<pre>
...
&lt;devices&gt;
&lt;interface type='direct' trustGuestRxFilters='yes'&gt;
&lt;source dev='eth0'/&gt;
&lt;mac address='52:54:00:5d:c7:9e'/&gt;
&lt;boot order='1'/&gt;
&lt;rom bar='off'/&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
There are several possibilities for specifying a network
interface visible to the guest. Each subsection below provides
more details about common setup options.
</p>
<p>
<span class="since">Since 1.2.10</span>),
the <code>interface</code> element
property <code>trustGuestRxFilters</code> provides the
capability for the host to detect and trust reports from the
guest regarding changes to the interface mac address and receive
filters by setting the attribute to <code>yes</code>. The default
setting for the attribute is <code>no</code> for security
reasons and support depends on the guest network device model as
well as the type of connection on the host - currently it is
only supported for the virtio device model and for macvtap
connections on the host.
</p>
<p>
Each <code>&lt;interface&gt;</code> element has an
optional <code>&lt;address&gt;</code> sub-element that can tie
the interface to a particular pci slot, with
attribute <code>type='pci'</code>
as <a href="#elementsAddress">documented above</a>.
</p>
<h5><a id="elementsNICSVirtual">Virtual network</a></h5>
<p>
<strong><em>
This is the recommended config for general guest connectivity on
hosts with dynamic / wireless networking configs (or multi-host
environments where the host hardware details are described
separately in a <code>&lt;network&gt;</code>
definition <span class="since">Since 0.9.4</span>).
</em></strong>
</p>
<p>
Provides a connection whose details are described by the named
network definition. Depending on the virtual network's "forward
mode" configuration, the network may be totally isolated
(no <code>&lt;forward&gt;</code> element given), NAT'ing to an
explicit network device or to the default route
(<code>&lt;forward mode='nat'&gt;</code>), routed with no NAT
(<code>&lt;forward mode='route'/&gt;</code>), or connected
directly to one of the host's network interfaces (via macvtap)
or bridge devices ((<code>&lt;forward
mode='bridge|private|vepa|passthrough'/&gt;</code> <span class="since">Since
0.9.4</span>)
</p>
<p>
For networks with a forward mode of bridge, private, vepa, and
passthrough, it is assumed that the host has any necessary DNS
and DHCP services already setup outside the scope of libvirt. In
the case of isolated, nat, and routed networks, DHCP and DNS are
provided on the virtual network by libvirt, and the IP range can
be determined by examining the virtual network config with
'<code>virsh net-dumpxml [networkname]</code>'. There is one
virtual network called 'default' setup out of the box which does
NAT'ing to the default route and has an IP range
of <code>192.168.122.0/255.255.255.0</code>. Each guest will
have an associated tun device created with a name of vnetN,
which can also be overridden with the &lt;target&gt; element
(see
<a href="#elementsNICSTargetOverride">overriding the target element</a>).
</p>
<p>
When the source of an interface is a network,
a <code>portgroup</code> can be specified along with the name of
the network; one network may have multiple portgroups defined,
with each portgroup containing slightly different configuration
information for different classes of network
connections. <span class="since">Since 0.9.4</span>.
</p>
<p>
When a guest is running an interface of type <code>network</code>
may include a <code>portid</code> attribute. This provides the UUID
of an associated virNetworkPortPtr object that records the association
between the domain interface and the network. This attribute is
read-only since port objects are create and deleted automatically
during startup and shutdown. <span class="since">Since 5.1.0</span>
</p>
<p>
Also, similar to <code>direct</code> network connections
(described below), a connection of type <code>network</code> may
specify a <code>virtualport</code> element, with configuration
data to be forwarded to a vepa (802.1Qbg) or 802.1Qbh compliant
switch (<span class="since">Since 0.8.2</span>), or to an
Open vSwitch virtual switch (<span class="since">Since
0.9.11</span>).
</p>
<p>
Since the actual type of switch may vary depending on the
configuration in the <code>&lt;network&gt;</code> on the host,
it is acceptable to omit the virtualport <code>type</code>
attribute, and specify attributes from multiple different
virtualport types (and also to leave out certain attributes); at
domain startup time, a complete <code>&lt;virtualport&gt;</code>
element will be constructed by merging together the type and
attributes defined in the network and the portgroup referenced
by the interface. The newly-constructed virtualport is a combination
of them. The attributes from lower virtualport can't make change
on the ones defined in higher virtualport.
Interface takes the highest priority, portgroup is lowest priority.
(<span class="since">Since 0.10.0</span>). For example, in order
to work properly with both an 802.1Qbh switch and an Open vSwitch
switch, you may choose to specify no type, but both
a <code>profileid</code> (in case the switch is 802.1Qbh) and
an <code>interfaceid</code> (in case the switch is Open vSwitch)
(you may also omit the other attributes, such as managerid,
typeid, or profileid, to be filled in from the
network's <code>&lt;virtualport&gt;</code>). If you want to
limit a guest to connecting only to certain types of switches,
you can specify the virtualport type, but still omit some/all of
the parameters - in this case if the host's network has a
different type of virtualport, connection of the interface will
fail.
</p>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
&lt;/interface&gt;
...
&lt;interface type='network'&gt;
&lt;source network='default' portgroup='engineering'/&gt;
&lt;target dev='vnet7'/&gt;
&lt;mac address="00:11:22:33:44:55"/&gt;
&lt;virtualport&gt;
&lt;parameters instanceid='09b11c53-8b5c-4eeb-8f00-d84eaa0aaa4f'/&gt;
&lt;/virtualport&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<h5><a id="elementsNICSBridge">Bridge to LAN</a></h5>
<p>
<strong><em>
This is the recommended config for general guest connectivity on
hosts with static wired networking configs.
</em></strong>
</p>
<p>
Provides a bridge from the VM directly to the LAN. This assumes
there is a bridge device on the host which has one or more of the hosts
physical NICs enslaved. The guest VM will have an associated tun device
created with a name of vnetN, which can also be overridden with the
&lt;target&gt; element (see
<a href="#elementsNICSTargetOverride">overriding the target element</a>).
The tun device will be enslaved to the bridge. The IP range / network
configuration is whatever is used on the LAN. This provides the guest VM
full incoming &amp; outgoing net access just like a physical machine.
</p>
<p>
On Linux systems, the bridge device is normally a standard Linux
host bridge. On hosts that support Open vSwitch, it is also
possible to connect to an Open vSwitch bridge device by adding
a <code>&lt;virtualport type='openvswitch'/&gt;</code> to the
interface definition. (<span class="since">Since
0.9.11</span>). The Open vSwitch type virtualport accepts two
parameters in its <code>&lt;parameters&gt;</code> element -
an <code>interfaceid</code> which is a standard uuid used to
uniquely identify this particular interface to Open vSwitch (if
you do not specify one, a random interfaceid will be generated
for you when you first define the interface), and an
optional <code>profileid</code> which is sent to Open vSwitch as
the interfaces "port-profile".
</p>
<pre>
...
&lt;devices&gt;
...
&lt;interface type='bridge'&gt;
&lt;source bridge='br0'/&gt;
&lt;/interface&gt;
&lt;interface type='bridge'&gt;
&lt;source bridge='br1'/&gt;
&lt;target dev='vnet7'/&gt;
&lt;mac address="00:11:22:33:44:55"/&gt;
&lt;/interface&gt;
&lt;interface type='bridge'&gt;
&lt;source bridge='ovsbr'/&gt;
&lt;virtualport type='openvswitch'&gt;
&lt;parameters profileid='menial' interfaceid='09b11c53-8b5c-4eeb-8f00-d84eaa0aaa4f'/&gt;
&lt;/virtualport&gt;
&lt;/interface&gt;
...
&lt;/devices&gt;
...</pre>
<p>
On hosts that support Open vSwitch on the kernel side and have the
Midonet Host Agent configured, it is also possible to connect to the
'midonet' bridge device by adding a
<code>&lt;virtualport type='midonet'/&gt;</code> to the
interface definition. (<span class="since">Since
1.2.13</span>). The Midonet virtualport type requires an
<code>interfaceid</code> attribute in its
<code>&lt;parameters&gt;</code> element. This interface id is the UUID
that specifies which port in the virtual network topology will be bound
to the interface.
</p>
<pre>
...
&lt;devices&gt;
...
&lt;interface type='bridge'&gt;
&lt;source bridge='br0'/&gt;
&lt;/interface&gt;
&lt;interface type='bridge'&gt;
&lt;source bridge='br1'/&gt;
&lt;target dev='vnet7'/&gt;
&lt;mac address="00:11:22:33:44:55"/&gt;
&lt;/interface&gt;
&lt;interface type='bridge'&gt;
&lt;source bridge='midonet'/&gt;
&lt;virtualport type='midonet'&gt;
&lt;parameters interfaceid='0b2d64da-3d0e-431e-afdd-804415d6ebbb'/&gt;
&lt;/virtualport&gt;
&lt;/interface&gt;
...
&lt;/devices&gt;
...</pre>
<h5><a id="elementsNICSSlirp">Userspace SLIRP stack</a></h5>
<p>
Provides a virtual LAN with NAT to the outside world. The virtual
network has DHCP &amp; DNS services and will give the guest VM addresses
starting from <code>10.0.2.15</code>. The default router will be
<code>10.0.2.2</code> and the DNS server will be <code>10.0.2.3</code>.
This networking is the only option for unprivileged users who need their
VMs to have outgoing access. <span class="since">Since 3.8.0</span>
it is possible to override the default network address by
including an <code>ip</code> element specifying an IPv4
address in its one mandatory attribute, <code>address</code>.
Optionally, a second <code>ip</code> element with a
<code>family</code> attribute set to "ipv6" can be
specified to add an IPv6 address to the interface.
<code>address</code>. Optionally, address
<code>prefix</code> can be specified.
</p>
<pre>
...
&lt;devices&gt;
&lt;interface type='user'/&gt;
...
&lt;interface type='user'&gt;
&lt;mac address="00:11:22:33:44:55"/&gt;
&lt;ip family='ipv4' address='172.17.2.0' prefix='24'/&gt;
&lt;ip family='ipv6' address='2001:db8:ac10:fd01::' prefix='64'/&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<h5><a id="elementsNICSEthernet">Generic ethernet connection</a></h5>
<p>
Provides a means to use a new or existing tap device (or veth
device pair, depening on the needs of the hypervisor driver)
that is partially or wholly setup external to libvirt (either
prior to the guest starting, or while the guest is being started
via an optional script specified in the config).
</p>
<p>
The name of the tap device can optionally be specified with
the <code>dev</code> attribute of the
<code>&lt;target&gt;</code> element. If no target dev is
specified, libvirt will create a new standard tap device with a
name of the pattern "vnetN", where "N" is replaced with a
number. If a target dev is specified and that device doesn't
exist, then a new standard tap device will be created with the
exact dev name given. If the specified target dev does exist,
then that existing device will be used. Usually some basic setup
of the device is done by libvirt, including setting a MAC
address, and the IFF_UP flag, but if the <code>dev</code> is a
pre-existing device, and the <code>managed</code> attribute of
the <code>target</code> element is also set to "no" (the default
value is "yes"), even this basic setup will not be performed -
libvirt will simply pass the device on to the hypervisor with no
setup at all. <span class="since">Since 5.7.0</span> Using
managed='no' with a pre-created tap device is useful because
it permits a virtual machine managed by an unprivileged libvirtd
to have emulated network devices based on tap devices.
</p>
<p>
After creating/opening the tap device, an optional shell script
(given in the <code>path</code> attribute of
the <code>&lt;script&gt;</code> element) will be run; this can
be used to do whatever extra host network integration is
required.
</p>
<pre>
...
&lt;devices&gt;
&lt;interface type='ethernet'&gt;
&lt;script path='/etc/qemu-ifup-mynet'/&gt;
&lt;/interface&gt;
...
&lt;interface type='ethernet'&gt;
&lt;target dev='mytap1' managed='no'/&gt;
&lt;model type='virtio'/&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<h5><a id="elementsNICSDirect">Direct attachment to physical interface</a></h5>
<p>
Provides direct attachment of the virtual machine's NIC to the given
physical interface of the host.
<span class="since">Since 0.7.7 (QEMU and KVM only)</span><br/>
This setup requires the Linux macvtap
driver to be available. <span class="since">(Since Linux 2.6.34.)</span>
One of the modes 'vepa'
( <a href="http://www.ieee802.org/1/files/public/docs2009/new-evb-congdon-vepa-modular-0709-v01.pdf">
'Virtual Ethernet Port Aggregator'</a>), 'bridge' or 'private'
can be chosen for the operation mode of the macvtap device, 'vepa'
being the default mode. The individual modes cause the delivery of
packets to behave as follows:
</p>
<p>
If the model type is set to <code>virtio</code> and
interface's <code>trustGuestRxFilters</code> attribute is set
to <code>yes</code>, changes made to the interface mac address,
unicast/multicast receive filters, and vlan settings in the
guest will be monitored and propagated to the associated macvtap
device on the host (<span class="since">Since
1.2.10</span>). If <code>trustGuestRxFilters</code> is not set,
or is not supported for the device model in use, an attempted
change to the mac address originating from the guest side will
result in a non-working network connection.
</p>
<dl>
<dt><code>vepa</code></dt>
<dd>All VMs' packets are sent to the external bridge. Packets
whose destination is a VM on the same host as where the
packet originates from are sent back to the host by the VEPA
capable bridge (today's bridges are typically not VEPA capable).</dd>
<dt><code>bridge</code></dt>
<dd>Packets whose destination is on the same host as where they
originate from are directly delivered to the target macvtap device.
Both origin and destination devices need to be in bridge mode
for direct delivery. If either one of them is in <code>vepa</code> mode,
a VEPA capable bridge is required.</dd>
<dt><code>private</code></dt>
<dd>All packets are sent to the external bridge and will only be
delivered to a target VM on the same host if they are sent through an
external router or gateway and that device sends them back to the
host. This procedure is followed if either the source or destination
device is in <code>private</code> mode.</dd>
<dt><code>passthrough</code></dt>
<dd>This feature attaches a virtual function of a SRIOV capable
NIC directly to a VM without losing the migration capability.
All packets are sent to the VF/IF of the configured network device.
Depending on the capabilities of the device additional prerequisites or
limitations may apply; for example, on Linux this requires
kernel 2.6.38 or newer. <span class="since">Since 0.9.2</span></dd>
</dl>
<pre>
...
&lt;devices&gt;
...
&lt;interface type='direct' trustGuestRxFilters='no'&gt;
&lt;source dev='eth0' mode='vepa'/&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
The network access of direct attached virtual machines can be
managed by the hardware switch to which the physical interface
of the host machine is connected to.
</p>
<p>
The interface can have additional parameters as shown below,
if the switch is conforming to the IEEE 802.1Qbg standard.
The parameters of the virtualport element are documented in more detail
in the IEEE 802.1Qbg standard. The values are network specific and
should be provided by the network administrator. In 802.1Qbg terms,
the Virtual Station Interface (VSI) represents the virtual interface
of a virtual machine. <span class="since">Since 0.8.2</span>
</p>
<p>
Please note that IEEE 802.1Qbg requires a non-zero value for the
VLAN ID.
</p>
<dl>
<dt><code>managerid</code></dt>
<dd>The VSI Manager ID identifies the database containing the VSI type
and instance definitions. This is an integer value and the
value 0 is reserved.</dd>
<dt><code>typeid</code></dt>
<dd>The VSI Type ID identifies a VSI type characterizing the network
access. VSI types are typically managed by network administrator.
This is an integer value.
</dd>
<dt><code>typeidversion</code></dt>
<dd>The VSI Type Version allows multiple versions of a VSI Type.
This is an integer value.
</dd>
<dt><code>instanceid</code></dt>
<dd>The VSI Instance ID Identifier is generated when a VSI instance
(i.e. a virtual interface of a virtual machine) is created.
This is a globally unique identifier.
</dd>
</dl>
<pre>
...
&lt;devices&gt;
...
&lt;interface type='direct'&gt;
&lt;source dev='eth0.2' mode='vepa'/&gt;
&lt;virtualport type="802.1Qbg"&gt;
&lt;parameters managerid="11" typeid="1193047" typeidversion="2" instanceid="09b11c53-8b5c-4eeb-8f00-d84eaa0aaa4f"/&gt;
&lt;/virtualport&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
The interface can have additional parameters as shown below
if the switch is conforming to the IEEE 802.1Qbh standard.
The values are network specific and should be provided by the
network administrator. <span class="since">Since 0.8.2</span>
</p>
<dl>
<dt><code>profileid</code></dt>
<dd>The profile ID contains the name of the port profile that is to
be applied to this interface. This name is resolved by the port
profile database into the network parameters from the port profile,
and those network parameters will be applied to this interface.
</dd>
</dl>
<pre>
...
&lt;devices&gt;
...
&lt;interface type='direct'&gt;
&lt;source dev='eth0' mode='private'/&gt;
&lt;virtualport type='802.1Qbh'&gt;
&lt;parameters profileid='finance'/&gt;
&lt;/virtualport&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...
</pre>
<h5><a id="elementsNICSHostdev">PCI Passthrough</a></h5>
<p>
A PCI network device (specified by the &lt;source&gt; element)
is directly assigned to the guest using generic device
passthrough, after first optionally setting the device's MAC
address to the configured value, and associating the device with
an 802.1Qbh capable switch using an optionally specified
&lt;virtualport&gt; element (see the examples of virtualport
given above for type='direct' network devices). Note that - due
to limitations in standard single-port PCI ethernet card driver
design - only SR-IOV (Single Root I/O Virtualization) virtual
function (VF) devices can be assigned in this manner; to assign
a standard single-port PCI or PCIe ethernet card to a guest, use
the traditional &lt;hostdev&gt; device definition and
<span class="since">Since 0.9.11</span>
</p>
<p>
To use VFIO device assignment rather than traditional/legacy KVM
device assignment (VFIO is a new method of device assignment
that is compatible with UEFI Secure Boot), a type='hostdev'
interface can have an optional <code>driver</code> sub-element
with a <code>name</code> attribute set to "vfio". To use legacy
KVM device assignment you can set <code>name</code> to "kvm" (or
simply omit the <code>&lt;driver&gt;</code> element, since "kvm"
is currently the default).
<span class="since">Since 1.0.5 (QEMU and KVM only, requires kernel 3.6 or newer)</span>
</p>
<p>
Note that this "intelligent passthrough" of network devices is
very similar to the functionality of a standard &lt;hostdev&gt;
device, the difference being that this method allows specifying
a MAC address and &lt;virtualport&gt; for the passed-through
device. If these capabilities are not required, if you have a
standard single-port PCI, PCIe, or USB network card that doesn't
support SR-IOV (and hence would anyway lose the configured MAC
address during reset after being assigned to the guest domain),
or if you are using a version of libvirt older than 0.9.11, you
should use standard &lt;hostdev&gt; to assign the device to the
guest instead of &lt;interface type='hostdev'/&gt;.
</p>
<p>
Similar to the functionality of a standard &lt;hostdev&gt; device,
when <code>managed</code> is "yes", it is detached from the host
before being passed on to the guest, and reattached to the host
after the guest exits. If <code>managed</code> is omitted or "no",
the user is responsible to call <code>virNodeDeviceDettach</code>
(or <code>virsh nodedev-detach</code>) before starting the guest
or hot-plugging the device, and <code>virNodeDeviceReAttach</code>
(or <code>virsh nodedev-reattach</code>) after hot-unplug or
stopping the guest.
</p>
<pre>
...
&lt;devices&gt;
&lt;interface type='hostdev' managed='yes'&gt;
&lt;driver name='vfio'/&gt;
&lt;source&gt;
&lt;address type='pci' domain='0x0000' bus='0x00' slot='0x07' function='0x0'/&gt;
&lt;/source&gt;
&lt;mac address='52:54:00:6d:90:02'/&gt;
&lt;virtualport type='802.1Qbh'&gt;
&lt;parameters profileid='finance'/&gt;
&lt;/virtualport&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<h5><a id="elementsTeaming">Teaming a virtio/hostdev NIC pair</a></h5>
<p>
<span class="since">Since 6.1.0 (QEMU and KVM only, requires
QEMU 4.2.0 or newer and a guest virtio-net driver supporting
the "failover" feature, such as the one included in Linux
kernel 4.18 and newer)
</span>
The <code>&lt;teaming&gt;</code> element of two interfaces can
be used to connect them as a team/bond device in the guest
(assuming proper support in the hypervisor and the guest
network driver).
</p>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='mybridge'/&gt;
&lt;mac address='00:11:22:33:44:55'/&gt;
&lt;model type='virtio'/&gt;
&lt;teaming type='persistent'/&gt;
&lt;alias name='ua-backup0'/&gt;
&lt;/interface&gt;
&lt;interface type='network'&gt;
&lt;source network='hostdev-pool'/&gt;
&lt;mac address='00:11:22:33:44:55'/&gt;
&lt;model type='virtio'/&gt;
&lt;teaming type='transient' persistent='ua-backup0'/&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
The <code>&lt;teaming&gt;</code> element required
attribute <code>type</code> will be set to
either <code>"persistent"</code> to indicate a device that
should always be present in the domain,
or <code>"transient"</code> to indicate a device that may
periodically be removed, then later re-added to the domain. When
type="transient", there should be a second attribute
to <code>&lt;teaming&gt;</code> called <code>"persistent"</code>
- this attribute should be set to the alias name of the other
device in the pair (the one that has <code>&lt;teaming
type="persistent'/&gt;</code>).
</p>
<p>
In the particular case of QEMU,
libvirt's <code>&lt;teaming&gt;</code> element is used to setup
a virtio-net "failover" device pair. For this setup, the
persistent device must be an interface with <code>&lt;model
type="virtio"/&gt;</code>, and the transient device must
be <code>&lt;interface type='hostdev'/&gt;</code>
(or <code>&lt;interface type='network'/&gt;</code> where the
referenced network defines a pool of SRIOV VFs). The guest will
then have a simple network team/bond device made of the virtio
NIC + hostdev NIC pair. In this configuration, the
higher-performing hostdev NIC will normally be preferred for all
network traffic, but when the domain is migrated, QEMU will
automatically unplug the VF from the guest, and then hotplug a
similar device once migration is completed; while migration is
taking place, network traffic will use the virtio NIC. (Of
course the emulated virtio NIC and the hostdev NIC must be
connected to the same subnet for bonding to work properly).
</p>
<p>
NB1: Since you must know the alias name of the virtio NIC when
configuring the hostdev NIC, it will need to be manually set in
the virtio NIC's configuration (as with all other manually set
alias names, this means it must start with "ua-").
</p>
<p>
NB2: Currently the only implementation of the guest OS
virtio-net driver supporting virtio-net failover requires that
the MAC addresses of the virtio and hostdev NIC must
match. Since that may not always be a requirement in the future,
libvirt doesn't enforce this limitation - it is up to the
person/management application that is creating the configuration
to assure the MAC addresses of the two devices match.
</p>
<p>
NB3: Since the PCI addresses of the SRIOV VFs on the hosts that
are the source and destination of the migration will almost
certainly be different, either higher level management software
will need to modify the <code>&lt;source&gt;</code> of the
hostdev NIC (<code>&lt;interface type='hostdev'&gt;</code>) at
the start of migration, or (a simpler solution) the
configuration will need to use a libvirt "hostdev" virtual
network that maintains a pool of such devices, as is implied in
the example's use of the libvirt network named "hostdev-pool" -
as long as the hostdev network pools on both hosts have the same
name, libvirt itself will take care of allocating an appropriate
device on both ends of the migration. Similarly the XML for the
virtio interface must also either work correctly unmodified on
both the source and destination of the migration (e.g. by
connecting to the same bridge device on both hosts, or by using
the same virtual network), or the management software must
properly modify the interface XML during migration so that the
virtio device remains connected to the same network segment
before and after migration.
</p>
<h5><a id="elementsNICSMulticast">Multicast tunnel</a></h5>
<p>
A multicast group is setup to represent a virtual network. Any VMs
whose network devices are in the same multicast group can talk to each
other even across hosts. This mode is also available to unprivileged
users. There is no default DNS or DHCP support and no outgoing network
access. To provide outgoing network access, one of the VMs should have a
2nd NIC which is connected to one of the first 4 network types and do the
appropriate routing. The multicast protocol is compatible with that used
by user mode linux guests too. The source address used must be from the
multicast address block.
</p>
<pre>
...
&lt;devices&gt;
&lt;interface type='mcast'&gt;
&lt;mac address='52:54:00:6d:90:01'/&gt;
&lt;source address='230.0.0.1' port='5558'/&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<h5><a id="elementsNICSTCP">TCP tunnel</a></h5>
<p>
A TCP client/server architecture provides a virtual network. One VM
provides the server end of the network, all other VMS are configured as
clients. All network traffic is routed between the VMs via the server.
This mode is also available to unprivileged users. There is no default
DNS or DHCP support and no outgoing network access. To provide outgoing
network access, one of the VMs should have a 2nd NIC which is connected
to one of the first 4 network types and do the appropriate routing.</p>
<pre>
...
&lt;devices&gt;
&lt;interface type='server'&gt;
&lt;mac address='52:54:00:22:c9:42'/&gt;
&lt;source address='192.168.0.1' port='5558'/&gt;
&lt;/interface&gt;
...
&lt;interface type='client'&gt;
&lt;mac address='52:54:00:8b:c9:51'/&gt;
&lt;source address='192.168.0.1' port='5558'/&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<h5><a id="elementsNICSUDP">UDP unicast tunnel</a></h5>
<p>
A UDP unicast architecture provides a virtual network which enables
connections between QEMU instances using QEMU's UDP infrastructure.
The xml "source" address is the endpoint address to which the UDP socket
packets will be sent from the host running QEMU.
The xml "local" address is the address of the interface from which the
UDP socket packets will originate from the QEMU host.
<span class="since">Since 1.2.20</span></p>
<pre>
...
&lt;devices&gt;
&lt;interface type='udp'&gt;
&lt;mac address='52:54:00:22:c9:42'/&gt;
&lt;source address='127.0.0.1' port='11115'&gt;
&lt;local address='127.0.0.1' port='11116'/&gt;
&lt;/source&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<h5><a id="elementsNICSModel">Setting the NIC model</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
&lt;target dev='vnet1'/&gt;
<b>&lt;model type='ne2k_pci'/&gt;</b>
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
For hypervisors which support this, you can set the model of
emulated network interface card.
</p>
<p>
The values for <code>type</code> aren't defined specifically by
libvirt, but by what the underlying hypervisor supports (if
any). For QEMU and KVM you can get a list of supported models
with these commands:
</p>
<pre>
qemu -net nic,model=? /dev/null
qemu-kvm -net nic,model=? /dev/null
</pre>
<p>
Typical values for QEMU and KVM include:
ne2k_isa i82551 i82557b i82559er ne2k_pci pcnet rtl8139 e1000 virtio.
<span class="since">Since 5.2.0</span>, <code>virtio-transitional</code>
and <code>virtio-non-transitional</code> values are supported.
See <a href="#elementsVirtioTransitional">Virtio transitional devices</a>
for more details.
</p>
<h5><a id="elementsDriverBackendOptions">Setting NIC driver-specific options</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
&lt;target dev='vnet1'/&gt;
&lt;model type='virtio'/&gt;
<b>&lt;driver name='vhost' txmode='iothread' ioeventfd='on' event_idx='off' queues='5' rx_queue_size='256' tx_queue_size='256'&gt;
&lt;host csum='off' gso='off' tso4='off' tso6='off' ecn='off' ufo='off' mrg_rxbuf='off'/&gt;
&lt;guest csum='off' tso4='off' tso6='off' ecn='off' ufo='off'/&gt;
&lt;/driver&gt;
</b>&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
Some NICs may have tunable driver-specific options. These are
set as attributes of the <code>driver</code> sub-element of the
interface definition. Currently the following attributes are
available for the <code>"virtio"</code> NIC driver:
</p>
<dl>
<dt><code>name</code></dt>
<dd>
The optional <code>name</code> attribute forces which type of
backend driver to use. The value can be either 'qemu' (a
user-space backend) or 'vhost' (a kernel backend, which
requires the vhost module to be provided by the kernel); an
attempt to require the vhost driver without kernel support
will be rejected. If this attribute is not present, then the
domain defaults to 'vhost' if present, but silently falls back
to 'qemu' without error.
<span class="since">Since 0.8.8 (QEMU and KVM only)</span>
</dd>
<dd>
For interfaces of type='hostdev' (PCI passthrough devices)
the <code>name</code> attribute can optionally be set to
"vfio" or "kvm". "vfio" tells libvirt to use VFIO device
assignment rather than traditional KVM device assignment (VFIO
is a new method of device assignment that is compatible with
UEFI Secure Boot), and "kvm" tells libvirt to use the legacy
device assignment performed directly by the kvm kernel module
(the default is currently "kvm", but is subject to change).
<span class="since">Since 1.0.5 (QEMU and KVM only, requires
kernel 3.6 or newer)</span>
</dd>
<dd>
For interfaces of type='vhostuser', the <code>name</code>
attribute is ignored. The backend driver used is always
vhost-user.
</dd>
<dt><code>txmode</code></dt>
<dd>
The <code>txmode</code> attribute specifies how to handle
transmission of packets when the transmit buffer is full. The
value can be either 'iothread' or 'timer'.
<span class="since">Since 0.8.8 (QEMU and KVM only)</span><br/><br/>
If set to 'iothread', packet tx is all done in an iothread in
the bottom half of the driver (this option translates into
adding "tx=bh" to the qemu commandline -device virtio-net-pci
option).<br/><br/>
If set to 'timer', tx work is done in qemu, and if there is
more tx data than can be sent at the present time, a timer is
set before qemu moves on to do other things; when the timer
fires, another attempt is made to send more data.<br/><br/>
The resulting difference, according to the qemu developer who
added the option is: "bh makes tx more asynchronous and reduces
latency, but potentially causes more processor bandwidth
contention since the CPU doing the tx isn't necessarily the
CPU where the guest generated the packets."<br/><br/>
<b>In general you should leave this option alone, unless you
are very certain you know what you are doing.</b>
</dd>
<dt><code>ioeventfd</code></dt>
<dd>
This optional attribute allows users to set
<a href='https://patchwork.kernel.org/patch/43390/'>
domain I/O asynchronous handling</a> for interface device.
The default is left to the discretion of the hypervisor.
Accepted values are "on" and "off". Enabling this allows
qemu to execute VM while a separate thread handles I/O.
Typically guests experiencing high system CPU utilization
during I/O will benefit from this. On the other hand,
on overloaded host it could increase guest I/O latency.
<span class="since">Since 0.9.3 (QEMU and KVM only)</span><br/><br/>
<b>In general you should leave this option alone, unless you
are very certain you know what you are doing.</b>
</dd>
<dt><code>event_idx</code></dt>
<dd>
The <code>event_idx</code> attribute controls some aspects of
device event processing. The value can be either 'on' or 'off'
- if it is on, it will reduce the number of interrupts and
exits for the guest. The default is determined by QEMU;
usually if the feature is supported, default is on. In case
there is a situation where this behavior is suboptimal, this
attribute provides a way to force the feature off.
<span class="since">Since 0.9.5 (QEMU and KVM only)</span><br/><br/>
<b>In general you should leave this option alone, unless you
are very certain you know what you are doing.</b>
</dd>
<dt><code>queues</code></dt>
<dd>
The optional <code>queues</code> attribute controls the number
of queues to be used for either
<a href="https://www.linux-kvm.org/page/Multiqueue"> Multiqueue
virtio-net</a> or <a href="#elementVhostuser">vhost-user</a> network
interfaces. Use of multiple packet processing queues requires the
interface having the <code>&lt;model type='virtio'/&gt;</code>
element. Each queue will potentially be handled by a different
processor, resulting in much higher throughput.
<span class="since">virtio-net since 1.0.6 (QEMU and KVM only)</span>
<span class="since">vhost-user since 1.2.17 (QEMU and KVM only)</span>
</dd>
<dt><code>rx_queue_size</code></dt>
<dd>
The optional <code>rx_queue_size</code> attribute controls
the size of virtio ring for each queue as described above.
The default value is hypervisor dependent and may change
across its releases. Moreover, some hypervisors may pose
some restrictions on actual value. For instance, latest
QEMU (as of 2016-09-01) requires value to be a power of two
from [256, 1024] range.
<span class="since">Since 2.3.0 (QEMU and KVM only)</span><br/><br/>
<b>In general you should leave this option alone, unless you
are very certain you know what you are doing.</b>
</dd>
<dt><code>tx_queue_size</code></dt>
<dd>
The optional <code>tx_queue_size</code> attribute controls
the size of virtio ring for each queue as described above.
The default value is hypervisor dependent and may change
across its releases. Moreover, some hypervisors may pose
some restrictions on actual value. For instance, QEMU
v2.9 requires value to be a power of two from [256, 1024]
range. In addition to that, this may work only for a subset of
interface types, e.g. aforementioned QEMU enables this option
only for <code>vhostuser</code> type.
<span class="since">Since 3.7.0 (QEMU and KVM only)</span><br/><br/>
<b>In general you should leave this option alone, unless you
are very certain you know what you are doing.</b>
</dd>
<dt>virtio options</dt>
<dd>
For virtio interfaces,
<a href="#elementsVirtio">Virtio-specific options</a> can also be
set. (<span class="since">Since 3.5.0</span>)
</dd>
</dl>
<p>
Offloading options for the host and guest can be configured using
the following sub-elements:
</p>
<dl>
<dt><code>host</code></dt>
<dd>
The <code>csum</code>, <code>gso</code>, <code>tso4</code>,
<code>tso6</code>, <code>ecn</code> and <code>ufo</code>
attributes with possible values <code>on</code>
and <code>off</code> can be used to turn off host offloading options.
By default, the supported offloads are enabled by QEMU.
<span class="since">Since 1.2.9 (QEMU only)</span>
The <code>mrg_rxbuf</code> attribute can be used to control
mergeable rx buffers on the host side. Possible values are
<code>on</code> (default) and <code>off</code>.
<span class="since">Since 1.2.13 (QEMU only)</span>
</dd>
<dt><code>guest</code></dt>
<dd>
The <code>csum</code>, <code>tso4</code>,
<code>tso6</code>, <code>ecn</code> and <code>ufo</code>
attributes with possible values <code>on</code>
and <code>off</code> can be used to turn off guest offloading options.
By default, the supported offloads are enabled by QEMU.
<span class="since">Since 1.2.9 (QEMU only)</span>
</dd>
</dl>
<h5><a id="elementsBackendOptions">Setting network backend-specific options</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
&lt;target dev='vnet1'/&gt;
&lt;model type='virtio'/&gt;
<b>&lt;backend tap='/dev/net/tun' vhost='/dev/vhost-net'/&gt;</b>
&lt;driver name='vhost' txmode='iothread' ioeventfd='on' event_idx='off' queues='5'/&gt;
<b>&lt;tune&gt;
&lt;sndbuf&gt;1600&lt;/sndbuf&gt;
&lt;/tune&gt;</b>
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
For tuning the backend of the network, the <code>backend</code> element
can be used. The <code>vhost</code> attribute can override the default vhost
device path (<code>/dev/vhost-net</code>) for devices with <code>virtio</code> model.
The <code>tap</code> attribute overrides the tun/tap device path (default:
<code>/dev/net/tun</code>) for network and bridge interfaces. This does not work
in session mode. <span class="since">Since 1.2.9</span>
</p>
<p>
For tap devices there is also <code>sndbuf</code> element which can
adjust the size of send buffer in the host. <span class="since">Since
0.8.8</span>
</p>
<h5><a id="elementsNICSTargetOverride">Overriding the target element</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
<b>&lt;target dev='vnet1'/&gt;</b>
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
If no target is specified, certain hypervisors will
automatically generate a name for the created tun device. This
name can be manually specified, however the name <i>should not
start with either 'vnet', 'vif', 'macvtap', or 'macvlan'</i>,
which are prefixes reserved by libvirt and certain hypervisors.
Manually specified targets using these prefixes may be ignored.
</p>
<p>
Note that for LXC containers, this defines the name of the interface
on the host side. <span class="since">Since 1.2.7</span>, to define
the name of the device on the guest side, the <code>guest</code>
element should be used, as in the following snippet:
</p>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
<b>&lt;guest dev='myeth'/&gt;</b>
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<h5><a id="elementsNICSBoot">Specifying boot order</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
&lt;target dev='vnet1'/&gt;
<b>&lt;boot order='1'/&gt;</b>
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
For hypervisors which support this, you can set a specific NIC to
be used for network boot. The <code>order</code> attribute determines
the order in which devices will be tried during boot sequence. The
per-device <code>boot</code> elements cannot be used together with
general boot elements in
<a href="#elementsOSBIOS">BIOS bootloader</a> section.
<span class="since">Since 0.8.8</span>
</p>
<h5><a id="elementsNICSROM">Interface ROM BIOS configuration</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
&lt;target dev='vnet1'/&gt;
<b>&lt;rom bar='on' file='/etc/fake/boot.bin'/&gt;</b>
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
For hypervisors which support this, you can change how a PCI Network
device's ROM is presented to the guest. The <code>bar</code>
attribute can be set to "on" or "off", and determines whether
or not the device's ROM will be visible in the guest's memory
map. (In PCI documentation, the "rombar" setting controls the
presence of the Base Address Register for the ROM). If no rom
bar is specified, the qemu default will be used (older
versions of qemu used a default of "off", while newer qemus
have a default of "on").
The optional <code>file</code> attribute is used to point to a
binary file to be presented to the guest as the device's ROM
BIOS. This can be useful to provide an alternative boot ROM for a
network device.
<span class="since">Since 0.9.10 (QEMU and KVM only)</span>.
</p>
<h5><a id="elementDomain">Setting up a network backend in a driver domain</a></h5>
<pre>
...
&lt;devices&gt;
...
&lt;interface type='bridge'&gt;
&lt;source bridge='br0'/&gt;
<b>&lt;backenddomain name='netvm'/&gt;</b>
&lt;/interface&gt;
...
&lt;/devices&gt;
...</pre>
<p>
The optional <code>backenddomain</code> element allows specifying a
backend domain (aka driver domain) for the interface. Use the
<code>name</code> attribute to specify the backend domain name. You
can use it to create a direct network link between domains (so data
will not go through host system). Use with type 'ethernet' to create
plain network link, or with type 'bridge' to connect to a bridge inside
the backend domain.
<span class="since">Since 1.2.13 (Xen only)</span>
</p>
<h5><a id="elementQoS">Quality of service</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
&lt;target dev='vnet0'/&gt;
<b>&lt;bandwidth&gt;
&lt;inbound average='1000' peak='5000' floor='200' burst='1024'/&gt;
&lt;outbound average='128' peak='256' burst='256'/&gt;
&lt;/bandwidth&gt;</b>
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
This part of interface XML provides setting quality of service. Incoming
and outgoing traffic can be shaped independently.
The <code>bandwidth</code> element and its child elements are described
in the <a href="formatnetwork.html#elementQoS">QoS</a> section of
the Network XML.
</p>
<h5><a id="elementVlanTag">Setting VLAN tag (on supported network types only)</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='bridge'&gt;
<b>&lt;vlan&gt;</b>
<b>&lt;tag id='42'/&gt;</b>
<b>&lt;/vlan&gt;</b>
&lt;source bridge='ovsbr0'/&gt;
&lt;virtualport type='openvswitch'&gt;
&lt;parameters interfaceid='09b11c53-8b5c-4eeb-8f00-d84eaa0aaa4f'/&gt;
&lt;/virtualport&gt;
&lt;/interface&gt;
&lt;interface type='bridge'&gt;
<b>&lt;vlan trunk='yes'&gt;</b>
<b>&lt;tag id='42'/&gt;</b>
<b>&lt;tag id='123' nativeMode='untagged'/&gt;</b>
<b>&lt;/vlan&gt;</b>
...
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
If (and only if) the network connection used by the guest
supports VLAN tagging transparent to the guest, an
optional <code>&lt;vlan&gt;</code> element can specify one or
more VLAN tags to apply to the guest's network
traffic <span class="since">Since 0.10.0</span>. Network
connections that support guest-transparent VLAN tagging include
1) type='bridge' interfaces connected to an Open vSwitch bridge
<span class="since">Since 0.10.0</span>, 2) SRIOV Virtual
Functions (VF) used via type='hostdev' (direct device
assignment) <span class="since">Since 0.10.0</span>, and 3)
SRIOV VFs used via type='direct' with mode='passthrough'
(macvtap "passthru" mode) <span class="since">Since
1.3.5</span>. All other connection types, including standard
linux bridges and libvirt's own virtual networks, <b>do not</b>
support it. 802.1Qbh (vn-link) and 802.1Qbg (VEPA) switches
provide their own way (outside of libvirt) to tag guest traffic
onto a specific VLAN. Each tag is given in a
separate <code>&lt;tag&gt;</code> subelement
of <code>&lt;vlan&gt;</code> (for example: <code>&lt;tag
id='42'/&gt;</code>). For VLAN trunking of multiple tags (which
is supported only on Open vSwitch connections),
multiple <code>&lt;tag&gt;</code> subelements can be specified,
which implies that the user wants to do VLAN trunking on the
interface for all the specified tags. In the case that VLAN
trunking of a single tag is desired, the optional
attribute <code>trunk='yes'</code> can be added to the toplevel
<code>&lt;vlan&gt;</code> element to differentiate trunking of a
single tag from normal tagging.
</p>
<p>
For network connections using Open vSwitch it is also possible
to configure 'native-tagged' and 'native-untagged' VLAN modes
<span class="since">Since 1.1.0.</span> This is done with the
optional <code>nativeMode</code> attribute on
the <code>&lt;tag&gt;</code> subelement: <code>nativeMode</code>
may be set to 'tagged' or 'untagged'. The <code>id</code>
attribute of the <code>&lt;tag&gt;</code> subelement
containing <code>nativeMode</code> sets which VLAN is considered
to be the "native" VLAN for this interface, and
the <code>nativeMode</code> attribute determines whether or not
traffic for that VLAN will be tagged.
</p>
<h5><a id="elementLink">Modifying virtual link state</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
&lt;target dev='vnet0'/&gt;
<b>&lt;link state='down'/&gt;</b>
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
This element provides means of setting state of the virtual network link.
Possible values for attribute <code>state</code> are <code>up</code> and
<code>down</code>. If <code>down</code> is specified as the value, the interface
behaves as if it had the network cable disconnected. Default behavior if this
element is unspecified is to have the link state <code>up</code>.
<span class="since">Since 0.9.5</span>
</p>
<h5><a id="mtu">MTU configuration</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
&lt;target dev='vnet0'/&gt;
<b>&lt;mtu size='1500'/&gt;</b>
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
This element provides means of setting MTU of the virtual network link.
Currently there is just one attribute <code>size</code> which accepts a
non-negative integer which specifies the MTU size for the interface.
<span class="since">Since 3.1.0</span>
</p>
<h5><a id="coalesce">Coalesce settings</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
&lt;target dev='vnet0'/&gt;
<b>&lt;coalesce&gt;
&lt;rx&gt;
&lt;frames max='7'/&gt;
&lt;/rx&gt;
&lt;/coalesce&gt;</b>
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
This element provides means of setting coalesce settings for
some interface devices (currently only type <code>network</code>
and <code>bridge</code>. Currently there is just one attribute,
<code>max</code>, to tweak, in element <code>frames</code> for
the <code>rx</code> group, which accepts a non-negative integer
that specifies the maximum number of packets that will be
received before an interrupt.
<span class="since">Since 3.3.0</span>
</p>
<h5><a id="ipconfig">IP configuration</a></h5>
<pre>
...
&lt;devices&gt;
&lt;interface type='network'&gt;
&lt;source network='default'/&gt;
&lt;target dev='vnet0'/&gt;
<b>&lt;ip address='192.168.122.5' prefix='24'/&gt;</b>
<b>&lt;ip address='192.168.122.5' prefix='24' peer='10.0.0.10'/&gt;</b>
<b>&lt;route family='ipv4' address='192.168.122.0' prefix='24' gateway='192.168.122.1'/&gt;</b>
<b>&lt;route family='ipv4' address='192.168.122.8' gateway='192.168.122.1'/&gt;</b>
&lt;/interface&gt;
...
&lt;hostdev mode='capabilities' type='net'&gt;
&lt;source&gt;
&lt;interface&gt;eth0&lt;/interface&gt;
&lt;/source&gt;
<b>&lt;ip address='192.168.122.6' prefix='24'/&gt;</b>
<b>&lt;route family='ipv4' address='192.168.122.0' prefix='24' gateway='192.168.122.1'/&gt;</b>
<b>&lt;route family='ipv4' address='192.168.122.8' gateway='192.168.122.1'/&gt;</b>
&lt;/hostdev&gt;
...
&lt;/devices&gt;
...
</pre>
<p>
<span class="since">Since 1.2.12</span> network devices and
hostdev devices with network capabilities can optionally be provided
one or more IP addresses to set on the network device in the
guest. Note that some hypervisors or network device types will
simply ignore them or only use the first one.
The <code>family</code> attribute can be set to
either <code>ipv4</code> or <code>ipv6</code>, and the
<code>address</code> attribute contains the IP address. The
optional <code>prefix</code> is the number of 1 bits in the
netmask, and will be automatically set if not specified - for
IPv4 the default prefix is determined according to the network
"class" (A, B, or C - see RFC870), and for IPv6 the default
prefix is 64. The optional <code>peer</code> attribute holds the
IP address of the other end of a point-to-point network
device <span class="since">(since 2.1.0)</span>.
</p>
<p>
<span class="since">Since 1.2.12</span> route elements can also be
added to define IP routes to add in the guest. The attributes of
this element are described in the documentation for
the <code>route</code> element
in <a href="formatnetwork.html#elementsStaticroute">network
definitions</a>. This is used by the LXC driver.
</p>
<pre>
...
&lt;devices&gt;
&lt;interface type='ethernet'&gt;
<b>&lt;source/&gt;</b>
<b>&lt;ip address='192.168.123.1' prefix='24'/&gt;</b>
<b>&lt;ip address='10.0.0.10' prefix='24' peer='192.168.122.5'/&gt;</b>
<b>&lt;route family='ipv4' address='192.168.42.0' prefix='24' gateway='192.168.123.4'/&gt;</b>
<b>&lt;source/&gt;</b>
...
&lt;/interface&gt;
...
&lt;/devices&gt;
...
</pre>
<p>
<span class="since">Since 2.1.0</span> network devices of type
"ethernet" can optionally be provided one or more IP addresses
and one or more routes to set on the <b>host</b> side of the
network device. These are configured as subelements of
the <code>&lt;source&gt;</code> element of the interface, and
have the same attributes as the similarly named elements used to
configure the guest side of the interface (described above).
</p>
<h5><a id="elementVhostuser">vhost-user interface</a></h5>
<p>
<span class="since">Since 1.2.7</span> the vhost-user enables the
communication between a QEMU virtual machine and other userspace process
using the Virtio transport protocol. A char dev (e.g. Unix socket) is used
for the control plane, while the data plane is based on shared memory.
</p>
<pre>
...
&lt;devices&gt;
&lt;interface type='vhostuser'&gt;
&lt;mac address='52:54:00:3b:83:1a'/&gt;
&lt;source type='unix' path='/tmp/vhost1.sock' mode='server'/&gt;
&lt;model type='virtio'/&gt;
&lt;/interface&gt;
&lt;interface type='vhostuser'&gt;
&lt;mac address='52:54:00:3b:83:1b'/&gt;
&lt;source type='unix' path='/tmp/vhost2.sock' mode='client'&gt;
&lt;reconnect enabled='yes' timeout='10'/&gt;
&lt;/source&gt;
&lt;model type='virtio'/&gt;
&lt;driver queues='5'/&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
The <code>&lt;source&gt;</code> element has to be specified
along with the type of char device.
Currently, only type='unix' is supported, where the path (the
directory path of the socket) and mode attributes are required.
Both <code>mode='server'</code> and <code>mode='client'</code>
are supported.
vhost-user requires the virtio model type, thus the
<code>&lt;model&gt;</code> element is mandatory.
<span class="since">Since 4.1.0</span> the element has an
optional child element <code>reconnect</code> which
configures reconnect timeout if the connection is lost. It
has two attributes <code>enabled</code> (which accepts
<code>yes</code> and <code>no</code>) and
<code>timeout</code> which specifies the amount of seconds
after which hypervisor tries to reconnect.
</p>
<h5><a id="elementNwfilter">Traffic filtering with NWFilter</a></h5>
<p>
<span class="since">Since 0.8.0</span> an <code>nwfilter</code> profile
can be assigned to a domain interface, which allows configuring
traffic filter rules for the virtual machine.
See the <a href="formatnwfilter.html">nwfilter</a> documentation for more
complete details.
</p>
<pre>
...
&lt;devices&gt;
&lt;interface ...&gt;
...
&lt;filterref filter='clean-traffic'/&gt;
&lt;/interface&gt;
&lt;interface ...&gt;
...
&lt;filterref filter='myfilter'&gt;
&lt;parameter name='IP' value='104.207.129.11'/&gt;
&lt;parameter name='IP6_ADDR' value='2001:19f0:300:2102::'/&gt;
&lt;parameter name='IP6_MASK' value='64'/&gt;
...
&lt;/filterref&gt;
&lt;/interface&gt;
&lt;/devices&gt;
...</pre>
<p>
The <code>filter</code> attribute specifies the name of the nwfilter
to use. Optional <code>&lt;parameter&gt;</code> elements may be
specified for passing additional info to the nwfilter via the
<code>name</code> and <code>value</code> attributes. See
the <a href="formatnwfilter.html#nwfconceptsvars">nwfilter</a>
docs for info on parameters.
</p>
<h4><a id="elementsInput">Input devices</a></h4>
<p>
Input devices allow interaction with the graphical framebuffer
in the guest virtual machine. When enabling the framebuffer, an
input device is automatically provided. It may be possible to
add additional devices explicitly, for example,
to provide a graphics tablet for absolute cursor movement.
</p>
<pre>
...
&lt;devices&gt;
&lt;input type='mouse' bus='usb'/&gt;
&lt;input type='keyboard' bus='usb'/&gt;
&lt;input type='mouse' bus='virtio'/&gt;
&lt;input type='keyboard' bus='virtio'/&gt;
&lt;input type='tablet' bus='virtio'/&gt;
&lt;input type='passthrough' bus='virtio'&gt;
&lt;source evdev='/dev/input/event1/&gt;
&lt;/input&gt;
&lt;/devices&gt;
...</pre>
<dl>
<dt><code>input</code></dt>
<dd>The <code>input</code> element has one mandatory attribute,
the <code>type</code> whose value can be 'mouse', 'tablet',
(<span class="since">since 1.2.2</span>) 'keyboard' or
(<span class="since">since 1.3.0</span>) 'passthrough'.
The tablet provides absolute cursor movement,
while the mouse uses relative movement. The optional
<code>bus</code> attribute can be used to refine the exact device type.
It takes values "xen" (paravirtualized), "ps2" and "usb" or
(<span class="since">since 1.3.0</span>) "virtio".</dd>
</dl>
<p>
The <code>input</code> element has an optional
sub-element <code>&lt;address&gt;</code> which can tie the
device to a particular PCI
slot, <a href="#elementsAddress">documented above</a>.
On S390, <code>address</code> can be used to provide a CCW address for
an input device (<span class="since">since 4.2.0</span>).
For type <code>passthrough</code>, the mandatory sub-element <code>source</code>
must have an <code>evdev</code> attribute containing the absolute path to the
event device passed through to guests. (KVM only)
<span class="since">Since 5.2.0</span>, the <code>input</code> element
accepts a <code>model</code> attribute which has the values 'virtio',
'virtio-transitional' and 'virtio-non-transitional'. See
<a href="#elementsVirtioTransitional">Virtio transitional devices</a>
for more details.
</p>
<p>
The subelement <code>driver</code> can be used to tune the virtio
options of the device:
<a href="#elementsVirtio">Virtio-specific options</a> can also be
set. (<span class="since">Since 3.5.0</span>)
</p>
<h4><a id="elementsHub">Hub devices</a></h4>
<p>
A hub is a device that expands a single port into several so
that there are more ports available to connect devices to a host
system.
</p>
<pre>
...
&lt;devices&gt;
&lt;hub type='usb'/&gt;
&lt;/devices&gt;
...</pre>
<dl>
<dt><code>hub</code></dt>
<dd>The <code>hub</code> element has one mandatory attribute,
the <code>type</code> whose value can only be 'usb'.</dd>
</dl>
<p>
The <code>hub</code> element has an optional
sub-element <code>&lt;address&gt;</code>
with <code>type='usb'</code>which can tie the device to a
particular controller, <a href="#elementsAddress">documented
above</a>.
</p>
<h4><a id="elementsGraphics">Graphical framebuffers</a></h4>
<p>
A graphics device allows for graphical interaction with the
guest OS. A guest will typically have either a framebuffer
or a text console configured to allow interaction with the
admin.
</p>
<pre>
...
&lt;devices&gt;
&lt;graphics type='sdl' display=':0.0'/&gt;
&lt;graphics type='vnc' port='5904' sharePolicy='allow-exclusive'&gt;
&lt;listen type='address' address='1.2.3.4'/&gt;
&lt;/graphics&gt;
&lt;graphics type='rdp' autoport='yes' multiUser='yes' /&gt;
&lt;graphics type='desktop' fullscreen='yes'/&gt;
&lt;graphics type='spice'&gt;
&lt;listen type='network' network='rednet'/&gt;
&lt;/graphics&gt;
&lt;/devices&gt;
...</pre>
<dl>
<dt><code>graphics</code></dt>
<dd>
<p>
The <code>graphics</code> element has a mandatory <code>type</code>
attribute which takes the value <code>sdl</code>, <code>vnc</code>,
<code>spice</code>, <code>rdp</code>, <code>desktop</code> or
<code>egl-headless</code>:
</p>
<dl>
<dt><code>sdl</code></dt>
<dd>
<p>
This displays a window on the host desktop, it can take 3 optional
arguments: a <code>display</code> attribute for the display to use,
an <code>xauth</code> attribute for the authentication identifier,
and an optional <code>fullscreen</code> attribute accepting values
<code>yes</code> or <code>no</code>.
</p>
<p>
You can use a <code>gl</code> with the <code>enable="yes"</code>
property to enable OpenGL support in SDL. Likewise you can
explicitly disable OpenGL support with <code>enable="no"</code>.
</p>
</dd>
<dt><code>vnc</code></dt>
<dd>
<p>
Starts a VNC server. The <code>port</code> attribute specifies
the TCP port number (with -1 as legacy syntax indicating that it
should be auto-allocated). The <code>autoport</code> attribute is
the new preferred syntax for indicating auto-allocation of the TCP
port to use. The <code>passwd</code> attribute provides a VNC
password in clear text. If the <code>passwd</code> attribute is
set to an empty string, then VNC access is disabled. The
<code>keymap</code> attribute specifies the keymap to use. It is
possible to set a limit on the validity of the password by giving
a timestamp <code>passwdValidTo='2010-04-09T15:51:00'</code>
assumed to be in UTC. The <code>connected</code> attribute allows
control of connected client during password changes. VNC accepts
<code>keep</code> value only <span class="since">since 0.9.3</span>.
NB, this may not be supported by all hypervisors.
</p>
<p>
The optional <code>sharePolicy</code> attribute specifies vnc
server display sharing policy. <code>allow-exclusive</code> allows
clients to ask for exclusive access by dropping other connections.
Connecting multiple clients in parallel requires all clients asking
for a shared session (vncviewer: -Shared switch). This is
the default value. <code>force-shared</code> disables exclusive
client access, every connection has to specify -Shared switch for
vncviewer. <code>ignore</code> welcomes every connection
unconditionally <span class="since">since 1.0.6</span>.
</p>
<p>
Rather than using listen/port, QEMU supports a <code>socket</code>
attribute for listening on a unix domain socket path
<span class="since">Since 0.8.8</span>.
</p>
<p>
For VNC WebSocket functionality, <code>websocket</code> attribute
may be used to specify port to listen on (with -1 meaning
auto-allocation and <code>autoport</code> having no effect due to
security reasons) <span class="since">Since 1.0.6</span>.
</p>
<p>
Although VNC doesn't support OpenGL natively, it can be paired
with graphics type <code>egl-headless</code> (see below) which
will instruct QEMU to open and use drm nodes for OpenGL rendering.
</p>
</dd>
<dt><code>spice</code> <span class="since">Since 0.8.6</span></dt>
<dd>
<p>
Starts a SPICE server. The <code>port</code> attribute specifies
the TCP port number (with -1 as legacy syntax indicating that it
should be auto-allocated), while <code>tlsPort</code> gives
an alternative secure port number. The <code>autoport</code>
attribute is the new preferred syntax for indicating
auto-allocation of needed port numbers. The <code>passwd</code>
attribute provides a SPICE password in clear text. If the
<code>passwd</code> attribute is set to an empty string, then
SPICE access is disabled. The <code>keymap</code> attribute
specifies the keymap to use. It is possible to set a limit on
the validity of the password by giving a timestamp
<code>passwdValidTo='2010-04-09T15:51:00'</code> assumed to be
in UTC.
</p>
<p>
The <code>connected</code> attribute allows control of connected
client during password changes. SPICE accepts <code>keep</code> to
keep client connected, <code>disconnect</code> to disconnect client
and <code>fail</code> to fail changing password . NB, this may not
be supported by all hypervisors.
<span class="since">Since 0.9.3</span>
</p>
<p>
The <code>defaultMode</code> attribute sets the default channel
security policy, valid values are <code>secure</code>,
<code>insecure</code> and the default <code>any</code> (which is
secure if possible, but falls back to insecure rather than erroring
out if no secure path is available).
<span class="since">Since 0.9.12</span>
</p>
<p>
When SPICE has both a normal and TLS secured TCP port configured,
it can be desirable to restrict what channels can be run on each
port. This is achieved by adding one or more <code>&lt;channel&gt;
</code> elements inside the main <code>&lt;graphics&gt;</code>
element and setting the <code>mode</code> attribute to either
<code>secure</code> or <code>insecure</code>. Setting the mode
attribute overrides the default value as set by
the <code>defaultMode</code> attribute. (Note that specifying
<code>any</code> as mode discards the entry as the channel would
inherit the default mode anyways.) Valid channel names include
<code>main</code>, <code>display</code>, <code>inputs</code>,
<code>cursor</code>, <code>playback</code>, <code>record</code>
(all <span class="since"> since 0.8.6</span>);
<code>smartcard</code> (<span class="since">since 0.8.8</span>);
and <code>usbredir</code> (<span class="since">since 0.9.12</span>).
</p>
<pre>
&lt;graphics type='spice' port='-1' tlsPort='-1' autoport='yes'&gt;
&lt;channel name='main' mode='secure'/&gt;
&lt;channel name='record' mode='insecure'/&gt;
&lt;image compression='auto_glz'/&gt;
&lt;streaming mode='filter'/&gt;
&lt;clipboard copypaste='no'/&gt;
&lt;mouse mode='client'/&gt;
&lt;filetransfer enable='no'/&gt;
&lt;gl enable='yes' rendernode='/dev/dri/by-path/pci-0000:00:02.0-render'/&gt;
&lt;/graphics&gt;</pre>
<p>
Spice supports variable compression settings for audio, images and
streaming. These settings are accessible via the <code>compression
</code> attribute in all following elements: <code>image</code> to
set image compression (accepts <code>auto_glz</code>,
<code>auto_lz</code>, <code>quic</code>, <code>glz</code>,
<code>lz</code>, <code>off</code>), <code>jpeg</code> for JPEG
compression for images over wan (accepts <code>auto</code>,
<code>never</code>, <code>always</code>), <code>zlib</code> for
configuring wan image compression (accepts <code>auto</code>,
<code>never</code>, <code>always</code>) and <code>playback</code>
for enabling audio stream compression (accepts <code>on</code> or
<code>off</code>). <span class="since">Since 0.9.1</span>
</p>
<p>
Streaming mode is set by the <code>streaming</code> element,
settings its <code>mode</code> attribute to one of
<code>filter</code>, <code>all</code> or <code>off</code>.
<span class="since">Since 0.9.2</span>
</p>
<p>
Copy &amp; Paste functionality (via Spice agent) is set by the
<code>clipboard</code> element. It is enabled by default, and can
be disabled by setting the <code>copypaste</code> property to
<code>no</code>. <span class="since">Since 0.9.3</span>
</p>
<p>
Mouse mode is set by the <code>mouse</code> element, setting its
<code>mode</code> attribute to one of <code>server</code> or
<code>client</code>. If no mode is specified, the qemu default will
be used (client mode). <span class="since">Since 0.9.11</span>
</p>
<p>
File transfer functionality (via Spice agent) is set using the
<code>filetransfer</code> element. It is enabled by default, and
can be disabled by setting the <code>enable</code> property to
<code>no</code>. <span class="since">Since 1.2.2</span>
</p>
<p>
Spice may provide accelerated server-side rendering with OpenGL.
You can enable or disable OpenGL support explicitly with
the <code>gl</code> element, by setting the <code>enable</code>
property. (QEMU only, <span class="since">since 1.3.3</span>).
Note that this only works locally, since this requires usage of
UNIX sockets, i.e. using <code>listen</code> types 'socket' or
'none'. For accelerated OpenGL with remote support, consider
pairing this element with type <code>egl-headless</code>
(see below). However, this will deliver weaker performance
compared to native Spice OpenGL support.
</p>
<p>
By default, QEMU will pick the first available GPU DRM render node.
You may specify a DRM render node path to use instead. (QEMU only,
<span class="since">since 3.1.0</span>).
</p>
</dd>
<dt><code>rdp</code></dt>
<dd>
<p>
Starts a RDP server. The <code>port</code> attribute specifies the
TCP port number (with -1 as legacy syntax indicating that it should
be auto-allocated). The <code>autoport</code> attribute is the new
preferred syntax for indicating auto-allocation of the TCP port to
use. In the VirtualBox driver, the <code>autoport</code> will make
the hypervisor pick available port from 3389-3689 range when the VM
is started. The chosen port will be reflected in the <code>port</code>
attribute. The <code>multiUser</code> attribute is a boolean deciding
whether multiple simultaneous connections to the VM are permitted.
The <code>replaceUser</code> attribute is a boolean deciding whether
the existing connection must be dropped and a new connection must
be established by the VRDP server, when a new client connects in
single connection mode.
</p>
</dd>
<dt><code>desktop</code></dt>
<dd>
<p>
This value is reserved for VirtualBox domains for the moment. It
displays a window on the host desktop, similarly to "sdl", but
using the VirtualBox viewer. Just like "sdl", it accepts
the optional attributes <code>display</code> and
<code>fullscreen</code>.
</p>
</dd>
<dt><code>egl-headless</code><span class="since">Since 4.6.0</span></dt>
<dd>
<p>
This display type provides support for an OpenGL accelerated
display accessible both locally and remotely (for comparison,
Spice's native OpenGL support only works locally using UNIX
sockets at the moment, but has better performance). Since this
display type doesn't provide any window or graphical console like
the other types, for practical reasons it should be paired with
either <code>vnc</code> or <code>spice</code> graphics types.
This display type is only supported by QEMU domains
(needs QEMU <span class="since">2.10</span> or newer).
<span class="Since">5.0.0</span> this element accepts a
<code>&lt;gl/&gt;</code> sub-element with an optional attribute
<code>rendernode</code> which can be used to specify an absolute
path to a host's DRI device to be used for OpenGL rendering.
</p>
<pre>
&lt;graphics type='spice' autoport='yes'/&gt;
&lt;graphics type='egl-headless'&gt;
&lt;gl rendernode='/dev/dri/renderD128'/&gt;
&lt;/graphics&gt;
</pre>
</dd>
</dl>
</dd>
</dl>
<p>
Graphics device uses a <code>&lt;listen&gt;</code> to set up where
the device should listen for clients. It has a mandatory attribute
<code>type</code> which specifies the listen type. Only <code>vnc</code>,
<code>spice</code> and <code>rdp</code> supports <code>&lt;listen&gt;
</code> element. <span class="since">Since 0.9.4</span>.
Available types are:
</p>
<dl>
<dt><code>address</code></dt>
<dd>
<p>
Tells a graphics device to use an address specified in the
<code>address</code> attribute, which will contain either an IP address
or hostname (which will be resolved to an IP address via a DNS query)
to listen on.
</p>
<p>
It is possible to omit the <code>address</code> attribute in order to
use an address from config files <span class="since">Since 1.3.5</span>.
</p>
<p>
The <code>address</code> attribute is duplicated as <code>listen</code>
attribute in <code>graphics</code> element for backward compatibility.
If both are provided they must be equal.
</p>
</dd>
<dt><code>network</code></dt>
<dd>
<p>
This is used to specify an existing network in the <code>network</code>
attribute from libvirt's list of configured networks. The named network
configuration will be examined to determine an appropriate listen
address and the address will be stored in live XML in <code>address
</code> attribute. For example, if the network has an IPv4 address in
its configuration (e.g. if it has a forward type of <code>route</code>,
<code>nat</code>, or no forward type (isolated)), the first IPv4
address listed in the network's configuration will be used.
If the network is describing a host bridge, the first IPv4 address
associated with that bridge device will be used, and if the network is
describing one of the 'direct' (macvtap) modes, the first IPv4 address
of the first forward dev will be used.
</p>
</dd>
<dt><code>socket</code> <span class="since">since 2.0.0 (QEMU only)</span></dt>
<dd>
<p>
This listen type tells a graphics server to listen on unix socket.
Attribute <code>socket</code> contains a path to unix socket. If this
attribute is omitted libvirt will generate this path for you.
Supported by graphics type <code>vnc</code> and <code>spice</code>.
</p>
<p>
For <code>vnc</code> graphics be backward compatible
the <code>socket</code> attribute of first <code>listen</code> element
is duplicated as <code>socket</code> attribute in <code>graphics</code>
element. If <code>graphics</code> element contains a <code>socket</code>
attribute all <code>listen</code> elements are ignored.
</p>
</dd>
<dt><code>none</code> <span class="since">since 2.0.0 (QEMU only)</span></dt>
<dd>
<p>
This listen type doesn't have any other attribute. Libvirt supports
passing a file descriptor through our APIs virDomainOpenGraphics() and
virDomainOpenGraphicsFD(). No other listen types are allowed if this
one is used and the graphics device doesn't listen anywhere. You need
to use one of the two APIs to pass a FD to QEMU in order to connect to
this graphics device. Supported by graphics type <code>vnc</code> and
<code>spice</code>.
</p>
</dd>
</dl>
<h4><a id="elementsVideo">Video devices</a></h4>
<p>
A video device.
</p>
<pre>
...
&lt;devices&gt;
&lt;video&gt;
&lt;model type='vga' vram='16384' heads='1'&gt;
&lt;acceleration accel3d='yes' accel2d='yes'/&gt;
&lt;/model&gt;
&lt;driver name='qemu'/&gt;
&lt;/video&gt;
&lt;/devices&gt;
...</pre>
<dl>
<dt><code>video</code></dt>
<dd>
<p>
The <code>video</code> element is the container for describing
video devices. For backwards compatibility, if no <code>video</code>
is set but there is a <code>graphics</code> in domain xml, then
libvirt will add a default <code>video</code> according to the guest
type.
</p>
<p>
For a guest of type "kvm", the default <code>video</code> is:
<code>type</code> with value "cirrus", <code>vram</code> with value
"16384" and <code>heads</code> with value "1". By default, the first
video device in domain xml is the primary one, but the optional
attribute <code>primary</code> (<span class="since">since 1.0.2</span>)
with value 'yes' can be used to mark the primary in cases of multiple
video device. The non-primary must be type of "qxl" or
(<span class="since">since 2.4.0</span>) "virtio".
</p>
</dd>
<dt><code>model</code></dt>
<dd>
<p>
The <code>model</code> element has a mandatory <code>type</code>
attribute which takes the value "vga", "cirrus", "vmvga", "xen",
"vbox", "qxl" (<span class="since">since 0.8.6</span>),
"virtio" (<span class="since">since 1.3.0</span>),
"gop" (<span class="since">since 3.2.0</span>),
"bochs" (<span class="since">since 5.6.0</span>), "ramfb"
(<span class="since">since 5.9.0</span>), or "none"
(<span class="since">since 4.6.0</span>, depending on the hypervisor
features available.
The purpose of the type <code>none</code> is to instruct libvirt not
to add a default video device in the guest (see the paragraph above).
This legacy behaviour can be inconvenient in cases where GPU mediated
devices are meant to be the only rendering device within a guest and
so specifying another <code>video</code> device along with type
<code>none</code>.
Refer to <a id="elementsHostDev">Host device assignment</a> to see
how to add a mediated device into a guest.
</p>
<p>
You can provide the amount of video memory in kibibytes (blocks of
1024 bytes) using <code>vram</code>. This is supported only for guest
type of "libxl", "vz", "qemu", "vbox", "vmx" and "xen". If no
value is provided the default is used. If the size is not a power of
two it will be rounded to closest one.
</p>
<p>
The number of screen can be set using <code>heads</code>. This is
supported only for guests type of "vz", "kvm", "vbox" and "vmx".
</p>
<p>
For guest type of "kvm" or "qemu" and model type "qxl" there are
optional attributes. Attribute <code>ram</code> (<span class="since">
since 1.0.2</span>) specifies the size of the primary bar, while the
attribute <code>vram</code> specifies the secondary bar size.
If <code>ram</code> or <code>vram</code> are not supplied a default
value is used. The <code>ram</code> should also be rounded to power of
two as <code>vram</code>. There is also optional attribute
<code>vgamem</code> (<span class="since">since 1.2.11</span>) to set
the size of VGA framebuffer for fallback mode of QXL device.
Attribute <code>vram64</code> (<span class="since">since 1.3.3</span>)
extends secondary bar and makes it addressable as 64bit memory.
</p>
<p><span class="since">Since 5.9.0</span>, the <code>model</code>
element may also have an optional <code>resolution</code> sub-element.
The <code>resolution</code> element has attributes <code>x</code> and
<code>y</code> to set the minimum resolution for the video device. This
sub-element is valid for model types "vga", "qxl", "bochs", and
"virtio".
</p>
</dd>
<dt><code>acceleration</code></dt>
<dd>
Configure if video acceleration should be enabled.
<dl>
<dt><code>accel2d</code></dt>
<dd>Enable 2D acceleration (for vbox driver only,
<span class="since">since 0.7.1</span>)</dd>
<dt><code>accel3d</code></dt>
<dd>Enable 3D acceleration (for vbox driver
<span class="since">since 0.7.1</span>, qemu driver
<span class="since">since 1.3.0</span>)</dd>
<dt><code>rendernode</code></dt>
<dd>Absolute path to a host's DRI device to be used for
rendering (for 'vhostuser' driver only, <span
class="since">since 5.8.0</span>). If none is specified,
libvirt will pick one available.</dd>
</dl>
</dd>
<dt><code>address</code></dt>
<dd>
The optional <code>address</code> sub-element can be used to
tie the video device to a particular PCI slot.
On S390, <code>address</code> can be used to provide the
CCW address for the video device (<span class="since">
since 4.2.0</span>).
</dd>
<dt><code>driver</code></dt>
<dd>
The subelement <code>driver</code> can be used to tune the device:
<dl>
<dt><code>name</code></dt>
<dd>
Specify the backend driver to use, either "qemu" or
"vhostuser" depending on the hypervisor features available
(<span class="since">since 5.8.0</span>). "qemu" is the
default QEMU backend. "vhostuser" will use a separate
vhost-user process backend (for <code>virtio</code>
device).
</dd>
<dt>virtio options</dt>
<dd>
<a href="#elementsVirtio">Virtio-specific options</a> can also be
set (<span class="since">Since 3.5.0</span>)
</dd>
<dt>VGA configuration</dt>
<dd>
Control how the video devices exposed to the guest using the
<code>vgaconf</code> attribute which takes the value "io", "on" or "off".
At present, it's only applicable to the bhyve's "gop" video model type
(<span class="since">Since 3.5.0</span>)
</dd>
</dl>
</dd>
</dl>
<h4><a id="elementsConsole">Consoles, serial, parallel &amp; channel devices</a></h4>
<p>
A character device provides a way to interact with the virtual machine.
Paravirtualized consoles, serial ports, parallel ports and channels are
all classed as character devices and so represented using the same syntax.
</p>
<pre>
...
&lt;devices&gt;
&lt;parallel type='pty'&gt;
&lt;source path='/dev/pts/2'/&gt;
&lt;target port='0'/&gt;
&lt;/parallel&gt;
&lt;serial type='pty'&gt;
&lt;source path='/dev/pts/3'/&gt;
&lt;target port='0'/&gt;
&lt;/serial&gt;
&lt;serial type='file'&gt;
&lt;source path='/tmp/file' append='on'&gt;
&lt;seclabel model='dac' relabel='no'/&gt;
&lt;/source&gt;
&lt;target port='0'/&gt;
&lt;/serial&gt;
&lt;console type='pty'&gt;
&lt;source path='/dev/pts/4'/&gt;
&lt;target port='0'/&gt;
&lt;/console&gt;
&lt;channel type='unix'&gt;
&lt;source mode='bind' path='/tmp/guestfwd'/&gt;
&lt;target type='guestfwd' address='10.0.2.1' port='4600'/&gt;
&lt;/channel&gt;
&lt;/devices&gt;
...</pre>
<p>
In each of these directives, the top-level element name (parallel, serial,
console, channel) describes how the device is presented to the guest. The
guest interface is configured by the <code>target</code> element.
</p>
<p>
The interface presented to the host is given in the <code>type</code>
attribute of the top-level element. The host interface is
configured by the <code>source</code> element.
</p>
<p>
The <code>source</code> element may contain an optional
<code>seclabel</code> to override the way that labelling
is done on the socket path. If this element is not present,
the <a href="#seclabel">security label is inherited from
the per-domain setting</a>.
</p>
<p>
If the interface <code>type</code> presented to the host is "file",
then the <code>source</code> element may contain an optional attribute
<code>append</code> that specifies whether or not the information in
the file should be preserved on domain restart. Allowed values are
"on" and "off" (default). <span class="since">Since 1.3.1</span>.
</p>
<p>
Regardless of the <code>type</code>, character devices can
have an optional log file associated with them. This is
expressed via a <code>log</code> sub-element, with a
<code>file</code> attribute. There can also be an <code>append</code>
attribute which takes the same values described above.
<span class="since">Since 1.3.3</span>.
</p>
<pre>
...
&lt;log file="/var/log/libvirt/qemu/guestname-serial0.log" append="off"/&gt;
...</pre>
<p>
Each character device element has an optional
sub-element <code>&lt;address&gt;</code> which can tie the
device to a
particular <a href="#elementsControllers">controller</a> or PCI
slot.
</p>
<p>
For character device with type <code>unix</code> or <code>tcp</code>
the <code>source</code> has an optional element <code>reconnect</code>
which configures reconnect timeout if the connection is lost.
There are two attributes, <code>enabled</code> where possible
values are "yes" and "no" and <code>timeout</code> which is in
seconds. The <code>reconnect</code> attribute is valid only
for <code>connect</code> mode.
<span class="since">Since 3.7.0 (QEMU driver only)</span>.
</p>
<h5><a id="elementsCharGuestInterface">Guest interface</a></h5>
<p>
A character device presents itself to the guest as one of the following
types.
</p>
<h6><a id="elementCharParallel">Parallel port</a></h6>
<pre>
...
&lt;devices&gt;
&lt;parallel type='pty'&gt;
&lt;source path='/dev/pts/2'/&gt;
&lt;target port='0'/&gt;
&lt;/parallel&gt;
&lt;/devices&gt;
...</pre>
<p>
<code>target</code> can have a <code>port</code> attribute, which
specifies the port number. Ports are numbered starting from 0. There are
usually 0, 1 or 2 parallel ports.
</p>
<h6><a id="elementCharSerial">Serial port</a></h6>
<pre>
...
&lt;devices&gt;
&lt;!-- Serial port --&gt;
&lt;serial type='pty'&gt;
&lt;source path='/dev/pts/3'/&gt;
&lt;target port='0'/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<pre>
...
&lt;devices&gt;
&lt;!-- USB serial port --&gt;
&lt;serial type='pty'&gt;
&lt;target type='usb-serial' port='0'&gt;
&lt;model name='usb-serial'/&gt;
&lt;/target&gt;
&lt;address type='usb' bus='0' port='1'/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<p>
The <code>target</code> element can have an optional <code>port</code>
attribute, which specifies the port number (starting from 0), and an
optional <code>type</code> attribute: valid values are,
<span class="since">since 1.0.2</span>, <code>isa-serial</code> (usable
with x86 guests), <code>usb-serial</code> (usable whenever USB support
is available) and <code>pci-serial</code> (usable whenever PCI support
is available); <span class="since">since 3.10.0</span>,
<code>spapr-vio-serial</code> (usable with ppc64/pseries guests),
<code>system-serial</code> (usable with aarch64/virt and,
<span class="since">since 4.7.0</span>, riscv/virt guests) and
<code>sclp-serial</code> (usable with s390 and s390x guests) are
available as well.
</p>
<p>
<span class="since">Since 3.10.0</span>, the <code>target</code>
element can have an optional <code>model</code> subelement;
valid values for its <code>name</code> attribute are:
<code>isa-serial</code> (usable with the <code>isa-serial</code> target
type); <code>usb-serial</code> (usable with the <code>usb-serial</code>
target type); <code>pci-serial</code>
(usable with the <code>pci-serial</code> target type);
<code>spapr-vty</code> (usable with the <code>spapr-vio-serial</code>
target type); <code>pl011</code> and,
<span class="since">since 4.7.0</span>, <code>16550a</code> (usable
with the <code>system-serial</code> target type);
<code>sclpconsole</code> and <code>sclplmconsole</code> (usable with
the <code>sclp-serial</code> target type). Providing a target model is
usually unnecessary: libvirt will automatically pick one that's suitable
for the chosen target type, and overriding that value is generally not
recommended.
</p>
<p>
If any of the attributes is not specified by the user, libvirt will
choose a value suitable for most users.
</p>
<p>
Most target types support configuring the guest-visible device
address as <a href="#elementsAddress">documented above</a>; more
specifically, acceptable address types are <code>isa</code> (for
<code>isa-serial</code>), <code>usb</code> (for <code>usb-serial</code>),
<code>pci</code> (for <code>pci-serial</code>) and <code>spapr-vio</code>
(for <code>spapr-vio-serial</code>). The <code>system-serial</code>
and <code>sclp-serial</code> target types don't support specifying an
address.
</p>
<p>
For the relationship between serial ports and consoles,
<a href="#elementCharSerialAndConsole">see below</a>.
</p>
<h6><a id="elementCharConsole">Console</a></h6>
<pre>
...
&lt;devices&gt;
&lt;!-- Serial console --&gt;
&lt;console type='pty'&gt;
&lt;source path='/dev/pts/2'/&gt;
&lt;target type='serial' port='0'/&gt;
&lt;/console&gt;
&lt;/devices&gt;
...</pre>
<pre>
...
&lt;devices&gt;
&lt;!-- KVM virtio console --&gt;
&lt;console type='pty'&gt;
&lt;source path='/dev/pts/5'/&gt;
&lt;target type='virtio' port='0'/&gt;
&lt;/console&gt;
&lt;/devices&gt;
...</pre>
<p>
The <code>console</code> element is used to represent interactive
serial consoles. Depending on the type of guest in use and the specifics
of the configuration, the <code>console</code> element might represent
the same device as an existing <code>serial</code> element or a separate
device.
</p>
<p>
A <code>target</code> subelement is supported and works the same
way as with the <code>serial</code> element
(<a href="#elementCharSerial">see above</a> for details).
Valid values for the <code>type</code> attribute are:
<code>serial</code> (described below);
<code>virtio</code> (usable whenever VirtIO support is available);
<code>xen</code>, <code>lxc</code> and <code>openvz</code>
(available when the corresponding hypervisor is in use).
<code>sclp</code> and <code>sclplm</code> (usable for s390 and
s390x QEMU guests) are supported for compatibility reasons but should
not be used for new guests: use the <code>sclpconsole</code> and
<code>sclplmconsole</code> target models, respectively, with the
<code>serial</code> element instead.
</p>
<p>
Of the target types listed above, <code>serial</code> is special in
that it doesn't represents a separate device, but rather the same
device as the first <code>serial</code> element. Due to this, there can
only be a single <code>console</code> element with target type
<code>serial</code> per guest.
</p>
<p>
Virtio consoles are usually accessible as <code>/dev/hvc[0-7]</code>
from inside the guest; for more information, see
<a href="http://fedoraproject.org/wiki/Features/VirtioSerial">http://fedoraproject.org/wiki/Features/VirtioSerial</a>.
<span class="since">Since 0.8.3</span>
</p>
<p>
For the relationship between serial ports and consoles,
<a href="#elementCharSerialAndConsole">see below</a>.
</p>
<h6><a id="elementCharSerialAndConsole">Relationship between serial ports and consoles</a></h6>
<p>
Due to hystorical reasons, the <code>serial</code> and
<code>console</code> elements have partially overlapping scopes.
</p>
<p>
In general, both elements are used to configure one or more serial
consoles to be used for interacting with the guest. The main difference
between the two is that <code>serial</code> is used for emulated,
usually native, serial consoles, whereas <code>console</code> is used
for paravirtualized ones.
</p>
<p>
Both emulated and paravirtualized serial consoles have advantages and
disadvantages:
</p>
<ul>
<li>
emulated serial consoles are usually initialized much earlier than
paravirtualized ones, so they can be used to control the bootloader
and display both firmware and early boot messages;
</li>
<li>
on several platforms, there can only be a single emulated serial
console per guest but paravirtualized consoles don't suffer from the
same limitation.
</li>
</ul>
<p>
A configuration such as:
</p>
<pre>
...
&lt;devices&gt;
&lt;console type='pty'&gt;
&lt;target type='serial'/&gt;
&lt;/console&gt;
&lt;console type='pty'&gt;
&lt;target type='virtio'/&gt;
&lt;/console&gt;
&lt;/devices&gt;
...</pre>
<p>
will work on any platform and will result in one emulated serial console
for early boot logging / interactive / recovery use, and one
paravirtualized serial console to be used eg. as a side channel. Most
people will be fine with having just the first <code>console</code>
element in their configuration, but if a specific configuration is
desired then both elements should be specified.
</p>
<p>
Note that, due to the compatibility concerns mentioned earlier, all the
following configurations:
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type='pty'/&gt;
&lt;/devices&gt;
...</pre>
<pre>
...
&lt;devices&gt;
&lt;console type='pty'/&gt;
&lt;/devices&gt;
...</pre>
<pre>
...
&lt;devices&gt;
&lt;serial type='pty'/&gt;
&lt;console type='pty'/&gt;
&lt;/devices&gt;
...</pre>
<p>
will be treated the same and will result in a single emulated serial
console being available to the guest.
</p>
<h6><a id="elementCharChannel">Channel</a></h6>
<p>
This represents a private communication channel between the host and the
guest.
</p>
<pre>
...
&lt;devices&gt;
&lt;channel type='unix'&gt;
&lt;source mode='bind' path='/tmp/guestfwd'/&gt;
&lt;target type='guestfwd' address='10.0.2.1' port='4600'/&gt;
&lt;/channel&gt;
&lt;!-- KVM virtio channel --&gt;
&lt;channel type='pty'&gt;
&lt;target type='virtio' name='arbitrary.virtio.serial.port.name'/&gt;
&lt;/channel&gt;
&lt;channel type='unix'&gt;
&lt;source mode='bind' path='/var/lib/libvirt/qemu/f16x86_64.agent'/&gt;
&lt;target type='virtio' name='org.qemu.guest_agent.0' state='connected'/&gt;
&lt;/channel&gt;
&lt;channel type='spicevmc'&gt;
&lt;target type='virtio' name='com.redhat.spice.0'/&gt;
&lt;/channel&gt;
&lt;/devices&gt;
...</pre>
<p>
This can be implemented in a variety of ways. The specific type of
channel is given in the <code>type</code> attribute of the
<code>target</code> element. Different channel types have different
<code>target</code> attributes.
</p>
<dl>
<dt><code>guestfwd</code></dt>
<dd>TCP traffic sent by the guest to a given IP address and port is
forwarded to the channel device on the host. The <code>target</code>
element must have <code>address</code> and <code>port</code> attributes.
<span class="since">Since 0.7.3</span></dd>
<dt><code>virtio</code></dt>
<dd>Paravirtualized virtio channel. Channel is exposed in the guest under
/dev/vport*, and if the optional element <code>name</code> is specified,
/dev/virtio-ports/$name (for more info, please see
<a href="http://fedoraproject.org/wiki/Features/VirtioSerial">http://fedoraproject.org/wiki/Features/VirtioSerial</a>). The
optional element <code>address</code> can tie the channel to a
particular <code>type='virtio-serial'</code>
controller, <a href="#elementsAddress">documented above</a>.
With qemu, if <code>name</code> is "org.qemu.guest_agent.0",
then libvirt can interact with a guest agent installed in the
guest, for actions such as guest shutdown or file system quiescing.
<span class="since">Since 0.7.7, guest agent interaction
since 0.9.10</span> Moreover, <span class="since">since 1.0.6</span>
it is possible to have source path auto generated for virtio unix channels.
This is very useful in case of a qemu guest agent, where users don't
usually care about the source path since it's libvirt who talks to
the guest agent. In case users want to utilize this feature, they should
leave <code>&lt;source&gt;</code> element out. <span class="since">Since
1.2.11</span> the active XML for a virtio channel may contain an optional
<code>state</code> attribute that reflects whether a process in the
guest is active on the channel. This is an output-only attribute.
Possible values for the <code>state</code> attribute are
<code>connected</code> and <code>disconnected</code>.
</dd>
<dt><code>xen</code></dt>
<dd> Paravirtualized Xen channel. Channel is exposed in the guest as a
Xen console but identified with a name. Setup and consumption of a Xen
channel depends on software and configuration in the guest
(for more info, please see <a href="http://xenbits.xen.org/docs/unstable/misc/channel.txt">http://xenbits.xen.org/docs/unstable/misc/channel.txt</a>).
Channel source path semantics are the same as the virtio target type.
The <code>state</code> attribute is not supported since Xen channels
lack the necessary probing mechanism.
<span class="since">Since 2.3.0</span>
</dd>
<dt><code>spicevmc</code></dt>
<dd>Paravirtualized SPICE channel. The domain must also have a
SPICE server as a <a href="#elementsGraphics">graphics
device</a>, at which point the host piggy-backs messages
across the <code>main</code> channel. The <code>target</code>
element must be present, with
attribute <code>type='virtio'</code>; an optional
attribute <code>name</code> controls how the guest will have
access to the channel, and defaults
to <code>name='com.redhat.spice.0'</code>. The
optional <code>address</code> element can tie the channel to a
particular <code>type='virtio-serial'</code> controller.
<span class="since">Since 0.8.8</span></dd>
</dl>
<h5><a id="elementsCharHostInterface">Host interface</a></h5>
<p>
A character device presents itself to the host as one of the following
types.
</p>
<h6><a id="elementsCharSTDIO">Domain logfile</a></h6>
<p>
This disables all input on the character device, and sends output
into the virtual machine's logfile
</p>
<pre>
...
&lt;devices&gt;
&lt;console type='stdio'&gt;
&lt;target port='1'/&gt;
&lt;/console&gt;
&lt;/devices&gt;
...</pre>
<h6><a id="elementsCharFle">Device logfile</a></h6>
<p>
A file is opened and all data sent to the character
device is written to the file.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="file"&gt;
&lt;source path="/var/log/vm/vm-serial.log"/&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<h6><a id="elementsCharVC">Virtual console</a></h6>
<p>
Connects the character device to the graphical framebuffer in
a virtual console. This is typically accessed via a special
hotkey sequence such as "ctrl+alt+3"
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type='vc'&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<h6><a id="elementsCharNull">Null device</a></h6>
<p>
Connects the character device to the void. No data is ever
provided to the input. All data written is discarded.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type='null'&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<h6><a id="elementsCharPTY">Pseudo TTY</a></h6>
<p>
A Pseudo TTY is allocated using /dev/ptmx. A suitable client
such as 'virsh console' can connect to interact with the
serial port locally.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="pty"&gt;
&lt;source path="/dev/pts/3"/&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<p>
NB special case if &lt;console type='pty'&gt;, then the TTY
path is also duplicated as an attribute tty='/dev/pts/3'
on the top level &lt;console&gt; tag. This provides compat
with existing syntax for &lt;console&gt; tags.
</p>
<h6><a id="elementsCharHost">Host device proxy</a></h6>
<p>
The character device is passed through to the underlying
physical character device. The device types must match,
eg the emulated serial port should only be connected to
a host serial port - don't connect a serial port to a parallel
port.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="dev"&gt;
&lt;source path="/dev/ttyS0"/&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<h6><a id="elementsCharPipe">Named pipe</a></h6>
<p>
The character device writes output to a named pipe. See pipe(7) for
more info.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="pipe"&gt;
&lt;source path="/tmp/mypipe"/&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<h6><a id="elementsCharTCP">TCP client/server</a></h6>
<p>
The character device acts as a TCP client connecting to a
remote server.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="tcp"&gt;
&lt;source mode="connect" host="0.0.0.0" service="2445"/&gt;
&lt;protocol type="raw"/&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<p>
Or as a TCP server waiting for a client connection.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="tcp"&gt;
&lt;source mode="bind" host="127.0.0.1" service="2445"/&gt;
&lt;protocol type="raw"/&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<p>
Alternatively you can use <code>telnet</code> instead
of <code>raw</code> TCP in order to utilize the telnet protocol
for the connection.
</p>
<p>
<span class="since">Since 0.8.5,</span> some hypervisors support
use of either <code>telnets</code> (secure telnet) or <code>tls</code>
(via secure sockets layer) as the transport protocol for connections.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="tcp"&gt;
&lt;source mode="connect" host="0.0.0.0" service="2445"/&gt;
&lt;protocol type="telnet"/&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
...
&lt;serial type="tcp"&gt;
&lt;source mode="bind" host="127.0.0.1" service="2445"/&gt;
&lt;protocol type="telnet"/&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<p>
<span class="since">Since 2.4.0,</span> the optional attribute
<code>tls</code> can be used to control whether a chardev
TCP communication channel would utilize a hypervisor configured
TLS X.509 certificate environment in order to encrypt the data
channel. For the QEMU hypervisor, usage of a TLS environment can
be controlled on the host by the <code>chardev_tls</code> and
<code>chardev_tls_x509_cert_dir</code> or
<code>default_tls_x509_cert_dir</code> settings in the file
/etc/libvirt/qemu.conf. If <code>chardev_tls</code> is enabled,
then unless the <code>tls</code> attribute is set to "no", libvirt
will use the host configured TLS environment.
If <code>chardev_tls</code> is disabled, but the <code>tls</code>
attribute is set to "yes", then libvirt will attempt to use the
host TLS environment if either the <code>chardev_tls_x509_cert_dir</code>
or <code>default_tls_x509_cert_dir</code> TLS directory structure exists.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="tcp"&gt;
&lt;source mode='connect' host="127.0.0.1" service="5555" tls="yes"/&gt;
&lt;protocol type="raw"/&gt;
&lt;target port="0"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<h6><a id="elementsCharUDP">UDP network console</a></h6>
<p>
The character device acts as a UDP netconsole service,
sending and receiving packets. This is a lossy service.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="udp"&gt;
&lt;source mode="bind" host="0.0.0.0" service="2445"/&gt;
&lt;source mode="connect" host="0.0.0.0" service="2445"/&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<h6><a id="elementsCharUNIX">UNIX domain socket client/server</a></h6>
<p>
The character device acts as a UNIX domain socket server,
accepting connections from local clients.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="unix"&gt;
&lt;source mode="bind" path="/tmp/foo"/&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<h6><a id="elementsCharSpiceport">Spice channel</a></h6>
<p>
The character device is accessible through spice connection
under a channel name specified in the <code>channel</code>
attribute. <span class="since">Since 1.2.2</span>
</p>
<p>
Note: depending on the hypervisor, spiceports might (or might not)
be enabled on domains with or without <a href="#elementsGraphics">spice
graphics</a>.
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="spiceport"&gt;
&lt;source channel="org.qemu.console.serial.0"/&gt;
&lt;target port="1"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<h6><a id="elementsNmdm">Nmdm device</a></h6>
<p>
The nmdm device driver, available on FreeBSD, provides two
tty devices connected together by a virtual null modem cable.
<span class="since">Since 1.2.4</span>
</p>
<pre>
...
&lt;devices&gt;
&lt;serial type="nmdm"&gt;
&lt;source master="/dev/nmdm0A" slave="/dev/nmdm0B"/&gt;
&lt;/serial&gt;
&lt;/devices&gt;
...</pre>
<p>
The <code>source</code> element has these attributes:
</p>
<dl>
<dt><code>master</code></dt>
<dd>Master device of the pair, that is passed to the hypervisor.
Device is specified by a fully qualified path.</dd>
<dt><code>slave</code></dt>
<dd>Slave device of the pair, that is passed to the clients for connection
to the guest console. Device is specified by a fully qualified path.</dd>
</dl>
<h4><a id="elementsSound">Sound devices</a></h4>
<p>
A virtual sound card can be attached to the host via the
<code>sound</code> element. <span class="since">Since 0.4.3</span>
</p>
<pre>
...
&lt;devices&gt;
&lt;sound model='es1370'/&gt;
&lt;/devices&gt;
...</pre>
<dl>
<dt><code>sound</code></dt>
<dd>
The <code>sound</code> element has one mandatory attribute,
<code>model</code>, which specifies what real sound device is emulated.
Valid values are specific to the underlying hypervisor, though typical
choices are 'es1370', 'sb16', 'ac97', 'ich6' and 'usb'.
(<span class="since">
'ac97' only since 0.6.0, 'ich6' only since 0.8.8,
'usb' only since 1.2.7</span>)
</dd>
</dl>
<p>
<span class="since">Since 0.9.13</span>, a sound element
with <code>ich6</code> model can have optional
sub-elements <code>&lt;codec&gt;</code> to attach various audio
codecs to the audio device. If not specified, a default codec
will be attached to allow playback and recording.
</p>
<p>
Valid values are:
</p>
<p>
<ul>
<li>'duplex' - advertise a line-in and a line-out </li>
<li>'micro' - advertise a speaker and a microphone </li>
<li>'output' - advertise a line-out
<span class="since">Since 4.4.0</span></li>
</ul>
</p>
<pre>
...
&lt;devices&gt;
&lt;sound model='ich6'&gt;
&lt;codec type='micro'/&gt;
&lt;/sound&gt;
&lt;/devices&gt;
...</pre>
<p>
Each <code>sound</code> element has an optional
sub-element <code>&lt;address&gt;</code> which can tie the
device to a particular PCI
slot, <a href="#elementsAddress">documented above</a>.
</p>
<h4><a id="elementsWatchdog">Watchdog device</a></h4>
<p>
A virtual hardware watchdog device can be added to the guest via
the <code>watchdog</code> element.
<span class="since">Since 0.7.3, QEMU and KVM only</span>
</p>
<p>
The watchdog device requires an additional driver and management
daemon in the guest. Just enabling the watchdog in the libvirt
configuration does not do anything useful on its own.
</p>
<p>
Currently libvirt does not support notification when the
watchdog fires. This feature is planned for a future version of
libvirt.
</p>
<pre>
...
&lt;devices&gt;
&lt;watchdog model='i6300esb'/&gt;
&lt;/devices&gt;
...</pre>
<pre>
...
&lt;devices&gt;
&lt;watchdog model='i6300esb' action='poweroff'/&gt;
&lt;/devices&gt;
&lt;/domain&gt;</pre>
<dl>
<dt><code>model</code></dt>
<dd>
<p>
The required <code>model</code> attribute specifies what real
watchdog device is emulated. Valid values are specific to the
underlying hypervisor.
</p>
<p>
QEMU and KVM support:
</p>
<ul>
<li>'i6300esb' - the recommended device,
emulating a PCI Intel 6300ESB </li>
<li>'ib700' - emulating an ISA iBase IB700 </li>
<li>'diag288' - emulating an S390 DIAG288 device
<span class="since">Since 1.2.17</span></li>
</ul>
</dd>
<dt><code>action</code></dt>
<dd>
<p>
The optional <code>action</code> attribute describes what
action to take when the watchdog expires. Valid values are
specific to the underlying hypervisor.
</p>
<p>
QEMU and KVM support:
</p>
<ul>
<li>'reset' - default, forcefully reset the guest</li>
<li>'shutdown' - gracefully shutdown the guest
(not recommended) </li>
<li>'poweroff' - forcefully power off the guest</li>
<li>'pause' - pause the guest</li>
<li>'none' - do nothing</li>
<li>'dump' - automatically dump the guest
<span class="since">Since 0.8.7</span></li>
<li>'inject-nmi' - inject a non-maskable interrupt
into the guest
<span class="since">Since 1.2.17</span></li>
</ul>
<p>
Note 1: the 'shutdown' action requires that the guest
is responsive to ACPI signals. In the sort of situations
where the watchdog has expired, guests are usually unable
to respond to ACPI signals. Therefore using 'shutdown'
is not recommended.
</p>
<p>
Note 2: the directory to save dump files can be configured
by <code>auto_dump_path</code> in file /etc/libvirt/qemu.conf.
</p>
</dd>
</dl>
<h4><a id="elementsMemBalloon">Memory balloon device</a></h4>
<p>
A virtual memory balloon device is added to all Xen and KVM/QEMU
guests. It will be seen as <code>memballoon</code> element.
It will be automatically added when appropriate, so there is no
need to explicitly add this element in the guest XML unless a
specific PCI slot needs to be assigned.
<span class="since">Since 0.8.3, Xen, QEMU and KVM only</span>
Additionally, <span class="since">since 0.8.4</span>, if the
memballoon device needs to be explicitly disabled,
<code>model='none'</code> may be used.
</p>
<p>
Example: automatically added device with KVM
</p>
<pre>
...
&lt;devices&gt;
&lt;memballoon model='virtio'/&gt;
&lt;/devices&gt;
...</pre>
<p>
Example: manually added device with static PCI slot 2 requested
</p>
<pre>
...
&lt;devices&gt;
&lt;memballoon model='virtio'&gt;
&lt;address type='pci' domain='0x0000' bus='0x00' slot='0x02' function='0x0'/&gt;
&lt;stats period='10'/&gt;
&lt;driver iommu='on' ats='on'/&gt;
&lt;/memballoon&gt;
&lt;/devices&gt;
&lt;/domain&gt;</pre>
<dl>
<dt><code>model</code></dt>
<dd>
<p>
The required <code>model</code> attribute specifies what type
of balloon device is provided. Valid values are specific to
the virtualization platform
</p>
<ul>
<li>'virtio' - default with QEMU/KVM</li>
<li>'virtio-transitional' <span class="since">Since 5.2.0</span></li>
<li>'virtio-non-transitional' <span class="since">Since 5.2.0</span></li>
<li>'xen' - default with Xen</li>
</ul>
See <a href="#elementsVirtioTransitional">Virtio transitional devices</a>
for more details.
</dd>
<dt><code>autodeflate</code></dt>
<dd>
<p>
The optional <code>autodeflate</code> attribute allows to
enable/disable (values "on"/"off", respectively) the ability of the
QEMU virtio memory balloon to release some memory at the last moment
before a guest's process get killed by Out of Memory killer.
<span class="since">Since 1.3.1, QEMU and KVM only</span>
</p>
</dd>
<dt><code>period</code></dt>
<dd>
<p>
The optional <code>period</code> allows the QEMU virtio memory balloon
driver to provide statistics through the <code>virsh dommemstat
[domain]</code> command. By default, collection is not enabled. In
order to enable, use the <code>virsh dommemstat [domain] --period
[number]</code> command or <code>virsh edit</code> command to add the
option to the XML definition. The <code>virsh dommemstat</code> will
accept the options <code>--live</code>, <code>--current</code>,
or <code>--config</code>. If an option is not provided, the change
for a running domain will only be made to the active guest. If the
QEMU driver is not at the right revision, the attempt to set the
period will fail. Large values (e.g. many years) might be ignored.
<span class='since'>Since 1.1.1, requires QEMU 1.5</span>
</p>
</dd>
<dt><code>driver</code></dt>
<dd>
For model <code>virtio</code> memballoon,
<a href="#elementsVirtio">Virtio-specific options</a> can also be
set. (<span class="since">Since 3.5.0</span>)
</dd>
</dl>
<h4><a id="elementsRng">Random number generator device</a></h4>
<p>
The virtual random number generator device allows the host to pass
through entropy to guest operating systems.
<span class="since">Since 1.0.3</span>
</p>
<p>
Example: usage of the RNG device:
</p>
<pre>
...
&lt;devices&gt;
&lt;rng model='virtio'&gt;
&lt;rate period="2000" bytes="1234"/&gt;
&lt;backend model='random'&gt;/dev/random&lt;/backend&gt;
&lt;!-- OR --&gt;
&lt;backend model='egd' type='udp'&gt;
&lt;source mode='bind' service='1234'/&gt;
&lt;source mode='connect' host='1.2.3.4' service='1234'/&gt;
&lt;/backend&gt;
&lt;!-- OR --&gt;
&lt;backend model='builtin'/&gt;
&lt;/rng&gt;
&lt;/devices&gt;
...
</pre>
<dl>
<dt><code>model</code></dt>
<dd>
<p>
The required <code>model</code> attribute specifies what type
of RNG device is provided. Valid values are specific to
the virtualization platform:
</p>
<ul>
<li>'virtio' - supported by qemu and virtio-rng kernel module</li>
<li>'virtio-transitional' <span class='since'>Since 5.2.0</span></li>
<li>'virtio-non-transitional' <span class='since'>Since 5.2.0</span></li>
</ul>
See <a href="#elementsVirtioTransitional">Virtio transitional devices</a>
for more details.
</dd>
<dt><code>rate</code></dt>
<dd>
<p>
The optional <code>rate</code> element allows limiting the rate at
which entropy can be consumed from the source. The mandatory
attribute <code>bytes</code> specifies how many bytes are permitted
to be consumed per period. An optional <code>period</code> attribute
specifies the duration of a period in milliseconds; if omitted, the
period is taken as 1000 milliseconds (1 second).
<span class='since'>Since 1.0.4</span>
</p>
</dd>
<dt><code>backend</code></dt>
<dd>
<p>
The <code>backend</code> element specifies the source of entropy
to be used for the domain. The source model is configured using the
<code>model</code> attribute. Supported source models are:
</p>
<dl>
<dt><code>random</code></dt>
<dd>
<p>
This backend type expects a non-blocking character device
as input. The file name is specified as contents of the
<code>backend</code> element. <span class='since'>Since
1.3.4</span> any path is accepted. Before that
<code>/dev/random</code> and <code>/dev/hwrng</code> were
the only accepted paths. When no file name is specified,
the hypervisor default is used. For QEMU, the default is
<code>/dev/random</code>. However, the recommended source
of entropy is <code>/dev/urandom</code> (as it doesn't
have the limitations of <code>/dev/random</code>).
</p>
</dd>
<dt><code>egd</code></dt>
<dd>
<p>
This backend connects to a source using the EGD protocol.
The source is specified as a character device. Refer to
<a href='#elementsCharHostInterface'>character device host interface</a>
for more information.
</p>
</dd>
<dt><code>builtin</code></dt>
<dd>
<p>
This backend uses qemu builtin random generator, which uses
<code>getrandom()</code> syscall as the source of entropy.
(<span class="since">Since 6.1.0 and QEMU 4.2</span>)
</p>
</dd>
</dl>
</dd>
<dt><code>driver</code></dt>
<dd>
The subelement <code>driver</code> can be used to tune the device:
<dl>
<dt>virtio options</dt>
<dd>
<a href="#elementsVirtio">Virtio-specific options</a> can also be
set. (<span class="since">Since 3.5.0</span>)
</dd>
</dl>
</dd>
</dl>
<h4><a id="elementsTpm">TPM device</a></h4>
<p>
The TPM device enables a QEMU guest to have access to TPM
functionality. The TPM device may either be a TPM 1.2 or
a TPM 2.0.
</p>
<p>
The TPM passthrough device type provides access to the host's TPM
for one QEMU guest. No other software may be using the TPM device,
typically /dev/tpm0, at the time the QEMU guest is started.
<span class="since">'passthrough' since 1.0.5</span>
</p>
<p>
Example: usage of the TPM passthrough device
</p>
<pre>
...
&lt;devices&gt;
&lt;tpm model='tpm-tis'&gt;
&lt;backend type='passthrough'&gt;
&lt;device path='/dev/tpm0'/&gt;
&lt;/backend&gt;
&lt;/tpm&gt;
&lt;/devices&gt;
...
</pre>
<p>
The emulator device type gives access to a TPM emulator providing
TPM functionality for each VM. QEMU talks to it over a Unix socket. With
the emulator device type each guest gets its own private TPM.
<span class="since">'emulator' since 4.5.0</span>
The state of the TPM emulator can be encrypted by providing an
<code>encryption</code> element.
<span class="since">'encryption' since 5.6.0</span>
</p>
<p>
Example: usage of the TPM Emulator
</p>
<pre>
...
&lt;devices&gt;
&lt;tpm model='tpm-tis'&gt;
&lt;backend type='emulator' version='2.0'&gt;
&lt;encryption secret='6dd3e4a5-1d76-44ce-961f-f119f5aad935'/&gt;
&lt;/backend&gt;
&lt;/tpm&gt;
&lt;/devices&gt;
...
</pre>
<dl>
<dt><code>model</code></dt>
<dd>
<p>
The <code>model</code> attribute specifies what device
model QEMU provides to the guest. If no model name is provided,
<code>tpm-tis</code> will automatically be chosen for non-PPC64
architectures.
<span class="since">Since 4.4.0</span>, another available choice
is the <code>tpm-crb</code>, which should only be used when the
backend device is a TPM 2.0. <span class="since">Since 6.1.0</span>,
pSeries guests on PPC64 are supported and the default is
<code>tpm-spapr</code>.
</p>
</dd>
<dt><code>backend</code></dt>
<dd>
<p>
The <code>backend</code> element specifies the type of
TPM device. The following types are supported:
</p>
<dl>
<dt><code>passthrough</code></dt>
<dd>
<p>
Use the host's TPM device.
</p>
<p>
This backend type requires exclusive access to a TPM device on
the host. An example for such a device is /dev/tpm0. The fully
qualified file name is specified by path attribute of the
<code>source</code> element. If no file name is specified then
/dev/tpm0 is automatically used.
</p>
</dd>
</dl>
<dl>
<dt><code>emulator</code></dt>
<dd>
<p>
For this backend type the 'swtpm' TPM Emulator must be installed on the
host. Libvirt will automatically start an independent TPM emulator
for each QEMU guest requesting access to it.
</p>
</dd>
</dl>
</dd>
<dt><code>version</code></dt>
<dd>
<p>
The <code>version</code> attribute indicates the version
of the TPM. By default a TPM 1.2 is created. This attribute
only works with the <code>emulator</code> backend. The following
versions are supported:
</p>
<ul>
<li>'1.2' : creates a TPM 1.2</li>
<li>'2.0' : creates a TPM 2.0</li>
</ul>
</dd>
<dt><code>encryption</code></dt>
<dd>
<p>
The <code>encryption</code> element allows the state of a TPM emulator
to be encrypted. The <code>secret</code> must reference a secret object
that holds the passphrase from which the encryption key will be derived.
</p>
</dd>
</dl>
<h4><a id="elementsNVRAM">NVRAM device</a></h4>
<p>
nvram device is always added to pSeries guest on PPC64, and its address
is allowed to be changed. Element <code>nvram</code> (only valid for
pSeries guest, <span class="since">since 1.0.5</span>) is provided to
enable the address setting.
</p>
<p>
Example: usage of NVRAM configuration
</p>
<pre>
...
&lt;devices&gt;
&lt;nvram&gt;
&lt;address type='spapr-vio' reg='0x00003000'/&gt;
&lt;/nvram&gt;
&lt;/devices&gt;
...
</pre>
<dl>
<dt><code>spapr-vio</code></dt>
<dd>
<p>
VIO device address type, only valid for PPC64.
</p>
</dd>
<dt><code>reg</code></dt>
<dd>
<p>
Device address
</p>
</dd>
</dl>
<h4><a id="elementsPanic">panic device</a></h4>
<p>
panic device enables libvirt to receive panic notification from a QEMU
guest.
<span class="since">Since 1.2.1, QEMU and KVM only</span>
</p>
<p>
This feature is always enabled for:
</p>
<ul>
<li>pSeries guests, since it's implemented by the guest firmware</li>
<li>S390 guests, since it's an integral part of the S390 architecture</li>
</ul>
<p>
For the guest types listed above, libvirt automatically adds a
<code>panic</code> element to the domain XML.
</p>
<p>
Example: usage of panic configuration
</p>
<pre>
...
&lt;devices&gt;
&lt;panic model='hyperv'/&gt;
&lt;panic model='isa'&gt;
&lt;address type='isa' iobase='0x505'/&gt;
&lt;/panic&gt;
&lt;/devices&gt;
...
</pre>
<dl>
<dt><code>model</code></dt>
<dd>
<p>
The optional <code>model</code> attribute specifies what type
of panic device is provided. The panic model used when this attribute
is missing depends on the hypervisor and guest arch.
</p>
<ul>
<li>'isa' - for ISA pvpanic device</li>
<li>'pseries' - default and valid only for pSeries guests.</li>
<li>'hyperv' - for Hyper-V crash CPU feature.
<span class="since">Since 1.3.0, QEMU and KVM only</span></li>
<li>'s390' - default for S390 guests.
<span class="since">Since 1.3.5</span></li>
</ul>
</dd>
<dt><code>address</code></dt>
<dd>
<p>
address of panic. The default ioport is 0x505. Most users
don't need to specify an address, and doing so is forbidden
altogether for s390, pseries and hyperv models.
</p>
</dd>
</dl>
<h4><a id="elementsShmem">Shared memory device</a></h4>
<p>
A shared memory device allows to share a memory region between
different virtual machines and the host.
<span class="since">Since 1.2.10, QEMU and KVM only</span>
</p>
<pre>
...
&lt;devices&gt;
&lt;shmem name='my_shmem0'&gt;
&lt;model type='ivshmem-plain'/&gt;
&lt;size unit='M'&gt;4&lt;/size&gt;
&lt;/shmem&gt;
&lt;shmem name='shmem_server'&gt;
&lt;model type='ivshmem-doorbell'/&gt;
&lt;size unit='M'&gt;2&lt;/size&gt;
&lt;server path='/tmp/socket-shmem'/&gt;
&lt;msi vectors='32' ioeventfd='on'/&gt;
&lt;/shmem&gt;
&lt;/devices&gt;
...
</pre>
<dl>
<dt><code>shmem</code></dt>
<dd>
The <code>shmem</code> element has one mandatory attribute,
<code>name</code> to identify the shared memory. This attribute cannot
be directory specific to <code>.</code> or <code>..</code> as well as
it cannot involve path separator <code>/</code>.
</dd>
<dt><code>model</code></dt>
<dd>
Attribute <code>type</code> of the optional element <code>model</code>
specifies the model of the underlying device providing the
<code>shmem</code> device. The models currently supported are
<code>ivshmem</code> (supports both server and server-less shmem, but is
deprecated by newer QEMU in favour of the -plain and -doorbell variants),
<code>ivshmem-plain</code> (only for server-less shmem) and
<code>ivshmem-doorbell</code> (only for shmem with the server).
</dd>
<dt><code>size</code></dt>
<dd>
The optional <code>size</code> element specifies the size of the shared
memory. This must be power of 2 and greater than or equal to 1 MiB.
</dd>
<dt><code>server</code></dt>
<dd>
The optional <code>server</code> element can be used to configure a server
socket the device is supposed to connect to. The optional
<code>path</code> attribute specifies the absolute path to the unix socket
and defaults to <code>/var/lib/libvirt/shmem/$shmem-$name-sock</code>.
</dd>
<dt><code>msi</code></dt>
<dd>
The optional <code>msi</code> element enables/disables (values "on"/"off",
respectively) MSI interrupts. This option can currently be used only
together with the <code>server</code> element. The <code>vectors</code>
attribute can be used to specify the number of interrupt
vectors. The <code>ioeventd</code> attribute enables/disables (values
"on"/"off", respectively) ioeventfd.
</dd>
</dl>
<h4><a id="elementsMemory">Memory devices</a></h4>
<p>
In addition to the initial memory assigned to the guest, memory devices
allow additional memory to be assigned to the guest in the form of
memory modules.
A memory device can be hot-plugged or hot-unplugged depending on the
guests' memory resource needs.
Some hypervisors may require NUMA configured for the guest.
</p>
<p>
Example: usage of the memory devices
</p>
<pre>
...
&lt;devices&gt;
&lt;memory model='dimm' access='private' discard='yes'&gt;
&lt;target&gt;
&lt;size unit='KiB'&gt;524287&lt;/size&gt;
&lt;node&gt;0&lt;/node&gt;
&lt;/target&gt;
&lt;/memory&gt;
&lt;memory model='dimm'&gt;
&lt;source&gt;
&lt;pagesize unit='KiB'&gt;4096&lt;/pagesize&gt;
&lt;nodemask&gt;1-3&lt;/nodemask&gt;
&lt;/source&gt;
&lt;target&gt;
&lt;size unit='KiB'&gt;524287&lt;/size&gt;
&lt;node&gt;1&lt;/node&gt;
&lt;/target&gt;
&lt;/memory&gt;
&lt;memory model='nvdimm'&gt;
&lt;source&gt;
&lt;path&gt;/tmp/nvdimm&lt;/path&gt;
&lt;/source&gt;
&lt;target&gt;
&lt;size unit='KiB'&gt;524288&lt;/size&gt;
&lt;node&gt;1&lt;/node&gt;
&lt;label&gt;
&lt;size unit='KiB'&gt;128&lt;/size&gt;
&lt;/label&gt;
&lt;readonly/&gt;
&lt;/target&gt;
&lt;/memory&gt;
&lt;memory model='nvdimm' access='shared'&gt;
&lt;source&gt;
&lt;path&gt;/dev/dax0.0&lt;/path&gt;
&lt;alignsize unit='KiB'&gt;2048&lt;/alignsize&gt;
&lt;pmem/&gt;
&lt;/source&gt;
&lt;target&gt;
&lt;size unit='KiB'&gt;524288&lt;/size&gt;
&lt;node&gt;1&lt;/node&gt;
&lt;label&gt;
&lt;size unit='KiB'&gt;128&lt;/size&gt;
&lt;/label&gt;
&lt;/target&gt;
&lt;/memory&gt;
&lt;/devices&gt;
...
</pre>
<dl>
<dt><code>model</code></dt>
<dd>
<p>
Provide <code>dimm</code> to add a virtual DIMM module to the guest.
<span class="since">Since 1.2.14</span>
Provide <code>nvdimm</code> model adds a Non-Volatile DIMM
module. <span class="since">Since 3.2.0</span>
</p>
</dd>
<dt><code>access</code></dt>
<dd>
<p>
An optional attribute <code>access</code>
(<span class="since">since 3.2.0</span>) that provides
capability to fine tune mapping of the memory on per
module basis. Values are the same as
<a href="#elementsMemoryBacking">Memory Backing</a>:
<code>shared</code> and <code>private</code>.
For <code>nvdimm</code> model, if using real NVDIMM DAX device as
backend, <code>shared</code> is required.
</p>
</dd>
<dt><code>discard</code></dt>
<dd>
<p>
An optional attribute <code>discard</code>
(<span class="since">since 4.4.0</span>) that provides
capability to fine tune discard of data on per module
basis. Accepted values are <code>yes</code> and
<code>no</code>. The feature is described here:
<a href="#elementsMemoryBacking">Memory Backing</a>.
This attribute is allowed only for
<code>model='dimm'</code>.
</p>
</dd>
<dt><code>source</code></dt>
<dd>
<p>
For model <code>dimm</code> this element is optional and allows to
fine tune the source of the memory used for the given memory device.
If the element is not provided defaults configured via
<code>numatune</code> are used. If <code>dimm</code> is provided,
then the following optional elements can be provided as well:
</p>
<dl>
<dt><code>pagesize</code></dt>
<dd>
<p>
This element can be used to override the default
host page size used for backing the memory device.
The configured value must correspond to a page size
supported by the host.
</p>
</dd>
<dt><code>nodemask</code></dt>
<dd>
<p>
This element can be used to override the default
set of NUMA nodes where the memory would be
allocated.
</p>
</dd>
</dl>
<p>
For model <code>nvdimm</code> this element is mandatory. The
mandatory child element <code>path</code> represents a path in
the host that backs the nvdimm module in the guest. The following
optional elements may be used:
</p>
<dl>
<dt><code>alignsize</code></dt>
<dd>
<p>
The <code>alignsize</code> element defines the page size
alignment used to mmap the address range for the backend
<code>path</code>. If not supplied the host page size is used.
For example, to mmap a real NVDIMM device a 2M-aligned page may
be required, and host page size is 4KB, then we need to set this
element to 2MB.
<span class="since">Since 5.0.0</span>
</p>
</dd>
<dt><code>pmem</code></dt>
<dd>
<p>
If persistent memory is supported and enabled by the hypervisor
in order to guarantee the persistence of writes to the vNVDIMM
backend, then use the <code>pmem</code> element in order to
utilize the feature.
<span class="since">Since 5.0.0</span>
</p>
</dd>
</dl>
</dd>
<dt><code>target</code></dt>
<dd>
<p>
The mandatory <code>target</code> element configures the placement and
sizing of the added memory from the perspective of the guest.
</p>
<p>
The mandatory <code>size</code> subelement configures the size of the
added memory as a scaled integer.
</p>
<p>
The <code>node</code> subelement configures the guest NUMA node to
attach the memory to. The element shall be used only if the guest has
NUMA nodes configured.
</p>
<p>
The following optional elements may be used:
</p>
<dl>
<dt><code>label</code></dt>
<dd>
<p>
For NVDIMM type devices one can optionally use
<code>label</code> and its subelement <code>size</code>
to configure the size of namespaces label storage
within the NVDIMM module. The <code>size</code> element
has usual meaning described
<a href="#elementsMemoryAllocation">here</a>.
For QEMU domains the following restrictions apply:
</p>
<ol>
<li>the minimum label size is 128KiB,</li>
<li>the remaining size (total-size - label-size) will be aligned
to 4KiB as default.</li>
</ol>
</dd>
<dt><code>readonly</code></dt>
<dd>
<p>
The <code>readonly</code> element is used to mark the vNVDIMM
as read-only. Only the real NVDIMM device backend can guarantee
the guest write persistence, so other backend types should use
the <code>readonly</code> element.
<span class="since">Since 5.0.0</span>
</p>
</dd>
</dl>
</dd>
</dl>
<h4><a id="elementsIommu">IOMMU devices</a></h4>
<p>
The <code>iommu</code> element can be used to add an IOMMU device.
<span class="since">Since 2.1.0</span>
</p>
<p>
Example:
</p>
<pre>
...
&lt;devices&gt;
&lt;iommu model='intel'&gt;
&lt;driver intremap='on'/&gt;
&lt;/iommu&gt;
&lt;/devices&gt;
...
</pre>
<dl>
<dt><code>model</code></dt>
<dd>
<p>
Supported values are <code>intel</code> (for Q35 guests) and,
<span class="since">since 5.5.0</span>, <code>smmuv3</code> (for
ARM virt guests).
</p>
</dd>
<dt><code>driver</code></dt>
<dd>
<p>
The <code>driver</code> subelement can be used to configure
additional options, some of which might only be available for
certain IOMMU models:
</p>
<dl>
<dt><code>intremap</code></dt>
<dd>
<p>
The <code>intremap</code> attribute with possible values
<code>on</code> and <code>off</code> can be used to
turn on interrupt remapping, a part of the VT-d functionality.
Currently this requires split I/O APIC
(<code>&lt;ioapic driver='qemu'/&gt;</code>).
<span class="since">Since 3.4.0</span> (QEMU/KVM only)
</p>
</dd>
<dt><code>caching_mode</code></dt>
<dd>
<p>
The <code>caching_mode</code> attribute with possible values
<code>on</code> and <code>off</code> can be used to
turn on the VT-d caching mode (useful for assigned devices).
<span class="since">Since 3.4.0</span> (QEMU/KVM only)
</p>
</dd>
<dt><code>eim</code></dt>
<dd>
<p>
The <code>eim</code> attribute (with possible values
<code>on</code> and <code>off</code>) can be used to
configure Extended Interrupt Mode. A q35 domain with
split I/O APIC (as described in
<a href="#elementsFeatures">hypervisor features</a>),
and both interrupt remapping and EIM turned on for
the IOMMU, will be able to use more than 255 vCPUs.
<span class="since">Since 3.4.0</span> (QEMU/KVM only)
</p>
</dd>
<dt><code>iotlb</code></dt>
<dd>
<p>
The <code>iotlb</code> attribute with possible values
<code>on</code> and <code>off</code> can be used to
turn on the IOTLB used to cache address translation
requests from devices.
<span class="since">Since 3.5.0</span> (QEMU/KVM only)
</p>
</dd>
</dl>
</dd>
</dl>
<h3><a id="vsock">Vsock</a></h3>
<p>A vsock host/guest interface. The <code>model</code> attribute
defaults to <code>virtio</code>. <span class="since">Since 5.2.0</span>
<code>model</code> can also be 'virtio-transitional' and
'virtio-non-transitional', see
<a href="#elementsVirtioTransitional">Virtio transitional devices</a>
for more details.
The optional attribute <code>address</code> of the <code>cid</code>
element specifies the CID assigned to the guest. If the attribute
<code>auto</code> is set to <code>yes</code>, libvirt
will assign a free CID automatically on domain startup.
<span class="since">Since 4.4.0</span></p>
<pre>
...
&lt;devices&gt;
&lt;vsock model='virtio'&gt;
&lt;cid auto='no' address='3'/&gt;
&lt;/vsock&gt;
&lt;/devices&gt;
...</pre>
<h3><a id="seclabel">Security label</a></h3>
<p>
The <code>seclabel</code> element allows control over the
operation of the security drivers. There are three basic
modes of operation, 'dynamic' where libvirt automatically
generates a unique security label, 'static' where the
application/administrator chooses the labels, or 'none'
where confinement is disabled. With dynamic
label generation, libvirt will always automatically
relabel any resources associated with the virtual machine.
With static label assignment, by default, the administrator
or application must ensure labels are set correctly on any
resources, however, automatic relabeling can be enabled
if desired. <span class="since">'dynamic' since 0.6.1, 'static'
since 0.6.2, and 'none' since 0.9.10.</span>
</p>
<p>
If more than one security driver is used by libvirt, multiple
<code>seclabel</code> tags can be used, one for each driver and
the security driver referenced by each tag can be defined using
the attribute <code>model</code>
</p>
<p>
Valid input XML configurations for the top-level security label
are:
</p>
<pre>
&lt;seclabel type='dynamic' model='selinux'/&gt;
&lt;seclabel type='dynamic' model='selinux'&gt;
&lt;baselabel&gt;system_u:system_r:my_svirt_t:s0&lt;/baselabel&gt;
&lt;/seclabel&gt;
&lt;seclabel type='static' model='selinux' relabel='no'&gt;
&lt;label&gt;system_u:system_r:svirt_t:s0:c392,c662&lt;/label&gt;
&lt;/seclabel&gt;
&lt;seclabel type='static' model='selinux' relabel='yes'&gt;
&lt;label&gt;system_u:system_r:svirt_t:s0:c392,c662&lt;/label&gt;
&lt;/seclabel&gt;
&lt;seclabel type='none'/&gt;
</pre>
<p>
If no 'type' attribute is provided in the input XML, then
the security driver default setting will be used, which
may be either 'none' or 'dynamic'. If a 'baselabel' is set
but no 'type' is set, then the type is presumed to be 'dynamic'
</p>
<p>
When viewing the XML for a running guest with automatic
resource relabeling active, an additional XML element,
<code>imagelabel</code>, will be included. This is an
output-only element, so will be ignored in user supplied
XML documents
</p>
<dl>
<dt><code>type</code></dt>
<dd>Either <code>static</code>, <code>dynamic</code> or <code>none</code>
to determine whether libvirt automatically generates a unique security
label or not.
</dd>
<dt><code>model</code></dt>
<dd>A valid security model name, matching the currently
activated security model. Model <code>dac</code> is not available
when guest is run by unprivileged user.
</dd>
<dt><code>relabel</code></dt>
<dd>Either <code>yes</code> or <code>no</code>. This must always
be <code>yes</code> if dynamic label assignment is used. With
static label assignment it will default to <code>no</code>.
</dd>
<dt><code>label</code></dt>
<dd>If static labelling is used, this must specify the full
security label to assign to the virtual domain. The format
of the content depends on the security driver in use:
<ul>
<li>SELinux: a SELinux context.</li>
<li>AppArmor: an AppArmor profile.</li>
<li>
DAC: owner and group separated by colon. They can be
defined both as user/group names or uid/gid. The driver will first
try to parse these values as names, but a leading plus sign can
used to force the driver to parse them as uid or gid.
</li>
</ul>
</dd>
<dt><code>baselabel</code></dt>
<dd>If dynamic labelling is used, this can optionally be
used to specify the base security label that will be used to generate
the actual label. The format of the content depends on the security
driver in use.
The SELinux driver uses only the <code>type</code> field of the
baselabel in the generated label. Other fields are inherited from
the parent process when using SELinux baselabels.
(The example above demonstrates the use of <code>my_svirt_t</code>
as the value for the <code>type</code> field.)
</dd>
<dt><code>imagelabel</code></dt>
<dd>This is an output only element, which shows the
security label used on resources associated with the virtual domain.
The format of the content depends on the security driver in use
</dd>
</dl>
<p>When relabeling is in effect, it is also possible to fine-tune
the labeling done for specific source file names, by either
disabling the labeling (useful if the file lives on NFS or other
file system that lacks security labeling) or requesting an
alternate label (useful when a management application creates a
special label to allow sharing of some, but not all, resources
between domains), <span class="since">since 0.9.9</span>. When
a <code>seclabel</code> element is attached to a specific path
rather than the top-level domain assignment, only the
attribute <code>relabel</code> or the
sub-element <code>label</code> are supported. Additionally,
<span class="since">since 1.1.2</span>, an output-only
element <code>labelskip</code> will be present for active
domains on disks where labeling was skipped due to the image
being on a file system that lacks security labeling.
</p>
<h3><a id="keywrap">Key Wrap</a></h3>
<p>The content of the optional <code>keywrap</code> element specifies
whether the guest will be allowed to perform the S390 cryptographic key
management operations. A clear key can be protected by encrypting it
under a unique wrapping key that is generated for each guest VM running
on the host. Two variations of wrapping keys are generated: one version
for encrypting protected keys using the DEA/TDEA algorithm, and another
version for keys encrypted using the AES algorithm. If a
<code>keywrap</code> element is not included, the guest will be granted
access to both AES and DEA/TDEA key wrapping by default.</p>
<pre>
&lt;domain&gt;
...
&lt;keywrap&gt;
&lt;cipher name='aes' state='off'/&gt;
&lt;/keywrap&gt;
...
&lt;/domain&gt;
</pre>
<p>
At least one <code>cipher</code> element must be nested within the
<code>keywrap</code> element.
</p>
<dl>
<dt><code>cipher</code></dt>
<dd>The <code>name</code> attribute identifies the algorithm
for encrypting a protected key. The values supported for this attribute
are <code>aes</code> for encryption under the AES wrapping key, or
<code>dea</code> for encryption under the DEA/TDEA wrapping key. The
<code>state</code> attribute indicates whether the cryptographic key
management operations should be turned on for the specified encryption
algorithm. The value can be set to <code>on</code> or <code>off</code>.
</dd>
</dl>
<p>Note: DEA/TDEA is synonymous with DES/TDES.</p>
<h3><a id="launchSecurity">Launch Security</a></h3>
<p>
The contents of the <code>&lt;launchSecurity type='sev'&gt;</code> element
is used to provide the guest owners input used for creating an encrypted
VM using the AMD SEV feature (Secure Encrypted Virtualization).
SEV is an extension to the AMD-V architecture which supports running
encrypted virtual machine (VMs) under the control of KVM. Encrypted
VMs have their pages (code and data) secured such that only the guest
itself has access to the unencrypted version. Each encrypted VM is
associated with a unique encryption key; if its data is accessed to a
different entity using a different key the encrypted guests data will
be incorrectly decrypted, leading to unintelligible data.
For more information see various input parameters and its format see the
<a href="https://support.amd.com/TechDocs/55766_SEV-KM_API_Specification.pdf">SEV API spec</a>
<span class="since">Since 4.4.0</span>
</p>
<pre>
&lt;domain&gt;
...
&lt;launchSecurity type='sev'&gt;
&lt;policy&gt;0x0001&lt;/policy&gt;
&lt;cbitpos&gt;47&lt;/cbitpos&gt;
&lt;reducedPhysBits&gt;1&lt;/reducedPhysBits&gt;
&lt;dhCert&gt;RBBBSDDD=FDDCCCDDDG&lt;/dhCert&gt;
&lt;session&gt;AAACCCDD=FFFCCCDSDS&lt;/session&gt;
&lt;/launchSecurity&gt;
...
&lt;/domain&gt;
</pre>
<dl>
<dt><code>cbitpos</code></dt>
<dd>The required <code>cbitpos</code> element provides the C-bit (aka encryption bit)
location in guest page table entry. The value of <code>cbitpos</code> is
hypervisor dependent and can be obtained through the <code>sev</code> element
from the domain capabilities.
</dd>
<dt><code>reducedPhysBits</code></dt>
<dd>The required <code>reducedPhysBits</code> element provides the physical
address bit reducation. Similar to <code>cbitpos</code> the value of <code>
reduced-phys-bit</code> is hypervisor dependent and can be obtained
through the <code>sev</code> element from the domain capabilities.
</dd>
<dt><code>policy</code></dt>
<dd>The required <code>policy</code> element provides the guest policy
which must be maintained by the SEV firmware. This policy is enforced by
the firmware and restricts what configuration and operational commands
can be performed on this guest by the hypervisor. The guest policy
provided during guest launch is bound to the guest and cannot be changed
throughout the lifetime of the guest. The policy is also transmitted
during snapshot and migration flows and enforced on the destination platform.
The guest policy is a 4 unsigned byte with the fields shown in Table:
<table class="top_table">
<tr>
<th> Bit(s) </th>
<th> Description </th>
</tr>
<tr>
<td> 0 </td>
<td> Debugging of the guest is disallowed when set </td>
</tr>
<tr>
<td> 1 </td>
<td> Sharing keys with other guests is disallowed when set </td>
</tr>
<tr>
<td> 2 </td>
<td> SEV-ES is required when set</td>
</tr>
<tr>
<td> 3 </td>
<td> Sending the guest to another platform is disallowed when set</td>
</tr>
<tr>
<td> 4 </td>
<td> The guest must not be transmitted to another platform that is
not in the domain when set. </td>
</tr>
<tr>
<td> 5 </td>
<td> The guest must not be transmitted to another platform that is
not SEV capable when set. </td>
</tr>
<tr>
<td> 6:15 </td>
<td> reserved </td>
</tr>
<tr>
<td> 16:32 </td>
<td> The guest must not be transmitted to another platform with a
lower firmware version. </td>
</tr>
</table>
</dd>
<dt><code>dhCert</code></dt>
<dd>The optional <code>dhCert</code> element provides the guest owners
base64 encoded Diffie-Hellman (DH) key. The key is used to negotiate a
master secret key between the SEV firmware and guest owner. This master
secret key is then used to establish a trusted channel between SEV
firmware and guest owner.
</dd>
<dt><code>session</code></dt>
<dd>The optional <code>session</code> element provides the guest owners
base64 encoded session blob defined in the SEV API spec.
See SEV spec LAUNCH_START section for the session blob format.
</dd>
</dl>
<h2><a id="examples">Example configs</a></h2>
<p>
Example configurations for each driver are provide on the
driver specific pages listed below
</p>
<ul>
<li><a href="drvxen.html#xmlconfig">Xen examples</a></li>
<li><a href="drvqemu.html#xmlconfig">QEMU/KVM examples</a></li>
</ul>
</body>
</html>