Setting VLAN tag (on supported network types only)
Isolating guests’ network traffic from each other
Modifying virtual link state
MTU configuration
Coalesce settings
IP configuration
vhost-user connection
vhost-user connection with passt backend
Traffic filtering with NWFilter
Input devices
Hub devices
Graphical framebuffers
Video devices
Consoles, serial, parallel & channel devices
Guest interface
Parallel port
Serial port
Console
Relationship between serial ports and consoles
Channel
Host interface
Domain logfile
Device logfile
Virtual console
Null device
Pseudo TTY
Host device proxy
Named pipe
TCP client/server
UDP network console
UNIX domain socket client/server
Spice channel
Nmdm device
Sound devices
Audio backends
None audio backend
ALSA audio backend
Coreaudio audio backend
D-Bus audio backend
Jack audio backend
OSS audio backend
PipeWire audio backend
PulseAudio audio backend
SDL audio backend
Spice audio backend
File audio backend
Watchdog devices
Memory balloon device
Random number generator device
TPM device
NVRAM device
panic device
Shared memory device
Memory devices
IOMMU devices
Vsock
Crypto
Pstore
Security label
Key Wrap
Launch Security
Example configs
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 driver docs
Element and attribute overview
The root element required for all virtual machines is named domain. It has two attributes, the type specifies the hypervisor used for running the domain. The allowed values are driver specific, but include “xen”, “kvm”, “hvf” (since 8.1.0 and QEMU 2.12), “qemu” and “lxc”. The second attribute is id which is a unique integer identifier for the running guest machine. Inactive machines have no id value.
General metadata
MyGuest4dea22b3-1d52-d8f3-2516-782e98ab3fa043dc0cf8-809b-4adb-9bea-a9abb5f3d90eA short description - title - of the domainSome human readable description....
...
name
The content of the name element provides a short name for the virtual machine. This name should consist only of alphanumeric 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. Since 0.0.1
uuid
The content of the uuid element provides a globally unique identifier for the virtual machine. The format must be RFC 4122 compliant, eg 3e3fce45-4f53-4fa7-bb32-11f34168b82b. If omitted when defining/creating a new machine, a random UUID is generated. Since 0.0.1
Since 0.8.7, it is also possible to provide the UUID via a SMBIOS System Information specification.
hwuuid
The optional hwuuid element can be used to supply an alternative UUID for identifying the virtual machine from the domain uuid above. The difference between using the hwuuid element and simply providing an alternative UUID via a SMBIOS System Information specification is that the hwuuid affects all devices that expose the UUID to the guest. Since 11.7.0 QEMU/KVM only
genid
Since 4.4.0, the genid 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 uuid. 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:
VM starts executing a snapshot
VM is recovered from backup
VM is failover in a disaster recovery environment
VM is imported, copied, or cloned
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-initialising its random number generator, etc.
The libvirt XML parser will accept both a provided GUID value or just 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.
title
The optional element title provides space for a short description of the domain. The title should not contain any newlines. Since 0.9.10.
description
The content of the description 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. Since 0.7.2
metadata
The metadata 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). Since 0.9.10
Operating system booting
There are a number of different ways to boot virtual machines each with their own pros and cons.
BIOS bootloader
Booting via the BIOS is available for hypervisors supporting full virtualisation. In this case the BIOS has a boot order priority (floppy, harddisk, cdrom, network) determining where to obtain/find the boot image.
...
hvm/usr/lib/xen/boot/hvmloader
...
...
hvm
...
...
hvm/usr/share/OVMF/OVMF_CODE.fd/var/lib/libvirt/nvram/guest_VARS.fd
...
...
hvm/usr/share/OVMF/OVMF_CODE.fd
...
...
hvm/usr/share/OVMF/OVMF_CODE.fd
</source>
...
...
hvm
...
...
hvm
...
firmware
The firmware attribute allows management applications to automatically fill and elements and possibly enable some features required by selected firmware. Accepted values are bios and efi. The selection process scans for files describing installed firmware images in specified location and uses the most specific one which fulfills domain requirements. The locations in order of preference (from generic to most specific one) are:
/usr/share/qemu/firmware
/etc/qemu/firmware
$XDG_CONFIG_HOME/qemu/firmware
For more information refer to firmware metadata specification as described in docs/interop/firmware.json in QEMU repository. Regular users do not need to bother. Since 5.2.0 (QEMU and KVM only) For VMware guests, this is set to efi when the guest uses UEFI, and it is not set when using BIOS. Since 5.3.0 (VMware ESX and Workstation/Player)
type
The content of the type element specifies the type of operating system to be booted in the virtual machine. hvm indicates that the OS is one designed to run on bare metal, so requires full virtualisation. linux (badly named!) refers to an OS that supports the Xen 3 hypervisor guest ABI. There are also two optional attributes, arch specifying the CPU architecture to virtualisation, and machine referring to the machine type. The Capabilities XML provides details on allowed values for these. If arch is omitted then for most hypervisor drivers, the host native arch will be chosen. For the test, ESX and VMWare hypervisor drivers, however, the i686 arch will always be chosen even on an x86_64 host. Since 0.0.1
firmware
Since 7.2.0 QEMU/KVM only
When using firmware auto-selection there are different features enabled in the firmwares. The list of features can be used to limit what firmware should be automatically selected for the VM. The list of features can be specified using zero or more feature elements. Libvirt will take into consideration only the listed features and ignore the rest when selecting the firmware.
feature
The list of mandatory attributes:
enabled (accepted values are yes and no) is used to tell libvirt if the feature must be enabled or not in the automatically selected firmware
name the name of the feature, the list of the features:
enrolled-keys whether the selected nvram template has default certificate enrolled. Firmware with Secure Boot feature but without enrolled keys will successfully boot non-signed binaries as well. Valid only for firmwares with Secure Boot feature.
secure-boot whether the firmware implements UEFI Secure boot feature.
loader
The optional loader 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 virtualised domains as well as setting the QEMU BIOS file path for QEMU/KVM domains. Xen since 0.1.0, QEMU/KVM since 0.9.12 Then, since 1.2.8 it's possible for the element to have two optional attributes: readonly (accepted values are yes and no) to reflect the fact that the image should be writable or read-only. The second attribute type accepts values rom and pflash. 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, type should be pflash. Moreover, some firmwares may implement the Secure boot feature. Attribute secure can be used to tell the hypervisor that the firmware is capable of Secure Boot feature. It cannot be used to enable or disable the feature itself in the firmware. Since 2.1.0. If the loader is marked as read-only, then with UEFI it is assumed that there will be a writable NVRAM available. In some cases, however, it may be desirable for the loader to run without any NVRAM, discarding any config changes on shutdown. The stateless flag (Since 8.6.0) can be used to control this behaviour, when set to yes NVRAM will never be created.
When firmware autoselection is enabled, the format attribute can be used to tell libvirt to only consider firmware builds that are in a specific format. Supported values are raw and qcow2. Since 9.2.0 (QEMU only)
nvram
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 either selected by the firmware autoselection process or defined in qemu.conf. If needed, the template attribute can be used to override the automatically chosen NVRAM template and templateFormat to specify the format for the template file (currently supported are raw and qcow2). When firmware auto-selection is in use the templateFormat field reflects the format of the picked template. Since 10.10.0 (QEMU only)
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). Since 1.2.8
Since 8.5.0, it's possible for the element to have type attribute (accepts values file, block and network) in that case the NVRAM storage is described by a sub-element with the same syntax as disk's source. See Hard drives, floppy disks, CDROMs. For block backed NVRAM images it may be necessary to ensure that the block device has the correct guest visible size based on hypervisor expectations. This may require use of non raw format image that allows arbitrary disk size.
Note: network backed NVRAM the variables are not instantiated from the template and it's user's responsibility to provide a valid NVRAM image.
This element supports a format attribute, which specifies the format of the NVRAM image. Since 9.2.0 (QEMU only) Note that hypervisors may not support automatic population of the nvram if format differs from templateFormat or may support only a specific format.
It is not valid to provide this element if the loader is marked as stateless.
boot
The dev attribute takes one of the values "fd", "hd", "cdrom" or "network" and is used to specify the next boot device to consider. The boot 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 Hard drives, floppy disks, CDROMs, Network interfaces, and Host device assignment sections below) were introduced and they are the preferred way providing full control over booting order. The boot element and per-device boot elements are mutually exclusive. Since 0.1.3, per-device boot since 0.8.8
smbios
How to populate SMBIOS information visible in the guest. The mode 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 virConnectGetSysinfo call can be used to see what values are copied), or "sysinfo" (use the values in the SMBIOS System Information element). If not specified, the hypervisor default is used. Since 0.8.7
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, rebootTimeout doesn't make sense for UEFI, useserial 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.
bootmenu
Whether or not to enable an interactive boot menu prompt on guest startup. The enable attribute can be either "yes" or "no". If not specified, the hypervisor default is used. Since 0.8.3 Additional attribute timeout 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 enable is set to "yes". Since 1.2.8
bios
This element has attribute useserial with possible values yes or no. It enables or disables Serial Graphics Adapter which allows users to see BIOS messages on a serial port. Therefore, one needs to have Serial port defined. Since 0.9.4. The rebootTimeout attribute (since 0.10.2 (QEMU only)) 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 65535 and special value -1 disables the reboot.
Host bootloader
Hypervisors employing paravirtualisation 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 pygrub with Xen. The Bhyve hypervisor also uses a host bootloader, either bhyveload or grub-bhyve.
...
/usr/bin/pygrub--append single
...
bootloader
The content of the bootloader 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. Since 0.1.0
bootloader_args
The optional bootloader_args element allows command line arguments to be passed to the bootloader. Since 0.2.3
Direct kernel boot
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 virtualised guests.
...
hvm/usr/lib/xen/boot/hvmloader/root/f8-i386-vmlinuz/root/f8-i386-initrdconsole=ttyS0 ks=http://example.com/f8-i386/os//path/to/shim.efi/root/ppc.dtb
...
type
This element has the same semantics as described earlier in the BIOS bootloader section.
loader
This element has the same semantics as described earlier in the BIOS bootloader section.
kernel
The contents of this element specify the fully-qualified path to the kernel image in the host OS.
initrd
The contents of this element specify the fully-qualified path to the (optional) ramdisk image in the host OS.
cmdline
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
shim
Use specified fully-qualified path to load an initial UEFI bootloader that handles chaining to a trusted full bootloader under secure boot environments.
dtb
The contents of this element specify the fully-qualified path to the (optional) device tree binary (dtb) image in the host OS. Since 1.0.4
Container boot
When booting a domain using container based virtualisation, instead of a kernel / boot image, a path to the init binary is required, using the init element. By default this will be launched with no arguments. To specify the initial argv, use the initarg element, repeated as many time as is required. The cmdline element, if set will be used to provide an equivalent to /proc/cmdline but will not affect init argv.
To set environment variables, use the initenv element, one for each variable.
To set a custom work directory for the init, use the initdir element.
To run the init command as a given user or group, use the inituser or initgroup elements respectively. Both elements can be provided either a user (resp. group) id or a name. Prefixing the user or group id with a + will force it to be considered like a numeric value. Without this, it will be first tried as a user or group name.
exe/bin/systemd--unitemergency.servicesome value/my/custom/cwdtester1000
If you want to enable user namespace, set the idmap element. The uid and gid elements have three attributes:
start
First user ID in container. It must be '0'.
target
The first user ID in container will be mapped to this target user ID in host.
count
How many users in container are allowed to map to host's user.
Common element configuration
These options apply to any form of booting of the guest OS.
...
...
/path/to/slic.dat
...
acpi
The table element contains a fully-qualified path to the ACPI table, with the type attribute dictating what data must be present in the file:
raw: a single ACPI table with header and data, with ACPI signature auto-detected from header (Since 11.2.0 (QEMU)).
rawset: concatenation of multiple ACPI tables with header and data, each with any ACPI signature, auto-detected from header (Since 11.2.0 (Xen)).
slic: a single ACPI table with header and data, providing software licensing information. The ACPI table signature in the header will be forced to SLIC (Since 1.3.5 (QEMU), mis-interpreted as rawset Since 5.9.0 (Xen)).
msdm: a single ACPI table with header and data, providing Microsoft Data Management information. The ACPI table signature in the header will be forced to MSDM (Since 11.2.0 (QEMU)).
Each type may be used only once, except for raw which can appear multiple times.
SMBIOS System Information
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 dmidecode command in the guest). The optional sysinfo element covers all such categories of information. Since 0.8.7
...
...
LENOVOFedoraVirt-Manager0.9.4LENOVO20BE0061MC0B98401 ProW1KS427111EDell Inc.2.1265X0XF240000101Type3Sku1myappname:some arbitrary dataotherappname:more arbitrary dataexample value
...
The sysinfo element has a mandatory attribute type that determine the layout of sub-elements, with supported values of:
smbios
Sub-elements call out specific SMBIOS values, which will affect the guest if used in conjunction with the smbios sub-element of the os element (see Operating system booting). Each sub-element of sysinfo names a SMBIOS block, and within those elements can be a list of entry elements that describe a field within the block. The following blocks and entries are recognised:
bios
This is block 0 of SMBIOS, with entry names drawn from:
vendor
BIOS Vendor's Name
version
BIOS Version
date
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.
release
System BIOS Major and Minor release number values concatenated together as one string separated by a period, for example, 10.22.
system
This is block 1 of SMBIOS, with entry names drawn from:
manufacturer
Manufacturer of BIOS
product
Product Name
version
Version of the product
serial
Serial number
uuid
Universal Unique ID number. If this entry is provided alongside a top-level uuid element (see General metadata), then the two values must match.
sku
SKU number to identify a particular configuration.
family
Identify the family a particular computer belongs to.
baseBoard
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:
manufacturer
Manufacturer of BIOS
product
Product Name
version
Version of the product
serial
Serial number
asset
Asset tag
location
Location in chassis
NB: Incorrectly supplied entries for the bios, system or baseBoard blocks will be ignored without error. Other than uuid validation and date format checking, all values are passed as strings to the hypervisor driver.
chassis
Since 4.1.0, this is block 3 of SMBIOS, with entry names drawn from:
manufacturer
Manufacturer of Chassis
version
Version of the Chassis
serial
Serial number
asset
Asset tag
sku
SKU number
oemStrings
This is block 11 of SMBIOS. This element should appear once and can have multiple entry 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. ( Since 4.1.0 )
fwcfg
Some hypervisors provide unified way to tweak how firmware configures itself, or may contain tables to be installed for the guest OS, for instance boot order, ACPI, SMBIOS, etc.
It even allows users to define their own config blobs. In case of QEMU, these then appear under domain's sysfs (if the guest kernel has FW_CFG_SYSFS config option enabled), under /sys/firmware/qemu_fw_cfg. Note, that these values apply regardless the mode under . Since 6.5.0
Please note that because of limited number of data slots use of fwcfg is strongly discouraged and should be used instead.
example value
The sysinfo element can have multiple entry child elements. Each element then has mandatory name attribute, which defines the name of the blob and must begin with opt/ and to avoid clashing with other names is advised to be in form opt/$RFQDN/$name where $RFQDN is a reverse fully qualified domain name you control. Then, the element can either contain the value (to set the blob value directly), or file attribute (to set the blob value from the file).
CPU Allocation
...
2
...
vcpu
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.
cpuset
The optional attribute cpuset 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 cputune. If the attribute emulatorpin of cputune is specified, the cpuset specified by vcpu here will be ignored. Similarly, for virtual CPUs which have the vcpupin specified, the cpuset specified by cpuset here will be ignored. For virtual CPUs which don't have vcpupin specified, each will be pinned to the physical CPUs specified by cpuset 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. Since 0.4.4
current
The optional attribute current can be used to specify whether fewer than the maximum number of virtual CPUs should be enabled. Since 0.8.5
placement
The optional attribute placement can be used to indicate the CPU placement mode for domain process. The value can be either "static" or "auto", but defaults to placement of numatune or "static" if cpuset is specified. Using "auto" indicates the domain process will be pinned to the advisory nodeset from querying numad and the value of attribute cpuset will be ignored if it's specified. If both cpuset and placement are not specified or if placement is "static", but no cpuset is specified, the domain process will be pinned to all the available physical CPUs. Since 0.9.11 (QEMU and KVM only)
vcpus
The vcpus element allows to control state of individual vCPUs. The id 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 vcpu element minus 1. The enabled attribute allows to control the state of the vCPU. Valid values are yes and no. hotpluggable 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 yes and no. order 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 initialisation 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. Since 2.2.0 (QEMU only)
IOThreads Allocation
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. Since 1.2.8 (QEMU only)
...
4
...
...
...
iothreads
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. Since 1.2.8
iothreadids
The optional iothreadids 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 iothreads defined for the domain. The id attribute is used to define the IOThread ID. The id attribute must be a positive integer greater than 0. If there are less iothreadids defined than iothreads defined for the domain, then libvirt will sequentially fill iothreadids starting at 1 avoiding any predefined id. If there are more iothreadids defined than iothreads defined for the domain, then the iothreads value will be adjusted accordingly. Since 1.2.15 The element has two optional attributes thread_pool_min and thread_pool_max which allow setting lower and upper boundary for number of worker threads for given IOThread. While the former can be value of zero, the latter can't. Since 8.5.0 Since 9.4.0 an optional sub-element poll with can be used to override the hypervisor-default interval of polling for the iothread before it switches back to events. The optional attribute max sets the maximum time polling should be used in nanoseconds. Setting max to 0 disables polling. Attributes grow and shrink override (or disable when set to 0 the default steps for increasing/decreasing the polling interval if the set interval is deemed insufficient or extensive.
defaultiothread
This element represents the default event loop within hypervisor, where I/O requests from devices not assigned to a specific IOThread are processed. The element then can have thread_pool_min and/or thread_pool_max attributes, which control the lower and upper boundary for number of worker threads of the default event loop. Emulator might be multithreaded and spawn so called worker threads on demand. In general neither of these attributes should be set (leaving the emulator use its own default values), unless the emulator runs in a real time workload and thus can't afford unpredictability of time it takes to spawn new worker threads. Since 8.5.0
CPU Tuning
...
20481000000-11000000-11000000-11000000-1
...
cputune
The optional cputune element provides details regarding the CPU tunable parameters for the domain. Note: for the qemu driver, the optional vcpupin and emulatorpin pinning settings are honored after the emulator is launched and NUMA constraints considered. This means that it is expected that other physical CPUs of the host will be used during this time by the domain, which will be reflected by the output of virsh cpu-stats. Since 0.9.0
vcpupin
The optional vcpupin element specifies which of host's physical CPUs the domain vCPU will be pinned to. If this is omitted, and attribute cpuset of element vcpu is not specified, the vCPU is pinned to all the physical CPUs by default. It contains two required attributes, the attribute vcpu specifies vCPU id, and the attribute cpuset is same as attribute cpuset of element vcpu. QEMU driver support since 0.9.0, Xen driver support since 0.9.1
emulatorpin
The optional emulatorpin 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 cpuset of element vcpu is not specified, "emulator" is pinned to all the physical CPUs by default. It contains one required attribute cpuset specifying which physical CPUs to pin to.
iothreadpin
The optional iothreadpin element specifies which of host physical CPUs the IOThreads will be pinned to. If this is omitted and attribute cpuset of element vcpu is not specified, the IOThreads are pinned to all the physical CPUs by default. There are two required attributes, the attribute iothread specifies the IOThread ID and the attribute cpuset specifying which physical CPUs to pin to. See the IOThreads Allocation section documenting valid values of iothread. Since 1.2.9
shares
The optional shares 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. The value should be in range [2, 262144] using cgroups v1, [1, 10000] using cgroups v2. Since 0.9.0
period
The optional period element specifies the enforcement interval (unit: microseconds). Within period, each vCPU of the domain will not be allowed to consume more than quota worth of runtime. The value should be in range [1000, 1000000]. A period with value 0 means no value. Only QEMU driver support since 0.9.4, LXC since 0.9.10
quota
The optional quota element specifies the maximum allowed bandwidth (unit: microseconds). A domain with quota 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, 17592186044415] 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. Only QEMU driver support since 0.9.4, LXC since 0.9.10
global_period
The optional global_period element specifies the enforcement CFS scheduler interval (unit: microseconds) for the whole domain in contrast with period which enforces the interval per vCPU. The value should be in range 1000, 1000000]. A global_period with value 0 means no value. Only QEMU driver support since 1.3.3
global_quota
The optional global_quota element specifies the maximum allowed bandwidth (unit: microseconds) within a period for the whole domain. A domain with global_quota 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, 17592186044415] or less than 0. A global_quota with value 0 means no value. Only QEMU driver support since 1.3.3
emulator_period
The optional emulator_period element specifies the enforcement interval (unit: microseconds). Within emulator_period, emulator threads (those excluding vCPUs) of the domain will not be allowed to consume more than emulator_quota worth of runtime. The value should be in range [1000, 1000000]. A period with value 0 means no value. Only QEMU driver support since 0.10.0
emulator_quota
The optional emulator_quota element specifies the maximum allowed bandwidth (unit: microseconds) for domain's emulator threads (those excluding vCPUs). A domain with emulator_quota 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, 17592186044415] or less than 0. A quota with value 0 means no value. Only QEMU driver support since 0.10.0
iothread_period
The optional iothread_period element specifies the enforcement interval (unit: microseconds) for IOThreads. Within iothread_period, each IOThread of the domain will not be allowed to consume more than iothread_quota worth of runtime. The value should be in range [1000, 1000000]. An iothread_period with value 0 means no value. Only QEMU driver support since 2.1.0
iothread_quota
The optional iothread_quota element specifies the maximum allowed bandwidth (unit: microseconds) for IOThreads. A domain with iothread_quota 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, 17592186044415] or less than 0. An iothread_quota with value 0 means no value. You can use this feature to ensure that all IOThreads run at the same speed. Only QEMU driver support since 2.1.0
vcpusched, iothreadsched and emulatorsched
The optional vcpusched, iothreadsched and emulatorsched elements specify the scheduler type (values batch, idle, fifo, rr) for particular vCPU, IOThread and emulator threads respectively. For vcpusched and iothreadsched the attributes vcpus and iothreads select which vCPUs/IOThreads this setting applies to, leaving them out sets the default. The element emulatorsched does not have that attribute. Valid vcpus values start at 0 through one less than the number of vCPU's defined for the domain. Valid iothreads values are described in the IOThreads Allocation section. If no iothreadids are defined, then libvirt numbers IOThreads from 1 to the number of iothreads available for the domain. For real-time schedulers (fifo, rr), 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). Since 1.2.13 emulatorsched since 5.3.0
cachetune Since 4.1.0
Optional cachetune 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 vcpus specifies to which vCPUs this allocation applies. A vCPU can only be member of one cachetune element allocation. The vCPUs specified by cachetune can be identical with those in memorytune, however they are not allowed to overlap. The optional, output only id attribute identifies cache uniquely. Supported subelements are:
cache
This optional element controls the allocation of CPU cache and has the following attributes:
level
Host cache level from which to allocate.
id
Host cache id from which to allocate.
type
Type of allocation. Can be code for code (instructions), data for data or both 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 both for host with CDP (code/data prioritisation) enabled.
size
The size of the region to allocate. The value by default is in bytes, but the unit attribute can be used to scale the value.
unit (optional)
If specified it is the unit such as KiB, MiB, GiB, or TiB (described in the memory element for Memory Allocation) in which size is specified, defaults to bytes.
monitor Since 4.10.0
The optional element monitor creates the cache monitor(s) for current cache allocation and has the following required attributes:
level
Host cache level the monitor belongs to.
vcpus
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.
memorytune Since 4.7.0
Optional memorytune 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 vcpus specifies to which vCPUs this allocation applies. A vCPU can only be member of one memorytune element allocation. The vcpus specified by memorytune can be identical to those specified by cachetune. However they are not allowed to overlap each other. Supported subelements are:
node
This element controls the allocation of CPU memory bandwidth and has the following attributes:
id
Host node id from which to allocate memory bandwidth.
bandwidth
The memory bandwidth to allocate from this node. The value is usually in percent (Intel) but can also be in MB/s (if resctrl is mounted with the mba_MBps option) or in 1/8 GB/s increments (AMD). The user is responsible for making sure the value makes sense on their system and configuration.
Memory Allocation
...
1524288524288524288
...
memory
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 unit, which defaults to "KiB" (kibibytes, 210 or blocks of 1024 bytes). Valid units are "b" or "bytes" for bytes, "KB" for kilobytes (103 or 1,000 bytes), "k" or "KiB" for kibibytes (1024 bytes), "MB" for megabytes (106 or 1,000,000 bytes), "M" or "MiB" for mebibytes (220 or 1,048,576 bytes), "GB" for gigabytes (109 or 1,000,000,000 bytes), "G" or "GiB" for gibibytes (230 or 1,073,741,824 bytes), "TB" for terabytes (1012 or 1,000,000,000,000 bytes), or "T" or "TiB" for tebibytes (240 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 NUMA is configured for the guest (See CPU model and topology) the memory element can be omitted. In the case of crash, optional attribute dumpCore can be used to control whether the guest memory should be included in the generated coredump or not (values "on", "off"). unit since 0.9.11, dumpCore since 0.10.2 (QEMU only)
maxMemory
The run time maximum memory allocation of the guest. The initial memory specified by either the element or the NUMA cell size configuration can be increased by hot-plugging of memory to the limit specified by this element. The unit attribute behaves the same as for . The slots 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. Since 1.2.14 supported by the QEMU driver.
currentMemory
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 memory element. The unit attribute behaves the same as for memory.
Memory Backing
...
...
The optional memoryBacking element may contain several elements that influence how virtual memory pages are backed by host pages.
hugepages
This tells the hypervisor that the guest should have its memory allocated using hugepages instead of the normal native page size. Since 1.2.5 it's possible to set hugepages more specifically per numa node. The page element is introduced. It has one compulsory attribute size which specifies which hugepages should be used (especially useful on systems supporting hugepages of different sizes). The default unit for the size attribute is kiB (multiplier of 1024). If you want to use different unit, use optional unit attribute. For systems with NUMA, the optional nodeset 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 NUMA Node Tuning.
nosharepages
Instructs hypervisor to disable shared pages (memory merge, KSM) for this domain. Since 1.0.6
locked
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 hard_limit (see Memory Tuning) on memory allocation suitable for the specific environment at the same time to mitigate the risks described above. Since 1.0.6
source
Using the type attribute, it's possible to provide "file" to utilise file memorybacking or keep the default "anonymous". Since 4.10.0, you may choose "memfd" backing. (QEMU/KVM only)
access
Using the mode attribute, specify if the memory is to be "shared" or "private". This can be overridden per numa node by memAccess.
allocation
Using the optional mode attribute, specify when to allocate the memory by supplying either "immediate" or "ondemand". Since 8.2.0 it is possible to set the number of threads that hypervisor uses to allocate memory via threads attribute. To speed allocation process up, when pinning emulator thread it's recommended to include CPUs from desired NUMA nodes so that allocation threads can have their affinity set.
discard
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 optimisation and is not guaranteed to work in all cases (e.g. when hypervisor crashes). Since 4.4.0 (QEMU/KVM only)
Memory Tuning
...
1128267108864
...
memtune
The optional memtune 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 . For backwards compatibility, output is always in KiB. unit since 0.9.11 Possible values for all *_limit parameters are in range from 0 to VIR_DOMAIN_MEMORY_PARAM_UNLIMITED.
hard_limit
The optional hard_limit 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 undecidable problem; that said, if you already set locked in Memory Backing because your workload demands it, you'll have to take into account the specifics of your deployment and figure out a value for hard_limit 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.
soft_limit
The optional soft_limit element is the memory limit to enforce during memory contention. The units for this value are kibibytes (i.e. blocks of 1024 bytes)
swap_hard_limit
The optional swap_hard_limit 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
min_guarantee
The optional min_guarantee 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.
NUMA Node Tuning
...
...
numatune
The optional numatune 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. Since 0.9.3
memory
The optional memory element specifies how to allocate memory for the domain process on a NUMA host. It contains several optional attributes. Attribute mode is either 'interleave', 'strict', 'preferred', or 'restrictive', defaults to 'strict'. The value 'restrictive' specifies using system default policy and only cgroups is used to restrict the memory nodes, and it requires setting mode to 'restrictive' in memnode elements (see quirk below). This exists solely for the purpose of being able to request movement of such memory for a running domain using virsh numatune or virDomainSetNumaParameters and is not guaranteed to happen. Attribute nodeset specifies the NUMA nodes, using the same syntax as attribute cpuset of element vcpu. Attribute placement ( since 0.9.12 ) can be used to indicate the memory placement mode for domain process, its value can be either "static" or "auto", defaults to placement of vcpu, or "static" if nodeset is specified. "auto" indicates the domain process will only allocate memory from the advisory nodeset returned from querying numad, and the value of attribute nodeset will be ignored if it's specified. If placement of vcpu is 'auto', and numatune is not specified, a default numatune with placement 'auto' and mode 'strict' will be added implicitly. Since 0.9.3 See virDomainSetNumaParameters for more information on update of this element.
memnode
Optional memnode elements can specify memory allocation policies per each guest NUMA node. For those nodes having no corresponding memnode element, the default from element memory will be used. Attribute cellid addresses guest NUMA node for which the settings are applied. Attributes mode and nodeset have the same meaning and syntax as in memory element. This setting is not compatible with automatic placement. Note that for memnode this will only guide the memory access for the vCPU threads or similar mechanism and is very hypervisor-specific. This does not guarantee the placement of the node's memory allocation. For proper restriction other means should be used (e.g. different mode, preallocated hugepages).
QEMU Since 1.2.7
Block I/O Tuning
...
800/dev/sda1000/dev/sdb50010000100002000020000
...
blkiotune
The optional blkiotune element provides the ability to tune Blkio cgroup tunable parameters for the domain. If this is omitted, it defaults to the OS provided defaults. Since 0.8.8
weight
The optional weight 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].
device
The domain may have multiple device elements that further tune the weights for each host block device in use by the domain. Note that multiple disks (See Hard drives, floppy disks, CDROMs) 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 element of a disk definition (See Hard drives, floppy disks, CDROMs) which can applies to an individual disk). Each device element has two mandatory sub-elements, path describing the absolute path of the device, and weight 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]. Since 0.9.8 Additionally, the following optional sub-elements can be used:
read_bytes_sec
Read throughput limit in bytes per second. Since 1.2.2
write_bytes_sec
Write throughput limit in bytes per second. Since 1.2.2
read_iops_sec
Read I/O operations per second limit. Since 1.2.2
write_iops_sec
Write I/O operations per second limit. Since 1.2.2
Resource partitioning
Hypervisors may allow for virtual machines to be placed into resource partitions, potentially with nesting of said partitions. The resource element groups together configuration related to resource partitioning. It currently supports a child element partition 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.
...
/virtualmachines/production
...
Resource partitions are currently supported by the QEMU and LXC drivers, which map partition paths to cgroups directories, in all mounted controllers. Since 1.0.5
Fibre Channel VMID
The FC SAN can provide various QoS levels and access control depending on the VMID. It can also collect telemetry data at per-VM level which can be used to enhance the IO performance of the VM. This can be configured by using the appid attribute of fibrechannel element. The attribute contains single string (max 128 bytes) and it is used by kernel to create VMID.
...
...
Using this feature requires Fibre Channel capable HW, kernel compiled with option CONFIG_BLK_CGROUP_FC_APPID and nvme_fc kernel module loaded. Since 7.7.0
CPU model and topology
Requirements for CPU model, its features and topology can be specified using the following collection of elements. Since 0.7.5
...
core2duoIntel
...
...
...
...
In case no restrictions need to be put on CPU model and its features, a simpler cpu element can be used. Since 0.7.6
...
...
cpu
The cpu element is the main container for describing guest CPU requirements. Its match attribute specifies how strictly the virtual CPU provided to the guest matches these requirements. Since 0.7.6 the match attribute can be omitted if topology is the only element within cpu. Possible values for the match attribute are:
minimum
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 host-model mode; the domain will not be created if the provided virtual CPU does not meet the requirements.
exact
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.
strict
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.
Since 0.8.5 the match attribute can be omitted and will default to exact. Sometimes the hypervisor is not able to create a virtual CPU exactly matching the specification passed by libvirt. Since 3.2.0, an optional check 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 check 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:
none
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 behaviour of QEMU is to emit warnings, but start the domain anyway.
partial
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.
full
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.
Since 0.9.10, an optional mode 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 mode attribute are:
custom
In this mode, the cpu element describes the CPU that should be presented to the guest. This is the default when no mode 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.
host-model
The host-model mode is essentially a shortcut to copying host-model CPU definition from domain 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 match attribute can't be used in this mode. Specifying CPU model is not supported either, but model's fallback attribute may still be used. Using the feature 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. (Since 1.1.1)
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 host-model may not work as expected. Since 3.2.0 and QEMU 2.9.0 this mode works the way it was designed and it is indicated by the fallback attribute set to forbid in the host-model CPU definition advertised in domain capabilities XML. When fallback attribute is set to allow in the domain capabilities XML, it is recommended to use custom mode with just the CPU model from the host capabilities XML.
Since 1.2.11 PowerISA allows processors to run VMs in binary compatibility mode supporting an older version of ISA. Libvirt on PowerPC architecture uses the host-model 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:
power7
...
host-passthrough
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 feature 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. Depending on hypervisor version the virtual CPU may or may not contain features which may block migration even to an identical host. Since 6.5.0 optional migratable attribute may be used to explicitly request such features to be removed from (on) or kept in (off) the virtual CPU. This attribute does not make migration to another host safer: even with migratable='on' migration will be dangerous unless both hosts are identical as described above.
maximum
When running a guest with hardware virtualisation this CPU model is functionally identical to host-passthrough, so refer to the docs above.
When running a guest with CPU emulation, this CPU model will enable the maximum set of features that the emulation engine is able to support. Note that even with migratable='on' migration will be dangerous unless both hosts are running identical versions of the emulation code.
Since 7.1.0 with the QEMU driver.
Both host-model and host-passthrough modes make sense when a domain can run directly on the host CPUs (for example, domains with type kvm or hvf). The actual host CPU is irrelevant for domains with emulated virtual CPUs (such as domains with type qemu). However, for backward compatibility host-model 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.
If an application does not care about a specific CPU, just wants the best feature set without a need for migration compatibility, the maximum model is a good choice on hypervisors where it is available.
model
The content of the model element specifies CPU model requested by the guest. The list of available CPU models and their definition can be found in directory cpu_map, 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. Since 0.9.10, an optional fallback attribute can be used to forbid this behaviour, in which case an attempt to start a domain requesting an unsupported CPU model will fail. Supported values for fallback attribute are: allow (this is the default), and forbid. The optional vendor_id attribute ( Since 0.10.0 ) 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".
vendor
Since 0.8.3 the content of the vendor 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 cpu_map/*_vendors.xml.
topology
The topology element specifies requested topology of virtual CPU provided to the guest. Its attributes sockets, dies (Since 6.1.0), clusters (Since 10.1.0), cores, and threads accept non-zero positive integer values. They refer to the total number of CPU sockets, number of dies per socket, number of clusters per die, number of cores per cluster, and number of threads per core, respectively. The dies and clusters attributes are 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 cpus element equals to the number of vcpus resulting from the topology. Moreover, not all architectures and machine types support specifying a value other than 1 for all attributes.
feature
The cpu element can contain zero or more feature elements 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 feature element depends on its policy attribute, which has to be set to one of the following values:
force
The virtual CPU will claim the feature is supported regardless of it being supported by host CPU.
require
Guest creation will fail unless the feature is supported by the host CPU or the hypervisor is able to emulate it.
optional
The feature will be supported by virtual CPU if and only if it is supported by host CPU.
disable
The feature will not be supported by virtual CPU.
forbid
Guest creation will fail if the feature is supported by host CPU.
Since 0.8.5 the policy attribute can be omitted and will default to require.
Individual CPU feature names are specified as part of the name attribute. For example, to explicitly specify the 'pcid' feature with Intel IvyBridge CPU model:
...
IvyBridgeIntel
...
deprecated_features
Since 11.0.0, S390 guests may utilise the deprecated_features attribute to specify toggling of CPU model features that are flagged as deprecated by the hypervisor. When this attribute is set to off, the active guest XML will reflect the respective features with the disable policy. When this attribute is set to on, the respective features will be enabled.
cache
Since 3.3.0 the cache element describes the virtual CPU cache. If the element is missing, the hypervisor will use a sensible default.
level
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 cache elements with the level attribute set and those without the attribute is forbidden.
mode
The following values are supported:
emulate
The hypervisor will provide a fake CPU cache data.
passthrough
The real CPU cache data reported by the host CPU will be passed through to the virtual CPU.
disable
The virtual CPU will report no CPU cache of the specified level (or no cache at all if the level attribute is missing).
maxphysaddr
Since 8.7.0 the maxphysaddr element describes the virtual CPU address size in bits. The hypervisor default is used if the element is missing.
mode
This mandatory attribute specifies how the address size is presented. The follow modes are supported:
passthrough
The number of physical address bits reported by the host CPU will be passed through to the virtual CPUs
emulate
The hypervisor will define a specific value for the number of bits of physical addresses via the bits attribute, (optional since 9.2.0) The number of bits cannot exceed the number of physical address bits supported by the hypervisor.
bits
The bits attribute is mandatory if the mode attribute is set to emulate and specifies the virtual CPU address size in bits.
limit
The limit attribute can be used to restrict the maximum value of address bits for passthrough mode, i.e. in case the host CPU reports more bits than that, limit is used. Since 9.3.0
Guest NUMA topology can be specified using the numa element. Since 0.9.8
...
...
...
...
Each cell element specifies a NUMA cell or a NUMA node. cpus specifies the CPU or range of CPUs that are part of the node. Since 6.5.0 For the qemu driver, if the emulator binary supports disjointed cpus ranges in each cell, the sum of all CPUs declared in each cell will be matched with the maximum number of virtual CPUs declared in the vcpu element. This is done by filling any remaining CPUs into the first NUMA cell. Users are encouraged to supply a complete NUMA topology, where the sum of the NUMA CPUs matches the maximum virtual CPUs number declared in vcpus, to make the domain consistent across qemu and libvirt versions. memory specifies the node memory in kibibytes (i.e. blocks of 1024 bytes). Since 6.6.0 the cpus attribute is optional and if omitted a CPU-less NUMA node is created. Since 1.2.11 one can use an additional unit attribute (See Memory Allocation) to define units in which memory is specified. Since 1.2.7 all cells should have id attribute in case referring to some cell is necessary in the code, otherwise the cells are assigned ids in the increasing order starting from 0. Mixing cells with and without the id attribute is not recommended as it may result in unwanted behaviour. Since 1.2.9 the optional attribute memAccess 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 cell element can have an optional discard attribute which fine tunes the discard feature for given numa node as described under Memory Backing. Accepted values are yes and no. Since 4.4.0
This guest NUMA specification is currently available only for QEMU/KVM and Xen.
A NUMA hardware architecture supports the notion of distances between NUMA cells. Since 3.10.0 it is possible to define the distance between NUMA cells using the distances element within a NUMA cell description. The sibling 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.
...
...
...
...
Describing distances between NUMA cells is currently only supported by Xen and QEMU. If no distances 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.
ACPI Heterogeneous Memory Attribute Table
...
...
...
...
Since 6.6.0 the cell element can have a cache child element which describes memory side cache for memory proximity domains. The cache element has a level attribute describing the cache level and thus the element can be repeated multiple times to describe different levels of the cache.
The cache element then has following mandatory attributes:
level
Level of the cache this description refers to.
associativity
Describes cache associativity (accepted values are none, direct and full).
policy
Describes cache write associativity (accepted values are none, writeback and writethrough).
The cache element has two mandatory child elements then: size and line which describe cache size and cache line size. Both elements accept two attributes: value and unit which set the value of corresponding cache attribute.
The NUMA description has an optional interconnects element that describes the normalised memory read/write latency, read/write bandwidth between Initiator Proximity Domains (Processor or I/O) and Target Proximity Domains (Memory).
The interconnects element can have zero or more latency child elements to describe latency between two memory nodes and zero or more bandwidth child elements to describe bandwidth between two memory nodes. Both these have the following mandatory attributes:
initiator
Refers to the source NUMA node
target
Refers to the target NUMA node
type
The type of the access. Accepted values: access, read, write
value
The actual value. For latency this is delay in nanoseconds, for bandwidth this value is in kibibytes per second. Use additional unit attribute to change the units.
To describe latency from one NUMA node to a cache of another NUMA node the latency element has optional cache attribute which in combination with target attribute creates full reference to distant NUMA node's cache level. For instance, target='0' cache='1' refers to the first level cache of NUMA node 0.
Events configuration
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 virDomainReboot , virDomainShutdown , or virDomainShutdownFlags . Using virsh reboot or virsh shutdown would also trigger the event.
...
destroyrestartrestartpoweroff
...
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.
on_poweroff
The content of this element specifies the action to take when the guest requests a poweroff.
on_reboot
The content of this element specifies the action to take when the guest requests a reboot.
on_crash
The content of this element specifies the action to take when the guest crashes.
Each of these states allow for the same four possible actions.
destroy
The domain will be terminated completely and all resources released.
restart
The domain will be terminated and then restarted with the same configuration.
preserve
The domain will be terminated and its resource preserved to allow analysis.
rename-restart
The domain will be terminated and then restarted with a new name. (Only supported by the libxl hypervisor driver.)
QEMU/KVM/HVF supports the on_poweroff and on_reboot events handling the destroy and restart actions, but the combination of on_poweroff set to restart and on_reboot set to destroy is forbidden.
The on_crash event supports these additional actions since 0.8.4.
coredump-destroy
The crashed domain's core will be dumped, and then the domain will be terminated completely and all resources released
coredump-restart
The crashed domain's core will be dumped, and then the domain will be restarted with the same configuration
Since 3.9.0, the lifecycle events can be configured via the virDomainSetLifecycleAction API.
The on_lockfailure element ( since 1.0.0 ) may be used to configure what action should be taken when a lock manager loses resource locks. The following actions are recognised 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.
poweroff
The domain will be forcefully powered off.
restart
The domain will be powered off and started up again to reacquire its locks.
pause
The domain will be paused so that it can be manually resumed when lock issues are solved.
ignore
Keep the domain running as if nothing happened.
Power Management
Since 0.10.2 it is possible to forcibly enable or disable BIOS advertisements to the guest OS. (NB: Only qemu driver support)
...
...
pm
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. 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).
Disk Throttle Group Management
Since 11.2.0 it is possible to create multiple named throttle groups and then reference them within throttlefilters``(sub-element of ``disk element) to form filter chain in QEMU for specific disk. The limits(throttlegroups) are shared within domain, hence the same group can be referenced by different filters.
...
limit010000000400000100000
...
throttlegroup
It has the same sub-elements as iotune (See Hard drives, floppy disks, CDROMs), The difference is that is required.
Hypervisor features
Hypervisors may allow certain CPU / machine features to be toggled on/off.
...
1648128
...
All features are listed within the features element, omitting a togglable feature tag turns it off. The available features can be found by asking for the capabilities XML and domain capabilities XML, but a common set for fully virtualized domains are:
pae
Physical address extension mode allows 32-bit guests to address more than 4 GB of memory.
acpi
ACPI is useful for power management, for example, with KVM or HVF guests it is required for graceful shutdown to work.
apic
APIC allows the use of programmable IRQ management. Since 0.10.2 (QEMU only) there is an optional attribute eoi with values on and off which toggles the availability of EOI (End of Interrupt) for the guest.
hap
Depending on the state attribute (values on, off) enable or disable use of Hardware Assisted Paging. The default is on if the hypervisor detects availability of Hardware Assisted Paging.
viridian
Enable Viridian hypervisor extensions for paravirtualizing guest operating systems
privnet
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.
hyperv
Enable various features improving behavior of guests running Microsoft Windows. Since 11.3.0 some of these flags are also available for Xen domains running Microsoft Windows.
Feature
Description
Value
Since
relaxed
Relax constraints on timers
on, off
1.0.0 (QEMU 2.0), 11.3.0 (Xen, always on)
vapic
Enable virtual APIC
on, off
1.1.0 (QEMU 2.0), 11.3.0 (Xen)
spinlocks
Enable spinlock support - retries attribute defines after how many failed acquisition attempts to notify the hypervisor
on, off; retries - between 4095 and 4294967295, the special value 4294967295 means to never notify the hypervisor (default if omitted)
1.1.0 (QEMU 2.0), never-notify mode 11.9.0 (QEMU 2.0)
vpindex
Virtual processor index
on, off
1.3.3 (QEMU 2.5), 11.3.0 (Xen, always on)
runtime
Processor time spent on running guest code and on behalf of guest code
on, off
1.3.3 (QEMU 2.5)
synic
Enable Synthetic Interrupt Controller (SynIC)
on, off
1.3.3 (QEMU 2.6), 11.3.0 (Xen)
stimer
Enable SynIC timers, optionally with Direct Mode support
on, off; direct - on,off
1.3.3 (QEMU 2.6), direct mode 5.7.0 (QEMU 4.1), 11.3.0 (Xen, on/off only)
reset
Enable hypervisor reset
on, off
1.3.3 (QEMU 2.5)
vendor_id
Set hypervisor vendor id
on, off; value - string, up to 12 characters
1.3.3 (QEMU 2.5)
frequencies
Expose frequency MSRs
on, off
4.7.0 (QEMU 2.12), 11.3.0 (Xen)
reenlightenment
Enable re-enlightenment notification on migration
on, off
4.7.0 (QEMU 3.0)
tlbflush
Enable PV TLB flush support
on, off; direct - on,off; extended - on,off
4.7.0 (QEMU 3.0), direct and extended modes 11.0.0 (QEMU 7.1.0), 11.3.0 (Xen, on/off only)
ipi
Enable PV IPI support
on, off
4.10.0 (QEMU 3.1), 11.3.0 (Xen)
evmcs
Enable Enlightened VMCS
on, off
4.10.0 (QEMU 3.1)
avic
Enable use Hyper-V SynIC with hardware APICv/AVIC
on, off
8.10.0 (QEMU 6.2)
emsr_bitmap
Avoid unnecessary updates to L2 MSR Bitmap upon vmexits.
on, off
10.7.0 (QEMU 7.1)
xmm_input
Enable XMM Fast Hypercall Input
on, off
10.7.0 (QEMU 7.1)
Since 8.0.0 (QEMU) Since 11.3.0 (Xen), the hypervisor can be configured further by setting the mode attribute to one of the following values:
custom
Set exactly the specified features.
passthrough
Enable all features currently supported by the hypervisor, even those that libvirt does not understand. Migration of a guest using 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. Depending on hypervisor version the virtual CPU may or may not contain features which may block migration even to an identical host.
host-model
Similar to the passthrough mode, except libvirt detects which enlightenments are supported by hypervisor and expands them on domain startup into the live XML. In a sense, this is similar to host-model CPU mode (See CPU model and topology). Since 11.9.0
The mode attribute can be omitted and will default to custom.
pvspinlock
Notify the guest that the host supports paravirtual spinlocks for example by exposing the pvticketlocks mechanism. This feature can be explicitly disabled by using state='off' attribute.
kvm
Various features to change the behavior of the KVM hypervisor.
Feature
Description
Value
Since
hidden
Hide the KVM hypervisor from standard MSR based discovery
on, off
1.2.8 (QEMU 2.1.0)
hint-dedicated
Allows a guest to enable optimizations when running on dedicated vCPUs
on, off
5.7.0 (QEMU 2.12.0)
poll-control
Decrease IO completion latency by introducing a grace period of busy waiting
on, off
6.10.0 (QEMU 4.2)
pv-ipi
Paravirtualized send IPIs
on, off
7.10.0 (QEMU 3.1)
dirty-ring
Enable dirty ring feature
on, off; size - must be power of 2, range [1024,65536]
8.0.0 (QEMU 6.1)
xen
Various features to change the behavior of the Xen hypervisor.
Feature
Description
Value
Since
e820_host
Expose the host e820 to the guest (PV only)
on, off
6.3.0
passthrough
Enable IOMMU mappings allowing PCI passthrough
on, off; mode - optional string sync_pt or share_pt
6.3.0
pmu
Depending on the state attribute (values on, off, default on) enable or disable the performance monitoring unit for the guest. Since 1.2.12
vmport
Depending on the state attribute (values on, off, default on) enable or disable the emulation of VMware IO port, for vmmouse etc. Since 1.2.16
gic
Enable for architectures using a General Interrupt Controller instead of APIC in order to handle interrupts. For example, the 'aarch64' architecture uses gic instead of apic. The optional attribute version specifies the GIC version; however, it may not be supported by all hypervisors. Accepted values are 2, 3 and host. Since 1.2.16
smm
Depending on the state attribute (values on, off, default on) enable or disable System Management Mode. Since 2.1.0
Optional sub-element tseg 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 unit 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.
If the VM is booting you should leave this option alone, unless you are very certain you know what you are doing.
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 pc-q35-2.9. Starting with pc-q35-2.10 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.
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.
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).
See Memory Allocation for more details about the unit attribute. Since 4.5.0 (QEMU only)
ioapic
Tune the I/O APIC. Possible values for the driver attribute are: kvm (default for KVM domains) and qemu which puts I/O APIC in userspace which is also known as a split I/O APIC mode. Since 3.4.0 (QEMU/KVM only)
hpt
Configure the HPT (Hash Page Table) of a pSeries guest. Possible values for the resizing attribute are enabled, which causes HPT resizing to be enabled if both the guest and the host support it; disabled, which causes HPT resizing to be disabled regardless of guest and host support; and required, 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. Since 3.10.0 (QEMU/KVM only).
The optional maxpagesize 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. Since 4.5.0 (QEMU/KVM only).
vmcoreinfo
Enable QEMU vmcoreinfo device to let the guest kernel save debug details. Since 4.4.0 (QEMU only)
htm
Configure HTM (Hardware Transactional Memory) availability for pSeries guests. Possible values for the state attribute are on and off. If the attribute is not defined, the hypervisor default will be used. Since 4.6.0 (QEMU/KVM only)
nested-hv
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 desirable 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 state attribute are on and off. If the attribute is not defined, the hypervisor default will be used. Since 4.10.0 (QEMU/KVM only)
msrs
Some guests might require ignoring unknown Model Specific Registers (MSRs) reads and writes. It's possible to switch this by setting unknown attribute of msrs to ignore. If the attribute is not defined, or set to fault, unknown reads and writes will not be ignored. Since 5.1.0 (bhyve only)
ccf-assist
Configure ccf-assist (Count Cache Flush Assist) availability for pSeries guests. Possible values for the state attribute are on and off. If the attribute is not defined, the hypervisor default will be used. Since 5.9.0 (QEMU/KVM only)
cfpc
Configure cfpc (Cache Flush on Privilege Change) availability for pSeries guests. Possible values for the value attribute are broken (no protection), workaround (software workaround available) and fixed (fixed in hardware). If the attribute is not defined, the hypervisor default will be used. Since 6.3.0 (QEMU/KVM only)
sbbc
Configure sbbc (Speculation Barrier Bounds Checking) availability for pSeries guests. Possible values for the value attribute are broken (no protection), workaround (software workaround available) and fixed (fixed in hardware). If the attribute is not defined, the hypervisor default will be used. Since 6.3.0 (QEMU/KVM only)
ibs
Configure ibs (Indirect Branch Speculation) availability for pSeries guests. Possible values for the value attribute are broken (no protection), workaround (count cache flush), fixed-ibs (fixed by serializing indirect branches), fixed-ccd (fixed by disabling the cache count) and fixed-na (fixed in hardware - no longer applicable). If the attribute is not defined, the hypervisor default will be used. Since 6.3.0 (QEMU/KVM only)
tcg
Various features to change the behavior of the TCG accelerator.
Feature
Description
Value
Since
tb-cache
The size of translation block cache size
an integer (a multiple of MiB)
8.0.0
async-teardown
Depending on the enabled attribute (values yes, no) enable or disable QEMU asynchronous teardown to improve memory reclaiming on a guest. Since 9.6.0 (QEMU only)
ras
Report host memory errors to a guest using ACPI and guest external abort exceptions when enabled (on). If the attribute is not defined, the hypervisor default will be used. Since 10.4.0 (QEMU/KVM and ARM virt guests only)
ps2
Depending on the state attribute (values on, off) enable or disable the emulation of a PS/2 controller used by ps2 bus input devices. If the attribute is not defined, the hypervisor default will be used. Since 10.7.0 (QEMU only)
aia
Configure aia (Advanced Interrupt Architecture) for RISC-V 'virt' guests. Possible values for the value attribute are aplic (one emulated APLIC device present per socket), aplic-imsic (one APLIC and one IMSIC device present per core), or none (no support for AIA). If the attribute is not defined, the hypervisor default will be used. Since 11.1.0 (QEMU/KVM and RISC-V guests only)
Time keeping
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'.
...
...
clock
The offset 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.
utc
The guest clock will always be synchronized to UTC when booted. Since 0.9.11 'utc' mode can be converted to 'variable' mode, which can be controlled by using the adjustment 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 adjustment is hypervisor specific.
localtime
The guest clock will be synchronized to the host's configured timezone when booted, if any. Since 0.9.11, the adjustment attribute behaves the same as in 'utc' mode.
timezone
The guest clock will be synchronized to the requested timezone using the timezone attribute. Since 0.7.7
variable
The guest clock will have an arbitrary offset applied relative to UTC or localtime, depending on the basis attribute. The delta relative to UTC (or localtime) is specified in seconds, using the adjustment 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. Since 0.7.7 Since 0.9.11 the basis attribute can be either 'utc' (default) or 'localtime'.
absolute
The guest clock will be always set to the value of the start attribute at startup of the domain. The start attribute takes an epoch timestamp. Since 8.4.0.
A clock may have zero or more timer sub-elements. Since 0.8.0
timer
Each timer element requires a name attribute, and has other optional attributes that depend on the name specified. Various hypervisors support different combinations of attributes.
name
The name attribute selects which timer is being modified, and can be one of "platform" (currently unsupported), "hpet" (xen, qemu, lxc), "kvmclock" (qemu), "pit" (qemu), "rtc" (qemu, lxc), "tsc" (xen, qemu - since 3.2.0 ), "hypervclock" (qemu - since 1.2.2 ) or "armvtimer" (qemu - since 6.1.0 ). The hypervclock timer adds support for the reference time counter and the reference page for iTSC feature for guests running the Microsoft Windows operating system.
track
The track attribute specifies what the timer tracks, and can be "boot", "guest", or "wall", or "realtime". Only valid for name="rtc" or name="platform".
tickpolicy
The tickpolicy 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.
delay
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.
catchup
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.
merge
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
discard
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.
If the policy is "catchup", there can be further details in the catchup sub-element.
catchup
The catchup element has three optional attributes, each a positive integer. The attributes are threshold, slew, and limit.
Note that hypervisors are not required to support all policies across all time sources
frequency
The frequency attribute is an unsigned integer specifying the frequency at which name="tsc" runs.
mode
The mode attribute controls how the name="tsc" timer is managed, and can be "auto", "native", "emulate", "paravirt", or "smpsafe". Other timers are always emulated.
present
The present attribute can be "yes" or "no" to specify whether a particular timer is available to the guest.
Performance monitoring events
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 perf element or enable them via virDomainSetPerfEvents API. The performance values are then retrieved using the virConnectGetAllDomainStats API. Since 2.0.0
...
...
event name
Description
stats parameter name
cmt
usage of l3 cache in bytes by applications running on the platform
perf.cmt
mbmt
total system bandwidth from one level of cache
perf.mbmt
mbml
bandwidth of memory traffic for a memory controller
perf.mbml
cpu_cycles
the count of CPU cycles (total/elapsed)
perf.cpu_cycles
instructions
the count of instructions by applications running on the platform
perf.instructions
cache_references
the count of cache hits by applications running on the platform
perf.cache_references
cache_misses
the count of cache misses by applications running on the platform
perf.cache_misses
branch_instructions
the count of branch instructions by applications running on the platform
perf.branch_instructions
branch_misses
the count of branch misses by applications running on the platform
perf.branch_misses
bus_cycles
the count of bus cycles by applications running on the platform
perf.bus_cycles
stalled_cycles_frontend
the count of stalled CPU cycles in the frontend of the instruction processor pipeline by applications running on the platform
perf.stalled_cycles_frontend
stalled_cycles_backend
the count of stalled CPU cycles in the backend of the instruction processor pipeline by applications running on the platform
perf.stalled_cycles_backend
ref_cpu_cycles
the count of total CPU cycles not affected by CPU frequency scaling by applications running on the platform
perf.ref_cpu_cycles
cpu_clock
the count of CPU clock time, as measured by a monotonic high-resolution per-CPU timer, by applications running on the platform
perf.cpu_clock
task_clock
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
perf.task_clock
page_faults
the count of page faults by applications running on the platform. This includes minor, major, invalid and other types of page faults
perf.page_faults
context_switches
the count of context switches by applications running on the platform
perf.context_switches
cpu_migrations
the count of CPU migrations, that is, where the process moved from one logical processor to another, by applications running on the platform
perf.cpu_migrations
page_faults_min
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
perf.page_faults_min
page_faults_maj
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
perf.page_faults_maj
alignment_faults
the count of alignment faults, that is when the load or store is not aligned properly, by applications running on the platform
perf.alignment_faults
emulation_faults
the count of emulation faults, that is when the kernel traps on unimplemented instructions and emulates them for user space, by applications running on the platform
perf.emulation_faults
Devices
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 devices element. Since 0.1.3
...
/usr/lib/xen/bin/qemu-dm
...
emulator
The contents of the emulator element specify the fully qualified path to the device model emulator binary. The capabilities XML specifies the recommended default emulator to use for each particular domain type / architecture combination.
To help users identifying devices they care about, every device can have direct child alias element which then has name 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: [a-zA-Z0-9_-]. Since 3.9.0
...
Hard drives, floppy disks, CDROMs
Any device that looks like a disk, be it a floppy, harddisk, cdrom, or paravirtualized driver is specified via the disk element.
...
</source>
10000000400000100000
...
...
...
</source>
</source>
<source protocol="http" name="url_path" query="foo=bar&baz=flurb>
somevalue
</source>
</source>
</source>
</source>
</source>
</source>
</source>
</source>
</source>
</source>
</source>
</source>
</source>
</source>
</backingStore>
</disk>
</devices>
...
disk
The disk element is the main container for describing disks and supports the following attributes:
type
Valid values are "file", "block", "dir" ( since 0.7.5 ), "network" ( since 0.8.7 ), or "volume" ( since 1.0.5 ), or "nvme" ( since 6.0.0 ), or "vhostuser" ( since 7.1.0 ), or "vhostvdpa" ( since 9.8.0 (QEMU 8.1.0)), or "ctl" ( since 12.0.0 ) and refer to the underlying source for the disk. Since 0.0.3
device
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".
Using "lun" ( since 0.9.10 ) is only valid when the type is "block" or "network" for protocol='iscsi' or when the type is "volume" when using an iSCSI source pool for mode "host" or as an NPIV 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'). Since 0.1.4
model
Indicates the emulated device model of the disk. Typically this is indicated solely by the bus property.
For bus "virtio" the model can be specified further with "virtio", "virtio-transitional" or "virtio-non-transitional". See virtio device models for more details. Since 5.2.0
For bus "usb" the model can be specified further with usb-storage or usb-bot. There is no difference between the two models for . However with usb-bot a device configured as is properly exposed as a cdrom device inside the guest OS. Unfortunately this configuration is not ABI compatible with usb-storage and thus it can't be interchanged during migration.
The QEMU hypervisor driver will pick usb-bot for cold starts or hotplug for cdrom devices to properly configure the devices. This is not compatible for migration to older versions of libvirt and explicit configuration needs to be used. Since 11.5.0; relevant only for QEMU hypervisor.
rawio
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, rawio 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). Since 0.9.10
sgio
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 device is 'lun'. The attribute exists Since 1.0.2, although currently it's no longer supported by any hypervisor.
snapshot
Indicates the default behavior of the disk during disk snapshots: internal requires a file format such as qcow2 that can store both the snapshot and the data changes since the snapshot; external will separate the snapshot from the live data; no means the disk will not participate in snapshots; and manual allows snapshotting done via an unmanaged storage provider. Read-only disks default to no, while the default for other disks depends on the hypervisor's capabilities. Some hypervisors allow a per-snapshot choice as well, during domain snapshot creation. Not all snapshot modes are supported; for example, enabling snapshots with a transient disk generally does not make sense. Since 0.9.5
source
Representation of the disk source depends on the disk type attribute value as follows:
file
The file attribute specifies the fully-qualified path to the file holding the disk. Since 0.0.3
Since 9.0.0 a new optional attribute fdgroup can be added instructing to access the disk via file descriptors associated to the domain object via the virDomainFDAssociate() API rather than opening the files. The files do not necessarily have to be accessible by libvirt via the filesystem. The filename passed via file can still be used to generate paths to write into image metadata when doing block operations but libvirt will not access these natively.
block
The dev attribute specifies the fully-qualified path to the host device to serve as the disk. Since 0.0.3
dir
The dir attribute specifies the fully-qualified path to the directory to use as the disk. Since 0.7.5
Note that most hypervisors that support dir disks do that by exposing an emulated block device with an emulated filesystem populated with contents of the configured directory. As guest operating system may cache the filesystem metadata, outside changes to the directory may not appear in the guest and/or may result in corrupted data being observable from the VM.
The format of the emulated filesystem is controlled by the format attribute of the driver element. Currently only the fat format is supported. Hypervisors may only support mode.
network
The protocol attribute specifies the protocol to access to the requested image. Possible values are "nbd", "iscsi", "rbd", "sheepdog", "gluster", "vxhs", "nfs", "http", "https", "ftp", ftps", "tftp", or "ssh".
For any protocol other than nbd an additional attribute name is mandatory to specify which volume/image will be used.
For "nbd", the name attribute is optional. TLS transport for NBD can be enabled by setting the tls attribute to yes. For the QEMU hypervisor, usage of a TLS environment can also be globally controlled on the host by the nbd_tls and nbd_tls_x509_cert_dir in /etc/libvirt/qemu.conf. ('tls' Since 4.5.0 ) Since 8.2.0 the optional attribute tlsHostname can be used to override the expected host name of the NBD server used for TLS certificate verification.
For "rbd", the name attribute could be two formats: the format of pool_name/image_name includes the rbd pool name and image name with default rbd pool namespace; for the customized namespace, the format is pool_name/namespace/image_name ( Since 11.6.0 and QEMU 5.0 ). The pool name, namespace and image are separated by slash.
For protocols http and https an optional attribute query specifies the query string. ( Since 6.2.0 )
For "iscsi" ( since 1.0.4 ), the name attribute may include a logical unit number, separated from the target's name by a slash (e.g., iqn.2013-07.com.example:iscsi-pool/1). If not specified, the default LUN is zero.
For "vxhs" ( since 3.8.0 ), the name is the UUID of the volume, assigned by the HyperScale server. Additionally, an optional attribute tls (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 vxhs_tls and vxhs_tls_x509_cert_dir or default_tls_x509_cert_dir settings in the file /etc/libvirt/qemu.conf. If vxhs_tls is enabled, then unless the domain tls attribute is set to "no", libvirt will use the host configured TLS environment. If the tls attribute is set to "yes", then regardless of the qemu.conf setting, TLS authentication will be attempted.
Since 0.8.7
volume
The underlying disk source is represented by attributes pool and volume. Attribute pool specifies the name of the storage pool (managed by libvirt) where the disk source resides. Attribute volume specifies the name of storage volume (managed by libvirt) used as the disk source. The value for the volume attribute will be the output from the "Name" column of a virsh vol-list [pool-name] command.
Use the attribute mode ( since 1.1.1 ) to indicate how to represent the LUN as the disk source. Valid values are "direct" and "host". If mode is not specified, the default is to use "host". Using "direct" as the mode value indicates to use the storage pool's source element host 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 mode 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 disk configured using type 'block' or 'network' and device of 'lun' with respect to how the LUN is presented to and may be used by the guest. Since 1.0.5
nvme
To specify disk source for NVMe disk the source element has the following attributes:
type
The type of address specified in address sub-element. Currently, only pci value is accepted.
managed
This attribute instructs libvirt to detach NVMe controller automatically on domain startup (yes) or expect the controller to be detached by system administrator (no).
namespace
The namespace ID which should be assigned to the domain. According to NVMe standard, namespace numbers start from 1, including.
The difference between and 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 /dev/nvme0n1 via and therefore lower latencies can be achieved.
vhostuser
Enables the hypervisor to connect to another process using vhost-user protocol. Requires shared memory configured for the VM, for more details see access mode for memoryBacking element (See Memory Backing).
The source element has following mandatory attributes:
type
The type of char device. Currently only unix type is supported.
path
Path to the unix socket to be used as disk source.
Note that the vhost server replaces both the disk frontend and backend thus almost all of the disk properties can't be configured via the XML for this disk type. Additionally features such as blockjobs, incremental backups and snapshots are not supported for this disk type.
vhostvdpa
Enables the hypervisor to connect to a vDPA block device. Requires shared memory configured for the VM, for more details see access mode for memoryBacking element (See Memory Backing).
The source element has a mandatory attribute dev that specifies the fully-qualified path to the vhost-vdpa character device (e.g. /dev/vhost-vdpa-0).
ctl
Enables the hypervisor to connect to the FreeBSD CAM Target Layer (CTL), where CAM is a Common Access Method Storage subsystem. Since 12.0.0, bhyve.
The source element has a mandatory attribute dev that specifies the fully-qualified path to the CTL device (e.g. /dev/cam/ctl).
With "file", "block", and "volume", one or more optional sub-elements seclabel (See Security label) can be used to override the domain security labeling policy for just that source file. (NB, for "volume" type disk, seclabel is only valid when the specified storage volume is of 'file' or 'block' type). since 0.9.9
The source element may also have the index attribute with same semantics the index attribute of backingStore.
The source element may contain the following sub elements:
host
When the disk type is "network", the source may have zero or more host sub-elements used to specify the hosts to connect. The host 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.
Protocol
Meaning
Number of hosts
Default port
nbd
a server running nbd-server
only one
10809
iscsi
an iSCSI server
only one
3260
rbd
monitor servers of RBD
one or more
librados default
sheepdog
one of the sheepdog servers (default is localhost:7000)
zero or one
7000
gluster
a server running glusterd daemon
one or more ( Since 2.1.0 ), just one prior to that
24007
vxhs
a server running Veritas HyperScale daemon
only one
9999
nfs
a server running Network File System
only one ( Since 7.0.0 )
must be omitted
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. nfs only supports the use of a "tcp" transport, and does not support using a port at all so it must be omitted.
snapshot
The name attribute of snapshot element can optionally specify an internal snapshot name to be used as the source for storage protocols. Supported for 'rbd' since 1.2.11 (QEMU only).
config
The file attribute for the config 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' since 1.2.11 (QEMU only).
auth
Since 3.9.0, the auth element is supported for a disk type "network" that is using a source element with the protocol attributes "rbd", "iscsi", or "ssh". If present, the auth element provides the authentication credentials needed to access the source. It includes a mandatory attribute username, which identifies the username to use during authentication, as well as a sub-element secret with mandatory attribute type, to tie back to a libvirt secret object 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 uuid attribute with the UUID of the secret object or a usage attribute matching the key that was specified in the secret object.
encryption
Since 3.9.0, the encryption can be a sub-element of the source element for encrypted storage sources. If present, specifies how the storage source is encrypted See the Storage Encryption page for more information. Note that the 'qcow' format of encryption is broken and thus is no longer supported for use with disk images. ( Since 4.5.0 )
reservations
Since 4.4.0, the reservations can be a sub-element of the source element for storage sources (QEMU driver only). If present it enables persistent reservations for SCSI based disks. The element has one mandatory attribute managed with accepted values yes and no. If managed 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 source, which currently accepts only the following attributes: type with one value unix, path path to the socket, and finally mode which accepts one value client specifying the role of hypervisor. It's recommended to allow libvirt manage the persistent reservations.
initiator
Since 4.7.0, the initiator element is supported for a disk type "network" that is using a source element with the protocol attribute "iscsi". If present, the initiator element provides the initiator IQN needed to access the source via mandatory attribute name.
address
For disk of type nvme this element specifies the PCI address of the host NVMe controller. Since 6.0.0
slices
The slices element using its slice sub-elements allows configuring offset and size of either the location of the image format (slice type='storage') inside the storage source or the guest data inside the image format container (future expansion). The offset and size values are in bytes. Since 6.1.0
ssl
For https and ftps accessed storage it's possible to tweak the SSL transport parameters with this element. The verify attribute allows to turn on or off SSL certificate validation. Supported values are yes and no. Since 6.2.0
cookies
For http and https accessed storage it's possible to pass one or more cookies. The cookie name and value must conform to the HTTP specification. Since 6.2.0
readahead
The readahead element has a size attribute which specifies the size of the readahead buffer in bytes for protocols which support it. Note that '0' is considered as if the value is not provided. Since 6.2.0
timeout
The timeout element has a seconds attribute which specifies the connection timeout in seconds for protocols which support it. Note that '0' is considered as if the value is not provided. Since 6.2.0
identity
When using an nfs protocol, this is used to provide information on the configuration of the user and group. The element has two attributes, user and group. The user can provide these elements as user or group strings, or as user and group ID numbers directly if the string is formatted using a "+" at the beginning of the ID number. If either of these attributes is omitted, then that field is assumed to be the default value for the current system. If both user and group are intended to be default, then the entire element may be omitted.
When using an ssh protocol, this element is used to enable authentication via ssh keys. In this configuration, the element has three possible attributes. The username attribute is required and specifies the name of the user on the remote server. ssh keys can be specified in one of two ways. The first way is by adding them to an ssh-agent and providing the path to the ssh-agent socket in the agentsock attribute. This method works for ssh keys with or without password protection. Alternatively, for ssh keys without a password, the ssh key can be specified directly by setting the keyfile attribute.
reconnect
For disk type vhostuser configures reconnect timeout if the connection is lost. This is set with the two mandatory attributes enabled and timeout. For disk type network and protocol nbd the QEMU NBD reconnect delay can be set via attribute delay:
enabled
If the reconnect feature is enabled, accepts yes and no
timeout
The amount of seconds after which hypervisor tries to reconnect.
delay
Only for NBD hosts. The amount of seconds during which all requests are paused and will be rerun after a successful reconnect. After that time, any delayed requests and all future requests before a successful reconnect will immediately fail. If not set the default QEMU value is 0.
knownHosts
For storage accessed via the ssh protocol, this element configures a path to a file that will be used to verify the remote host. This file must contain the expected host key for the remote host or the connection will fail. The location of the file is specified via the path attribute. Since 9.8.0
dataStore
This element describes external data store, which is storage holding the actual data blocks of the given storage image. In such case the disk source image holds only the metadata. This feature is currently supported only by the qcow2 format. Since 10.10.0
The following attribute is supported in dataStore:
type
The type attribute represents the type of storage used by the data store, see disk type attribute above for more details and possible values.
Moreover, dataStore supports the following sub-elements:
format
The format element contains type attribute which specifies the internal format of the data store. Only raw value is supported.
source
This element has the same structure as the source element in disk. It specifies which file, device, or network location contains the data of the described data store.
For a "file" or "volume" disk type which represents a cdrom or floppy (the device attribute), it is possible to define policy what to do with the disk if the source file is not accessible. (NB, startupPolicy is not valid for "volume" disk unless the specified storage volume is of "file" type). This is done by the startupPolicy attribute ( since 0.9.7 ), accepting these values:
mandatory
fail if missing for any reason (the default)
requisite
fail if missing on boot up, drop if missing on migrate/restore/revert
optional
drop if missing at any start attempt
Since 1.1.2 the startupPolicy 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.
backingStore
This element describes the backing store used by the disk specified by sibling source element. Since 1.2.4. If the hypervisor driver does not support the backingStoreInput ( Since 5.10.0 ) domain feature the backingStore is ignored on input and only used for output to describe the detected backing chains of running domains. If backingStoreInput is supported the backingStore is used as the backing image of source or other backingStore overriding any backing image information recorded in the image metadata. An empty backingStore 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 backingStore:
type
The type attribute represents the type of disk used by the backing store, see disk type attribute above for more details and possible values.
index
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 virDomainBlockRebase API). For example, vda[2] refers to the backing store with index='2' of the disk with vda target.
Moreover, backingStore supports the following sub-elements:
format
The format element contains type attribute which specifies the internal format of the backing store, such as raw or qcow2.
The format element can contain metadata_cache subelement, which has identical semantics to the identically named subelement of driver of a disk.
source
This element has the same structure as the source element in disk. It specifies which file, device, or network location contains the data of the described backing store.
backingStore
If the backing store is not self-contained, the next element in the chain is described by nested backingStore element.
mirror
This element is present if the hypervisor has started a long-running block job operation, where the mirror location in the source sub-element will eventually have the same contents as the source, and with the file format in the sub-element format (which might differ from the format of the source). The details of the source sub-element are determined by the type attribute of the mirror, similar to what is done for the overall disk device element. The job attribute mentions which API started the operation ("copy" for the virDomainBlockRebase API, or "active-commit" for the virDomainBlockCommit API), since 1.2.7. The attribute ready, if present, tracks progress of the job: yes if the disk is known to be ready to pivot, or, since 1.2.7, abort or pivot if the job is in the process of completing. If ready is not present, the disk is probably still copying. For now, this element only valid in output; it is ignored on input. The source sub-element exists for all two-phase jobs since 1.2.6. Older libvirt supported only block copy to a file, since 0.9.12; for compatibility with older clients, such jobs include redundant information in the attributes file and format in the mirror element.
target
The target element controls the bus / device under which the disk is exposed to the guest OS. The dev 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 bus attribute specifies the type of disk device to emulate; possible values are driver specific, with typical values being "ide", "scsi", "virtio", "xen", "usb", "sata", "sd", or "nvme" "sd" since 1.1.2, "nvme" since 11.5.0. 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 tray 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 tray could be updated while the domain is running. The optional attribute removable sets the removable flag for USB or SCSI disks, and its value can be either "on" or "off", defaulting to "off". The optional attribute rotation_rate sets the rotation rate of the storage for disks on a SCSI, IDE, or SATA bus. Values in the range 1025 to 65534 are used to indicate rotational media speed in revolutions per minute. A value of one is used to indicate solid state, or otherwise non-rotational, storage. These values are not required to match the values of the underlying host storage. Since 0.0.3; bus attribute since 0.4.3; tray 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; "rotation_rate" attribute value since 7.3.0 The optional attribute dpofua (Since 11.10.0, only QEMU driver) controls the support of DPO(Disable Page Out) and FUA(Force Unit Access) properties of a SCSI disk cache access (both must be present or absent). If the value is omitted hypervisor default is applied (which may depend on the machine type version) and is the suggested setting.
throttlefilters
The optional throttlefilters element provides the ability to provide additional per-device throttle chain Since 11.2.0 For example, if we have four different disks and we want to limit I/O for each one and we also want to limit combined I/O of all four disks, we can leverage throttlefilters to achieve this goal by setting two throttlefilter for each disk: disk's own filter(e.g. limit2) and combined filter(e.g. limit012). The order of such throttlefilter doesn't matter within throttlefilters. throttlefilters and iotune should be used exclusively.
throttlefilter
The optional throttlefilter element is to reference defined throttle group.
iotune
The optional iotune element provides the ability to provide additional per-device I/O tuning, with values that can vary for each device (contrast this to the blkiotune element (See Block I/O Tuning), 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. Since 0.9.8
total_bytes_sec
The optional total_bytes_sec element is the total throughput limit in bytes per second. This cannot appear with read_bytes_sec or write_bytes_sec.
read_bytes_sec
The optional read_bytes_sec element is the read throughput limit in bytes per second.
write_bytes_sec
The optional write_bytes_sec element is the write throughput limit in bytes per second.
total_iops_sec
The optional total_iops_sec element is the total I/O operations per second. This cannot appear with read_iops_sec or write_iops_sec.
read_iops_sec
The optional read_iops_sec element is the read I/O operations per second.
write_iops_sec
The optional write_iops_sec element is the write I/O operations per second.
total_bytes_sec_max
The optional total_bytes_sec_max element is the maximum total throughput limit in bytes per second. This cannot appear with read_bytes_sec_max or write_bytes_sec_max.
read_bytes_sec_max
The optional read_bytes_sec_max element is the maximum read throughput limit in bytes per second.
write_bytes_sec_max
The optional write_bytes_sec_max element is the maximum write throughput limit in bytes per second.
total_iops_sec_max
The optional total_iops_sec_max element is the maximum total I/O operations per second. This cannot appear with read_iops_sec_max or write_iops_sec_max.
read_iops_sec_max
The optional read_iops_sec_max element is the maximum read I/O operations per second.
write_iops_sec_max
The optional write_iops_sec_max element is the maximum write I/O operations per second.
size_iops_sec
The optional size_iops_sec element is the size of I/O operations per second.
Throughput limits since 1.2.11 and QEMU 1.7
group_name
The optional group_name provides the ability 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.
group_name since 3.0.0 and QEMU 2.4
total_bytes_sec_max_length
The optional total_bytes_sec_max_length element is the maximum duration in seconds for the total_bytes_sec_max burst period. Only valid when the total_bytes_sec_max is set.
read_bytes_sec_max_length
The optional read_bytes_sec_max_length element is the maximum duration in seconds for the read_bytes_sec_max burst period. Only valid when the read_bytes_sec_max is set.
write_bytes_sec_max
The optional write_bytes_sec_max_length element is the maximum duration in seconds for the write_bytes_sec_max burst period. Only valid when the write_bytes_sec_max is set.
total_iops_sec_max_length
The optional total_iops_sec_max_length element is the maximum duration in seconds for the total_iops_sec_max burst period. Only valid when the total_iops_sec_max is set.
read_iops_sec_max_length
The optional read_iops_sec_max_length element is the maximum duration in seconds for the read_iops_sec_max burst period. Only valid when the read_iops_sec_max is set.
write_iops_sec_max
The optional write_iops_sec_max_length element is the maximum duration in seconds for the write_iops_sec_max burst period. Only valid when the write_iops_sec_max is set.
Throughput length since 2.4.0 and QEMU 2.6
driver
The optional driver element allows specifying further details related to the hypervisor driver used to provide the disk. Since 0.1.8
If the hypervisor supports multiple backend drivers, then the name attribute selects the primary backend driver name, while the optional type 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".
The optional cache attribute controls the cache mechanism, possible values are "default", "none", "writethrough", "writeback", "directsync" (since 0.9.5; like "writethrough", but it bypasses the host page cache) and "unsafe" (since 0.9.7; host may cache all disk io, and sync requests from guest are ignored). Since 0.6.0
The optional error_policy attribute controls how the hypervisor will behave on a disk read or write error, possible values are stop (suspend/pause the domain on error), report (report the error to the guest OS; since 0.9.7), ignore (ignore the error and try to continue), and enospace (suspend/pause the domain only if host storage is full; report the error to the guest OS otherwise).
The default is left to the discretion of the hypervisor. Since 0.8.0.
The optional rerror_policy attribute controls behavior for read errors only. If no rerror_policy is given, error_policy is used for both read and write errors. If rerror_policy is given, it overrides the error_policy for read errors. Also note that "enospace" is not a valid policy for read errors, so if error_policy is set to "enospace" and no rerror_policy is given, the read error policy will be left at its default. Since 0.9.7
The optional io attribute controls specific policies on I/O; qemu guests support "threads" and "native" Since 0.8.8, io_uring Since 6.3.0 (QEMU 5.0).
The optional ioeventfd attribute allows users to set domain I/O asynchronous handling 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. Since 0.9.3 (QEMU and KVM only) In general you should leave this option alone, unless you are very certain you know what you are doing.
The optional event_idx 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. Since 0.9.5 (QEMU and KVM only) In general you should leave this option alone, unless you are very certain you know what you are doing.
The optional copy_on_read 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. Since 0.9.10 (QEMU and KVM only)
The optional discard 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). Since 1.0.6 (QEMU and KVM only)
The optional detect_zeroes 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 discard above (it will act as "on" if discard is set to "ignore"). NB enabling the detection is a compute intensive operation, but can save file space and/or time on slow media. Since 2.0.0
The optional iothread attribute assigns the disk to an IOThread as defined by the range for the domain iothreads value. (See IOThreads Allocation). Multiple disks may be assigned to the same IOThread and are numbered from 1 to the domain iothreads value. Available for a disk device target configured to use "virtio" bus and "pci" or "ccw" address types. Since 1.2.8 (QEMU 2.1) Note: iothread is mutually exclusive with iothreads.
The optional iothreads sub-element allows specifying multiple IOThreads via the iothread sub-element with attribute id the disk will use for I/O operations. The virt queues (see queues attribute below) are automatically distributed among the configured iothreads.
Optionally the iothread element can have multiple queue subelements with mandatory id attribute specifying that the iothread should be used to handle given virt queue. If queue mapping is present the queues attribute of driver must be configured and all configured virt queues must be included in the mapping. The virtio-blk device exposes request virt queues 0 to N-1 where N is the number of queues configured for the device.
Since 10.0.0 (QEMU 9.0, virtio disks only).
Examples:
The optional statistics sub-element allows configuring statistics collection in configurable intervals for the given disk. Intervals are configured by sub-elements with interval attribute configuring the collection window duration in seconds. The statistics are available via the bulk statistics API.
Example:
Since 11.9.0 (QEMU 10.2, virtio, ide, scsi disks only).
The optional queues attribute specifies the number of virt queues for virtio-blk ( Since 3.9.0 ) or vhost-user-blk ( Since 7.1.0 )
The optional queue_size attribute specifies the size of each virt queue for virtio-blk or vhost-user-blk. ( Since 7.8.0 )
For virtio disks, Virtio-related options can also be set. ( Since 3.5.0 )
The optional metadata_cache subelement controls aspects related to the format specific caching of storage image metadata. Note that this setting applies only on the top level image; the identically named subelement of backingStore's format element can be used to specify cache settings for the backing image.
Since 7.0.0 the maximum size of the metadata cache of qcow2 format driver of the qemu hypervisor can be controlled via the max_size subelement (see example below).
The optional discard_no_unref attribute can be set to control the way the qemu hypervisor handles guest discard commands inside the qcow2 image. When enabled, a discard request from within the guest will mark the qcow2 cluster as zero, but will keep the reference/offset of that cluster. But it will still pass the discard further to the lower layer. This will resolve fragmentation within the qcow2 image. Since 9.5.0 (QEMU 8.1)
In the majority of cases the default configuration used by the hypervisor is sufficient so modifying this setting should not be necessary. For specifics on how the metadata cache of qcow2 in qemu behaves refer to the qemu qcow2 cache docs
Example:
12341234
backenddomain
The optional backenddomain element allows specifying a backend domain (aka driver domain) hosting the disk. Use the name attribute to specify the backend domain name. Since 1.2.13 (Xen only)
boot
Specifies that the disk is bootable. The order 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 loadparm 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 loadparm to select a boot entry. Since 3.5.0 The per-device boot elements cannot be used together with general boot elements in BIOS bootloader section. Since 0.8.8
encryption
since:Since 3.9.0 the encryption element is preferred to be a sub-element of the source element. If present, specifies how the volume is encrypted using "qcow". See the Storage Encryption page for more information.
readonly
If present, this indicates the device cannot be modified by the guest. For now, this is the default for disks with attribute device='cdrom'.
shareable
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.
transient
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, snapshots, or blockjobs. Only supported in vmx hypervisor (Since 0.9.5) and qemu hypervisor (Since 6.9.0).
In cases where the source image of the disk is supposed to be shared between multiple concurrently running VMs the optional shareBacking attribute should be set to yes. Note that hypervisor drivers may need to hotplug such disk and thus it works only with configurations supporting hotplug. Since 7.4.0
Hypervisors may need to store a temporary file containing the data written by the domain while running, which may be stored in the same location as the original source of the disk (The qemu driver stores the temporary files as $(origsource).TRANSIENT-$(vmname), where $(origsource) is the full original path of the disk source and $(vmname) is the name of the domain).
serial
If present, this specify serial number of virtual hard drive. For example, it may look like WD-WMAP9A966149. Not supported for scsi-block devices, that is those using disk type 'block' using device 'lun' on bus 'scsi'. Also not supported for multiple NVMe devices on the same controller since those have serial number per controller and not per disk. Since 0.7.1
Note that depending on hypervisor and device type the serial number may be truncated silently. IDE/SATA devices are commonly limited to 20 characters. SCSI devices depending on hypervisor version are limited to 20, 36 or 247 characters.
Hypervisors may also start rejecting overly long serials instead of truncating them in the future so it's advised to avoid the implicit truncation by testing the desired serial length range with the desired device and hypervisor combination.
wwn
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. Since 0.10.1
vendor
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. Only for 'scsi' bus.:since:Since 1.0.1
product
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 for 'scsi' (Since 1.0.1). For 'sata' or 'ide' not longer than 40 printable characters (Since 11.1.0). Other bus is not supported.
address
If present, the address element ties the disk to a given slot of a controller (the actual device can often be inferred by libvirt, although it can be be explicitly specified. See Controllers). The type attribute is mandatory, and is typically "pci" or "drive". For a "pci" controller, additional attributes for bus, slot, and function must be present, as well as optional domain and multifunction (since 0.9.7). Multifunction defaults to 'off'. For a "drive" controller, additional attributes controller, bus, target (since 0.9.11), and unit are available, each defaulting to 0.
auth
Since 3.9.0, the auth element is preferred to be a sub-element of the source element. The element is still read and managed as a disk sub-element. It is invalid to use auth as both a sub-element of disk and source. Since 0.9.7
geometry
The optional geometry element provides the ability to override geometry settings. This mostly useful for S390 DASD-disks or older DOS-disks. Since 0.10.0
cyls
The cyls attribute is the number of cylinders.
heads
The heads attribute is the number of heads.
secs
The secs attribute is the number of sectors per track.
trans
The optional trans attribute is the BIOS-Translation-Modus (none, lba or auto)
blockio
If present, the blockio element allows to override any of the block device properties listed below. Since 0.10.2 (QEMU and KVM)
logical_block_size
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.
physical_block_size
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.
discard_granularity
The smallest amount of data that can be discarded in a single operation. It impacts the unmap operations and it must be a multiple of a logical_block_size. This is usually properly configured by the hypervisor.
Filesystems
A directory on the host that can be accessed directly from the guest. since 0.3.3, since 0.8.5 for QEMU/KVM
...
...
...
filesystem
The filesystem attribute type specifies the type of the source. The possible values are:
mount
A host directory to mount in the guest. Used by LXC, OpenVZ (since 0.6.2) and QEMU/KVM (since 0.8.5). This is the default type if one is not specified. This mode also has an optional sub-element driver, with an attribute type='path' or type='handle' (since 0.9.7). The driver block has an optional attribute wrpolicy that further controls interaction with the host page cache; omitting the attribute gives default behavior, while the value immediate means that a host writeback is immediately triggered for all pages touched during a guest file write operation (since 0.9.10) . Since 6.2.0, type='virtiofs' is also supported. Using virtiofs requires setting up shared memory, see the guide: Virtiofs
template
OpenVZ filesystem template. Only used by OpenVZ driver.
file
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.
block
A host block device to mount in the guest. The filesystem format will be autodetected. Only used by LXC driver (since 0.9.5).
ram
An in-memory filesystem, using memory from the host OS. The source element has a single attribute usage which gives the memory usage limit in KiB, unless units are specified by the units attribute. Only used by LXC driver. (since 0.9.13)
bind
A directory inside the guest will be bound to another directory inside the guest. Only used by LXC driver (since 0.9.13)
The filesystem element has an optional attribute accessmode which specifies the security mode for accessing the source (since 0.8.5). Currently this only works with type='mount' for the QEMU/KVM driver. For driver type virtiofs, only passthrough is supported. For other driver types, the possible values are:
passthrough
The source is accessed with the permissions of the user inside the guest. This is the default accessmode if one is not specified. More info
mapped
The source is accessed with the permissions of the hypervisor (QEMU process). More info
squash
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. More info
Since 5.2.0, the filesystem element has an optional attribute model with supported values "virtio", "virtio-transitional" or "virtio-non-transitional". See virtio device models for more details.
The filesystem element has optional attributes fmode and dmode. These two attributes control the creation mode for files and directories when used with the mapped value for accessmode (since 6.10.0, requires QEMU 2.10 ). If not specified, QEMU creates files with mode 600 and directories with mode 700. The setuid, setgid, and sticky bit are unsupported.
The filesystem element has an optional attribute multidevs which specifies how to deal with a filesystem export containing more than one device, in order to avoid file ID collisions on guest when using 9pfs ( since 6.3.0, requires QEMU 4.2 ). This attribute is not available for virtiofs. The possible values are:
default
Use QEMU's default setting (which currently is warn).
remap
This setting allows guest to access multiple devices per export without encountering misbehaviours. Inode numbers from host are automatically remapped on guest to actively prevent file ID collisions if guest accesses one export containing multiple devices.
forbid
Only allow to access one device per export by guest. Attempts to access additional devices on the same export will cause the individual filesystem access by guest to fail with an error and being logged (once) as error on host side.
warn
This setting resembles the behaviour of 9pfs prior to QEMU 4.2, that is no action is performed to prevent any potential file ID collisions if an export contains multiple devices, with the only exception: a warning is logged (once) on host side now. This setting may lead to misbehaviours on guest side if more than one device is exported per export, due to the potential file ID collisions this may cause on guest side in that case.
driver
The optional driver element allows specifying further details related to the hypervisor driver used to provide the filesystem. Since 1.0.6
If the hypervisor supports multiple backend drivers, then the type attribute selects the primary backend driver name, while the format 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".
For virtio-backed devices, Virtio-related options can also be set. ( Since 3.5.0 )
For virtiofs, the queue attribute can be used to specify the queue size (i.e. how many requests can the queue fit). ( Since 6.2.0 )
QEMU supports mtp which exposes a virtual USB MTP device to the guest. ( Since 10.2.0 )
binary
The optional binary element can tune the options for virtiofsd. All of the following attributes and elements are optional. The attribute path can be used to override the path to the daemon. Attribute xattr enables the use of filesystem extended attributes. Caching can be tuned via the cache element, possible mode values being none and always. Locking can be controlled via the lock element - attributes posix and flock both accepting values on or off. ( Since 6.2.0 ) The sandboxing method used by virtiofsd can be configured with the sandbox element, possible mode values being namespace and chroot, see the virtiofsd documentation for more details. ( Since 7.2.0 ) Element thread_pool accepts one attribute size which defines the maximum thread pool size. A value of "0" disables the pool. The thread pool helps increase the number of requests in flight when used with storage that has a higher latency. However, it has an overhead, and so for fast, low latency filesystems, it may be best to turn it off. ( Since 8.5.0 ) Element openfiles accepts one attribute max which defines the maximum number of file descriptors. Non-positive values are forbidden. The upper bound on the number of open files is implementation defined. ( Since 10.6.0 )
source
The resource on the host that is being accessed in the guest. The name attribute must be used with type='template', and the dir attribute must be used with type='mount'. For virtiofs, the socket attribute can be used to connect to a virtiofsd daemon launched outside of libvirt. In that case, the target element does not apply and neither do most virtiofs-related options, since they are controlled by virtiofsd, not libvirtd. The usage attribute is used with type='ram' to set the memory limit in KiB, unless units are specified by the units attribute.
target
Where the source 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.
idmap
For virtiofs, an idmap element can be specified to map IDs in the user namespace. See the Container boot section for the syntax of the element. Since 10.0.0
readonly
Enables exporting filesystem as a readonly mount for guest, by default read-write access is given (works for QEMU/KVM driver, Since 11.0.0, requires virtiofs 1.13.0 ).
space_hard_limit
Maximum space available to this guest's filesystem. Since 0.9.13 Only supported by the OpenVZ driver.
space_soft_limit
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. Since 0.9.13 Only supported by the OpenVZ driver.
Device Addresses
Many devices have an optional 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.
Every address has a mandatory attribute type 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 device uses type='drive', while a device would use type='pci' on i686 or x86_64 guests, or type='spapr-vio' on PowerPC64 pseries guests. Each address type has further optional attributes that control where on the bus the device will be placed:
pci
PCI addresses have the following additional attributes: domain (a 2-byte hex integer, not currently used by qemu), bus (a hex value between 0 and 0xff, inclusive), slot (a hex value between 0x0 and 0x1f, inclusive), and function (a value between 0 and 7, inclusive). Also available is the multifunction attribute, which controls turning on the multifunction bit for a particular slot/function in the PCI control register ( since 0.9.7, requires QEMU 0.13 ). multifunction defaults to 'off', but should be set to 'on' for function 0 of a slot that will have multiple functions used. ( Since 4.10.0 ), PCI address extensions depending on the architecture are supported. For example, PCI addresses for S390 guests will have a zpci child element, with two attributes: uid (a hex value between 0x0001 and 0xffff, inclusive), and fid (a hex value between 0x00000000 and 0xffffffff, inclusive) used by PCI devices on S390 for User-defined Identifiers and Function Identifiers. Since 1.3.5, some hypervisor drivers may accept an 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). The relationship between the PCI addresses configured in the domain XML and those seen by the guest OS can sometime seem confusing: a separate document describes how PCI addresses work in more detail.
drive
Drive addresses have the following additional attributes: controller (a 2-digit controller number), bus (a 2-digit bus number), target (a 2-digit target number), and unit (a 2-digit unit number on the bus).
virtio-serial
Each virtio-serial address has the following additional attributes: controller (a 2-digit controller number), bus (a 2-digit bus number), and slot (a 2-digit slot within the bus).
ccid
A CCID address, for smart-cards, has the following additional attributes: bus (a 2-digit bus number), and slot attribute (a 2-digit slot within the bus). Since 0.8.8.
usb
USB addresses have the following additional attributes: bus (a hex value between 0 and 0xfff, inclusive), and port (a dotted notation of up to four octets, such as 1.2 or 2.1.3.1).
spapr-vio
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: reg (the hex value address of the starting register). Since 0.9.9.
ccw
S390 guests with a machine value of s390-ccw-virtio use the native CCW bus for I/O devices. CCW bus addresses have the following additional attributes: cssid (a hex value between 0 and 0xfe, inclusive), ssid (a value between 0 and 3, inclusive) and devno (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. Since 1.0.4
virtio-mmio
This places the device on the virtio-mmio transport, which is currently only available for some armv7l and aarch64 virtual machines. virtio-mmio addresses do not have any additional attributes. Since 1.1.3 If the guest architecture is aarch64 and the machine type is virt, 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. Since 3.0.0
isa
ISA addresses have the following additional attributes: iobase and irq. Since 1.2.1
unassigned
For PCI hostdevs, 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. is an invalid address type for all other device types. Since 6.0.0
Virtio-related options
QEMU's virtio devices have some attributes related to the virtio transport under the driver element: The iommu attribute enables the use of emulated IOMMU by the device. The attribute ats controls the Address Translation Service support for PCIe devices. This is needed to make use of IOTLB support (see IOMMU devices). Possible values are on or off. Since 3.5.0
The attribute packed controls if QEMU should try to use packed virtqueues. Compared to regular split queues, packed queues consist of only a single descriptor ring replacing available and used ring, index and descriptor buffer. This can result in better cache utilization and performance. If packed virtqueues are actually used depends on the feature negotiation between QEMU, vhost backends and guest drivers. Possible values are on or off. Since 6.3.0 (QEMU and KVM only)
This optional attribute page_per_vq controls the layout of the notification capabilities exposed to the guest. When enabled, each virtio queue will have a dedicated page on the device BAR exposed to the guest. It is recommended to be used when vDPA is enabled on the hypervisor, as it enables mapping the notification area to the physical device, which is only supported in page granularity. The default is determined by QEMU. Since 7.9.0 (QEMU 2.8) Note: In general you should leave this option alone, unless you are very certain you know what you are doing.
Virtio device models
Virtio devices come in several variants, some of which are only applicable to certain machine types or scenarios. The variant can be chosen via the model attribute, which supports the following values:
virtio
This is the recommended choice in the absence of guest OS specific constraints, as it will will generally work correctly across a large range of architectures, machine types and libvirt versions.
Since 5.2.0, the following values can additionally be used with machine types based on PCI (either conventional PCI or PCI Express):
virtio-transitional
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.
virtio-non-transitional
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.
While the information outlined above applies to most virtio devices, there are a few exceptions:
for SCSI controllers, there is no virtio model available due to historical reasons: use virtio-scsi instead, which behaves the same as virtio does for other devices. Both virtio-transitional and virtio-non-transitional work with SCSI controllers;
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 virtio, which will result in a non-transitional device.
For more details see the qemu patch posting and the virtio-1.0 spec.
Controllers
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 PCI topology for guests where device hotplug is expected.
...
...
...
</devices>
...
Each controller has a mandatory attribute type, which must be one of 'ide', 'fdc', 'scsi', 'sata', 'usb', 'ccid', 'virtio-serial' or 'pci', and a mandatory attribute index which is the decimal integer describing in which order the bus controller is encountered (for use in controller attributes of elements). Since 1.3.5 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:
virtio-serial
The virtio-serial controller has two additional optional attributes ports and vectors, which control how many devices can be connected through the controller. Since 5.2.0, it supports an optional attribute model which can be 'virtio', 'virtio-transitional' or 'virtio-non-transitional'. See virtio device models for more details.
scsi
A scsi controller has an optional attribute model, which is one of 'auto', 'buslogic', 'ibmvscsi', 'lsilogic', 'lsisas1068', 'lsisas1078', 'virtio-scsi', 'vmpvscsi', 'virtio-transitional', 'virtio-non-transitional', 'ncr53c90' (as builtin implicit controller only), 'am53c974', 'dc390'. See virtio device models for more details.
usb
A usb controller has an optional attribute model, 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, since 0.10.0, if the USB bus needs to be explicitly disabled for the guest, model='none' may be used. Since 1.0.5, no default USB controller will be built on s390. Since 1.3.5, USB controllers accept a ports attribute to configure how many devices can be connected to the controller.
ide
Since 3.10.0 for the vbox driver, the ide controller has an optional attribute model, which is one of "piix3", "piix4" or "ich6".
xenbus
Since 5.2.0, the xenbus controller has an optional attribute maxGrantFrames, which specifies the maximum number of grant frames the controller makes available for connected devices. Since 6.3.0, the xenbus controller supports the optional maxEventChannels attribute, which specifies maximum number of event channels (PV interrupts) that can be used by the guest.
nvme
Supported Since 11.5.0, the nvme controller can be used to support NVMe disks. It has an optional serial sub-element just like regular disks do.
Note: The PowerPC64 "spapr-vio" addresses do not have an associated controller.
For controllers that are themselves devices on a PCI or USB bus, an optional sub-element can specify the exact relationship of the controller to its master bus, with semantics described in the Device Addresses section.
An optional sub-element driver can specify the driver specific options:
queues
The optional queues attribute specifies the number of queues for the controller. For best performance, it's recommended to specify a value matching the number of vCPUs. Since 1.0.5 (QEMU and KVM only)
cmd_per_lun
The optional cmd_per_lun attribute specifies the maximum number of commands that can be queued on devices controlled by the host. Since 1.2.7 (QEMU and KVM only)
max_sectors
The optional max_sectors 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. Since 1.2.7 (QEMU and KVM only)
ioeventfd
The optional ioeventfd attribute specifies whether the controller should use I/O asynchronous handling or not. Accepted values are "on" and "off". Since 1.2.18
iothread
Supported for controller type scsi using model virtio-scsi for address types pci and ccw since 1.3.5 (QEMU 2.4). The optional iothread attribute assigns the controller to an IOThread as defined by the range for the domain iothreads (See IOThreads Allocation). Each SCSI disk assigned to use the specified controller will utilize the same IOThread. If a specific IOThread is desired for a specific SCSI disk, then multiple controllers must be defined each having a specific iothread value. The iothread value must be within the range 1 to the domain iothreads value.
iothreads
Supported for virtio-scsi controllers using address types pci and ccw. since 11.2.0 (QEMU 10.0). Mutually exclusive with iothread.
The optional iothreads sub-element allows specifying multiple IOThreads via the iothread sub-element with attribute id the virtio-scsi controller will use for I/O operations. The virt queues (see queues attribute of driver) are automatically distributed among the configured iothreads.
Optionally the iothread element can have multiple queue subelements with mandatory id atribute specifying that the iothread should be used to handle given virt queue. If queue mapping is present the queues attribute of driver must be configured and all configured virt queues must be included in the mapping. The virtio-scsi device exposes request virt queues 0 to N-1 where N is the number of queues configured for the device.
Example:
<driver queues='4>
</driver>
virtio options
For virtio controllers, Virtio-related options can also be set. ( Since 3.5.0 )
USB companion controllers have an optional sub-element 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 index value should be equal. Not all controller models can be used as companion controllers and libvirt might provide some sensible defaults (settings of master startport and function of an address) for some particular models. Preferred companion controllers are ich-uhci[123].
...
...
...
PCI controllers have an optional model attribute; possible values for this attribute are
pci-root, pci-bridge ( since 1.0.5 )
pcie-root, dmi-to-pci-bridge ( since 1.1.2 )
pcie-root-port, pcie-switch-upstream-port, pcie-switch-downstream-port ( since 1.2.19 )
pci-expander-bus, pcie-expander-bus ( since 1.3.4 )
pcie-to-pci-bridge ( since 4.3.0 )
The root controllers (pci-root and pcie-root) have an optional pcihole64 element specifying how big (in kiB, or in the unit specified by pcihole64's unit 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). Since 1.1.2 (QEMU only)
PCI controllers also have an optional subelement with an attribute name. 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 attribute. In almost all cases, you should not manually add a subelement to a controller, nor should you modify one that is automatically generated by libvirt. Since 1.2.19 (QEMU only).
PCI controllers also have an optional subelement 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 subelement to a controller, nor should you modify the values in the those that are automatically generated by libvirt. Since 1.2.19 (QEMU only).
chassisNr
PCI controllers that have attribute model="pci-bridge", can also have a chassisNr attribute in the 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.
chassis
pcie-root-port and pcie-switch-downstream-port controllers can also have a chassis attribute in the 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.
port
pcie-root-port and pcie-switch-downstream-port controllers can also have a port attribute in the 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.
hotplug
pci-root (Since 7.9.0), pcie-root-port (Since 6.3.0) and pcie-switch-downstream-port controllers (Since 6.3.0) can also have a hotplug attribute in the subelement, which is used to disable hotplug/unplug of devices on a particular controller. For the pci-root controller, the setting affects the ACPI based hotplug. For the rest, the setting affects both ACPI based hotplug as well as PCIE native hotplug. The default setting of hotplug is on; it should be set to off to disable hotplug/unplug of devices on a particular controller.
busNr
pci-expander-bus and pcie-expander-bus controllers can have an optional busNr 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).
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).
node
Some PCI controllers (pci-expander-bus for the pc machine type, pcie-expander-bus for the q35 machine type and, since 3.6.0, pci-root for the pseries machine type) can have an optional subelement within the 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).
index
pci-root controllers for pSeries guests use this attribute to record the order they will show up in the guest. Since 3.6.0
memReserve
Some PCI devices have non-prefetchable memory bar larger than 2MiB. Use this attribute to override value computed by firmware and thus make controller reserve more memory (in KiB) so that such PCI device can be hot plugged. For cold plugged PCI devices, the firmware will automatically reserve the correct amount of memory. Since 10.3.0
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.
...
...
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 pcie-to-pci-bridge controller will automatically be added: this controller, which plugs into a pcie-root-port, provides 31 usable PCI slots (1-31) with hotplug support ( since 4.3.0 ). If the QEMU binary doesn't support the corresponding device, then a dmi-to-pci-bridge 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. ( since 1.1.2 ).
Domains with an implicit pcie-root can also add controllers with model='pcie-root-port', model='pcie-switch-upstream-port', and model='pcie-switch-downstream-port'. 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). ( since 1.2.19 )
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. ( since 1.2.19 )
...
...
Device leases
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.
...
...
someareasomekey
...
...
lockspace
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.
key
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.
target
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.
Host device assignment
USB / PCI / SCSI devices
USB (since 0.4.4), PCI (since 0.6.0, KVM only) and SCSI (since 1.0.6, KVM only) devices attached to the host can be passed through to the guest using the hostdev element.
...
</source>
...
or:
...
</source>
...
or:
...
</source>
...
or:
...
</source>
...
or:
...
...
or:
...
</source>
</source>
...
hostdev
The hostdev element is the main container for describing host devices. For each device, the mode is always "subsystem" and the type is one of the following values with additional attributes noted.
usb
USB devices are detached from the host on guest startup and reattached after the guest exits or the device is hot-unplugged.
pci
For PCI devices, when managed 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 managed is omitted or "no", the user is responsible to call virNodeDeviceDetachFlags (or virsh nodedev-detach before starting the guest or hot-plugging the device and virNodeDeviceReAttach (or virsh nodedev-reattach) after hot-unplug or stopping the guest. Since 10.3.0 an optional display attribute may be used to enable using a vgpu device as a display device for the guest. Supported values are either on or off (default). There is also an optional ramfb attribute with values of either on or off (default). When enabled, the ramfb attribute provides a memory framebuffer device to the guest. This framebuffer allows the vgpu to be used as a boot display before the gpu driver is loaded within the guest. ramfb requires the display attribute to be set to on.
scsi
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 sgio ( since 1.0.6, but currently no longer supported by any hypervisor driver ) 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 rawio (since 1.2.9 ) attribute indicates whether the lun needs the rawio capability. Valid settings are "yes" or "no". See the rawio description within the Hard drives, floppy disks, CDROMs section. If a disk lun in the domain already has the rawio capability, then this setting not required.
scsi_host
since 2.5.0 For SCSI devices, user is responsible to make sure the device is not used by host. This type passes all LUNs presented by a single HBA to the guest. Since 5.2.0, the model attribute can be specified further with "virtio", "virtio-transitional" or "virtio-non-transitional". See virtio device models for more details.
mdev
For mediated devices ( Since 3.2.0 ) the model attribute specifies the device API which determines how the host's vfio driver will expose the device to the guest. Currently, model='vfio-pci', model='vfio-ccw' ( Since 4.4.0 ) and model='vfio-ap' ( Since 4.9.0 ) is supported. MDEV section provides more information about mediated devices as well as how to create mediated devices on the host. Since 4.6.0 (QEMU 2.12) an optional display 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 Video devices. This attribute is limited to model='vfio-pci' only. Supported values are either on or off (default is 'off'). It is required to use a graphical framebuffer (See Graphical framebuffers) in order to use this attribute, currently only supported with VNC, Spice and egl-headless graphics devices. Since version 5.10.0, there is an optional ramfb attribute for devices with model='vfio-pci'. Supported values are either on or off (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.
Note: There are also some implications on the usage of guest's address type depending on the model attribute, see the address element below.
Note: The managed attribute is only used with type='pci' and is ignored by all the other device types, thus setting managed explicitly with other than a PCI device has the same effect as omitting it. Similarly, model attribute is only supported by mediated devices and ignored by all other device types.
source
The source element describes the device as seen from the host using the following mechanism to describe:
usb
The USB device can either be addressed by vendor / product id using the vendor and product elements or by the device's address on the host using the address element.
Since 1.0.0, the source element of USB devices may contain startupPolicy 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:
mandatory
fail if missing for any reason (the default)
requisite
fail if missing on boot up, drop if missing on migrate/restore/revert
optional
drop if missing at any start attempt
Since 8.6.0, the source element can contain guestReset attribute with the following value:
off
all guest initiated device reset requests are ignored
uninitialized
device request is ignored if device is initialized, otherwise reset is performed
on
device is reset on every guest initiated request
This attribute can be helpful when assigning an USB device with a firmware that crashes on reset.
pci
PCI devices can only be described by their address. Since 6.8.0 (Xen only), the source element of a PCI device may contain the writeFiltering attribute to control write access to the PCI configuration space. By default Xen only allows writes of known safe values to the configuration space. Setting writeFiltering='no' will allow all writes to the device's PCI configuration space.
scsi
SCSI devices are described by both the adapter and address elements. The address element includes a bus attribute (a 2-digit bus number), a target attribute (a 10-digit target number), and a unit attribute (a 20-digit unit number on the bus). Not all hypervisors support larger target and unit values. It is up to each hypervisor to determine the maximum value supported for the adapter.
Since 1.2.8, the source element of a SCSI device may contain the protocol attribute. When the attribute is set to "iscsi", the host device XML follows the network disk device (See Hard drives, floppy disks, CDROMs) using the same name attribute and optionally using the auth element to provide the authentication credentials to the iSCSI server.
Since 6.7.0, the optional initiator sub-element controls the IQN of the initiator ran by the hypervisor via it's <iqn name='iqn...' subelement.
scsi_host
Since 2.5.0, multiple LUNs behind a single SCSI HBA are described by a protocol attribute set to "vhost" and a wwpn attribute that is the vhost_scsi wwpn (16 hexadecimal digits with a prefix of "naa.") established in the host configfs.
mdev
Mediated devices ( Since 3.2.0 ) are described by the address element. The address element contains a single mandatory attribute uuid.
vendor, product
The vendor and product elements each have an id 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.
boot
Specifies that the device is bootable. The order attribute determines the order in which devices will be tried during boot sequence. The per-device boot elements cannot be used together with general boot elements in BIOS bootloader section. Since 0.8.8 for PCI devices, Since 1.0.1 for USB devices.
rom
The rom element is used to change how a PCI device's ROM is presented to the guest. The optional bar 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"). Since 0.9.7 (QEMU and KVM only). The optional file 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). Since 0.9.10 (QEMU and KVM only). The optional enabled attribute can be set to no 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 bar or file attributes will be rejected. Since 4.3.0 (QEMU and KVM only).
address
The address element for USB devices has a bus and device 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 lspci or with virsh nodedev-list. 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 model attribute of element hostdev, e.g. any address type other than PCI for vfio-pci device API or any address type other than CCW for vfio-ccw device API will result in an error. See the Device Addresses section for more details on the address element.
driver
PCI hostdev devices can have an optional driver subelement that specifies which host driver to bind to the device when preparing it for assignment to a guest. Since 10.0.0 (useful for QEMU and KVM only). This is done by setting the element's model attribute, for example:
...
...
tells libvirt to bind the driver "vfio-pci-igb" to the device on the host before handing it off to QEMU for assignment to the guest. Normally libvirt will bind the device to the "best match" VFIO-type driver that it finds in the kernel's modules.alias file (based on matching the corresponding fields of the device's modalias file in sysfs) or to the generic "vfio-pci" driver if no better match is found (vfio-pci is always used prior to libvirt 10.0.0), but in cases when the correct driver isn't listed in modules.alias then the desired device-specific driver can be forced by setting driver name, or if the device-specific driver that is found is "problematic" in some way, the generic vfio-pci driver similarly be forced.
(Note: Since 1.0.5, the name attribute has been described to be used to select the type of PCI device assignment ("vfio", "kvm", or "xen"), but those values have been mostly useless, since the type of device assignment is actually determined by which hypservisor is in use. This means that you may occasionally see or in a domain's status XML, or more rarely in config, but those specific values are essentially ignored.)
readonly
Indicates that the device is readonly, only supported by SCSI host device now. Since 1.0.6 (QEMU and KVM only)
shareable
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. Since 1.0.6, but only works as expected since 1.2.2.
ACPI Generic Initiators
A host device may include an element to create ACPI Generic Initiator objects for the device in QEMU.
This can be used for NVIDIA Multi-Instance GPU (MIG) configurations, where a physical GPU is partitioned into multiple isolated instances, each associated with one or more virtual NUMA nodes.
By attaching an <acpi nodeset=.../> element to the MIG device in the domain XML, the guest will configure the correct partitioning for that instance.
...
</source>
Attributes of :
nodeset
A list of NUMA node IDs that will be associated with the device. Each node in the set causes libvirt to create an acpi-generic-initiator object in QEMU, tied to this device.
The value uses the standard libvirt nodeset syntax (e.g. 0-3,5).
If the element is omitted, no acpi-generic-initiator objects are created for the device.
Block / character devices
Block / character devices from the host can be passed through to the guest using the hostdev element. This is only possible with container based virtualization. Devices are specified by a fully qualified path. since after 1.0.1 for LXC:
...
/dev/sdf1
</source>
...
...
/dev/input/event3
</source>
...
...
eth0
</source>
...
hostdev
The hostdev element is the main container for describing host devices. For block/character device passthrough mode is always "capabilities" and type is "storage" for a block device, "misc" for a character device and "net" for a host network interface.
source
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.
Redirected devices
USB device redirection through a character device is supported since after 0.9.5 (KVM only):
...
...
redirdev
The redirdev element is the main container for describing redirected devices. bus must be "usb" for a USB device. An additional attribute type is required, matching one of the supported serial device types (See Consoles, serial, parallel & channel devices), to describe the host side of the tunnel; type='tcp' or type='spicevmc' (which uses the usbredir channel of a SPICE graphics device (See Graphical framebuffers)) are typical.
The redirdev element has an optional sub-element which can tie the device to a particular controller. Further sub-elements, such as , may be required according to the given type, although a sub-element is not required (since the consumer of the character device is the hypervisor itself, rather than a device visible in the guest).
boot
Specifies that the device is bootable. The order attribute determines the order in which devices will be tried during boot sequence. The per-device boot elements cannot be used together with general boot elements in BIOS bootloader section. ( Since 1.0.1 )
redirfilter
Theredirfilterelement is used for creating the filter rule to filter out certain devices from redirection. It uses sub-element to define each filter rule. class 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 vendor and product attributes. version is the device revision from the bcdDevice field (not the version of the USB protocol). These four attributes are optional and -1 can be used to allow any value for them. allow attribute is mandatory, 'yes' means allow, 'no' for deny.
Smartcard devices
A virtual smartcard device can be supplied to the guest via the smartcard 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. Since 0.8.8
...
cert1cert2cert3/etc/pki/nssdb/
...
The element has a mandatory attribute mode. 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.
host
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 sub-element.
host-certificates
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 certutil -d /etc/pki/nssdb -x -t CT,CT,CT -S -s CN=cert1 -n cert1, and the resulting three certificate names must be supplied as the content of each of three sub-elements. An additional sub-element can specify the absolute path to an alternate directory (matching the -d option of the certutil command when creating the certificates); if not present, it defaults to /etc/pki/nssdb.
passthrough
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 type is required, matching one of the supported serial device types (See Consoles, serial, parallel & channel devices), to describe the host side of the tunnel; type='tcp' or type='spicevmc' (which uses the smartcard channel of a SPICE graphics device (See Graphical framebuffers)) are typical. Further sub-elements, such as , may be required according to the given type, although a sub-element is not required (since the consumer of the character device is the hypervisor itself, rather than a device visible in the guest).
Each mode supports an optional sub-element (See Device Addresses), which fine-tunes the correlation between the smartcard and a ccid bus controller. For now, qemu only supports at most one smartcard, with an address of bus=0 slot=0.
Network interfaces
...
...
There are several possibilities for specifying a network interface visible to the guest. Each subsection below provides more details about common setup options.
Since 1.2.10 ), the interface element property trustGuestRxFilters 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 yes. The default setting for the attribute is no 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.
Each element has an optional sub-element that can tie the interface to a particular pci slot, with attribute type='pci' as documented in the Device Addresses section.
Since 6.6.0, one can force libvirt to keep the provided MAC address when it's in the reserved VMware range by adding a type="static" attribute to the element. Note that this attribute is useless if the provided MAC address is outside of the reserved VMWare ranges.
Since 11.2.0, the element can optionally contain currentAddress attribute (output only), which contains new MAC address if the guest changed it. This is currently implemented only for the model type virtio in QEMU/KVM and requires setting trustGuestRxFilters to yes.
Since 7.3.0, one can set the ACPI index against network interfaces. With some operating systems (eg Linux with systemd), the ACPI index is used to provide network interface device naming, that is stable across changes in PCI addresses assigned to the device. This value is required to be unique across all devices and be between 1 and (16*1024-1).
Virtual network
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 definition Since 0.9.4 ).
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 element given), NAT'ing to an explicit network device or to the default route (), routed with no NAT (), or connected directly to one of the host's network interfaces (via macvtap) or bridge devices ( Since 0.9.4)
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 'virsh net-dumpxml [networkname]'. 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 192.168.122.0/255.255.255.0. Each guest will have an associated tun device created with a name of vnetN, which can also be overridden with the element (see Overriding the target element).
When the source of an interface is a network, a portgroup 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. Since 0.9.4.
When a guest is running an interface of type network may include a portid 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. Since 5.1.0
Also, similar to direct network connections (described below), a connection of type network may specify a virtualport element, with configuration data to be forwarded to a vepa (802.1Qbg) or 802.1Qbh compliant switch ( Since 0.8.2 ), or to an Open vSwitch virtual switch ( Since 0.9.11 ).
Since the actual type of switch may vary depending on the configuration in the on the host, it is acceptable to omit the virtualport type attribute, and specify attributes from multiple different virtualport types (and also to leave out certain attributes); at domain startup time, a complete 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. ( Since 0.10.0 ). 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 profileid (in case the switch is 802.1Qbh) and an interfaceid (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 ). 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.
...
...
...
Bridge to LAN
This is the recommended config for general guest connectivity on hosts with static wired networking configs.
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 attached. The guest VM will have an associated tun device created with a name of vnetN, which can also be overridden with the element (see Overriding the target element). The tun device will be attached to the bridge. The IP range / network configuration is whatever is used on the LAN. This provides the guest VM full incoming & outgoing net access just like a physical machine.
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 to the interface definition. ( Since 0.9.11 ). The Open vSwitch type virtualport accepts two parameters in its element - an interfaceid 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 profileid which is sent to Open vSwitch as the interfaces "port-profile".
...
...
...
...
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 to the interface definition. ( Since 1.2.13 ). The Midonet virtualport type requires an interfaceid attribute in its element. This interface id is the UUID that specifies which port in the virtual network topology will be bound to the interface.
...
...
...
...
Userspace connection using SLIRP
The user interface type connects the guest interface to the outside via a transparent userspace proxy that doesn't require any special system privileges, making it usable in cases when libvirt itself is running with no privileges (e.g. libvirt's "session mode" daemon, or when libvirt is run inside an unprivileged container).
By default, this user proxy is done with QEMU's SLIRP driver, a userspace proxy built into QEMU that has DHCP & DNS services that give the guest an IP address of 10.0.2.15, a default route of 10.0.2.2 and DNS server at 10.0.2.3.
Since 3.8.0 it is possible to override the guest's default network address by including an ip element specifying an IPv4 address in its one mandatory attribute, address. Optionally, a second ip element with a family attribute set to "ipv6" can be specified to add an IPv6 address to the interface. address. Optionally, an address prefix can be specified.
...
...
...
These settings are surprisingly not used by SLIRP to set the exact IP address; instead they are used to determine what network/subnet the guest's IP address should be on, and the guest will be given an address in that subnet, but the host portion of the address will still be the host portion of "10.0.2.15" (based on the configured prefix (or a prefix of 24 if no prefix is specified). The DNS and default gateway addresses given to the guest will be similarly based on the network portion of the configuration-provided combined with the host portion of SLIRPs default settings for DNS/gateway (10.0.2.3/10.0.2.2). To help resolve the confusion of the previous sentences, the table below shows examples of the settings that will be provided to the guest (via a DHCP response) to use for its interface config (ip/prefix, DNS, default gateway) for various settings of element address and prefix in libvirt's config:
libvirt element
guest ip/prefix
guest DNS
guest default gateway
(unspecified)
10.0.2.15/24
10.0.2.3
10.0.2.2
address='172.17.1.1' prefix='16'
172.17.2.15/16
172.17.2.3
172.17.2.2
address='172.17.1.1' prefix='24'
172.17.1.15/24
172.17.1.3
172.17.1.2
address='172.17.1.1' prefix='8'
172.0.2.15/16
172.0.2.3
172.0.2.2
address='172.17.1.1' prefix='23'
172.17.0.15/23
172.17.0.3
172.17.0.2
Userspace connection using passt
Since 9.0.0 (QEMU and KVM only) an alternate backend implementation of the user interface type can be selected by setting the interface's subelement type attribute to passt. In this case, the passt transport (details here) is used. passt is run as a separate process from QEMU - the passt process handles the details of forwarding network traffic back and forth to the physical network (using userspace proxies and a separate network namespace to provide outgoing UDP/TCP/ICMP sessions, and optionally redirecting incoming traffic destined for the host toward the guest instead), and a socket between passt and QEMU forwards that traffic on to the guest (and back out, of course).
Since 11.1.0 (QEMU and KVM only) you may prefer to use the passt backend with the more efficient and performant type='vhostuser' rather than type='user'. All the options related to passt in the paragraphs below here also apply when using the passt backend with type='vhostuser'; any other details specific to vhostuser are described here.
Similar to SLIRP, passt has an internal DHCP server that provides a requesting guest with one ipv4 and one ipv6 address. There are default values for both of these, or you can use the element (described above, with behavioral differences as outlined below) to configure one IPv4 and one IPv6 address that passt's DHCP server can provide to the guest.
Unlike SLIRP, when no address is specified, passt will by default provide the guest with an IP address, DNS server, etc. that are identical to those settings on the host itself (through the magic of the proxies and a separate network namespace, this doesn't create any conflict).
Also different from SLIRP's behavior: if you do specify IP address(es), the exact address and netmask/prefix you specify will be provided to the guest (i.e. passt doesn't interpret the settings as a network address like SLIRP does, but as a host address). In the table of examples given above, the guest IP would be set to exactly 172.17.1.1 in all cases (the DNS and default gateway will be set the same as they are on the host).
Just as with SLIRP, though, once traffic from the guest leaves the host towards the rest of the network, it will always appear as if it came from the host's IP.
There are a few other options that are configurable only for the passt backend. For example, the subelement's attribute logFile can be used to tell the passt process for this interface where to write its message log (since 9.0.0)[*], while the hostname attribute is used to set the hostname sent to the guest in a DHCPv4 response (using option 12) (since 11.8.0), and fqdn sets the "fully qualified domain name" sent to the guest in DHCPv4 response option 81 and DHCPv6 response option 39 (since 11.8.0). Also, the subelement attribute dev can tell passt a particular host interface to use when deriving the routes given to the guest for forwarding traffic upstream.
[*] Due to the design decisions of passt, when using SELinux on the host, it is recommended that the log file reside in the runtime directory of the user under which the passt process will run, most probably ``/run/user/$UID`` (where ``$UID`` is the UID of that user), e.g. ``/run/user/1000``. Be aware that libvirt does not create this directory if it does not already exist to avoid possible, however unlikely, issues with orphaned directories or permissions, etc. The logfile attribute is meant mostly for debugging, so it shouldn't be set under normal circumstances.
Additionally, when passt is used, multiple elements can be added to forward incoming network traffic for the host to this guest interface. Each must have a proto attribute (set to tcp or udp), optional original address (if not specified, then all incoming sessions to any host IP for the given proto/port(s) will be forwarded to the guest), and an optional dev attribute to limit the forwarded traffic to a specific host interface.
The decision of which ports to forward is described with zero or more subelements of (if there is no then all ports for the given proto/address will be forwarded). Each has a start and optional end attribute. If end is omitted then a single port will be forwarded, otherwise all ports between start and end (inclusive) will be forwarded. If the port number(s) should remain unmodified as the session is forwarded, no further options are needed, but if the guest is expecting the sessions on a different port, then this should be specified with the to attribute of - the port number of each forwarded session in the range will be offeset by "to - start". A element can also be used to specify a range of ports that should not be forwarded. This is done by setting the range's exclude attribute to yes. This may not seem very useful, but can be when it is desirable to forward a long range of ports with the exception of some subset.
...
...
...
Generic ethernet connection
Provides a means to use a new or existing tap device (or veth device pair, depending 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).
The name of the tap device can optionally be specified with the dev attribute of the 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 dev is a pre-existing device, and the managed attribute of the target 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. Since 5.7.0 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.
After creating/opening the tap device, an optional shell script (given in the path attribute of the